Empirical And Analytical Advances

Studies in the History of the English Language IV: Empirical and Analytical Advances in the Study of English Language Change

Edited by Susan M. Fitzmaurice Donka Minkova

Mouton de Gruyter

Studies in the History of the English Language IV

≥

Topics in English Linguistics 61

Editors

Elizabeth Closs Traugott Bernd Kortmann

Mouton de Gruyter Berlin · New York

Studies in the History of the English Language IV Empirical and Analytical Advances in the Study of English Language Change Edited by

Susan M. Fitzmaurice Donka Minkova

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Studies in the history of the English language IV : empirical and analytical advances in the study of English language change / edited by Susan M. Fitzmaurice , Donka Minkova. p. cm. ⫺ (Topics in English linguistics ; 61) Includes bibliographical references and index. ISBN 978-3-11-020587-9 (alk. paper) 1. English language ⫺ History. 2. English language ⫺ Grammar, Historical. I. Fitzmaurice, Susan M. II. Minkova, Donka, 1944⫺ III. Title: Studies in the history of the English language 4. IV. Title: Empirical and analytical advances in the study of English language change. PE1075.S885 2008 420.9⫺dc22 2008040565

ISBN 978-3-11-020587-9 ISSN 1434-3452 Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. ” Copyright 2008 by Walter de Gruyter GmbH & Co. KG, 10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Christopher Schneider, Berlin. Typesetting: ptp ⫺ berlin gmbH, Berlin. Printed in Germany.

Richard Hogg was the UK ambassador to the first SHEL meeting at UCLA in 2000. As the grand-marshal of the just finished 6-volume Cambridge History of the English Language, he brought the accumulated wisdom, good will, and best wishes of the ICEHL community to North America. Later he continued his enthusiastic involvement with the SHEL series as a presenter and reviewer, and only his sudden death in September 2007 stopped him from attending SHEL-5. As the invited commentator for the Old English section of the SHEL-4 collection he was characteristically prompt, incisive, and generous in supporting the enterprise. We dedicate this volume to his memory with deep appreciation of his lasting contributions to our field, of his collegiality and friendship.

Table of contents

Dedication to Richard M. Hogg

v

Tabula Laudatoria

ix

Introduction: Heuristics and evidence in studying the history of the English language Susan Fitzmaurice and Donka Minkova

1

Triggering events William Labov

11

What’s new in Old English? Richard M. Hogg

55

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose to investigate the syntax-pragmatics interface Elizabeth Closs Traugott and Susan Pintzuk

61

Anglian dialect features in Old English anonymous homiletic literature: A survey, with preliminary findings R.D. Fulk

81

The elusive progress of prosodical study Thomas Cable Fidelity in versification: Modern English translations of Beowulf and Sir Gawain and the Green Knight Jennifer Anh-Thu Tran Smith

101

121

Response to Tom Cable’s comments Jennifer Anh-Thu Tran Smith

153

Metrical evidence: Did Chaucer translate The Romaunt of the Rose? Xingzhong Li

155

Trochees in an iambic meter: Assumptions or evidence? Xingzhong Li

181

viii

Table of contents

“Ubbe dubbede him to knith”: The scansion of Havelok and Middle English -es, -ed, and -ed(e) Christina M. Fitzgerald

187

A response to Tom Cable Christina M. Fitzgerald

205

Patterns and productivity David Denison

207

Borrowed derivational morphology in Late Middle English: A study of the records of the London Grocers and Goldsmiths Chris C. Palmer

231

Fixer-uppers and passers-by: Nominalization of verb-particle constructions Don Chapman

265

Words and constructions in grammaticalization: The end of the English impersonal construction Graeme Trousdale

301

Variation in Late Modern English: Making the best use of ‘bad data’ Joan C. Beal

327

English/French bilingualism in nineteenth century Louisiana: A social network analysis Connie Eble

337

Taking permissible shortcuts? Limited evidence, heuristic reasoning and the modal auxiliaries in early Canadian English Stefan Dollinger

357

‘What strikes the ear’: Thomas Sheridan and regional pronunciation Raymond Hickey

387

Author index

413

Subject index

423

Tabula Laudatoria

Richard Hogg Cindy Allen Leslie Arnovick Joan Beal Ricardo Bermúdez-Otero Douglas Biber Derek Britton Thomas Cable Don Chapman Claire Cowie Anne Curzan David Denison Teresa Fanego Maurizio Gotti Jonathan Hope Raymond Hickey Patrick Honeybone Marianne Hundt

Dieter Kastovsky Christian Kay Ans van Kemenade Peter Kitson Bill Kretzschmar Ian Lancashire Katie Lowe Emma Moore Lynda Mugglestone Christian Mair Derek Pearsall Malcolm Richardson Nikolaus Ritt Jeremy Smith Robert Stockwell Graeme Trousdale Ilse Wischer Wim van der Wurff

Introduction: Heuristics and evidence in studying the history of the English language*

Susan Fitzmaurice and Donka Minkova This volume continues the project of initiating and energizing the conversations among historians of the English language fostered by the series of conferences on studying the history of the English language (SHEL), begun in 2000 at UCLA. It follows in the footsteps of three high-profile SHELbased collections of peer-reviewed research papers and point-counterpoint commentaries. In the current volume, we invited our contributors to reflect upon their approaches and practices in undertaking historical studies, focusing particularly on the methods deployed in selecting and analyzing data. The essays in this volume represent interests in the study of linguistic change in English that range across different periods, genres, and aspects of the language and show different approaches and use of evidence to deal with the subject. They also represent the current state of research in the field and the nature of the debates in which scholars and historians engage as regards the nature of the evidence adduced in the explanation of change and the robustness of heuristics. We approach the history of the English language from different perspectives. One of us (DM) works on phonology, morphology, and meter, principally in Old and Middle English, while the other (SF) works on grammatical and semantic-pragmatic change, principally in seventeenth and eighteenth century English. Despite these different orientations, however, we share a strong interest in examining the evidence that informs and grounds research in our fields at the same time as interrogating the heuristics employed by our colleagues for the histories they present. The contributions to the volume give expression to these interests. Our first contribution is an essay by SHEL-4 plenary speaker William Labov that explores the nature of what he calls ‘triggering events’. His principal concern is to identify the immediate ‘triggers’ of changes in vowel systems that tend to be represented as chain shifts. His empirical foundation rests both on historical sound changes in English and on data drawn from speakers whose dialects have provided the basis for the

2

Susan Fitzmaurice and Donka Minkova

detailed account of regional accents provided in the Atlas of Northern American English (ANAE). At issue in this article is the nature of ‘causes’ as opposed to triggers: the difference, in Labov’s view, is that “there are bends in the chain of causality at which triggering events are located. Around the bend there are further chains of causality, but they are often orthogonal to the question that drives the original search”. The “triggering events”, as we understand them, can be system-internal, driven by factors such as dispersion or functional load, and externally motivated. The paper demonstrates ANAE’s potential to reveal new relationships between ongoing sound changes and throw new light on the long-standing discussion of the distinction between “proximate mechanisms” and “causal explanations” (Minkova 1999), a central concern for historical linguists. The rest of the volume is organized into four sections, partly along chronological lines, and partly following the topics of the contributions. For each section, we invited a colleague whose own work is related to the topics and approaches represented by the contributions to write an introduction to the papers. The result is interesting conversation within each section with quite different outcomes in terms of our contributors’ responses to the section introductions and the shape and interaction of the essays in that section. Section 1 concerns developments in the study of Old English. In the last completed publication before his untimely death, Richard Hogg’s introduction poses the question, ‘What’s new in Old English studies?’ and finds much in the two studies in the section to help him ponder the answer and conclude optimistically that “[the contributions on Old English] create new avenues to explore and I hope that my own comments have suggested yet other avenues. Old English is alive and well.” For example, the availability of the York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE) has significantly enhanced work on Old English syntax. Recent interest in the syntax-pragmatics interface, most especially with respect to information structure, led Elizabeth Traugott and Susan Pintzuk to address the challenge of building on the syntactically parsed corpus to make frequency counts of these factors possible. Their ultimate aim is to enable as detailed an account as is feasible of Old English information-structure, including the function, status, and frequency of clause-initial elements. Their more immediate purpose in their contribution is to discuss the method used in coding for information status and to exemplify some of the potentials of adding this kind of information. Their methodology builds on the notion of antecedence (Prince 1992, Birner and Ward 1998), which identifies referents that are “discourse-old” as opposed to “hearer-old” rather than just “given”. This

Introduction

3

allows for a considerably more fine-grained account of Topicalization and Left-dislocation than has been used in the past, and paves the way for expanding the coding to other nominals and for further detailed comparisons of information structure not only with other stages of English but also with other dead languages available to us only through manuscripts. The premise of Rob Fulk’s paper “Anglian dialect features in Old English anonymous homiletic literature: A survey, with preliminary findings” is the observation that there is much disagreement and confusion about the cause of the admixture of seemingly Anglian dialect features frequently found in Late West Saxon prose. The phenomenon has been attributed alternately to diatopic variation within Wessex, stylistic considerations, and Anglian origins for the greater part of the corpus of late Old English prose (the last possibility perhaps having implications for the dating of the relevant texts). The stylistic explanation, which continues to hold considerable sway, is very plausible only in regard to homiletic works, in which the elevated tone associated with Anglian dialect characteristics would be appropriate. Homilies (including saints’ legends), however, comprise most of the relevant corpus of late anonymous prose, and so uncertainty in regard to these texts is the greatest obstacle to explaining the phenomenon. As a preliminary step in the research required to resolve these issues, then, the corpus of anonymous homilies, amounting to about a quarter of a million words, is surveyed to determine the incidence and distribution of 54 distinctive Anglian (or non-West Saxon) features, including 26 items of vocabulary, and the results are presented for each of the nearly 150 texts. The results justify the preliminary conclusions that (1) reliable Anglian features are commoner in anonymous homilies than has generally been recognized, (2) their incidence shows considerable variability from one text to the next, and (3) features of different linguistic types (phonological/graphemic, morphological, syntactic, and lexical) invariably co-occur in texts that display more than a few Anglian characteristics. Settling the larger question about the cause of seeming dialect mixture will depend upon the close analysis of individual texts. The three studies in Section 2 bring the evidence from older verse into focus. Meter can be a friend or a foe; verse evidence is sometimes deliberately disregarded in syntactic reconstruction. Smith, Li, and Fitzgerald show how verse can be a friend in the undergraduate classroom, in scholarly debates about authorship, and in reconstructing language change. Thomas Cable surveys these contributions with a critical eye; his essay “The Elusive Progress of Prosodical Study” is an excellent example of how

4

Susan Fitzmaurice and Donka Minkova

problematic the area of English historical metrics can be. Here a wellestablished scholar takes issue with both the technical details and the larger conceptual problems in the work of younger scholars who inhabit a parallel theoretical universe. The field of English historical metrics is notoriously divisive, and Cable outlines clearly the historical demarcation between literary and generative prosodists; this in itself is a good lesson for the next generation of English historical metrists. Although the combative tone has gone from our exchanges and we do not start our statements with “I repeat regretfully, respectfully, but peremptorily and irrevocably, that it is impossible to argue with persons who say that …” (Saintsbury 1923: 145, n.), the passions and sometime unbridgeable differences are still there. As volume editors, we take a detached stance; we believe that all four contributions in this section add much to our understanding of the interaction between meter, language, and literature. They offer insights into the advantages and the problems of developing new research methodologies in historical reconstruction. There is clearly both room and need for more work that will further and deepen our knowledge of the older verse traditions, their literary and linguistic settings, and their integration into linguistic theory. Jennifer Smith’s paper “Fidelity in Versification: Modern English Translations of Beowulf and Sir Gawain and the Green Knight” attempts to assess the extent to which modern translators have been faithful in their reproduction of the elements of Old and Middle English alliterative verse. In place of the sense of holistic fidelity that most scholars use to evaluate translations of Old and Middle English verse, she adopts an approach comparing frequencies. Through the examination of six present day English translations of Beowulf and seven present day English translations of Sir Gawain and the Green Knight she is able to conduct an objective or independent assessment of the fidelity of one aspect of verse translation. Her method of analysis consists of measuring how much the scansion of the Modern English poetry violates or deviates from the Old and Middle English standards. She defines alliterating violations as failure to alliterate and alliteration in the wrong position; positional violations are defined as deviations from the four (or five) stressed line of alliterative verse. Ultimately, what she finds is that the attitude of the translators, their perspective on the text, what they see as most salient, be it the sound or the pacing, the alliteration or the epic themes, permeate the Modern English versions entirely. In exploring how modern translators have made choices in their own translations, she seeks to clarify some of the prosodic qualities of OE, ME, and ModE, as well as to contribute to the awareness that the collective power of

Introduction

5

many translations are in and of themselves powerful forces of literary work both in their ability to mediate their source text and in their ability to revitalize linguistically inaccessible texts. Xingzhong Li’s study in this section re-examines the long-standing controversy about the authorship of the ME poem The Romaunt of the Rose translated from the original French poem the Roman de la Rose. Over a century, Chaucerian scholars have used a range of evidence of diction, rhyme, grammar of Chaucerian English, and other information in determining the authorship. The accepted view, dating back to 1900, has been that Chaucer translated only the first 1,705 of the 7,692 lines of the poem; this view has dominated the literature on The Romaunt authorship. Li’s study adopts a comparative approach and exploits new and independent metrical evidence to test the authorship as well as the claims of earlier scholars. The findings strongly support the accepted view that Fragment A of the poem is by Chaucer while Fragment B is not, but they disagree with the centennial hypothesis on the authenticity of Fragment C. In the fourth study involving metrical evidence, Christina Fitzgerald tackles the question of when Middle English -es, -ed, and -ede ceased to be syllabic. Following G. V. Smithers’ (1983) study of scansion and the use of ME -en and -e in Havelok the Dane, Fitzgerald uses scansion of the same early fourteenth century poem – a text with authorial origins in Lincolnshire and soon after copied in Norfolk – to elicit information about the pronunciation of -es, -ed, and -ede. Her study is concerned with scansion and metrical stress as tools to unlock linguistic information rather than with reconstructing all of the prosodic features of the poet’s language. Her analysis shows that 32.67 % of regular and unambiguous metrical environments of -es, -ed, and -ede in the poem show syncopation. Though the pronunciation of inflectional endings outweighs their syncopation by a ratio of approximately 3 : 2, the numbers make it clear that syncopation is by no means an infrequent or isolated phenomenon. Fitzgerald concludes that in the early fourteenth century East Midlands dialect of the Havelok poet, the inflectional endings -es, -ed, and -ede had begun to lose their syllabicity, though they could still be used for metrical purposes in alternating stress poetry. Section 3 of the volume represents the diversity of current approaches to morphosyntactic change in English. David Denison’s introduction “Patterns and Productivity” explores how the three contributions approach the question of the patterns that speakers draw upon in lexical and morphosyntactic innovation, assessing the relative productivity of particular patterns.

6

Susan Fitzmaurice and Donka Minkova

His essay raises important questions about the extent to which researchers can make reasonable inferences about patterns on the basis of different kinds of evidence – from the immensely rich source of the internet to the surviving records of small groups of speakers in particular historical periods to selected examples from the Old English corpus. The three essays themselves adopt rather different perspectives to this question. The methods adopted by the group are more speculative in some respects than they are analytical – Chapman in terms of treating the internet as an open-ended corpus, Palmer in terms of extrapolating from a very small dataset, and Trousdale in terms of being truly speculative. Denison’s essay provides an insightful and energetic critique of their effectiveness and their differences. Chris C. Palmer’s paper “Borrowed Derivational Morphology in Late Middle English: A Study of the Records of the London Grocers and Goldsmiths” seeks to characterize the morphological status of several derivational suffixes borrowed from French and Latin within the records of two communities in the fifteenth century. It compares the use of native nominal affixes (-ness, -ship, and -hood) with borrowed, potential affixes (-cion, -ance, -ity, -age, and -ment) throughout the English portions of these multilingual texts. Attempting to locate evidence of the naturalization of these forms – the process by which these endings become derivational morphemes in the general English lexicon – Palmer develops the notion of local productivity. This measure combines both quantitative and qualitative data to show that, even in smaller corpora, historical linguists can find evidence of the morphological status of different potential affixes for communities within particular historical moments. Palmer finds that despite variation in the use of -age between the two communities the majority of borrowed potential affixes were in the early stages of naturalization. They had limited productivity within a restricted subset of the lexis, and speakers were beginning to see them as individual units. Ultimately, Palmer argues that diachronic studies should consider such data to better understand the social and linguistic mechanisms that may have led to the increasing productivity of borrowed derivational affixes in the English language. The -er suffix has been one of the most productive derivational suffixes in English, regularly forming nominalizations of verbs like farmer, teacher and writer. The verb + particle construction, like wake up or find out has also been long established in English. But the combination of these two constructions, that is the -er suffixation of a verb + particle construction has not been as well established. There are four possibilities, namely patterns like on-looker, looker-on, looker-onner, and look-onner, and in his study

Introduction

7

“Fixer-uppers and passers-by: Nominalization of verb-particle constructions”, Don Chapman examines the distribution of each of these in the history of English. This sort of examination is difficult to conduct using standard corpora because these nominalizations retain an ad hoc or ludic feel to them and rarely show up in published writings. The Oxford English Dictionary (OED) and the internet present two tools for finding less established constructions like this one. The results of examining the distribution of these nominalizations in the OED are that the pattern looker-on has been the most widely used, followed by the pattern on-looker. Both however dip sharply in the twentieth century, the same century that the pattern lookeronner began to appear. The results of Chapman’s internet search confirm the trend of the twentieth century in the OED, namely that looker-onner has increased in usage while on-looker and looker-on have decreased. He observes that looker-onner appears to be the most productive form by far, but looker-on continues to be used, while on-looker has become only minimally productive. The form that did not occur in the OED, look-onner, is used as much as looker-on in the internet searches. Chapman speculates that if this represents a trend, perhaps one day look-onner will supplant looker-onner. This trend would present a progression of looker-on > looker-onner > look onner, such that the suffix gradually moves from the verb where it properly belongs to the end of the term, where suffixes usually go, with an intermediate stage in which the suffix occurs in both positions. Graeme Trousdale’s article “Words and constructions in grammaticalization: the end of the English impersonal construction” is concerned with the remnants of the impersonal construction in early Modern English, and specifically with the role of grammaticalization in the development of both this construction and the transitive construction in English. An informative outline of the history of the impersonal construction up to and including the early Modern period is followed by a discussion of impersonal and transitive constructions using some of the theoretical apparatus from cognitive linguistics in general, and Cognitive Grammar in particular. Finally, he accounts for the change in terms of patterns of grammaticalization, developing some of the arguments of Meillet (1958 [1912]) whose work is traditionally cited in discussions of the grammaticalization of lexical items, but who in fact also considered changes in word order and phrase structure as potential instances of grammaticalization. The article proposes that constructional accounts of language structure can both inform and be informed by grammaticalization theory.

8

Susan Fitzmaurice and Donka Minkova

Section 4 concludes the volume. In this section, three authors provide different perspectives on aspects of variation in late modern English (LModE). The title of Joan Beal’s introduction “Variation in Late Modern English: making the best use of ‘bad data’” echoes Roger Lass’ (1990) famous “How to do things with junk”. She provides a cogent research context for examining the complex question of how to extract answers from evidence that is scanty and holey. The papers by Eble and Dollinger point up the problem very clearly as they deal with varieties of English that have not been studied from a sociohistorical perspective. In contrast, Hickey scrutinizes the evidence for assessing the extent to which the pronouncements of an influential commentator such as Sheridan might shape language history. Connie Eble’s essay “English/French bilingualism in nineteenth-century Louisiana: A social network analysis” applies the notion of social network to an archive of family papers to explore why French disappeared as the language of public life in northwestern Louisiana by the time of the Civil War. The social network itself and the language practices of a cohesive group of white creoles living in rural Natchitoches Parish during the nineteenth century are inferred from the writings preserved in the Prudhomme Family Papers. For a half century, dense and strong ties of kinship preserved French as the language of the descendants of the founder population, while restricting it increasingly to the domains of personal life and religious practice. At the same time, weaker connections favoured the addition of English for communication outside the creole network that aided economic prosperity. Sending members of the younger generation beyond the local area to learn to speak, read, and write English made knowledge of English a property of their ties to each other and made them conduits of linguistic change to their creole network. Eble finds that by the beginning of the twentieth century, French had disappeared entirely from the Prudhomme Family Papers and presumably from the lives of the creole network of the area. Stefan Dollinger asks provocatively: “Progressive colonial English?” The paper deals with a ‘bad’ data problem specific to colonial Englishes in the Late Modern English period. Considering the complex sociolinguistic situation of newly-formed colonial varieties in the LModE period, Dollinger argues, in light of the present suboptimal resources, for the adoption of heuristic methods of approximation, of ‘good-enough’ estimates that have proved useful in other disciplines. These methods may provide a feasible shortcut for English historical linguistics in general, but particularly for the

Introduction

9

characterization of colonial varieties. The approach is illustrated by the semantic development of CAN and MAY in Ontario English in terms of their progressive, respectively conservative, behaviour in comparison to British English, before being applied to a larger set of modal auxiliaries. In the LModE colonial context, the limitations of statistical testing are discussed and a solution is suggested by combining LModE findings, based on limited data, with twentieth-century findings on the modal auxiliaries, which allows the assessment of features of colonial varieties with a certain degree of confidence. Raymond Hickey’s paper “‘What strikes the ear’: Thomas Sheridan and regional pronunciation” concludes the volume. The study turns our attention to the contemporary description of late modern English by considering the role of the elocutionist and grammarian Thomas Sheridan in the rise of sociolinguistic censure. He examines Sheridan’s attitude to non-standard features in both southeastern British English and Irish English in the late eighteenth century to track how prescriptive notions of language use seemed to be fleshed out during this time. He looks in some detail at what present researchers might glean about the nature of regional pronunciations in the late modern period from Sheridan’s negative comments on the speech of his fellow Irishmen. Finally, he also considers the possible influence of Sheridan’s strictures on the development of Irish English during the nineteenth-century. We close with acknowledgments and thanks to the individuals and organizations that helped the progress of the conversations captured in this volume. The expertise of our reviewers (named in the Tabula Laudatoria) informed the work performed by new and established scholars alike in developing their contributions.

Notes *

The term heuristics is understood in different ways in humanities. In our understanding, heuristics refers to methods of discovery, which may include empirical, analytical and speculative methods.

10

Susan Fitzmaurice and Donka Minkova

References Birner, Betty J., and Gregory Ward 1998 Information Status and Non-Canonical Word Order in English. (Studies in Language Companion Series 40.) Amsterdam/Philadelphia: John Benjamins. Lass, Roger 1990 How to Do Things with Junk: Exaptation in Language Evolution. Journal of Linguistics 26.1: 79–109. Meillet, Antoine 1958 Linguistique historique et linguistique général. Paris. Champion. [1912] Minkova, Donka. 1999 Proximate mechanisms vs. causal explanations, Zeitschrift für Sprachwissenschaft 18.2: 226–230. Prince, Ellen F. 1992 The ZPG letter: Subjects, definiteness, and information-status. In Discourse Description: Diverse Analyses of a Fundraising Text, William C. Mann and Sandra A. Thompson (eds.), 295–325. (Pragmatics and Beyond, New Series 16.). Amsterdam/Philadelphia: John Benjamins. Saintsbury, George 1923 A History of English Prosody. Vol. I: From the Origins to Spenser. London: Macmillan and Co. Limited. Smithers, G.V. 1983 The scansion of Havelok and the use of ME –en and –e in Havelok and by Chaucer. In Middle English studies presented to Norman Davis in honour of his seventieth birthday, Douglas Gray and E.G. Stanley (eds.), 195–234. Oxford: Clarendon Press.

Triggering events

William Labov There is general agreement that the heart of the study of language change is the search for causes.1 It is what we generally mean by the explanation of change. And while we would like to apply to this search the universal principles that govern grammar as a whole, it is also understood, following Meillet (1921), that no universal principles can account for the sporadic course of change, in which particular changes begin and end at a given time in history. The actuation problem (Weinreich, Labov and Herzog 1968) demands that we search for universals in particulars. However, the pursuit of the causes of any given change might on further reflection involve us in an unsatisfactory and endless recursion. It goes without saying that any given state of a language is the outcome of a previous state of that language, and so on backward in time as far into the past as our knowledge can carry us. The title of this chapter then needs some justification if it refers to linguistic events. In such an endless chain of causes, every state of the language is a triggering event for the one that follows. Even if there is no change in a given system, it has a cause: the state of equilibrium that was reached in the preceding period. And when there is change, as Martinet (1955) has argued, the evolving system reflects a series of earlier readjustments that spiral backward in time. I would like to defend the concept of “triggering event” by arguing that this sequence of preceding causes is not a smooth and uniform sequence. Rather, there are bends in the chain of causality at which triggering events are located. Around the bend there are further chains of causality, but they are often orthogonal to the question that drives the original search (Figure 1). A nonlinguistic example may illustrate the point. We are all interested in the pre-history that gave rise to mammalian evolution, and in this causal sequence we encounter the extinction at the K-T boundary between the Cretaceous and Tertiary periods of the dinosaurs, along with plesiosaurs, mosasaurs, and a majority of all other existing families. What caused this massive extinction? The most strongly supported theory is that of Luis

12

William Labov

and Walter Alvarez, originally proposed in the 1980s: that the K-T extinction was the result of the impact of a large meteor with the earth. While the exact killing mechanisms may or may not yet have been identified, all the data – including the rate of extinction, the nature of the recovery, and the patterns of survivorship – are concordant with the hypothesis of extinction by asteroid impact. (Fastovsky and Sheehan 2004).

What were the causes of this intersection of asteroid and earth? It is an important question for the future of the human race, which would be profoundly influenced by a major impact of this kind. The answer would involve calculations of celestial mechanics that are not immediately relevant to the later history of biological evolution. The hypothesis of a meteor impact, if it continues to be supported, provides a satisfactory answer to the question, what was the triggering event that gave rise to mammalian predominance in the evolutionary sequence? The linguistic triggering event that we are looking for may also be the result of a variety of factors concatenated by historical accident.

Figure 1. A bend in the chain of causality

Chain shifts are a natural subject for the study of causal sequences, and the search for triggering events. Table 1 lists six chain shifts studied in the Atlas of North American English (Labov, Ash and Boberg 2006; henceforth ANAE). The most recent events are listed on the right, and the events preceding them in the two columns on the left. Some shifts link two or three events, one shows five. In each case we are led to ask, what was the triggering event that was responsible for this shift? We might think, again following Martinet, that this triggering event must be an external event impinging on the linguistic process, like the Norman invasion or World War II, outside of the realm of autonomous lin guistic explanation. For some shifts, this is the case.2 But in others, it will

Triggering events

13

Table 1. Six chain shifts described in ANAE. The Canadian Shift (Ch. 15) The Pittsburgh Shift (Ch. 19) The Northern Cities Shift (Ch. 14) The Southern Shift (Ch. 18) The Back Upglide Shift (Ch. 18) Back Chain Shift before /r/ (Ch. 19)

Entering → /e/ → /æ/ → /¡ / → 

/o/

Leaving→ /o/

→

→

→

/¡ / →

/oh/ →

/o/

/iy/ →

/ey/ →

/ay/ →

/ah/

/nw/ →

/aw/ →

/ahr/ →

/ohr/ →

/e/



→

/æ/ →

appear that there are linguistic bends in the chain of causality. I will argue that there are triggering events of a purely linguistic character. Their explanation calls upon a different set of principles than those that operate on the changes they initiate. First however I’d like to show that bends in the linguistic chain are essential characteristics of chain shifts. In fact, without such shifts of direction it will be difficult to defend the very concept of a chain shift. Consider the simplest kind of chain shift (1)

BĺAĺ

A is the leaving element and B is the entering element following the notation of Table 1.3 A causal connection might be said to exist if A moves away because B approached A, reducing the margin of security, or if A moved away, increasing the margin of security, and B consequently moved in the direction of A. However, such chain shift events are subject to an alternative interpretation. The movement of A may be generalized to B, just as the change of a front vowel may be generalized to the corresponding back vowel without any relevant change in margins of security. In (2), if A is a vowel /e/ moving in the vowel space from mid to high, and B is a low vowel /æ/ that moves from low to mid behind it, one could argue that the movements of A and B are causally related. But this can also be conceived as a single expression (3) in which all front vowels undergo a loss of one

14

William Labov

degree of openness. Thus whatever factor C acted on /e/ to make it less open came to act equally on /æ/ so that the causal relationship is seen as (4) rather than (1). (2)

(3)

A

eĺi

B

æĺe

Į open ĺ Į -1 open / _____ [+ant]

(4)

C

B

A

Option (4) is not available if A and B are different kinds of linguistic processes. Thus in the Southern Shift, A is the monophthongization of /ay/ and B is the lowering and centralization of the nucleus of /ey/ (ANAE: Ch. 18), as represented in (5). In A, /ay/ is a vowel from the subsystem of front upgliding vowels that moves to the system of long and ingliding vowels, while B is an adjustment entirely within the set of front upgliding vowels. (5)

A

ay ĺ ah4

B

ey ĺ ay

Here we must accept a chain shift of type (1) since there is no single process that can be generalized to unite the behavior of A and B. The causal relationship seems clear: the removal of /ay/ from the front upgliding system led to a readjustment by the well recognized principle of maximal dispersion – the tendency of vowels to achieve equidistant positions within a subsystem (Martinet 1955, Liljencrants and Lindblom 1972, Disner 1980, Lindblom 1988). Figure 2 sums up the characterization of these two situations: generalizable shifts within a subsystem, and sequential shifts across subsystems. The type of causal explanation applied to chain shifts is not in question here. In this search for triggering events, one may take a teleological posi-

Triggering events

15

Figure 2. Generalizable and sequential chain shifts across subsystems.

tion, like Martinet (1955) or Jakobson (1972), and argue that speakers shift their vowels to minimize misunderstanding. Or one can attribute these linked movements to the mechanical effects of misunderstanding on the probability matching of the language learner (Labov 1994, Ch. 20). Evidence for the causal link may come from temporal sequencing, geographic nesting, or internal correlation (ANAE Chs. 14, 18). However, the order of events is crucial to the present discussion: whether we are dealing with a drag chain or a push chain will be decisive in the search for triggering events. Subsystems of English vowels. Much of the logic of chain shifting involves movements out of and into subsystems. The binary notation of ANAE Ch. 2, is designed to characterize these subsystems in a coherent and systematic manner. Table 2 shows the four subsystems of North American English: short vowels, front upgliding vowels, back upgliding vowels, and the smaller set of long and ingliding vowels.5 In r-pronouncing dialects, this fourth subset consists primarily of the /ah/ class in father, ma, pa, pajama, etc., and /oh/ in law, talk, off, cloth, etc. In r-less dialects, the marginal members of this subset /ih/, /eh/, /uh/ are greatly expanded to include beer, bare, boor, etc. The notation does not describe the set of contrasts in any one dialect, but rather the initial position from which present-day dialects can be derived. In that sense, the individual units are historical word classes comparable to the lexical key words presented in Wells (1982).6 The principles of maximal dispersion and maintenance of margins of security developed in Martinet (1955) operate within subsystems (see also Liljencrants and Lindblom 1972, Lindblom 1988). Labov (1994, Ch. 9) presented data from misunderstandings in spontaneous speech which show that confusions occur primarily within members of a subsystem, rather than across subsystems. There is for example more confusion between /i/ and /e/

16

William Labov

than between /e/ and /ey/, and more between /ey/ and /ay/ than between /ay/ and /aw/. Table 2. ANAE notation for four subsystems of vowels of North American English, with type words. LONG SHORT

Front upgliding

Back upgliding

Vy

Vw

V nucleus high

mid

low

1.

front

back

Ingliding

Upgliding

front

back

front

Vh back

/i/

/u/

/iy/

/iw/

/uw/

bit

put

beat

suit

boot

unrounded

rounded

/e/

/¡/

/ey/

/oy/

/ow/

/oh/

bet

but

bait

boy

boat

bought

/æ/

/o/

/ay/

/aw/

/ah/

bat

cot

bite

bout

balm

The Canadian Shift

The first of the North American chain shifts in Table 1 is the Canadian Shift as in (6). (6)

/e/ → /æ/ → /o/ →

This chain shift was first described by Clarke et al. in 1995 on the basis of word lists read by 16 college students, and has since been confirmed by several further studies of Canadian English (ANAE Ch.16; Boberg 2004). It is the most consistent marker of the Canadian English dialect in ANAE, and it is the basis for the isogloss defining the Canada region (including all points in Canada outside of the Atlantic Provinces). Figure 3 compares the Canada dialect region with the combined means for all others for the vowels involved in the shift. There is no notable difference for /i/. But the Canadian /e/ is significantly lower than the general mean, and an even greater difference appears for /æ/ in both F1 and F2. One can also observe that Canadian /o/ is well back of the general mean.

Triggering events

17

Figure 3. Mean values of vowels in the Canadian Shift for the Canada region [N = 25] and all other dialects combined [N = 414]. Source: ANAE Ch. 15]

It was clear from the outset that the Canadian shift of the short front vowels was a response to the low back merger of /o/ and /oh/ in cot and caught, Don and dawn, etc., well established in Canada. (6a) is therefore a more complete representation of the Canadian Shift. /oh/ is not a leaving element, but collapses with /o/. (6a) /e/ → /æ/ → /o/ → /oh/ To which subsystem do we assign the collapsed vowel phonologically? The decision is dictated by phonological facts. While the original short-o was a checked vowel, which cannot occur in stressed word-final position, the merged vowel occurs in free as well as checked position: that is, the vowel of cot is now an allophone of the vowel of caw. Though both vowels may shift position in the course of the merger,7 it is /o/ that moves to the long and ingliding subsystem rather than /oh/ to the short subsystem. Figure 4 embeds the Canadian Shift in the acoustically defined phonological space characteristic of the Germanic language family, with a peripheral region enclosing a nonperipheral region.8 By the principles of chain shifting developed in Labov (1994, Ch. 5–6), tense or long vowel nuclei rise along the peripheral track and lax or short nuclei fall along the nonperipheral track. A shift from a short to a long subsystem appears as a movement from a nonperipheral to a peripheral track, as indicated in Figure 4. The remaining

18

William Labov

short vowels then readjust their positions along the nonperipheral track to achieve maximal dispersion.

Figure 4. Shift of subsystems in the Canadian Shift

The temporal relations of the low back merger and the Canadian shift are consistent with the causal assignment to the merger as prior. As noted above, the first report of the shift of /e/ and /æ/ date from 1995. The low back merger in Canada was firmly documented well before then (Scargill and Warkentyne 1972, Gregg 1957). Chambers (1993: 11–12) cites literary sources for the merger in the middle of the 19th century. The geographic distribution of the Canadian Shift and the low back merger are also consistent with the causal connection inferred; here we encounter the nesting relation that plays an important role in the application of dialect geography to historical sequencing. Figure 5 maps the distribution of ANAE subjects who satisfy the acoustic criteria for the Canadian Shift (grey symbols), and the isogloss that defines the region in which these symbols predominate. The homogeneity of this isogloss – the proportion of speakers within the area who satisfy the criteria – is .84. Twenty-one of the 25 Canadians within the isogloss do so, producing a more reliable definition of the Canadian dialect than Canadian raising, the best known stereotype of Canadian English (ANAE, Ch. 15). However, consistency – the proportion of speakers who show the trait who are within the isogloss – is quite low, since the same forces are operating wherever the low back merger is found. The implicational relation between the Canadian Shift and the low back merger is evident in that only three of the 60 speakers who show the Canadian Shift have /o/ and /oh/ distinct. The important geographic relation is that the Canadian Shift isogloss is strictly contained within the low back merger isogloss (the oriented line on Figure 5). The low back merger extends to a much wider territory, covering the West, Western Pennsylvania and Eastern New England. A total of 123 speakers

Triggering events

19

produced /o/ and /oh/ the same in minimal pair tests, and only 60 showed the back shifting of /e/ and /æ/. At the same time, the Canadian Shift does appear among a minority in other low back merger areas: twelve in the West, five in Western Pennsylvania, four in Texas, where the merger is reported in progress (Bailey et al. 1991); and seven in the Midland where the merger is generally in transition. However only two grey symbols appear within the dashed isoglosses: these outline the areas of greatest resistance to the merger: in the Inland North, the Mid-Atlantic States and the South.

Figure 5. Nesting of Canadian Shift within the Low Back Merger isogloss

Both temporal and spatial evidence reinforce the general principles of chain shifting to indicate that the low back merger creates the conditions for the backing of /æ/ and accompanying backing and lowering of /e/. In removing /o/ from the subset of short vowels,9 it acts as the triggering event for the Canadian Shift.

20

2.

William Labov

The Pittsburgh Shift

ANAE reports for the first time a chain shift in the city of Pittsburgh, as indicated in (7). As in (6a), we add the third element /oh/, indicating the low back merger that is missing in Table 2. (7) /¡ / → /o/ → /oh/ 

Figure 6 presents the Pittsburgh Shift in the same framework as Figure 4. The low back merger is solidly entrenched in Pittsburgh, as it is in Canada. But in Pittsburgh, the phoneme /¡/ moves downward on the nonperipheral track from mid, back of center position, while /æ/ remains in place in the low front area. Figure 7 provides a detailed view of this downward movement in the vowel system of a 35-year-old man from Pittsburgh, interviewed in 1996. On the left, the short-a vowels follow the nasal system: words with nasal codas are raised to mid and upper-mid position, while all others are in a tight cluster in low front position. In the back, /o/ is clearly merged with /oh/ in the same lower mid back position as in Canada. Between /æ/ and /o/~/oh/ are located the majority of the tokens of /¡ /. Words with /¡ / before /n/ are particularly low (sun, fun, months), but the token of duck is regularly judged to be dock by speakers of other dialects. Figure 8 places this Pittsburgh development against the mean values of the low vowels of Canada and 18 other North American dialects.10 It can be seen that the mean /æ/ of Canada is the furthest back of all dialects, while Pittsburgh /æ/ is in normal low front, unraised position.11 At right, both Canada and Pittsburgh show the merger of /o/ and /oh/ in lower mid back position (the two Canada tokens practically coincide). In the center, the Pittsburgh mean for /¡ / is much lower than any other dialect, not far from the general /o/ distribution. 





Figure 6. Shift of subsystems in the Pittsburgh Shift

Triggering events

21

Figure 7. The Pittsburgh Shift in the vowel system of Kenneth K., 35 [1996], TS 545.

Figure 8. Mean positions of low vowels for 20 ANAE dialects, with Canadian Shift labeled for Canada [CA], Pittsburgh Shift labeled for Pittsburgh [PI] and Northern Cities Shift labeled for the Inland North [IN].

22

William Labov

The low back merger is evidently the conditioning event for the Pittsburgh Shift, just as it is for the Canadian Shift. Here, however we have the same cause with two different effects. In the search for causes of linguistic change, it seems reasonable to expect that the same cause will have similar or comparable effects. Why is it that /¡ / instead of /æ/ moved into the empty space created by the back shift of /o/ and its merger with /oh/? Among North American sound changes, there are other cases of two neighboring phonemes competing to fill the empty space in the pattern.12 One might say that these are two equally likely possibilities, and it is a matter of chance which was realized. But these choices are not equiprobable: there are 60 communities which show evidence of the Canadian Shift and only one city with the Pittsburgh Shift.13 To account for the unique Pittsburgh development, it is not unreasonable to turn to the other unique feature of the Pittsburgh dialect: the monophthongization of /aw/. The Pittsburgh long monophthong in down, town, south, out and house is located in low central position, partially overlapping with /¡ /. There is no danger of confusion between /¡ / and /aw/, however, since monophthongized /aw/ (now /ah/) has twice the length of /¡ /, so that typically the longest /¡ / is shorter than the shortest /aw/ (ANAE p. 273). One hypothesis is that the lowering of /¡ / is the result of a change in the organization of the vowel system of Pittsburgh speakers in which /¡ / is re-analyzed as /a/, the short counterpart of /ah/. This would oppose the long and short pairs down ~ dun, about ~ but, howl ~ hull as /dahn ~ dan, ‹baht ~ bat, hahl ~ hal/. If further evidence supports such an abstract re-analysis, then both the low back merger and the monophthongization of /aw/ appear to be triggering events for the Pittsburgh Shift. Both are movements of word classes into the long and ingliding subsystem from other subsystems. 













3.

Causes of the low back merger

Given our understanding of the effect of the low back merger on other linguistic events, the question that naturally arises is: what are the causes of the low back merger? Herold (1990, 1997) has provided a convincing social account of the actuation of the low back merger in Eastern Pennsylvania – the influx of large numbers of immigrants from Eastern Europe into coal-mining communities. However, no linguistic mechanism for a substratum effect has yet been staked out, and the inquiry we are conducting here calls for a much more general solution. We must account for the linguistic

Triggering events

23

antecedents of the collapse of /o/ and /oh/ in more than half of the North American continent with a variety of vowel systems, and in Scotland as well. Why then is the distinction between /o/ and /oh/ so likely to collapse? If there is a linguistic answer to this question, then the low back merger is not the triggering event we are looking for, but it is only a link in the causal chain. A first thought about the cause of a merger is the functional load of the distinction. In the case of /o/ and /oh/, there is no problem in finding minimal pairs. We can generate sizeable numbers in the style of (8). (8)

cot rot tot sot cotter dotter Don yon pond fond hock stock

caught wrought taught sought caught her daughter dawn yawn pawned fawned hawk stalk

cock tock odd nod cod mod sod Sol moll collar holler odd ability

caulk talk awed gnawed cawed Maud sawed Saul maul caller hauler audibility

This proliferation of minimal pairs masks, however, the odd skewing in the distribution of /o/ and /oh/ that appears in Table 3. Almost all of the contrast between /o/ and /oh/ is before a set of five apical consonants /t,d,s,n,l/ and one non-apical /k/, as indicated by the bold lines. There is no contrast before labials or palatals. Occurrences of /o/ before /z/ are limited to special lexical items and words in which intervocalic /s/ is voiced. In the lower half of Table 3, there are six environments where /oh/ is not represented at all, and one – final position – where /o/ does not appear. Three sets of /oh/ words in Table 3 are italicized. These are /o/ words that are tensed in American English before front voiceless fricatives and nasals, the same core phonetic conditioning that operates in the tensing of short-a in the Mid-Atlantic region and broad-a in Britain (Ferguson 1975, Labov 1989).14 This tensing process typically proceeds by lexical diffusion, but does not substantially increase contrast between /o/ and /oh/. There are a total of six environments in which one side or the other is represented by a small number of learned, colloquial or specialized vocabulary, so that in

24

William Labov

twelve environments, contrast is marginal; monosyllabic and minimal pairs are not to be found.15 Table 3. ANAE notation for four subsystems of vowels of North American English, with type words. /o/

/oh/

APICALS t

cot, tot, hot, got, dot

caught. bought, taut, fought

d

odd, hod, god, sod

awed, hawed, gaud, sawed

s

toss, moss, floss, cost, loss

sauce, exhaust, caustic

z

(Oz, positive)

cause, clause, hawser, pause, paws

n

don, Ron, pond

dawn, awn, yawn, lawn

l

doll, moll, collar

all, tall, maul, caller

NON-APICALS p

hop, pop, top, sop

b

rob, hob

F 

Scotch, botch, watch

----------

j

lodge, dodge, Roger

----------

g

log. hog, cog, dog

(auger, augment, augur, August)

k

stock, hock, clock

stalk, hawk, talk

f

(boff, toff)

→ off, doff, scoff (cough, trough)

7

(Goth)

→ cloth, moth

6

(gosh, bosh)

(wash)

'

(bother)

] 

----------

m

bomb, Tom, prom

1

(pong, Kong)

#

----------

---------(daub, bauble)

---------------------------→ strong. song, wrong, strong law, saw, flaw, thaw, claw

Triggering events

25

Figure 9. Historical development of the long open-o word class

In order to see how this bizarre distribution came about, it may be helpful to review the historical formation of this word class, as summarized schematically in Figure 9. Proceeding from left to right, the diagram shows 1. an original /aw/ diphthong in Old English (thaw, straw, claw) 2. additions to Old English /aw/ through Old English sound changes a. breaking and rounding of strong verb preterits before velars in complex codas (fought, taught) b. vocalization and rounding of /l/ in complex codas (talk, call, all) c. vocalization of coda /g/ (maw, saw, draw) 3. additions to /aw/ in Middle English through vocalization of /v/ (hawk, laundry) accretion of new /aw/ forms from Old French loan words a. original OF back upgliding diphthongs (applaud, because) b. collapse of bisyllabic /a + u/ words to single syllables (pawn, brawn) c. denasalization and rounding of nasal vowels (lawn, spawn) 4. smoothing (monophthongization) of /aw/ to /oh/.

26

William Labov

5. lengthening of /o/ to /oh/ in Early Modern English before voiceless fricatives and velar nasals (cloth, off, loss, lost, strong, song, wrong, long) 6. lexically irregular rounding of /a/ after /w/ (water, warrant, walrus) The O.E. /aw/ class traced here is not a reflex of PGmc /aw/, which is realized in Old English as ƝƗ. It was cobbled together by a series of conditioned sound changes so that its shape is a matter of historical accident. The general sound change that set the stage for the low back merger was the smoothing of ME /aw/ to /oh/.16 It must have taken place before the shift of Middle English /o/ to /oh/ by compensatory lengthening in thought and brought. We can also argue that it must have also preceded the completion of the Great Vowel Shift in the back vowels, by which ME u: diphthongized to /aw/. The smoothing of /aw/ created the juxtaposition of /o/ and /oh/ – two lower back mid vowels differentiated only by length,17 which is unstable on two counts. First, it is well established that length distinctions without accompanying differences in vowel quality tend to collapse, in English and many other languages (Chen and Wang 1975). Second is the asymmetrical distribution of Table 3. Given this situation, the merger of the opposition is a likely outcome unless qualitative differences develop to support it. Such qualitative differentiation of /o/ and /oh/ did develop in three areas outlined by the dotted isoglosses of Figure 5 (ANAE Ch. 11): (1) the unrounding and fronting of /o/ in Western New England and New York State;18 (2) the raising of /oh/ to upper mid position (east coast dialects from Providence to Baltimore); (3) restoration of the back upglide of /oh/ in the South.19 Outside of these areas, the low back merger is either complete or in transition. It follows that the juxtaposition of long and short open-o by the smoothing of /aw/ was the triggering event of the low back merger. What is the relationship of the other events of Figure 9 to the low back merger? The /aw/ class originated in final position, where it could not contrast with short open /o/. The changes that followed were largely conditioned by the vocalization of /l, g, x/ in complex codas before /k/, /l/, /t/. They created the limited contrasts which resisted the merger to a certain extent; however, one would have to say that it was the absence of sound changes conditioned by other consonants that favored the merger. If the smoothing of /aw/ was the triggering event for the low back merger and ultimately the Canadian Shift and the Pittsburgh Shift, we must ask if it in turn had a relevant predecessor. I argued that it must have pre-

Triggering events

27

ceded the completion of the Great Vowel Shift on the assumption that it was a drag chain. But it is also possible that a push chain was involved, and that the descending diphthong [8u] → [ԥu] → in out, south, down, etc. reduced the margin of security of /aw/ realized as [#u] in a way that promoted the shift to [n:]. If that is the case, we would have to push our inquiry into the triggering event of the Great Vowel Shift, a question that has been much discussed (Luick 1903, Martinet 1955, Stockwell and Minkova 1997). There is not enough evidence to pursue this connection here, except to emphasize the possibility of a chain of linguistic triggering events receding into the past. In any case, there is no reason to believe that any one external event intervened to produce these chain shifts.

4.

The fronting of /uw/

In the two cases just studied, the low back merger was seen to set the conditions for subsequent changes in the vowel system, responding to the tendency of subsystems to maintain equidistant spacing or maximal dispersion. We will now consider a sound change that appears to be inconsistent with previous explanation based on these principles. This is the fronting of /uw/, an ongoing shift that covers 90% of the North American continent. The various phonetic forms involved are shown in (9). (9)

u

7

u

Œ

üu

ü

7A +

ü

u

Martinet (1955) advanced an explanation for what is now recognized as a general principle of chain shifting: that back vowels move to the front.20 He argues that the repeated fronting of /u/ and /o/ is the result of the fact that even though there is a strong tendency to front-back symmetry in the vowel system, there is physically less room in the back than in the front. Such fronting is then the result of pressure to relieve overcrowding among the back vowels. Specifically, this happens when through one linguistic process or another, a vowel system develops four degrees of height among the back vowels. Haudricourt and Juilland (1949) applied this logic to a wide range of sound changes in western Europe and confirmed Martinet’s prediction in every case. Labov (1991), defining three major dialects of English, argued that the third dialect, characterized by the low back merger, would be sta-

28

William Labov

ble, and resist the fronting of /uw/ and /ow/ that is predominant in the South and the Midland. Figure 10 shows that the completed ANAE data does not satisfy this expectation. The grey symbols identify speakers for whom /uw/ after coronal consonants – in do, dew, too, two, soon, noon, etc. – is front of center, that is, mean F2 is greater than the midpoint of 1550 Hz in this normalized system. This includes 89% of the population: there are only 49 of the 439 ANAE subjects for whom this is not the case. Furthermore, these 49 are concentrated in two narrowly circumscribed areas: New England and the Minnesota-Wisconsin. In general, Eastern New England is a conservative area in regard to the fronting of /uw/ and /ow/, and its behavior is consistent with what we would expect from the low back merger in that area. The Minnesota-Wisconsin area shows considerable variation in regard to the low back merger. But the conservative character of the vowel system, with back /uw/ and /ow/ often monophthongal, must be accounted for by a strong Scandinavian and German substratum (Allen 1973).

Figure 10. Fronting of /uw/ after coronal consonants. Grey symbols = Second formant > 1550 Hz.

Triggering events

29

Once we have dispensed with these two areas, we are faced with the fact that /uw/ is fronted without exception in all other regions: in the Midland, the Mid-Atlantic States, in the South, and most importantly in three areas where the low back merger is complete: Canada, The West, and Western Pennsylvania. It is not possible to account for this massive, continentalwide fronting as a response to overcrowding among the back vowels. Although the structural approach to the causes of /uw/ fronting in North America seems to fail in this case, we can open an inquiry into the causes of this phenomenon from another structural direction. Because /uw/ fronting is so widespread in North America, it is unlikely that we will find a specific population movement like the migration of Slavic coal miners into Eastern Pennsylvania identified by Herold (1990). The antecedent event must be one of great generality. One clue to the problem may be found in the extraordinary difference between the fronting of /uw/ after coronal consonants, examined in Figure 10, and the same word class after non-coronal consonants in roof, boots, coop, food, move, etc. While 390 ANAE subjects shifted /uw/ after coronal consonants front of center, only 130 did so for the non-coronal class. Table 4 (columns 2 and 3) demonstrates this extraordinary effect of coronal onset in a regression analysis of all 4,747 tokens of /uw/ measured acoustically. The age coefficient in Table 4 indicates vigorous change in progress in apparent time. The figure –101 in column 1, row 1 indicates that the expected value of F2 for speakers 25 years older than the mean is 101 Hz less than the general mean of F2 for /uw/, all other things being equal. For the generation 25 years younger than the mean age, the fronting of /uw/ is advanced by 101 Hz. As in most sound changes in progress, women lead: in this case by the effect of half a generation. Among the internal constraints, the effect of a preceding coronal stands out at 480 Hz, more than twice the effect of any other. This means that for the average speaker with a mean F2 for /uw/ after coronal consonants of 1800 Hz, the value of /uw/ after noncoronals is around 1300 Hz, half way between a back and a center vowel. This preponderant effect of preceding coronals is a striking exception to the general rule that English vowels are influenced by the following environment much more than the preceding one.21 It is not difficult to explain the tendency for preceding coronals to promote the fronting of /uw/, which is a widespread effect. It appears strongly in Lennig’s (1978) analysis of sound change in progress in Paris. The F2 locus of apical consonants ranges closely around 1800 Hz, so that when a following back /uw/ requires

30

William Labov

Table 4. Regression coefficients for F2 of /uw/ and /ow/ for all of North America. Vowels before /l/ excluded.a Age * 25 years represents the age coefficient times 25. /uw/ [N=4747] Coefficient Constant

Probability

1547

/ow/ N=6736] Coefficient

Probability

1386

SOCIAL Age * 25

–101

<.0001

–24

<.0001

42

<.0001

46

<.0002

Coronal

480

<.0001

94

<.0001

Velar

181

<.0001

43

<.0001

Liquid

151

<.0001

–

n.s.

Obstruent+Liquid

164

<.0001

–

n.s.

Labial

104

<.0001

–70

<.0001

Nasal

–54

.0020

None

–

n.s.

31

<.0003

Coronal

70

<.0001

–

n.s.

Nasal

–193

<.0001

–101

<.0001

Fricative

–137

<.0001

–21

.0023

–89

<.0001

–39

<.0002

40

.0095

–

n.s.

–

n.s.

–75

<.0001

Female PHONETIC Onset

Coda

Stop Voiced Following syllables a

Vowels before /l/ are excluded, since outside of the South, they are in extreme back position, and even in the South, fronting is quite limited. Because there many powerful effects, only those with a probability < .01 are shown.

Triggering events

31

a rapid transition of 1000 Hz from that locus to F2 of the vowel nucleus. Articulatory ease will favor the raising of this second formant. If sound change begins to front /uw/, allophones after coronals will be in advance of others. Yet the size of this effect – 480 Hz – is more than one would expect from a phonetically motivated influence. One way of evaluating the coronal effect on /uw/ is to compare it to the coronal effect on the fronting of the mid-back vowel /ow/. This parallel fronting is not as widespread as the fronting of /uw/, but is vigorously in progress throughout the Midland, the South and the Mid-Atlantic States (ANAE Ch. 12). The right hand side of Table 4 presents the age coefficients for /ow/. To ensure comparability for phonetic effects, all regions of North America are included, even though there is no active fronting for about half the population. The coefficients for /ow/ are therefore generally lower, since for the regions where there is very little fronting, the effects are much smaller. In general, the effects on /uw/ and /ow/, both external and internal, are in the same direction. The point of interest is the relation of coefficient for preceding coronals to other effects on /ow/. While the /uw/ coefficient is 2½ times greater than any other, the /ow/ coefficient comparable to other phonetic effects, and less than the influence of following nasals. If the effect of a preceding coronal on /uw/ was the result of the same mechanism as the /ow/ effect, we would expect it to be only 20% greater, since the distance between second formants and the apical locus for extreme back /ow/ is only 20% greater – 1000 Hz as opposed to 800 Hz. It follows that mechanical effects are not likely to account for the 480 Hz coronal coefficients for /uw/. It seems likely that this is a phonological effect, not a phonetic one. The search for phonological effects leads us to the /yuw/ class of high rising diphthongs, which is historically quite distinct from the falling /uw/. The /yuw/ class was derived from a variety of different sources (Jespersen 1949: 3.8). – – – – – – –

OE iw as in Tiwesdæg ‘Tuesday’ OE e:ow as in e:ow, ‘you’ French iu, as in riule ‘rule’ French unstressed e+u, as n seur ‘sure’ French u, as in rude, French. ui, as in fruit French iv, as in OF sivre -> M.E. sewe, ‘sue’

32

William Labov

In modern English, these seven were joined by an eighth, which was distinct in Middle English –

OE e:a as in de:aw, ‘dew’

Although some scholars believe that this vowel was once equivalent to French front rounded [y], Jespersen argues that it has consistently been a rising diphthong /ju/, which in terms of ANAE notation is /yuw/. The /y/ glide is generally maintained after labials and velars, except in Norfolk and a few other sites in England (Trudgill 1974, 1986). In North America, the glide has long been variable after apicals. In many cities, it became a marker of refined speech and varied according to the preceding context: the probability of a /j/ glide is greatest after /t/ in tune, etc. and least after /l/ and /r/ in lewd and rude (where it is also frequently deleted in British English).22 The development of the /yuw/ class is closely aligned to the problem under study. In current North American English, the historical /y/ glide has all but disappeared after coronal consonants in tune, dew, suit, stupid, etc. In the middle of the 20th century, Kurath and McDavid (1961) found widespread use of the glide after coronals in the South, while the characteristic Northern form was [iu], an unrounded front vowel moving back towards a high back target (see also Kenyon and Knott (1953) who represent this vowel generally as [iu]). This vocalic realization set up the contrast indicated in Table 2 as /iw/ vs. /uw/, exemplified by such minimal pairs as dew and do, lute and loot, tutor and tooter. ANAE (Ch. 8) investigated the contrast with the minimal pair dew ~ do, and mapped both word classes in spontaneous speech as well. Figure 11 shows that the distinction has almost disappeared in North America. It is mainly confined to two limited areas in the South: one in central North Carolina, the other in the smaller cities of the Gulf States. Where the distinction is found, it is almost always in the first mora as a vocalic nucleus: [KW] vs. [U
Triggering events

33

Figure 11. Retention of the /iw ~ /uw/ contrast in North America. Grey symbols and solid isogloss: speakers with /iw/ and /uw/ distinct in production and perception of minimal pair tests. Dashed isoglosses enclose communities where acoustic measurements show a significant difference between /iw/ and /uw/ in spontaneous speech. Solid isogloss defines the South as the area where /ay/ is monophthongal before obstruents.

Figure 12 shows the most conservative dialect in regard to the fronting of /uw/ and /ow/: Providence, Rhode Island. (In this and the diagrams to follow, /Tuw/ indicates /uw/ after coronals, and /Kuw/ after noncoronals.) Here the means for all vowels are back of center, including /iw/ in stupid and Tuesday. The vowels after non-coronals are further back, not far from the bench mark of vowels before /l/ (not included in the calculation of /Kuw/ means). Figure 13 shows more advanced fronting in three different patterns. Typical for the North, Canada and West is Figure 13a, the /uw/ and /iw/ vowels of a speaker from Alberta. The mean for /Tuw/ is more than 2000 Hz, well front of the center mark of 1550 Hz, and there is no differentiation of /Tuw/ and /iw/. But the mean of Kuw in roof, boots, etc. is well back of center, lower than 1400 Hz. This differentiation by 500 Hz is the phonetic

34

William Labov

Figure 12. High back upgliding vowels of a conservative speaker from Providence, Rhode Island: Alex S., 42 [1996], TS 47

realization of the regression coefficient of 480 Hz in Table 4. Figure 13b, a speaker from Lexington, Kentucky, is a fully fronted system, where /iw/, /Tuw/ and /Kuw/ are indistinguishable, in high front rounded position, some 900 Hz fronter than /uw/ before /l/. Figure 13c shows the high vowels of a speaker from Charlotte, NC, who maintains the distinction between /iw/ and /uw/. The /iw/ class in new, dew, Tuesday, Duke, shoes is tightly clustered around a mean at 400, 2094 Hz, while /Tuw/ shows an equally tight cluster at 493,1789 Hz. Both F1 and F2 differences are significant at the .001 level. The fact that /Tuw/ is only slightly front of center indicates that the distinction between /iw/ and /Tuw/ is maintained only by inhibiting the fronting of /Tuw/. In other words, the merger of /iw/ and /Tuw/ is necessarily correlated with the full fronting of /Tuw/. Table 5 allows us to compare the means, age and coronal onset coefficients of /uw/ for eight major ANAE dialects. The regional mean values show that the South and the Midland are the most advanced and the North the least advanced. The array of negative age coefficients indicate that all dialects except the Mid-Atlantic are engaged in change in progress in apparent time, but the size of the age gradient varies widely. Though the South is advanced in fronting, the age coefficient is quite low, and most notably, the coronal onset coefficient is only a small fraction of that found for other dialects. It is less than a third of the coefficient for the equally advanced Midland dialect, reflecting the Southern tendency to retain the /iw/ ~ /uw/ distinction. The fully fronted /Kuw/ in Figure 13b reflects the general merger of /iw/ with /uw/ as a whole, even though /iw/ has no allophones in common with /Kuw/. The phonological effects of this merger are comparable to the phonological effect of the merger of /o/ and /oh/, and (in the discussion to follow), the merger of /o/ and /ah/.

Triggering events

35

a) Differentiation of /uw/ after coronals (Tuw) and noncoronals (Kuw): Brent M., 25 [1997], Edmonton, Alberta, TS 654

b) Consolidation of /iw, and /uw/ in front position: Fay M., 34 [1995]. Lexington, KY, TS 283

c) Maintenance of /iw/ ~ /uw/ distinction: Charlotte, NC. Matthew D., 45 [1996], Charlotte NC TS 483

Figure 13. Three fronting patterns of the high back upgliding vowels

36

William Labov

Table 5. Regression analyses of F2 of /uw/ of vowels not before /l/ by region. All coefficients significant at p < .0001 level.

Midland South ENE Mid-Atlantic Western Pa West Canada North

N 580 1107 116 190 161 468 521 1062

Mean F2(uw) 1713 1703 1584 1534 1529 1520 1492 1359

Age*25 –107 –86 –244 –119 –76 –155 –83

Coronal Onset 442 141 456 338 362 469 514

Figure 14 traces the history of these developments in a single overview, showing the origins of the /yuw/ class and its eventual merger with /uw/ in the course of the fronting process. As in the case of the low back merger, there is no external triggering event, but rather a series of interconnected changes across a long history. As with /oh/, historical accidents led to the formation of a highly skewed and marginal contrast of /iw/ vs. /uw/. It is proposed here that the triggering event for the fronting of /uw/ is the collapse of the /iw/~/uw/ distinction. That distinction was the result of the loss of /y/ after coronals. Moving forward in time, we begin with the very general allophonic effect of /y/ on the following high vowel as in (10). The triggering event was the loss of the conditioning factor (11) which created the opposition /iw/ vs. /uw/ by secondary split. The collapse of this opposition (12) led to a general fronting (13). (10) (11) (12) (13)

[+high] → [-back] / y ____ w /y/ → 0 / [+cor] _____ [+high] → [-back] /[+cor] _____ w [+high] → [-back] / ______w

The triggering event (11) is one of the many deletions of the “peripheral phonemes” of modern English, /w, h, y/ (Vachek 1964).23 It is not likely that any further inquiry into the causes of the loss of this glide will illuminate our understanding of the fronting of /uw/ in North America.

Triggering events

37

Figure 14. Development of /uw/

5.

The Northern Cities Shift

The Northern Cities Shift [NCS] was first described in 1972, and its various stages have been traced by a number of scholars (Labov, Yaeger & Steiner 1972, Labov 1991, Eckert 1999, Gordon 2001). ANAE (Chapter 14) shows that it is the dominant vowel system of the Inland North, a territory of some 88,000 square miles with 34,000,000 speakers. The rotation of six vowels (five originating in the short subsystem) is notable given the relative stability of the English short vowel system from Old English to the present day. Figure 15 displays the ordering of events that is most generally accepted, although some points of order are still open to question.24 Table 1 suggested two sets of causal relationships. Working backwards in time towards a triggering event, we observe that the most recent shift is backing of /¡ /, to the point that Inland North bus can sound like boss in other dialects. This seems to be a joint response to two preceding events: the increased margin of security in the back caused by lowering of /oh/ and a decrease in the 

38

William Labov

Figure 15. Northern Cities Shift with generally accepted ordering (ANAE Ch. 14)

margin of security caused by the backward shift of /e/ towards the /¡/ region, so that Inland North desk sounds like dusk to speakers of other dialects. The major change in /i/ is backing towards a central position, which seems to be a generalized response to the backing of /e/. The lowering of /oh/ appears to be entailed by the fronting of /o/. This in turn is generally accepted to be the result of the vacancy in the low front area following the general raising and fronting of /æ/. One causal link is missing from this account: what is responsible for the lowering and backing of /e/? As noted in the discussion of the Pittsburgh Shift, it can happen that the shift of a given vowel can attract two neighboring vowels into the region vacated. In this case, early evidence indicates that /e/ first moved downward into the low central area vacated by /æ/ at the same time that /o/ moved forward, creating considerable overlap of /e/ and /o/ for many Inland North speakers (Labov and Baranowski 2006). Although this overlap has continued, the predominant tendency in the following decades was for /e/ to shift back towards /¡/ (Eckert 1999).25 The current situation in the ANAE records of the 1990s can be seen by returning to Figure 8, where the means for the Inland North are labeled against the background of 20 other dialects. The IN mean for /æ/ (not including vowels before nasals) is higher and fronter than any other /æ/ means. A corresponding shift is seen for the IN mean for /o/, which is considerably fronter than any other. The diamond representing /e/ for the Inland North is further back than any other, and the IN wedge is at the rear edge of the /¡/ distribution. We do not see a marked lowering of /oh/ in this display, however. In this account of the NCS, the initial event is clearly the general raising of /æ/, marked “1” on Figure 15. The temporal evidence favors this interpretation. The earliest records from the 1960s show both fronting of /o/ and raising of /æ/, but no evidence of the other sound changes (Fasold 1969;

Triggering events

39

Figure 16. Tensing of /o/ and /æ/ as initiating events of the NCS.

Labov, Yaeger & Steiner 1972). The geographic evidence for ordering is not as clear as in the Canadian Shift, since the complexity of the NCS requires that its geographic outlines be established by relations among its components taken pairwise.26 There seems little doubt that the general raising of /æ/ is the triggering event for the NCS. In the spirit of our current inquiry, we ask what in turn triggered the raising of /æ/? Figure 16 displays these six events on the acoustically defined map of phonological space with peripheral and nonperipheral tracks, as in Figures 4 and 6. We observe that /i/ and /e/ fall along the non-peripheral track as general principles predict (Labov 1994).27 The peripheral track is occupied by /oh/, and tense vowels that were originally /o/ and /æ/. The shifting of /o/ to the peripheral track is the result of the logic that was presented in the discussion of the Canadian Shift. In the Inland North, /o/ does not merge with /oh/ but with /ah/ in father, ma, pa, bra, pajama, etc. This merged vowel is best represented as /ah/ since it occurs in both checked and free position. The differentiation of peripheral and nonperipheral tracks for the front and back vowels is entirely a matter of F2 in acoustic terms. However, F1 does not consistently differentiate peripheral/nonperipheral, tense/lax or long/short among the low vowels, as might be suggested by Figure 16. Instead, the low peripheral vowels are largely marked by duration. The phonetic reflex of the phonological status of merged /o/ appears as an increase in mean duration.28 The /æ/ symbol that appears on the peripheral track is not written as /æh/, since it has not merged with any free vowel. Nevertheless, the Inland North /æ/ behaves like a tense vowel. Not only do all tokens undergo raising and fronting, but it has clearly developed a second mora. The Inland North has developed a high frequency of “Northern Breaking” in which the peripheral nucleus shifts abruptly to a full second mora in lower mid position (ANAE Ch. 13), with a considerable increase in duration.

40

William Labov

Figure 17. Natural break map for mean F1 of /æ/ (four ranges) showing range of 445 to 684 Hz. [Vowels before nasal consonants not included].

Although the generalized raising of short-a is found throughout the Inland North, it is unique in the English-speaking world. No other dialect shows such a generalized tensing and raising, affecting even unstressed realizations of the function word that and polysyllables like athletic and animadversion. All other dialects which show raising will differentiate vowels before nasals from others, but in the Inland North this difference is rarely significant. The unique character of this general raising emerges in Figure 17, based on an analysis of F1 of /æ/ in North America into four “natural break” categories.29 The black circles show the category with lowest F1 range: from 445 to 684 Hz. The natural break algorithm automatically isolates the Inland North, with all of the cities around the Great Lakes and the Erie Canal area of New York State, along with the later extension along the corridor to St. Louis, and a small scattering of isolated speakers in the upper regions of the South. While /æ/ is raised and fronted in some contexts by almost all speakers of North American English, some historical process

Triggering events

41

in this particular area has eliminated all contextual conditions, in a process that may be represented as (11). (11) [+low, +ant] → [+tense] The local character of this phenomenon – its heavy concentration in the Inland North – shifts the inquiry to the question of who were the people involved in this event and what were the tensing conditions for the dialects they spoke?

Figure 18. Cities on the Erie Canal

The ANAE maps of the NCS in western New York State focus on the series of cities strung out on a line from east to west in New York State: Schenectady, Syracuse, Rochester, Buffalo. They were founded as small villages by New England settlers in the 18th century and developed as major cities early in the 19th century under the impact of the Erie Canal, shown in Figure 18. The Erie Canal realized an ambitious plan to open a waterway to the west, connecting New York City with the Great Lakes.30 It was begun in 1817 and completed in 1825, with extraordinary economic consequences for western New York State. Before the canal, the cost to ship one ton of goods from Buffalo to New York City was $100; using the canal, it could be shipped for $10 (McKelvey 1949a). The great drop in cost of transportation prompted westward migration and the development of farmland throughout the Inland North. Fresh produce and vast quantities of wheat were shipped to the metropolitan areas of the east and consumer goods were shipped west (McKelvey 1949a, 1949b).

42

William Labov

The consequences for urban growth were correspondingly great. At the end of the War of 1812, Rochester had a population of 331, largely of New England origin. The construction of the Canal required large numbers of laborers, and a number of Irish arrived, forming the section of the city called Dublin. From 1820 to 1830, the population grew from 1507 to 9207 (McKelvey 1949a). The proportion of the population drawn from New England dropped steadily (10% in 1845, 5% in 1855), with the new immigrants from Ireland, Germany and Great Britain. All of the major cities in New York, with the exception of Binghamton and Elmira, fall along the trade route established by the Erie Canal, from New York City to Albany, through Schenectady, Utica and Syracuse, to Rochester and Buffalo. Nearly 80% of upstate New York’s population lives within 25 miles of the Erie Canal. Figure 19 shows that the growth of Rochester followed a logarithmic increase from 1820 to 1930. But this spectacular expansion was small compared to the growth of population in surrounding Monroe County and in the seven neighboring counties from 1810 to 1830, reaching a peak in 1850. This was the type of 10-fold increase that is required to defeat the Principle of First Effective Settlement (Zelinsky 1992) – that the first group arriving in an area sets the cultural pattern for those who arrive later. It is the type of explosive growth that Herold (1990) documented in small towns of Eastern Pennsylvania as the trigger for the low back merger there. But we are not dealing here with coal mining villages in the Allegheny Plateau; a much larger and more diverse migration created the population of upstate New York with at least four types of short-a systems. (1) There is general agreement that the original settlers of New York State were from New England. Current evidence on the short-a class in New England points to the dominance of the nasal system: that is, an allophonic tensing of all /æ/ before nasal consonants in both open and closed syllables, and nowhere else (ANAE Ch. 13, 16). (2) Settlers from Eastern New England introduce the assignment of a variable set of short-a words to the broad-a class: aunt, can’t, half, past, etc. (3) It is evident that the main commercial traffic, freight and passenger, passed to and from New York City, whose new predominance as a port of entry coincided with the opening of the Canal. The New York City short-a is split into two phonemic classes, with tensing in syllables closed by voiced stops, voiceless fricatives and front nasals, along with many grammatical and lexical specifications (Trager 1930, 1934, 1942; ANAE Ch. 13; Labov 2006). (4) We must note

Triggering events

43

Figure 19. Growth of population in western New York State, 1800–1950. Source: McKelvey 1949.

the sudden admixture of large numbers of speakers from Ireland, where short-a is normally low front or low central. The end result of such dialect mixture is very often the formation of a koiné (Trudgill 1986: 107–110), involving leveling (elimination of marked variants) and simplification. Three types of simplifications of these mixed short-a systems are (a) no tensing, as in Britain, (b) the nasal system, and (c) 100 % tensing as in (10), which is in fact what transpired. Though we cannot be certain exactly when this linguistic development occurred, it seems most likely that it happened during this population explosion and before the system was exported to the Great Lakes region in the continued westward expansion. The direction of westward movement can be seen in Figure 20, taken from Kniffen and Glassie’s (1966) study of house building patterns. Here the stream of New England settlement is traced by the construction of frame houses; when log cabins are built, they are temporary and use simple false corner-timbering instead of dove-tailing notches. The lexical and phonological isoglosses that define the North follow the line of Northern

44

William Labov

Figure 20. Westward settlement streams as shown by building material. Source: Kniffen and Glassie 1966: Fig.27.

settlement as precisely as the detail of Figure 20 allows. Figure 21 maps the westward extension of the Northern Cities Shift. The grey symbols (and black isogloss) identify speakers who satisfy the EQ criterion of the NCS (ANAE, Ch. 11, 14). For these speakers the general tensing, raising and fronting of /æ/, combined with the lowering and backing of /e/, has reversed the relationship between these two phonemes found in other dialects. The southern limit of the NCS coincides with the barred black-andwhite isogloss, the division between North and Midland established in Carver (1990): on the basis of 13 lexical oppositions (Northern darning needle vs. Midland snake feeder, belly-flop vs. belly-buster, stone boat vs. mud boat, sawbuck vs. trestle, blat vs. bawl, etc.). The North-Midland line

Triggering events

45

Figure 21. Extension of the Northern Cities Shift to the Great Lakes region of the Inland North. Grey symbols and solid black isogloss identify speakers for whom /æ/ is higher and fronter than /e/. Barred black-and-white isogloss is the lexical line separating North from Midland based on DARE data (Carver 1990:). Dotted isogloss indicates recent extension of NCS to St. Louis.

extends westward from New York State, passes south of the Western Reserve in Ohio, runs close to the northern border of Indiana, and then turns south to include the northern third of Illinois. The lexical features that identify the North are largely rural terms, many obsolete and unknown to city dwellers today. Like the Kniffen and Glassie data, they reflect directly the agricultural practices of the mid-19th century period when the Inland North was settled: clearing land, building stone walls, and framing houses. But as noted above, the earliest evidence of the NCS sound changes dates from the 1960s. If the triggering linguistic event took place during the NY State population expansion of the first half of the 19th century, its effects must have lasted for a century before coming to the attention of linguists. This is not unlikely, if we calculate the time required to reach the present level of /æ/ raising. The initial tensing to the peripheral track actually has the effect of lowering /æ/ in terms of higher F1,31 and from studies of current sound changes in progress we can expect that the raising from low to upper mid position would take three generations32 (Labov 2001). The raising of /æ/ has reached its maximum in this area today,

46

William Labov

as shown by the age coefficients of Table 6, drawn from a regression analysis of vowels from 63 speakers in the Inland North (ANAE Table 16). There is no correlation of age and the first formant of /æ/, even at the p < .10 level of significance. This indicates that this process has been active for some time and has reached its limit. Table 6. Age coefficients for five elements of the NCS in regression analysis by vowel tokens [N=1479] for all Inland North speakers. All figures show younger speakers favoring the change. Coefficient

Probability

First formant of /æ/

–

–

Second formant of /o/

–12

< .05

Second formant of /oh/

–24

< .001

Second formant of /e/

68

< .001

Second formant of /¡/

17

< .10

The specific hypothesis that is supported here is that the triggering event for the general raising of /æ/ was a koiné formation in western New York State in the first half of the 19th century. This event was the result of a variety of contingent historical processes, so that further inquiry into its linguistic antecedents will not materially increase our understanding of the evolution of the Northern Cities Shift. That said, we continue to explore the dialectology of Western New England, the point of origin of the initial settlement of the area, where many of the components of the NCS can be found in an incipient form (Boberg 2001). The match was struck by builders of the Erie Canal, but the timber that burned was grown in New England.

6.

An overview of triggering events

We started with the proposition that a clear demonstration of the causal character of a chain shift required a bend in the chain of linguistic causality. It turned out that there were many such bends in the history of the sound changes in progress in North America. They generally involve the removal of a vowel from one subsystem and its insertion into another. An under-

Triggering events

47

standing of the development of the complex English vowel system requires the concept of the linguistic subsystem, within which the principles of chain shifting and maximal dispersion are defined. Mergers across subsystems play a particularly important role in these developments. Evidence for the reality of the subsystem concept is drawn from both phonological distribution and phonetic differentiation, where the ability to distinguish phonetic from phonological effects is crucial. Some of the triggering events encountered were linked with a chain of other triggering events, receding into the indefinite linguistic past with no obvious break in the chain of causality. The low back merger was linked to the odd composition of long open-o, which has been a source of instability in English for many centuries. In two cases we met sharp discontinuities in the succession of events. The fronting of /uw/ seems to have been triggered by the loss of the initial glide after coronal consonants in the oddly formed /yuw/ class, and though we pursued the consequences of that loss it did not seem fruitful to pursue its antecedents. The social and economic ferment centered on the building of the Erie Canal created sharp linguistic and social discontinuities that triggered the revolutionary chain shifts of the 20th century. No matter how carefully we probe into its parentage, it seems that a new linguistic world was born in western New York State in the first third of the 19th century. To some extent, these findings are conditioned by the complex character of the English vowel system, with its sixteen phonemes far out in the upper tail of the distribution of vowel inventories. Here the organization into subsystems plays a role that can hardly be replicated in the modal type of five vowel languages. But such hierarchical organization is not difficult to find in languages with nasal vowels, glottalized vowels, creaky vowels, long and short vowels, stressed and unstressed subsystems. Bradley (1969) describes elaborate chain shifts within and across glottal-tone and open-tone subsystems in Akha, a Lolo-Burmese language (Labov 1994, Ch. 5). Latvian dialects provide a dazzling array of chain shifts across ingliding, monophthongal, upgliding and short vowel subsystems (Labov 1994, Ch. 5). Britannic chain shifts cross long and short, monophthongal and diphthongal subsystems (McCone 1996). Such hierarchical organization is of course even more common in consonantal systems. The dialectology of the New World offers an attractive opportunity to study linguistic changes in progress. The events I have chronicled here are new sound changes, written on the tabula rasa of the frontier. As we follow their antecedents backward in time we encounter the dialectology and

48

William Labov

language contacts of the Old World, where many layers of intersecting influence accumulate over the centuries. The record is blurred and overlaid, but it is worth deciphering. Tracing history as it is being made is exhilarating, but it is always helpful to know where we came from.

Notes 1. The original form of this paper was given at Studies in the History of English4 (SHEL-4) on October 1, 2005 at Flagstaff, AZ. 2. World War II appears to be the triggering event for the re-orientation of norms of r-pronunciation to favor constricted [r] in New York City (Labov 1966) and elsewhere (Labov, Ash and Boberg 2006). 3. This notation is neutral as to which moves first, that is, whether (1) is a push chain or a drag chain. 4. See Table 2 for this notation 5. The expression “long and ingliding” refers to the fact that members of this set are normally monophthongal in low or lower-mid position, but glide towards a center target when their nucleus is in upper mid or high position. In the course of linguistic change, inglides regularly develop when vowels raise from lower mid to upper mid position. 6. Though Wells’ notation is lexical, the tabular organization he presents reflects the same subclasses as in Table 2 (ANAE Ch. 2). 7 The phonetic position of the merged class varies considerably throughout the merged areas; in Canada it is quite far back and rounded. 8. This display shows the positions of the vowel nuclei only, without tracing the direction of movement that would differentiae the upgliding and ingliding subsystems. 9. See Labov and Baranowski (2006) for evidence that the merger of /o/ with /ah/ leads to a phonologically determined increase in length of approximately 50 msec. 10. The mean values of Figure 8 are all normalized with the log mean normalization (Nearey 1978; ANAE, Ch. 5.7). 11. In all such mean calculations, the mean of /æ/ before nasals is calculated separately from the main distribution. The figures shown here are the means for /æ/ not before nasals. 12. Most notable is the simultaneous downward shift of /e/ and forward shift of /o/ into the low front position vacated by /æ/ in the Northern Cities Shift (Labov and Baranowski 2006). 13. Many of these communities are of course linked historically, and the low back merger is not an independent development in all 60. However, there is good

Triggering events

14. 15.

16. 17.

18.

19.

20.

21.

22.

23.

24.

49

reason to believe that the merger did occur independently in at least five areas: Eastern New England, Canada, western Pennsylvania, eastern Pennsylvania and the West. In the conditioning of the tensing of short-a, the front nasals /m,n/ are selected; in the tensing of short-o, the velar nasal. The examples shown are from my own speech, where all common words before voiceless fricatives and front nasals are tense, but uncommon words like Goth, Gothic, wroth, gosh, bosh, tosh, ping pong, King Kong, ding dong are lax. I use the traditional term ‘smoothing’ here to avoid confusion with the recent monophthongization of /aw/ to /ah/ in Pittsburgh discussed in the last section. In lower mid or low back position, there is no inglide differentiating the two vowels; such an inglide develops only as /oh/ rises to upper mid or high position. To the best of my knowledge, this was first noted by the spelling reformer Michael Barton (1830), who found that his own New York State speech differed from the New England dictionary writers in just this respect. Although it seems possible that the South retained the original /aw/, this does not seem likely since the Southern back upglide appears in lost, often, cloth, which never participated in the original back upglide of Old and Middle English. The shift to an articulatory framework (ANAE Ch. 2) converts this third principle to the first. The fronting of [u] to [ü] is there seen as an example of peripheral vowels becoming less open, parallel to the raising of [a] to [i]. Martinet’s argument can be applied in this framework as well. This is seen here in the effect of following laterals, which are excluded from this analysis, and the effect of following nasals. The influence of preceding nasals and laterals is always much less than the effect of the same environment following. The status of words with initial palatal consonants is not always clear. The coarticulatory effect on /uw/ in choose and shoes may be strong enough to eliminate the difference between this /uw/ and /iw/ after palatals in juice, chew, etc. This use of “peripheral” is of course distinct from its use above in the description of English vowels. Vachek meant that the glides /h, w, y/ were peripheral in lack of phonological integration. The unrounding and fronting of /o/ in Western New England appears to date from the beginning of the 19th century (Barton 1830) and may be almost contemporaneous with the raising of /æ/. The lowering and fronting of /oh/ is quite variable across speakers in many areas, and for some speakers may be the last stage. Though the backing of /¡/ appears as the most recent stage in Eckert (1999), a relatively back position for /¡/ can be observed as far east as Providence and as far west as South Dakota.

50

William Labov

25. The shift in the direction of /i/ shows a parallel re-direction. Early studies of Chicago indicated a strong tendency towards the lowering of /i/ but in more recent studies /i/ shifts back towards a more central position (Labov, Yaeger & Steiner 1972). 26. The EQ criterion for the NCS (ANAE Chs. 11, 14) defines those speakers for whom /æ/ is higher and fronter than /e/, and this necessarily is a superset of dialects in which the raising of /æ/ has reached a mean position above the midline. 27. In an articulatory framework, the backing and lowering of /e/ are both favored by its lax character. 28. In the Inland North, many /o/ tokens overlap with /e/ on the F1/F2 plane, As noted in footnote 9 above, the mean duration of this overlapped /o/ is 50 msec greater than the mean for /e/ (Labov and Baranowski 2006). 29. This function of the Mapinfo program is defined as: “The range breaks are determined according to an algorithm such that the difference between the data values and the average of the data values is minimized on a per range basis.” 30. Since the Hudson River is actually an ocean estuary, ice-free year long, it was a more practical route than the Delaware, which freezes over in the winter. From the time that the Erie Canal was completed, New York City rapidly surpassed Philadelphia as the leading city of the United States. 31. This lowering was in fact noted by Labov, Yaeger & Steiner (1972) in the speech of the oldest New York City speaker, for whom short-a was lengthened 32. A new and vigorous change such as the raising of /aw/ in Philadelphia shows an F1 age coefficient in regression analysis of 3.0 Hz per year. Thus three generations (75 years) will lower F1 by 225 Hz, which would bring a low front vowel with an F1 mean of 800 Hz to an upper mid front vowel with an F1 mean of 575 Hz. The actual time course will be longer, since 3 Hz per year is at the midpoint of the s-shaped curve with maximum velocity of change.

References Allen, Harold 1973 The use of Atlas informants of foreign parentage. In Lexicography and Dialect Geography: Festgabe for Hans Kurath. Harald Scholler and John Reidy (eds.), (Wiesbaden: Franz Steiner Verlag [Zeitschrift für Dialektologie und Linguistik], Beihefte, Neue Folge Nr. 9 der Zeitschrift für Mundartforschung]), 17-2. Bailey, Guy, Tom Wikle and Lori Sand 1991 The Focus of linguistic innovation in Texas. English World-Wide 12(2): 195–221

Triggering events

51

Bailey, Guy, Tom Wikle, Jan Tillery and Lori Sand. 1991 The apparent time construct. Language Variation and Change 3: 241–264. Barton, Michael 1830 Something New, comprising a New and Perfect Alphabet. Boston/ Harvard: Marsh, Capen and Lynn. Boberg, Charles 2001 The Phonological Status of Western New England. American Speech 76: 3–29. Bradley, David 1969 Problems in Akha phonology: Synchronic and diachronic. Unpublished paper. Carver, Craig M. 1990 American Regional Dialects: A Word Geography. Ann Arbor: University of Michigan Press. Chambers, J. K. 1993 ‘Lawless and vulgar innovators’: Victorian views of Canadian English. In Focus on Canada. S. Clarke (ed.). Amsterdam: John Benjamins. Chen, Matthew and William S.-Y. Wang 1975 Sound Change: actuation and implementation. Language 51: 255–81. Clarke, Sandra, Ford Elms and Amani Youssef 1995 The third dialect of English: Some Canadian evidence. Language Variation and Change 7: 209–228. Disner, Sandra 1980 Insights on vowel spacing: results of a language survey. University of California Working Papers in Phonetics 50: 70–92. Eckert, Penelope 1999 Linguistic Variation as Social Practice. Oxford: Blackwell. Fasold, Ralph 1969 A sociolinguistic study of the pronunciation of three vowels in Detroit speech. Unpublished ms. Fastovsky, David E. and Peter M. Sheehan 2000 The extinction of the dinosaurs in North America. GSA Today 15: 4– 10. Ferguson, Charles A. 1975 ‘Short a’ in Philadelphia English. In Studies in linguistics in honor of George L. Trager. M. Estellie Smith (ed.), 259–274. The Hague: Mouton. Gordon, Matthew J. 2001 Small-town values and big-city vowels: A study of the Northern cities Shift in Michigan. Publications of the American Dialect Society 8.

52

William Labov

Gregg, R. J. 1957 Notes on the pronunciation of Canadian English as spoken in Vancouver, B.C. Journal of the Canadian Linguistic Association 3: 20–26. Haudricourt, A. G. and A. G. Juilland 1949 Essai pour une histoire structurale du phonétisme français. Paris: C. Klincksieck. Herold, Ruth 1990 Mechanisms of merger: The implementation and distribution of the low back merger in Eastern Pennsylvania. University of Pennsylvania dissertation. 1997 Solving the actuation problem: Merger and immigration in eastern Pennsylvania. Language Variation and Change 9: 165–189. Jakobson, Roman 1972 Principles of historical phonology. In A Reader in Historical and Comparative Linguistics. A. R. Keiler (ed.), 121–138. New York: Holt, Rinehart and Winston. Jespersen, Otto 1949 A Modern English Grammar on Historical Principles. Part I: Sounds and Spellings. London: George Allen & Unwin. Kenyon, John and Thomas Knott 1953 A Pronouncing Dictionary of American English. Springfield, MA: G. C. Merriam. Kniffen, Fred B. and Henry Glassie 1966 Building in wood in the Eastern United States. Geographic Review 56: 40–66. Kurath, Hans and Raven I. McDavid, Jr. 1961 The Pronunciation of English in the Atlantic States. Ann Arbor: University of Michigan Press. Labov, William 1966 The Social Stratification of English in New York City. 2nd edition. [2006] Cambridge: Cambridge University Press. 1989 The exact description of the speech community: short a in Philadelphia. In Language Change and Variation. R. Fasold and D. Schiffrin (eds.), 1–57. Washington, Georgetown University Press. 1991 The three dialects of English. In New Ways of Analyzing Sound Change. P. Eckert (ed.), 1–4. New York: Academic Press. 1994 Principles of Linguistic Change. Volume 1: Internal factors. Oxford: Basil Blackwell Publishers. 2001 Principles of Linguistic Change. Volume 2: Social factors. Oxford: Blackwell.

Triggering events

53

Labov, William, Sharon Ash and Charles Boberg 2006 Atlas of North American English: Phonology and Sound Change. Berlin: Mouton de Gruyter. Labov, William and Maciej Baranowski 2006 50 msec. Language Variation and Change 18: 223–240. Labov, William, Malcah Yaeger and Richard Steiner 1972 A Quantitative Study of Sound Change in Progress. Philadelphia: U.S. Regional Survey. Lennig, Matthew 1978 Acoustic measurement of linguistic change: the modern Paris vowel system. University of Pennsylvania dissertation. Liljencrants, Johan and Lindblom, Bjorn 1972 Numerical simulation of vowel quality systems: The role of perceptual contrast. Language 48: 839–862. Lindblom, Bjorn 1988 Phonetic content in phonology. In Phonological 1988: Proceedings of the 6th International Phonology Meeting. W. Dressler et al. (eds.), 181–196. Cambridge: Cambridge University Press. Luick, Karl 1903 Studien zur Englischen Lautgeschichte. Vienna and Leipzig. Martinet, André 1955 Economie des changements phonétiques. Berne: Francke. McCone, Kim 1996 Towards a relative chronology of ancient and medieval Celtic sound change. Maynooth Studies in Celtic Linguistics 1. Maynooth: Department of Old Irish, St. Patrick’s College. McKelvey, Blake 1949a A panoramic view of Rochester’s history. Rochester History 11: 1. 1949b Rochester and the Erie Canal. Rochester History 11: 3. Meillet, Antoine 1921 Linguistique historique et linguistique générale. Paris: La société linguistique de Paris. Nearey, Terrance 1978 Phonetic Feature System for Vowels. Bloomington, Indiana: Indiana University Linguistics Club Scargill, M.H. and H.J. Warkentyne 1972 The Survey of Canadian English: a report. English Quarterly 5: 47–10 Stockwell, Robert and Donka Minkova 1997 On drifts and shifts. Studia Anglica Posnaniensia 31: 283–303. Trager, George L. 1930 The pronunciation of ‘short A’ in American Standard English. American Speech 5: 396–400.

54

William Labov 1934

What conditions limit variants of a phoneme? American Speech 9: 313–315. 1942 One phonemic entity becomes two: the case of ‘short a’. American Speech 17: 30–41. Trudgill, Peter 1974 The Social Differentiation of English in Norwich. Cambridge: Cambridge University Press. 1986 Dialects in Contact. Oxford and New York: Basil Blackwell. Vachek, Josef 1964 On peripheral phonemes of Modern English. Brno studies in English, 7–109. Weinreich, Uriel, William Labov and Marvin Herzog 1968 Empirical foundations for a theory of language change. In Directions for Historical Linguistics. W. Lehmann and Y. Malkiel (eds.), 97– 195. Austin: University of Texas Press. Wells, J. C. 1982 The Accents of English. Volume 1: An Introduction. Cambridge: Cambridge University Press. Zelinsky, Wilbur 1992 The Cultural Geography of the United States. A Revised Edition Englewood Cliffs, NJ: Prentice Hall.

What’s new in Old English?

Richard M. Hogg “Why do we have to study Old English? Surely that’s all been done.” Perhaps all of us have heard such comments, especially from rather bright undergraduates. And it is a question which deserves an answer, for there is a respectable argument which would support the underlying scepticism. For no one can deny that in the last quarter of the nineteenth century, from the time of Sweet and Cosijn to that of Bülbring and Lindelöf, the fundamentals of Old English were systematically laid down and subsequently codified by Luick. Later Campbell set out the structures of phonology and morphology for those uncomfortable with German and Mitchell published his major study of syntax. The two papers in this part of the present volume, however, both demonstrate, in different ways, that not all has been done. And they each do more than this, in that each points interested scholars in directions which deserve and will reward study, because their subjects are relatively unexplored and at the same time raise issues critical to Old English studies. The papers have similarities, in that both are dealing with corpora, yet despite having this in common, they are otherwise very different. Traugott and Pintzuk (henceforth T&P) are working with an existing corpus for which they are concerned to establish a method for analysing syntactic and pragmatic features; Fulk’s paper, on the other hand, is concerned with analysing the phonological, morphosyntactic and lexical features in a part of the available Old English corpus which is relatively neglected (at least by linguists, as opposed to traditional philologists). The creation and use of historical corpora is, of course, by no means restricted to Old English, or indeed, to the history of English as a whole. Other languages have seen the development of corpora as well, both present-day and historical. And in the context of this collection both papers are also attempting to confront issues both of what a corpus should contain and of how it should be used. For T&P the corpus itself is less relevant than an understanding of how its evidence can be used, thus requiring them to develop methods or, rather, a heuristics, which can be tested against a corpus

56

Richard M. Hogg

which was originally used for simpler tasks than they wish to analyse. Fulk, too, uses the existing corpus, or, rather, a part of that corpus. For him the problem is one of how to cope with what is normally described as ‘dialect mixture’, and this not only requires a philological analysis but also theoretical methods for distinguishing such ‘mixed’ texts. If one considers only the references to corpora in T&Ps paper, what is most immediately obvious is both the amount of work being done and the globalisation of the corpora ‘industry’, see for example the list of institutions involved in Pintzuk et al (2000). However, one thought strikes me immediately: the emergence of so many corpora, despite the evident depth of collaboration, has its own perils. In particular, I would suggest, there is some danger in this hectic advance of being lost amidst the historical mists of time. It is true, of course, that some of the relevant detail is easily recovered, but there also much that is in danger of being lost. Let me give an example. It remains simple to trace the origins of English language corpora back to the Brown Corpus first published in 1961. Hence it is equally easy to progress to the Lancaster-Oslo-Bergen (LOB) Corpus and then to further later corpora. So far so good, and obviously there is information widely available, as, for example, presented in Meyer (2002). For historical corpora the first place to look is quite obviously the Helsinki Corpus. Later evolutions, particularly the use of parsed corpora, are in some respects well-documented. Yet it seems to me that what we now need is a thoroughgoing historical account. For example, what role was played in the emergence of the historical corpora by other studies and scholars? Thus a key figure in the development of Old English corpora was surely the late Angus Cameron, for his determination to create a new Dictionary of Old English and to do so using the best mixture of traditional and contemporary methods certainly, and at the very least, enabled others to appreciate the possibility of electronic historical corpora. It strikes me that another area which deserves consideration, even at this early point, is the development of parsed and annotated corpora. This is an issue which is very relevant to T&P’s paper in this volume. To what extent, for example, are variant parsings compatible with one another and what attempts are made to ensure that parsing strategies do not vary unpredictably from one corpus to another related corpus? One area which might well be explored here are the many problematic parsings which are noted in Mitchell (1985) or in many other papers by Mitchell, see for example Mitchell (1988: ch. 26, inter alia). Indeed I was slightly surprised to find no references at all to Mitchell’s work in the paper. Doubtless there are other questions on which it would useful

What's new in Old English?

57

to have an historical account of how corpora have got to the place they now have, as essential tools for the historical study of English. To some this may seem somewhat premature, but is seems to me that in such a fast-moving situation it is all too likely that important information could be lost. The paper by T&P demonstrates rather well the point I am trying to make. The issue of how to add pragmatic coding to a syntactically parsed corpus such as the York-Toronto-Helsinki-Parsed Corpus of Old English Prose (YCOE) is both highly innovative and (partly because it is innovative) fraught with difficulty. In their paper T&P assess the possibilities of what might be coded and how. But it strikes me that, simply because this involves a substantial range of pragmatic features and because there is less agreement, or perhaps more interactive discussion, about pragmatic features in comparison with morphosyntactic features, it will be difficult to come to a conclusion on which everyone will agree. The paper by Fulk fits into an altogether different approach to corpus material, one that is much more philologically, rather than linguistically, focused. Apart from its inherent value, which is considerable, this is an excellent demonstration of how varied corpus work can be. In his paper, which like T&P’s paper, reports on preliminary work, Fulk has identified a very important, and difficult, area which has to a large extent been ignored by successive generations of linguists, namely the wide variety of material generally recognised as Late West Saxon, but which is quite clearly not part of the linguistic canon. Typical of the type of comments made about the material Fulk is looking at is the following quotation from Campbell (1959: §17): “Even when West-Saxon had become a well-established literary dialect, and was used as something of a standard written language, many manuscripts display a considerable non-West-Saxon element in their orthography and inflexions”. As Fulk notes, it is the case that we no longer have much use for the concept of a Saxon patois favoured, for example, by Bülbring, yet it has to be confessed that most linguists, as opposed to literary historians, are still reluctant to explore the textual material beyond that canon (let me be clear about this, I am at least as guilty in this matter as anyone). The texts which Fulk is studying are those later homiletic texts, from the time of Ælfric and later, which are often dismissed as the result of ‘dialect mixture, that is to say, they are treated as essentially Late West Saxon, but cast aside as “nicht streng westsächsische”, see for example Brunner (1965: §2, Anm 3). The problems with such texts are well-presented by Fulk, and it must be obvious to all that those problems are neither insignificant nor

58

Richard M. Hogg

easily resolved. They involve quite general issues such as dialect contact and literary tradition. Above all, what the issues discussed by Fulk remind me of is the concept of Missprachen which is so well explored in LALME (McIntosh, Samuels and Benskin 1986) and further in Benskin and Laing (1981) and later papers. There must surely be scope for an investigation into whether or not it might be possible to apply, in some form, the concept of Missprachen to Old English, despite the difficulties which will be faced. It is helpful, of course, that already Meg Laing has explored ways in which this can be done for early Middle English, see for example her paper on anchor texts which can help to illuminate some of the problems which also face late Old English (Laing 1991). In my own small way I have attempted to demonstrate how the application of LALME methodology can be applied to later Old English, see Hogg (2004). One real difficulty which Fulk’s paper raises is the perennial, but as yet unsolved, question of what we mean by terms such as “Late-West-Saxon” and “Anglian”. As can be seen by the above remarks, Fulk places his comments into a binary structure, West Saxon or Anglian, and then he has to confront the issue of material which is obviously neither canonical West Saxon nor canonical Anglian (whatever that might be). Surely the time has come for us all to recognise that the traditional terminology is based on a framework which was established in the late nineteenth century. That framework has been immensely useful, but, I would like to suggest, there must be questions asked about whether or not this traditional terminology may now have reached its ‘sell-by’ date. Is it not the case that for as long as we adhere to it we will not be able to assess properly the kind of point I raised in the previous paragraph? I might add that the continued use of such terms simply fails to recognise properly the structures of the society in which our texts were written. I am convinced that Fulk himself is more than aware of the problematic issues which the traditional terminology presents us with. And I recognise that in the present context there was no room to present a better alternative which would begin fully to resolve these matters. I would suggest, however, that this is certainly an area in which Fulk could make a major contribution, for, given its pioneering nature, the paper provides a model and a target for further work and discussion. In these comments above I have attempted to make suggestions about further avenues for research which these papers have highlighted. This is not the place, I believe, to comment on the papers themselves, for that can be left for another occasion and, I hope, in many cases, to other scholars. But to answer the hypothetical student, these papers show very well that it

What's new in Old English?

59

has not all been done, they both create new avenues to explore and I hope that my own comments have suggested yet other avenues. Old English is alive and well.

Editorial note We invited the authors of the papers in this section to consider responding to the issues raised by Richard Hogg. Elizabeth Traugott wrote to us: “I do agree that the syntactic parsing is an issue, and I brought it up in a paper on Left-dislocations that is forthcoming in FLH.” The concluding paragraph of Traugott and Pintzuk’s paper in the volume was revised in response to Richard Hogg’s Introduction; both authors expressed thanks to Richard Hogg for his comments. Robert Fulk wrote to us to say that he was “extremely grateful for the many kind things that Richard says about the paper” and added: “The Anglian/West Saxon binary is a genuine problem, but since I’ve already indicated in the paper that an attempt to resolve such issues will have to be reserved for another publication, I don’t think there’s any need for me to say more at this point. I’m certainly grateful for the suggestion about the possibility of Mischsprachen in Late OE, with the accompanying references, and this is a possibility I’ll be sure to pursue later.”

References Benskin, Michael and Margaret Laing 1981 Translations and Mischsprachen in Middle English manuscripts. In So meny people longages and tonges. M. Benskin and M. L. Samuels (eds.), 55–106. Edinburgh: Benskin and Samuels. Brunner, Karl 1965 Altenglische Grammatik, nach der Angelsächsischen Grammatik von Eduard Sievers. Tübingen: Niemeyer. Campbell, Alistair 1959 Old English Grammar. Oxford: Clarendon Press. Hogg, Richard M. 2004 North Northumbrian and South Northumbrian: a geographical question? In English Historical Dialectology. M. Dossena and R. Lass (eds.), 241–255. München: Peter Lang.

60

Richard M. Hogg

Laing, Margaret 1991 Anchor texts and literary manuscripts in early Middle English. In Regionalism in Late Medieval Manuscripts and Texts. F. Riddy (ed.), 27–52. Woodbridge: D.S. Brewer. McIntosh, A., M. L. Samuels and M. Benskin 1986 A Linguistic Atlas of Late Medieval English. Aberdeen: Aberdeen University Press. Meyer, Charles M. 2002 English Corpus Linguistics. An Introduction. Cambridge: Cambridge University Press. Mitchell, Bruce 1985 Old English Syntax. Oxford: Clarendon Press. 1988 On Old English. Oxford: Basil Blackwell. Pintzuk, Susan, Eric Haeberli, Ans van Kemenade, Willem Koopman, and Frank Beths (eds.) 2000 The Brooklyn-Geneva-Amsterdam-Helsinki Parsed Corpus of Old English. York: Department of Language and Linguistic Sciences, University of York, UK. Taylor, Ann, Anthony Warner, Susan Pintzuk, and Frank Beths (eds.) 2003 The York-Toronto-Helsinki Parsed Corpus of Old English Prose. Department of Language and Linguistic Science, University of York, UK. Distributed through the Oxford Text Archive.

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose to investigate the syntaxpragmatics interface*

Elizabeth Closs Traugott and Susan Pintzuk 1.

Introduction

The availability first of the Brooklyn-Geneva-Amsterdam-Helsinki Parsed Corpus of Old English (Pintzuk et al. 2000), and more recently of the YorkToronto-Helsinki Parsed Corpus of Old English Prose (Taylor et al. 2003, henceforth YCOE) has significantly enhanced work on Old English syntax. In particular, it has enabled extensive data-driven investigations of clausal syntax within formal generative frameworks (see e.g. chapters in Fischer et al. 2000 and articles in Taylor and van der Wurff 2005). There is increasing interest in the additional potential for using the YCOE and corpora of later stages of English, including the Penn-Helsinki Parsed Corpus of Middle English (Kroch and Taylor 2000) to determine the information structure as well as the syntax of Old English in comparison with that of later English, particularly with respect to the development of subject-oriented syntax (see Kemenade and Los 2006; and, in similar vein, Whitton 2004, Speyer 2005, but with the limited objective of accounting for the decline in the frequency of Topicalization in texts during the history of English). In the present paper we discuss issues that arise in using the YCOE for purposes of analyzing the syntax-pragmatics interface in Old English, with particular attention to Topicalization and Left-dislocation. The immediate question for this paper is how to address the challenge of building on the syntactically parsed corpus to make such work on information structure possible. The ultimate aim is to enable as detailed an account as is feasible of Old English (OE) information structure, including the function, status, and frequency of clause-initial elements. This would allow comparisons not only throughout the history of English but also with other languages of similar periods. It would also inform the debate on the proper analysis of aspects of OE clausal syntax, e.g. object-verb and verb-object

62

Elizabeth Closs Traugott and Susan Pintzuk

orders, or the proposed clitic status of preverbal pronouns, and would allow study of information patterns associated with genre. Our work to date on addressing this problem has targeted preposed objects in a selected set of OE texts, see Traugott and Pintzuk (In Preparation) on “Topicalization”, and Traugott (2007) on “Left-dislocation”.1 This has allowed us to refine a coding system for pragmatic analysis that is relevant for information structure studies on OE in general, while restricting the initial task to a definable subset of examples which can be compared with Present Day English (PDE) data. While we have not coded for Topic and Focus, the coding could be extended to do so, given adequately operationalized definitions of these terms. Here we discuss what needs to be coded (section 2), the coding strategies used for the OE data (section 3), and the potential value of coding the YCOE to analyze information structure (section 4).

2.

Assessing the relevant coding parameters

The first question is of course what to code. The literature on information structure is vast, and derives from different theoretical traditions. There is no complete consensus on which factors are the most significant for pragmatic analysis, and terminology is used in different ways. We have worked primarily with analyses developed in traditions associated with the following: Prince (1981a, 1981b, 1997); Vallduví (1992); Gundel, Hedberg and Zacharski (1993); Lambrecht (1994); Kiss (1995); Birner and Ward (1998). However, little of this literature has been operationalized for coding of electronic corpora. One exception is Godfrey, Holliman and McDaniel (1992). Their data are PDE telephone conversations in the Switchboard Corpus, so comparisons of results with written texts of earlier periods of the language are necessarily problematic. Nevertheless, Gregory and Michaelis’s (2001) analysis of Topicalization in this corpus suggested a solution to the problem of operationalizing the distinction between material that is discourse-old/-new (having an antecedent or not), and hearer/addresseeold/-new (situated on the “Givenness” hierarchy). All studies of older languages present problems of textual interpretation (see Herring, van Reenen, and Schøsler 2000). A fundamental problem for studying the information structure of any written text is the absence of direct information regarding prosody (but see Schlüter 2005, Speyer 2005 for some suggestions concerning how to access evidence for the role of stress

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

63

clash in grammatical change). Yet most contemporary work on Topic and Focus crucially identifies aspects of information structure with prosodic patterns that are either largely constructed (e.g. Kiss 1995) or, in the case of spoken corpora, amenable to intonational analysis with ToBI techniques developed by Beckman and Pierrehumbert (1986) (e.g., Frascarelli and Hinterhölzl 2006). A second problem is that the traditional tests for the scope of Focus, such as questions of the type What happened?, Who did it?, are rarely found in written data, especially in OE. Evidence from the exceptions, Boethius’ Consolation of Philosophy and Ælfric’s Colloquy, cannot easily be projected directly onto texts that lack such question-answer pairs, especially if a particular type has not been attested. Furthermore, many of the pairs that do occur are elliptical and do not provide the full clausal structure that we expect in test examples. However, a few are reminiscent of PDE examples: (1)

Canst þu ænig þing? Do you know any thing? Ænne cræft ic cann. One skill I know. Hwylcne? Which one? Hunta ic eom Huntsman I am. (Ælfric Colloquy 50)

A third problem is that work on information status relies on inferences about what addressees may be expected to treat as “shared” information (e.g. such factors as motivate use of the with sun, president, etc., or the introduction of the menu or the waitress in a restaurant scenario), and these inferences are less robustly recovered for a society that flourished over a millennium ago. Given such problems, we have to admit a degree of indeterminacy, and we must pay close attention to the discourse context in order to make reasonable interpretations. In other words, we are more reliant on textual evidence than may be customary for analyzing information structure in contemporary languages, although even for historical data there are crossdialectal and cross-cultural differences to be aware of.

64

Elizabeth Closs Traugott and Susan Pintzuk

With these issues in mind, we developed a relatively theory-neutral coding system designed to make it possible to retrieve the factors described below. Although recognizing that manual coding is subject to individual discrepancies, we hope the following definitions will encourage coding that is reasonably replicable.2 a) Pragmatic Topic. This is the constituent which the speaker intends the utterance to be “about” (Reinhart 1981, Lambrecht 1994, Kiss 1995). This is usually referred to as “Topic” in the information-structure literature. We call it “Pragmatic Topic” to distinguish it from the syntactic topic that is the subject of OE syntax studies. b) Topic Persistence (aka “topic continuity”, “topic promotion”, see Givón 1983). This concerns the extent to which a referent is used in subsequent discourse as Pragmatic Topic. c) Focus. This is of two types: i) “Unmarked” (Vallduví 1992, Kiss 1998, Gundel and Fretheim 2004), which denotes new information. In PDE unmarked Focus is typically clause-final O (object) or VO (the predicate); or the entire sentence may be Focus, as in There is a rotten plank in the trellis in answer to a question like What’s the problem? ii) “Marked”, including “identificational focus”, used in answer to such questions as Who did it? (in English this carries stress, cf. BILL did it), and “contrastive focus” (see Object Topicalization in f) below. d) Givenness. This concerns the degree to which a referent is represented as identifiable by the addressee/reader and is “hearer/addressee-old”. Drawing on Prince (1981a), Gundel, Hedberg and Zacharski (1993: 275) propose the following implicational cognitive status hierarchy for PDE: (ident = “identifiable”, indef = “indefinite”): (2)

in > focus

{it}

activated

¦£¦ that ¦²¦ ¦¦ ¦ ¤ this ¦» ¦¦ ¦ ¦¥¦this N ¦¦¼¦

>

familiar >

uniquely > ident

referential > type ident

{that N}

{the N}

{indef }

{a N}

this N}

e) Antecedence. This concerns whether a referent is overtly available in prior text and the degree to which this referent is lexically identifiable with

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

65

the preposed object. Antecedence identifies referents that are “discourseold” as opposed to “hearer-old” (Prince 1992, Birner and Ward 1998). Gregory and Michaelis (2001) call it “Anaphoricity” and distinguish it from Givennness as a “backward look” that makes reference to text rather than the hearer’s cognitive status. Thus, although I and you are situationally evoked and, in most interactive contexts, given and “in focus” (see (2)), they are not necessarily anaphoric in the text. Since our work to date has been on Object Topicalization and Left-dislocation (both subject- and object-), we have coded only antecedents of subjects of Left-dislocations and preposed objects, but the coding of Antecedence could be generalized to any nominal. f) Object Topicalization. This concerns constraints on preposed objects, and is a concept that has undergone considerable refinement over the last three decades. Three factors are of particular relevance: i) Poset and open proposition. There are two corner-stones of most of Prince’s and Birner and Ward’s work on Topicalization. One is a partially ordered set (“poset”): the preposed object triggers the inference that the entity represented by the NP is in a partially ordered set relation to an entity or entities already evoked in the discourse-model. The other is a “salient open proposition”: the preposed object is part of a presupposed proposition that has an unspecified variable (X), as in (3): (3)

I’ll have to introduce two principles. One I’m going to introduce now and one I’m going to introduce later. (Birner and Ward 1998: 78)

Here the open proposition is I’m going to introduce X; X (one) is a member of the poset two principles, and now/later are Focus. ii) Topic-Topicalization and Focus-Topicalization. Gundel (1974), Prince (1981b), Lambrecht (1994), and others distinguish Topic-Topicalization (Topic-TOP) in which the preposed object is anaphoric and Pragmatic Topic (“discourse-old”) (see one … one … in (3)) from Focus-Topicalization (Focus-TOP) in which the preposed object is non-anaphoric (“discoursenew”) and Focus: (4)

I made a lot of sweetbreads. A couple of pounds I think I made for her. (Birner and Ward 1998: 84)

66

Elizabeth Closs Traugott and Susan Pintzuk

Prince calls this “Focus Movement”. In (4) there is a poset relationship (amount of sweetbreads), an open proposition (I made X amount [of sweetbreads]), and the preposed object is Focus (a couple of pounds). iii) Contrastive Topicalization. A subset of Topic-TOPs and Focus-TOPs can be contrastive. Contrastive Topic-TOPs involve a poset and saliently available (anaphoric) referents (Lambrecht 1994: 291), as in (5): (5)

A. What do you think of John and Mary? B. JOHN/HIM I like, MARY/HER I detest.

while Contrastive Focus-TOP has no anaphoric reference, as in (6): (6)

A. What are you thinking about? B. JOHN I like, MARY I detest.

g) Yiddish Movement.3 This is similar to Focus-TOP, except that the open proposition has only a “plausible” and not a “salient” relationship to the poset (Prince 1981b, Birner and Ward 1998). (7b) is felicitous only under the assumption that it is shared knowledge that B’s son steals cars: (7) a. A. How’s your son? B. Don’t ask. A sportscar he wants. b. A. How’s your son? B. Don’t ask! A sportscar he stole! (Prince 1981b: 260)

3.

The coding used for the OE data

The process of coding the data begins with retrieving the data (in our case, clauses with objects in initial position) using CorpusSearch, a search engine designed for annotated corpora like the YCOE. Once the data have been retrieved, the clauses can be coded manually for a combination of structural, lexical, semantic, discourse, and pragmatic characteristics. In this section, we first describe the semantic, discourse, and pragmatic coding that we implemented, and then mention briefly those factors for which we did not code, and why.

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

67

3.1. Factors coded4 a) Animacy. This is a semantic, not pragmatic, factor, but has often been invoked in studies of pragmatic Topichood (e.g. Givón 1983), and therefore needs to be added to relevant pronouns (personal and demonstrative) and nominals. If the analysis, like ours, is limited to sentences with preposed objects, then only preposed objects need be coded for Animacy, but in a study of Pragmatic Topic and Focus, all nominal phrases would need to be coded. b) Antecedence. Relevant distinctions for our studies of preposed objects and of Left-dislocations were found to be as follows (“referent” should be understood as “referent of preposed object or of subject of Leftdislocation”): i) No referent is mentioned in prior text. ii) The referent is either a clause or previous relevant narrative, e.g., in (8) the preposed object þis refers to the prior quotation (here and in subsequent examples, the preposed object is italicized and bolded): (8)

“We ferdon þurh fyr & wæter & þu us læddest on celincnge”. “We passed through fire and water and thou us ledst into coolness”. Þis geaxode se dema, & þider efste mid his wife. This heard that judge and thither hurried with his wife. ‘“We have passed through fire and water, and you have led us into a cool place”. The judge heard this and hurried there with his wife’. (coaelive,ÆLS_[Julian_and_Basilissa]: 341.1148)

iii) The referent is a member of a set at least part of which is mentioned in prior text, e.g., in (9) the twegen dælas are two of the three parts into which the property is divided: (9)

Þa nydde se bisceop þæt hi namon þæne þryddan dæl & þa Then forced that bishop that they took that third part and those twegen dælas he dyde to þære cyrcean, & to þæs mynstræs neode. two parts he gave to that church’s and to that minster’s need.

68

Elizabeth Closs Traugott and Susan Pintzuk

‘Then the bishop compelled them to take the third part and the other two parts he gave for the needs of the church and the minster’. (coaelive,ÆLS_[Basil]: 286.640) iv) The referent is mentioned in prior text, but the form is different, e.g. “NP … pronoun”, where NP is the referent of the object pronoun, as in (10): (10) An æþelboren wif wearð micclum geswenct mid langsumre A nobly-born woman became greatly afflicted with lengthy untrymnesse. & hire ne mihte nan læcecræft fremian. sickness and her not could no doctor heal. ‘A woman of noble birth became greatly afflicted with a long-term illness, and no doctor was able to heal her’. (cocathom2,ÆCHom_II,_2: 14.79.325) v) The referent is mentioned in prior text, and the two nominal phrases (the preposed object and the referent) have the same lexical head, although case and modifiers may be different. An example is given in (11), where Crassus is introduced as the subject in the nominative case and is then referred to again as the (indirect) object in the dative case: (11) Lucinius Crassus se consul … he gefor mid firde ongean Lucinius Crassus that consul … he marched with army against Aristonocuse þæm cyninge … Crassuse wæron monege cyningas Aristonocus that king … To-Crassus were many kings of monegum landum to fultume cumene. from many lands to help come. ‘The consul Lucinius Crassus … marched with his army against King Aristonicus … Many kings from many lands came to help Crassus’. (coorosiu,Or_5: 4.118.5.2473) Prior text for anaphoricity was restricted to the preceding ten finite clauses.

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

69

c) Givenness. OE had no articles. NPs could occur without any determiner, although demonstrative pronouns were frequently used to indicate definiteness or specificity. This raises the question whether Gundel, Hedberg, and Zacharski’s (1993) hierarchy in (2) can be projected onto a language with a different structure from PDE. They appear to think so, since they give figures using their hierarchy not only for English but also for Chinese and Japanese, However, Gundel, Hedberg, and Zacharski’s hierarchy seems to be over-specific for OE and we reduced it to three categories: hearer/ addressee-old, uniquely identifiable, and referential indefinite. Gundel, Hedberg, and Zacharski’s term “in focus” is unfortunate, since “Focus” is an information status, so we call this member of the hierarchy “hearer/ addressee-old”. Because Gundel, Hedberg, and Zacharski’s “activated” vs. “familiar” would require a (currently unavailable) detailed understanding of the use of demonstratives in OE that is consistent with information status theory and that does not project PDE assumptions about scenario familiarity, the categories “activated”, “familiar” and “uniquely identifiable” were merged as “uniquely identifiable”. Likewise, since there is no indefinite article, “referential” and “type identifiable” were merged as “referential indefinite”. Referentially indefinite NPs are those that do not contain personal pronouns or demonstratives, including those that are modified by quantifiers of all types (e.g. manige ‘many’, sum ‘a certain’). Our Givenness hierarchy is in (12). (12) hearer/addressee-old > uniquely identifiable > referential indefinite {P, D} {Det N} {(Quant) N} d) Poset relationship of the preposed object to a referent in the prior context. The set-subset or part-whole relationship was coded separately from the identity relationship, since, although identity is considered a poset relationship by Birner and Ward (1998), it is sufficiently different from the setsubset and part-whole relationships to merit separate status. In (9) there is set-subset relationship, in (10) and (11) an identity relationship (identity of referent in both cases, as well as identity of form in (11)). e) Contrast relationship of the preposed object (or subject of a Leftdislocation) with a nominal constituent in the prior or following context. The nominals in question must belong to a lexical set. (13) illustrates a contrast with a prior referent that is not preposed (blindum mannum), and with a following referent that is preposed (deofulseocum and wodum):

70

Elizabeth Closs Traugott and Susan Pintzuk

(13) & he forgeaf blindum mannum gesihðe … dumbum he forgeaf and he gave blind men sight … to-dumb he gave getincnysse & deafum heorcnunge. Deofulseocum & wodum he speech and to-deaf hearing. To-possessed-by-the-devil and mad he sealde gewyt. gave sense. ‘and he gave blind men sight … the dumb he gave speech and the deaf hearing. He gave sense to those who were possessed by the devil or mad’. (cocathom1,ÆCHom_I,_1:187.254.269–274) A contrastive object or subject is identifiable with Focus-TOP if it has no antecedent, as in (13), or as Topic-TOP if it has one (see (15) below. f) Topic Persistence. If the referent of a preposed object (or subject of a Left-dislocation) is mentioned in the following ten finite clauses, it is considered to be persistent. Subsequent Mention is distinguished from Topic Persistence in that Subsequent Mention simply requires a referent to be referred to again, while Topic Persistence requires that the referent be used as overt Pragmatic Topic (in the sense of what the clause is “about”) and subject within what is arguably the same “episode”, i.e. in a coherent section of text in which place and time remain relatively constant. To distinguish between Subsequent Mention and Topic Persistence, we coded whether the referent serves as subject or object in a following clause, and whether that clause is main or subordinate. In (14) the slaves are the subject of the following subordinate, not main, clauses and are therefore only Subsequent Mentions, not examples of Topic Persistence (Subsequent Mentions are underlined). (14) & ealle þa men þe hie on þeowdome hæfdon hie gefreodon, on þæt and all those men that they in slavery had they freed on that gerad þæt he him aðas sworan þæt hie him condition that they they oaths swear-SUBJUNCT that they them æt þæm gewinnum gelæsten. at that battle support-SUBJUNCT.

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

71

‘and they (the Romans) freed all the slaves on condition that they (the slaves) would swear that they would fight with them (the Romans)’. (coorosiu,Or_4:9.102.2.2099) By contrast, in (15) the preposed object (Julius Caesar) appears twice in Subsequent Mention, and is used as Pragmatic Topic in the third subsequent clause. Because there is a switch to Pragmatic Topic within ten clauses, we coded the clause for Topic Persistence. (15) & hiene seolfne mon gesette þæt he wæs hierra þonne consul: þæt and him himself one appointed that he was higher than consul: this he heton tictator. Æfter þæm he for on Affrice. they called Dictator. After that he went into Africa. ‘and they appointed him himself higher than consul, a position they called dictator. After that he went to Africa’. (coorosiu,Or_5:12.128.10.2711) 3.2. Factors not coded a) Pragmatic Topic. We did not code for Pragmatic Topic, except in the case of Topic Persistence, because it was not relevant to the study. However, this could be done. b) Unmarked Focus. Unmarked Focus, and especially its scope, is difficult to identify consistently absent prosody and standard tests used in contemporary languages where intuitions can be tapped. However, it can be considered to be the default status when other information statuses do not apply. c) Open proposition. The example of an open proposition is very clear in example (3), because of the repetition of the pattern I QUASI-AUX introduce X. However, most examples are not so transparent (see Birner and Ward 1998), and evoke (often culturally) shared extralinguistic information. In our desire to avoid projecting assumptions from the present onto the past, we did not code for open proposition. As a consequence, there is no coding specifically for Yiddish-movement (it depends in large part on judgments that the open proposition is in a “plausible”, not “salient” relationship to the poset).

72

Elizabeth Closs Traugott and Susan Pintzuk

d) Adverbs, prepositional phrases and other constituents regarded as “syntactic topics” in the syntactic literature on OE. In the initial stages of the study we wanted to investigate strictly OVS and OSV structures, on the hypothesis that these might show rather different properties from other socalled “topics”, and that a full understanding of relatively simple constituent orders should precede the investigation of more complicated ones. However, the next stage of the research will be to code adverbs and prepositional phrases. This will require coding for semantic adverbial categories such as locative, directional, temporal, manner, and epistemic. As in PDE, many adverbs and prepositional phrases are potentially ambiguous, e.g. her is locative ‘here’, but in the Chronicles it is largely used as a temporal ‘in this year’. Likewise, nu, þa, etc. are not only temporal ‘now, then’ respectively, but also discourse particles signaling narrative textual organization ‘at this point in the discourse’.

4.

The potential value of coding the corpus

While it is clearly possible to do research on the information status of OE word order without electronic corpora, pragmatic coding of the YCOE allows rigorous analysis of correlations between word order and pragmatic and semantic factors that prior studies have not been able to achieve, since quantitative analysis can be done of any possible combination of factors that the researcher deems relevant. Furthermore, given the current availability of the parsed Middle and Early Modern English corpora (Kroch and Taylor 2000; Kroch, Santorini, and Delfs 2004; Taylor et al. 2006; respectively), a full-fledged history of changes in information status is feasible. Likewise, if comparable coding (both syntactic and pragmatic) is provided, comparative studies of information status will also be possible with other languages that have case morphology, such as Latin (see Spevak 2005), or have both case morphology and V2 syntax, such as Old French (Prévost 2005). In earlier work, several factors have often been studied together, potentially confounding possible distinctions. For example, Kohonen (1978) investigates Topicalization and Left-dislocation in OE and Early Modern English (Homilies I, Vices and Virtues, and Sawles Warde), with attention to preposed objects, prepositional objects, negative ne, and adverbs. He combines not only a variety of syntactic factors but also a number of disparate pragmatic ones. Drawing primarily from the then-dominant theories of

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

73

information status (Prague School Functional Sentence Perspective, and Hallidayan Functional Grammar), he takes “Givenness” as “a status that is decided by the speaker” (p. 138). His methodology includes the investigation of a large number of factors including anaphora, repetition, synonym, antonymy, paraphrase, pragmatically known and new information (pp. 138–139). Using these, Kohonen found that in Homilies I all preposed direct and indirect objects are “given” (although later he acknowledged “a small margin of uncertainty” in Kohonen (1982: 188, cited in Masayuki 2002: 251, ft. 5). Specifically, Kohonen (1978) found 50 objects and 14 indirect objects that are “new”. He said that 0 out of the 64 (0%) objects and indirect objects that are new occur in “initial field”, i.e. are preposed (he calls this “fronted”). By contrast, we found that in Homilies I, 21 out of 198 (10.6%) preposed direct and indirect objects have 0 anaphora, i.e. are discourse-new. With respect to Givenness, Kohonen found 156 objects and 99 indirect objects that are “given”. 11% of the 156 given objects and 18% of the 99 given indirect objects are preposed (pp. 145–146). However, we found that in Homilies I 89 out of 198 (44.9%) preposed objects are indisputably “given”, i.e. personal pronouns, demonstrative pronouns, or uniquely identifiable. The remaining 109 out of 198 (55.1%) are indefinite or quantified referential indefinites (referential, but hearer/addressee new). Masayuki (2002) investigates only OVS clauses in Homilies I, and identifies several important characteristics of preposed objects in this text. One characteristic he mentions is that if there is Topic Persistence (Masayuki’s “forward looking center”), it is frequently not in an immediately adjacent clause (p. 245–246). Another characteristic is that preposed objects can be used to introduce a digression or summarize an episode (p. 247). Furthermore, he shows that, despite Kohonen’s finding that all preposed objects in Homilies I are “given”, nevertheless there are some that are presumably Focus (therefore “new”); however, he gives no frequencies on this. The limited data and lack of comparison of OVS with OSV structures render this particular study suggestive only. Masayuki also hypothesizes that some examples “might be comparable” to Yiddish-Movement (p. 249–250) because of the weakness of the open proposition relationship. However, the connection to Yiddish-Movement seems far-fetched because the pragmatic incongruity of Yiddish-Movement absent shared knowledge does not appear to be in evidence. Moreover, it does not help to account for one major difference between OE and PDE Topicalization: the frequent use of noncontrastive personal pronouns in OE: 24.1% of preposed objects in our data base are personal pronouns, 43.9% in OVS clauses, 13.1% in OSV clauses.

74

Elizabeth Closs Traugott and Susan Pintzuk

One possible line of research that emerged from our preliminary studies is the potential for detailed analysis of the information structure patterns associated with particular genres. For example, Table 1 shows that the frequency of OVS vs. OSV is different between the Orosius and West Saxon Gospels on the one hand, and Ælfric’s writings, especially the Lives, on the other. Table 1. Clauses with initial objects in five Old English texts, by constituent order Text

OVS

OSV

Total

%OVS

Lives

62

77

139

44.6%

Homilies I

71

124

195

36.4%

Homilies II

64

109

173

37.0%

Orosius

18

43

61

29.5%

WS Gospels

24

76

100

24.0%

Total

239

429

668

35.8%

The difference between the patterns in the Orosius and the writings of Ælfric may not be too surprising given that the former is a historical and secular work, while the latter are in the homiletic tradition. The Gospels belong to yet another genre, though one might expect them to be closer than they are to Ælfric’s writings, which refer extensively to the Gospels. However, as Table 2 below shows, the overall percentages obscure more fine-grained differences that emerge when the type of preposed object is considered (personal pronoun, demonstrative pronoun, or full nominal). Particularly striking in Table 2 is the difference between Ælfric’s Lives and Homilies I and II. This suggests that genre and register differences may be considerable even in the work of one author. In OVS clauses in Lives, 60% of preposed objects are Ps, and only 21% are Ns, figures that distinguish it from the other Ælfrician texts: compare 28% Ps and 58% Ns in OVS clauses in Homilies I and 31% Ps and 63% Ns in Homilies II. In this respect, Lives appears to be rather closer to the Orosius (40% Ps and 33% Ns in OVS clauses) than to other texts by Ælfric. Similar but not quite so dramatic differences can be seen in OSV clauses. Note that in Table 2, percentages are rounded and so may not add up to 100%:

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

75

Table 2. Type of preposed object Text

Order

Lives

OVS

37

60%

12

19%

13

21%

62

OSV

18

23%

9

12%

50

65%

77

OVS

20

28%

10

14%

41

58%

71

OSV

4

3%

11

9%

109 88%

124

OVS

20

31%

4

6%

40

63%

64

OSV

8

7%

9

8%

92

84%

109

OVS

7

40%

5

28%

6

33%

18

OSV

14

33%

3

7%

26

61%

43

OVS

21

88%

2

8%

1

4%

24

OSV

12

16%

9

11%

55

72%

76

Homilies I Homilies II Orosius WS Gospels

P

D

N

Total

One question raised by the distributions in Tables 1 and 2 is the correspondence between word order, Pragmatic Topic, and syntactic topic. If it were the case that the first position in the clause is always occupied by the Pragmatic Topic, regardless of the order and position of the remaining clausal constituents, then we might expect that the distribution of object types (P, D, and N) would be the same in OVS and OSV clause. However, Table 2 shows that this expectation is not fulfilled. Furthermore, as shown in section 3.1.d., some preposed objects are contrastive and Focus-TOPs. We plan in future research to investigate the syntactic and pragmatic characteristics of clause-initial constituents to determine whether we can indeed find a correspondence between pragmatic function and syntactic position, and thus use the pragmatics to understand the syntax. However extensive the new possibilities, it must always be recalled that our access to the information status of ancient texts will be limited by our as yet incomplete understanding of information structure in contemporary languages. Even more crucially, it will be limited by the absence of intonational information in written texts and the impossibility of fully understanding the encyclopedic knowledge of authors and audiences of the past.

76

Elizabeth Closs Traugott and Susan Pintzuk

Notes *

1.

2.

3.

4.

Many thanks to Joanne Close for her assistance with coding and running the statistics. Thanks also to our audiences at various presentations of related work (Traugott and Pintzuk 2005 a, b), to two anonymous reviewers, and to Richard Hogg for valuable comments and suggestions. The texts are: Ælfric’s Lives of Saints (Lives), Ælfric’s Catholic Homilies I and II (Homilies I, Homilies II), Orosius, and the West-Saxon Gospels (WS Gospels). Automatic coding for most factors will not be possible: the YCOE is morphologically and syntactically annotated, but it is not annotated for factors such as Pragmatic Topic, Animacy, or indexicality. In addition, searches and automated coding are possible only within sentences using the search engines currently available, so that automated coding for factors such as Topic Persistence is not possible. While one anonymous reviewer was understandably concerned about the problems of manual coding done by a variety of different researchers, possibly with different theoretical assumptions, it is important to recognize that these same types of problems arise when researchers manually collect and interpret sets of examples from either historic texts or modern corpora. Yiddish-Movement is included because it has sometimes been suggested that OE shows features of Yiddish-Movement, specifically the weakness of the presuppositional link between the open proposition and the preposed object (see Masayuki 2002). The following grammatical abbreviations are used: D = demonstrative pronoun, Det = demonstrative determiner, N = full nominal, P = personal pronoun, Quant = quantifier, SUBJUNCT = subjunctive.

References Beckman, Mary E., and Janet Pierrehumbert 1986 Intonational structure in Japanese and English. Phonology Yearbook 3: 255–309. Birner, Betty J., and Gregory Ward Information Status and Non-Canonical Word Order in English. 1998 (Studies in Language Companion Series 40.) Amsterdam/Philadelphia: Benjamins. Fischer, Olga, Ans van Kemenade, Willem Koopman, and Wim van der Wurff (eds.) The Syntax of Early English. (Cambridge Syntax Guides.) Cam2000 bridge: Cambridge University Press.

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

77

Frascarelli, Mara, and Roland Hinterhölzl 2006 Types of topics in German and Italian. To appear in On Information Structure, Meaning and Form. Susanne Winkler and Kerstin Schwabe (eds.). Amsterdam/Philadelphia: Benjamins. Givón, T. (ed.) Topic Continuity in Discourse: A Quantitative Cross-Language 1983 Study. (Typological Studies in Language 3.). Amsterdam/Philadelphia: Benjamins. Godfrey, J., E. Holliman, and J. McDaniel 1992 SWITCHBOARD: Telephone speech corpus for research and development. International Conference on Acoustics, Speech and Signal Processing 92: 517–520. And see http://www.cis.upenn.edu/~treebank/home.html Gregory, Michelle L., and Laura A. Michaelis 2001 Topicalization and Left-dislocation: A functional opposition revisited. Journal of Pragmatics 33: 1665–1706. Gundel, Jeannette 1974 The Role of Topic and Comment in Linguistic Theory. PhD diss., University of Texas. Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski 1993 Cognitive status and the form of referring expressions in discourse. Language 69: 272–307. Gundel, Jeanette K., and Thorstein Fretheim 2004 Topic and Focus. In The Handbook of Pragmatics. Laurence R. Horn and Gregory Ward (eds.), 175–196. (Blackwell Handbooks in Linguistics.) Oxford/Malden, MA: Blackwell. Herring, Susan C., Pieter van Reenen, and Lene Schøsler (eds.) Textual Parameters in Older Languages. (Amsterdam Studies in the 2000 Theory and History of Linguistic Science 195.) Amsterdam/Philadelphia: Benjamins. Kemenade, Ans van Syntactic Case and Morphological Case in the History of English. 1987 Dordrecht: Foris. Kemenade, Ans van, and Bettelou Los 2006 Discourse adverbs and clausal syntax in Old and Middle English. In Handbook of the History of English. Ans van Kemenade and Bettelou Los (eds.), 224–248. (Blackwell Handbooks in Linguistics.) Oxford/Malden, MA: Blackwell. Kiss, Katalin É. 1995 Discourse configurational languages: Introduction. In Discourse Configurational Languages. Katalin É. Kiss (ed.), 3–27. (Oxford

78

Elizabeth Closs Traugott and Susan Pintzuk

Studies in Comparative Syntax.) New York/Oxford: Oxford University Press. 1998 Identificational focus versus information focus. Language 74: 245– 273. Kohonen, Viljo 1978 On the Development of English Word Order in Religious Prose around 1000 and 1200 A.D: A Quantitative Study of Word Order in Context. PhD. diss., University of Åbo. 1982 A note on factors affecting the position of accusative objects and complements in Ælfric’s Catholic Homilies I. In Approaches to Word Order. Nils Erik Enkvist and Viljo Kohonen (eds.), Åbo: Åbo Akademi. Koopman, Willem F. 1997 Another look at clitics in Old English. Transactions of the Philological Society 95: 73–93. Kroch, Anthony, and Ann Taylor (eds.) The Penn-Helsinki Parsed Corpus of Middle English. Philadelphia: 2000 Department of Linguistics, University of Pennsylvania, 2nd edition. Kroch, Anthony, Beatrice Santorini, and Lauren Delfs Penn-Helsinki Parsed Corpus of Early Modern English. Philadel2004 phia: Department of Linguistics, University of Pennsylvania. Lambrecht, Knud Information Structure and Sentence Form: Topic, Focus and the 1994 Mental Representations of Discourse Referents. (Cambridge Studies in Linguistics 71.) Cambridge: Cambridge University Press. Masayuki, Ohkado 2002 On object fronting in Old English. In Studies in English Historical Linguistics and Philology: A Festschrift for Akio Oizumi. Jacek Fisiak (ed.), 239–253. (Studies in English Medieval Language and Literature 2.) Frankfurt am Main: Lang. Pintzuk, Susan Phrase Structures in Competition: Variation and Change in Old 1999 English Clause Structure. (Outstanding Dissertations in Linguistics.) New York: Garland. Pintzuk, Susan, Eric Haeberli, Ans van Kemenade, Willem Koopman, and Frank Beths (eds.) The Brooklyn-Geneva-Amsterdam-Helsinki Parsed Corpus of Old 2000 English. York: Department of Language and Linguistic Sciences, University of York, UK. Prévost, Sophie 2005 An informational approach to the sentence-initial and left-detached positions in relation with the position of subject. Paper presented at

Coding the York-Toronto-Helsinki Parsed Corpus of Old English Prose

79

the Workshop on Sentence-initial and Sentence-final Position, Societas Linguistica Europea (SLE) 38, Valencia, Sept. 7th–10th 2005. Prince, Ellen F. 1981a Toward a taxonomy of given-new information. In Radical Pragmatics. Peter Cole (ed.), 223–255. New York: Academic Press. 1981b Topicalization, Focus-Movement, and Yiddish-Movement: A pragmatic differentiation. In Proceedings of the Seventh Annual Meeting of the Berkeley Linguistics Society. Danny K. Alford et al. (eds.), 249–264. Berkeley: University of California, Berkeley Linguistics Society. 1992 The ZPG letter: Subjects, definiteness, and information-status. In Discourse Description: Diverse Analyses of a Fundraising Text. William C. Mann and Sandra A. Thompson (eds.), 295–325. (Pragmatics and Beyond, New Series 16.) Amsterdam/Philadelphia: Benjamins. 1997 On the functions of Left-dislocation in English discourse. In Directions in Functional Linguistics. Akio Kamio (ed.), 117–143. (Studies in Language Companion Series 36.) Amsterdam/Philadelphia: Benjamins. Reinhart, Tanya 1981 Pragmatics and linguistics: An analysis of sentence topic. Philosophica 27: 53–94. Schlüter, Julia Rhythmic Grammar: The Influence of Rhythm on Grammatical Varia2005 tion and Change in English. (Topics in English Linguistics 46.) Berlin/New York: Mouton de Gruyter. Spevak, Olga 2005 The focus position in Latin: A confrontation between Latin and Czech. Paper presented at the Workshop on Sentence-initial and Sentence-final Position, Societas Linguistica Europea (SLE) 38, Valencia, Sept. 7th–10th 2005. Speyer, Augustin 2005 Topicalization and the trochaic requirement. University of Pennsylvania Working Papers in Linguistics 10.2: 243–256. Taylor, Ann, Arja Nurmi, Anthony Warner, Susan Pintzuk, and Terttu Nevalainen (annotators) Parsed Corpus of Early English Correspondence, parsed version. 2006 Compiled by the CEEC Project Team. York: University of York and Helsinki: University of Helsinki. Distributed through the Oxford Text Archive.

80

Elizabeth Closs Traugott and Susan Pintzuk

Taylor, Ann, and Wim van der Wurff (eds.) 2005 Aspects of OV and VO order in the history of English. English Language and Linguistics 9. 1 (special issue). Taylor, Ann, Anthony Warner, Susan Pintzuk, and Frank Beths (eds.) The York-Toronto-Helsinki Parsed Corpus of Old English Prose. 2003 Department of Language and Linguistic Science, University of York, UK. Distributed through the Oxford Text Archive. Traugott, Elizabeth Closs 2007 Old English Left-dislocations: Their structure and information status. Folia Linguistica: Acta Societatis Linguisticae Europaeae 14: 405– 441. Traugott, Elizabeth Closs, and Susan Pintzuk 2005a The discourse function of object topicalization in Old English. Paper presented at Societas Linguistica Europea (SLE) 38, Valencia, Sept. 7th–10th. 2005b The information-status of Old English Left-dislocations. Paper presented at the Society for the History of English Linguistics (SHEL) 4, Flagstaff, Sept. 30th–Oct 1st 2005. In Prep. The discourse function of object topicalization in Old English. Vallduví, Enric The Informational Component. (Outstanding Dissertations in Lin1992 guistics.) New York: Garland. Whitton, Laura 2004 The relationship between the pragmatics of preposed objects and the decline of topicalization in the history of English. University of Pennsylvania Working Papers in Linguistics 10: 187–200.

Anglian dialect features in Old English anonymous homiletic literature: A survey, with preliminary findings

R.D. Fulk 1.

Introduction

On the standard view of Old English dialectologists, Late West Saxon is best represented by the language of the authentic works of Æthelwold, bishop of Winchester (963–84), and of his student, the homilist Ælfric, writing from 990 to c. 1010.1 Yet the greater part of the surviving corpus of Late West Saxon prose shows an admixture of dialect features foreign to the language of Æthelwold and Ælfric. This corpus of seemingly dialectally mixed prose comprises a variety of textual types, including wisdom and travel literature, penitentials, medical books, chronicles, works of theology, letters, and many shorter medical recipes, prognostics, liturgical and penitential texts, inscriptions, glosses, records, and charters, as well as all the Psalter glosses.2 The chief representatives of such texts, however, are anonymous homilies, by far the commonest textual type of any real substance in this category. These homilies thus must play a central role in any discussion of what the cause of seeming dialect mixture in Late West Saxon texts is – whether it is, for example, due to imperfect “translation” of texts from one dialect into another, to regional or social variation within West Saxon, to competition among different artificial, literary standards or focused varieties, to register, to scribal idiosyncrasies, or to a combination of such factors. It was at one time widely believed that Anglian (or specifically West Mercian) features in Late West Saxon prose should be attributed to local variation within Wessex (see Bülbring 1902: §§ 23, 27, Sprockel 1965: xxvi, n. 2). This view passed out of favor in the course of the last century, in part because it came to be understood that West Saxon was the literary standard written in all parts of England, not just in Wessex, in the tenth and eleventh centuries. Vleeskruyer (1953) argues that there must have been a

82

R.D. Fulk

Mercian prose tradition in the vernacular before the reign of Alfred the Great (871–899) in order to account for the Mercian features in The Life of St. Chad (LS 3), and he compiles a list (51–62) of texts likely to have been transmitted through this tradition. Most of the texts on this list are now broadly acknowledged to be of Anglian origin, and so when Wilson (1959: 293) remarks, “Vleeskruyer spoils his case by claiming for it too many texts on far too slender evidence,” very likely the inclusion of some anonymous homilies on the list (55, 58–61) is a good part of the reason for this assessment. A particular source of doubt is the observation that evidence like Vleeskruyer’s “will not inspire confidence at least in the case of the extensive collections of homilies for which Vleeskruyer claims a Mercian origin, for their language is often poetical and archaic, and hence apparently Mercian” (Campbell 1955: 56).3 Hence some scholars tend to see the Anglian elements in such homilies, or even in Late West Saxon prose in general, as artificial and literary – and indeed, even Ælfric uses the occasional poeticism (see, e.g., Godden 1980: 217–219; Frank 1994). And yet although there seems even now to be general agreement that Vleeskruyer’s claims are not fully supportable, there has developed over the years a degree of consensus – chiefly among linguists rather than literary scholars or textual editors – that there must have been a Mercian tradition of writing in the vernacular before the rise of Wessex in the course of the ninth century, or at least that some anonymous Late West Saxon prose must derive from Mercian originals.4 Yet doubts remain. Benskin (1994: 172–3), assuming variation in the mixture of dialect features to be regional in nature, regards the uneasy consensus about Anglian origins as methodologically insupportable. Smith (1996: 65–7) considers it as unrealistic to regard Late West Saxon as a monolithic standard; it should instead be thought of as an ideal toward which scribes in varying degrees strove.5 Weighing the merit of the alternative explanations will demand careful and detailed attention both to individual dialect features and to the specific texts that contain them. It is fruitless to speculate about causes without reference to linguistic and textual specifics, because they often do or do not lend themselves transparently to particular explanations. For example, variation between forms of libban and lifian ‘to live’ cannot be solely regional in nature, since in Ælfric the former is the unmarked form and the latter is used only in varieties of a formula meaning ‘the living God’. The distribution of lifian thus reinforces the general assumption that some Anglian forms may be associated with an elevated register, and the historical circumstances that could have produced

Anglian dialect features in Old English anonymous homiletic literature

83

this result will probably be of no small concern in evaluating the claim that the rise of a tradition of vernacular writing in ninth-century Wessex had no antecedent. Likewise, a text evincing few Anglian features accords better with the explanation of regional variation within Wessex than does a text with a wide variety of Anglian features, especially when some light may be shed on the regional distribution of relevant features on the basis of Middle English evidence. To the end of weighing the plausibility of competing explanations it is therefore imperative not to offer generalizations in a philological vacuum but to ground conclusions in the probabilities suggested by close examination of particular features and texts. The process of examining such features and texts in detail, naturally, is well beyond the scope of what can be attempted in this small space – Vleeskruyer’s study of the St. Chad homily is an example of the degree of detail that will prove most beneficial, as is the excellent study by Roberts (1986) – or indeed by a single scholar, and no such attempt will be made here. But prior to and as a complement to that process of examination, there may be some worth in presenting a broad overview of seemingly non-West-Saxon features found in anonymous homiletic prose. Such an overview may shed light on directions that it would be most useful for close examination of features and texts to take, both by dialectologists and by textual editors.

2.

The survey

A survey of the corpus of anonymous homilies will offer several benefits: (1) It will help to determine how widespread and various the phenomenon of seeming dialect mixture is. Any particular homiletic text can then be compared to the remainder of the corpus, revealing the relative extent of its deviation from Ælfrician standards. (2) It will show how much variation there is from text to text in regard to the density of Anglian features. This should facilitate comparison of anonymous texts with non-homiletic texts generally acknowledged to be of Anglian origin, to judge whether the density of such features is comparable. This may help to resolve the question whether such features could be aspects of literary style rather than evidence of dialect origins (whether “dialect” is conceived in regional, social, or structural terms). (3) It will serve to identify which of these Anglian features are and are not found with any frequency in the corpus. This is a consideration that is also of relevance to determining whether such features are attributable to literary convention, and whether the incidence of a feature is

84

R.D. Fulk

limited enough to be ascribed plausibly to regional variation. (4) It will reveal which features tend to co-occur, again with implications for their status as evidence. For example, it is plain in the data that syncope in the present tense, second and third persons singular, of strong verbs and longstemmed verbs of the first weak class, as well as in passive participles of weak verbs of the first class with stems that end in a dental stop (feature 12), is found in a wide range of texts, including many that display few or no other Anglian features. Since it seems fairly implausible that so many supposedly Anglian texts should have been thoroughly Saxonized except in regard to this one feature, it is probably best regarded in homilies as stylistic rather than indicative of origins in a particular region or scribal tradition.6 (5) In a few instances, probable dates for the composition of certain anonymous homilies can be divined, and a survey may suggest chronological factors in the distribution of Anglian features. The Anglian features surveyed are listed in Appendix A.7 These are by no means all the Anglian characteristics observable in the corpus, and even some familiar features have been omitted. For example, Anglian smoothing (which is all too often indistinguishable from Late West Saxon smoothing) is too rare a feature to be regularly surveyed, though a few probable instances are noted in the presentation of the data. And West Mercian second fronting of æ to e is not regularly surveyed because of interference from the Kentish development of æ to e – though once again, frequent e for æ is sometimes noted in the data when there is no other strong evidence of Kentish features in a text, such as confusion of y and e. In general, aside from such special cases, the Anglian features selected for inclusion are all those that are well attested in the corpus and which are generally regarded in the literature as most plainly diagnostic. But there is variability in the degree to which the features may be regarded as reliable, and indeed, it is to be hoped that the inclusion of some seemingly less reliable features (like the unsyncopated verb forms mentioned above) will contribute fruitfully to determining their relative worth. The texts examined are listed in Appendix B, along with lists of the surveyed features found in them. The corpus comprises anonymous homilies for specified (HomS) and unspecified (HomU) occasions, as well as miscellaneous homilies (HomM). Also included are all anonymous saints’ legends (LS), since many of these served a liturgical purpose, and they are often cast as homilies.8 Generally excluded are the Blickling and Vercelli homilies, and recensions of them, though a few representative homilies from this body of material are included for general comparative purposes.9 The texts

Anglian dialect features in Old English anonymous homiletic literature

85

examined are those in the Dictionary of Old English Corpus in Electronic Form (2004 release), and thus, variants from different manuscripts of the same text are not taken into account. Detailed studies of individual texts could fruitfully add to our understanding by examining the manuscript variants. The corpus of texts surveyed amounts to approximately a quarter of a million words.

3.

Findings

The findings to be derived from the survey are preliminary in more than one respect. The methodological issues raised above will need to be addressed in detail before any firm conclusions can be defended. Moreover, no conclusions about individual texts should be drawn without careful study of those texts themselves, as certain dangers attend analyzing the given data in isolation. For example, although LS 29 displays a variety of the listed features, few of them cannot be analyzed as Kentish rather than Mercian. Likewise, the relative lateness of a manuscript text may produce seeming anomalies. For instance, LS 22 is almost certainly Kentish in origin, and yet it shows retraction of æ to a before checked l (feature 7), a feature that should be regarded as distinctively Anglian if the text did not contain so many late spellings. An important additional consideration is that many anonymous homilies are composites, incorporating parts of prior works, and this practice may lead to linguistic heterogeneity within a homily.10 Also, due to the fragmentary nature of the extant witnesses to Old Kentish, some features are difficult to identify as Anglian to the exclusion of Kentish, especially lexical ones. Whether two homilies were copied by the same scribe is also a consideration worthy of note. The incidence of a feature relative to a text’s length, too, is a significant datum, though it is the sort of information that can be evaluated more productively in a detailed study of an individual text than in a preliminary and exploratory survey. At all events, a count of forms may be less objective a datum than it at first seems, since a discussion of certain ambiguities may be necessary to present a realistic picture, and so incidence is not for the most part recorded here, except in connection with feature 22, to which it is directly relevant. It is not possible to mount here an extended discussion of such issues, or even of the five issues raised above on which these data may shed some light. But a few very general observations on the data may be offered:

86

R.D. Fulk

(1) Anglian features appear to be much commoner in the corpus than has generally been recognized. Nearly all anonymous homilies contain some, and yet many textual editors in particular tend to underassess their numbers and possible significance. For example, in his edition of the legend of St. Machutus, Yerkes (1984: xxxvi) maintains that the text displays no phonological or morphological features that cannot be regarded as Late West Saxon, with the exception only of instances of ¬ for Early West Saxon ȯe (features 2, 3, and 8, though he does not distinguish the three). Now it can be seen that the text evinces a considerable variety of non-Ælfrician features, and with further study it may well prove necessary to conclude that the translation was not originally made at Winchester, after all, but that its distinctive Winchester vocabulary is due to redaction in the course of transmission. Comparison of the anonymous homilies with other texts that are generally acknowledged to be of Anglian origin will most likely reveal that many show an equal pervasiveness of such features. (2) There is great variability in the number of features discoverable. Compare HomS 41, showing none of the listed Anglian characteristics, to HomS 22, in which many different sorts are to be found, though both homilies exceed 200 lines in length. Closer examination of such texts may suggest which of the possible reasons (such as different degrees of authorial competence at the use of elevated language, different degrees of Saxonization of the text, and different local origins) best explains such disparities. (3) The variety of types of features found in a great many of these texts is also remarkable. For example, HomS 33 evinces Anglian phonological/graphemic, morphological, syntactic, and lexical features. Variety of this kind is probably no small consideration in determining whether such dialect characteristics could be due to artificial literary conventions. These data are likely to repay closer examination, particularly when individual texts, and more especially related ones, are compared. A more detailed examination of the implications of these data, both in general and in regard to particular texts, is in preparation.

Appendix A: Anglian features surveyed 1. o for a before nasal consonants, e.g. noma ‘name’, lond ‘land’, song ‘sang’ = Late West Saxon (LWS) nama, land, sang. 2. ¬ for Early West Saxon (EWS) ȯe, LWS ‰, ȯ as the front mutation of ¬a, e.g. hƝran ‘hear, obey’ and ƗlƝsde ‘released’.

Anglian dialect features in Old English anonymous homiletic literature

87

3. Failure of diphthongization by initial palatal consonants, e.g. forgetan ‘forget’, gƝt ‘yet, still’, cald ‘cold’ = EWS forgietan, gƯet, ceald, LWS forgitan, g©t, ceald. (Must be distinguished from the effects of LWS smoothing.) 4. Non-WS examples of back mutation, e.g. seoððan, sioþþan ‘after’ and andswearade ‘answered’. 5. Other unusual instances of back mutation, e.g. teala ‘well’, neomaþ ‘take’ = usual LWS tela, nimaþ. 6. Ɲ for WS ® as the reflex of Gmc. Ɲ1, e.g. dƝd ‘deed’, lƝgon ‘lay’ = WS d®d, l®gon. 7. Retraction (rather than breaking) of prehistoric æ before covered l, e.g. haldan ‘hold’, all ‘all’ = WS healdan, eall. 8. Front mutation of the vowel produced under no. 7 as æ or e, e.g. onheld ‘inclined’, welde ‘boiled’. 9. Breaking in Anglian seolf ‘self’ = WS self, sylf. 10. Anglian ætƝawan ‘show’ = WS ætƯewan, æt©wan. Note also that æteow- is rare in WS in the present tense. 11. Anglian and EWS unmutated dƝagol, dƝogol ‘hidden’; cf. LWS dƯgel. (Cf. also non-LWS dƝgol, dƝgel.) 12. Absence of syncope (and usually of front mutation) in the present tense, second and third persons singular, of strong verbs and longstemmed verbs of the first weak class, as well as in passive participles of weak verbs of the first class with stems that end in a dental stop, e.g. cumeð ‘comes’, brnjceð ‘uses’, fyllest ‘fill’, seted ‘set’, l®ded ‘led’ = WS cymð, br©cð, fylst, set(t), l®d(d). 13. Anglian hafaþ, nafaþ, hafast, nafast ‘(not) have’ (WS hæfþ, næfþ, hæfst, næfst). 14. Anglian sagaþ (or segeð?) ‘says’ (WS sægð, segþ). 15. Distinctive accusative personal pronouns þec, mec, njsic. 16. Use of the accusative case with mid ‘with’. 17. Forms of lifigan ‘live’ (WS libban), except in formulas (þƗm lifigendan Gode, ðone lifiende GƗst, etc.). 18. Anglian forms of ‘to be’, e.g. earon ‘are’ (WS, Kentish sind(on)), eam ‘am’ (also Kentish; WS eom). 19. Preterite forms sƝgon, gesegen to sƝon ‘see’ (WS sƗwon, gesewen). 20. Preterite cwǀm(-) to cuman ‘come’. 21. Preterite heht to hƗtan ‘promise, call’ (WS hƝt). 22. Frequent use of the suffix -ad(-) in preterite forms of verbs of the second weak class without ƒ in the root (also Kentish).

88

23. 24. 25. 26. 27. 28. 29.

30.

R.D. Fulk

Masculine s® ‘sea’ (WS usually feminine). Possessive pronoun njs(s)- ‘our’ rather than njr-. Inceptive prefix in- = WS on-. Co-occurrence of a demonstrative and a possessive pronoun, e.g. his þƗm hƗlgum ‘his the saints’. Consistent use of syncopated cwyð, cwyst ‘says, say’ in a text marked by unsyncopated verb forms. ac = Lat. nonne? Anglian words not normally found in West Saxon: (a) Ɨcweðan ‘say’; (b) bewerian in the sense ‘forbid, prohibit’ rather than ‘defend’; (c) byrgan ‘taste’; (d) fand ‘found’ (WS funde); (e) fƝogan ‘hate’; (f) unstressed fore ‘for’ (WS for); (g) gƝn ‘yet, still’ (WS ƥƯet, g©t); (e) unstressed in (WS on); (i) Ưren ‘iron’ (WS Ưse(r)n); (j) n®nig ‘no, none’; (k) nymðe and (Mercian) nefne ‘unless’ (WS bnjtan); (l) oferhygd ‘pride’ (WS ofermǀd); (m) sceþþan ‘harm’; (n) scua ‘shadow’; (o) semninga, samnunga ‘suddenly’ (WS f®ringa); (p) smƯreness ‘anointment’; (q) symbelness ‘festival, festivity’; (r) werig ‘accursed’; (s) ymbsellan ‘surround’. Anglian words not normally found in Late West Saxon, though they occur in Early West Saxon: (a) bebicgan ‘purchase’; (b) carcern ‘prison’; (c) (ge)fƝon ‘rejoice’ (LWS fægnian); (d) frignan ‘ask’; (e) infinitive gangan ‘go’; (f) snyttru ‘wisdom’ (LWS wƯsdǀm).

Appendix B: Texts surveyed After each short title (see n. 1), in square brackets are given the numbers, from Appendix A, corresponding to the Anglian features discoverable in the text. Some notable features other than those listed in Appendix A are also occasionally remarked, and if the homily is known to be composite, this information is supplied. The notation “{Ker 331.44; 136}” (and similar, at the end of an entry) indicates that the text is manuscript item 331.44 in Ker 1957, and the edited text comes to (approximately) 136 lines. This information is supplied to facilitate the identification of texts contained in the same manuscript and to provide some indication whether the incidence of Anglian features may be to any great extent influenced by the length of the text. The notation “(HomS 4 = Vercelli 9)” indicates that the homily in question is a Vercelli homily (or a closely related recension of one) and is thus omitted from the survey, as explained above.

Anglian dialect features in Old English anonymous homiletic literature

89

HomS 1 (ScraggVerc 5) [1, 7, 12, 17, 22(1×), 24, 29(e, j); hƯe, sƯe (cf. tƯedernesse with inorganic Ưe)] {Ker 394.5; 204} HomS 2 (ScraggVerc 16) [1(conone = canone), 2, 5, 12, 13, 16, 17, 20, 22, 23, 29(b, e, l); hƯe, sƯe, but also geƯecan (cf. unetymological sieððan, and see Scragg 1992: xlviii)] {Ker 394.18; 205} HomS 3 (ScraggVerc 8) [1, 5, 12, 17, 26, 27, 29(e, j); sƯen, hƯe; eorne, eallan = georne, geallan] {Ker 394.10; 102} (HomS 4 = Vercelli 9) HomS 5 (Willard) [2, 12, 22, 29(h)] {Ker 153.4; 92} HomS 6 (Ass 14) [12, 13, 22(1×), 29(h, j); composite] {Ker 56.11; 141} HomS 7 [12, 13, 19, 22(1×), 26, 29(h)] {Ker 153.6; 206} HomS 8 (BlHom 2) [1, 2, 3, 4, 6, 12, 22(2×), 29(e, h, j); hƯe; bendas (masc.)?; ah = ac] {Ker 382.2; 212} HomS 9 [none of the listed Anglian features] {Ker 45B.20; 69} HomS 10 (BlHom 3) [1, 2, 4, 5, 7, 8, 12, 13, 14, 17, 22(2×), 26, 29(j, l); sƯe, hƯe; ah = ac; mehte; andrysnlicu, andrysnum] {Ker 382.3; 217} (HomS 11.1, 11.2 = Vercelli 3) HomS 12 [2, 12, 17, 26, 29(h, l)] {Ker 331.44; 136} HomS 13 (Ass 11) [2(1×); otherwise none of the listed Anglian features] {Ker 68.14; 144} HomS 14 (BlHom 4) [1, 2, 3, 4, 5(feala), 7, 12, 13, 17, 29(e, j); hƯe; senna = synna; ah = ac] HomS 15 (Belf 6) [1, 2, 6, 7, 12; the language is late] {Ker 310.29; 143} HomS 16 (Ass 12) [2(1×: nƝde), 5, 22(1×)] {Ker 48.22; 158} HomS 17 (= Blickling 5) [1, 2, 3, 4, 10, 12, 22(1×), 30(a); ah = ac and swƯgende = swƯgiende; also hƯe several times, but no other ] {Ker 382.5; 199} HomS 18 [2, 5, 12, 29(h), 30(e); Anglian features sparse] {Ker 309.23; 266} HomS 19 (Schaefer) [12, 13, 14, 17, 20, 22(1×), 29(h, j, p)] {Ker 32.18; 259} (HomS 20 (not in corpus); HomS 21 = Blickling 6) HomS 22 (CenDom 1) [2, 3, 5, 11, 12, 13, 14, 22, 26, 27, 28, 29(j, l, q); eorness ‘anger’] {Ker 48.22; 272} HomS 23 (CenDom 2) [none of the listed Anglian features] {Ker 45B.21; 125} (HomS 24 (24.1, 24.2) = Vercelli 1) HomS 25 [2, 3, 5, 11, 12, 13, 16, 17, 22, 29(b, j, l, p, q), 30(c, e, f)] {Ker 309.26; 434}

90

R.D. Fulk

(HomS 26 = Blickling 7) HomS 27 [12, 17(?), 26, 29(j); Anglian features sparse; includes extracts from Ælfric] {Ker 38.32; 174} HomS 28 [1, 2, 6, 12, 17, 19, 22; swƯgende; composite] {Ker 338.33; 217} (HomS 29 (not in corpus)) HomS 30 (TristrApp 2) [1, 2, 12, 13, 26, 29(j, s); composite, chiefly of Ælfric and Wulfstan] {Ker 331.52; 199} HomS 31 (Willard) [1, 2, 3, 5, 7, 12, 22, 26, 27, 29(h, j), 30(b)] {Ker 331.53; 139} HomS 32 (Baz-Cr) [1, 2, 5, 10, 12, 17, 22(1×), 23, 29(h, j, q, r)] {Ker 331.54; 148} HomS 33 (Först) [1, 2, 6, 7, 12, 19, 26, 27, 29(f, h, j, k, q), 30(b, e, f); many examples of e for æ, but none of e for y, and hence the feature is West Mercian rather than Kentish] {Ker 333.26; 215} (HomS 34 = Vercelli 19) HomS 35 (Tristr 4) [5, 12, 13, 17, 22(1×), 29(j, p); gp. dem. þeara (2×, as in the Vespasian Psalter)] {Ker 56.31; 304} (HomS 36 = Vercelli 11) HomS 37 (Baz-Cr) [4, 10, 12, 23, 26, 27, 29(h, j), 30(f)] {Ker 331.55; 108} (HomS 38 = Vercelli 20; HomS 39 = Vercelli 12; HomS 40.1, 40.2, 40.3 = Blickling 9, Vercelli 10) HomS 41 (Baz-Cr) [none of the listed Anglian features; a composite Ælfrician homily, with material from a version of HomU 27] {Ker 21.28; 201} HomS 42 (Baz-Cr) [3, 5, 7, 12, 19, 30(e); cf. also weastmes, Ɲafen, ealc, nƝafre] {Ker 57.45; 114} (HomS 43 = Vercelli 13) HomS 44 (Baz-Cr) [2, 3, 10, 12, 27, 29(j, o, q), 30(e, f)] {Ker 38.37; 162} HomS 45 (Tristr 3) [1(lƯchomena)] {Ker 38.38; 263} (HomS 46 = Blickling 11; HomS 47 = Blickling 12) HomS 48 (TristrApp 3) [2, 3, 5, 7, 12(rare), 19; geof- rather than gyf-; but ansƯene (vs. ansƝone), giefð; the language is late] {Ker 57.39; 160} HomS 49 (Brot 2) [1, 3, 6, 17; ciercean (often); Anglian features sparse] {Ker 364.c; 251} HomS 50 (KerTibC 1) [12, 19, 27] {Ker 197.a; 88} LS 1.1 (AndrewBright) [1, 2, 5, 10, 12, 18, 23, 28, 29(h, j), 30(b, c, e); hƯe, sƯe(n) (but also hiera = heora and -gieton = -gieten (2×), gehƯer- (4×), andgiet, ƗflƯemde, onsƯene (2×), hiere, gelƯefon = gelƯefen); meahtes =

Anglian dialect features in Old English anonymous homiletic literature

91

-est; gesƯe (non-WS: Campbell 1959: §237.3)] {Ker 48.64; 333} (LS 1.2 = Blickling 18 (xix).) LS 2 (DepAugust) [none of the listed Anglian features] {Ker 38.55; 18} LS 3 (Chad) [1, 2, 3, 4, 6, 7, 8, 9, 12, 13, 16, 18(earun), 19, 21, 22, 28, 29(f, h, j, k, l), 30(c, d); ic onginnu, gewƯtu; ah = ac; bolde, selde; (-)nƝow-; nalde, walde, warhte, margene (Campbell 1959: § 156); heordes = hƯredes; getihhade, betwƯh, werc (Angl. smoothing); gelƝadan; slƝnne, sl®ð (forms of slƝan); h®h-; eorne ‘angry’; dƝgulran; very frequent e for æ, but no distinctively Kentish features] {Ker 333.1; 245} LS 4 (Christoph) [1, 12, 30(e); hƯe (3×) but no other ] {Ker 216.1; 176} LS 5 (InventCrossNap) [1, 2, 3, 4, 6, 7, 9, 12, 13, 16, 18, 19, 23, 29(i), 30(b); but the language is late (e.g., has ME for OE , frequent examples of e for æ); = Ɨl®dan, dælas] {Ker 310.12; 560} LS 6 (InventCrossMor) [2, 3, 4, 5, 6, 8, 18, 19, 22; eorre; the language shows some signs of lateness; includes extracts from Ælfric] {Ker 297.a; 275} LS 7 (Euphr) [1, 6, 29(q); Anglian features sparse] {Ker 162.44; 334} LS 8 (Eust) [1, 3, 7, 10, 17, 29(h, q); Ɲghwanon; Anglian features sparse] {Ker 162.41; 473} LS 9 (Giles) [2, 3, 5, 12, 13, 19, 23, 29(d, h); note also nƝor ‘nearer’] {Ker 57.26; 548} LS 10 (Guth) [1, 2, 3, 4, 15, 17, 24, 29(h, j, o, r), 30(c); hƯe, but no other words containing ; rƝc ‘smoke’; ondrysenlic ‘terrible’; smƯcan ‘reek’; ofergeotol ‘forgetful’] {Ker 394.29; 256} LS 10.1 (Guth) [1, 2, 3, 7, 8, 12, 13, 16, 17, 25, 26, 28, 29(h, j, o, r), 30(d, f)] {Ker 344.5; 1327} LS 11 (James) [2, 3, 12; some late features] {Ker 209.11; 150} LS 12 (NatJnBapt) [1, 2, 3, 5, 9, 12, 17, 27, 29(j, k, m), 30(c, e, f); hƯe, but no other instances of except for gehƯered; dƝgol] {Ker 382.14; 164} LS 13 (Machutus) [1, 2, 3, 5, 7, 8, 12(rare), 17, 22(frequent), 23, 27(?), 29(g, h, j, k), 30(e); ending -s = -st on verbs; hƯe; fƝras with Anglian smoothing] {Ker 168; 839} LS 14 (Margaret CCCC 303) [1, 2, 3, 4, 5, 7, 8, 12(rare), 16, 18, 19, 26, 29(d), 30(b); eorre; feagre, þƝar; relatively late language] {Ker 57.23; 310}

92

R.D. Fulk

LS 15 (MargaretKer) [none of the listed Anglian features] {Ker 177C; 13} LS 16 (Margaret Cot. Tib. A. iii) [1, 2, 3, 4, 5, 7, 12, 18, 22, 27, 29(d, h), 30(b)] {Ker 186.15; 290} (LS 17.1 (MartinMor), LS 17.2 (MartinVerc 18) = Blickling 17 (xviii)) LS 18.1 (NatMaryAss 10N) [1, 4, 6, 7, 8, 9, 12, 17, 18, 19, 22(rare), 29(c, h, j, k, n), 30(c); naldæst; many late spellings] {Ker 310.16; 705} LS 18.2 (NatMaryAss 10J) [2, 12, 17, 27, 29(j, n), 30(c)] {Ker 331.72; 710} (LS 19 = Vercelli 17; LS 20 (AssumptMor) = Blickling 13) LS 20.1 (AssumptWillard) [1, 2, 29(o); westu ‘be you’; answorede; hƯe but no other ] {Ker 48.54; 25} LS 21 (AssumptTristr) [1, 4, 6, 7 (þistro aldor, wuldres aldor), 12, 14, 20, 29(h, j, k, o); sƯe (5×), hƯe (5×), but cf. hiere, hiera (once each), plus hƯe = hƝo(2×)] {Ker 32.11; 279} LS 22 (InFestisMarie) [2, 5(feala), 7(wal), 18(eam); late language (translation of a sermon composed in early 12th century)] {Ker 209.44; 182} LS 23 (Mary of Egypt) [1, 2, 4, 6, 12(not frequent), 22(rare), 29(b, h, j, q); syo (Kentish); few nonstandard spellings in stressed syllables] {Ker 162.31; 960} LS 24 (MichaelTristr) [1, 12, 14, 23, 25, 29(h, k, q), 30(c, f); fultumendum (Angl.); cynelecra; heorde = hƯrede] {Ker 32.17; 187} (LS 25 (MichaelMor) = Blickling 16 (xvii)) LS 26 (MildredCockayne) [3, 12] {Ker 138.3; 100} LS 27.1 (MildredFörst) [none of the listed Anglian features] {Ker 281; 26} LS 27.2 (SeaxburghFörst) [5(feala), 12] {Ker 281; 27} LS 28 (Neot) [2, 3, 5(feala), 12(rare), 18(eam); eorre; late forms (mid-12th cent. MS); “On þan time wæs ælfred king.”] {Ker 209.43; 177} LS 29 (Nicholas) [2, 3, 5(feola), 8, 12(rare), 19, 22(1×), 29(a); feager, getƝalan, Ɲahte, gesƝawe, also h®fde; untƝne (Kent.); heara = heora; many late forms, few distinctively Anglian ones] {Ker 57.34; 581} LS 30 (Pantaleon) [1, 2, 3, 5, 8, 13, 17, 19, 22, 29(j, k), 30(e); walde = wolde and heordum= hƯredum] {Ker 222.14; 500} LS 31 (Paulinus) [none of the listed Anglian features] {Ker 309.75; 17} (LS 32 (Peter&Paul) = Blickling 15) LS 33 (Quintin) [none of the listed Anglian features] {Ker 215.4; 8} LS 34 (Seven Sleepers) [1, 2, 5(feala), 8, 12(rare), 17, 26, 30(e)] {Ker 162.30; 772} LS 35 (VitPatr) [1, 2, 8, 12, 13, 14, 15, 17, 19, 22, 29(h, j, o), 30(a, e)] {Ker 182.2; 424}

Anglian dialect features in Old English anonymous homiletic literature

93

HomU 1 (Irv 5) [1, 2, 3, 6, 7, 9, 12; walde; ealc, Ɲafre; e for WS æ (but no y for e); ah = ac; bearn; a late text (so commonness of features like 1 is notable), with frequent u for y] {Ker 310.78; 190} HomU 2 (Irv 6) [1(frequent), 2, 6, 7, 12, 19; walde; ansƝone; HƝalend (cf. h®num); hyo; late language (same MS as the preceding)] {Ker 310.79; 268} HomU 3 (Irv 7) [1, 2, 6, 7, 8, 12, 17, 29(l); Ɲafre (cf. r®deste (® = Ɲa)); domselt = dǀmsetl/dǀmseld; geornlucost (etc.); late language (same MS as the preceding); composite] {Ker 310.80; 167} HomU 4 (Belf 13) [1, 6, 9; late] {Ker 32.82; 25} HomU 5.1 (Buch A) [1, 3, 4, 6, 7, 12; late. Note that HomU 5.1–5.7 all belong to the same text, in the same MS, all brief; all show e for æ but no e for y.] {Ker 398.3; 347 lines in toto} HomU 5.2 (Buch B) [1, 2, 5, 6, 7, 12, 18(eam); scal(t)] HomU 5.3 (Buch C) [1, 2, 3, 5, 6, 7, 12, 13, 29(h); scal(t)] HomU 5.4 (Buch D) [1, 2, 3, 4, 6, 7, 12, 13; scalt] HomU 5.5 (Buch E) [1, 2, 3, 5, 6, 7, 12, 13; scalt] HomU 5.6 (Buch F) [1, 2, 4, 6, 7, 13, 18(am)] HomU 5.7 (Buch G) [1, 2, 3, 4, 5, 6, 7, 13, 29(h); scal; Anglian smoothing in drƝigen] (HomU 6 = Vercelli 15; HomU 7 = Vercelli 22; HomU 8 = Vercelli 2; HomU 9 = Vercelli 4; HomU 10 = Vercelli 6; HomU 11 = Vercelli 7) HomU 12.1 (Först) [1, 6, 9, 12, 17, 22(2×), 24, 27, 29(h, j)] {Ker 32.12; 127} HomU 12.2 (Willard) [1, 12, 13, 27, 29(h, j)] {Ker 32.12 (part); 66} HomU 12.3 (CCCC41) [12, 24, 29(h, j, k, r); eorre] {Ker 32.12 (end); 32} HomU 13 (Hall) [2, 5, 6, 12] {Ker 398.2; 24} HomU 14 (Holt) [12] {Ker 99; 12} HomU 15 (Robinson) [1, 3, 4, 6, 12, 29(h); Kentish features: ¬ for ‰ (and ‰ for ¬ by hypercorrection); geh®rdan = LWS geh©rdon; sƯe (but no other ȯe); frequent ȯo for ¬o] {Ker 186.18; 81} HomU 16 (Kluge) [2, 5, 12; frequent e for æ, but no e for y] {Ker 186.24; 53} HomU 17.1 (Kluge E) [1, 12, 13, 14, 29(h)] {Ker 48.67; 104} HomU 17.2 (Kluge G) [2, 5, 6, 12] {Ker 209.51; 86} (HomU 18 = Blickling 1; HomU 19 = Blickling 8; HomU 20 = Blickling 10) HomU 21 (Nap 1) [1, 12, 14, 29(j), 30(b); partly Wulfstanian] {Ker 331.1; 81}

94

R.D. Fulk

HomU 22 (Nap 23) [none of the listed features] {Ker 68.2 (part); 43} HomU 23 (Nap 24) [2, 22(1×), 29(h)] {Ker 68.2 (part); 7} HomU 24 (Nap 25) [12; Wulfstanian] {Ker 331.21, 186.19i (parts); 36} HomU 25 (Nap 27) [none of the listed Anglian features; Wulfstanian] {Ker 49B.12; 41} HomU 26 (Nap 29) [1, 5, 12, 23(unger©dre s®s), 29(j, q); composite] {Ker 331.22; 266} HomU 27 (Nap 30) [1, 12, 13; Anglian features sparse; composite, with extracts from Wulfstan and the Vercelli Homilies] {Ker 331.23; 283} HomU 28 (Nap 31) [none of the listed Anglian features] {Ker 331.24; 23} HomU 29.1 (Nap 36) [5; ah = ac(4×); sennan = synnan; cf. also frequent i(e)orne = georne; Wulfstanian] {Ker 186.19e; 75} HomU 29.2 (Nap 35) [7; Wulfstanian] {Ker 331.29; 61} HomU 30 (Nap 38) [none of the listed Anglian features] {Ker 49B.15; 17} (HomU 31 (not in corpus)) HomU 32 (Nap 40) [12, 27, 29(h); bl®st (Anglian); expansion of a version of HomU 8, with extracts from Wulfstan] {Ker 144.1; 144} HomU 33 (Nap 41) [4(gesweogian, forsweogað: Kentish?)] {Ker 49B.39; 4} HomU 34 (Nap 42) [7, 12, 27] {Ker 68.1; 330} HomU 35.1 (Nap 43) [2, 3, 12, 18, 22(1×), 29(h, j, r)] {Ker 68.2; 315} HomU 35.2 (Nap 44) [1, 2, 3, 5, 12, 13, 14, 18, 22, 29(h, r)] {Ker 186.17; 351} HomU 36 (Nap 45) [2, 3, 4, 5, 12, 17, 18, 22(1×), 29(h, r)] {Ker 68.3; 195} HomU 37 (Nap 46) [3, 12, 22(1×); composite] {Ker 68.11; 279} HomU 38 (Nap 47) [2, 12; composite of Wulfstan and Byrhtferth, prob. by the latter; mentions that the millennium has passed] {Ker 69.7; 81} HomU 39 (Nap 48) [none of the listed Anglian features; prob. by Byrhtferth] {Ker 288.3; 104} HomU 40 (Nap 50) [1, 12, 17; chiefly Wulfstanian] {Ker 69.10; 255} HomU 41 (Nap 51) [ah = ac; iorne = georne (in the same MS as HomU 29.1)] {Ker 186.19h; 32} HomU 42 (Nap 52) [ah = ac (in the same MS as HomU 29.1); lƝacnian = l®cnian] {Ker 186.19k; 24} HomU 43 (Nap 53) [5; ah = ac (in the same MS as HomU 29.1)] {Ker 186.19l; 20} HomU 44 (Nap 55) [1, 2, 3, 7, 17; cf. þicgeð; Anglian features sparse; composite, chiefly of Ælfric and Wulfstan] {Ker 331.43; 207} HomU 45 (Nap 56) [6, 7] {Ker 310.37; 68}

Anglian dialect features in Old English anonymous homiletic literature

95

HomU 46 (Nap 57) [17, 22, 29(l); Anglian features sparse] {Ker 283.4, 5; 255} HomU 47 (Nap 58) [2, 8; hylle = helle; fragments of composite homilies, heavily Wulfstanian] {Ker 177A.18; 221} HomU 48 (Nap 59) [none of the listed Anglian features; HomU 48–50 are in one MS] {Ker 402.b(i); 72} HomU 49 (Nap 60) [6; Wulfstanian; mentions the Danelaw] {Ker 402.b(ii); 31} HomU 50 (Nap 61) [none of the listed Anglian features] {Ker 402.b(iii); 32} HomU 51 (NapDeInfant) [none of the listed Anglian features] {Ker 333.24; 30} HomU 52 (NapDeCogita) [none of the listed Anglian features] {Ker 333.25; 23} HomU 53 (NapSunEpis) [1, 2, 12] {Ker 38.4; 152} HomU 54 (Priebsch) [13, 18] {Ker 35.4; 110} HomU 55 (Thorpe) [1, 12, 13, 22(2×); andrisnlican] {Ker 50.2; 142} HomU 56 (Warn 43) [2, 6] {Ker 209.45; 22} HomU 57 (Warn 44) [2, 3, 8, 29(d)] {Ker 209.46; 25} HomU 58 (Nap 16) [2, 6, 8] {Ker 332.18; 77} HomU 59 (Nap 37) [none of the listed Anglian features] {Ker 177A.16; 37} HomM 1 (Healey) [1(ondryslic), 2, 3, 7, 12, 18(eam), 29(m); lyoman, syo yorðe, etc.(Kentish), hylware = helware; hige, hyge = hƯe and syge = sƯe; ondryslic; not a few late forms] {Ker 336.4; 175} HomM 2 [29(h); otherwise none of the specified Anglian features] {Ker 332.18; 37} (HomM 3 (not in corpus)) HomM 4 (McDougall) [1, 2, 9, 11, 12, 13, 14, 29(h, j); efenhiafden (Kentish); wibede (archaic, Mercian); some late forms] {Ker 182.4; 106} HomM 5 (Willard) [1, 2, 3, 12, 13, 17, 18, 27, 29(h, j), 30(c, e)] {Ker 336.6; 565} HomM 6 (Ker Otho B 10) [2, 12, 29(h)] {Ker 178; 10} HomM 7 (KerTibC 1) [none of the highlighted Anglian features] {Ker 197.g; 57} HomM 8 (Murfin) [12, 29(j, l); thoroughly Saxonized, but with some late forms] {Ker 18.40; 209} HomM 9 [2, 7, 29(r)] {Ker 57.40; 24} HomM 10 [2] {Ker 338.17; 61}

96

R.D. Fulk

(HomM 11 = Vercelli 14) HomM 12 (KerVitD 7) [none of the listed Anglian features] {Ker 407; 2} HomM 13 (ScraggVerc 21) [1, 5, 12, 26, 29(l, r); hƯe, sƯe, but also hiera; ondrysenlic] {Ker 394.26; 257} HomM 14.1 (Healey) [1, 2, 12, 29(h); lufodesðu; Kentish features] {Ker 336.2; 16} HomM 14.2 (Healey) [1, 2, 3, 4, 12, 13, 18, 22(1×), 27, 29(e, h, j, l); cf. also neahtum ‘nights’ (as in Bede, Mart, Lch, Marv)] {in Ker 336 (fols. 12r/1–17r/5); 152} (HomM 15 (Wanley): incipits and explicits of 13 homilies from a lost MS, all with very late language. {Ker 173})

Notes 1. The most relevant texts are Æthelwold’s translation of the Rule of St. Benedict (BenR) in Oxford, Corpus Christi College 197 and Cambridge, Corpus Christi College 178; Ælfric’s Catholic Homilies (ÆCHom) in London, British Library, Royal 7 C. xii and Cambridge University Library Gg. 3. 28; other of Ælfric’s homilies (ÆHom), particularly in Cambridge, Corpus Christi College 188 and London, British Library, Cotton Vitellius C. v; Ælfric’s Lives of Saints in London, British Library, Cotton Julius E. vii; Ælfric’s Grammar (ÆGram) and Glossary (ÆGl) in Oxford, St. John’s College 154 and Durham, Cathedral B. III. 32; and Ælfric’s preface to his translation of Genesis (ÆGenPref), along with the portions of the translation of the Hexateuch attributable to him (on which see Crowley 1980: 158–159, n. 84). Here and throughout, the short titles of Old English texts are those recommended by Mitchell et al. (1975, 1979) and employed by the Dictionary of Old English (Cameron et al. 2007). 2. The most substantial non-homiletic works in Late West Saxon that display Anglian features are Solomon and Saturn I & II (Sol I, II), Scriftboc (Conf 1.1 (Spindler)), the Herbarium and Medicina de quadrupedibus (Med 1.1 (de Vriend)), Lacnunga (Med 3), Bald’s Leechbook (Lch II), the translation of Boniface’s letter to Eadburga (Let 1 (Sisam)), Alexander’s Letter to Aristotle (Alex), and The Marvels of the East (Marv). Less frequent Anglian features are evident in the Worcester and Peterborough Chronicles (Chron D, E), and in the translation of Chaps. 1–16 of Alcuin’s De virtutibus et vitiis in London, British Library, Cotton Vespasian D. xiv (Alc (Warn 35)). Mercian features are especially frequent in some earlier texts: Bishop Wærferth’s translation of Gregory the Great’s Dialogi (GD), the Old English Bede (Bede), and the various fragments of the Old English Martyrology (Mart).

Anglian dialect features in Old English anonymous homiletic literature

97

3. On the reasons for associating Mercianisms with the poetic and the archaic, see Fulk 1992: §§ 337–339. 4. See, for example, Flasdieck 1924: 385–413; Hedberg 1945 6; Sisam 1956: 119–128; Campbell 1959: §§ 16, 258; Pope 1967–1968: 1.83; Gneuss 1972: 82; Scragg 1973; Roberts 1986. The counterarguments of Bately 1988, however, merit especial attention. 5. Smith does not, apparently, mean that the language of Æthelwold and Ælfric is more various than generally supposed. What he says is, “Close examination of the written record soon reveals that ‘Standard [i.e., Late] West Saxon’ admitted of a good deal more variation than standard written Present-Day English. Not only are there certain lexical habits which are restricted to certain monastic scriptoria, but there also seem to be persistent orthographic distinctions between the various scriptorial outputs. Codification of these latter features remains an important task for Anglo-Saxonists” (1996: 67). In referring to lexical habits he seems to be comparing Æthelwoldian/Ælfrician West Saxon to Late West Saxon varieties without the distinctive Winchester vocabulary – that is, varieties such as those examined in this survey. If that is the case, rather than raising a methodological obstacle, he is in the final sentence calling for the very research to which the present survey aims to contribute. His view of the cause of variation in Late West Saxon, however, is plainly in greater accord with Benskin’s suppositions than with Vleeskruyer’s. 6. The artificiality of some unsyncopated forms is also suggested by þicgeð in HomU 44. The proper form would be þigeð, and þicgeð may be a mere guess at the Anglian equivalent. 7. Space does not permit a full account of the literature on each of these features, but each is treated in one or more of the following studies: Campbell 1959, Fulk 1992, 2004, Jordan 1906, Kleinman 1997, Roberts 1986, Schabram 1965, Vleeskruyer 1953, and Wenisch 1979. 8. On the liturgical use of saints’ lives, see the introduction to Upchurch 2007. 9. Completion of the survey in time for inclusion in this volume required some selectivity, and these collections present some particular difficulties. Moreover, their language has already received extended attention: see especially Hardy 1899, Menner 1949, Peterson 1953, Schabram 1965: 77–87, and Scragg 1992: xliii–lxxi. The last is especially enlightening. 10. For information on composite homilies and the manuscript contexts of anonymous homilies and saints’ lives, see Scragg 1979, 1996.

98

R.D. Fulk

References Bately, Janet 1988 Old English prose before and during the reign of Alfred. AngloSaxon England 17: 93–138. Benskin, Michael 1994 Descriptions of dialect and areal distributions. In Speaking in Our Tongues: Medieval Dialectology and Related Disciplines. Margaret Laing and Keith Williamson (eds.), 169–87. Cambridge: D. S. Brewer. Bülbring, Karl 1902 Altenglisches Elementarbuch. Vol. I: Lautlehre. Heidelberg: Winter. Cameron, Angus, Ashley Crandell Amos, and Antonette diPaolo Healey (eds.) 2007 Dictionary of Old English: A to G Online. Toronto: Pontifical Institute of Medieval Studies. Campbell, Alistair 1955 Review of Vleeskruyer 1953. Medium Ævum 24: 52–56. 1959 Old English Grammar. Oxford: Clarendon. Crowley, Joseph Patrick 1980 The study of Old English dialects. PhD. dissertation, Department of English, University of North Carolina at Chapel Hill. Flasdieck, Hermann 1924 Zur Charakteristik der sprachlichen Verhältnisse in altenglischer Zeit. Beiträge zur Geschichte der deutschen Sprache und Literature 48: 376–413. Frank, Roberta 1994 Poetic words in late Old English prose. In From Anglo-Saxon to Early Middle English: Studies Presented to E. G. Stanley. Malcolm Godden et al. (eds.), 87–107. Oxford: Clarendon. Fulk, R. D. 1992 A History of Old English Meter. Philadelphia: University of Pennsylvania Press. 2004 Old English werg-, wyrg- ‘accursed’. Historische Sprachforschung 117: 315–322. Gneuss, Helmut 1972 The origin of standard Old English and Æthelwold’s school at Winchester. Anglo-Saxon England 1: 63–83. Godden, Malcolm 1980 Ælfric’s changing vocabulary. English Studies 61: 206–23. Hardy, Ashley K. 1899 Die Sprache der ‘Blickling Homilien’. Leipzig diss. Leipzig: Glausch.

Anglian dialect features in Old English anonymous homiletic literature

99

Hedberg, Johannes 1945 The Syncope of the Old English Present Endings: A Dialect Criterion. Lund Studies in English 12. Lund: Gleerup. Jordan, Richard 1906 Eigentümlichkeiten des anglischen Wortschatzes. Anglistische Forschungen 17. Heidelberg: Winter. Ker, N. R. 1957 Catalogue of Manuscripts Containing Anglo-Saxon. Oxford: Clarendon. Kleinman, Scott 1997 Iron-clad evidence in early mediaeval dialectology: Old English ‘isern’, ‘isen’, and ‘iren’. Neuphilologische Mitteilungen 98: 371–390. Menner, Robert J. 1949 The Anglian vocabulary of the Blickling Homilies. In Philologica: The Malone Anniversary Studies. Thomas A. Kirby and Henry Bosley Woolf (eds.), 56–64. Baltimore: The Johns Hopkins Press. Mitchell, Bruce, et al. 1975 Short titles of Old English texts. Anglo-Saxon England 4: 207–221. 1979 Short titles of Old English texts: addenda and corrigenda. AngloSaxon England 8: 331–333. Peterson, Paul W. 1953 Dialect grouping in the unpublished Vercelli Homilies. Studies in Philology 50: 559–565. Pope, John C. (ed.) 1967– Homilies of Ælfric: A Supplementary Collection. 2 vols. Early Eng1968 lish Text Society o.s. 259, 260. Oxford: Oxford University Press. Roberts, Jane 1986 The Old English prose translation of Felix’s Vita sancti Guthlaci. In Studies in Earlier Old English Prose. Paul E. Szarmach (ed.), 363– 379. Albany: State University of New York Press. Schabram, Hans 1965 Superbia: Studien zum altenglischen Wortschatz. Teil I: Die dialektale und zeitliche Verbreitung des Wortguts. Munich: Fink. Scragg, D. G. 1973 The compilation of the Vercelli Book. Anglo-Saxon England 2: 189– 207. 1979 The corpus of vernacular homilies and prose saints’ lives before Ælfric. Anglo-Saxon England 8: 223–277. 1996 The corpus of anonymous lives and their manuscript context. In Holy Men and Holy Women: Old English Prose Saints’ Lives and Their Contexts. Paul E. Szarmach (ed.), 209–230. Albany: State University of New York Press.

100

R.D. Fulk

Scragg, Donald G. (ed.) 1992 The Vercelli Homilies. Early English Text Society o.s. 300. Oxford: Oxford University Press. Sisam, Kenneth 1956 Canterbury, Lichfield, and the Vespasian Psalter. Review of English Studies n.s. 7: 1–10, 113–131. Smith, Jeremy 1996 An Historical Study of English: Function, Form and Change. London: Routledge. Sprockel, C. 1965 The Language of the Parker Chronicle. Vol. I: Phonology and Accidence. The Hague: Nijhoff. Upchurch, Robert K. (ed.) 2007 Ælfric’s Lives of the Virgin Spouses with Modern English ParallelText Translations. Exeter: University of Exeter Press. Vleeskruyer, Rudolf (ed.) 1953 The Life of St. Chad: An Old English Homily. Amsterdam: NorthHolland. Wenisch, Franz 1979 Spezifisch anglisches Wortgut in den nordhumrischen Interlinearglossierungen des Lukasevangeliums. Anglistische Forschungen 132. Heidelberg: Winter. Wilson, R. M. 1959 The provenance of The Vespasian Psalter gloss. In The AngloSaxons: Studies in Some Aspects of their History and Culture Presented to Bruce Dickins. Peter Clemoes (ed.), 292–310. London: Bowes and Bowes. Yerkes, David (ed.) 1984 The Old English Life of Machutus. Toronto: University of Toronto Press.

The elusive progress of prosodical study

Thomas Cable 1.

Overview

The three essays in this section are valuable contributions to specific problems of English historical metrics even as they raise intractable questions about general prosody. The essay by J. A. T. Smith, “Fidelity in versification: Modern English translations of Beowulf and Sir Gawain and the Green Knight,” studies the patterns of Old and Middle English alliterative meters and their incorporation into meters of Modern English. Her essay asks about the ways that linguistic stress and the abstract units of meter coincide, as does Xingzhong Li’s “Metrical Evidence: Did Chaucer translate The Romaunt of the Rose?” Whereas Smith’s essay deals with the native meters of English at two different periods, Li’s subject is a meter transported into English from French. All three meters – OE alliterative, ME alliterative, and ME iambic tetrameter – regulate lines that occur in time, but only the iambic tetrameter is clear enough to draw a useful distinction between linguistic stress and metrical beat. The present critique draws this distinction in a way that Li’s analysis does not. For Old and Middle English alliterative meter, a comparison of the originals with their ModE translations leaves one with the sense that, despite the remarkable scholarly progress of the past thirty-five years, there are aspects of both these forms that remain elusive. Christina M. Fitzgerald’s “‘Ubbe dubbede him to knith’: The Scansion of Havelok and ME -es, -ed, and -ed(e)” returns us to a methodology that predates the discoveries on which Li and Smith base their analyses. It is arguable – and argued here – that this older, traditional methodology yields results that are useful and enduring, especially as exemplified in the essay that Fitzgerald takes as her model, Smithers 1983. In his careful study of -e and -en in Havelok and the implications that he draws for Chaucer’s meter, Smithers provides an indispensable account of fourteenth-century accentual-syllabic meter. Fitzgerald’s equally careful extension of the methodology to other inflectional endings is suggestive; however, a lack of compati-

102

Thomas Cable

bility in the phonological contexts of the two essays means that a direct application of Smithers’ methodology uses both more and fewer categories than are needed. A reorganization of the categories would make for simplicity and clarity and would foreground the discoveries that Fitzgerald makes.

2.

Smith on Beowulf

Smith’s essay focuses on modern adaptations of Old and Middle English alliterative meters and in doing so contributes to our knowledge of the models. As she makes clear, her evaluations are not intended to judge the aesthetic value – or even the metrical effectiveness – of the thirteen translations, six of Beowulf and seven of Sir Gawain and the Green Knight, but to assess their fidelity to the original meters. Stating the problem this way immediately raises the question of what the most accurate statement of these medieval forms is. To measure deviation from a model, one must know the model. The benefit of this kind of analysis flows in two directions: the exercise of summarizing the medieval forms and casting them into modern forms gives the older forms salience and tangibility. At the same time, this exercise – especially the process of extracting the most relevant features – can change our understanding of the form with which we began. It may be that the modern version shows that the statement of the original needs revision, which in turn would require a revised modern imitation, at least for the goal of fidelity. OE meter is harder to understand than ME meter. Aspects of OE meter, as Smith says, are impossible to imitate in ModE, mainly because OE had phonemic length. To a lesser degree the lack of reduced vowels in OE, which ModE has in abundance, also sets up a barrier. Even so, it can be argued that Smith’s analysis does a better job of tracking fidelity in the translations of Beowulf than in those of SGGK. Her essay is an admirable first attempt at both, and it is something that metrists can build on and revise if they can show that other criteria have higher priority. Because Smith’s analysis ranks Ruth P. M. Lehmann’s translation of Beowulf highest in fidelity to Old English meter, I am immediately predisposed in favor of its criteria. It was Lehmann who gave me my introduction to medieval English metrics – and who wisely advised against intruding into scholarly writing in the first person singular. On two other matters in this critique it will be necessary to refer to my own work.

The elusive progress of prosodical study

103

A surprising result of Smith’s analysis is the low fidelity to OE meter of Chickering’s translation, a widely used text, as Smith notes. His edition is respected among Anglo-Saxonists for its scholarship, and it probably gives the impression to most specialists of being about as faithful to OE meter as any other. Part of this impression results from his crafty handling of the syntactic phrases that form the hemistichic structure, even if the hemistichs often have more or fewer than four metrical positions (the abstract units into which syllables are grouped). The accuracy of the translation and the appropriateness of the ModE diction also add to the overall feel of authenticity. Even among Anglo-Saxonists, few readers will try to scan a translation using Sievers’ Five Types, and they won’t subject Chickering’s translation to that. Obviously the greatest deviance, as Chickering himself noted (and as Smith quoted), is his disregard for alliteration. If the widespread lack of alliteration is something that does not register on the casual reader, one may wish to reconsider the priority normally given to alliteration in any ranking of features. A similar reconsideration might apply to the rules for metrical positions that Smith posits. She cautiously notes that “the rules that we currently have for metrical positions are not yet complete” (p. 131). The two examples that she gives in evaluating the results and identifying “mistakes” in the OE verse suggest the extent of this incompleteness. She scans line 736a with a stress on ðicgean and on niht: (1)

(s) (x x x x) (s) ðicgean ofer þa niht

However, Klaeber, Bliss, Pope, Fulk, and Suzuki all scan the line with primary stress on þa and secondary stress on niht. The cue of alliteration and the requirement of a four-position verse should lead the reader to the emphatic stress on the determiner: “No longer … after that night.” Similarly, for the violation in 731a: (2)

mynte þæt he gedælde

If mynte were given ictus, it would be in violation of the rule that the first lift of the off-verse determines alliteration, which here is on the [d] of gedælde. However, some rules are clearly more violable than others, and the rule for a minimum of four positions to a normal verse (and a maximum of five) should have precedence over any tendencies of alliteration, however strong they may be.

104

Thomas Cable

The main determinant of OE meter by Smith’s analysis is relative stress within a template of metrical positions. While it is reasonable to exclude phonemic length in a comparison of ModE translations, it should be noted that an unavoidable asymmetry results: the ModE translations can be compared against each other in the proximity of their meters to OE meter, yet paradoxically even the closest of the ModE meters cannot be compared in any profound way with OE meter. There is no measuring rod to say how far Lehmann’s meter diverges from OE meter. This is because the features of OE meter that are salient from our modern view are probably epiphenomenal from an Anglo-Saxon view. If a system of quantity is as pervasive as recent work in Kaluza’s Law has suggested, then this system would interact with the system of stress and with a system of syllable count in ways that ModE translations cannot possibly replicate. If it turns out that the quantitative elements are essentially in the service of a syllable-count meter, it may well be that the stressed syllables of OE poetry are too. It may be that stressed syllables serve to demarcate points in the hemistich where variations from the four-syllable norm are allowed. Historically, it may be that a four-syllable hemistich (a Germanic reflex of an Indo-European syllablecount meter) evolved into a four-position hemistich. All this is subject for further research, despite the considerable work that has been done during the past thirty years. The one thing that is clear is that no satisfactory description of OE meter currently exists. No description can grandly be said to have “explanatory adequacy,” or even “descriptive adequacy” (see Chomsky 1964, pp. 62–63). Yet the paradox is this: because the various elements are so tightly integrated, we can go a long way by beginning with any one of the various superficial parts, as modern theories have done, and showing how the other parts unfold systematically. Smith’s collection of features has the great virtue of resting on a bedrock of metrical positions. Any advances in our understanding of OE meter will have to refer to this abstract unit. The fact that Lehmann’s translation comes out at the top and Raffel’s at the bottom suggests that parts of the analysis will last. Her essay is valuable for providing the opportunity to reexamine our assumptions about OE meter. The features that she selects are basic ones for our understanding at the moment and for future insight.

The elusive progress of prosodical study

3.

105

Smith on Sir Gawain and the Green Knight

The analysis of ME meter is more problematic. Whereas Smith finds great divergence of fidelity to the metrical norm among the translators of OE, she finds the translators of ME clustered within a narrow range. If this general fidelity does not seem right impressionistically, and to this reader it does not, then the criteria that led to the conclusion might be re-examined. (Of course, the modern reader’s impressions may be wrong, as they often are.) For the b-verse, Smith uses the pattern that the present author and Hoyt Duggan arrived at independently, although through quite different assumptions. Duggan’s analysis of 5,952 lines assumed that final -e was silent, and it included no occurrences of -e in the scansions. My own analysis of 6,100 lines assumed that final -e occurred in all instances where it was historical. The implications from these different assumptions show up especially at the end of the b-verse and in the a-verse generally. My own system sees the b-verse as ending on an unstressed syllable and the a-verse as longer or weightier than the b-verse. The first of these entailments, the unstressed syllable at the end of the line, Smith disregards “because ModE pronunciation no longer vocalizes the final e in most words and because many of the final e’s that are necessary to [Cable’s] argument are reconstructed rather than extant.” Even with the loss of final -e, feminine endings in ModE are easy to get. (The previous sentence has seven words stressed / x.) Nevertheless, it is fair enough to eliminate this particular feature from the comparisons, since none of the translators adhere to it. It is the a-verse that’s a problem. The most important single metrical feature of ME alliterative verse is the asymmetry between the a-verse and the b-verse. (There are notable asymmetries in OE: A3’s, some D’s, and other subtypes, but the majority of hemistichs in OE poetry, alliteration aside, could occur in either half of the line.) In ME poetry, the a-verse is longer or heavier than the b-verse, or both longer and heavier. The b-verse is tightly and precisely constrained. The rule that Smith takes from Duggan – that the a-verse must have at least one long dip – does nothing to capture this difference. The b-verse must have a long dip too. The rule does not explain why Tolkien’s (1975) translation has the authentic feel of ME verse and Merwin’s (2002) does not. Smith attributes to my system, and then rejects, the rule that the a-verse must have two long dips. However, neither my system nor any other has such a rule. My own rule calls for either two long dips or three lifts in the a-verse. This is not a categorical rule but a gradient rule, with about 6 per-

106

Thomas Cable

cent exceptions. There is doubtless a better and fuller statement to be made, as will be suggested below, but meanwhile such a generalization gives some specificity as to how the a-verse can be longer or heavier or both. Merwin’s translation, whatever its merits, pays little attention to this asymmetry: (3)

He lifts up a corner of the curtain a little 1185 And peers out cautiously to see what it might be. It was the lady, the loveliest one to look at, Who shut the door behind her without making a sound And turned toward the bed, and the knight was embarrassed And lay back to look as though he were asleep, 1190 And with hushed steps she stole to his bed …

In six of the seven verses, the b-verse has more syllables than the a-verse, the reverse of normal ME practice. Three lines violate the Duggan-Cable bverse rule (1185, 1188, 1189), and five lines violate my version of the averse rule (1185, 1187, 1189, 1190, 1191). The fact that there is no violation of Smith’s a-verse rule raises questions about the adequacy of that rule. A metrical description should have the strength to filter out, or radically reduce, verses that are unmetrical or marginal.

4.

Li on The Romaunt of the Rose

Li’s essay is a sensitive and carefully nuanced analysis of selected metrical patterns in the three divisions of The Romaunt of the Rose. Its purpose is to provide evidence toward determining whether the third section, C, is by Chaucer. Prevailing scholarly opinion includes A in Chaucer’s corpus, definitely excludes B, and less definitely excludes C. Li’s metrical tests are profiles of the poetic line that measure trochaic substitutions, headless lines, extrametrical syllables, anapestic feet, and other similar patterns. His analysis of these features leads to the conclusion that C is Chaucerian. Although the essay begins with trochaic substitutions and gives more attention to that than to any of the other features, it is the most problematic. This critique will focus on three of Li’s half dozen tests: anapestic feet, extrametrical syllables, and trochaic substitution. The consideration of trochaic substitutions at the end will open the topic to generalizations about English prosody that can only be sketched.

The elusive progress of prosodical study

107

Anapestic substitutions, or unelidable trisyllabic feet, have an objectivity and certainty that lend themselves to a statistical analysis more readily than most other metrical patterns. Tarlinskaja in the two works cited by Li and in 1987 and 1993 has used Russian statistical methods to reveal the structure of mixed meter (dol’nik, by her importation of the Russian term) in poets from Middle English to the present, including among many others, Shakespeare, Tennyson, and Frost. With Chaucer, of course, there is always uncertainty about the text and about the possibility of elision. As Li acknowledges, it is sometimes difficult to decide whether two syllables are “fully contracted, elided, resolved, or … separate, albeit ‘weak’ syllables.” Even so, his results are impressive, both for revealing four different internal geometries of the anapestic foot in Chaucer and for demonstrating the aberrancy of the B section of the Romaunt. Under the rubric “extrametrical syllables,” Li considers the extra unstressed syllable before a syntactic break, manifested most clearly at the end of the line (traditionally known as a “feminine ending”) but also within the line. Whereas anapestic feet have a tangibility that clearly affects the rhythm of the line, unstressed syllables at the end of the line are less intrusive. This is especially true if the reading of these “feminine endings” excludes final -e, as Li’s does. Among the three sections of The Romaunt of the Rose, The House of Fame, and The Book of the Duchess, the proportion of lines with a feminine ending ranges from 1.7 percent to 5.4 percent. The fact that the B section of the Romaunt is the lowest of the five is interesting, although including final -e would surely give a fuller and better index. The motive for the present method seems to be to keep the features as objective as possible. An appreciation of that motive is clear when we turn to trochaic substitution, which is harder to determine objectively. Here, the readings proposed below diverge radically from Li’s readings. Whatever the merits of either, the point is that analysis is difficult when modern readers are not in agreement. Li has two tests involving trochees – their distribution in the line and their internal makeup – which are considered together here because of their common problem. Identifying trochaic substitutions in Chaucer is always tricky. Between stress doublets and the uncertainty of monosyllabic words, alternative readings are usually possible. In examples (6) to (8) of his essay, Li scans thirteen lines with trochaic substitutions. Ten of these can be read differently, some more plausibly than others. In the alternative readings

108

Thomas Cable

below, the slash notation (/) indicates not the most neutral pattern of linguistic stress, whatever that may be, but a reading of the rhythmical beat. The difference between stress and beat will be discussed after the examples. That difference leads to a consideration of some of the most basic problems in general prosody and the reef on which English prosody of the past two hundred years has split. Li rightly sets aside as ambiguous certain categories: Romance words, words like also and therefore, suffixes like -nesse and -yng(e), which can attract stress, and proper names like Thopas and Adams. Even so, much ambiguity remains. He identifies trochees with the notation S W, as in these lines. Explanations for the revised scansions are given at the end. (4)

S W Anoon therwith whan y sawgh this

BD 500

However, why not this? / / / / Anoon therwith whan y sawgh this (5)

S W S W Nor of a thousand full scarsly

RomB 5460

Why not? / / / / Nor of a thousand full scarsly (6)

S W And of her myshappe hem diffame

RomB 5500

Why not? / / / / And of her myshappe hem diffame

(7)

SW Aftir to thenken on hir sorwe

BD 100

The elusive progress of prosodical study

But why not this? / / / / Aftir to thenken on hir sorwe (8)

S W If that I were god of richesse

RomC 6010

Why not? / / / / If that I were god of richesse (9)

S W Of which fyve in his right hond were

RomA 940

Why not? / / / / Of which fyve in his right hond were (10)

S W SW Feble as a forwounded man

RomA 1830

Why not? / / / / Feble as a forwounded man

(11)

S

W

Ryght as this drem wol tel us alle

RomA 30

Why not this? / / / / Ryght as this drem wol tel us alle (12)

S W Loo, to the Hous of Fame yonder

HF 1070

109

110

Thomas Cable

Why not? / / / / Loo, to the Hous of Fame yonder (13)

SW Debonaire, goode, glade, and sadde

BD 860

These examples are listed in the approximate order of plausibility of the revised reading. Here briefly are the rationalizations. For the first two, the final syllable of the line is the rhyming syllable: in (4) this rhymes with hys; in (5) scarsly rhymes with certeynly. It is hard to understand what the marking of the sequence SW indicates. In no usual meaning of the term can it be a “trochee.” The emphasis on the rhyme would be lost, and the rhythm of the line would collapse into the rhythm of ordinary talk: (14)

/ x Anoon therwith whan y sawgh this / x Nor of a thousand full scarsly

What can these scansions mean at any level of description? The first line is an example of the great variability of phrasal stress from Middle English to the present and its dependence on context, including metrical context. One can imagine that “when I saw this” would be at least as normal as “when I saw this,” although even raising the question of a phrasal norm misses the point when there are clear-cut norms of meter (as Li effectively argues) that can override what we take to be the most neutral patterns of stress. The second example is simply the kind of stress shift within a word that occurs during the fourteenth century, but we shall return shortly to the distinction between stress and beat. For (6) stress was variable on words with mis-, mys- in Chaucer’s English. It is plausible to assume that in this instance the stress is on mys-. Example (7) has the word aftir, which modern readers assume is always stressed as in ModE. With evidence from the laboriously regular verse of Orm, Cable 2002 suggested the possibility of ME stress shift in after, although so far no one seems to have taken up the idea.

The elusive progress of prosodical study

111

Examples (8) to (12) all involve the promotion of a function word to metrical stress, sometimes at the expense of a lexical word. This reading may strike modern ears as artificial, but otherwise modern ears will have to contend with a breakdown of rhythm, especially in (8) and (9) where the trochaic substitution does not follow a syntactic break. Example (10) is problematic by any scansion. Giving the metrical beat to the article a is not an especially satisfying solution, but neither is starting the line with two trochees; moreover the syllables on either side of a are normally unstressed. Examples (11) and (12) have monosyllabic words as possible initial trochees, the most usual place in the line to get trochees, and the alternative readings are given with no great sense of certainty. Example (13) with line-initial Debonaire seems to be a certain example of the trochaic pattern locked within a single word, especially since stress on the first syllable is Chaucer’s normal way of doing it. Still, one wonders about the possibility of reversion to the historical source, de bon aire, and a lack of stress on de. The last three of the thirteen examples are difficult. Two of them support Li’s argument for the aberrancy of B; the other does not support his argument for the Chaucerian style of C. (15)

/ / / / A cherl chaungeth that curteis is

RomB 4030

There is no obvious way to get a regular spacing of beats in this line. It very well could have been composed by a poet other than Chaucer. Similarly, RomB 4650, With what lord thou haddist to do, is too clunky for Chaucer’s normal metrical style – in accord with Li’s argument. The following line from the C portion is also problematic: (16)

S W To have double absolucioun

RomC 6410

The fact that Chaucer does not normally pronounce the -e in the phrase to have makes this look like a trochaic substitution in second position. As Li’s tables show, this line would have more in common with the B portion of the Romaunt than with the A portion. It is necessary to step back from all these details and ask what the subject of prosodic study is. The difference between the view of the present critique and Li’s essay is not so much over whether a particular prefix re-

112

Thomas Cable

ceives stress or not. It really has to do with what constitutes a line of poetry. In his masterly annotated bibliography of writings since George Gascoigne, Brogan 1986 traced the two main streams of English prosody: temporal metrics and stress metrics. Li’s essay, like all modern linguistic prosody, belongs firmly to stress metrics. To say that a line of poetry takes place in time evokes the response from stress metrists that matters of timing are matters of performance, and therefore idiosyncratic and not a part of the enduring structure of the line. To be sure, temporal metrists over the past two centuries have done little to advance their intuition of the line as a temporal object by tripping it out with bars and measures, crotchets and quavers, and detailed scores for performance. The essence of the temporal metrist’s insight is very simple. There should be a beat. The easiest way of identifying the beat is by tapping, and tapping can be done only in time. The usual configurations of S and W positions don’t do the job, nor does the assignment of degrees of linguistic stress by the various systems. The beat in English poetry is obviously determined mainly by linguistic stress, but the two systems are not isomorphic. The normal placement of stress can destroy the beat. To give these generalizations specific force, one can try tapping four beats in the first three lines from the trochaic discussion above (4–6). One of the four beats should be placed on the syllable that Li identifies as the stressed syllable of a trochee – or the two syllables in (18): (17)

(18)

(19)

/ Anoon therwith whan y sawgh this

BD 500

/ / Nor of a thousand full scarsly

RomB 5460

/ And of her myshappe hem diffame

RomB 5500

One should precede the experiment by reading a hundred lines of Chaucer to get the tune in the head. Then it should be impossible to stress the lines as marked. Li recognizes that trochaic substitutions are “destructive” and “disruptive” to the meter. The question is whether the patterns of ordinary talk disrupt the meter or the meter disrupts the patterns of ordinary talk, whatever ordinary talk might have been for Chaucer.

The elusive progress of prosodical study

5.

113

Fitzgerald on Havelok

If the generalizations offered above seem to come from a land where only metrists live, it should be clear that they indicate principles that have been held, often implicitly, by many traditional philologists and editors. Christina Fitzgerald’s essay is an extension of work by a distinguished editor, G. V. Smithers. In both Fitzgerald’s and Smithers’ essays, words such as “beat,” “offbeat,” and “pulse” indicate the orientation. Smithers wrote: “There are occasional signs that in the metre of Hav., as of Chaucer’s fivebeat line, it is the rhythmic pulse, and not linguistic stresses, that must have primacy” (p. 197). His example is: (20) x / x / x/x / x It was a king bi are-dawes x / x / x / x / x Þat in his timҽ werҽ gode lawes

27–28

Neither the notion of a “rhythmic pulse” nor this scansion would have a place in any version of linguistic poetics of the past forty years. Fitzgerald’s point of departure is promising, because there is much in Smithers’ essay that has not been attended to since its publication in 1983. However, there is a problem in completely adopting Smithers’ six phonological-metrical contexts for determining the syllabicity of -es, -ed, and -ed(e). Smithers’ focus on -e and -en meant that his main questions were whether -e elided with the first syllable of the next word and whether -en, which would have blocked such an elision, was scribal; for example, line 1002 with the -n on komen as it appears in the manuscript, and as Smithers would emend it: (21)

x / x / x x / x / And he gart komen into þe tun x / x / (x) x / x / And he gart kome into þe tun

As is well known, during the fourteenth century final -n was often dropped, leaving -e as the final syllable. This variable usage in the language set up options in a line of poetry and the possibility of elision.

114

Thomas Cable

The questions that Fitzgerald asks are at least as interesting as Smithers’ questions (When did ‘lov-ed’ become ‘lov’d’?), and to some extent they overlap, but the wholesale importation of Smithers’ methodology brings to mind Charles Lamb’s “Dissertation upon Roast Pig.” For a century in China after the accidental discovery of the dish the method of obtaining it was the conflagration of a whole house. For the question of fourteenthcentury syllabicity, one does not need Smithers’ full architecture, which has already done its work. The contexts for -e(n) that Smithers analyzes are not analogous with the contexts for -es, whether genitive or plural, because the -s could not be dropped. Smithers’ six categories describe contexts for elision of -e, or the blocking of elision by -n, and as such they are irrelevant for the loss of syllabicity of -es, which involves syncopation, not elision: knightes, for example, becoming knights. Elision, by contrast, occurs in To take oure wey, where the -e of take forms a single syllable with oure, a situation familiar in Chaucer. For Smithers’ contexts to be relevant, it would have to be shown that the syllable of a following word affected syncopation – something, as far as I know, that no one has suggested. Instead, all that is needed for Fitzgerald’s purposes is the conclusion that Smithers came to: the meter of Havelok (like the meter of Chaucer) is very regular in its even alternation of stress, with occasional “inverted first feet” but no “trisyllabic feet” anywhere else in the line: (22) x / x / x / x / For the verbal endings -ed and -ede Fitzgerald raises the interesting question of their ambiguity. Was it “Ubbe dubb-ed” or “Ubbe dubb-de”? Whereas her whole analysis would be clarified by fewer phonologicalmetrical categories, it would also be helped, in the account of verbal endings, by more and finer-grained morphological distinctions. Specifically, an analysis of weak verbs might take into account their historical classes and sub-classes and their typical development from OE to ME. For example, Fitzgerald says of her first category, “orthographic syncope” that the “spellings suggest that the -e- of these inflectional endings was unpronounced linguistically (or ‘naturally’) as well as metrically” (p. 191); an example is demd in 2766: (23)

x / x / x / x / Til knithes hauҽden demd him rith

The elusive progress of prosodical study

115

However, demd (OE infinitive deman) is a weak verb of class 1 with a long stem. The OE preterit was demde. The ending -ed was not syncopated because it was never the regular form. (See Campbell 1959, §§ 748–49; Mossé 1952, §§ 89, 91.) Whether the ending -de was used naturally by the Havelok author cannot be determined by this line alone, because if the -e occurred in ordinary speech it would be elided before the h of him. If a full survey showed verbs like demd occurring only in elidable contexts or else where a final -e (written or not) was needed metrically, a plausible inference would be that -de was the natural form for the author. Similarly, the next four categories taken from Smithers tend to obscure the emerging generalizations by providing too many environments for elision and too few distinctions of grammar. Except for the preterits of weak verbs of class 1 with long stems like demde and the preterits of weak verbs of class 3, the consonant and vowel configurations of the next word in the line do not matter. For Environment 2, it does not matter that sittes is followed by the vowel of in; nor for Environment 3 that knictes is followed by the h of he; nor for Environment 4 that wordes is followed by the consonant of ne; nor for Environment 5 that Godardes is followed by the vowel of alþerbeste. The ending -es never elides with the first syllable of a following word. These categories make a difficult subject downright daunting. Like a spit for roasting a pig, there is a simpler way of getting at the loss of syllabicity of -es. Just begin with Smithers’ conclusion about the strict verse form, the armature of te tum te tum te tum te tum. The evidence for Fitzgerald’s important observations appears piecemeal in her essay, but putting it together is tough going. For the verbal endings -ed and -ede, the contexts for elision are relevant. However, Fitzgerald identifies these endings by the spelling of the scribe, not by their historical sources and their development from OE to ME. The point of Smithers’ essay is that the scribe of MS Laud Misc. 108 is unreliable. With the use of meter Smithers eliminates dozens of occurrences of spurious -n. For -ed and -ede we can be guided not only by the meter but also by the normal development of the forms. Fitzgerald considers a total of 166 occurrences of -ed and -ede in the seven environments and the “ambiguous” category (listed in notes 14–17, 19–20, 24). A sizable proportion of these past tense verbs is accounted for by the single word havede: 72 instances altogether, or about 43 percent of the total. In OE the third person singular of this weak verb of class 3 was hæfde, and the normal development was hafde, hadde, as in Chaucer’s “So hadde I spoken with hem everichon.” Despite the scribe’s usual, ambiguous spelling of hauede, there

116

Thomas Cable

is little reason to posit haued for the Havelok poet. In every instance that Smithers considers, he assumes syncopation of the first e. From the data that Fitzgerald has analyzed the only evidence that I can find for haued is in Environment 7 (note 20), where hauede precedes a stressed syllable beginning with a vowel. There are four instances that suggest -ed rather than -de (932, 1001, 2267, 2747); for example 932: (24)

/ x / x /x x / Huan he hauede eten inow

The evidence is never as neat as one would wish. (Despite my complaint about Fitzgerald’s proliferation of categories, they turn out to be useful, not least in helping to locate evidence contrary to my point of view!) Smithers would not be bothered by the hiatus of -e and -eten (though he would be bothered by the double offbeat in the fourth foot). In considering the choice of -e or -en (parallel to -de or -ed here), he writes that the idea that “the final -n prevents hiatus … is hardly conclusive, since incontestable cases of hiatus are very common in Hav.” He lists 26 in the first 500 lines and continues, “here again -e could provide the required unaccented syllable (with hiatus) just as well as -en” (p. 206). As Minkova (1991: 111) says, “Avoidance of hiatus is emphatically not a universal phonological principle.”

6.

Conclusions

Here are some generalizations that may both oversimplify and clarify the questions that these three essays raise. There are two main traditions of English poetry, and there are three main theoretical orientations that study the two traditions. The two traditions of poetry are the native meters (studied by Smith) and the imported meters (studied by Li and Fitzgerald). The origin of these meters is less relevant than their adaptation to the English language, and to the changing English language. “Native” and “foreign” are convenient shorthand references. The three main theoretical orientations of English prosody are harder to sort out because of their overlapping assumptions. There is a clear division between traditional (or literary) prosodists and generative (or linguistic) prosodists. Traditional prosodists in turn divide into two groups: stress and temporal. By the view of the present critique, all three schools have made sophisticated and significant contributions, and none has got it exactly right.

The elusive progress of prosodical study

117

The main issue between literary and linguistic prosodists is the assumption that literary prosodists make about the number of beats in a line of poetry in the “imported” meters: five beats in iambic pentameter, four in iambic tetrameter. This assumption, held by W. K. Wimsatt, caused him frustration in his reading of the earliest generative studies (see Wimsatt 1970 on Halle and Keyser 1966); G. V. Smithers, the editor of Havelok, was in agreement with Wimsatt on generative prosody. From the other direction, Bruce Hayes acknowledged the two cultures, literary and linguistic, and found it frustrating, his word, that Derek Attridge, a literary prosodist with a knowledge of linguistics, claimed “that exactly five beats are perceived in every line” (Hayes 1984: 919). Hayes criticized Attridge for relying on his own intuitions. This disagreement is so basic it is hard to imagine the kind of empirical evidence that would resolve it. From the view of the literary prosodist (a view that I share) the sine qua non of iambic pentameter is the occurrence of five beats, which can be tapped with a finger, a foot, a pencil. Certain configurations of syllables pronounced in a certain way and (in temporal metrics) occasional, specific kinds of pauses will allow those five beats. In most instances the sequence of syllables can be pronounced in another way that will not allow the five beats. It is arguable that this alternative pronunciation is not a different performance of the same line of poetry; it is a performance of something else, a performance of the same sequence of syllables that is not the line of poetry that the poet heard. This is an ontological difference, but it is not mysteriously ontological. It may sound Borgesian, but it is straightforward. By this understanding, to ignore the line in time is like studying a score by Beethoven by attending only to the notes and ignoring the rests. (The fact that the Western writing system does not systematically indicate the various kinds of pauses in prose and poetry is another story.) The juxtaposition of the issues raised in these three essays throws light in several directions. Perhaps the most interesting light falls on the alliterative line of OE and of ME as seen from the line of The Romaunt of the Rose and of Havelok. What was perceived and counted as a metrical beat in Beowulf? What was perceived and counted as a metrical beat in Piers Plowman?

118

Thomas Cable

References Bliss, A. J. 1967 The metre of Beowulf. Rev. ed. Oxford: Blackwell. Brogan, T. V. F. 1981 English versification, 1570–1980. Baltimore: Johns Hopkins University Press. Cable, Thomas 1991 The English alliterative tradition. Philadelphia: University of Pennsylvania Press. 2002 Issues for a new history of English prosody. In: Studies in the history of the English language: a millennial perspective. Donka Minkova and Robert Stockwell (eds.), 125–151. Berlin: Mouton de Gruyter. Campbell, Alistair 1959 Old English grammar. Oxford: Clarendon Press. Chickering, Howell D. 1977 Beowulf. Garden City: Anchor Press/Doubleday. Chomsky, Noam 1964 Current issues in linguistic theory. In: The structure of language: readings in the philosophy of language. Jerry A. Fodor and Jerrold J. Katz (eds.), 50–118. Englewood Cliffs, NJ: Prentice Hall. Fulk, R. D. 1992 A history of Old English meter. Philadelphia: University of Pennsylvania Press. Halle, Morris, and Samuel Jay Keyser 1966 Chaucer and the study of prosody. College English 28: 187–219. Hayes, Bruce 1984 Review of Derek Attridge, The rhythms of English poetry. Language 60: 914–923. Klaeber, F., ed. 1950 Beowulf and the Fight at Finnsburg. 3rd ed. Boston: Heath. Lehmann, Ruth P. (trans.) 1988 Beowulf. Austin: University of Texas Press. Merwin, W.S. (trans.) 2002 Sir Gawain and the Green Knight. New York: Alfred A. Knopf. Minkova, Donka 1991 The history of final vowels in English: the sound of muting. Berlin: Mouton de Gruyter. Mossé, Fernand 1952 A handbook of Middle English, trans. James A. Walker. Baltimore: Johns Hopkins Press.

The elusive progress of prosodical study

119

Oakden, J. P. 1930– Alliterative poetry in Middle English. 2 vols. Manchester: Manches1935 ter University Press. Pope, John Collins 1966 The rhythm of Beowulf. 2nd ed. New Haven: Yale University Press. Smithers, G. V. 1983 The scansion of Havelok and the use of ME -en and -e in Havelok and by Chaucer. In: Middle English studies presented to Norman Davis in honour of his seventieth birthday. Douglas Gray and E. G. Stanley (eds.), 195–234. Oxford: Clarendon Press. Suzuki, Seiichi 1996 The metrical organization of Beowulf: prototype and isomorphism. Berlin: Mouton de Gruyter. Tarlinskaja, Marina 1987 Shakespeare’s verse: iambic pentameter and the poet’s idiosyncrasies. New York: Peter Lang. 1993 Strict stress-meter in English poetry compared with German and Russian. Calgary: University of Calgary Press. Tolkien, J. R. R. (trans.) 1975 Sir Gawain and the Green Knight, Pearl, Sir Orfeo. Boston: Houghton Mifflin Company. Wimsatt, W. K. 1970 The rule and the norm: Halle and Keyser on Chaucer’s meter. College English 31: 774–788.

Fidelity in versification: Modern English translations of Beowulf and Sir Gawain and the Green Knight

Jennifer Anh-Thu Tran Smith I would ask [those more learned], as scholars, to consider whether a literal prose version of a verse epic is, properly, a translation. – Alexander 49

1.

Introduction: Purpose, selection of the texts, and parameters of study

When Seamus Heaney won the Whitbread Book of the Year in 2000, beating out the tremendously popular Harry Potter and the Prisoner of Azkaban by J.K. Rowling, the deciding sentiment seemed to be one of literary salvation. As Dr. Eric Anderson, the judges’ chairman expressed it, “We felt this was a master poet breathing new life into a work that has only been known to a tiny number of academics in the past … [It retrieved] for the reading public a buried treasure.” A great accomplishment indeed. Yet, for the “tiny number of academics” to whom Beowulf has never been lost, Dr. Anderson’s sentiments may have come as a surprise. For as Howell D. Chickering, himself a translator of Beowulf, noted, “roughly one new translation” of Beowulf has appeared “every two years” since the turn of the century (2001: 6–7). In translation, at least, Beowulf is anything but inaccessible. Yet, Dr. Anderson’s sentiments may have more merit than they appear to initially. For a great number of the reading public and for all of the students who will henceforth read his translation of Beowulf in their university approved Norton Anthologies, Heaney has resurrected and made cool an obscure, old poem. The translation’s success on the popular market highlights anew what the glut of translations confined to the academic arena could not – that translation itself, not the original text, is the most important part of maintaining the accessibility of literature’s oldest, most revered, and least read works. In many ways, translations are more important than the original itself.

122

Jennifer Anh-Thu Tran Smith

The great responsibility that they have, therefore, in this mediating capacity necessitates a greater discretion when it comes to merit than popularity alone can offer. In texts that purport to translate (not adapt) a story, fidelity to the meter and meaning of a poem is not just important, but necessary. Though the kind of holistic and impressionist evaluations that most frequently speak to this notion of fidelity are useful, this paper seeks to challenge what it means to be a faithful translation of verse through the use of quantitative modes of linguistic analysis. In the dichotomy of meter and meaning described above, it is a focused study on what it means to be faithful to the meter. In particular, it is a study on what it means to be faithful to the requirements of Old and Middle English alliterative verse – both instances of foreign or semi-foreign languages that students of English must encounter in the academic classroom. It is not a study of the fidelity of meaning or sentiment; it does not attempt to quantify in any way, whatsoever, that even greater part of the translator’s task. The immediate limitations of this approach, therefore, should be apparent. The results are intended to speak to the ability of Modern English translators to inhabit a foreign poetic medium. They are intended as a way to illuminate the rules that govern alliterative verse, a way to enter the conversation about how different Old, Middle, and Modern English are, and a way to reconsider what it means to read a translation in verse. It is not intended to undermine the legitimacy and value of any of the translations, but rather an attempt to introduce a new methodological approach to evaluating translation and verse. It is my hope that this particular analysis will prove salient to not just teachers and students of Old and Middle English literature, but also to scholars of verse and to scholars of translation in general. Perhaps the next time a Nobel Prize winning poet decides to translate a piece of medieval literature, or even in the near future when Michael Drout publishes J.R.R. Tolkien’s translation of Beowulf, that in addition to one’s sense of goodness that such a text may have, there is also a real knowledge of fidelity for at least one aspect of the poem. I have chosen Beowulf and Sir Gawain and the Green Knight as the original texts against which to measure their alliterative translations. I have selected these two texts not just because they are prominent and frequently taught pieces from medieval England, but also for a couple of other expedient factors. The first is that only one extant manuscript of each of these works exists. Though a hindrance in most other comparative studies, the fact that each text only has a single source aids in attributing variations

Fidelity in versification

123

between the ModE translations to the translators themselves and not to their source material. The second helpful characteristic is the variety and availability of translations available for each text. This variety allows for an analysis that can determine whether changes in verse are due to the qualities of Modern English prosody or to the abilities and attitudes of the translators themselves. In order to restrict this examination to those translations that would be of most value, I confined the texts to those which possessed at least the first, and as many of the proceeding characteristics as possible listed here: 1) the translation is in alliterative verse, 2) the translation either is or has been a popular edition for classroom use, and 3) the translations are distributed roughly over the last half of the century. The reason for the first requirement should be obvious. The second is important for its pedagogical implications. The third is simultaneously important for its historical currency and its historical continuity. The limitation of approximately sixty years ensures that the modern texts are not so linguistically divergent that they show substantial differences among themselves. The historical continuity is necessary to clarify whether changes in fidelity are due to prosodic differences or to changes in the philosophical climate of translation itself – that is, whether the translators are liberal or conservative with their translations due to the language or to some other larger trend within the academic community. This last point, which is difficult if not impossible to measure, may account for the inclusion of a facing page copy of the original text in Chickering, Alexander, Finch, Vantuono, and Merwin’s translations. Translators themselves may take more liberties when they know beforehand that the reader may avail him or herself of a facing page translation. As Casey Finch explains in the introduction to his translation of SGGK, “The presence of these editions of the original poems gives me encouragement that the reader will never be led astray by any liberties I might have taken in the translation” (ix–x). In assessing the final results of this metrical study, therefore, one should take into account that there is this other factor influencing the translators’ decisions to deviate from the original texts. Following these guidelines, I selected translations of Beowulf by Charles W. Kennedy (1940), Burton Raffel (1963), Michael Alexander (1973), Howell D. Chickering (1977), Ruth Lehmann (1988), and finally, Seamus Heaney (2000). Translations of Sir Gawain and the Green Knight include those by Brian Stone (1959), Marie Borroff (1967), J.R.R. Tolkien (1975), Keith Harrison (1983), Casey Finch (1993), William Vantuono (1999), and

124

Jennifer Anh-Thu Tran Smith

W.S. Merwin (2002)1. Of the above, all of the translations of Beowulf and those by Borroff, Tolkien, and Finch have been popular within the classroom. Lehmann, Tolkien, and Borroff are all notable scholars of alliterative verse. Heaney, Tolkien, and Merwin, furthermore, are notable authors and poets in their own rights. As for the limits of my survey, I have evaluated 392 lines of poetry from two different passages in Beowulf. The first passage is from the scene in which Grendel enters Heorot the night of Beowulf’s arrival in Denmark. In the original OE, this passage consisted of 27 lines (710–736), although the lengths of the translations do vary. The second passage from Beowulf is from the scene of Beowulf’s death, and it consists of 30 lines (2694–2723) in the original Old English, also with varying lengths in the translations. For SGGK, the scope of my scansion includes two stanzaic paragraphs from each translation and from the original ME for a total of 344 alliterating lines of poetry. The first stanzaic paragraph is from the beginning of the work and describes the court of King Arthur just before the Green Knight enters (37–59). The second stanzaic paragraph is from the first morning that Gawain awakes in the castle of Sir Bertilak and Bertilak’s wife attempts to seduce him (1178–1207). As a study of alliterative verse, I have excluded from the verse analysis the bob and wheel at the end of each stanzaic paragraph in SGGK. For purposes of narrative clarity, however, the bob and wheel are included in the line numbers cited above.

2.

Methodology: Stress, scansion, and metrical violations

As a quantitative analysis of fidelity, this examination is largely based upon a deviational methodology. Or simply put, how much the scansion of the Modern English poetry deviates from the Old English and Middle English standards. Because these violations are so important in the final determination of verse fidelity, the stability of the scansion is of immediate consequence to the final analysis. How I value the various qualities of the meter and the inevitable variation inherent in scansion will affect the final results of this survey. To address the first issue, I have taken a conservative approach. I have scanned each line of verse so that as few violations as possible are counted. For example, in Lehmann’s line 2721:

Fidelity in versification

(1)

125

(w s)(w) (s) (w) / (s) (w) (s) (w) the battle-weary, blood-stained chieftain

I have scanned the compound words in two different ways in order to reduce the number of violations. In the on-verse, I have scanned the line so that the second component of the compound has equal stress. In the offverse, I have scanned the second component of the compound so that it has no stress at all. Although Sievers’ types D and E allow for secondary stress on compound words, [s w / s s w] and [s s w / s ø ] respectively, the pattern [s s / s w] is not allowable. With the more conservative method, the resulting verse is [s w / s w]. The pattern falls into Sievers’ Type A, the most well-attested verse line, and therefore, also the most preferable. The only exception to this policy of conservative scansion is in the subordination of positional and alliterative qualities to the sense units that determine the beginning and end of each verse.2 In order to address the variation in scanning, I have attempted to reduce the inevitable variation that individual judgment plays in this study by following very methodical rules of scansion. I have scanned each line of verse using the same rules, which I follow in the same order each time. I began by scanning each line of the excerpted passages identifying the hemistiches, the alliteration, and the stresses, and only lastly did I determine what violations existed. This analytical breakdown from Kennedy’s translation of Beowulf serves as a straightforward example: (2) 2704

(w s) (w w) (s) (w)/ (s) (w w) (s) He wore on his corselet, cutting and keen, on-verse: lift, dip, lift, dip off-verse: lift, dip, lift alliteration: c/k (3x) violation(s): missing one dip in off-verse initial finite verb counted as strong last lift alliterates

I first began by scanning the line of verse for strong, “s,” and weak, “w,” syllables following the conservative rules for stress determination as described by Tom Cable (1991). I looked predominantly at lexical category rather than alliterating consonant, scanning some syllables as strong even when they did not alliterate on the basis of the following rules: 1) nouns, adjectives (except indefinite and interrogative pronominal adjectives), in-

126

Jennifer Anh-Thu Tran Smith

finitives and participles always receive metrical stress, 2) finite verbs and adverbs may or may not receive metrical stress, and 3) articles, prepositions, conjunctions, auxiliary verbs, linking verbs, the verb have, pronominal adjectives, and pronouns (personal, demonstrative, indefinite, relative, an interrogative) do not receive metrical stress unless they occur at the end of the half-line (Cable 1991: 80). Next, I divided the strong and weak stresses into metrical positions, as is indicated by the parentheses. If not already done for me by the translator, I also indicated the division between hemistiches with a forward slash. Then I parsed those positions into a series of lifts and dips below the line of scanned verse. I determined the alliterating consonant, which I indicated by bolding the applicable letters, and I noted that consonant below. After this preliminary work, only then did I begin to determine which violations, if any, the verse possessed. Though the rules which describe OE and ME norms are not yet fully described, for the purposes of my analysis, I have had to work on the assumption that the set of rules that I initially decided on were infallible. Only in the analysis was I able to evaluate the validity of the initial descriptive patterns. Though I have been and will continue to use the term “rule” to describe the theoretical requirements therefore, I must emphasize that the rules are no more than tendencies. Most of them are well supported by metrical evaluations, but certainly not all. The “circular reasoning” that Stockwell and Minkova (1997: 59–60) and Hoyt Duggan (2001: 479) have all highlighted – that the text is only as good as our knowledge of the rules and our rules only as good as the text – is the challenge of finding out what governs alliterative verse. It is, after all, very difficult to “distinguish the unmetrical from the rare” (Duggan, 2001: 482). With that said, however, the rules for the OE verse are actually quite well described. I have based my rules on the combination of Sievers’s and Heusler’s work that Minkova and Stockwell have formulated. What follows is a list of violations derived from those rules. There is less consensus on the ME alliterative rules, especially in the onverse. While I began my analysis therefore, basing all of the rules on Tom Cable’s patterns, I eventually discarded one of them as inapplicable. The rule in question is whether or not a regular on-verse must have two strong dips.3 Hoyt Duggan’s assessment is that it need have only one strong dip. (I will explain how I came to the conclusion that Duggan was correct later in my analysis.) The following chart listing the violations for the ME alliterative verse, therefore, are a combination of both Cable and Duggan’s rules.

Fidelity in versification

127

Table 1. Old English Positional Violations Moderate: one (two) too many lifts in on-verse one (two) too many dips in on-verse one (two) too many lifts in off-verse one (two) too many dips in off-verse missing one (two) lift(s) in on-verse missing one (two) dip(s) in on-verse missing one (two) lift(s) in off-verse missing one (two) dip(s) in off-verse initial finite verb counted as strong Alliterating Violations Major: no alliteration Moderate: no alliteration in on-verse no alliteration in off-verse Minor: alliteration not from first lift both lifts do not alliterate in on-verse w/anacrusis first lift of off-verse does not alliterate last lift alliterates

Violations of the verses fall into two broad categories: positional and alliterating. Positional violations are defined as deviations from the four (or five) stressed line of verse, and alliterating violations are defined as missing or misplaced alliterating syllables. These broad categories are further divided into major, moderate, and minor subdivisions each with an assigned weighted value of 2, 1, and .5 respectively. Because not all violations are equal in terms of their effect upon the lines of verse, these weighted values allow for a more effective comparison of the verse fidelity of each author both within the linguistic groups of Old and Middle English and across the linguistic and metrical divide of OE alliterative verse and the form that it takes during the Alliterative Revival. All graphed values at the end of this paper and the subsequent ordering of performance, therefore, will reflect final weighted scores rather than an absolute number of violations.

128

Jennifer Anh-Thu Tran Smith

Table 2. Middle English Positional Violations Moderate: Too many lifts in on-verse Too many lifts in off-verse Missing lifts in on-verse Missing lifts in off-verse Minor: Missing one strong dip in on-verse One lift in off-verse not preceded by strong dip Two lifts in off-verse preceded by strong dips Final dip in off-verse is strong Alliterating Violations Major: No Alliteration Moderate: No alliteration in on-verse No alliteration in off-verse Minor: First lift of off-verse does not alliterate Alliteration on unstressed syllable

As a cross-linguistic comparative work, however, there are a few rules that I could not include in a study of fidelity due to inherent prosodic or practicable differences. These exceptions included resolution, silent beats, hypermetric verses (i.e. Type 3), and Type A3 verses in the OE poetry and Cable’s requirement of one and only one final unstressed syllable in the ME poetry. I did not take into account resolution in the metrical analysis because ModE has no such linguistic quality. Allowing its use in the scansion of the OE, therefore, and not in the scansion of the ModE is inequitable. I have excluded silent beats, hypermetric verses, and Type A3 verses in the OE analyses because these rules are still tendentious and most of the translators would not have included them as requirements. I have excluded Cable’s rule of a final unstressed syllable because ModE pronunciation no longer vocalizes the final e in most words and because many of the final e’s that are necessary to his argument are reconstructed rather than extant. At its fundamental level, the weights of each violation are dependent upon the number of positions that such a deviation affects. Since most posi-

Fidelity in versification

129

tional violations only affect one position, their weighted score is just one. Those positional violations that are less severe because they depend not on the absence or presence of extra positions so much as the presence of the wrong types are weighted 0.5. The same logic applies to the absence of alliteration in either the on-verse or the off-verse. Either earns a weighted value of 1 and both earn a weighted value of 2. Like the positional violations, alliteration that is generally incorrect in use, though not wholly absent, is assigned a weighted value of 0.5. In the rules for the ME alliterative verse, although dips are regulated, I have excluded the absence of dips as violations because the rules that necessitate the weight of the dips assume their presence. Counting both the absence of the dips and their violation of type in such an instance would result in a double counting. The violations that read “one lift in off-verse not preceded by strong dip” and “two lifts in off-verse preceded by strong dips” actually result from a single rule, that one and only one lift is preceded by a strong dip in the off-verse. The first rule then is the absence of any strong dips at all, and the second the presence of too many. Some of the raw calculations of violations on the proceeding charts have fractional numbers. For example, under the raw tabulation of Kennedy’s “No alliteration in on-verse,” he receives a score of 6.5 for both passages. This fractional number is how I have dealt with the incidence of two types of alliteration in a single line of poetry – that is, an alliteration that follows the scheme aabb or abba. For example, the violations for Kennedy’s line 728 and 2699 fall into these respective categories: (3)

The savage monster was minded to sever alliteration: m (2x); s (2x) violation(s): if s, last lift alliterates if s, first lift of off-verse does not alliterate if m, alliteration not from first lift A little lower the stalwart struck alliteration: l (2x); st (2x) violation(s): if l, no alliteration in off-verse if st, no alliteration in on-verse if st, last lift alliterates

Because the Old and Middle English verse does not exhibit such a phenomenon and both of these patterns violate other metrical rules, in looking

130

Jennifer Anh-Thu Tran Smith

at the modern translations, I have counted violations that result from a line with double alliteration as half a violation in order to give preference to neither one set of alliteration nor another.

3.

Results

3.1. OE positional fidelity Beginning with the first passage of the Beowulf analysis, two results are immediately noteworthy. The first is the fact that the original OE is not without mistakes and has earned a score of 5. All of these violations, however, fall under the category of missing positions, an indication that the use of silent beats or Type 3 verses would in fact be quite beneficial to the scansion of this passage. For example, line 736 of the OE has the following conservative scansion: (4) 736

(s) (w w w w) (s) / (s) (s) (w) (s) ðicgean ofer þá niht. Þrýð-swýð behéold, on-verse: lift, dip, lift off-verse: lift, dip, lift alliteration: Þ (2x) violation(s): missing one dip in on-verse

If, however, an empty position is counted before “ðicgean,” or the first syllable of “ofer” is counted as stressed, the line could be considered metrically regular. If I had considered A3 verses as an allowable, but rare metrical type, line 731 in the OE would have also come out without any violations: (5) 731

(s)(w w w w)(s)(w) / (w w) (s) (s)(w) mynte þæt hé gedǙlde, Ǚr þon dæg cwóme on-verse: lift, dip, lift, dip off-verse: dip, lift, lift, dip alliteration: d (2x) violation(s): alliteration not from first lift

As is, however, the evident alliteration on the second lift of the on-verse results in a conflict with the seeming requirement that all alliteration is determined by the first lift of the on-verse.

Fidelity in versification

131

In addition to the anomalies of the OE, the analysis of the first passage also shows Raffel’s translation to be substantially higher in positional violations than the others. For the first passage alone, he earns a score of 33.5 as compared to the next highest of 19.8 (Heaney) or the lowest of 8.75 (Lehmann). Raffel continues this trend in the second passage, as well, where he earns a score of 37.5 as compared to Chickering’s second highest score of 24.5 and Lehmann’s lowest score of 7.5. What is most interesting about the second passage, however, is that Lehmann’s score is actually better than the original, with a score of just 7.5 as opposed to 9.5. Unlike the first passage, the types of violations that occur in the OE encompass extra-metrical positions in addition to missing positions. The relative prominence of three lift verses, the hypermetric verse, even in just this short passage is quite compelling evidence in support of them as rare but regular features of OE alliterative verse. Hutcheson has calculated that these verses exist in 2.3 % of Beowulf’s on-verses and 4.8 % of the off-verses. The fundamental problem with the present rules is that a ModE translator, limited by subject matter, trying to translate the story of Beowulf, should not be able to reproduce verse that is more accurate in form than the original, especially when our descriptive rules for OE verse are derived from the original itself. Such an event indicates that the rules that we currently have for metrical positions are not yet complete, or at least not complete without the addition of silent beats and hypermetric verses. The cumulative results for both passages follow very much the same patterns as the individual passages. Lehmann is close behind the original OE, followed by Kennedy, Alexander, Heaney, Chickering, and lastly, Raffel. 3.2. OE alliterating fidelity In the data for alliterating violations, Chickering leads the group as the negative outlier with a total score of 86. The next highest number is Raffel’s with 56.5, and the lowest is Lehmann’s 2.5. The reason for Chickering’s poor score is that for a substantial number of lines he fails to alliterate at all, thereby earning him many major violations with a weighted value of 2. Raffel, the outlier in the last category, does not do as poorly in this comparison, though his performance is still less than middling. On the other hand, Lehmann is still the positive outlier in this group of translators. She is

132

Jennifer Anh-Thu Tran Smith

the most consistent in her reproduction of alliteration according to the metrical norms. What is very striking here is the relatively low number of alliterating violations in the Old English verse. Unlike the results for the positional violations, the OE is definitively the best at following the rules of OE metrics, which indicates that, unlike the OE positional rules, the OE alliterating rules are indeed an accurate schematic. The only violation of the OE alliterative verse rules is for the rule “alliteration not from first lift.” This instance, the same as discussed above under positional violations, indicates the relevance of the A3 verse type. Overall, Lehmann is the most able to translate the alliterative verse, followed by Alexander, Kennedy, Heaney, Raffel, and Chickering. 3.3. Overall fidelity of the OE translators In the cumulative results of the survey, the most faithful of the translators by a large margin is Ruth Lehmann. She is followed by Alexander, Kennedy, Heaney, Raffel, and finally Chickering. Lehmann’s remarkable results indicate that Beowulf can indeed be translated with a faithful rendering of the verse. Yet, the less than faithful performances of Raffel and Chickering (still one of the most popular translations) indicates that verse form alone does not constitute the aforementioned popularity factor. The last set of comparative data in the appendix, which shows the percentage of lines without any violations at all, is illuminating in a different way. Naturally, the Old English does the best; however, it is Chickering’s that is most interesting. For, in just under sixty lines of verse, not a single line follows all of the metric rules. This outcome is consistent with what Chickering himself says of his philosophy in translating the text. He claims that “long stretches” of alliteration would “stupefy the most ardent reader” and that “the form of the original is best conveyed by a four-stress line with a heavy caesura” (1977: xi). His preferencing of the stress over the alliteration however does not extend to a precise translation of the number of stresses per line either. In an effort to “follow the natural emphasis of the Modern English sense,” there are “sometimes … five beats in a line instead of four” (1977: xi). That Chickering veers from these fundamental metrical points is especially important because his comments are easily overlooked by the individual simply looking for a verse translation of Beowulf. Although his translation is ostensibly written in a verse form, as Chickering

Fidelity in versification

133

himself says, “The translation is not a metrical imitation” (1977: xi). It is another kind of verse altogether. 3.4. ME positional fidelity As expected, the results of the survey of ME alliterative verse are less distinctive than the preceding study on the ModE translations of Beowulf. Though Tolkien performs the best out of the translators, the greatest spread between the translators is just 7 points. As I already alluded to earlier in this paper, however, the analysis demonstrated that one of the original rules that I included was not effective. This rule required that a regular on-verse must have two heavy dips if it was not extended (extra-metrical). When I initially scanned the verse using this requirement, I tabulated violations in two categories: two dips not strong in regular on-verse (2x) and two dips not strong in regular on-verse (1x). The first was a moderate violation; the second was a minor violation. What I found, however, was that most of the Gawain Poet’s, and indeed most of everyone’s violations, resulted from “two dips not strong in regular on-verse (x1).” When tabulated according to these criteria, the fact that the Gawain Poet has fifteen instances of such a case in just these cited lines, when his next highest tally is less than half that, indicates that this rule is likely a faulty one. Taking into consideration Duggan’s suggestion that the rule governing the on-verse necessitate only a single strong dip, (491) the absolute quantity of lines affected by such a violation would be considerably less. Duggan and Cable are in agreement about the relatively greater weight of the on-verse as compared to the off-verse: A-verses … tend to be somewhat longer than b-verses, to incorporate into their structure more polysyllabic words, and usually to carry more semantic weight. The majority have two staves with a varying number of unstressed syllables in the dips. Some appear to be metrically much heavier than others, and some metrists consider it an open question as to whether there are only two metrically stressed syllables in the a-verse or whether some verses have three or even four. (Duggan, 2001: 481–2)

Duggan also observed, however, that “most Middle English alliterative poets wrote two-stressed a-verses without two strong dips in fewer than fifteen percent of their lines” (2001: 488). When taking into consideration the fact that the Gawain Poet would have 0 such violations as compared to

134

Jennifer Anh-Thu Tran Smith

the 15 based on Cable’s more stringent rule, and that such a reevaluation of the rules governing the verse line would result in all of the poets performing much better in this survey, I decided to recalculate all of the data counting only the violation of “missing strong dip in on-verse.” The new tabulation resulted in Stone having only 1 violation as opposed to 13 in this category, Borrof having only 2 as opposed to 15, Tolkien only 1 as opposed to 15, Harrison 4 instead of 20, Finch none instead of 13, Vantuono 3 instead of 16, and Merwin 4 instead of 16. Whether all of these authors would be able to address the second half of Duggan’s rule, that both lifts in the on-verse alliterate, is not measured here. Likely, some of those poets, especially Harrison and Merwin who do not consistently alliterate, would perform less well in the overall evaluation of alliterating violations. 3.5. ME alliterative fidelity The most distinctive results in alliterative fidelity are those of Tolkien, Harrison, and Merwin. Tolkien, with a final weighted score of 1, actually performs better than the Gawain Poet, whose score is 2. The difference between the two, however, is minimal. (A larger survey of verse would likely sway the results back in the Gawain Poet’s favor.) Harrison and Merwin, however, unlike all of the other translators, all of whom score under 10, score 21.5 and 33 respectively. Most of their violations are due not to the misuse of alliteration, but to the absence of alliteration altogether. Like Chickering in his OE translation, alliteration is not the distinguishing metrical element of their translations. Most of the translators, however, still chose to include alliteration even if it was on an unstressed syllable. Though Borroff states that the Gawain Poet does on occasion alliterate on unstressed syllables, this practice is considered a variant rather than a standard practice (1967: 56). In the lines examined, however, this variant practice has been employed by all of the translators except Tolkien and Harrison. Tolkien seemingly did not need such a metrical loophole and Harrison seemed to prefer no alliteration over alliteration on an unstressed syllable. The increase in use of such a metrical practice in the ModE translations is likely due to the visual as opposed to oral nature of the poetic form. Because these translators are composing on the page more than they are necessarily composing out loud, the visual repetition of “sounds” is as impor-

Fidelity in versification

135

tant as the stress that governs alliterative verse. For example, Vantuono’s translation of line 53 clearly alliterates on the /k/ sound in “king,” “commands,” and “court.” Table 3.

(w w w w) (s) (w) (s) / (w w) (s) (w) (s) And he, the most gracious king, who commands the court; on-verse: DIP, lift, dip, lift off-verse: DIP, lift, dip, lift alliteration: k/c (3x) violations: two dips not strong in regular on-verse (x1) alliteration on unstressed syllable first lift of off-verse does not alliterate Yet, in modern pronunciation, the stress of “commands” rightfully belongs to the second syllable of the word. If “commands” had been in any other position, the alliteration on the “c” could be ignored as coincidence rather than necessary alliteration. Since the first lift of the off-verse must always alliterate, however, and in this instance it does not, but the first semantically prominent word in the off-verse does, albeit on an unstressed syllable, the alliteration seems to be taking a more dominant role. Note that the alliteration on unstressed syllables only occurs on the front end of words. Consonance alliteration on unstressed internal syllables is not a factor. This restriction reinforces the understanding that alliteration for modern translators is as much a visual phenomenon as an aural one. 3.6 Overall Fidelity of the ME Translators In terms of overall lines without any violations at all, only twelve percent separate Tolkien from the Gawain Poet and all of the poets are able to write at least a few lines without any metrical violations. The relative success of these poets contrasts with the OE translators, one of whom, Chickering, does not have any lines without violations, and one of whom, Raffel, only manages one line without violations.

136

4.

Jennifer Anh-Thu Tran Smith

A comparison of the ModE translations of OE and ME verse

The fact that all of the ME translators are able to achieve some degree of success in translating the verse form is likely due to the higher coincidence of ME and ModE words. They simply did not have to look as far afield to find words that alliterate. For example, compare the ModE translations of line 1185 from SGGK: Table 4.

Gawain Poet

A corner of þe cortyn he caٕt vp a lyttel,

Brian Stone

Caught up the corner of the curtain a little

Marie Borroff

A corner of the curtain he caught back a little

J.R.R. Tolkien

a corner of the curtain he caught up a little,

Keith Harrison

Lifts back the corner of the curtain, just a little,

Casey Finch

And, catching a corner of the curtain, lifts it up.

William Vantuono

A corner of the curtain he caught up a little,

W.S. Merwin

He lifts up a corner of the curtain a little

“Corner” and “curtain” appear in every single translation and “little” appears in six out of seven of the translations. Some version of “caught” appears in five out of seven of the translations, and the final two both use “lift.” The striking similarity of vocabulary in a line without proper nouns (for these are invariable in any language), indicates the increased facility in maintaining the metrical requirements. Though on occasion the translations of Beowulf are close in syntax, they never show as close a coincidence of words and syntax simultaneously. Line 711 from Beowulf represents the closest that translations come to one another through OE: Table 5.

OE poet

Grendel gongan,

Godes yrre bær,

Charles W. Kennedy

Grendel came creeping, accursed of God,

Burton Raffel

Hills and bogs, bearing God’s hatred,4

Fidelity in versification

Michael Alexander

Grendel stalking; God’s brand was on him.

Howell D. Chickering

Grendel came walking;

Ruth Lehmann

Grendel came gliding;

Seamus Heaney

God-cursed Grendel came greedily loping.

137

he bore God’s wrath. God’s wrath he bore,

Assuming that translators never chose to avoid a word to simply avoid translating a line as one of their predecessors did, the fact that each translator chooses a different word to render the same OE word is quite striking. The only words that are consistent in the following line are the proper nouns of Grendel and God. To describe “gongan,” the translators came up with “creeping,” “stalking,” “walking,” “gliding,” and “loping.” For “yrre,” they came up with “accursed,” “hatred,” “brand,” “wrath,” “wrath,” and “cursed.” The only repetition occurs with the word “wrath,” which both Lehmann and Chickering use. Compared to the ME translations, however, the degree of variation between newly rendered texts is substantially greater. The fact that the ME is a much friendlier source for translation is also reflected in the final quantity of lines translated. In the ModE translations of SGGK, the quantity of lines in the original exactly match the quantity translated. In the ModE translations of Beowulf, only two out of the six translators translated the same quantity of lines as the original. Three of the translations resulted in fewer lines and one resulted in more lines. Even the modern editions of SGGK are more uniform in their editorial principles. None of them show the half-line division in the middle of each line, both in the ModE translations and in the ME where applicable. The only editorial variation is in Tolkien’s translation, which chooses not to capitalize the first letter of every line. His use of the majuscule is restricted to the beginning of sentences. Four of the six Beowulf texts, on the other hand, like Tolkien’s translation of SGGK, only capitalize at the beginning of sentences. All of the texts with facing pages in OE also follow this same practice. Tolkien, the only translator to have worked extensively either in ME or OE, perhaps associates some qualities of ME verse more closely with OE than the other translators of SGGK. This essential uniformity of modern layout contrasts with the varying kinds practiced by translators of Beowulf. Of the six Beowulf translations examined, two choose to show the half-line divisions and those with facing page translations, Seamus Heaney’s and Howell D. Chickering’s, also show

138

Jennifer Anh-Thu Tran Smith

half-line divisions in the OE. Heaney, interestingly, does not maintain the visual divisions in his own ModE translation, except in two instances when a scop within the narrative itself sings the story of Sigemund and Hnaef. In these two cases, not only does Heaney show the caesurae within his own translation, but he also italicizes the text. In the case of the song of Hnaef, furthermore, he places each verse on different lines, though still maintaining a consistent marginal justification. For example, a standard line of translation from Heaney would look as follows: (6)

They sang then and played to please the hero, words and music for their warrior prince, harp tunes and tales of adventure: (ll.1062–4) By contrast, Sigemund’s song is laid out as: Sigemund’s name was known everywhere. He was utterly valiant and venturesome, a fence round his fighters and flourished therefore (ll.897–9) Hnaef’s song is laid out in yet another fashion: Hildeburh had little cause to credit the Jutes: son and brother, she lost them both on the battlefield. (ll.1070–2) In Heaney’s translation, the moments that are explicitly aural reveal the hemistichic structure of the verse, whereas the rest of the textual moments remain unmarked by either italicization or line breaks. Among the translations, there appears to be a general trend to present each hemistich separately in the more aural text. That is, Beowulf, especially when rendered in the OE, tends to show the division. Though it was perhaps performed aloud, the authors and editors working with SGGK do not register it as necessarily aural. As such, the hemistich remains a point of metrical obscurity rather than metrical prominence. Readers who would visually notice the breaks in a translation of Beowulf, may never realize that SGGK is subject to similar metrical rules.

Fidelity in versification

5.

139

Conclusion

Though translating from ME is less difficult than translating from OE, the fact that poetic and literal decisions were made by Tolkien, Borroff, Finch, and the others is still paramount to understanding what sort of translations they have created. In her introduction, Borroff highlights the fact that she has “constantly had to compromise, sometimes forced away from literal rendition by the exigencies of the meter, sometimes foregoing an attractive phrase or cadence for the sake of a more faithful rendition” (1967: xiii). This sense of balance between meter and content is not as obvious in Tolkien’s introduction, in which he states: The main object of the present translations is to preserve the metres, which are essential to the poems as wholes; and to present the language and style, nonetheless, not as they may appear at a superficial glance, archaic, queer, crabbed and rustic, but as they were for the people to whom they were addressed: if English and conservative, yet courtly, wise, and well-bred – educated, indeed learned. (1975: 14)

Such an attitude is certainly reflected in his success in reproducing the forms of the ME alliterative verse. Yet, the point of view, that one should “[keep] as close to the original as possible,” (Vantuono, 1999: xl) seems in recent years to be giving way to a more liberal approach to translation. The importance of the translator as independent poet rather than lesser translator is most explicit in the two most recent translations of Beowulf and SGGK, by Heaney and Merwin, respectively. Both of these men, poets first and foremost, highlight the personal nature of their translation experience. Merwin explains: As I tried to hear something of the poem, line by line, I came to notice, or imagine that I noticed, in the fullness and articulation of the diction, the hint of an accent that seemed familiar … What I thought I was overhearing was an intonation that I recognized … from my childhood in the mining city of Scranton, Pennsylvania. There, above my head, I had heard among my elders the sounds of the Welsh language and the intonations of the Welsh accent in English. (2002: xx)

This account of coming to understand the sound of SGGK is reminiscent of what Seamus Heaney writes in his introduction to Beowulf. For Heaney, it was a kind of familial inspiration that brought him closer to the feel of the

140

Jennifer Anh-Thu Tran Smith

original. At every stage of composition, Heaney’s family members inject themselves into the text, whether it be through the vocabulary of his Irish aunt or the “big voiced” men in his family. Heaney constantly frames the translation in terms of his heritage. He even states that he “consider[s] Beowulf to be part of [his] voice-right” (2000: xxiv). The attitude of the translators, their perspective on the text, what they see as most relevant be it the sound or the pacing, the alliteration or the epic themes, permeate the ModE versions entirely. As readers, we must ask ourselves if we want a text “that is itself good poetry and that at the same time carries a reasonable measure of the force and flavor of the original” (Raffel, 1963: xxi), or whether we are looking for a translation that is guided by a kind of familial inspiration that leads to poetic intimacy. Perhaps what these final observations about authorial intent reveal in light of the formal results of the analysis is that what makes a good translation and what makes a faithful one are two very different qualities. Their role as mediator to the original text and as purveyor of new literary material must be simultaneously examined. Together, as mediators and as new creations, they form a collection of texts, replete with inaccuracies and infidelities, that most medievalists would indeed find very familiar – a grouping of texts much more like the medieval model of manuscript production, with each translator gleaning from those who preceded, deviating in new ways, and re-conceiving the text. Perhaps it is in that collectivity that the translations have their greatest value. For, although only one Old English version of Beowulf survives and only one Middle English version of SGGK survives, these stories still have their own libraries of versions and possibilities – each to be studied individually and as a whole in decades to come.

Fidelity in versification

141

Appendix GRENDEL’S ENTRANCE

w

Alliterating Violations No Alliteration No alliteration in on-verse Alliteration not from first lift Both lifts do not alliterate in on-verse w/ anacrusis No alliteration in off-verse First lift of off-verse does not alliterate Last lift alliterates SUM

ei gh t O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 6a. Adjusted alliteration scores for passage one from Beowulf. “Grendel’s Entrance”.

2 1 0,5 0,5 1 0,5 0,5

0 0 1

0 3 4 0,5 9,5 13,5

0 0 0 0 4,5 2,5 0 2 6,5 0 7 10 0,5 17,8 24

1 0 7

16 1 6

0 0 0

4 0 8

0 0 1 1 6,5

0 1 4 4 41

0 2 0 0 2

0 1 6 5 18,5

Positional Violations Too many lifts in on-verse Too many dips in on-verse Too many lifts in off-verse Too many dips in off-verse Missing lifts in on-verse Missing dips in on-verse Missing lifts in off-verse Missing dips in off-verse Initial finite verb counted as strong SUM

w ei gh t O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 6b. Adjusted position scores for passage one from Beowulf. “Grendel’s Entrance”.

1 1 1 1 1 1 1 1 0,5

0 0 0 0 3 1 0 1 0 5

2,5 3 0 0 0 3 0 3 0 3 2 7 0 5 2 7 4,5 5 8,75 33,5

1 0 1 0 1 6 0 7 2 17

3 0 3 1 1 5 2 6 0 21

2 0 5 0 1 0 0 5 2 14

1 0 1 1 0 6 1 8 3,5 19,8

O

No Violations Lines without any violations Total Lines %

ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 6c. Total percentage of lines without violations for passage one from Beowulf. “Grendel’s Entrance”.

22 3 0 9 0 15 4 27 23 27 27 27 27 26 81% 13% 0% 33% 0% 56% 15%

142

Jennifer Anh-Thu Tran Smith

BEOWULF’S DEATH

Alliterating Violations No Alliteration No alliteration in on-verse Alliteration not from first lift Both lifts do not alliterate in on-verse w/ anacrusis No alliteration in off-verse First lift of off-verse does not alliterate last lift alliterates SUM

w ei gh t O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 7a. Adjusted alliteration scores for passage two from Beowulf. “Beowulf’s Death”.

2 1 0,5

0 0 0

1 2,5 9

6 3 8

2 0 12

0,5 1 0,5 0,5

0 0 0 0 0

1 0 1 4,5 5 1 8 5 0 12,5 12 0 24,3 32,5 11,5

16 3 5 1 2 5 5 45

0 0 1

4 3 13

0 1 0 1 0 7 0 10 0,5 27,5

Positional Violations Too many lifts in on-verse Too many dips in on-verse Too many lifts in off-verse Too many dips in off-verse Missing lifts in on-verse Missing dips in on-verse Missing lifts in off-verse Missing dips in off-verse Initial finite verb counted as strong SUM

w ei gh t O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 7b. Adjusted position scores for passage two from Beowulf. “Beowulf’s Death”.

1 1 1 1 1 1 1 1 0,5

0 0 1 0 3 2 0 3 1 9,5

1 0 3 0 0 2 1 7 6 17

6 1 2 2 1 0 0 0 0 0 0 0 2 0 2 1 2 2 0 0 3 1 1 0 1 8 4 5 2 5 9 2 2 0 2 9 2 9 2 7 3 3 3 3 9 37,5 13,5 24,5 7,5 22,5

No Violations Lines without any violations Total Lines %

O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 7c. Total percentage of lines without violations for passage two from Beowulf. “Beowulf’s Death”.

23 4 1 10 0 22 2 30 28 29 30 31 30 30 77% 14% 3% 33% 0% 73% 7%

Fidelity in versification

143

BOTH PASSAGES

Alliterating Violations No Alliteration No alliteration in on-verse Alliteration not from first lift Both lifts do not alliterate in on-verse w/ anacrusis No alliteration in off-verse First lift of off-verse does not alliterate Last lift alliterates SUM

w ei gh t O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 8a. Cumulative adjusted alliteration scores for passages one and two from Beowulf. “Grendel’s Entrance” and “Beowulf’s Death”.

2 1 0,5 0,5 1 0,5 0,5

0 0 1

1 9 6,5 3,5 18,5 21,5

3 0 19

32 4 11

0 0 1

8 3 21

0 1 0 0 9 7,5 0 10 11,5 0 19,5 22 0,5 42 56,5

1 1 1 1 18

1 3 9 9 86

0 2 0 0 2,5

1 2 13 15 46

Positional Violations Too many lifts in on-verse Too many dips in on-verse Too many lifts in off-verse Too many dips in off-verse Missing lifts in on-verse Missing dips in on-verse Missing lifts in off-verse Missing dips in off-verse Initial finite verb counted as strong SUM

w ei gh t O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm g an H ea n ne y

Table 8b. Cumulative adjusted position scores for passages one and two from Beowulf. “Grendel’s Entrance” and “Beowulf’s Death”.

1 1 1 1 1 1 1 1 0,5

0 3,5 0 0 1 3 0 0 6 0 3 4 0 1 4 9 1 10,5 14,5 25,8

9 0 3 4 6 15 14 16 8 71

2 5 4 2 0 0 0 0 1 5 5 3 2 3 0 1 2 2 1 1 10 10 2 11 2 4 0 3 9 15 7 15 5 3 5 12,5 30,5 45,5 21,5 42,3

No Violations Lines without any violations Total Lines %

O ri gi n K al en ne R dy af fe l A le xa nd C hi er ck e Le rin hm an H ea n ne y

Table 8c. Cumulative percentage of lines without violations for passages one and two from Beowulf. “Grendel’s Entrance” and “Beowulf’s Death”.

45 7 1 19 0 37 6 57 51 56 57 58 57 56 79% 14% 2% 33% 0% 65% 11%

144

Jennifer Anh-Thu Tran Smith

FIRST AGE

Alliterating Violations No Alliteration No alliteration in on-verse No alliteration in off-verse First lift of off-verse does not alliterate Alliteration on unstressed syllable SUM

w ei gh t G aw ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 9a. Adjusted alliteration scores for passage one from Sir Gawain and the Green Knight. “First Age”.

2 1 1

0 0 0

1 0 0

0 0 0

0 0 0

0 0,5 4,5

0 0 0

0,5 0,5

0 0 0

0 1 2,5

0 1 0,5

0 0 0

2 0 6

0 0 0

0 0 2

3 1 3

1 2 2 1 3,5 11,5

Positional Violations Too many lifts in on-verse Too many lifts in off-verse Missing lifts in on-verse Missing lifts in off-verse Missing strong dip in on-verse strong dip Two lifts in off-verse preceded by strong dips Final dip in off-verse is strong SUM

w ei gh t G aw ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 9b. Adjusted position scores for passage one from Sir Gawain and the Green Knight. “First Age”.

1 1 1 1 1 0,5

0 0 0 0 0 2

0 0 0 1 0 4

0 0 0 3 2 10

0 0 0 0 0 4

0 1 0 1 2 6

0 2 0 0 0 2

0 1 0 0 3 4

0 2 1 0 1 4

0,5 0,5

0 0 1

0 1 3,5

0 2 11

1 0 2,5

1 0 7,5

4 0 5

0 1 6,5

4 1 8,5

aw G

No Violations Lines without any violations Total Lines %

ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 9c. Total percentage of lines without violations for passage one from Sir Gawain and the Green Knight. “First Age”.

16 12 5 12 3 11 8 1 18 18 18 18 18 18 18 18 89% 67% 28% 67% 17% 61% 44% 6%

Fidelity in versification

145

GAWAIN’S SEDUCTION

Alliterating Violations No Alliteration No alliteration in on-verse No alliteration in off-verse First lift of off-verse does not alliterate Alliteration on unstressed syllable SUM

w ei gh t G aw ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 10a. Adjusted alliteration scores for passage two from Sir Gawain and the Green Knight. “Gawain’s Seduction”.

2 1 1

0 0 1

1 0 3

1 0 2

0 0 1

3 0,5 8,5

0 0 2

0,5 0,5

1 1 2

2 2 7

1 0 4,5

0 0 1

1 2 0 5 15,5 5,5

1 0 1

7 1,5 5,5

2 1 3 0 5,5 21,5

w

Positional Violations Too many lifts in on-verse Too many lifts in off-verse Missing lifts in on-verse Missing lifts in off-verse Missing strong dip in on-verse strong dip Two lifts in off-verse preceded by strong dips Final dip in off-verse is strong SUM

ei gh t G aw ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 10b. Adjusted position scores for passage two from Sir Gawain and the Green Knight. “Gawain’s Seduction”.

1 1 1 1 1 0,5 0,5 0,5

0 0 1 1 0 4

0 4 1 3 1 10

0 0 0 6 0 9

0 4 0 5 3 2 6,5 17,5 11,5

0 0 0 3 1 4

0 1 0 2 2 10

1 6 1 3 0 8

6 0 9

3 3 13

6 4 20

1 2 1 2 0 4

0 0 1 4 3 1

2 4 5 0 11,5 10,5

aw G

No Violations Lines without any violations Total Lines %

ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 10c.Total percentage of lines without violations for passage two from Sir Gawain and the Green Knight. “Gawain’s Seduction”.

15 5 11 14 5 5 9 5 25 25 25 25 25 25 25 25 60% 20% 44% 56% 20% 20% 36% 20%

146

Jennifer Anh-Thu Tran Smith

BOTH PASSAGES

Alliterating Violations No Alliteration No alliteration in on-verse No alliteration in off-verse First lift of off-verse does not alliterate Alliteration on unstressed syllable SUM

w ei gh t G aw ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 11a. Cumulative adjusted alliteration scores for passages one and two from Sir Gawain and the Green Knight. “First Age” and “Gawain’s Seduction”.

2 1 1

0 0 1

2 0 3

1 0 2

0 0 1

0,5 0,5

1 1 2

2 3 9,5

1 1 5

0 0 1

3 1 13

0 0 2

3 2 0 5 21,5 5,5

1 0 3

10 2,5 8,5

3 5 9

3 1 33

w

Positional Violations Too many lifts in on-verse Too many lifts in off-verse Missing lifts in on-verse Missing lifts in off-verse Missing strong dip in on-verse One lift in off-verse not preceded by strong dip Two lifts in off-verse preceded by strong dips Final dip in off-verse is strong SUM

ei gh G aw ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 11b. Cumulative adjusted position scores for passages one and two from Sir Gawain and the Green Knight. “First Age” and “Gawain’s Seduction”.

1 1 1 1 1

0 0 1 1 0

0 4 1 4 1

0 0 0 9 2

0 0 0 3 1

0 2 0 3 4

1 8 1 3 0

1 3 1 2 3

0 2 2 4 4

0,5

6

14

19

8

16

10

8

5

0,5 0,5

0 5 7,5

4 4 21

0 7 4 10 4 0 3 4 22,5 11,5 20,5 25

2 6 18

8 1 19

aw G

No Violations Lines without any violations Total Lines %

ai St n P oe on t e Bo rr of To f lk ie H n ar ri s Fi on nc h V an tu M ono er w in

Table 11c. Cumulative percentage of lines without violations for passages one and two from Sir Gawain and the Green Knight. “First Age” and “Gawain’s Seduction”.

31 17 16 26 8 16 17 6 43 43 43 43 43 43 43 43 72% 40% 37% 60% 19% 37% 40% 14%

Fidelity in versification

147

BEO Comparative View 100 90

86 ,0

Weighted Sum Total Violation

80 71 70 60

56,5

50

46,0

45,5

42,25

42 ,0 40 30,5 30

25,75 21,5 18 ,0

20

14,5

10 2,5

0,5 0 Original

Kennedy

Raffel

Alexander

Alliterating Violations

Chickering

Lehmann

Heaney

Positional Violations

Figure 1. Comparative view of adjusted scores for alliterating and positional violations for passages one and two from Beowulf. “Grendel’s Entrance” and “Beowulf’s Death”.

SGGK Comparative View 90

80

Weighted Sum Total Violation

70

60

50

40 33 30

25 22,5

21

21 , 5 2 0,5 18

20

7, 5

10

19

1 1 ,5

9 ,5

9 5,5

5 2

1

0 Gawain Poet

Stone

Borroff

Tolkien Alliterating Violations

Harrison

F inc h

Vantuono

Merwin

Positional Violations

Figure 2. Comparative view of adjusted scores for alliterating and positional violations for passages one and two from Sir Gawain and the Green Knight. “First Age” and “Gawain’s Seduction”.

148

Jennifer Anh-Thu Tran Smith

BEO Cumulative Violations 140

120

Weighted Sum Total Violations

45,5 100 71

80

42,25 60 25,75

86 40 30,5 56,5 46

42

20

21,5 18

14,5 0

2,5

0 ,5

Original

Kennedy

Raffel

Alexander

Alliterating Violations

Chickering

Lehmann

Heaney

Positional Violations

Figure 3. Total adjusted violation scores for passages one and two from Beowulf. “Grendel’s Entrance” and “Beowulf’s Death”.

SGGK Cumulative Violations 140

Weighted Sum Total Violations

1 20

100

80

60

19

40 20,5 20

21 22,5

25

18

5, 5

9

Finch

Vantuono

33

21,5 0

7,5 2

9,5

Gawain Poet

Stone

11 ,5 5

1

Borroff

Tolkien Alliterating Violations

Harrison

M er w in

Positional Violations

Figure 4. Total adjusted violation scores for passages one and two from Sir Gawain and the Green Knight. “First Age” and “Gawain’s Seduction”.

Fidelity in versification

149

BEO % of Lines with No Violations 100%

90%

% of Lines w/o Violations

80%

79%

70%

6 5%

60%

50%

40% 33% 30%

20% 14% 11% 10% 2%

0%

0% Original

Kennedy

Raffel

Alexander

Chickering

Lehmann

Heaney

Figure 5. Total percentage of lines without violations for passages one and two from Beowulf. “Grendel’s Entrance” and “Beowulf’s Death”.

SGGK % of Lines with No Violations 100%

90%

80%

% of Lines w/o Violations

7 2% 7 0% 60% 60%

50% 40%

40%

37%

37%

40%

30% 19%

20%

14% 10%

0% Gawain Poet

Stone

Borroff

Tolkien

Harrison

F in c h

Vantuon o

M e r w in

Figure 6. Total percentage of lines without violations for passages one and two from Sir Gawain and the Green Knight. “First Age” and “Gawain’s Seduction”.

150

Jennifer Anh-Thu Tran Smith

Notes 1. Although my original goal was to look at six translations as I did for the Old English study, W.S. Merwin’s facing page translation came to my attention after I had already finished scanning the other six texts. Because Merwin is a popular poet in his own right and also because he has translated over twenty other works, however, I did not want to exclude him from my final analysis. The study of ME alliterative verse, therefore, is based on seven translations rather than six. 2. Within this abstract examination of fidelity, therefore, is an underlying consciousness of the structural capacity that semantic units play in the alliterative verse form. 3. A strong dip is a dip with two or more unstressed syllables. It is also often termed a long dip. 4. Taken out of context, as it is here, Raffel’s “hills and bogs” may seem out of place in comparison to the other translators’ lines. It is, however, a feature of all good translations. Rather than translate each line of OE to ModE in a one to one correspondence, Raffel translated several lines at a time, allowing flexibility in placement. Given different passages, all of the translations would exhibit this same movement. A more accurate representation of the entire grammatical unit would have to encompass four lines of verse. Ðá cóm of móre under mist-hleoþum Grendel gongan, Godes yrre bær, mynte se mán-scaða manna cynnes sumne besyrwan in sele þám héan. Out from the marsh, from the foot of misty Hills and bogs, bearing God’s hatred, Grendel came, hoping to kill Anyone he could trap on this trip to high Herot. Because the point of my comparison, however, is to discuss the translators’ decisions on how to render “gongan” and “yrre,” the line as quoted should be perfectly sufficient.

Fidelity in versification

151

References Alexander, Michael (trans.) 1973 Beowulf. Baltimore: Penguin Books. BBC News 2006 Heaney wins second Whitbread prize. http://news.bbc.co.uk/1/hi/uk/618887.stm Bjork, Robert E. and John D. Niles, eds. 1997 A Beowulf Handbook. Lincoln: University of Nebraska Press. Borroff, Marie (trans.) 1967 Sir Gawain and the Green Knight. New York: W. W. Norton and Co. Cable, Thomas 1991 The English Alliterative Tradition. Philadelphia: University of Philadelphia Press. Chickering, Howell D. 1977 Beowulf. Garden City: Anchor Press/Doubleday. Chickering, Howell D. 2001 From ‘Lo’ to ‘So’: Modern poetic paraphrases of Beowulf. 14 May 1999. Supplement to Yazyk i rechevaya deyatel’nost’, Nicolay Yakovlev (ed.), Vol. 4. Duggan, Hoyt 2001 Some aspects of a-verse rhythms in Middle English alliterative poetry. In Speaking Images: Essays in Honor of V. A. Kolve. Robert F. Yeager, and Charlotte Morse (eds.), 479–503. Ashville, North Carolina: Pegasus Press. Finch, Casey (trans.) 1993 The Complete Works of the Pearl Poet. Malcolm Andrew, Ronald Waldron, and Clifford Peterson (eds.). Berkeley: University of California Press. Harrison, Keith (trans.) 1983 Sir Gawain and the Green Knight. London: The Folio Society. Heaney, Seamus (trans.) 2000 Beowulf. New York: W. W. Norton & Company. Hutcheson, B.R. 1995 Old English Poetic Metre. Cambridge: D. S. Brewer. Kennedy, Charles W. (trans.) 1940 Beowulf. New York: Oxford University Press. Lehmann, Ruth P. (trans.) 1988 Beowulf. Austin: University of Texas Press. Merwin, W.S. (trans.) 2002 Sir Gawain and the Green Knight. New York: Alfred A. Knopf.

152

Jennifer Anh-Thu Tran Smith

Raffel, Burton (trans.) 1963 Beowulf. New York: The New American Library. Stone, Brian (trans.) 1974 Sir Gawain and the Green Knight. Middlesex: Penguin Books, Ltd. Stockwell, Robert and Donka Minkova 1977 Prosody. In A Beowulf Handbook. Robert Bjork and John Niles (eds.), 55–85. Lincoln: University of Nebraska Press. Tolkien, J.R.R. 1991 Beowulf: The monsters and the critics. In Interpretations of Beowulf R. D. Fulk (ed.), 14–44. Bloomington: Indiana University Press. Tolkien, J.R.R. (trans.) 1975 Sir Gawain and the Green Knight, Pearl, Sir Orfeo. Boston: Houghton Mifflin Company. Vantuono, William (trans.) 1999 Sir Gawain and the Green Knight. Notre Dame: University of Notre Dame Press.

Response to Tom Cable’s comments

Jennifer Anh-Thu Tran Smith I very much appreciate Tom Cable’s thoughtful and critical analysis of my essay. Several points that he highlights do indeed need further consideration, namely how to take into account Kaluza’s law when analyzing OE verse and how better to capture the requirement of the asymmetry between the a-verse and the b-verse in ME verse. There are, however, a couple of points that I would like to address. I would like to challenge Cable’s suggestion that alliteration is something that the casual reader does not register. The absence of alliteration should stand out as markedly as the absence of rhyme for a casual reader by virtue of its repetition, or rather, non-repetition. While most modern verse is not based on alliteration, it is still a potent prosodic feature in common rhymes like “Peter Piper picked a peck of pickled peppers” and “Sally sells seashells down by the seashore.” That is, alliteration is a familiar enough feature in modern language that its absence should at least be registered. I think it likely that the general preference for Chickering’s translation is not in its fidelity to verse but rather in “the accuracy of the translation and the appropriateness of the ModE diction” that Cable highlights. While it is sometimes easy to forget when embarking on a study of verse fidelity, the weight of a story’s effectiveness is in the meaning not in the rhythm. I would also like to address the difference in my scansion of line 736a and that of Klaeber, Bliss, Pope, Fulk, and Suzuki (a formidable group indeed). My own deviation speaks to my methodology, the general subject of this volume. In my section on methodology, I explain the order in which and the way in which I scanned each line following those rules set out by Cable, rules based on lexical category. This explanatory section speaks to my concern, one that I had from the very start of this project, about the role that personal subjectivity could play in the scansion of verse. So while the traditional scansion of line 736a which places stress on “ða” or “that” makes sense in delivery, it would represent a deviation from the lexical categories that typically receive stress. It would be a subjective choice given for emphasis and delivery. I decided that my rigidity in adhering to

154

Jennifer Anh-Thu Tran Smith

those rules would remove that element of subjective judgment making for a more consistent and accurate final assessment, surely a goal that we all strive to attain. I do recognize, however, that my choice to remain faithful to these rules of scansion is itself a choice. I only hope that how I came to my conclusions and why I did so remains clear to the reader.

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?*

Xingzhong Li 1.

Introduction

This study reconsiders a long-standing controversy on the authorship of the Middle English poem The Romaunt of the Rose (hereafter The Romaunt) translated from the original French poem entitled the Roman de la Rose (hereafter Le Roman). A century ago, Chaucerian scholars used evidence of diction, rhyme, grammar of Chaucerian English, cohesion of sub-stories in the poem, and other information, in determining The Romaunt authorship. The dominant view, proposed by Rev. Walter W. Skeat in 1900 and subsequently accepted in the literature, has been that Chaucer translated only the first 1,705 of the 7,692 lines in the poem. Adopting a comparative approach and exploiting new and independent metrical evidence, this study reexamines the authorship as well as Skeat’s claims. The findings strongly support the accepted view that Fragment A of the poem is by Chaucer while Fragment B is not, but they disagree with Skeat’s centennial hypothesis on the authenticity of Fragment C.

2.

The background

The Romaunt is a fragmentary translation of Le Roman. Co-authored by Guillaume de Lorris and Jean de Meun and edited by Félix Lecoy (1970), Le Roman is 21,750 lines long. Lorris composed the first 4,028 lines from the year 1225 through 1230, and then Meun took it over, amplified it, and finished the poem between the year 1269 and 1279 (Lecoy 1970: VIII). However, only portions amounting to about one-third of the original lines were translated into Middle English. As Skeat informs us (1900: 65, 76–78; 1965: 65, 76–78), Lindner (1888), shortly after, Kaluza (1891), and then Skeat himself, divided The Romaunt into three fragments, conveniently

156

Xingzhong Li

called A, B, C. Fragment A, a highly literary translation running from line 1 though line 1,705, corresponds to lines 1 through 1,678 in Le Roman. Fragment B, including lines from 1,706 to 5,810, covers original lines 1,679 through 5,169 in Le Roman, that is, all of Lorris’ work and a bit more than the first thousand lines of de Meun’s continuation. Fragment C, running from lines 5,811 up to 7,692, corresponds to lines 10,716 to 12,564 in Le Roman, leaving some 5,500 lines untranslated. So the three fragments consist of 7,692 lines, which are often included in Chaucer anthologies, such as in The Riverside Chaucer (Benson 1987), which this study uses as the test text. Concerning the genuineness of the three fragments, Benson states that most Chaucerian scholars have accepted Fragment A as an early work of Chaucer’s and rejected Fragment B as his work. With respect to Fragment C, most scholars are reluctant to give Chaucer credit (1987: 686). A further study shows a clearer historical picture of credit and discredit. According to Skeat, “Writing in 1892, Prof. [sic.] Lounsbury was disposed to claim for Chaucer the whole of the existing English translation of the Romaunt of the Rose; but later investigations have shown that this view is no longer tenable” (1900 and 1965: vi). Then Skeat rejects Max Kaluza’s view that Fragment C is Chaucer’s genuine work: “It is much to be regretted that Dr. Kaluza, to whom we owe so much for his separation of Fragment A from the rest, has taken up the untenable position of claiming Fragment C for Chaucer, as well as Fragment A.” Immediately following this rejection, Skeat also refutes John Koch’s view that Chaucer translated neither of Fragments A and C: “Koch takes advantages of this [i.e., Kaluza’s view] at once, arguing that if Fragments A and C are by the same author, it is obvious that Chaucer wrote neither of them, because it is impossible that Fragment C can be genuine” (149). Skeat’s own view is that only Fragment A is Chaucer’s genuine work. With abundance of evidence he concludes, “The more closely we examine Fragment A, the more obvious its genuineness becomes” (72). As for Fragment B, he convincingly justifies his claim that this fragment was written “by a Northerner, who imitated Chaucer’s diction and grammar rather carefully, but reverted to what was habitual to him whenever he forgot what Chaucer’s system of rimes demanded” (79). Concerning Fragment C, Skeat also rejects it, but with less faith and weaker evidence than that for crediting Fragment A and discrediting Fragment B: “C is far nearer than B to Chaucer, and has a better chance of being genuine. Nevertheless, I am reluctantly compelled to say that I fear it is not his” (90). Skeat provides several arguments, of which the self-integration of Fragment C as a complete story on its own is

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

157

the primary one: “In a word, the translator of Fragment C selected a fair specimen of Jean de Meun at his best, and has given us his account of Hypocrisy with some spirit, in as complete a form as was practicable. And this is the best reason for supposing that this translator had nothing to do with Chaucer or his Northern continuator” (93). However, this argument could be made conversely. Fragment C’s very separation from Fragment B by more than 5,000 lines of Le Roman and its self-completion of the story could be the very reason why Chaucer chose to translate it. To me, both sides of this argument sound speculative. In examining the translatorship of The Romaunt, past Chaucerian scholars, Skeat included, typically and scrupulously examined the dialects, rhyme words, and syntax (case endings in particular) in the fragments and accepted only those portions which reflect similar features in Chaucer’s genuine poems. The comparative methodology itself is flawless and, actually, is the one that this paper is going to adopt, too. However, few Chaucerian scholars have used metrical features as evidence to test the translatorship, hence this preliminary study. The next section tells more about this research method.

3.

The methodology

First of all, this study concurs with earlier Chaucerians’ division of The Romaunt into three fragments. Then it assumes that metrical features are much more pervasive than, say, rhyme words and, therefore, constitute solid and valid evidence. It further assumes that if metrical patterns in a given fragment potentially accord with or deviate from those in Chaucer’s own poems, there is reason to accept or reject the translatorship. Finally, this paper adopts a statistical approach by setting up a reliable corpus of data for metrical analysis. 3.1. The corpus of data Given the fact that Chaucer almost exclusively used iambic tetrameter in his genuine early poems and the fact that all three fragments in The Romaunt were written in the same meter, these two groups of verse are metrically comparable. For the purpose of this paper, I adopted the “systematic sampling” method used in Li (2004) to set up a corpus of data for analysis

158

Xingzhong Li

because “like random samples, systematic samples are also representative samples. … [T]he findings based on the [systematic] samples are not biased and … are generalizable … to the parent population” (Hopkins, Glass, and Hopkins 1987: 117). According to these authors, “Selecting every hundredth word in a dictionary would yield a 1 percent systematic sample of words” (117). Because “the size of a sample can never compensate for bias or a lack of representativeness,” the sample size is less important than the sample itself, and “a representative sample of 100 is generally preferable to an unrepresentative sample of 1,000,000!” (113). So I systematically selected every tenth line in each of Chaucer’s iambic tetrameter and pentameter poems, the ratio of the sample size n to the size of the whole population N being 10%. Subtracting editor-emended lines, this sampling method yielded 4,146 lines out of Chaucer’s over 41,000 lines of verse, of which 1,126 are in iambic tetrameter and 3,020 in iambic pentameter. It is these lines that this paper uses for prosodic scrutiny and comparison. Among all the sampled iambic verses are 763 lines from The Romaunt, completed before 1369, 131 lines from The Book of the Duchess (henceforth BD) completed in 1369, 215 lines from The House of Fame (henceforth HF) completed between 1378–80, 3 lines from the short poem “Anelida and Arcite” (Strophe 5 and Antistrophe 5, henceforth “Anel”), and 14 lines from “Sir Thopas” in The Canterbury Tales (henceforth “Thop”). Although dating Chaucer’s complete poems is very difficult, it is generally accepted that Chaucer translated before he composed, and that he composed tetrameter verse before pentameter verse. So the sampled iambic tetrameter lines for this study represent Chaucer’s early poetic achievements. The results of the metrical analysis of these lines are, when necessary, compared to those of the 3,020 sampled iambic pentameter lines from Chaucer’s Troilus and Criseyde and The Canterbury Tales. 3.2. Metrical features and their measures Metrically, how a line of verse conforms to (or deviates from) a canonical metrical pattern determines that line’s prosodic features or complexity. Youmans (1989: 341) claims that the task of metrical theorists is to establish a metrical prototype and to determine degrees of deviation from this prototype rather than to define a precise boundary between metrical and unmetrical lines. Following Youmans’ claim and adopting the notation of

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

159

Liberman and Prince (1977), Li (2004: 316) established a central metrical prototype for English iambic tetrameter verse, drawing evidence from Chaucer’s octosyllabic lines in particular. I reproduce this metrical prototype here as (1) below, for subsequently it is used to gauge metrical features. (1)

Prototypical iambic tetrameter line: Line: Hemistichs:

W

Metrical feet:

S

W

S

W

S

Metrical positions: W

S

W

S

W

S

W

S

Syllable [±stress]:

+s 2

–s 3

+s 4

–s 5

+s 6

–s 7

+s 8

–s 1

This prototype is a frequency-driven, statistics-based, hierarchically structured, saliency-ranked model for the centuries-old English iambic tetrameter verse in general. It claims that iambic tetrameter is both linear and hierarchical in structure. It also claims that Chaucer’s iambic tetrameter lines are prototypically footed, which argues against the claim that Chaucerian metrics is unfooted (Halle and Keyser 1966, 1971; Cable 2002) though preChaucerian (Minkova 1996) and post-Chaucerian iambic lines of verse (Tarlinskaja 1976) in early ME, ME, and ModE are all footed. In particular, Minkova (1996: 101) argues for a basic iambic pattern in the preChaucerian poem Ormulum, which contains invariably seven feet divided 4:3 across the line. The same idea is also exploited in Minkova (2000). Similarly, Tarlinskaja (1976: 259–260, 279–280) provides stress profiles of iambic tetrameter and pentameter verse composed by Chaucer and many post-Chaucerian poets, and the profiles in general exhibit foot-based meters. Of course, this does not mean that proponents of metrical footing, in Chaucer and elsewhere, have not been interested in the different properties of metrical positions depending on their linear structure. In fact, they have exploited various types of mismatch between linguistic stress and metrical positions to discover patterns of deviations away from underlying metrical templates (Tarlinskaja 1974, 1976, 2006; Kiparsky 1975, 1977; Youmans 1989, 1996; Minkova 1996; Youmans and Li 2002; Li 2004). It is those accepted patterns of deviations that this paper employs to test the transla-

160

Xingzhong Li

torship of the Romaunt, and it is those metrical principles represented in this prototype that this paper relies on for measuring metrical styles in the sampled lines of verse. Specifically, this prototype suggests that the eight metrical positions differ from one another in their metrical saliency. After discussing the saliency dominance relationships in this prototype, Li (2004: 317–319) concludes that the prototypical saliency in the eight metrical positions is ranked as in (2): (2)

The ranked prototypical saliency in the eight metrical positions: 15372648 Least salient

Most salient

It is widely accepted that the more salient a metrical constituent is, the more constrained it becomes, the lower the frequency of metrical deviations that it incurs, and the more disruptive the deviations will be. This also means that the stress patterning of [WS.WS.WS.WS] would tend to be, in general, less regular toward the beginning of a line and more regular toward the end of a line. Conversely, the deviation of the stress patterning would be more common toward the beginning, but less frequent toward the end. Aside from the ranked saliency of the eight metrical positions in iambic tetrameter, Li (2004: 319–320) also claims that some metrical feet in the prototype are more salient than others, and their saliency is ranked as in (3): (3)

The ranked prototypical saliency in the metrical feet:

Least salient

F1  F3  F2  F4

Most salient

That is, the fourth foot is the most salient, and the second foot the next most salient, followed by the third foot; the first foot is the least salient. Predictably then, metrical deviations would likely occur the most frequently in the first foot, the next most frequently in the third foot; deviations would be the least frequent in the fourth foot and the next least frequent in the second foot.

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

161

Additionally, Li (2004: 320) claims that this prototype defines a ranked saliency principle with regard to hemistichs (i.e. half lines), reproduced here as (4): (4)

The ranked prototypical saliency in the hemistichs: Less salient

H1  H2

More salient

Predictably, metrical deviations in the first hemistich would be more frequent, but would be less rhythmically disruptive if at all, than those in the second hemistich.

4.

Patterns of metrical deviations

Once the metrical prototype is established and the metrical principles presented, the next step is to define the patterning of deviations from this prototype. “Individual poets vary widely in this respect – metrical deviations that are innocuous for one poet will be costly for another” (Youmans 1989: 349). Thus, if deviations in the translated fragments under discussion differ from those in Chaucer’s own poems, we could use the differences as tests. Specifically, this study examines deviations or metrical parameters including trochaic substitution, trochaic makeup, extrametricality, headlessness, and metrical resolution to attest the translatorship of The Romaunt. 4.1. Trochaic substitutions As a result of the Norman Conquest, a huge number of French words entered the English vocabulary. Chaucer’s verse thus inevitably contains many Romance words, which follow different stress patterns. His verse also contains stress doublets in words like “also”, “honour”, and “therefore” as well as words suffixed with “-nesse” and “-yng(e),” which attract word stress (Halle and Keyser 1966: 194–196) although it is more likely that such suffixes may be promoted to secondary stress, for rhyming purposes in particular. Also contained in his verse are disyllabic proper names like “Thopas” and “Adams,” which may be stressed both ways, according to Chaucerian scholars. I excluded from my statistics such cases that would otherwise be treated as trochaic substitutions. For example,

162

Xingzhong Li

(5)

W S Therfore y kan hem now beleve.

HF 990

WS And of the goddess of slepyng.

BD 230

In assigning word stress, I have followed Hayes’ definition of a stress peak, which, simply put, is any syllable with a stress greater than at least one of its neighbors (1989: 227). This definition takes no heed of the opposition between major and minor category words that is maintained in the HalleKeyser theory (1971), which in general treats stressed syllables in polysyllabic minor category words as W, and only stressed syllables in major category words as S. By Hayes’ definition, any mapping of a stress peak and its following syllable with a lesser stress in the sequence of odd-even metrical positions is counted as a trochaic substitution. Therefore, “In a book that …” (RomB 5650) is considered as two successive trochees, the first being weak and the second strong, though a weak trochee is often traditionally analyzed as a pyrrhic. The word “aftir” in (6), because it frequently occupies SW positions elsewhere in this dataset, is also treated as a trochee.1 (6)

SW Aftir to thenken on hir sorwe.

BD 100

In thinking about trochees, it is also necessary to distinguish between stress within a lexical word and stress within the utterance as a whole. The former is rigidly constrained by the phonological system of the language whereas the latter is primarily constrained by performance considerations and can be very variable, more so with sequences of monosyllables. Thus, nearly every trochee consisting of monosyllabic words in (7) below, such as “ryght as”, “god of”, may alternatively be considered as non-trochees or even iambs. Such trochees, which lack the rigid constraint on stress in polysyllabic words in isolation, however, tend to be constrained on the phrase level, where monosyllabic function words are often cliticized. Insofar as poets tend to map stressed monosyllables of major word categories into even metrical positions, I have opted for analyzing trochees of monosyllables as indicated in (7). If a given fragment in The Romaunt ignores the trend, that would help reveal its distinct metrical style.

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

163

In this set of data, I identified a total of 182 trochaic substitutions, as sampled in (7) (where RomA, RomB, RomC, as used in Benson 1993, correspond to Fragments A, B, C in The Romaunt): (7)

S W Ryght as this drem wol tel us alle.

RomA 30

S W A cherl chaungeth that curteis is.

RomB 4030

S W If that I were god of richesse,

RomC 6010

S W Anoon therwith whan y sawgh this — BD 500 SWSW Feble as a forwoundid man,

RomA 1830

S W S W With what lord thou haddist to do

RomB 4650

S W S W Nor of a thousand full scarsly,

RomB 5460

The distribution of trochees seems to be constrained, because in this set of data no successive trochees occur in F3 and F4, nor do any lines have more than two trochees. Additionally, trochees rarely occur at the end of a line. Understandably so, because the absence of such cases is just what the prototype predicts: the second half of a line is constrained more strictly than the first half. Table 1 below shows more clearly how the 182 trochaic substitutions are constrained by, or deviate from, the metrical prototype in (1). Taken in the aggregate, the first foot allows the most trochaic substitutions (6.7%), followed by the third foot (4.8%), then by the second foot (4.0%). The fourth foot nearly prohibits trochaic substitutions (0.6%). To state this another way: of the total of 182 trochaic substitutions identified, about 41.8% (76 cases) are found in the first foot, 24.7% in the second foot, 29.7% in the third foot, and only 3.8% in the fourth foot. Clearly, then,

164

Xingzhong Li

Chaucer’s meter is freer at the beginning of each hemistich than at the end – just as the ranked prototypical foot saliency in (3), reproduced here as F1  F3  F2  F4, would lead us to expect. Table 1. The distribution of trochaic substitutions in the sampled tetrameter lines

Poems (# of lines) RomA (169) RomB (408) RomC (186) BD (131) HF (215) Thop (14) Anel (3) Total lines (1,126) Total trochees (182)

F1 (%) 9 (5.3%) 33 (8.1%) 11 (5.9%) 11 (8.4%) 12 (5.6%) – – 76 (6.7%) 76 (41.8%)

F2 (%) 6 (3.6%) 19 (4.7%) 7 (3.8%) 3 (2.3%) 10 (4.7%) – – 45 (4.0%) 45 (24.7%)

F3 (%) 8 (4.7%) 15 (3.7%) 9 (4.8%) 9 (6.9%) 12 (5.6%) – 1 (33.3%) 54 (4.8%) 54 (29.7%)

F4 (%) 1 (0.6%) 3 (0.7%) 1 (0.5%) 1 (0.8%) 1 (0.5%) – – 7 (0.6%) 7 (3.8%)

But Table 1 tells more: RomB stands out in that its incidence of trochees from the initial to the last foot shows a sagging tendency, from 8.1% to 4.7% to 3.7% to 0.7%, with a ranking of F1 << F2 << F3 << F4, which goes against the foot saliency principle. This contrasts sharply with the fallrise-fall trend of the incidence in all other poems. Using the letter H for a high incidence and L for a low incidence, we can render this finding more tangible, as demonstrated in Table 2 below, where “//” stands for hemistich boundary: Table 2. The contrasts of the high-low incidences of the trochaic substitutions

Poems RomA RomB RomC BD HF Thop Anel

F1 H H H H H – –

F2 L L L L L – –

//

F3 H L H H H – –

F4 L L L L L – –

The HLHL pattern is prototypically canonical. However, RomB ignores it, for it exhibits a HLLL pattern. Thus, the statistics in Tables 1 and 2 for

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

165

RomA, RomC, BD, and HF support the hypothesis of a dipodic structure emphasized in the prototype, whereas RomB does not. These initial metrical facts corroborate a major part of Skeat’s hypothesis that RomA is Chaucer’s work and RomB is not. They, however, do not support his hypothesis that RomC is not part of the Chaucer canon though more data are required to reach a firm conclusion. 4.2. Trochaic makeup Could the makeup of trochees shed some light and lead to a different result? A meticulous study of trochaic structures helped me identify five types of them in the data set: (1) trochees consisting of two monosyllabic words; (2) trochees of two comma-separated monosyllabic words; (3) trochees of the second syllable in an end-stressed disyllable followed by a monosyllabic function word; (4) trochees of an initially-stressed disyllable; and (5) trochees of the first two syllables in a trisyllable with a SWS stress pattern. I provide illustrative examples in (8): (8)

S W Type 1: Of which five in his right hond were.

RomA 940

S W Type 2: Loo, to the House of Fame yonder,

HF 1070

S W Type 3: And of her myshappe hem diffame;

RomB 5500

S W Type 4: To have double absolucioun.

RomC 6410

SW Type 5: Debonaire, goode, glade, and sadde,

BD 860

Acknowledging metrical significance of syntactic and phonological breaks such as in Type 2 and for the purpose of this study, I just simplify the discussion by collapsing Types 1 and 2 as the “monosyllabic word” (mw) category and grouping the remaining three types as the “disyllabic word” (dw) category and display their distributions in Table 3.

166

Xingzhong Li

Table 3. Distributions of the trochees of monosyllabic words (mw) and disyllabic words (dw)

Poems (# of lines) RomA (169) RomB (408) Rom C (186) BD (131) HF (215) Thop (14) Anel (3) Total lines (1,126) Total trochees (182)

mw (%) 19 (11.2%) 43 (10.5%) 21 (11.3%) 18 (13.7%) 30 (14.0%) – 1 (33.3%) 132 (11.7%) 132 (72.5%)

dw (%) 5 (3.0%) 27 (6.6%) 7 (3.8%) 6 (4.6%) 5 (2.3%) – – 50 (4.4%) 50 (27.5%)

ratio (mw : dw) 3.8 1.6 3.0 3.0 6.0 – – 2.6 2.6

Again, RomB exhibits peculiar features: its “mw” percentage of 10.5% sags below the average 11.7% in contrast with all other poems whose values are either above or stay closer to the average, whereas its “dw” percentage of 6.6% stands well above the average of 4.4% in contrast with all other poems whose values, 3.0, 3.8, 4.6, 2.3 respectively, are either lower or just slightly higher than the average. The ratios are even more telling. The lower a ratio is, the higher percentage of dw there is in a particular poem. RomB’s ratio, 1.6, is significantly lower than the average of 2.6. By contrast, the ratios in other poems, 3.8, 3.0, 3.0, 6.0 respectively, all stand higher or significantly higher than the average. These statistics are significant to metrical theory because they demonstrate that trochees of disyllables are of sharper accentual contrast than trochees of monosyllables, are thus more destructive to iambic meters and, therefore, are more constrained and occur less frequently. They are also significant because they classify RomA and RomC into one group of highly similar metrical styles (i.e., for mw, 11.2% in RomA : 11.3% in RomC; for dw, 3.0% in RomA : 3.8% in RomC) and convincingly show that RomB does not conform to what may be called the Sensitivity Principle of Trochees as closely as all the other poems do: (9)

The Sensitivity Principle of Trochees: Trochees of disyllables are metrically more sensitive and disruptive than those of monosyllables in iambic meters.

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

167

4.3. Headless lines It is metrically permissible to delete a weakly stressed syllable at the beginning of a line or, less frequently, at the beginning of a hemistich to create a headless line or headless hemistich. There is no doubt that Chaucer was conscious of syllable count or deletion in his composition because he acknowledged that “som vers fayle in a sillable,” and that he did “no diligence / To shewe craft, but o sentence” (HF 1098–1100). In the sampled lines, I identified 92 lines (8.2%) that lack a metrically desirable syllable. Below are some examples: (10)

1 Techith for to robbe and stele 1 2

3

4 5 6

7

8

1 Glad and mery for to be, 1

2

3

3

2

3

4 5

2

3

4 5

BD 70

6 7 8

1 We ben shrewes, every wyght 1

RomC 5870

4 5 6 7 8

1 Such a tempest gan to rise 1

RomB 2290

45 6 7 8

1 So that Venus be present 1 2

RomA 190

HF 1830

6 7 8

A purs that heng 1 by a hand 1 2 3 4 5 6 7 8

RomA 240

Salue thee first, 1 be not domm, 12 3 4 5 6 7 8

RomB 2220

The first four examples are straightforward headless lines, and so is HF 1830, where the word every is used as a disyllable. The last two lines each are analyzed as containing a headless second hemistich with a pause on the fifth position. Of course, they might be alternatively analyzed as headless lines by placing a missing syllable at the beginning of each line, but that would require a reader to place stress on weak syllables a and thee, demote the lexical word purs, and reverse the stress pattern on salue, a less plausi-

168

Xingzhong Li

ble treatment. In general, missing syllables are clear-cut, and ambiguous positions are rare, as given in (11): (11) a.

1 I was ryght yong, soth to say, 12 3

4

5

BD 1090

6 7 8

b. I was ryght yong, 1 soth to say, 1 2 3 4 5 6 7 8 c. The righte weye ben 1 goon. (?) 1 2 3 4 5 6 7 8

RomB 4000

d. The righte weye 1 ben goon. (?) 1 2 3 4 56 7 8 e. Anoon-ryght 1 I wente ner; 1 2 3 45 6 7 8

(?)

f. Anoon-1-ryght I wente ner; 1 2 3 4 5 6 7 8

(?)

BD 450

I counted (11a) in my statistics because it conforms more closely to the Chaucerian norm as shown in Table (4) below.2 I also counted (11d) into my computation because iambic phrases like “ben goon” are common at the end of a line. For the third case, I took (11f) because it is prosodically preferred though (11e) is preferred morphologically. This analysis can be supported by the fact that a short pause can be optionally inserted between two adjacent stresses in a phrase, especially when this is a rhythmic cue, for example, “a little dog” vs. “a young ø dog.” The distributions of the 92 lines with missing syllables are represented in Table 4 below, where MP stands for metrical position. Given the metrical position saliency principle in (2), even-numbered positions are in general more salient than odd-numbered ones and, therefore, less likely to harbor missing syllables, especially toward the end of a line. All the poems in the data abide by this principle except for RomB, whose missing syllables spread to four different positions, marking it as less normative. This conclusion is further corroborated by the fact that in the 3,020 sampled lines from Chaucer’s iambic pentameter, there is no single poem whose distribution pattern of missing syllables parallels that in RomB, and all missing syllables in that data set occur in odd-numbered positions.

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

169

Table 4. Distribution patterns of the missing syllables

Poems (# of lines) RomA (169) RomB (408) RomC (186) BD (131) HF (215) Thop (14) Anel (3) Total line (1,126) Total cases (92)

MP1 (%) 10 (5.9%) 19 (4.7%) 7 (3.8%) 19 (14.5%) 21 (9.8%) – – 76 (6.7%) 76 (82.6%)

MP3 (%) 1 (0.6%) – – 2 (1.5%) 1 (0.5%) – – 4 (0.4%) 4 (4.3%)

// MP5 (%) 2 (1.2%) 7 (1.7%) –

MP6 (%) –

MP7 (%) –

1 (0.2%) –

1 (0.2%) –

1 (0.8%) –

–

–

–

–

– – 10 (0.9%) 10 (10.9%)

– – 1 (0.1%) 1 (1.1%)

– – 1 (0.1%) 1 (1.1%)

When measured by the hemistich saliency principle in (4), we ought to expect fewer cases of deviations in the second hemistich than in the first. However, of the 12 occurrences in the second hemistich in Table (4), 9 (75%) of them happen in RomB, once more singling out this fragment of the translation as being non-Chaucerian. 4.4. Extrametrical syllables Just as a metrical W may be omitted at the beginning of a line, and less often, at the beginning of a hemistich, so, too, an extra W position may be added at the end of a line and, less often, at the end of a hemistich. Prototypically, this extra W can be represented as resulting from a “split” S position in MP 8 (or less often in MP 4, 2, or 6): (12)

S S

(W)

170

Xingzhong Li

From the tetrameter data, I identified 60 cases of extrametrical syllables. This number was derived by omitting lines containing possible contractions, elisions, resolutions, and the word-final -e. All the identified occurrences appear at the end of a line, of the first hemistich, and of the first foot, as seen in (13): RomA 1540

(13) Ye that ageyns youre love mistakith, 1 2 3 4 5 6 7 8 x Or ellis, shortly, this is the eende, 1 2 3 4 x 5 6 7 8

RomB 1960

My lady, that is so fair and bryght! 1 2 x 3 4 5 6 7 8

BD 1180

Although line-final extrametrical syllables are not as significant as they are in other positions, they were nevertheless included in my counting since they are statistically significant in differentiating metrical styles of the poems under discussion. Table 5 shows how these extrametrical syllables are distributed in the data set. Table 5. Distribution patterns of the extrametrical syllables

Poems (# of lines) RomA (169) RomB (408) RomC (186) BD (131) HF (215) Thop (14) Anel (3) Total lines (1,126) Total cases (60)

F1 (%) – 1 (0.2%) – 1 (0.8%) 1 (0.5%) – – 3 (0.3%) 3 (5.0%)

F2 (%) 1 (0.6%) 11 (2.7%) 1 (0.5%) 4 (3.1%) 1 (0.5%) – – 18 (1.6%) 18 (30.0%)

F3 (%) – – – – – – – – –

F4 (%) 6 (3.6%) 7 (1.7%) 10 (5.4%) 6 (4.6%) 10 (4.7%) – – 39 (3.5%) 39 (65.0%)

Two significant findings are perceived in the table. First, no lines are found that include extrametrical syllables at the end of the third foot. Thus, headlessness and extrametricality are mirror-image principles in Chaucer’s iambic tetrameter verse: metrical W positions are omitted most often at the beginning of a line and next most often at the beginning of the second hemistich; conversely, metrical W positions are added most often at the end

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

171

of the line and next most often at the end of the first hemistich. Taken together, these tendencies lend strong support to the dipodic structure of Chaucer’s verse indicated in the metrical prototype (1). Second, the distribution pattern in RomB ignores the overall trend at the end of the first hemistich as well as of the line, as shown in Table 6: Table 6. The high-low incidence of the extrametrical syllables

Poems RomA RomB RomC BD HF Thop Anel

F1

F2 L (0.6%) H (2.7%) L (0.5%) L (3.1%) L (0.5%) – –

(0.2%) (0.8%) (0.5%) – –

F3 – – – – – – –

F4 H (3.6%) L (1.7%) H (5.4%) H (4.6%) H (4.7%) – –

ratio (F4 : F2) 6.0 0.6 10.8 1.5 9.4

The canonical incidence trend is clearly LH, as in all other poems, which evidentially puts both RomA and RomC into the group of Chaucer’s genuine poems. The ratios (F4 : F2) are more contrastive. While those in all other poems are one or two digits high, the ratio in RomB is only a fraction. These obvious contrasts in both the trend and the ratio mark off RomB as aberrant and least conforming to the metrical prototype in (1). 4.5. Syllables subject to resolution and anapestic feet Setting aside the lines with elidable and extrametrical syllables leaves other lines with different kinds of extra syllables. Traditionally, lines such as these are counted as including anapestic substitutions. Prototypically, anapestic feet take the following two forms, with (14a) being far more common than (14b): (14) a.

b.

F

F

S

W

W

W

S

S

W

S

by

the

dawn’s

ear

ly

light

172

Xingzhong Li

In individual cases, however, it is sometimes difficult to decide whether two adjacent unstressed syllables are fully contracted, elided, resolved, or whether they are separate, albeit “weak”, syllables. For example, (15)

W W S She wep[eth the tyme] that she hath wasted,

RomB 5010

W W S And of ribaw[dis shalt be] my king,

RomC 7300

These lines include geminate consonant series: “wepeth] [the tyme” [wép‹ș] [ð‹ tƯm] and “of ribawdis] [shalt be” [-dIs][šalt] or [-dIš][šalt] due to the phonological assimilation. They are likely to be pronounced [wép‹ș‹ tím] and [dIšalt] with deletion of one of the geminate members, thus weakening any potential phonological phrase boundary. Consequently, I treated geminate consonant series such as those in (15) as examples of resolution. Among the sampled lines, I identified 28 cases of anapestic feet. Interestingly, they exhibit four stress patterns instead of two, listed here in the order from the most to the least canonical: WWS, SWS, WWW, WSS. Illustrative examples are shown in (16): (16)

W W S And mat [in the myd] poynt of the checker 1 2 3 4 5 6 7 8 x

BD 660

S WS And though they die, they sette [not a lek]. 1 2 3 4 5 6 7 8

RomB 5730

W W W [And in hir] servise her hertes set. 1 2 3 4 5 6 7 8

RomB 3710

W S S “A Goddes half, [in good tyme]!” quod I, 1 2 3 4 5 6 7 8

BD 370

W S S [That so swe]tely smelleth in my nose. 1 2 3 4 5 6 7 8

RomB 3660

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

173

Table 7 below exhibits the distributions of the stress patterns in these identified anapests: Table 7. Distributions of the four stress patterns in the 28 anapests

Poems

WWS (%)

SWS (%)

WWW (%)

WSS (%)

Total (%)

RomA RomB RomC BD HF Thop Anel Total

– 4 (14.3%) 4 (14.3%) 6 (21.4%) 2 (7.1%) – – 16 (57.1%)

– 3 (10.7%) 1 (3.6%) – – – – 4 (14.3%)

– 1 (3.6%) – – – – – 1 (3.6%)

– 6 (21.4%) – 1 (3.6%) – – – 7 (25.0%)

– 14 (50.0%) 5 (17.9%) 7 (25.0%) 2 (7.1%) – – 28 (100.0%)

Once more, RomB is not in line with the metrical norm evident in other poems not only because it alone contains 50.0% of all the anapests but also because the eccentric WWW and WSS anapestic patterns occur nearly exclusively in it. This evidence becomes even more decisive if we look at Chaucer’s practice in his iambic pentameter lines. Of course, one may argue that Chaucer’s iambic pentameter is a completely different kind of verse and, therefore, it is statistically invalid to compare it directly with his iambic tetrameter verse since pentameter lines are 25% longer and, therefore, they might be expected to have 25% more metrical variations overall. This argument makes sense; however, it applies to some variations but not to all of them. This is because some variations, such as syntactic inversions, are more likely to have line-wide, couplet-wide, or global effects whereas other variations have only local effects and are thus less affected by line length. For example, “trochees only affect two adjacent syllables rather than the whole line; missing syllables only affect one weak metrical position, as do resoluble syllables. And extrametrical syllables do not affect underlying metrical position at all” (Li, 1995: 232). Of course, even in the case of local variations, it remains invalid to compare their total incidence per line because longer lines are more likely to contain more variations and shorter lines fewer variations. However, the contrary reasoning is more convincing. That is, if on an average, shorter lines contain more variations than longer lines, then it is valid to claim that the former is less regular than the latter. It

174

Xingzhong Li

is with this reasoning that I compare anapests in Chaucer’s iambic tetrameter with those in his iambic pentameter. Below, I list more revealing statistics to show metrical peculiarity in RomB, which is excluded from other tetrameter lines: Table 8. Distribution patterns of anapests in RomB, Chaucer’s tetrameter verse, and his pentameter verse

Data Sets (# of lines) RomB (408) Tetrameter (718) Pentameter (3,020)

Anapests (%) 14 (3.4%) 14 (1.9%) 15 (0.5%)

WWS (%)

SWS (%)

WWW (%)

4 (28.6%) 3 (21.4%) 1 (7.1%) 12 (85.7%) 14 (93.3%)

WSS (%) 6 (42.9%)

1 (7.1%)

–

1 (7.1%)

1 (6.7%)

–

–

In the 718 sampled tetrameter lines, there are only 14 anapestic feet, and their ratio with the total number of sampled lines, excluding RomB, is as low as 1.9%, which is much closer to the ratio of 0.5% in Chaucer’s 3,020 sampled iambic pentameter lines, where only 15 cases of anapests were found, than that in RomB, which is 3.4% high. A more telling fact that clinches all the preceding arguments is that 100% (15 : 15) of the anapests in the pentameter lines and 92.9% (13 : 14) of the anapests in the tetrameter lines show the prototypical WWS and SWS patterns whereas only 50% (7 : 14) in RomB exhibit them. If we just focus on the non-canonical WWW and WSS patterns, their frequency in the tetrameter and pentameter lines taken together is an insignificant 3.4% (1 : 29), but the frequency in RomB runs up to 50% (7 : 14). Statistics such as these, which illustrate that Chaucer’s stress norm for anapests is WWS, or less frequently, SWS, and that his underlying metrical pattern for iambic verse remains duple rather than triple even if he occasionally permits split positions, do present a striking contrast between RomB and Chaucer’s other lines of verse.

5.

Other tests

After we have examined metrical evidence, let us look at other tests Skeat used over one hundred years ago. In determining the authorship of RomC,

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

175

Skeat (1900) uses three main tests: (1) dialects, (2) imperfect rhymes, and (3) the diffuse correspondence between the original line numbers in Le Roman and the translated line numbers in The Romaunt. For the first test, Skeat claims, “That Fragment C has nothing to do with B is easily seen; for the traces of [RomB’s] Northumbrian dialect at once disappear, and we return to a dialect which does not materially differ from that of Fragment A” (90). In terms of the second test, Skeat lists several types of unsatisfactory rhymes that are not in Chaucer’s genuine poems, including words ending in -y from those ending in –ye (90–92). He found six examples of -y : -ye rhymes in RomC, but five of them become the -y : -y perfect rhymes except for companye : I (RomC 6875–6876) in Benson’s The Riverside Chaucer though there is one more case that Skeat does not list: sobrely: die (RomC 7405–7406). Skeat lists 21 examples of the -y : -ye imperfect rhyme that he found in RomB; however, all of them become perfect rhymes though I have found another mismatch in RomB but not in RomA: bitraye : Nay, 4551– 4552. The textual discrepancies cast some uncertainty on the rhyme test and warrant further studies. However, we might ascribe the discrepancies to editorial emendation. Anyway, imperfect rhyming is Skeat’s major reason to have excluded RomC from the Chaucer canon. Next, let’s take a look at Skeat’s diffuseness test, the results of which I reproduce here: the ratio for RomA is 101.6 (The Romaunt) : 100 (Le Roman), for RomB 117.5 : 100, and for RomC 102.1 : 100 (77, 90). He certainly takes the closeness between A and C as an argument for treating the latter as having a better chance of being genuine before he rejects C as Chaucer’s work, and his evidence for rejection is as weak as this argument is strong. Since I used Benson’s Riverside Chaucer, my calculations, given in Table 9, are slightly different from Skeat’s in regard to B and C: Table 9

Author Skeat Li

Ratios between the English lines and the corresponding French lines

RomA 101.6 : 100 101.6 : 100

RomB 117.5 : 100 117.6 : 100

RomC 102.1 : 100 101.8 : 100

RomC’s diffuseness in Skeat is a little higher than RomA’s, but only to the extent of 0.5%, as Skeat admits (90). Mine shows an even lower degree of diffuseness between A and C, the difference being only 0.2%. Though this

176

Xingzhong Li

fact is more literary than strictly metrical, it does place RomA and RomC into one category.

6.

Conclusion

Table 10 below summarizes what we have found, including Skeat’s criteria, where a “¥” represents the Chaucerian style and an “x” a deviant style: Table 10. A summary of the findings

Metrical Deviations Trochaic substitutions Trochaic makeup Headless lines Extrametrical syllables Anapestic feet Dialects Rhymes Diffuseness

RomA RomB RomC BD HF Thop Anel ¥ x ¥ ¥ ¥ – – ¥ x ¥ ¥ ¥ – – ¥ x ¥ ¥ ¥ – – ¥ x ¥ ¥ ¥ – – ¥ x ¥ ¥ ¥ – – ¥ x ¥ n/a n/a – – ¥ x x n/a n/a n/a n/a ¥ x ¥ n/a n/a n/a n/a

These findings in the aggregate show that RomA wins all eight votes and is thus evidentially Chaucerian in style, because metrically it deviates from the iambic tetrameter prototype in a way remarkably similar to The Book of the Duchesse and The House of Fame. RomB, however, deviates from the prototype in quite peculiar ways, fails to win a single vote, and is therefore evidentially non-Chaucerian in style. It is not by Chaucer. These two claims altogether corroborate the existing scholarship that Chaucer translated RomA but not RomB. RomC wins all of the five metrical votes as well as the two votes of dialects and diffuseness but fails to win the rhyme vote, and that presents a little uncertainty. However, all features considered, it is about 88% Chaucerian. A result such as this throws new light on an old uncertainty, encapsulated in Benson’s statement (1993: 686) that although RomC is “Chaucerian in language and manner,” it has been “rejected by most scholars”. To conclude, the metrical evidence in this study offers new and independent confirmation of a centennial claim by Skeat which has, since 1900, dominated the literature on The Romaunt’s authorship that RomA is indeed Chaucer’s work while RomB is decisively not. Nevertheless, the results of

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

177

this study disagree with Skeat’s hypothesis that RomC is not Chaucer’s work. It calls for Chaucerian scholars to reevaluate the authenticity of RomC and consider revoking the death penalty imposed on it and accepting it as part of the Chaucer canon.

Notes *

I’d like to thank the editors and the anonymous referees for their support of publishing this article and for their critical comments and rigorous editorial work, which benefited this article. All errors, however, remain mine. 1. Cable (2002: 132) provides a tentative proposal that several prepositions, including “after”, could be undergoing a stress shift during the thirteenth and fourteenth centuries. The proposal warrants further studies. 2. The choice of this example as a headless line was due to Thomas Cable’s suggestion, but I am alone responsible for the analysis.

References Benson, Larry D. (ed.) 1987 The Riverside Chaucer. 3rd ed. Boston: Houghton Mifflin Company. 1993 A Glossarial Concordance to the Riverside Chaucer. Vol. 2. New York: Garland Publishing, Inc. Cable, Thomas 2002 Issues for a new history of English prosody. In Studies in the History of the English Language: A Millennial Perspective. Donka Minkova and Robert Stockwell (eds.), 125–151. Berlin/New York: Mouton de Gruyter. Halle, Morris and Samuel J. Keyser 1966 Chaucer and the study of prosody. College English 28: 187–219. 1971 English Stress: Its Form, Its Growth, and Its Role in Verse. New York: Harper and Row. Hayes, Bruce 1989 The prosodic hierarchy in meter. In Phonetics and Phonology: Rhythm and Meter I. Paul Kiparsky and Gilbert Youmans (eds.), 201–260. San Diego: Academic Press. Hopkins, Kenneth D., Gene V. Glass, and B. R. Hopkins 1987 Basic Statistics for the Behavioral Sciences. 2nd ed. Englewood Cliffs, New Jersey: Prentice Hall.

178

Xingzhong Li

Kaluza, Max (ed.) 1891 Guillaume, Geoffrey Chaucer, Guillaume, Jean. The Romaunt of the Rose from the unique Glasgow ms, parallel with its original, Le Roman de la Rose. London: Pub. for the Chaucer Society by K. Paul, Trench, Trübner & Co. Kiparsky, Paul 1975 Stress, syntax, and meter. Language 51: 576–616. 1977 The rhythmic structure of English verse. Linguistic Inquiry 8: 189– 247. Lecoy, Félix, ed. 1970 Le Roman de le Rose, by Guiliaume de Lorris et Jean de Meun. Paris: Librairie Honoré Chompion, 1970. 3 vols. Li, Xingzhong 1995 Chaucer’s meters. PhD. diss., Department of English, University of Missouri, Columbia. 2004 A central metrical prototype for English iambic tetrameter verse: Evidence from Chaucer’s octosyllabic lines. In Studies in the History of the English Language: Unfolding Conversations. Anne Curzan and Kimberly Emmons (eds.), 315–341. Berlin/New York: Mouton de Gruyter. Liberman, Mark and Alan Prince 1977 On stress and linguistic rhythm. Linguistic Inquiry 8: 249–336. Minkova, Donka 1996 Non-primary stress in early Middle English accentual-syllabic verse. In English Historical Metrics. C. B. McCully and J. J. Anderson (eds.), 95–119. Cambridge: Cambridge University Press. 2000 Middle English prosodic innovations and their testability in verse. In Placing Middle English in Context. Irma Taavitsainen, Terttu Nevalainen, Päivi Pahta, and Matti Rissanen (eds.), 431–461. Skeat, Rev. Walter W. 1900 The Chaucer Canon: With a Discussion of the Works Associated with the Name of Geoffrey Chaucer. New York: Haskell House. Rpt., 1965. New York: Haskell House. Tarlinskaja, Marina 1974 Meter and rhythm of pre-Chaucerian rhymed verse. Linguistics 121: 65–87. 1976 English Verse: Theory and History. The Hague: Mouton. 2006 What is metricality? English iambic pentameter. In Formal Approaches to Poetry: Recent Developments in Metrics. B. Elan Dresher and Nila Friedberg (eds), 53–74. Berlin/New York: Mouton de Gruyter.

Metrical evidence: Did Chaucer translate The Romaunt of the Rose?

179

Youmans, Gilbert 1989 Milton’s meter. In Phonetics and Phonology: Rhythm and Meter I. Paul Kiparsky and Gilbert Youmans (eds.), 341–379. San Diego: Academic Press. 1996 Reconsidering Chaucer’s prosody. In English Historical Metrics. C. B. McCully and J. H. Anderson (eds.), 185–209. Cambridge: Cambridge University Press. Youmans, Gilbert and Xingzhong Li 2002 Chaucer: Folk poet or littérateur? In Studies in the History of the English Language: A Millennial Perspective. Donka Minkova and Robert Stockwell (eds.), 153–175. Berlin: Mouton de Gruyter.

Trochees in an iambic meter: Assumptions or evidence?

Xingzhong Li Cable’s commentary in this volume on my authorship article states that anapestic feet and extrametrical syllables are tangible and lend themselves to a statistical analysis more readily than trochaic substitutions. Consequently, he proposes radically different readings of trochees on the assumption that Chaucer rather freely altered the normal linguistic stress of words to conform to a fixed WSWS alternation. However, statistical evidence favors just the opposite.

1.

Trochees of disyllables

Cable contends that Chaucer used more stress doublets and reads disyllabic words such as those ending in -ly as iambs rather than trochees at the end of a line: (1)

Nor of a thousand full scarsly

RomB 5460

There are only six instances of scarsly in Chaucer’s verse as a whole, one at a line’s beginning and four in SW metrical positions. The only one that occurs line-finally is (1). I take it as further evidence of RomB’s metrical aberrancy because Chaucer never placed it line-finally in his own verse, be it a trochee or iamb. Cable also contends that the word after/aftir should be read iambically, as in (2): (2)

Aftir to thenken on hir sorwe

BD 100

By contrast, I scan it as an onset trochee, taking it as evidence of the permissiveness of a line-initial SWWS cadence in Chaucer’s verse. If Cable

182

Xingzhong Li

were correct, the word should occur in rhyming position as well, but none of its 452 instances in the Chaucer canon employs -tir as a rhyming syllable except as a one-time feminine rhyme with rafter, shown in (3), suggesting that after is incontrovertibly stressed SW: (3)

And by assaut he wan the citee after, 1 2 3 4 5 6 7 8 9 10 x

KnT 989

In Chaucer’s tetrameter verse per se, this word occurs 48 times, and 44 (91.7%) either occupy SW positions or, less frequently, appear lineinitially. These positions are putatively recognized by both poets and metrists as expected and permitted. Though the word occupies verse-internal WS positions 4 times (8.3%), as in (4), (4)

She longed so after the king,

BD 83

it exclusively occurs at the beginning of the second hemistich, the second least constrained position in a line. Interestingly, this figure closely parallels Tarlinskaja’s statistics of 9.4% for the fifth position when she constructed stress profiles for Chaucer’s The House of Fame (1976: 259). Hence, I conclude that after in (2) and (4) are trochaic substitutions (a trochaic word with an extrametrical syllable in (3)), and that they are permissible variations in Chaucer’s verse rather than prohibited cadences. Cable correctly suggests that words with mis-, mys- in Chaucer’s English had variable stress, but that reading would efface the metrical oddity of RomB 5500, where the SWWS pattern follows a morphological prefix but not a more usual phonological break. In fact, true stress doublets like hónour or honóur occur more or less freely anywhere in Chaucer’s lines, whereas words like clepen, which is stressed SW, occur regularly in SW positions, with a few predictable exceptions (Youmans and Li 2002).1 Minkova 2000 reaches the same conclusion about other possible stress doublets such as author and labor. If the stress patterns of trochaic words such as after can be altered freely, why did not Chaucer scatter them at random throughout his lines, the way he did with honour? That is, how does Cable’s theory account for the statistical distribution of trochaic words? That Cable collapses the two tests involving trochees – their distribution in the line and their internal makeup – is an oversight. Of the 50 disyllabic trochees, RomB alone contains 27 (54.0%). In light of the ratios between

Trochees in an iambic meter: Assumptions or evidence?

183

disyllabic trochees and the total lines of individual poems examined, RomB is out of line (6.6%), but RomC parallels RomA closely (3.8 : 3.0%). This demonstrates that the author of A and C was more sensitive to disyllabic trochees than was the author of B.

2.

Trochees of monosyllables

Trochees of this type lack the rigid constraint on stress in polysyllables and are thus subject to stress variability. Insofar as metrical deviations occur frequently at the beginning of a line (in the first foot) and of the second hemistich in major iambic poets, I follow most metrists and scan monosyllabic sequences of a content word with an encliticized function word, such as “Nor of …” RomB 5460, “Ryght as …” RomA 30, “Loo, to …” HF 1070, “… god of …” RomC 6010 in WS positions, as trochees. Sweepingly, Cable promotes all monosyllabic function words in even positions, for example, (5) a. If that I were god of richesse b. Feble as a forwoundid man

RomC 6010 RomA 1830

In discussing the significance of the Chaucerian metrical alternation, Minkova and Stockwell aptly state that three weak syllables in a row promote the middle one to metrical strong (2003: 130). This rule, while justifying Cable’s promotion on that in (5a), though his promotion is irrelevant to my argument since “if that” is not analyzed as a trochee in my article, blocks his promotion on of because it is not flanked by two weak syllables. The rule also effectively blocks his promotion on the schwa a, because it is surrounded with three instead of two weak syllables and loses its middleposition status. Statistics also support this analysis. Of the 264 instances of of in RomA, 212 (80.3%) occupy odd positions; the rest occupy even positions flanked by two unstressed syllables or, rarely, by one weak and one strong syllable. Thus, of is an example of an unstressed or weakly stressed syllable occupying an odd-numbered position. Regarding the schwa, it occurs 177 times in RomA, 173 (97.7%) occupy odd positions. A few exceptions occur in a permissible SWWS sequence. Consequently, I conclude that “god of” in (5a) is a strong trochee and “as a” in (5b) a weak one.

184

3.

Xingzhong Li

Conclusion

Of course, it is possible to read any sequence of eight or ten syllables in a strict WS alternating pattern if we ignore normal linguistic stress. Then what is the basis for Cable’s argument that Chaucer expected us to read his verse this way? Cable’s answer is: “There should be a beat” and “the normal placement of stress can destroy the beat.” But both “temporal metrics” and “stress metrics” accept the existence of rhythmic syncopation – rhythmic variation – especially in literary verse like Chaucer’s. Without it, verse tends to become sing-song doggerel. The question then becomes, what are the constraints on rhythmic variation? One example accepted by Chaucer and every other major iambic poet in English was a SWWS cadence, as discussed above. Finally, even if one accepts Cable’s position, my statistical evidence on trochees is still relevant to authorship studies because I used the same criteria for assigning relative stress in all the verse samples, so statistical differences are still diagnostic. I claim that distinctively different rhythmic patterns are found in RomB and the set of all other poems under scrutiny; whereas Cable might argue that normal linguistic stress can be altered in distinctively different kinds of words in the two samples. Either way, the poetic practice is different, providing evidence that the authors might be different, too.

Note 1. Editorial comment: the ratio of initial-to-final stress on honour in Chaucer is 41 :12; we would therefore recommend initial stress as the preferred scansion, following Minkova (2000: 448).

Trochees in an iambic meter: Assumptions or evidence?

185

References Minkova, Donka 2000 Constraint ranking in Middle English stress-shifting. English Language and Linguistics 1(1): 135–175. Middle English prosodic innovations and their testability in verse. In Placing Middle English in Context. Irma Taavitsainen, Terttu Nevalainen, Päivi Pahta, and Matti Rissanen, (eds.), 431–461. Berlin: Mouton de Gruyter. Minkova, Donka, and Robert Stockwell 2003 Emendation and the Chaucerian metrical template. In Chaucer and the Challenges of Medievalism: Studies in Honor of H. A. Kelly. Donka Minkova and Theresa Tinkle (eds.), 129–139. Frankfurt am Main: Peter Lang. Tarlinskaja, Marina 1976 English Verse: Theory and History. The Hague: Mouton. Youmans, Gilbert, and Xingzhong Li 2002 Chaucer: Folk poet or littérateur? In Studies in the History of the English Language: A Millennial Perspective. Donka Minkova and Robert Stockwell (eds.), 153–175. Berlin/New York: Mouton de Gruyter.

“Ubbe dubbede him to knith”: The scansion of Havelok and ME -es, -ed, and -ede*

Christina M. Fitzgerald When did “lov-ed” become “lov’d”? That is, when did the -ed cease to be syllabic? For that matter, when did plurals and genitives ending in -es, or third person singular, present-tense verbs ending in -es lose their syllabicity? It is possible that the changes were in full progress as early as the first decades of the fourteenth century. Following G. V. Smithers’ study of scansion and the use of ME final -en and -e in Havelok the Dane (Smithers, 1983), I am using scansion of the same early fourteenth century poem to elicit information about the pronunciation or syllabicity of -es and -ed(e), at least in East Midlands English of the period.1 I will argue that together onethird of these endings in the poem show syncopation of the internal -enecessary for metrical regularity, and that such a significant number of instances of necessary syncopation suggest that it is “legitimate” for the poet and thus available in his spoken ME dialect. Furthermore, I will argue that the frequently ambiguous cases of the -ede ending, as in my title – is it “Ubbe dubb-ed” or “Ubbe dubb-de”? – actually give us additional evidence for the loss of syllabic -ed in the early fourteenth century. First, why Havelok? As Smithers discovered, Havelok is an ideal poem for using metrical analysis to elicit linguistic information because the composition date is narrow, the provenance is relatively certain, the manuscript date is fairly close to the composition date, and the meter throughout is predictably regular. The only complete version of Havelok is in Bodleian MS Laud Misc. 108, a compilation manuscript completed ca. 1450, but portions of which are much earlier, including the booklet that contains Havelok, which is written in a ‘textura’ hand of the early fourteenth century (Smithers, 1987: xii). The date of composition can be limited to a range between 1301 and 1310, according to Smithers, since the “parlement” at Lincoln referred to twice in the text (ll. 1007, 1180) suggests the 1301 parliament, and a later version of the story, part of Le Petit Bruit of Rauf de Bohun, appears to be modeled on the ME version and was commissioned in

188

Christina M. Fitzgerald

1310. Both the poet and the text’s copyist use East Midland varieties of Middle English of the early fourteenth century.2 Thus Havelok makes an excellent source of linguistic information for Middle English (especially of East Midlands varieties) at the beginning of the fourteenth century, a period of transformation between early and late ME, and might also be used in other studies. What is more, as mentioned, Havelok appears to be reliably regular in its meter. As Smithers points out, the poet gives us internal evidence for his line-type when he uses “Benedicamus Domino” from the Mass as a whole line (l. 20), which “gives us, in a wholly unambiguous spelling, a norm of eight syllables…hence nine for those with feminine rhymes” and seven for headless lines (Smithers, 1983: 199,201). A typical feminine line looks like this: (1)

x / x / x / x / x l. 467 But tok þe maydnes boþe samen

The rate of the poet’s use of feminine lines is hard to determine, given that many end in final -e or other inflectional endings that might be deleted or syncopated. Headless lines, on the other hand are very common (about a third of my sample); a typical one looks like this: (2) l. 621

Ø / x / x / x / Þine cherles, þine hine

The norm of eight syllables in four iambic feet can also be seen in lines made up of monosyllabic words, where no question of syncopation, apocopation, or elision of inflectional endings is concerned; for example: (3) l. 1808

x / x / x / x / And at a dint he slow hem þre.

The other main variant lines in Havelok involve inversion of stress. The most common of this type is, predictably, inversion in the first foot3: (4) l. 2552

/x x / x / x / Brini on bac, and sheld and spere.

“Ubbe dubbede him to knith”

189

Inversion in other positions is rare, but does occur at syntactical breaks in the line, in the third foot, as in the following example (which also happens to be an example of necessary syncopation in an inflectional ending): (5) l. 2325

/ x x / (x) / x x / Wrastling with laddes, putting of ston.

In all these variants, the meter still consists of four “stresses” or “on-beats,” with only the number of “unstressed syllables” or “off-beats” varying; that variation, however, is tightly constrained by the position of the off-beat in the line: a missing off-beat at the head of the line or an additional one at the right edge. The underlying abstract metrical pattern in Havelok is thus iambic tetrameter. Of course, problematic lines still remain, but at a significantly low rate; in the lines I scanned for this paper, 537, only 4.28% were unscannable as written. In short, Havelok’s meter is almost oppressively regular, a quality that once dissuaded literary critics of its merits, but is a boon to historical linguists. Before I move on to my findings, I would like to emphasize that in my analysis of Havelok, I am not concerned with reconstructing all of the prosodic features of the poet’s language, but rather with scansion and metrical stress as tools to unlock linguistic information. As Roger Lass writes, “If we have reason to believe that a poet follows a strict metrical scheme, his verse may be evidence for things like syllable count and stress placement” (Lass, 1992: 28). I am following Smithers in using Havelok and scansion in this way, and should also say something about his findings regarding final -en and -e , especially as they affect some of the -ede endings. In brief, by isolating these endings in their specific metrical environments, Smithers finds that where the inflections would cause an extrametrical unstressed syllable, their apocopation or elision makes the line perfectly regular.4 He also shows that final -n is often spurious, most likely a faulty “emendation” by the scribe, without which final -e can be apocopated or elided. Where the final -n is grammatically required – for example, in infinitives – Smithers shows that it is also metrically required. Likewise, where the pronunciation of word final -e is grammatically legitimate, as in the inherited difference between a “definite” (OE weak) adjective and an “indefinite” (OE strong) form, it is also metrically required (Smithers, 1983). Where Smithers marks these endings as unpronounced or spurious, I do so also. A word now on my notational system, which, for the most part, is standard. On-beats and off-beats are marked with a / and an x respectively,

190

Christina M. Fitzgerald

positioned above the vowel or diphthong of the syllable. Headless lines are marked at the beginning by a zero with a line through it. Unpronounced syllables other than the inflectional endings which I am investigating are marked with a dot under the vowel of the syllable. If Smithers has argued that a final -en is spurious and should be -e, I italicize the -n and, where appropriate, mark the elision or apocope of the -e with a dot under it. For clarity, I have marked any unstressed syllable that I am arguing should be syncopated with an x in parentheses: (x). I have ignored line-final syllables, except where rhymes tell us that they must certainly be pronounced. And finally, I have underlined the word with the inflectional ending under discussion in each line. In extending Smithers’s methodology to the inflectional endings -es and -ed(e) I use his categories of their metrical environments (Smithers, 1983). I have also added a handful of my own categories. First, for each inflectional ending, I have pointed out the existence of de facto “orthographic syncope,” where the poet or the scribe has left out the -e- before -s, -d, or -de, where we might expect one. Second, in the -es section I have isolated those plural nouns of French origin which form the plural with only an -s or a -z, and which may have had some influence on English. Third, in both the -es and the -ede sections there are trisyllabic words in which syncopation or apocopation/elision is required, but which are ambiguous because it is unclear which -e- is affected. Furthermore, because of this widespread ambiguity in -ede, I have separated -ed and -ede into their own sections. Finally, I have discovered that the numerous ambiguous cases of -ede may actually tell us the most about the status of the pronunciation of inflectional endings and their obscuration by scribal practices of “correcting” poetry, which I will discuss at the end of this paper. Limits of space prevent me from presenting all instances of -es and -ed(e) from Havelok, so I will here give only sample lines from each category and each inflectional ending. My complete set of evidence, in separate appendices, one for each inflectional ending sorted by their metrical categories, is available upon request. I have also noted line citations of every instance of each inflection in each metrical category, for the reader’s reference. The complete evidence shows that in the regular, non-ambiguous cases, each inflectional ending can be found in the same set of metrical environments, behaving in the same way in each environment, regardless of grammatical function. The first category, as mentioned, is “orthographic syncope.” For example:

“Ubbe dubbede him to knith”

(6) l. 252

x / x / x / x / And in þe castels leth he do5

l. 2766

x / x / x / x / Til knithes hauҽden demd him rith6

l. 209

x / x / x / x / And preidҽ he shulde yemҽ hirҽ wel7

l. 436

x / x / x / x / Of Christ þat maude monҽ and sunne!8

191

Although they are few – in fact, these are the only examples – these spellings suggest that the -e- of these inflectional endings was unpronounced linguistically (or “naturally”) as well as metrically, and allow for the possibility of the metrical syncopation of -e- in other instances. In the second environment (Smithers’ first), -es, -ed, and -ede are followed by an unstressed syllable with an initial vowel or diphthong.9 For example: (7) l. 1317 l. 1862 l. 23 l. 2873

x / x / x /(x) x / Þat wite þw þat sittes in trone! Ø / x / (x) x / x / For þe laddes on ilke wise x / x / (x) x / x / Þe rym is maked of Hauelok Ø / x / (x) x / x / Hauҽ Ich lived into þis day

Ø / x / x / (x) x / l. 680 Godard stod and lokedҽ on him l. 2365

x / x / (x) x / x / Yet hwan he hauedҽ of al þe lond

In each of these examples, without syncopation, two successive off-beats are formed; however, with syncopation, the lines are perfectly regular. The only exception is in a case of an inverted first and second foot, with a word in which the phonotactic constraints make syncopation impossible:

192

Christina M. Fitzgerald

(8) l. 2105

/x / x x / x / Glotuns, reures, or wicke þeues.

Clearly “reures” – or “revres” – cannot be further reduced, and even if it could, the inverted feet require an off-beat in the position of the word’s second syllable. Otherwise, every other case in this category presumes syncopation of the inflectional ending. In the third category, the inflectional endings are followed by an unaccented syllable with an initial h- + vowel (which does not include hw-).10 For example: (9) l. 253 l. 2132 -ed l. 1386 l. 1753

x / (x) x / x / x / Þe knictes he micte tristen to Ø / x / (x) x / x / In his armes his brithe bride no examples Ø / x / (x) x / x / Þannҽ he hauedҽ his bede seyd x / x / x / (x) x / And bad him als he louedҽ his lif

As in Environment 2, syncopations of -e- would form perfect lines. The lack of examples in the -ed category, meanwhile, has more to do with the overall paucity of this form than with a lack of syncopation. Environment 4 follows the same pattern as Environments 2 and 3. In it, the inflections are followed by an unaccented syllable with an initial consonant (other than h-) plus a vowel or a diphthong.11 For example: (10) l. 1923 l. 2976

x / x / (x) x / x / Als it werҽ dogges þat weren henged Ø / x / x / (x) x / Neuerҽ yete wordes ne grewe

l. 2004

x / x / x /(x) x / Of þat Ich was þus greþed tonith.

l. 231

x / (x) x / x / x / And deyedҽ biforn his heymen alle.

“Ubbe dubbede him to knith”

l. 1863

193

Ø / x /(x) x / x / Him asayledҽn wit grete dintes.

Again, in these examples, syncopation of the marked inflectional endings would make these lines metrically perfect. We have another exception in this category, like the one in Environment 2, which involves both inverted feet and an additional, significant phonotactic constraint: (11) l. 688

/ x / x || x / x / To þe galues, so God me rede!

Clearly “galues” cannot be any more reduced than it already is, and even if it could, the inverted first and second feet require an off-beat in the position that the second syllable of “galues” holds. Otherwise, in every other example in this category, syncopation of the inflectional ending seems presumed by the poet. And in this category we have real evidence of language change, rather than metrical use alone, as here these syncopations are occurring in environments before consonants as well as before vowels and h+vowel, as in categories 2 and 3. Because these cases before consonants run against the usual expectations, the syncopation of these inflectional endings must therefore have been held as “legitimate” by the poet, as part of his spoken language. The fifth environment is divided into three sub-categories and differs slightly from previous examples. Here, each inflectional ending is the second of two successive unaccented syllables, and is followed by an accented syllable, except in one case involving an inverted first foot, where the syllable following the inflectional ending is unaccented.12 This category is by far the scantiest one in all cases, especially sub-divisions (b) and (c). The three divisions are: (12) (a) Followed by an accented syllable with an initial consonant: x / x / x / x (x) / l. 821 With-held he nouth a ferþinges nok l. 434

/ x(x) x /x / x / Waried wrþe he of norþ and suth13

l. 1989

x / x / x (x) / x / But siþen he hauedҽ lauth þe sor

194

Christina M. Fitzgerald

(b) Followed by an accented syllable with an initial h-: Ø / x / x(x) / x / l. 785 In þe se-weres he oftҽ setes -ed

no examples

-ede

no examples

(c) Followed by an accented syllable with an initial vowel or diphthong: x / x (x) / x / x / l. 2416 Of Godardes alþerbeste men -ed

no examples

l. 658

x / x (x) / x / x / Hwan he hauedҽ eten and was fed

In each of the above examples, the inflectional endings again cause two successive off-beats unless they are syncopated, in which case the lines would be absolutely regular. As these categories show thus far, the possibility of syncopation occurs where we would expect it metrically and where no prior phonotactic constraints prevent it. Where syncopation would be constrained – for example, in a word such as “wounded” – the poet constructs lines where the syllabicity of the inflectional ending is metrically required as well. The final categories show the poet using syllabic inflectional endings as off-beats in his metrically regular lines. Beginning with Environment 6, we have cases in which syncopation is not necessary because the inflectional endings form an off-beat between two on-beats. In this category, the endings are followed by an accented syllable with an initial consonant:14 (13) l. 63

Ø / x / x / x / He was Engelondes blome

l. 143

x / x / x / x / In harde bondes nicth and day

l. 1369

Ø / x / x / x / Bad he Grim hauҽ drenched me

“Ubbe dubbede him to knith”

x / x / x / x / And haued me to sorwe brouth.

l. 1373 l. 655

195

Ø / x / x /x / For him hungredҽ swiþe sore x / x / x / x / Þat it ne doutedҽ sond ne krike

l. 709

There is one exception in this category, and once again it involves an inverted foot: (14)

/ x x / (x) || / x x / Wrastling with laddes, putting of ston

l. 2325

In this case, the inversion of the third foot after the syntactical and metrical break means that the inflectional ending -es need not be and should not be pronounced, even between two stressed syllables. Otherwise, every other inflectional ending in Environment 6 is syllabic. In the final regular category, Environment 7, the inflections are also syllabic, for they are followed by an accented syllable with one of two initial environments:15 (15) (a) with an initial vowel or diphthong: Ø /x / x / x / l. 515 Louerdinges after me x / x / x / x / But he werҽ king or kinges eyr

l. 1116 l. 670

Ø

/x / x / x / He is drenched in þe flod

l. 1430

x / x / x / x / And heyҽ ben henged on a tre

l. 37

Ø / x / x /x / Ricthwisҽ men he louedҽ alle

(b) with an initial h- + vowel or diphthong: Ø /x / x / x / l. 2241 Lokes hwarҽ he stondes her!

196

Christina M. Fitzgerald

l. 2613

x / x / x / x / Þe helmes heyҽ on heued sette

l. 40

x /x / x / x / And hated hem so man doth galle

l. 305 l. 1155

Ø / x / x / x / Hic hauҽ yemed hirҽ to softe x / x / x / x / And grauntedҽ him al þat he bad.

As in the previous category, there is no syncopation in these examples, for it would ruin the meter of the lines. And in many cases – such as “hated” (l. 40) – syncopation is phonotactically constrained. There are no exceptions in this category.

Discussion/Conclusions The sheer regularity of the way the inflectional endings fit into all these categories, and the overall regularity of the meter that the use of syncopation produces, coupled with the evidence of orthographic syncope, however slight, not only suggest that syncope is metrically allowable and necessary, but also suggest a change in the poet’s spoken Middle English. As my statistical analysis will show shortly, syncopation, though not dominant in the poem, is nevertheless so common throughout that it must be legitimate and natural to the producers of this poetic text. There are, however, a significant number of ambiguous lines, especially in the cases of endings -es and -ede, as well as a much smaller number of wholly abnormal lines in every case, which make a totalizing argument impossible but not improbable. First, in the -es category, there are twentyone instances of lines that can be scanned in different ways, making the fate of -es ambiguous.16 Eight of these instances involve a trisyllabic word ending in -e+consonant+es; for example: (16) l. 82

/ x / x x || x / x / And in feteres ful faste festen

“Ubbe dubbede him to knith”

197

This particular line also involves two inverted feet, but some syncopation in “feteres” is still required, either in the second or third syllable; it cannot be absolutely determined which one. We can also see this ambiguity at work in the spelling of the word “casteles” (l. 397) as also “castels” (l. 252) and “castles” (l. 1294). Nevertheless, in these ambiguous cases, syncopation of one -e- or the other is necessary, even if it does not tell us anything conclusive about final -es. In eleven cases of ambiguous -es, the word under discussion forms the first foot in a line that could either be headless or complete with an inverted first foot.17 (This is also the general reason for ambiguity in -ed and for some of the -ede words.) For example: (17)

Ø? l. 371

/ (x) x / x / x / Knictes an sweynes bi herҽ siden

Is it “knict’s an sweynes” or “knictes an sweynes”? Finally, the remaining ambiguous -es cases are merely lines where one or more word could be the word undergoing inflectional ending syncopation or other loss.18 The ambiguous category in the case of -ede is much larger, much more complex, and, I think, more interesting than that of -es. It includes sixty-six examples, sixty in which a line internal -ede word must be two syllables, but where either -e- could be syncopated (subcategory a); for example: (18)

x /x x / x / x / He louede God with al his micth

l. 35

The other six examples (subcategory b) involved an -ede word at the beginning of a line that could be either headless or begin with an inverted foot, as in this example: (19)

Ø? l. 899

/ (x) x / x / x / Sparedҽ he neyþer tos ne heles

In subcategory (a) forty eight instances occur before stressed syllables beginning with consonants, twelve are before stressed syllables with h+vowel, and none are before stressed syllables beginning with vowels.19 In all of these examples, either the first -e- is syncopated or the final -e is apocopated; either way, the line would be regular. Where the stressed syllable following the -ede begins with a vowel, nothing is ambiguous: we can

198

Christina M. Fitzgerald

assume that the final -e undergoes elision, leaving a pronounced but unstressed -ed. (These instances, in Environment 7(a), are far fewer than before consonants – there are only six total.) Although -ede is commonly attested, it may be that the extrametrical final -e was added by the scribe when he assumed that -ed was syncopated and therefore that another syllable was needed, and he followed suit even in the six cases before vowels. Thus the line in this article’s title would be “Ubbe dubb-de him to knith” according to the scribe, but originally “Ubbe dubb-ed him to knith” in the poet’s ears. That the scribe frequently “corrected” the poet – for example, according to Smithers, adding final -n where the scribe thought it was needed to prevent hiatus – suggests strongly that here, too, he is correcting the poet and that -ed was not naturally pronounced in his spoken ME. This also accounts for the fact that there are 139 cases of -ede, almost half of which are ambiguous, and only 46 cases of -ed alone. Add to that my findings that of the scannable, unambiguous instances of -ede, 46 out of 61 instances, or 75.41%, require syncopation for regularity – by far the highest rate of any of the inflectional endings – and you have solid evidence that -ed(e) especially was non-syllabic in spoken East Midlands Middle English ca. 1300. Finally, in each inflectional ending group, there are abnormal lines that do not scan properly without emendation; however, their numbers are surprisingly low. Together, these lines represent a meager 4.28% of all the examples of the inflections under consideration. And in some cases, information about the behavior of the inflectional endings can still be elicited even though the entire line cannot be scanned. Before I move on to a discussion of the remaining quantitative data, I want to briefly note one last category that I have distinguished from the -es examples, that is, French loan words which form their plurals with -s or -z only (Zink, 1989: 9–21). Twenty-two separate lines are affected in this category, although only ten individual words are involved.20 As Fernand Mossé suggests, it is possible that French morphology and phonology affected English. As he writes, “As for the plural, since it was almost always in -s (or sometimes in -z) in Old French, it was quite naturally identified with the plural in -(e)s which was being generalized for English words (OF barons > ME barouns). ” (Mossé, 1952, 1968: 53). This influence may be showing its affect here in Havelok, with the syncopation of -e- before -s. As my examples so far have demonstrated – and which a perusal through the entire dataset would also make clear – there is a consistent pattern to the Havelok poet’s use of syncopation; numbers will make it even

“Ubbe dubbede him to knith”

199

more clear. Note the following statistics regarding the combined behavior of the inflectional endings -es, -ed, and -ede, followed by the statistics for each ending alone: Table 1. Total inflectional endings counted: 537 – Ambiguous lines: 91 – Abnormal lines: 23 – French loans w/ -s or -z: 22

Regularity of scansion: Total: 537 – Abnormal: 23 = Total regular environments: 514 (95.72%) = Total instances for statistical purposes below: 401

Total # of syncopated inflectional endings:

135 (32.67% of regular & unambiguous environments) Orthographic: 4 ( 2.96% of syncopated endings) Before vowels: 57 (42.22%) Before h-: 27 (20.00%) Before consonants: 47 (34.81%) Total # of unsyncopated inflectional endings: 266 (66.33% of regular & unambiguous environments) Before vowels: 60 (22.56% of unsyncopated endings) Before h-: 20 ( 7.32%) Before consonants: 186 (69.92%) Total Metrical Promotion:

55 instances (13.72% of regular & unambiguous environments) In syncopated environments: 10 (7.41% of 134 total syncopated instances) In unsyncopated environments: 45 (16.92% of 266 total unsyncopated instances)

Breakdown for -es: 330 instances (excluding 22 instances of -s or -z alone in French-derived words) – Ambiguous lines: 21 – Abnormal lines: 8 = Total regular instances: 301 Syncopated instances: 77 (25.58% of all regular & unambiguous environments for -es) Orthographic: 1 (1.30% of syncopated -es endings) Before vowels: 37 (48.05%) Before h-: 13 (16.88%) Before consonants: 26 (33.77%)

200

Christina M. Fitzgerald

Unsyncopated instances:

224 (74.42% of all regular & unambiguous environments for -es) Before vowels: 41 (18.30%) Before h-: 15 (6.70%) Before consonants: 175 (75.00%) Breakdown for -ed: 46 instances – Ambiguous lines: 2 – Abnormal lines: 5 = Total regular instances: 39 Syncopated instances: 12 (30.77% of all regular & unambiguous environments for -ed) Orthographic: 1 (8.33% of syncopated -ed endings) Before vowels: 8 (66.67%) Before h-: 0 (0%) Before consonants: 3 (25.00%) Unsyncopated instances: 27 (69.23% of all regular & unambiguous environments for -ed) Before vowels: 12 (44.44% of unsyncopated -ed endings ) Before h-: 2 (7.41%) Before consonants: 13 (48.15%) Breakdown for -ede: 139 instances – Ambiguous lines: 68 – Abnormal lines: 10 = Total regular instances: 61 Syncopated instances: 46 (75.41% of all regular & unambiguous environments for -ede) Orthographic: 2 (4.35% of syncopated -ede endings) Before vowels: 12 (26.09%) Before h-: 14 (30.43%) Before consonants: 18 (39.13%) Unsyncopated instances: 15 (24.59% of all regular & unambiguous environments for -ede) Before vowels: 7 (46.67%) Before h-: 3 (20.00%) Before consonants: 5 (33.33%)

In figuring the percentage of syncopation and non-syncopation, I have not included the abnormal or ambiguous examples in the total field, since no absolutely conclusive information can be derived one way or another from them. Though the syllabicity of inflectional endings outweighs their synco-

“Ubbe dubbede him to knith”

201

pation by a ratio of almost exactly 2:1, the numbers make it clear that syncopation is by no means an infrequent or isolated phenomenon. Rather, it is one-third of the total instances of the regular and unambiguous inflectional environments and it occurs at similar rates before vowels and consonants, with h-+vowel environments lower only because such a specific category would necessarily be smaller. I have also noted the frequency of metrical promotion (here only 13.71%). As Smithers has already argued, the infrequency of promotion suggests a general correspondence in the poet’s language between “natural” or linguistic stress and metrical stress, adding to the likelihood that his frequent use of syncopation is also justified by such syncopation in spoken ME (Smithers, 1983: 200–202). In general, I think it is clear from this data – both the examples I’ve shown and the statistics – that in the early fourteenth century East Midlands dialect of the Havelok poet, the inflectional endings -es, and -ed(e) had begun to lose their syllabicity, though they could still be used, with “poetic license,” for metrical purposes in footed poetry, the way Chaucer uses final -e, or even as Shakespeare will continue to use, occasionally, a syllabic -ed.

Notes *

I wish to thank Donka Minkova, Dorothy Siegel, and my anonymous reviewers for their help and suggestions for revisions of this paper. I would also like to thank collectively my interlocutors at the SHEL 4 conference, as well as Susan Fitzmaurice and her staff for organizing and hosting it. 1. There is one remaining inflectional ending to be discussed: -eth. I have excluded -eth from this study for a few reasons. The -eth marker of third person singular verbs in the present tense appears only 23 times in Havelok, and its presence in a text with authorial provenance in Lincolnshire (see discussion below) suggests that these instances are scribal translations of -es, rather than belonging to the original. And since it is the meter of the original and what it tells us about schwa-loss with which I am concerned, I have left these few scribal instances of -eth aside. 2. Angus McIntosh finds that the Laud Misc. 108 Havelok was “written in a kind of Middle English which in all main essentials is characteristic of Norfolk,” and that the Cambridge fragment (CUL Additional 4407) is “dialectically (though not of course textually) very close” to the Laud version. Although McIntosh does “not insist that the scribe [of the Laud Havelok] was a Norfolk man,” he nevertheless finds it “highly probable” (McIntosh, 1976: 36). However, Smithers argues that the authorial dialect “can be localized with certainty

202

3.

4. 5. 6. 7. 8. 9.

10.

11.

12.

13. 14.

Christina M. Fitzgerald and in fairly specific terms” as an “EMidl. type, in a Northerly variety,” specifically of Lincolnshire, perhaps even Lincoln itself (Smithers, 1987: lxxxix). Moreover, Smithers adds that “the core of it is surely close to English speech of that area and time, not a bookish or literary product. There is no direct mention of a single literary work in the poem” (Smithers, 1987: lxxxix). In my assumptions about footed meter in late Middle English verse forms, particularly tetrameter, I follow Donka Minkova, who gives a succinct overview. (Minkova, 2005). This is also in line with Minkova’s findings (Minkova, 1991). Frequently spelled “casteles,” presumably by analogy to English plurals, e.g., l. 397, or “castles,” e.g., l. 1294. Note: “haueden” is ambiguous regarding which syllable is syncopated, but that doesn’t affect “demd” which is clearly one syllable. Instead of “preyede,” e.g., l. 211. Instead of “makede,” e.g., l. 542. Examples of each inflectional ending in the second environment can be found in the following lines. For -es: ll. 33, 39, 41, 152, 213, 239, 268, 303, 645, 770, 897, 1147, 1317, 1361, 1402, 1445, 1862, 1901, 1912, 1958, 1981, 1983, 1985, 2105, 2261, 2262, 2584, 2585, 2660, 2812, 2904, 2908, 2951. For -ed: ll. 23, 58, 336, 365, 381, 972, 2514, 2873. For -ede: ll. 71, 176, 350, 437, 680, 1112, 1314, 2162, 2365. Examples of each inflectional ending in the third environment can be found in the following lines. For -es: ll. 253, 266, 936, 940, 1903, 1904, 1987, 2132, 2133, 2242. For -ed: no examples in Environment 3. For -ede: ll. 90, 660, 771, 829, 884, 1386, 1407, 1675, 1708, 1753, 1917, 2266, 2413, 2525. Examples of each inflectional ending in the fourth environment can be found in the following lines. For -es: ll. 355, 688, 763, 982, 1168, 1394, 1401, 1923, 1990, 2028, 2056, 2121, 2356, 2530, 2684, 2799, 2926, 2976, 2981. For -ed: l. 2004. For -ede: ll. 134, 226, 231, 542, 1042, 1169, 1737, 1738, 1830, 1863, 2045, 2054, 2313, 2502, 2933. Examples of each inflectional ending in the fifth environment can be found in the following lines. For -es: (a) ll. 39, 360, 689, 821, 1173, 2280, 2611; (b) ll. 266, 785, 2240; (c) ll. 239, 430, 1339, 2416. For -ed: (a) ll. 434, 743; (b) no examples; (c) no examples. For -ede: (a) ll. 348, 1989, 2621; (b) no examples; (c) ll. 181, 658, 815. This line involves an exception: an inverted first foot, followed by unaccented syllable. Examples of each inflectional ending in the sixth environment can be found in the following lines. For -es: ll. 2, 33 (two examples), 41, 63, 86, 95, 136, 143, 215, 243, 258, 261, 262 (two examples), 263, 280, 282, 350, 359, 360, 366, 371, 383, 390, 397, 400, 407, 467, 538, 548, 572, 621, 656, 712, 717, 718, 781, 782, 866, 881, 883, 884, 891, 897, 915, 920, 957, 1025, 1069, 1093, 1124,

“Ubbe dubbede him to knith”

15.

16.

17. 18. 19.

20.

203

1141, 1167, 1197, 1217, 1232, 1233, 1235, 1239, 1268, 1284, 1294, 1295, 1298, 1304, 1316, 1344, 1366, 1393 (two examples), 1400, 1413, 1438, 1445, 1647, 1727, 1739, 1747, 1770, 1819, 1838, 1842, 1846, 1865, 1875, 1884, 1886, 1898, 1899, 1900, 1901, 1918, 1920, 1924, 1925, 1941, 1953, 1978, 2005, 2014, 2020, 2037, 2039, 2057, 2126, 2136, 2151, 2170, 2195 (two examples), 2196 (two examples), 2205, 2232, 2241, 2257, 2278 (two examples), 2301, 2325, 2329, 2331, 2332, 2345, 2367, 2371, 2395, 2415, 2419, 2431, 2438, 2441, 2442, 2462, 2478, 2485, 2490, 2582, 2584, 2585, 2596, 2597, 2606, 2612, 2617, 2665, 2666, 2682, 2692, 2698, 2707, 2741, 2782, 2789, 2811, 2813, 2817, 2835, 2284, 2894, 2899, 2908, 2915, 2917, 2980, 2984, 3000. For -ed: ll. 225, 424, 563, 715, 1176, 1241, 1267, 1369, 1373, 2275, 2450, 2481, 2929. For -ede: ll. 655, 709, 713, 1244, plus l. 2888, where the final -e in spusede gets secondary stress. Examples of each inflectional ending in the seventh environment can be found in the following lines. For -es: (a) ll. 2, 55, 137, 211, 375, 515, 645, 703, 711, 858, 862, 969, 971, 1063, 1116, 1268, 1294, 1329, 1344, 1379, 1383, 1400, 1631, 1749, 1847, 1901, 1911, 1918, 1968, 2101, 2127, 2163, 2303, 2444, 2466, 2602, 2614, 2802, 2940, 2981, 2983; (b) ll. 150, 153, 587, 1987, 2154, 2241, 2293, 2304, 2393, 2396, 2515, 2581, 2613, 2766, 2991. For -ed: (a) ll. 670, 674, 972, 993, 1430, 1435, 2145, 2240, 2440, 2932, 2963, plus 2668, where the -ed in shivered gets secondary stress; (b) ll. 40, 305. For -ede: (a) ll. 37, 932, 1001, 2267, 2439, 2747; (b) ll. 30, 1155, 2743. These ambiguous -es cases can be found in ll. 82, 239, 243, 244, 371, 397 (two examples), 782, 784, 1107, 1410, 1786, 1840, 2051, 2117, 2354, 2366, 2522, 2595, 2760, 2980. See ll. 82, 243, 371, 782, 1107, 1840, 2117, 2522, 2595, 2980. See ll. 239, 244, 397 (two examples), 784, 1410, 1786, 2051, 2354, 2366, 2760. The lines in subcategory (a) are: ll. 30, 35, 67, 98, 211, 260, 343, 349, 382, 403, 420, 439, 482, 503, 545, 551, 569, 573, 602, 634, 650, 707, 729, 776, 836, 902, 930, 931, 962, 964, 979, 1166, 1181, 1256, 1276, 1282, 1428, 1650, 1711, 1928, 1995, 2001, 2058, 2176, 2190, 2228, 2284, 2315, 2358, 2556, 2638, 2676, 2755, 2766, 2771, 2793, 2857, 2892, 2907, 2970. In subcategory (b), the lines are: 899, 1168, 1429, 1431, 1644, 2354. The lines affected are: ll. 138, 261, 266, 267, 761, 780 (two examples), 806, 898, 1016, 1329, 2013, 2067, 2089, 2092, 2105, 2117, 2195, 2196, 2351, 2362, 2372, 2466. The ten words are: bar(o)uns, bedels, sergeans/serganz/seriaunz, paniers, wastels, simenels, laumprees, chaunpiouns, burgeys, and glotuns. Although “castels” is also a French loan, because it frequently appears as “casteles” and “castles” in Havelok, I have treated it as an -es word.

204

Christina M. Fitzgerald

References Lass, Roger 1992 Phonology and Morphology. In Cambridge History of the English Language, Vol. 2. Norman Blake (ed.), 23–156. Cambridge, England: Cambridge University Press. McIntosh, Angus 1976 The Language of the Extant Versions of Havelok the Dane. Medium Aevum 45.1: 36–49. Minkova, Donka 1991 The History of Final Vowels in English: The Sound of Muting. Berlin: Mouton de Gruyter. 2005 Chaucer’s language: Pronunciation, morphology, metre. In Chaucer: An Oxford Guide. Steve Ellis (ed.), 149–155. Oxford: Oxford University Press. Mossé, Fernand 1952, A Handbook of Middle English, James A. Walker, trans. Baltimore: 1968 Johns Hopkins University Press. Smithers, G.V. 1983 The scansion of Havelok and the use of ME -en and -e in Havelok and Chaucer. In Middle English Studies Presented to Norman Davis in Honour of His Seventieth Birthday. Douglas Gray and E. G. Stanley, (eds.), 195–234. Oxford: Clarendon. Smithers, G.V. (ed.) 1987 Havelok. Oxford: Oxford University Press. Zink, Gaston 1989 Morphologie du francais médiéval. Paris: Presses Universitaires de France.

A response to Tom Cable

Christina M. Fitzgerald I am grateful to Tom Cable for the careful reading he has given my article and for the helpful comments he has provided both in his introduction presented here and in correspondence with editor Donka Minkova. While recognizing that my article makes valuable contributions to the study of English historical metrics, Cable is critical of my selection of analytical categories. The general complaint – that I follow Smithers’ categories too slavishly, given that the objects of our respective studies are not the same – is perhaps fair. In correspondence not reproduced here, Cable further elucidates the problem by noting that while Smithers uses these categories to show where elision (of -e and -en) does not occur, I use them to show where syncope does occur. This difference could be potentially confusing to a reader who knows Smithers’ article. I use these categories because I found the linguistic environments in which the endings -es, -ed, and -ede appear important to the analysis of their behavior, as I discuss below. And further, these categories as a group are useful for sorting, finding, and analyzing the linguistic information that my study seeks to uncover. Cable’s position that the categories are “irrelevant for the loss of syllabicity of -es” is unnecessarily dismissive. Although syncopation and not elision is the operating phenomenon here, the environment for the syncopation of -es is still important. One expects a phrase such as “strikes it” and “strikes him” to be disyllabic more often than “strikes them” since the -s may re-syllabify before either a vowel or an h-, leaving the -e- open to apocopation in the newly formed stri-ke-sit or stri-ke-s(h)im (thus strik-sit or strik-s(h)im). But in “strikes them,” an <sth> consonant cluster (*stri-kesthem) could be an illegal onset, so the three syllables (stri-kes-them) are more likely at first to remain intact.1 The effect of the environment should therefore not be ruled out, and is confirmed by my finding that only onethird of the syncopations of -es occur before consonants other than initial h-, while the remainder occur before vowels and h-. At the same time that Cable finds that I provide “too many environments” for syncopation, he also charges me with making “too few distinc-

206

Christina M. Fitzgerald

tions of grammar” and needing “more and finer-grained morphological distinctions.” This is perhaps a fair complaint, though I originally decided against such finer distinctions precisely because of a proliferation of categories. Clearly there is more work to be done on this issue, and I invite other scholars to pick up where I have left off. Does genitive -es behave differently from plural -es? And in the case of -ed(e), what is the impact of the weak verbs’ historical classes and sub-classes and their development from OE to ME on the behavior of ME -ed(e)? These questions could perhaps be answered by further work with a wider range of evidence from more sources than Havelok alone.

Notes 1. I owe this point to Donka Minkova, personal communication.

Patterns and productivity

David Denison Introduction All three morphosyntax papers in this volume can be seen as attempts to tackle the following, perennial puzzle in historical linguistics – indeed in all linguistics: what patterns are speakers using as the basis of production of new words and sentences? The pre-theoretical term pattern conveniently covers all sorts of rule-based behaviour, subregularity and partial regularity. Chris Palmer, Don Chapman and Graeme Trousdale have gone about this puzzle in very different ways, and the following musings are prompted by their research. In response to a first version of this paper, Palmer revised his own contribution, Chapman submitted some follow-up comments, and Trousdale did both, and I have adjusted my contribution accordingly. First let me summarise the papers. Palmer investigates the naturalisation of abstract noun-forming suffixes in late Middle English in two sets of records kept by trade communities. He devises measures of ‘local productivity’, including a quantitative measure of aggregation of new types (i.e. words) introduced in successive subperiods, while various formal and textual clues are weighed up as qualitative evidence of naturalisation. Chapman investigates four patterns of addition of the suffix -er to a phrasal or prepositional verb to form a noun, three of them currently productive: pickerup, pick-upper and picker-upper. Historical evidence comes largely from the OED citation database, current data from the internet. Trousdale discusses the gradual loss of the OE impersonal construction as it falls victim to the transitive construction. The framework is Construction Grammar, and there is reference to Roschian levels of categorisation and to grammaticalisation not of lexemes but of constructions.

208

David Denison

Palmer on productivity and naturalisation Palmer chooses to make a detailed textual investigation of two relatively small and undigitised corpora, each representing the language of a professional community over a period of century or so, though only three decades’ worth are actually used. He distinguishes productivity from naturalisation for the possible case where an affix becomes productive in English with Latinate bases only. The measure of productivity of an affix preferred by Baayen (1992) is to count hapaxes, but Palmer shows that this is inappropriate: no available historical corpus constructed on a principled basis is big enough. Instead Palmer advocates the detailed study of even smaller corpora, making virtues of what might seem drawbacks: their small size, which means they can be read rather than having to be searched electronically, and their multilingual nature. Palmer offers counts of tokens for eight nominalising suffixes in each corpus, but in the event, not much is made of token counts. He turns to counts of types, and better yet, the aggregation of types over five-year periods. His notion of local productivity attempts to address the impact of frequency on derivational change. Hybrid formations (native base + foreign suffix or vice versa) turn out to be too infrequent to be of statistical use – and they had been expected to provide the main evidence for naturalisation. What is more, since hybrids with a foreign suffix are vanishingly rare, the limited evidence of hybrids suggests that the native affixes are more productive than the foreign, a result directly opposite to that suggested by the aggregation measure. With numbers everywhere so small, only a limited amount of light has been shed on productivity and naturalisation in these corpora, despite the careful and methodical approach. However, what conclusions there are can be supplemented in an interesting way, as close reading of textual juxtapositions allows Palmer to make deductions about the independent status of certain affixes for speakers and hearers in those communities. We have, then, the beginnings of a fine-grained sociolinguistic account of word formation among a fairly small group of people in one place over a limited time.1 Palmer cites Cowie (2000) as demonstrating that ‘intentionality clearly did not preclude productivity’. In fact both morphological papers in this volume mention the slightly self-conscious aspect of some kinds of word formation – indeed the possible playfulness – which doesn’t apply so readily to most other domains of language. Earlier Palmer had drawn attention to an idiosyncratic recent formation with -age, ownage, and the entirely

Patterns and productivity

209

playful tippage. I too can attest some playful new -age activity in the speech of student informants: (1)

chattage, faffage, foodage, liftage, peppage, sleepage, tun(e)age, workage, etc.

The words have an activity meaning and are often used with social peers in some such context as There was some general Xage. This creativity plays on the deliberate choice of the ‘wrong’ affix, yet used in the ‘right’ way insofar as attached to a suitable base. Consider OED’s definition of that suffix, parts of which are quoted below: -age suffix of abstr. nouns, originally in words adopted from Fr., afterwards a living Eng. formative. […] 1. From names of things, indicating that which belongs to or is functionally related to, […] passing into the whole functional apparatus collectively […] 2. […] 3. From verbs expressing action […] All of the coinages I have listed are on the face of it possible words of English, formed by addition of -age either to a noun (sense 1) or to an action verb (sense 3) or to a stem which could be either.2 Of course these nonce usages would normally be blocked by other formations on these bases which are already established, for example (2)

chat/chatting/chatter, faffing (about),3 feeding/food, lift(ing), pepping (up), sleep, work(ing)

The deliberate choice of -age in (1) is a social phenomenon, and yet it is possible that the special marking one can detect at present (such as in-group usage, playful style, etc.) could become lost over time, with some of the words entering the general vocabulary. My British informants employ it in a merely playful way, but I am told that a similar fashion for -age coinages was confined in one New England university in the late 1980s to drugrelated contexts. If the current British student usage derives from that American one (and of course it may well be an independent development), then already some generalisation has taken place. It is probably quite com-

210

David Denison

mon for an individual coinage or a coining pattern to spread outwards through social networks, accompanied in some instances by generalisation of meaning or of pragmatic context.4 It should be possible to find and document a present-day analogue for Palmer’s discussion of medieval word formation tendencies within a limited milieu. And the importance of register is clear: Bauer’s monograph (2001) reveals no awareness of the possibility that morphological categories might be more productive in some registers than in others, and the potential consequences of such stylistic forces for the weight of structural constraints in explanations of productivity. However, the little work that has been done in this area shows unambiguously that, unsurprisingly, different genres recruit different morphological categories to very different degrees. (R. H. Baayen 2006: 20)

Chapman on nominalisations Chapman is investigating a phenomenon barely attested in conventional corpora, so that searches must be conducted in improvised corpora, bringing in their wake all sorts of questions of reliability, with particular difficulties for frequency counts. Here the problem is not so much smallness of sample as uncountableness – certainly of the overall size of the sample, but even of the number of examples found as well. Chapman discusses some conventional measures of productivity, including hapax legomena, type or token frequency, first occurrences and elicitation tests. (For a fuller range of options, see Baayen (2006).) None seem to be wholly practicable in this case. For a rough-and-ready historical source of data, Chapman uses the quotation database of the OED, stating that the picker-upper pattern occurs mostly in the 1940s and later (though the online version of OED has since been updated with citations of picker-upper from 1913 and 1937, and of fixer-upper from 1932).5 We can supplement this from the Google News Archive, which gives examples of builder-upper, dragger-downer from 1931; lifter-upper, finder-outer, fixer-upper from 1934; helper-outer, picker-outer, waker-upper, warmer-upper from 1935; pepper-upper, stepper-upper from 1936; giver-upper, locker-upper, looker-oner, tearerdowner from 1937; opener upper from 1938. It does look as if the pattern enters American newspapers in some force in the 1930s.6

Patterns and productivity

211

For present-day usage, Chapman conducts Google searches of the internet, ingeniously semi-automating the choice of search terms by gathering contiguous verb-particle sequences from the (tagged) British National Corpus (BNC) and then trying all four kinds of -er formation for each pair found. Regrettably, as he says, token counts on such material are not practicable. Now the historical section of the paper had explicitly included both phrasal and prepositional verbs (e.g. pick up and look at, respectively), which the lists of forms in his Appendix 1 confirm. For the contemporary investigation it was not clear what tags were searched for in BNC. The adverbial particle of a phrasal verb is classified in BNC as AVP, the prepositional particle of a prepositional verb as PRP, which is the general tag for a preposition (and an ‘ambiguity tag’ like AVP-PRP can be found when the tagger fails to give a clear analysis). Apart from a handful of on and over pairs which could be either, the vast majority of forms in his Appendix 2 are self-evidently phrasal verbs. Don Chapman has confirmed that his original internet search was effectively limited to phrasal verbs, that being the construction readily picked out by the BNC tagging, but he has since examined some verb-preposition combinations too; see the Appendix below. I mention here a handful of minor quibbles with Chapman’s paper. The opening paragraphs are a little unclear as to whether the -er suffix under investigation is always agentive, though Chapman’s later discussion of Ryder (2000) includes many non-agentive uses – Chapman confirms that all senses were included. It is by no means just in and on which can appear in both phrasal and prepositional verbs, as implied in the paper: in fact all of the particles listed as occurring in prepositional verbs – about, along, around, round, and through – can appear in phrasal verbs too. On this point see now the Appendix below. Incidentally, double marking can be attested in nominalisations involving over: (3)

(4)

Situation: one person runs by and shows [sic; read shoves DD] another person such that the second person knocks into an elderly person and causes this person to suffer a broken hip. The “shover” keeps running is not caught, but the “knocker-overer” is caught. (http://gfp.typepad.com/the_garden_of_forking_pat/2004/10/ hard_determinis.html, 22 Apr. 2007) That’s right, I’m a motorcycle-knocker-overer. (http://www.poundy.com/2006/03/31/thats-right-im-a-motorcycleknocker-overer/, 22 Apr. 2007)

212

David Denison

Chapman’s note 11 had suggested to me that he thought overer impossible, though in fact he had meant only that it was less frequent than over. Having presented the historical and contemporary evidence he has gathered, Chapman speculates on the historical development, suggesting that the picker-upper pattern may be a transitional stage on the way from picker-up to pick-upper. That is certainly possible, but I want to canvass another option which invokes phonological factors. For legitimate phrasal verbs in typical sentence patterns, the sentential stress distribution between verb and particle is roughly equal, though clause-final position as in (6)b, a focus position, tends to elevate the stress on the particle. (5) (6)

Jim was picking up his daughter. a. Jim can pick up the beer. b. Jim can pick the beer up.

Both verb and particle have primary word stress. As far as the formation of phrasal verbs is concerned, a rhythmic template has developed which demands (roughly speaking) that the verbal formative be monosyllabic or an initial-stressed disyllable (Fraser 1976: 13, 1966, Kennedy 1920: 29), hence the expected absence of annoy off, resuscitate back.7 The phonological constraint on the form of the verb has not always played such a large part in phrasal verb formation: I have shown that it was absent or at least very much weaker in the fifteenth-century Paston Letters and in my own Corpus of Late Eighteenth-Century Prose (Denison 1981: 148, 2007: 124), so it is probably quite a recent development. In fact one can plausibly argue that phrasal verbs in present-day English are coming to represent a construction with its own syntactic, lexical, morphological, pragmatic, stylistic and phonological effects, but it is the rhythmic factor I wish to stress here. It does not seem fanciful to suppose that a similar rhythmic template might be developing in the derived agent nouns, first disfavouring uppicker and now increasingly pick-upper too, both of which force primary stress onto the particle (and incur stress-clash too).8 Then picker-up could come to the fore as a nominal formation with a similar distribution of stress between deverbal stem and particle as is found in the source phrasal verb. The current tendency to supplant even this formation with the doublymarked pattern, picker-upper, would then satisfy a new rhythmic template, one in which the relative weights of verb stem and particle are respected as

Patterns and productivity

213

well as the need to have -er as final element of an agent noun (which is both a rhythmic and a formal matter). Evidence in support of the rhythmic template comes from OED’s very first citation for picker-upper: (7)

1913 Chicago Sunday Tribune 30 Nov. VI. 6/1 arounder … there is a busy little picker-upper.

For every fling-

Notice that there is no need for double marking on the parallel flingarounder, since an unstressed dip between verb-stem and particle is effectively provided by the first syllable of around.9 Of the two possible replacements for picker-up, on this analysis pickerupper is more comfortable rhythmically than pick-upper, even with the apparent redundancy of double marking. Notice too that the alternating ´×´× does not incur the stress-clash of ´´× .10 Whether these rhythmic factors will be enough to ensure that picker-upper wins out, who can say? One indication that it might, is that beside that doubly-marked construction, there is even a form with triple marking, not mentioned by Chapman. There are a modest number of attestations on the internet11 compared to the hundreds of thousands for picker-upper, for example: (8)

Its employee, Mr Cockburn, was driving a tractor fitted with a ‘plastic picker upperer’ implement, which rolled up plastic that had been laid in rows in a … (http://www.findlaw.com.au/article/13040.htm, sampled 18 Apr. 2007) (9) yes i agree – complain, my dad works as a pool fixerupperer and tell them to put more acid solution/ph solution i forget which one in to balance it out asap. (http://malaysia.answers.yahoo.com/question/index?qid=200702052 13726AAqlQjN, sampled 21 Apr. 2007) (10) I nominate me for late catcher-onerer of the year! (http://www.vbforums.com/archive/index.php/t-193614.html, sampled 22 Apr. 2007) This too must be factored in. It seems to me that such usages are simultaneously quite natural (since I can and do use them myself!) and yet somewhat self-conscious, in that they so obviously violate norms of usage involving horror aequi or avoidance of haplology (Rohdenburg 2003) – and note the scare quotes in (8).

214

David Denison

In fact Bauer states flatly that ‘in English no suffix can be added to a base that already ends in the same suffix’ (1983: 92). Bauer’s claim seems on the face of it reasonable, so long as ‘the same’ is interpreted strictly to cover one form in one function. Thus in the selection of examples in (11) and (12) it is perfectly possible to add agentive or comparative -er to bases which happen to end in a non-morphemic -er,12 but an -er comparative is not usually possible for an adjective which already contains a final -er which is etymologically comparative, even though the possibility of a superlative -est proves that the adjective is not synchronically comparative, as shown in (13): (11) blabberer, botherer, chatterer, offerer, sufferer, wanderer (12) bitterer, cleverer, ?eagerer (13) *innerer, innerest, *outerer, outerest, *utterer (as adjective), utterest But the double termination of (8)–(10) shows that the claim cannot be upheld as an absolute block. Why add yet another -er on the end when a form has two already? And if that is possible, why not double up the final -er on pick-upper? – as far as I know, pick-upperer is very rare indeed (one hit for that particular form on Google). I have argued that a rhythmic need for an unstressed dip in second position has been gathering strength (which despite my analysis of flingarounder in (7) does not preclude double marking in the attested messerabouter, mucker-arounder, where the dip covers two syllables). Absence of a dip will then help explain the rarity of pick-upperer. Rhythm doesn’t exclude picker-upperer, but it doesn’t explain it either. One can only speculate. If a speaker is already intent on producing a doubly-marked form like picker-upper, does the occasional final addition of yet another -er syllable happen unconsciously through some kind of involuntary articulatory reduplication, or (semi-)consciously from a desire to distinguish a noun formative from a homophonous comparative adjective (upper, inner, outer, rounder)? Chapman’s paper raises the fascinating problem of what speakers do when faced with two potentially contradictory pulls.13 Sometimes one or other has to be favoured, as with the competing demands for clause-initial position of negative imperative don’t and first person imperative let’s (Denison 1998: 253–4). Sometimes neither can be satisfied, as argued in Hudson’s explanation for the absence of amn’t from most varieties of English:

Patterns and productivity

215

The word concerned needs to be an example of the negative of the present tense of BE, which (by default) is aren’t; but it also needs to be an example of the first-person of the present tense of BE, which is am. It cannot have both of these forms at once, so the conflict must be resolved; but we cannot resolve it in the normal way, by giving priority to the more specific alternative, because neither pattern is more specific than the other. Therefore the conflict remains unresolved, we don’t know how to pronounce (or write) the word, and we can’t use it. (Hudson 2000: 298)

Sometimes the outcome is one of evasion, as apparently in the favouring of never as negator for used to, avoiding an awkward choice between didn’t use(d) and use(d)n’t (Denison 1998: 197). Sometimes it involves trying to satisfy both desiderata simultaneously, as with the sixteenth- and seventeenth-century not say pattern, interpretable as a compromise between the older pattern of negation without an auxiliary, say not, and the incoming pattern of negator before the lexical verb, do not say (Ukaji 1992). It is this last possibility that seems most relevant here: that is, it is plausible to regard double marking with -er as a compromise between opposing demands. (Double -er does not seem explicable as emphatic reinforcement, as some kinds of linguistic doubling certainly are.)

Trousdale on impersonals Trousdale’s paper is rather different from the other two under consideration, being on syntax and on loss – of the OE and ME impersonals. Since the loss is attributed to the strengthening of another construction, again we have a choice between rival patterns and the need to explain the way that choice plays out. Trousdale presents a partial Construction Grammar analysis in which only four actual clause types are listed for Old English. What they have in common, as is eventually revealed, is that all involve precisely two argument NPs and are therefore, presumably, to be regarded as a reasonably self-contained subsystem of the grammar of OE. In the hope of clarifying the nomenclature, I give here a simplified version of Trousdale’s Figure 3:

216

David Denison

(14)

TrnCxn

ExpCxn

Type N

Type T

Type I

Type II

The overarching two-NP type is the Transitive Construction, TrnCxn. This abstract schema has just two subtypes, ExpCxn (unlabelled in Trousdale’s diagram, but referred to subsequently) and Type T. ExpCxn is the Experiencer Construction. It in turn sanctions three subtypes usually involving impersonal verbs. One of them is the most prototypical exemplar of ExpCxn – hence the emboldened box around it – namely Type N, the (true) impersonal construction; the others are extensions of Type N, namely Type I and Type II. To illustrate the four types I use Trousdale’s examples (5) and (7) as the basis of (15) – (18) below: (15) Type N Case of Experiencer: Dative or Accusative Case of Source: Genitive him ofhreow þæs mannes 3SM-dat pity-3SPast the-gen man-gen to-him pitied because-of-the-man ‘He pitied the man’ Or: ’The man caused pity in him’ (ÆCHom I XIII.281.12) (16) Type I Case of Experiencer: Dative Case of Source: Nominative him ne ofhreow na þæs deofles hryre 3SM-dat not pity-3SPast not the-gen devil-gen fall to-him not pitied not the devil’s fall ‘He did not pity the devil’s fall’ Or: ‘The devil’s fall did not bring about pity in him’ (ÆCHom I XIII.281.14)

Patterns and productivity

217

(17) Type II Case of Experiencer: Nominative Case of Source: Genitive ofhreow se mæssepreost þæs mannes the-nom priest-nom the-gen man-gen pity-3SPast the priest because-of-the-man pitied The priest pitied the man Or: ‘The priest felt pity because of the man’ (COE) ÆLS (Oswald 262) (18) Type T Case of Agent: Nominative Case of Patient: Accusative acwealde þone dracan he 3MS-nom kill-1/3SPast the-acc dragon-acc ‘he killed the dragon’ (Ælfric Homilies (Supp.), XXI, 455) The ‘impersonal construction’ is referred to by Trousdale but is not represented on any of his diagrams. The box which I have chosen to label as ‘ExpCxn’ in my (14) is not co-extensive with the impersonal construction, since some of its instances are not impersonal (see Trousdale’s (4) = my (19) below). I take it, therefore, that ‘impersonal construction’ refers generally to Type N, the prototypical instantiation of ExpCxn. Having long ago looked at OE impersonal and non-impersonal patterns myself (Denison 1990), where I considered relating them by means of Quirk’s serial relationship, I later suggested that ‘[t]hese ideas might now perhaps be reframed in terms of Prototype Theory or Grammatical Construction Theory’ (Denison 1993: 96)! Since precisely those two approaches lie at the heart of Trousdale’s paper, I welcome the opportunity to see how well they fare. I have a number of comments, some of them more in the way of suggestions for future work than criticisms of what is still only an indicative sketch. Trousdale apparently regards the historical change in the morphosyntax of case marking not as a contributory cause of the decline of the impersonal construction but as a symptom of the grammaticalisation of the transitive. I think it fair to say that no explanation is given for why the impersonal construction should have declined when it did: all that is claimed is ‘the gradual disfavouring of a set of constructional subtypes’, which is merely an alternative way of describing the change, and that the decline is ‘associated’ with the grammaticalisation of TrnCxn. There is some discussion of

218

David Denison

how patterns could become reclassified, with an interesting comment on ‘the ambiguous – and sometimes non-existent – force dynamics of the processes associated with impersonal constructions’ as explanation for alternative outcomes with different verbs. The value of the paper lies in new insight not so much into the impersonal as the history of transitives. If the whole process can be related to grammaticalisation, then grammaticalisation and Construction Grammar are shown to interact fruitfully and each receives a measure of support by its successful application to a new set of data. A fuller account on these lines will need to resolve a number of questions. One is the role of word order. It is not made clear whether Figures 3 and 4 in Trousdale’s paper specify the word order of constructions. All constructions are represented by formulas of the type NP-V-NP, but the definitions seem to be semantic and morphological: ‘one of the arguments is the experiencer of some sort of psychological state or process’ (Type N); ‘the NP functioning as patient is case-marked as accusative. The verb denotes an action involving the transfer of physical energy’ (Type T). Trousdale’s examples (4) (OE) and (12) (eModE), reproduced below, are both apparently Type I: (19) hu him se sige gelicade (20) […] to gete that thee and thyne behoueth The order of both is NP-NP-V, which implies that construction types merely stipulate argument structure.14 Yet a crucial element in Allen’s (1995) analysis, cited by Trousdale, concerns Experiencers that are preposed. To what extent, then, is word order criterial for the construction type(s) under discussion here? Which level is basic at a given period? In ModE apparently the TrnCxn is increasingly a superordinate category, showing ‘mutual distinctiveness with respect to other superordinate categories’ like the copula construction: ‘while the copula allows non-NP subjects (Over there is fine), few verbs [note omitted] in the TrnCxn do’. In his first version Trousdale had written ‘no verbs in the TrnCxn do’, in response to which I offered (21): (21) a. Over there suits me fine. b. Travelling slowly always irritates me. c. That the sun seems to move doesn’t disprove Copernicus’s theory.

Patterns and productivity

219

His note 9 now addresses my quibble. More to the point, if the TrnCxn has been becoming more schematic, are we to assume that it was (more of) a basic level category in OE or merely more substantive? Presumably the latter, since we are told that the basic level categories of OE were ExpCxn and Type T, not TrnCxn. There seems to be some unclarity as to which transitive construction is being followed: ‘By Modern English, the range of distinctive constructional types has reduced, and the TrnCxn licenses an increasingly larger [sic] range of tokens’ – but TrnCxn is also defined for OE. And why is the overarching two-NP construction of OE called TrnCxn, when one of its two apparently equal subtypes – neither has an emboldened box to suggest prototypicality – is not transitive or agentive at all? That seems to be anticipating the diachronic outcome which is presented for post-OE times. Another problem concerns the relationship between prototypical subtypes and extensions. In OE it is said to be ExpCxn which has both kinds, whereas Type T has only the former. In ModE it is TrnCxn which has developed extensions. These claims look on the face of it rather simplistic. Only one solitary instance of Type T is given for OE, and that is described as prototypically transitive. However, Type T has variants in OE too, and it would be interesting to know whether these would be described as nonprototypical extensions, or what. Here is one (many more could be given): (22) Gif he geeuenlæcð deofle on manlicum dædum (ÆCHom II, 13 129.74) if he imitates Devil[DAT] in sinful deeds The NP deofle in (22) is only glancingly a Patient, and it is not marked accusative, while the action is not telic or punctual (though it is volitional and actional). Trousdale mentions the fact that verbs like lufian ‘love’ are not covered by his classification. We could muddy the waters much more, unfortunately. How are we to classify common examples like the following? (23) þa hæþenan … mid anum swencge slogon him of þæt heafod. (ÆLS (Edmund) 123) the heathens … with one blow struck him[DAT] off the head[ACC] Example (23) is about as prototypically transitive as can be, yet there is a dative, him, which – if not analysed as a possessive within the Patient NP – may be called a dative of disadvantage, which is closely akin to an Experi-

220

David Denison

encer. Once again we need to know how this will be classified. In my discussion of the difficulty of separating impersonals from non-impersonals in OE (Denison 1993: 93–6), I drew attention to such facts as lystan ‘desire’ (impersonal), gitsian ‘covet’ and friclan ‘desire’ (non-impersonal) all taking a genitive of the Source object; sceamian ‘shame’ (impersonal) sharing a genitive Source object by coordination reduction with fægnian ‘rejoice in’ (non-impersonal); likewise egl(i)an ‘ail’ (impersonal) sharing a dative Experiencer/Patient object with derian ‘harm, injure’ (non-impersonal). What is the significance of frequency? Trousdale cites Thompson & Hopper as showing that ‘clauses which are canonically “transitive” – i.e. telic, punctual, volitional on the part of the agent, and so on – are surprisingly infrequent in naturally-occurring conversation’ in American English and cross-linguistically too, as are two-NP clauses generally. How does this bear on the story being told about the history of English? It seems to imply that Type T should be infrequent in OE, and we are told that Type N was ‘already rare in Old English’. Between them they are the only two-NP constructions listed. If frequencies are so low, how does this relate to the claim that extensions tend to involve high type-frequency constructions? Some numbers would at least put the offered fragment of grammar in context. Furthermore, are two-NP constructions really a self-contained domain? Some impersonal verbs can occur in one-place as well as two-place use: (24) Gyt me twynað (ÆCHom I, 4 72.30) yet me is-in-doubt The same goes for transitive verbs being used intransitively, though it is probably true that OE makes a clearer separation than does ModE between transitive and intransitive verbs. It has even been suggested that some impersonals may occur in three-place use (Denison 1993: 73) A possible reaction to the difficulties of examples like (22)–(24) is to allow multiple inheritance (Goldberg 1995: 97-8) from more than one superordinate construction, rather than always having a given construction sanctioned by a single superordinate construction. Although Croft (2001) does not seem to comment explicitly on this, I take it, for instance, that an impersonal clause in the interrogative would have to be sanctioned by both ExpCxn and an Interrogative Construction: (25) Hu þyncð eow nu, cwæð Orosius, … ? (Or 8.52.15) how seems you[DAT] now, said Orosius, …

Patterns and productivity

221

However, introducing multiple inheritance could necessitate many changes to the grammaticalisation story. In any case a fuller diachronic account may not seem so monotonic as the ‘gradual disfavouring’ of impersonal constructions offered here, since some verbs only developed impersonal uses in ME before losing them again (Denison 1993: 71–2 and references given there). Would that be linked to a partial grammaticalisation too? What I would advocate as a way of fleshing out this account is to use the Penn Parsed Corpora of OE, ME and eModE to make some estimates of the relative proportions of the different kinds of constructions in different periods, to see whether the transitive construction does indeed expand at the expense of ExpCxn in the way suggested. As to self-conscious creativity, discussed above in relation to wordformation, this is not so easy to find in the realm of syntax. Or rather, it is very easy indeed to find in the style of an individual writer – for example, the zeugma of ‘He left in high dudgeon and a hansom cab’ – but less easy to identify as a habit shared by a social network and spreading to the population at large. An often-cited example is final Not! added after a pause, sarcastically negating what precedes, at first a catchphrase of the show ‘Saturday Night Live’ (OED s.v. not adv. n. and int. C). A more general case might be the early diffusion of the progressive passive, although the sociolinguistic account of its origin does not insist on conscious awareness of its novelty on the part of early users (Pratt & Denison 2000). In Construction Grammar there is no principled difference between lexemes and syntactic constructions, so better examples might perhaps be adduced.

Category strengthening Conscious creativity is an intriguing but probably minor part of language change overall. If we return to the more general points under discussion in these three papers, what emerges forcefully as a common theme is the notion of category strengthening, a sort of snowball effect or positive feedback whereby identification of a pattern by hearers leads to its productive use by speakers, which in turn further strengthens the pattern, and so on. Yet of course it does not always happen like that: either the snowball never starts rolling at all, or it may take off and then come to a halt or even (to push the metaphor to its limit) melt away. And remember too that none of the morphological and syntactic phenomena discussed in these papers are of particularly high frequency – if anything, the contrary. For the idea of

222

David Denison

category strengthening to have real explanatory value and not merely post hoc plausibility, we will need to know how to measure the ‘strength’ of a category and also what other factors can interfere with the process of strengthening.15

Evidence and heuristics Finally, I consider the three papers in relation to the theme that runs through this volume. The choice of evidence in Palmer’s paper is an interesting one, out of tune with the current predisposition towards large evidential databases and computerised searching and counting, though somewhat akin to recent historical research on social networks, on which see for example Tieken-Boon van Ostade (2000) and Fitzmaurice (2007). Given the significance of register, given the unevenness of available records in late Middle English, and given Palmer’s concern with developments within specific communities, the choice of corpus is understandable, and the data in turn impose limitations on the analytic methods that can be applied. Chapman tries the conventional historical corpora in the ICAME collection and finds little of use: the nominalisations he is interested in are probably too colloquial and ephemeral to figure much in such corpora. His mining of the OED works quite well, at least as an indicator, and as we have seen, other historical collections are coming online in ever-growing size and numbers. For current usage the internet is the obvious source to turn to, despite the evident drawbacks of such an uncontrolled resource; useful references here include Keller, Lapata & Ourioupina (2002), Hundt, Nesselhauf & Biewer (2007), Rosenbach (2007). As a heuristic, the device of using the tagging of the BNC to help generate search strings for internet searches is ingenious, and the results are useful. As for triple marking by -er, who could have predicted such a bizarre form? You have to have the luck to stumble upon it, or to use it yourself. Trousdale’s paper is more programmatic than the other two I have discussed, and the direct evidence in it is limited, the data being borrowed from the work of others. What Trousdale is working towards has more to do with types of explanation: as far as data is concerned, aiming to relate the loss of one pattern (impersonals) to the rise of another (transitives), and in theoretical terms, marrying one approach (grammaticalisation, diachronic) with another (Construction Grammar, synchronic). These too are heuristics, and commendable ones, but at a more abstract level.

Patterns and productivity

223

Appendix: On the response by Don Chapman I have made some adjustments to my discussion above of Chapman’s paper in the light of his (unpublished) response to my first version. In three cases, though, he made rather detailed comments and even introduced some new findings which would not fit comfortably in the body of my paper. I reproduce them here with brief comment. I had noted the discrepancy between the historical survey, which covered both phrasal and prepositional verbs, and the internet data. Chapman responds: As prepositional verbs are notoriously difficult to identify – in many ways they constitute more a continuum of idiomaticity than a discrete category – I did not try to sort them out. For this response, I have examined a sample of fifteen such verb-preposition combinations: Table 1. Nominalisations of prepositional verbs (Chapman)

look after aim at look at account for ask for look for pay for wait for benefit from depend on apply to belong to agree with deal with work with

picker-up pattern x x x x x x x x x x x x x x x

picker-upper pattern x

pick-upper pattern x

x

x

x x x x

x

uppicker pattern x

x x

x x x x

x

As Chapman goes on to comment: Several prepositions are examined in this sample that were not included in Appendix 2, namely after, at, for, from, to, and with. As the table shows, the picker-up pattern is the most used. Space limits further analysis.

224

David Denison

I have been able to confirm with Google the existence of most, if not every one, of the combinations checked in Chapman’s Table 1. In response to my observation that in most cases the particles16 that occur in prepositional verbs can also occur in phrasal verbs, he writes: But if one speaks of tendencies rather than categories, the division remains strong. O’Dowd (1998) examined the percentage of time that certain particles/prepositions were used in phrasal verbs versus prepositional verbs. When these percentages are added to the distributions of the particles/ prepositions among the three nominalization patterns, the correspondences are striking. There is a strong tendency for those particles/prepositions that occur more typically in phrasal verbs to be more typical of the picker-upper pattern and those that are more typical of prepositional verbs to be typical of the picker-up pattern. Table 2. Correlation between nominalisation and use in verbal patterns (Chapman) particle / prepositions about along around back down in off on out over round through up

% used in % used in prepositional phrasal verbs verbs 0% 97%

–

–

66%

34%

–

–

94% 18% 79% 15% 98% 73%

4% 81% 21% 83% 1% 24%

–

–

31% 98%

65% 2%

picker-up pattern 4 1 2 7 18 16 12 13 31 8 4 2 33

pickerupper pattern 0 0 0 22 37 19 20 11 74 1 2 0 84

pick-upper pattern 0 0 0 6 21 10 10 3 23 0 2 0 67

It is indeed interesting to find that agent nominalisations of prepositional verbs behave differently from those of phrasal verbs. Finally in his response, Chapman agrees that the -er formations are infrequent. ‘But why that should be so calls for explanation, since the input (multi-word verbs) and the process (-er nominalization) are both very fre-

Patterns and productivity

225

quent.’ Perhaps the distinct nature of multi-word verbs plays a part. He checked on Google and found affixes other than -er (tense suffixes, past participle marking and -ing) repeated or misplaced on multi-word verbs. He offers these examples with pick up: The pacing […] picks ups about a third of the way into the book; he pick ups whatever the trouble is; rental items pick-upped or returned; has to be picked upped and turned on its side; delayed from pick upping Mel; a lot of the picking-upping. These forms do not appear to be as frequent as the -er forms, but they still show that the picker-upper phenomenon is not limited to the -er suffix. Nominalization and perhaps even suffixation show what interesting constructions multi-word verbs are. Whatever leads speakers to picker-upper and pick-upper might lead further.

Indeed so. The uncertainty as to the attachment point for affixes that is evident in the nominalisation even manifests itself occasionally with the verb. That uncertainty plus limitations of register will presumably hold the nominalisations back.

Notes 1. There is detailed discussion of five noun-forming suffixes of French origin in ME and eModE in Lloyd (2005). 2. Bauer’s definition of playful (1983: 264–5, 2001: 57–8) is not appropriate to the present case, being more suited to formations which play on sound without invoking the normal morphosemantics of a formative, though I suppose it could be argued that in standard PDE, Xage does not mean ‘activity of X’. More relevant to the present case are recent discussions of other coining fashions by Zwicky and Liberman on Language Log (e.g. http://itre.cis.upenn.edu/ %7Emyl/languagelog/archives/004254.html, http://itre.cis.upenn.edu/~myl/ languagelog/archives/004489.html). Bauer’s discussion of creativity rejects the notion of intentional or conscious coining as a usable criterion (2001: 68) and winds up unable to draw a clear distinction between creativity and productivity. 3. Faff v. intr. To fuss, to dither. Often with about. Also as n., fuss, ‘flap’ (OED s.v.). 4. Note in this connection Marchand’s comment: ‘4.4.12. Derivatives in -age are technical terms in that they bear the mark of a special milieu, professional or otherwise, in which they were coined. This does not, however, prevent the words from having general currency. The “result” group is especially common in this respect.’ (1969: 236).

226

David Denison

5. Chapman is right that Marchand does not mention the relevant patterns, though among his haphazard examples of -er derivatives from syntactic phrases there is stand-patter (1969: 280). 6. There are possible citations of picker-oner in this archive from 1905 and looker-inner from 1914, but the context could not be checked (without payment of a subscription, anyway), and the newspapers in question are often very poorly scanned. Other antedatings are similarly suspect. The dated citations from the 1930s given in the text look fairly secure, however. 7. Except that they are not entirely absent: annoys me off gets thousands of Google webhits, and resuscitate back gets a few dozen. Internet data can be rather inconvenient for assumptions about what is possible in language. 8. As noted above, prepositional verbs were silently dropped from Chapman’s original discussion in moving from the OED to internet data; the pattern laughatter would be even more awkward phonologically for a prepositional verb. Table 1 subsequently provided by Chapman (see Appendix) shows that some such forms do occur. 9. McIntyre (2004) credits Rosta with a rhythm-based explanation for the duplication in picker-upper. Discussions of double marking to be added to Chapman’s bibliography include McIntyre’s summary on LINGUIST (2004), cited by Elenbaas (2006). 10. Schlüter has presented evidence of a general tendency towards rhythmic alternation of stressed and unstressed syllables (Schlüter 2003: 88–100, 2005), although not specifically of phrasal verb nominalisations. She points out, however, that such alternation does play a part in word formation processes, citing 'solid vs. so'lidi‫ۈ‬ty; *cor'rup‫ۈ‬tize, etc. (p.c. 8 Nov. 2007). 11. For example, upperer – chosen over offerer, outerer as more likely to exemplify the kind of nominalisation in question – gets ‘about 910’ Google hits (22 Apr. 2007); 25 of the first 30 examined were triple-marked agent nouns, and the remainder included such a phrase as a secret office washing upperer, which at least has the double termination on up. 12. See also Marchand (1969: 277–8) on the -erer suffix of fruiterer. 13. Of course, the resolution of competing tendencies is the basis of Optimality Theory. 14. Note too that it is the first NP which is Experiencer in (19) but the second in (20). And no reason is given for classifying (20) as Type I rather than Type N: the pronoun that is not unambiguously subject of the nominal relative clause. 15. In this connection Elizabeth Traugott reminds me of the recent dissertation by Hilpert (2007), which adds a historical dimension to the ‘collostructional’ model of Gries & Stefanowitsch (2006). 16. I use particle as a cover term for both transitive (prepositional) and intransitive (adverbial) uses, whereas Chapman confines it to the latter only.

Patterns and productivity

227

References Allen, Cynthia L. 1995 Case marking and reanalysis: Grammatical relations from Old to Early Modern English. Oxford: Clarendon Press. Baayen, Harald 1992 Quantitative aspects of morphological productivity. In Yearbook of morphology 1991. G. E. Booij and Jaap van Marle (eds.), 109–149. Dordrecht: Kluwer. 2006 Corpus linguistics in morphology: Morphological productivity. http://www.ualberta.ca/~baayen/publications/BaayenCorpus Linguistics2006.pdf . Bauer, Laurie 1983 English word-formation. Cambridge: Cambridge University Press. 2001 Morphological productivity (Cambridge Studies in Linguistics 95). Cambridge: Cambridge University Press. Chapman, Don this vol. Fixer-uppers and passers-by: Nominalization of verb-particle constructions. Cowie, Claire 2000 The discourse motivations for neologising: Action nominalization in the history of English. In Lexicology, semantics and lexicography: Selected papers from the Fourth G. L. Brook Symposium, Manchester, August 1998 (Current Issues in Linguistic Theory 194). Julie Coleman and Christian J. Kay (eds.), 179–207. Amsterdam and Philadelphia PA: John Benjamins. Croft, William 2001 Radical Construction Grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press. Denison, David 1981 Aspects of the history of English group-verbs: With particular attention to the syntax of the Ormulum. DPhil dissertation, University of Oxford. 1990 The Old English impersonals revived. In Papers from the 5th International Conference on English Historical Linguistics: Cambridge, 6–9 April 1987 (Current Issues in Linguistic Theory 65). Sylvia Adamson, Vivien A. Law, Nigel Vincent and Susan Wright (eds.), 111–140. Amsterdam and Philadelphia: John Benjamins. 1993 English historical syntax: Verbal constructions (Longman Linguistics Library). London and New York: Longman.

228

David Denison

1998

Syntax. In The Cambridge history of the English language, vol. 4, 1776–1997. Suzanne Romaine (ed.), 92–329. Cambridge: Cambridge University Press. 2007 Syntactic surprises in some English letters: The underlying progress of the language. In Germanic language histories ‘from below’ (1700–2000) (Studia Linguistica Germanica 86). Stephan Elspaß, Nils Langer, Joachim Scharloth and Wim Vandenbussche (eds.), 115–127. Berlin and New York: Walter de Gruyter. Elenbaas, Marion 2006 Affix doubling and the English verb-particle combination. Paper presented at Langwidge Sandwidge, Manchester. Fitzmaurice, Susan 2007 Questions of standardization and representativeness in the development of social networks based corpora: The story of the Network of Eighteenth-Century English Texts. In Creating and digitizing language corpora, 2 vols, vol. 2, Diachronic databases. Joan C. Beal, Karen P. Corrigan and Hermann Moisl (eds.), 49–81. Basingstoke and New York: Palgrave. Fraser, Bruce 1976 The verb-particle combination in English. New York and London: Academic Press. Fraser, James B [Bruce] 1966 Some remarks on the verb-particle construction in English. In Report of the seventeenth annual Round Table Meeting on Linguistics and Language Studies: Problems in semantics, history of linguistics, linguistics and English (Monograph series on languages and linguistics 19). Francis P. Dinneen (ed.), 45–61. Washington DC: Georgetown University Press. Goldberg, Adele E. 1995 Constructions: A Construction Grammar approach to argument structure. Chicago and London: University of Chicago Press. Gries, Stefan Th. & Anatol Stefanowitsch (eds.) 2006 Corpora in cognitive linguistics: Corpus-based approaches to syntax and lexis (Trends in Linguistics / Studies and Monographs 172). Berlin and New York: Mouton de Gruyter. Hilpert, Martin 2007 Germanic future constructions: A usage-based study of grammaticalization. PhDissertation, Rice University. Hudson, Richard 2000 *I amn’t. Language 76, 297–323.

Patterns and productivity

229

Hundt, Marianne, Nadja Nesselhauf and Carolin Biewer (eds.) 2007 Corpus linguistics and the web (Language and Computers – Studies in Practical Linguistics 59). Amsterdam: Rodopi. Keller, Frank, Maria Lapata and Olga Ourioupina 2002 Using the Web to overcome data sparseness. In Proceedings of the Conference on Empirical Methods in Natural Language Processing. J. Hajic and Y. Matsumoto (eds.), 230–237. Philadelphia PA: Kennedy, A. G. 1920 The Modern English verb-adverb combination (Stanford University Publications University Series Language and Literature, I 1). Stanford: Stanford University. Lloyd, Cynthia 2005 Some Latinate deverbal suffixes in Middle English: Their integration, productivity and semantic coherence. PhD dissertation, University of Leeds. Marchand, Hans 1969 The categories and types of Present-Day English word-formation: A synchronic-diachronic approach, 2nd edition. Munich: Beck’sche Verlagsbuchhandlung. McIntyre, Andrew 2004 English affix reduplication. LINGUIST 15.1346, 15.1929. . O’Dowd, Elizabeth M. 1998 Prepositions and particles in English. New York: Oxford University Press. Palmer, Chris C. this vol. Borrowed derivational morphology in Late Middle English: A study of the records of the London Grocers and Goldsmiths. Pratt, Lynda and David Denison 2000 The language of the Southey-Coleridge circle. Language Sciences 22: 401–422. Rohdenburg, Günter 2003 Cognitive complexity and horror aequi as factors determining the use of interrogative clause linkers in English. In Determinants of grammatical variation in English (Topics in English Linguistics 43). Günter Rohdenburg and Britta Mondorf (eds.), 205–249. Berlin and New York: Mouton de Gruyter. Rosenbach, Anette 2007 Exploring constructions on the web: A case study. In Corpus linguistics and the web (Language and Computers – Studies in Practical Linguistics 59). Marianne Hundt, Nadja Nesselhauf and Carolin Biewer (eds.), 167–189. Amsterdam: Rodopi.

230

David Denison

Ryder, Mary Ellen 2000 Complex -er nominals: Where grammaticalization and lexicalization meet? In Between grammar and lexicon (Current Issues in Linguistic Theory 183). Ellen Contini-Morava and Yishai Tobin (eds.), 291– 331. Amsterdam: John Benjamins. Schlüter, Julia 2003 Phonological determinants of grammatical variation in English: Chomsky’s worst possible case. In Determinants of grammatical variation in English (Topics in English Linguistics 43). Günter Rohdenburg and Britta Mondorf (eds.), 69–118. Berlin and New York: Mouton de Gruyter. 2005 Rhythmic grammar: The influence of rhythm on grammatical variation and change in English (Topics in English Linguistics 46). Berlin and New York: Mouton de Gruyter. Tieken-Boon van Ostade, Ingrid 2000 Social network analysis and the history of English. European Journal of English Studies 4: 211–216. Trousdale, Graeme this vol. Words and constructions in grammaticalization: the end of the English impersonal construction. Ukaji, Masatomo 1992 ‘I not say’: Bridge phenomenon in syntactic change. In History of Englishes: New methods and interpretations in historical linguistics (Topics in English Linguistics 10). Matti Rissanen, Ossi Ihalainen, Terttu Nevalainen and Irma Taavitsainen (eds.), 453–462. Berlin: Mouton de Gruyter.

Borrowed derivational morphology in Late Middle English: A study of the records of the London Grocers and Goldsmiths*

Chris C. Palmer In Present Day English (PDE), a number of affixes that were originally restricted to borrowings from other languages are now productively used to coin a variety of new words. The suffix -age, for example, has recently shown up in a number of innovations in a range of registers and communities. In the 1970s in North America, signage (‘signs collectively, esp. public signs on facia boards, signposts, etc.’) seems to have emerged in official, governmental contexts and spread to a more general usage.1 In computer gaming communities, the word ownage has been coined to express one opponent’s strong victory over another: ‘The act or state of perpetrating fierce and unholy domination against another, typically in a videogame setting, resulting in shame and embarassment [sic] for the victim and his/her family until the end of time.’2 Currently, the use of ownage has been extended beyond gamer communities; one now finds real-life instances of ownage (people pulling pranks on one another, people falling while skateboarding, etc.) on such sites as http://www.ownagevideos.com/ (October 2006). Even though they are less widespread, individual playful applications of -age appear. In 2005 the employees at a deli in Ann Arbor, Michigan labeled their tip jar with “TIPPAGE: Supporting Counter Intelligence since 1738.” While this specific example may not gain wider currency, these varied innovations demonstrate how much PDE speakers (particularly in North America) perceive -age to be a productive, useful affix. In the history of the English language, -age did not always have the productive, morphemic status it has today. Indeed, like other derivational affixes such as -(c)ion, -ance, -ity, and -ment, it appeared on a restricted set of borrowings from Latin and French in the medieval period, such as baronage, frontage, and baggage (Marchand 1969: 234–6). Because these endings were likely seen initially as mere word-endings rather than potentially meaningful suffixes, they were not immediately combined with native

232

Chris C. Palmer

bases to produce innovative forms. How, then, did speakers and writers come to perceive -age and other endings as (potential) suffixes? This small-scale analysis of the records of two medieval London communities adds to a small but growing set of studies that shed light on such diachronic morphological questions. Previous work on the subject has produced an impressive catalogue of the meanings and uses of derivational morphemes in the history of English (Gadde 1910, Marchand 1969, Miller 2006), as well as a thorough yet general overview of the frequency, combinability, and semantics of different forms within texts of the medieval portion of the Helsinki Corpus (Dalton-Puffer 1996). Fleischman details the history of the suffix -age in Romance languages and English, with particular attention to the sociolinguistic impact of medieval feudal relations on its morphological and semantic development (Fleischman 1977). Emphasizing that there has been “a shortage of detailed diachronic studies of individual affixes which measure changes in productivity,” Cowie (1998a: 38) provides an analysis of the varying productivities of -ness, -(t)ion, and -ity in Early Modern and Modern English, taking into account sociolinguistic features such as register. But less attention has been devoted to the identification and analysis of some of the mechanisms that might have led to the eventual productivity – and suffixal status – of different endings on borrowings coming into English during the medieval period. Furthermore, these large-scale studies attempt to explain morphological processes within the general grammar of English in different historical periods; little attention has been paid to the morphological developments of individual affixes within specific communities. Hence, several critical questions about historical English morphology remain underexplored. What types of textual evidence potentially reveal the processes by which endings of borrowings became eventual suffixes? How do the records of individual communities help to complete the picture of such morphological developments? And which linguistic methods prove most useful in exploring such questions about borrowed derivational morphology? To address these questions, this article analyzes borrowings within two multilingual textual records in the late fourteenth to fifteenth centuries: the accounts of the London Grocers and the account/minute books of the London Goldsmiths. Primarily, this study compares the use of native nominal affixes (-ness, -ship, and -hood) with borrowed, potential affixes (-cion, -ance, -ity, -age, and -ment) throughout the English portions of these texts. Attempting to locate evidence of the naturalization of these forms – the process by which

Borrowed derivational morphology in Late Middle English

233

these endings become derivational morphemes in the general English lexicon – the paper develops the notion local productivity. This measure combines both quantitative and qualitative data to show that, even in smaller corpora, historical linguists can find evidence of the morphological status of different potential affixes for communities within particular historical moments. Specifically, this study finds that within these communities, the borrowed potential affixes were in the early stages of naturalization – they had limited productivity within a restricted subset of the lexis, though speakers were beginning to see them as individual units. There was a potential recognition of the similarity in form of these word-endings as well as their potential separability from their bases. Furthermore, the data indicate that there may in fact be variation in the derivational development of these affixes: -age is more productive and more naturalized for the Grocers than for the Goldsmiths. This study not only adds to our understanding of borrowings and derivational morphology in the medieval period. It also argues for the necessity of analyzing smaller sets of texts closely – e.g., examining the local productivities of affixes – alongside our larger, computer-assisted corpus studies. And finally, it reflects on some of the theoretical implications for our understanding of productivity and language change when we (as present day historical linguists) read and interpret this sort of data in specifically written examples.

Burnley’s description of the naturalization3 of borrowed morphemes In his chapter on “Lexis and Semantics” from the 1066–1476 section of the Cambridge History of the English Language, David Burnley (1992) broadly describes this general process of naturalization of borrowed morphemes: Foreign words may be adopted with affixes as part of their structure, and these affixes may become productive in English. Here it is necessary to distinguish three successive stages. Firstly, the word containing affixes is adopted into English and assimilated into the grammatical systems of the language. Secondly, after analysis of the word structure, there follows a period during which the word is stylistically differentiated from the rest of the lexis. It is synchronically recognisable by speakers of the language as for-

234

Chris C. Palmer

eign, and its affixes may be used to produce new formations with a restricted set of bases also perceived to be foreign. Such affixes are productive only within a subset of the lexis. Finally, as coinages become more numerous, the affix ceases to be considered exotic, and is used to coin words on bases of any origin. At this point the affix has become part of the general derivational system of the language. (445–6)

Burnley’s discussion of morpheme naturalization can be distilled into three primary stages. First, there is a period of adoption and assimilation of the borrowings containing the potential affix. Next, there is an analysis of the word structure, in which the ending formerly seen as merely a word ending is reanalyzed as an affix. This reanalysis is accompanied by the stylistic differentiation of these forms and the potential production of new forms within a foreign subset of the lexis. In the final stage, formations including the affix increase in frequency as the affix attaches to a wider set of bases, including native bases. It is not immediately clear from this description, however, what “stylistic differentiation” would actually look like and how the “analysis of word structure” could be identified in real language use. In other words, what textual and contextual clues can the historical linguist identify in order to pinpoint when and how Middle English readers and writers were analyzing the complex structure of their own words? The qualitative section of this paper will address this question specifically. Furthermore, Burnley writes that “as coinages become more numerous, the affix ceases to be considered exotic.” Implicit in this claim is that frequency of usage itself has a direct impact on the language user’s perceptions of words – their potential “nativeness” as well as their morphological structure. Both the quantitative and qualitative analyses in this paper will address the potential impact of frequent usage on derivational change.

Working towards a methodology: Some key terms As is clear from Burnley’s description and the preceding discussion, linguists must rely on two critical concepts in order to discuss processes related to the borrowing of derivational morphology: productivity and naturalization. Naturalization, as I use the term, refers simply to the increasing loss of speaker awareness of the foreignness of some loan words or morphemes.

Borrowed derivational morphology in Late Middle English

235

Naturalization can perhaps be distinguished from nativization, in which borrowings begin to adopt native patterns of phonology. An example of a nativized form in PDE would be [hamԥdž] for homage, where the initial [h] is pronounced, the first syllable stressed, and the final vowel reduced. An example of a borrowing that has been naturalized without such overt nativization is tax, at one time considered a hard word needing to be glossed in Cawdrey’s Table Alphabetical in the Early Modern period.4 In PDE this word has lost its apparent hardness and foreignness, perhaps due to frequency of usage rather than to native phonological changes. The interesting diachronic question for this study, of course, is how this naturalization process took place historically for derivational morphemes. Unfortunately, affixes did not have entries like the words in Cawdrey’s table. Because of this lack of meta-data, linguists must infer medieval readers’ understanding of these affixes from actual textual usage in available written records. In the case of derivational morphology, we want to know how (and if) borrowed affixes such as -age, -(a)cion, and -ance become more naturalized and integrated into the broader derivational system of English. Productivity has been variously defined, both on a theoretical/conceptual level and an empirical/operationalizable level. As Bauer (2005: 317) points out, Schultink provides one of the earliest attempts to theorize productivity: “By productivity we understand the possibility for language users, by means of a morphological process which underpins a form-meaning correspondence in some words they know, to coin, unintentionally, a number of new formations which is in principle infinite” (1961: 113, Bauer’s translation). This definition is not uncontroversial: it is virtually impossible to consistently and reliably distinguish creative, playful formations from unconscious applications of a word-formation rule.5 Moreover, Cowie (1998b) finds that nominalizations in -(t)ion were both productively and intentionally employed by scientific writers to construct the professional register of scientific discourse during the seventeenth to nineteenth centuries. In her study, intentionality clearly did not preclude productivity. The notions of productivity and naturalization, while related, should not be conflated. Affixes might exhibit lexically restricted productivity, in which they produce coinages within a restricted sub-set of the lexis (e.g., Latinate terminology), or wider productivity, in which coinages are produced without restriction to few identifiable sub-lexicons. In the case of affixes restricted to Latinate lexical items, the endings of borrowings can potentially become productive suffixes without being fully naturalized (that

236

Chris C. Palmer

is, without becoming part of “the general derivational system” of English, in Burnley’s terms). Morphological productivity can occur without full naturalization, but any assessment about the naturalization of a borrowed morpheme should consider its level of productivity. A lexically restricted affix is considered less naturalized, while a widely productive affix – one that attaches to both native and non-native stems – is considered more fully naturalized in English. To characterize the level of naturalization and the types of productivity of different affixes, this paper will inspect hybrid formations, in which a Latinate affix attaches to a native base or vice versa. These lexical items will be discussed later in the qualitative analysis. But how is one to quantify productivity, to measure the rate of coinages of a certain word formation pattern, particularly in a diachronic context? Harald Baayen (1992) has provided the most widely adopted and adapted measure of productivity. By examining very large corpora, he has found that the number of coinages in a language correlates strongly to the number of hapax legomena – words that occur exactly once in a large corpus. The assumption is that less productive processes create fewer words, all of which will eventually appear in a corpus, often more than once. More productive processes will generate a wider range of forms, many of which will only appear exactly one time in vast stretches of language. The primary formula Baayen provides is P = n1 / N

where n1 is the number of hapaxes formed by a particular process (e.g., -able) within a large corpus, N is the total number of tokens formed by this same process in this corpus, and P is the productivity value. There are two important clarifications to note here about the actual meaning of this value. First of all, the measure is an indirect account of the productivity of a certain affix. By counting hapaxes, one is not necessarily tallying actual coinages. The theory is that the hapaxes merely correlate with coinages generated from a certain word formation process. Secondly, the productivity has no inherent value beyond the corpus from which it is generated. It must be compared to other values (for other word formation processes, for example) to determine a relational measure of productivity. Up to this point, Baayen’s quantitative measure has never been successfully applied to historical corpora. One of the primary problems is the small size of available historical corpora such as ARCHER and the Helsinki Corpus. In small corpora, productive and non-productive processes alike produce so few hits that there will be an overgeneration of hapaxes, inflating

Borrowed derivational morphology in Late Middle English

237

the productivity counts of all processes.6 At the same time, some less productive processes available in a certain time period may appear infrequently (if at all) in a corpus full of small samples. This undergeneration would lead a historical linguist to underestimate the productivity of a number of emergent, less common processes. Clearly, to study the productivity of derivational morphemes historically, linguists must adjust their approach to measuring this phenomenon. One approach might involve applying Baayen’s and other frequency-based measures to larger, unprincipled historical corpora, which have become increasingly available in recent years (e.g., the Middle English Compendium, Early English Books Online). But linguists should not abandon the study of smaller corpora completely, especially since in some periods (such as Middle English) we must rely on the sparse resources we have available to us. In this study, I argue that the use of small corpora, a small body of texts, despite its apparent disadvantages, is a desirable, perhaps even necessary component to understand the development of derivational morphology in the history of English and the naturalization of borrowings. By focusing on the use of native and foreign derived nominalizations7 in the records of two medieval London communities, I attempt to discover – as much as possible – two Middle English communities’ understanding and usage of some potential affixes.

The importance of reading the records of multilingual communities While studying large, digital corpora for linguistic research has its evident advantages, much can be gained by reading through entire texts with a careful eye.8 To illustrate the advantages of such slower, philological reading, I refer to the two corpora9 explored in the present study: (1) the records of the Grocers’ Company; and (2) the Wardens’ Accounts and Court Minute Books of Goldsmiths’ Mistery of London. These fourteenth to fifteenth century accounts appear in printed editions with direct transcriptions of all verbal material.10 A significant advantage of these texts – unlike the majority of available larger corpora – is that they are multilingual. Each community kept records either in French, Latin, or (increasingly) English. Because of the multilingual nature of these records, we can be certain that the scribes and some portion of the community were familiar with multiple languages and had the linguistic resources to employ borrowings from

238

Chris C. Palmer

French and Latin into their English usage. William Rothwell emphasizes the need to consider such texts for linguistic evidence, asserting that “the rise of English in the fifteenth century … did not take place in a vacuum, but against the background of the dominant commercial, diplomatic and legal language of the time – French” (2001: 549). It is thus necessary for us to look at multilingual records to characterize how French (and even Latin) might have influenced English usage and grammar in particular communities, especially in the case of derivational morphology: French contact had an important impact on the morphology and the lexicon of Middle English. In addition to having some assurance about the language contact situation of these texts, there are other major advantages in choosing to analyze and compare these specific corpora. The editions of the texts both span approximately the same time period in the same location, mid-fourteenth to mid-fifteenth century London. They employ similar discourses, including much economic and legal lexis. Part of this lexical overlap is due to the daily business and record-keeping practices of both companies. The medieval Goldsmiths initially formed a guild in order to regulate the standards of gold and silver quality and exchange; over time, they began to take on such additional functions as renting and managing tenements.11 The medieval Grocers performed several activities: they were involved in the import and export of traded goods, including their storage and inventory; they maintained weights and measures at the Port of London; and they participated in religious ceremonies.12 The Grocers’ accounts include a number of ledgers with associated fees for different services; the Goldsmiths include such ledgers to a lesser extent. Both the Goldsmiths and Grocers also include a number of other genres, including inventories, defaults, ordinances, memoranda, and even occasional records of abuse. The range of genres allows for questions about the impact of register/discourse on derivational use. Furthermore, the records are precisely dated, allowing the researcher to track developments and changes throughout the textual history of each community. An individual year usually has several different entries, often in the same hand. The use of an entire corpus from a community also invites questions about social network considerations of the membership of that community (Milroy 1980). Unfortunately, it is nearly impossible to construct a proper network analysis using historical texts from the medieval period. We are unable to determine the density and multiplexity of social ties, particularly when specific writers are unidentifiable.13 Even so, historical information

Borrowed derivational morphology in Late Middle English

239

about the Goldsmiths and Grocers provides some hints of their linguistic influences. Reddaway and Walker (1975) describe the varied social forces that either constituted or else interacted with the Goldsmiths’ Company of the early fifteenth century: In a city as full of activity and combativeness as London in the forty years from 1404 to 1444 the Company’s position could hardly remain static. It had rivals to watch, border-line trades such as the refiners of precious metals (the finers) and the jewellers to draw into their partnership or subjection, alien goldsmiths to seek out and bring within the Company’s jurisdiction, and reasonable relations to maintain with a Crown and a parliament … (95)

The community was in flux, its increasing membership and daily social contacts including not only native goldsmiths but a substantial number of Dutchmen, immigrants from the Low Countries, the Rhineland, several parts of Germany, France, and even Italy and Spain (120). To ensure craftsmanship and proper goldsmith-customer relations, the Company dealt with a growing number of individual immigrants and communities of immigrants throughout London (121–123). During the early fifteenth century, the network ties among the Goldsmiths’ membership were becoming increasingly dynamic and multiplex. Their dealings throughout London with other crafts and communities may have increased their ability to absorb and spread linguistic innovations. As for the Grocers, Pamela Nightingale (2005) explains how in the early fifteenth century they made an active effort to recruit merchants from provincial towns into their general membership.14 The company wanted to maintain social ties to different regions in order to maintain dominance over distributive trade within England (395). By the 1430s, the Grocers were building a new hall, but their community by this time was “dispersed so widely throughout the City” that, even after the hall was erected, “their personal ties with their parish church and with their families and friends outside the craft … were much stronger than their business relationships within it” (429–430). The looser ties in the business community was a stark contrast to earlier centuries, when the Grocers had a more tight-knit community in which they “had lived, traded, and worshipped together in the neighborhood of Sopers Lane,” their former home (431). These historical facts may have linguistic implications for this study, since the loosening of community ties among the London Grocers, alongside their increasing ties

240

Chris C. Palmer

to regions throughout the country, may have made the members more likely to take up, to introduce, and to spread innovations – including new borrowings or newly coined words. Because specific network connections cannot be easily traced in these communities, any declarations about network influences must remain speculative at best. Even so, attention to such historical information provides a general sense of the possible linguistic influences of these communities in the larger geographic contexts of London and England.

Local productivity: A quantitative measure As discussed above, it is unhelpful to calculate productivity scores for affixes in these communities’ records: the corpora are too small to rely on counts of hapax legomena. It is thus necessary to retool general productivity measures for smaller sets of texts. In this section I introduce the notion of local productivity, a quantitative and qualitative measure of the independence, use, and attachability of potential affixes in small, historical corpora. The qualitative measure considers the cotext and context of individual examples to establish relative productivities of potential affixes. As will be shown in the following section, when the qualitative measure is applied to borrowed (potential) affixes, linguists can more accurately evaluate the lexical status(es) of these forms and determine how naturalized they are for specific communities in specific historical moments. The quantitative measure extends a methodology developed by Cowie and Dalton-Puffer in their article, “Diachronic word-formation and studying changes in productivity over time: Theoretical and methodological considerations” (2002). They emphasize that productivity measures of existing words, whether synchronic or diachronic, always rely on some account of the different types and tokens of a single affix (or other morphological process).15 In this sense, every productivity measure is quantitative. But the critical point is that this attention to type counts reveals an underlying assumption about productivity which is not entirely unproblematic: This [reliance on type/token counts] reflects the intuition that the productivity of an affix correlates directly with the number of different types containing it. One of the problems is that the existence of a large number of types may of course not be the result of current productivity but of an aggregation through productivity in the past. (416)

Borrowed derivational morphology in Late Middle English

241

Here Cowie and Dalton-Puffer critique productivity studies that assume a direct correlation between type frequency counts and synchronic productivity. Consider a section of Dalton-Puffer’s study of the Middle English suffixes -ness, -ite, and -acioun: Table 1. Types and Tokens for selected nominal derivatives from ME section of the Helsinki corpus (Dalton-Puffer 1996, and re-presented in Cowie & Dalton-Puffer 2002) Helsinki Corpus Subperiods Sub-Corpus word count -ness types/tokens -ite types/tokens -acioun types/ tokens

ME1: 1150–1250 113,010

ME2: 1250–1350 97,480

ME3: 1350–1420 184,230

124/468 7/12 4/10

60/289 20/57 20/56

108/575 71/365 138/533

The -ness types show a general decline in frequency over time (with a substantial drop-off in ME2), while the borrowed -ite and -acioun types show a consistent increase in frequency throughout the period. But can we conclude, based on these data alone, that -ness becomes less productive while -ite and -acioun become more productive? While this is likely the case, we do not have the complete picture here. It is impossible to know, for example, if -acioun is truly more productive than -ness by the ME3 period since we have no account of the new types appearing in each period. (The type counts of -ness may in fact consist of a higher number of innovative forms than the -acioun types.) Cowie and Dalton-Puffer argue that diachronic studies can directly respond to the need to measure productivity by accounting for newly introduced types rather than type frequencies alone. They introduce the concept of “aggregation,” which tracks “in each subperiod [the] new types [that] are added to the types of the previous period” alongside general increases and decreases in overall type frequencies (428). The aggregation of new types in a body of texts over time, the general trends in the use of different types, provides some sense of what is and what is not productive. While it may be possible to extrapolate from the data drawn from small corpora, the aggregation measure more accurately reflects the local productivity of different forms within a specific body of texts – and within the community that has produced and received these texts.

242

Chris C. Palmer

Tokens, types, and new types In my quantitative analysis of the Goldsmiths’ and Grocers’ records, which appear in Tables 2–5 below, I have counted the token and type frequencies (including new types) of nominal derivations in 5-year subperiods from 1415-1444, the overlapping time period in which both the Grocers and Goldsmiths used English in the published editions of these texts. Because these texts were not digitized, it was impractical to tabulate overall word counts. In my study, a token is any one occurrence of a form containing the ending listed at the head of a column.16 A type corresponds to a lexeme – the set of possible orthographic and inflectional variations of one lexical item. Thus, because of orthographic variation and inflectional morphological change, multiple tokens will correspond to the same type: e.g., payment and payementes are two tokens of type PAYMENT. Table 2. Token counts for selected nominal derivatives during 5-year periods of the Goldsmiths’ books Goldsmiths 1415–19 1420–24 1425–29 1430–34 1435–39 1440–44 Total # Tokens

-(c)ion 1 1 5 4 9 8 28

-ance 2 7 5 8 10 12 44

-ity 1 0 2 3 1 3 10

-age 0 0 1 0 0 0 1

-ment 0 0 2 2 7 2 13

-ness 0 3 5 0 0 5 13

-ship 1 7 5 2 1 4 20

-hood 0 0 1 1 2 2 6

Table 3. Token counts for selected nominal derivatives during 5-year periods of the Grocers’ books Grocers 1415–19 1420–24 1425–29 1430–34 1435–39 1440–44 Total # Tokens

-(c)ion 9 0 24 17 16 18 84

-ance 10 1 26 8 1 3 49

-ity 30 0 7 1 0 0 38

-age 0 0 13 12 18 8 51

-ment 7 0 22 14 21 15 79

-ness 0 0 2 1 0 0 3

-ship 10 9 13 12 10 3 57

-hood 13 0 0 1 1 1 16

Borrowed derivational morphology in Late Middle English

243

Table 4. Type and “new type” counts (# of newly occurring types in parentheses) for selected nominal derivatives during 5-year periods of the Goldsmiths’ books Goldsmiths 1415–19 1420–24 1425–29 1430–34 1435–39 1440–44 Total # Types

-(c)ion 1 (1) 1 (1) 5 (5) 3 (3) 7 (6) 6 (5) 21

-ance 1 (1) 4 (3) 3 (2) 3 (1) 2 (0) 7 (4) 11

-ity 1 (1) 0 (0) 2 (2) 2 (1) 1 (1) 1 (1) 6

-age 0 (0) 0 (0) 1 (1) 0 (0) 0 (0) 0 (0) 1

-ment 0 (0) 0 (0) 1 (1) 1 (1) 3 (1) 2 (2) 5

-ness 0 (0) 3 (3) 3 (2) 0 (0) 0 (0) 4 (2) 7

-ship 1 (1) 3 (2) 3 (1) 2 (1) 1 (0) 2 (1) 6

-hood 0 (0) 0 (0) 1 (1) 1 (1) 1 (0) 1 (1) 3

Table 5. Type and “new type” counts (# of newly occurring types in parentheses) for selected nominal derivatives during 5-year periods of the Grocers’ books Grocers 1415–19 1420–24 1425–29 1430–34 1435–39 1440–44 Total # Types

-(c)ion 7 (7) 0 (0) 12 (8) 10 (4) 6 (2) 8 (3) 24

-ance 6 (6) 1 (0) 9 (6) 4 (1) 1 (0) 3 (1) 14

-ity 3 (3) 0 (0) 2 (2) 1 (0) 0 (0) 0 (0) 5

-age 0 (0) 0 (0) 5 (5) 6 (2) 6 (1) 4 (0) 8

-ment 5 (5) 0 (0) 8 (7) 9 (3) 8 (2) 7 (3) 20

-ness 0 (0) 0 (0) 1 (1) 1 (1) 0 (0) 0 (0) 2

-ship 3 (3) 1 (0) 4 (1) 2 (0) 2 (1) 2 (0) 5

-hood 3 (3) 0 (0) 0 (0) 1 (0) 1 (0) 1 (0) 3

I have included both borrowed forms (-(c)ion, -ance, -ity, -age, and -ment) alongside native ones (-ness , -ship, and -hood.) Underneath the type heading in each column in Tables 4 and 5 appears the number of different types of each suffix during each 5-year subperiod alongside the number of “new” types in each period in parentheses. To clarify, in this study new does not necessarily nor even usually mean a neologism. Rather, it indicates the first use of a particular lexical item in English within each community’s entire records. So in this sense, each new type is a sort of “local neologism,” either the earliest diffusion of a borrowing into English writing for the Grocers or the Goldsmiths, or else a new derivation with an affix within these communities. As Cowie (1998a) emphasizes, for Latinate lexical items, there is unfortunately “no way of distinguishing between a loanword and a

244

Chris C. Palmer

derivation in a Latinate affix” (75–6). Even with the help of a historical dictionary, one cannot reliably identify these new types as either borrowings or derivations. It should be noted that my methodology here diverges from Cowie and Dalton-Puffer in two significant ways. First, the authors encourage the use of a longer range for subperiods (e.g., 50 years each). This is of course not possible for these specific texts: not only is the material in publicly available editions limited in time span, but even if these texts spanned several centuries, a long-range count could not be feasibly accomplished by employing a slow, full reading of texts. Second, Cowie and Dalton-Puffer recommend using a “starting lexicon,” deriving forms from a dictionary or an earlier period of a corpus, to determine whether a type is new to the corpus in later periods. Such a starting lexis was not necessary for the present study, since in the Grocers’ and Goldsmiths’ records there is a complete record of their emergent English usage. It is thus not too difficult to track local neologisms, the first uses of each type within the English sections of each community’s records.17 Tables 2 and 3 provide a general sense of the usage of each form in different periods within these records. One immediately apparent difference is the Grocers’ high use of -age forms and the Goldsmiths’ infrequent usage. But as the discussion above has emphasized, the important trends in this quantitative analysis can be seen more clearly by considering the trends in new-form usage over time. Aggregation trends are useful in understanding some general derivational patterning within these texts, but because the numbers are so small, one must be careful not to overextend the analysis. Even so, by looking up and down individual columns in Tables 4 and 5 – focusing particularly on the type frequencies of individual affixes as well as the (in)consistent aggregation of new types over time – we can draw several tentative conclusions: –

–

-(c)ion and -ance consistently aggregate throughout this period in both communities. The form -(c)ion continues to follow the trend seen in earlier ME (cf., Dalton-Puffer’s study, Table 1) as it outpaces -ity, which shows little (if any) aggregation or productive growth. The native forms show little aggregation in either community. This is unsurprising, except for how little -ness is employed compared to the borrowed suffixes (especially for the Grocers). The general trend in ME (as seen in the Helsinki corpus, Table 1) previously shows -ness to be eventually slightly outpaced by -(c)ion during 1350–1420. But in terms

Borrowed derivational morphology in Late Middle English

–

245

of overall frequency in the larger corpus study, -ness is clearly more frequent than -ity in every single subperiod from 1150–1420. Perhaps the multilingual nature of the Grocers’ and Goldsmiths’ communities explains their divergence from the more general trends in the ME period. Their knowledge of Latinate or French nominals ending in -(c)ion or -ite certainly gave them more opportunities to employ borrowed forms rather than nominals ending in -ness. This is not to say that for every -(c)ion form the Grocers and Goldsmiths had available a competing native nominal form, although we do find rebelions in the Goldsmiths’ records and rebelness in the Grocers’. But even in this case, the terms are not remotely synonymous: rebelion signifies an event, whereas rebelness conveys a general characteristic or personality trait. The effects of genre and their related lexical sets – as well as the wide availability of borrowed lexical items due to language contact – must have driven these communities’ use of non-native nominals over the native ones. The Goldsmiths use -ness in characterizing the occasional misdeeds of individuals (e.g., falsnesse, cursidnesse from 1425). But even in these very records of complaint they use many more legal and economic terms (e.g., axion, allegance, juggement, seurtee, from the same 1425 entry). The Grocers use only two -ness types during this entire period, perhaps because there were fewer records of complaint. A curious difference in the derivational uses of these communities is the differences in -ment and -age. These are fairly aggregating forms for the Grocers. For the Goldsmiths, -ment shows a very slight aggregation, but -age is almost entirely absent. In fact, their only usage of -age is usage. If we consider the everyday business of the Grocers, we find some explanation as to why their use of -age makes social sense. The Grocers Company, as part of their mercantile commitments, had to deal regularly with transport, storage, and the bookkeeping of inventory. Hence, in their records we commonly see forms such as cariage, portage, wharfage, and surplusage. According to Marchand, many of these -age forms suggest not only the physical amount of goods (the inventory), but also the toll or duty associated with the privilege or service in transporting and storing them (234–6). Thus the Grocers’ accounts reveal a sublexis of -age formations that signal the use of mercantile discourse, an emergent professional economic discourse in the late Middle English period. Similarly, -ment seems to function as a marker of economic and legal discourse for the Grocers: lexical items include inventory of capital (ornaments, vestiment, tenement), financial transactions

246

Chris C. Palmer

(payement, mersyments), and legal actions (judgement, amendment, arbeterment, testament, agrement). It should be noted that, because the overall period covered in this study is relatively small, these results cannot expose larger trends in morphological change. Even so, the analysis of the aggregation of nominals within the records reveals much about the local productivity of these forms within these communities. According to this quantitative account, the borrowed nominal forms are more productively employed than the native ones, almost across the board. The most consistently aggregating forms are clearly -(c)ion and -ance, and to a lesser extent, -age (for the Grocers) and -ment. An additional analysis of the -age and -ment types themselves reveals that social and discursive forces have promoted their increasing use.18

Local productivity and naturalization: A qualitative analysis As shown in the previous section, a quantitative analysis focused on the aggregation of new morphological types in two small corpora can provide generally useful information about the relative productivity of different affixal forms. But one of the big questions that this quantitative analysis has not answered, and perhaps cannot answer, is the following: how do we know if medieval people – or more specifically, these Grocers and Goldsmiths and their scribes – recognized a potential affix as an affix rather than just part of a borrowing? Are these foreign forms naturalized, marked as borrowed, or somewhere in between? And what are our criteria for making this decision? In order to answer these questions, the linguist cannot rely on quantitative data alone: s/he must conduct a qualitative analysis of specific textual examples. Interpreting data from the entire English records of the Grocers and Goldsmiths, I examine the internal morphological structure of individual lexical items, their co-text, and their wider context to ascertain the local productivity and naturalization of borrowed derivational morphemes.

Individual complex lexical items: Hybrid formations As Dalton-Puffer demonstrates, one way to evaluate the productivity and naturalization of borrowed morphemes is to examine hybrid formations,

Borrowed derivational morphology in Late Middle English

247

“complex words which mix elements from the native Germanic part of the vocabulary with elements from the borrowed Romance part of the vocabulary” (1996: 211). Hybrid formations qualitatively demonstrate which affixes are productive beyond a lexis restricted by source language – a key component of the final stage of Burnley’s description of naturalization. In Middle English, hybrid formations are primarily of two types: (a) Romance base + Germanic suffix; or (b) Germanic base + Romance suffix. DaltonPuffer reminds us that each type conveys a different characteristic of the language contact scenario, even a different conclusion we can draw about the derivational system of English: … there is a qualitative difference between a language adopting from another language names for objects, concepts or actions (often with their extralinguistic referents) which can then serve as derivational base [sic], and that language borrowing elements (suffixes) that more properly belong to its “mechanics.” In other words, there is a qualitative difference between the two types of hybrids. Type [(a)] (with a borrowed derivational base) is much more common and the result of “ordinary” borrowing. Type [(b)] (borrowed affix) is generally assumed to be the result of much closer language contact (cf. Thomason-Kaufman 1988) (211)

In other words, type (a) hybrids demonstrate the naturalization of borrowed bases, which integrate further into English as native derivational morphology begins to attach to them. Type (b) hybrids reveal the naturalization of borrowed affixes themselves, which begin to attach to a wider lexis than a smaller, borrowed sub-lexis. Both hybrid types indicate affixal productivity unrestricted by differences in the Latinate or native etymology of the base. While hybrids tell us much about the naturalization of borrowings and productivity of different affixes, they are unfortunately relatively infrequent in the Middle English period – particularly type (b) hybrids. Having searched all borrowed suffixes throughout the entire ME portion of the Helsinki Corpus (608,570 words), Dalton-Puffer finds only 14 type (b) hybrids. Miller (1997), however, finds a significant number of type (b) hybrids in Middle English by examining a wider range of texts than those in the Helsinki Corpus, at least 100 different types before 1450. In the multilingual records of the Grocers and Goldsmiths, we might expect to find a higher proportion of type (b) formations. But this is not the case. The set of hybrid forms from the English sections of the Goldsmiths’ and Grocers’ records in this period is listed below:

248

Chris C. Palmer

Goldsmiths: feithful, prively, masiterschip, dispitously, unbuxunesse, apprentishood, effectuely, duely, vilenously, generally, wardeynschipe, rentership, entirely, notably Grocers: quarterly, maisterschipe, Flaundrissh, curteysly, rebelness, prentyshodys, unresonable, grevowsly, condissionally, Remembrancer, Indyfferently, lynyally, wharfage, cranage Note that the sheer infrequency of hybrids in this period, particularly tokens of type (b), precludes a larger quantitative account of different hybridization patterns. But by inspecting these few formations individually, we can draw several conclusions about the naturalization of borrowings and the productivity of the derivational morphology in these communities: –

– –

–

The widely productive affixes are primarily native: -ful, -ly, -ship, -ness, and -hood. This conclusion is particularly important since the earlier quantitative analysis suggested -ship, -ness, and -hood had low local productivity relative to the borrowed forms. In this case, the qualitative data help to temper any conclusions drawn from the earlier quantitative data about the lack of new types produced by native affixes. In the list of hybrids -ness is clearly a productive form, active in coinages such as the infrequently unattested unbuxunesse ‘disobedience?’; the OED provide only six citations for this item (and does not include the Goldsmiths’ use). Conversely, the non-native forms are clearly not widely productive, almost entirely restricted to borrowed lexis. Because we have so many type (a) hybrids, we can conclude much about the naturalization of several borrowed stems and bases. Items such as faith, priv-, due, and reasonable have taken native derivational morphology, further integrating them into the English language. The only type (b) forms we get are wharfage and cranage, which occur in the Grocers’ records.

What can we determine about the productivity and naturalization of -age from these data? As mentioned above in the quantitative analysis, -age showed small but consistent aggregation for the Grocers. Because it is the only borrowed form that attaches to native stems, -age must be more widely productive than its borrowed nominal peers (at least for the Grocers). Attaching to a base other than a borrowing, it is in the final stages of the naturalization process.

Borrowed derivational morphology in Late Middle English

249

Even so, -age appears in only two hybrid types in these records: one must be careful not to assume it was widely productive for these communities. Here I diverge from Burnley, who claims simply that -age is “fully assimilated in Middle English” (449). There is not enough data in the records of these communities to indicate full assimilation, particularly since most -age formations are restricted to borrowed bases. Even though (according to Burnley) other hybrid forms such as barnage ‘infancy’ appear elsewhere as early as 1325, these records indicate that the Grocers use -age almost entirely within a restricted borrowed sub-lexis – namely, AngloFrench economic terminology. The evidence from the Grocers’ records corroborates Fleischman’s general contention that -age was used widely in the Middle Ages for deriving or borrowing words designating taxes, fees, and dues, with the suffix exhibiting “a decided affinity for commercial and nautical terms” (148, 407). And this fact is perhaps unsurprising since the records left by the medieval Grocers deal mostly with their financial transactions surrounding commercial trade via ports and wharves. In any case, rather than claiming full assimilation (i.e., complete naturalization and/or wide productivity) for -age in this period, it is perhaps more accurate to state that in certain medieval communities (such as the Grocers), the suffix -age is only partially assimilated and moderately productive.

Co-text The co-text and wider context of individual examples of borrowed derivational forms reveal more crucial information about the medieval readers’/writers’ awareness (most reliably among those in the Grocers’ and Goldsmiths’ mysteries) of the potential morphological status of such borrowed suffixes as -age, -ance, -(c)ion, -ment, and -ous. In attempting to analyze writers’ and readers’ understanding of potential affixes, I interpret examples of formal arrangement and textual cohesion as evidence for a paradigmatic recognition of similar forms. I argue that this recognition of formal similarity must be in place cognitively – that is, in speakers’ linguistic consciousness – in order for a borrowed ending to be perceived as a potentially productive suffix. Consider the following example: in the ledgers of the Grocers’ accounts, the scribes sometimes include -age forms within the same noun phrase on the same ledger line:

250

Chris C. Palmer

(1)

Grocers (1427):

Item for stapulton stoon Cranage Wharuage and cariage … Summa xxj li xij d

Here the scribe has brought together three words of similar shape – Cranage, Wharuage, and cariage. All three have the same endings, perform the same syntactic function, and belong to a similar semantic/lexical class of -age formations. Moreover, the physical shape of their presentation – their juxtaposition on the same ledger line – highlights their similarities.19 But the question of linguistic choice is critical here: has the scribe intentionally written the ledger in this fashion due to a consciousness about words containing -age? That is, does he have a mental paradigm that links these -age formations together? Amanda Pounder (2000) summarizes how and why the lining up of similar formations in texts often indicates word-formation paradigmaticity: In addition to psycholinguistic and historical evidence, a number of researchers have shown that paradigmatic relations between word-formations sharing a lexemic root are exploited as a cohesive device in texts… . the distribution and order of occurrence of words related through word-formation can be consciously manipulated as a stylistic device or as support of the development of an argument or thought … the “usual”, lexically fixed formmeaning combination can be replaced by another in order to obtain a series of similarly sounding formations (e.g., all stems with the same affix, with the same “ablaut” etc.); this strategy is connected to another sort of wordformation paradigmaticity, namely that of the set of lexemes in the same lexico-semantic class created by the same operation or at least by means of the same form rule. (83–4)

Pounder’s description of paradigms assumes that word-formation rules (i.e., rules that dictate the attachability of affixes to bases) pre-exist the creation of the text. They are a pattern a writer can exploit for different textual effects. In the case of Middle English morphology, we do not necessarily know a priori that the forms including -age from the ledger above were necessarily driven by a word-formation rule. In other words, since we have little direct evidence of -age existing as a productive suffix in this or earlier periods, it is unwise to interpret this ledger entry as an instance of creative application of a productive word-formation rule. It is not as if the scribe had a number of synonyms to choose from, and he merely picked those ending in -age to convey the pricing. In fact, these may have been the only linguistic options available.

Borrowed derivational morphology in Late Middle English

251

However, it is probable that the writer has recognized the similarity of form in writing Cranage, Wharuage, and cariage. And even if he has not, the readers of this text will likely see the formal resemblance, especially because the pragmatics of the ledger make the similarity so salient. The use of these forms creates a moment for potential re-analysis: even if the -age forms are not produced by a word-formation rule, these words suddenly look to a reader as if they might have been. Rather than perceiving these words as whole borrowings, readers may instead take them to be part of one suffixal paradigm – that is, the same suffix attached to three different stems. The recognition of analogous forms, as written and received in such textual examples, helps to solidify -age forms as a mental paradigm, a crucial step in the development of a productive affix. In 1436, the Grocers provide another example in which the ledger lines highlight the similarity of form of -age: (2)

Grocers (1436):

Also payde for Caryage of Ragge chalk and aschler … Also payde for Warvage of Ragge aschler chalk … .

In this juxtaposition of lines, the reader can visually see the analogous form and function of these two -age forms. This instance increases the likelihood that readers could interpret these words as part of the same suffixal paradigm. Elsewhere in the manuscript, the -age paradigm is even coordinated with a native, morphologically complex nominal within the same noun phrase. An entire ledger line at the top of one page reads as follows:

(3)

Grocers (1436): Also paid For costis Freigh[t] cariage Wharuage and pilyng up of ij shippes

Because the nouns cariage, Wharuage, and pilyng are syntactically parallel, the writer (and perhaps his readers) may intuit that the ending -age, like the proximate native suffix -yng, functions grammatically as a nominal marker. In syntactically parallel co-occurrences of borrowed and native forms, any

252

Chris C. Palmer

transparency in the productive, native form may reinforce (or at least suggest) the transparency and potential productivity of the borrowed ending. Similarly, we can read outward from other textual examples to try to reconstruct the mental lexical status of other potential affixes in borrowings. Consider the following example from the Goldsmiths: (4)

Goldsmiths (1436): … Robert Boso[un], citecein and goldsmyth of London, vilenously with malicious, sclandrous and heynous wurdis revylid and lyed John Pattyslee … . (italics mine)

On one hand, the Goldsmiths consistently employ -ous lexemes with negative semantic associations, most often in accounts of abuse. With one exception – the use of gracious in a copy of a letter from King Henry VI – the Goldsmiths employ -ous forms strictly to mark the abuse genre. In the example above, the -ous lexemes help the text cohere, linking the adverbial hybrid form vilenously with the adjectival series malicious, sclandrous, and heynous. Ultimately, though, the employment of these discourse-motivated lexemes and textually cohesive effects reveals a recognition of -ous forms similar in syntactic function, semantic sense, and morphological shape. These co-textual examples reveal a certain paradigmaticity for similar endings in borrowings. There are two possible implications of the recognition of analogous forms as seen in these texts: (1) English writers already had these forms mentally stored in a paradigm and were employing them because of their similar semantics and syntactic functions; and/or (2) readers of these records were encouraged to see the words as part of the same paradigm, even if they weren’t already mentally stored in this way. In this sense, these texts were part of a dynamic, diachronic force which helped readers within this community construct the linguistic consciousness of these borrowed word-endings as potential suffixes. This is an essential step in the analysis of suffixes – the awareness that a repeated form may potentially be a suffix.20

Context In addition to co-text, we can also look at the wider context of derivational examples to locate additional evidence for the potential affixal status of borrowed morphology. Consider the following examples from the Gold-

Borrowed derivational morphology in Late Middle English

253

smiths, the first of which occurs early in a 1436 memorandum and the second of which occurs many sentences later in the same text [italics mine]: (5)

Goldmiths (1436): … Þe same Robert Bosoun Þanne & Þere submiyttid & putte hym to stonde & obeye to Þe warde, ordinaunse, determinacion & dome of Þe said John Sutton, John Waryn, Robert Boteler with Þe assent & good avyce of Þe good men of Þe said cumpanye for Þe offense afforeseid.

(6)

Goldsmiths (1436): The said John Sutton, John Waryn, Robert Boteler, by assent and good avyce of many goode men of Þe cumpanye aforesaid, warde, ordeyne, determine & deme Þat Þe saide Robert Bousoun … .

Clearly the italicized forms have been stylistically ordered according to early legal rhetorical conventions. But the repetition of different formations involving the same bases, while helping different portions of the same text lexically cohere, also provides the modern reader with some insight into the status of the nominal endings for this medieval community. In examples (5) and (6) we see a series of nominals contrasted with their lexically equivalent verbal “roots,” indicating different word-formation processes: warde (n.) matches with warde (v.) (conversion), ordinaunse and determinacion complement ordeyne and determine (affixation), and dome contrasts with deme (vowel change). The implication here is that -aunse and -acion must be suffixes – they are detachable and directly linked to their bases through textual cohesion. More subtle contextual examples appear throughout these records: paymente appears within the same account as payde, meyntene appears in the same text as meyntenance, etc. Such examples suggest that these English speakers had a mental paradigm built around the same lexeme. The ordeyne paradigm, for example, includes both the verb ordeyne and its nominalization ordinaunse. The presence of this paradigm not only allows for the exploitation of cohesive textual effects, but it also reveals that -aunse has some sort of lexical status as a detachable unit. The Grocers’ records provide a similar example for -ment and -aunce, although it occurs in 1448 (slightly outside of the bounds of the quantitative study presented earlier). In the middle of a ledger, the scribe writes [italics mine]: (7)

Grocers (1448):

Item Payed to John Plomer for Alowaunce …

254

Chris C. Palmer

Then, further down the page, one sees the following written in smaller handwriting (though likely by the same hand probably after the page was finished): (8)

Grocers (1448):

Item Resseyued of the seyd John Blanche and John Plomer that was disalowed In the paymentes afore sayd …

In examples (7) and (8) the writer has employed textual cohesion for a significant additional entry not only with the metadiscursive afore sayd but also by employing two derivational paradigms: Payed gets linked to paymentes, and Alowaunce contrasts with disalowed. From this example the modern reader can infer the detachability of -ment, -aunce, and also the potential prefix dis-. This sort of evidence demonstrates another type of analysis of borrowings – the detachability of affixes. This analysis relies on a different sort of paradigm than the earlier co-textual examples: the repeated endings suggest a suffixal paradigm, whereas detachability of affixes depends on a lexemedriven paradigm, multiple words which share the same stem. While it is clear that the writer must be aware that these words are part of the same paradigms, these examples must also impact readers’ consciousness of the detachability of these endings. As language users increasingly see such endings as detachable, they are more likely to see them as re-attachable to different bases. Hence, potential detachability must be another analytical criterion in place for an ending to gain the status of a (productive) affix.

The lexical status of borrowed derivational endings for the Goldsmiths and Grocers Having considered the qualitative and quantitative analysis of different potential affixes in the records of these two communities, we are left with the following question: what is the lexical status of these (potential) borrowed affixes? The local productivity analysis above – which considered the aggregation of forms alongside their analyzability by medieval writers in specific contexts – suggests that -ance, -ment, -(c)ion, and -ous are in the middle stages of naturalization. They are limitedly productive within their restricted borrowed realms, but they are seen as analyzable, patterned, paradigmatic – not merely the random endings of borrowed words. These

Borrowed derivational morphology in Late Middle English

255

endings unmistakably have some sort of lexical status as morphological, or perhaps premorphological units. The ending -ity may have a premorphological status similar to its nominal peers within these communities, but this study did not find enough data to support this claim. Nevertheless, within the Grocers’ community, it is possible that -age, compared to its borrowed peers, is in fact further along the naturalization process and the path towards a wider domain of productivity. The ending has a fairly consistent aggregation, it has its own paradigm (realized in textual examples as stylistic parallelism and lexical cohesion) and, most interestingly, it is seen attached to two different native stems. For the Grocers it seems that -age is the most advanced in its potential as a suffix. There is a crucial implication to this conclusion about -age. I am not only arguing that -age is more naturalized and productive than the other nominal endings. I am also claiming that its lexical status in this time period differs in each community. There is evidence that it is recognized as a potentially productive ending for the Grocers, while there is no real evidence that it is anything other than part of a borrowing for the Goldsmiths. This claim is critically dependent on the notion of usage and its impact on grammar and the lexicon, especially as seen in Cognitive Grammatical (CG) accounts of language. In “Cognitive Approach to Word-Formation,” David Tuggy argues that The units of a language are conventional. That is, they are established by usage as shared by a community of people. All of language is in this sense usage-based, and usage is a central, not a peripheral concern of linguistics. (2005: 234)

He adds in a footnote, “Besides being central for conventionality, usage is crucial for the establishment of units in individuals’ minds.” According to CG, sheer frequency of use has a direct impact on the likelihood that a certain word-form will be stored as a unit in the speaker’s mind. Furthermore, Hay (2003) finds that the relative frequencies of derived forms to their base forms correlate strongly with the transparency and parsibility of complex words as well as the potential productivities of individual affixes. And Bybee (2007) argues that both token and type frequencies have direct effects on cognitive representations of morphology and, consequently, morphological development in language. Noting that “repetition strengthens memory representations for linguistic forms and makes them more accessible,” she asserts that high token frequency encourages forms to resist analogical

256

Chris C. Palmer

change and remain less interconnected in paradigmatic organizations within the mental lexicon (2007: 10–14). High type frequency is claimed to be a major determinant in the productivity of forms: The contribution of type frequency to productivity is the fact that when a construction is experienced with different items occupying a position, it enables the parsing of the construction. If happiness is learned by someone who knows no related words, there is no way to infer that it has two morphemes. If happy is also learned, then the learner could hypothesize that – ness is a suffix, but only if it occurs on other adjectives would its status as a suffix become established. (2007: 15)

According to Bybee, whenever a reader or listener encounters a suffixal paradigm in a text, s/he is further compelled to perceive the ending as a potentially productive suffix. Each written use of a lexical and suffixal paradigm has the potential to effect change in the status of relevant affixes in the minds of readers and listeners. Any assessment of the morphological status and productivity of affixes in different historical periods should thus consider usage patterns and token/ type frequencies in available records, and not simply because these texts reflect the grammar and lexicon of the literate individuals and communities who first produced them. Observable frequency patterns in written texts can be treated either as direct evidence of literate language use or, more cautiously, as indirect/hypothetical evidence of the types of lexical items available in certain forms of spoken discourse within a community at a particular point in time. The records of the Grocers and Goldsmiths are, after all, less “literary” and at times more colloquial than those of Chaucer and Gower. As such, accounts of frequency in non-canonical written material may also indicate the potential effects of usage not only on readers, but also on others in oral contact with the communities who have produced and used those texts. Under the light of these theories on usage and frequency, the potential affixes in the records of the Grocers and Goldsmiths may be examined with the following assumption: the more frequent and apparent a certain form is, the more likely it will achieve an independent status in the minds of medieval speakers within those communities.21 In the case of -age, it is hence not insignificant that the Goldsmiths use the form far less than the Grocers, who employ 51 -age tokens with 8 different types. Because the Grocers use the form more often – and because their ledgers emphasize the similarities

Borrowed derivational morphology in Late Middle English

257

in different types ending in -age – their texts are more likely to make -age salient as a potentially meaningful and/or productive unit in the lexicon. The Goldsmiths’ written records cannot have this impact on their readers, even if their spoken practice includes more common -age usage. The dissimilarity in written usage between these communities strongly suggests a different lexical status for this ending.

Final considerations Applying both quantitative and qualitative methods to study the local productivity of potential derivational affixes in two medieval economic records, this article draws the following conclusions: (1) The endings -(c)ion, -ance, -ity, and -ment are not fully naturalized as affixes in the English of the Grocers and Goldsmiths. But even though they are not yet seen as suffixes, -cion, -ance, and -ment are beginning to be analyzed as independent endings. In this period, they are premorphemes. (2) For the Grocers, -age is even closer to becoming a widely productive suffix than the other nominal endings. For the Goldsmiths, -age is less likely to be perceived as a potentially productive affix. (3) Productivity itself is a wider historical notion than the simple application of rules or the creation of words or, methodologically, counts of neologisms or hapax legomena as a surrogate for neologisms. The productivity of borrowed derivational morphology must begin in its early stages primarily via the processes of reanalysis and analogy on the part of speakers, listeners, writers, and readers. In the early stages of productive processes – the premorphological stages – the linguist must ask: how did this particular form come to be seen as a potentially meaningful unit? In later morphological stages, s/he would then ask: what kinds of stems or other morphemes can this form attach to, and can these combinatorial properties be explained? Descriptions and explanations of the (pre)morphology of different forms can be illuminated and expanded by studies of the local productivities of forms in different textual communities. (4) In determining the (pre)morphological status of affixes, it is hard to deduce earlier speakers’ perception of the status of these word-endings with quantitative data alone. But token, type, and new type counts are

258

(5)

(6)

(7)

(8)

Chris C. Palmer

critical in showing the general trends in usage of these forms. And usage, of course, must be active in a community in order for a certain form to be perceived as potentially productive, as part of the same suffixal paradigm. Whereas the quantitative analysis reveals historical trends in the individual types used within a community, the qualitative analysis is necessary for describing the level of naturalization and productivity of these endings. The types of evidence that help to characterize these endings’ morphological status include: (a) the structure and types of hybrid forms; (b) the textual linking of forms with the same ending (suffixal paradigms); and (c) the textual linking of forms with the same stem (lexemic paradigms), indicating the potential detachability of the endings. This study provides a methodology in which the evidence for the morphological status of affixes derives primarily from qualitative data and is reinforced by usage trends observed in the quantitative analysis. Nevertheless, there is a degree to which qualitative analyses should also inform future quantitative studies of such questions. To further study borrowed derivational morphemes, a next step might include statistical counts of suffixal and lexical paradigm occurrences themselves, to determine the aggregate impact of these larger discursive structures in texts as a whole. The analysis of the local productivity of potential suffixal forms has the most explanatory power for these specific communities within this time period. One should not immediately assume that the results can be generalized to every other community in the fifteenth century. However, linguists can certainly employ a similar methodology to investigate in other communities’ texts the sorts of specific evidential types outlined above. That is, a variety of close readings of texts may help us complete – albeit in small steps – the overall picture of the diachronic development of borrowed derivational morphology. Historical (written) texts not only reflect linguistic consciousness; they effect it.

This final point has important implications for how historical linguists think of language change as a more general phenomenon. Because all we have from earlier stages of history is written records, we must assume in some way that these written usages of language reflect the consciousness (and grammar) of the writer.22 But we must not overlook the impact writing can

Borrowed derivational morphology in Late Middle English

259

have on shaping linguistic consciousness. While the readership of such records is likely small, we do know that the Grocers read regularly from their records to the entire membership: the ordinances written in English “were read to the assembled members of the Company on every quarter day” (Nightingale 385). When a scribe writes three forms of -age together on a ledger, or links determine to determinacion in a rhetorical flourish, we can assume that he is writing his recognition of the detachability and similarity of word-endings onto the page, even if this is not his primary intent. But when readers view or audiences hear these examples, they are given the chance to reanalyze and rethink these forms – to recognize, even in an intuitive sense, that these endings may actually be meaningful, independent units in the language.

Notes *

1. 2. 3. 4. 5.

6.

7.

I would like to thank the editors and readers for their suggestions for revising this article. The readers’ comments resulted in an extension of the discussion of the social context of the Grocers’ and Goldsmiths’ texts, including considerations of social network theory. Elizabeth Traugott referred me to the excellent study by Fleischman (1976) on the cultural dimensions of -age usage. And David Denison's comments encouraged further discussion of the impact of token frequency on the lexical status of morphological forms, including references to Hay (2003) and Bybee (2007). Oxford English Dictionary (OED) Urban Dictionary. Entry 3. By Ghostpigeon, July 23, 2005. http://www.urbandictionary.com/define.php?term=ownage (March 2006) Burnley does not provide a label for this general process, so I have referred to it as naturalization. http://www.library.utoronto.ca/utel/ret/cawdrey/cawdrey0.html (March 2006) And furthermore, intentional formations do provide some sense of the domain of applicability of a certain morpheme. The playful formation tippage (listed above) shows semantic and morpho-syntactic similarities to the more widely used signage. A related problem with small corpora is the case in which one particular text or author exhibits a particular form in larger-than-usual numbers, possibly skewing the results of a supposedly “representative” study of this form in a certain period. There will be occasional reference to verbal, adverbial, and adjectival affixes as well.

260

Chris C. Palmer

8. Cf. Curzan and Palmer (2006) for a fuller discussion of complementary studies of principled vs. unprincipled, digital vs. non-digital, and large vs. small corpora. 9. Here I deliberately define corpus in the more traditional sense of any ‘body of texts,’ whether they are in digital, print, or manuscript form. This terminological choice reinforces my commitment to rethinking corpus linguistics as a discipline that considers all sorts of textual uses of language. 10. The Grocers’ accounts also provide accompanying mimeographs of the original manuscript folios. 11. While their records make these functions clear in a general sense, the company’s website also provides confirmation of their history: http://www.thegoldsmiths.co.uk/company/index.htm 12. http://www.grocershall.co.uk/company.html (March 2006). According to the company site, the “grossers” were originally the wholesale keepers of inventory, and the small shopkeepers bought from them and sold to customers at retail. Eventually, these smaller store owners came to be known as grocers. 13. Alexander Bergs (2005) has managed to adapt social network theory to examine morphosyntactic questions in the case of the Paston Letters. But the genre of letters is singular in its allowance of analysis of non-anonymous, individual uses of language within a community context. 14. The “provincial members were from places as far afield as Banbury, Shrewsbury, Ipswich, Cambridge, Cornwall, Essex, and Conventy, while later there was one from Bristol” (Nightingale 1995: 395). 15. In this context, suggestion and reparacioun are considered two different types of the affix -(c)ion since the lexemes are different. Under this schema, reparacioun, reparacion, reparaciouns are considered tokens of the same type, since each form is an orthographic or inflectional variation of the same lexeme (i.e., each has the same stem). 16. The OED explains that endings -ance and -ence are Latinate orthographic variants of the same form. While it may be possible that some speakers perceived these endings as distinct morphemes, in this study I have presented all tokens of these endings under one heading: -ance. Furthermore, occurrences of almost every type in the entire chart appeared as word-endings within these records – in rare cases (Grocers: Worshipfull), the suffix was followed by an additional suffix. 17. While dictionaries such as the Oxford English Dictionary (OED) or the Middle English Dictionary (MED) are useful in dating occurrences of new words in English, they are not necessary for tracking newly introduced forms within a community’s records. Of course, historical dictionaries offer early attestation dates for lexical items and could provide, in theory, a general picture of the Grocers’ and Goldsmiths’ neologisms relative to occurrences of the same or similar forms in other texts within the same period. However, as Cowie

Borrowed derivational morphology in Late Middle English

18.

19.

20.

21.

22.

261

(1998a: 71–72) argues, dictionaries do not reliably reflect all text types, nor do they exhaustively catalogue all instances of productive word-formation patterns available in the language of different periods and communities. Hence, in this study dictionaries were consulted for incidental evidence and for background information on selected (especially ambiguous) types. The effect of individual scribes, with varying idiolects and idiosyncracies, must have also impacted the use of these borrowings. Nevertheless, it is difficult to paint a clear picture of the exact impact a scribe would have in the aggregate patterning of word-endings. In the MS, the ledger line is broken so that cariage continues onto the next line. Even though all three -age derivations are not collinear, their immediate co-occurrence on the ledger line is still salient. We must acknowledge, however, that the readership for these texts was likely small, restricted to the communities that produced them. But it is still significant to point out the potential local effects for this community, particularly if other Middle English texts employed similar uses of analogous forms. Historical linguists will need to consider the aggregate effects of these texts on English speakers’ perceptions of these endings of borrowed words. At present, this assumption needs more theoretical backing from language processing studies that specifically investigate the cognitive status of derivational morphemes. Much of Bybee (2007) treats inflectional morphology with far less attention to derivational morphology. Moreover, her claims about the effects of token frequency on mental representations and the autonomy of forms revolves much more around the status of words and phrases than the mental status of individual bound affixes. The added difficulty, of course, is that speech exhibits different grammatical features than does writing. Cf. Biber et al. (1999). At some point, though, all historical linguists must assume that some features from writing must have occurred in speech as well. But ultimately, we must be comfortable with our fundamental inability to know the grammar of speech in times past. We should in fact be more attuned to the specifics of writing as a distinct and viable form of language.

262

Chris C. Palmer

References Baayen, Harald 1992 Quantitative Aspects of Morphological Productivity. In Yearbook of Morphology 1991. G. Booij and J. van Marle (eds.), 109–149. Dordrecht: Kluwer. Bauer, Laurie 2005 Productivity: Theories. In Handbook of Word-Formation. Pavol Stekauer and Rochelle Lieber (eds.), 315–334. (Studies in Natural Language & Linguistic Theory.) Dordrecht: Springer. Bergs, Alexander 2005 Social Networks and Historical Sociolinguistics: Studies in Morphosyntactic Variation in the Paston Letters (1421–1503). Berlin/New York: Mouton de Gruyter. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan 1999 Longman Grammar of Spoken and Written English. Harlow, England/New York: Longman. Burnley, David 1992 Lexis and Semantics. In The Cambridge History of the English Language, Volume II: 1066–1476. Norman Blake (ed.), 409–499. New York/Cambridge: Cambridge UP. Bybee, Joan 2007 Frequency of Use and the Organization of Language. Oxford/New York: Oxford UP. Cowie, Claire 1998a Diachronic word-formation: a corpus-based study of derived nominalizations in the history of English. PhD. diss., Cambridge: University of Cambridge. 1998b The Discourse Motivations for Neologising: Action Nominalization in the History of English. In Lexicology, Semantics, and Lexicography: Selected Papers from the Fourth G. L. Brook Symposium, Manchester, August 1998. Julie Coleman and Christian J. Kay (eds.), 179–207. (Amsterdam Studies in the Theory and History of Linguistic Science 194.) Amsterdam/Philadelphia: John Benjamins. Cowie, Claire and Christiane Dalton-Puffer 2002 Diachronic Word-Formation and Studying Changes in Productivity over Time: Theoretical and Methodological Considerations. In A Changing World of Words. Javier E. Diaz Vera (ed.), 410–437. Amsterdam/New York: Rodopi.

Borrowed derivational morphology in Late Middle English

263

Curzan, Anne and Chris C. Palmer 2006 The Importance of Historical Corpora, Reliability, and Reading. In Corpus-based Studies of Diachronic English. Roberta Facchinetti and Matti Rissanen (eds.), 17–34. Bern: Peter Lang. Dalton-Puffer, Christiane 1996 The French Influence on Middle English Morphology: A Corpusbased Study of Derivation. Berlin/New York: Mouton de Gruyter. Fleischman, Suzanne 1977 Cultural and Linguistic Factors in Word Formation: An Integrated Approach to the Development of the Suffix -age. Berkeley/Los Angeles/London: U of California P. Gadde, Frederik 1910 On the history and use of the suffixes -ery (-ry), -age, and -ment in English. PhD. diss., Lund: Berlingska boktryckeriet. Goldsmiths’ Mistery 2003 Wardens’ Accounts and Court Minute Books of the Goldsmiths’ Mistery of London 1334–1446. Lisa Jefferson (ed.) Woodbridge: Boydell. Grocers’ Company 1886 Facsimile of first volume of MS. archives of the Worshipful Company of Grocers of the city of London, A.D. 1345–1463. London: Grocers Company (Richard Clay). Hay, Jennifer 2003 Causes and Consequences of Word Structure. New York/London: Routledge. Marchand, Hans 1969 The Categories and Types of Present-day English Word-formation. Second edition. Munich: Beck. Miller, D. Gary 1997 The morphological legacy of French: Borrowed suffixes on native bases in Middle English. Diachronica 14 (2): 233–264. 2006 Latin Suffixal Derivatives in English and their Indo-European Ancestry. Oxford/New York: Oxford UP. Milroy, Lesley 1980 Language and Social Networks. Baltimore: University Park. Nightingale, Pamela 1995 A Medieval Mercantile Community: The Grocers’ Company & the Politics & Trade of London, 1000–1485. New Haven: Yale University. Pounder, Amanda 2000 Processes and Paradigms in Word-Formation Morphology. Berlin/ New York: Mouton de Gruyter.

264

Chris C. Palmer

Reddaway, T. F. and Lorna E. M. Walker 1975 The Early History of the Goldsmiths’ Company, 1327–1509. London: Arnold. Rothwell, William 2001 English and French in England after 1362. English Studies: A Journal of English Language and Literature 82 (6): 539–59. Schultink, Henk 1961 Produktiviteit als morfologisch fenomeen. Forum der Letteren 2: 110–125. Thomason, Sarah Grey and Terrence Kaufman 1988 Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California. Tuggy, David 2005 Cognitive approach to word-formation. In Handbook of WordFormation. Pavol Stekauer and Rochelle Lieber (eds.), 233–265. (Studies in Natural Language & Linguistic Theory.) Dordrecht: Springer.

Fixer-uppers and passers-by: Nominalization of verb-particle constructions

Don Chapman 1.

Introduction

The -er suffix has been one of the most productive derivational suffixes in English. Throughout its history, it has regularly been used to create nominalizations of verbs, usually as agentives like farmer (< to farm), teacher, banker, researcher, writer.1 The suffix is so productive today that the main restrictions on it seem to be pragmatic or blocking restrictions: practically any verb that can have an agentive referent and that does not already have one (like cook < to cook) can receive the -er suffix. But the productivity of -er is less established with one verb type that has received increasing attention, namely the multi-word verbs, like pick up or fight back.2 English speakers ought to be able to use the -er suffix with these verbs, but where does the suffix go? With the verb (picker-up, fighter-back), since -er regularly attaches to simple verbs? With both the verb and particle, as attested in such well-established terms like fixer-upper and perhaps picker-upper (a term made famous in an advertizing campaign)? With just the particle (pick-upper), since the suffix usually attaches to the end of a term? How speakers form these nominalizations today and in the past is the question this paper seeks to answer. But it is not an easy question to investigate; -er nominalizations of multi-word verbs have proven elusive in the standard corpora of historical and contemporary English, as will be discussed below. It is not that the inputs are rare: both multi-word verbs and -er suffixes have been common for centuries. It is the combination that appears to be an odd construction – one that will require less typical sources of evidence and methods to find and examine. In a volume focused on methods and evidence, this paper exemplifies the challenges of investigating a construction that has remained unconventional enough to elude most public writing. Instead less formal, less established writings must be found and used for purposes that they

266

Don Chapman

were not designed for. In jury-rigging such unconventional corpora, as will be discussed below, several compromises will have to be made in method; finding token frequency or hapax legomena or first occurrences of wordformation patterns may simply not be feasible or at least meaningful, so typical strategies of measuring productivity may not be available. But under the assumption that something is better than nothing, especially for an understudied phenomenon like picker-upper constructions, this present study derives less conventional methods, as it still tries to measure, or at least suggest, the productivity of the different nominalization patterns. Not much has been written on this topic. Some studies have noted other nominalizations of multi-word verbs, like picking up, and pick-up (Lindelöf 1938; Sørensen 1986; Fraser 1976: 27–29), but apparently very little has been said about the -er nominalizations. The standard reference works on word-formation mention this construction, but some of them just barely. Jespersen (1942) lists four possibilities: (1)

1) Prepose the particle and add -er to the verb: on-looker 2) Leave the particle after the verb and add -er to the verb: looker-on 3) Leave the particle after the verb and add -er to the particle: comeouter 4) Omit the particle and add -er to the verb: listener (from listen to/in), waiter (from wait on) (Jespersen 1942: 234–35)

Jespersen does not even mention the pattern picker-upper. Neither do later handbooks: Marchand (1969), Adams (1973), and Quirk et al. (1985) only mention Jespersen’s 1) on-looker and 2) looker-on. It isn’t until 1983 that Bauer gives picker-upper as another type of nominalization, when he mentions two more possibilities: 5) add -er to both the verb and the particle: breaker-inner, cleanerupper 6) omit -er from both the verb and the particle (conversion): drop-in (Bauer 1983: 288-89) Beyond these short references in the handbooks, there does not seem to be much else said about this construction. Of the six possibilities listed by Jespersen and Bauer, two can be dismissed in this paper because they lack either the particle or the -er suffix, namely the patterns picker (1: 4) and pick-up (1: 6).3 These two patterns

Fixer-uppers and passers-by

267

will not be considered further in this paper. The four remaining patterns which use an -er suffix and a particle and which will be treated in this paper are by-stander (1: 1), picker-up (1: 2), pick-upper (1: 3), and picker-upper (1: 5).4 The multi-word verbs that can serve as inputs for these nominalizations are wide-ranging and varied (Claridge 2000: 46–82), but for this paper only those with a single particle will be treated. Following the terminology of Biber et al. (1999), such multi-word verbs are called intransitive phrasal verbs, transitive phrasal verbs (or separable phrasal verbs), and prepositional verbs (or inseparable phrasal verbs). The intransitive phrasal verbs have no object, like wake up. The transitive phrasal (or separable) verbs have an object that can occur before or after the particle, as in tear down a wall vs. tear it down, and finally the prepositional (or inseparable) phrasal verbs take an object which always follows the particle (consult with clients, consult with them). In principle we have four patterns of nominalizations (by-stander, picker-up, picker-upper, picker-up) and three subcategories of multi-word verbs that serve as inputs.

2.

Multi-word verb nominalizations in ICAME and BNC

The initial question of this paper is which of all these patterns have occurred in the history of English. A recent word-formation study offers some slight evidence for the popularity of the picker-upper pattern today.5 Two hundred respondents were asked to choose a term for the following prompt: “A person who frequently interrupts other people when they are talking: a. interrupter b. interruptist c. interruptant

d. e. f. g. h.

butt-in butter-in butter-inner butt-innist butt-insky

i. cut-in j. cutter-in k. cutter-inner l. cutt-innist m. cutman

n. interposer o. interposist p. other: _____

Thirty-two of these respondents chose a nominalization of a multi-word verb, according to the following distribution:

268

Don Chapman

Table 1. butter-in butter-inner other (butt-inner) cutter-in cutter-inner TOTAL

Native 1 12 1 1 1 16

Near Native 0 3 0 0 2 5

Non-Native 3 3 0 4 1 11

TOTAL 4 18 1 5 4 32

The overwhelming majority of native or near-native speakers chose the picker-upper pattern, while the picker-up pattern was chosen mainly by non-native speakers. But just how popular has the picker-upper pattern been, in both the past and present? The responses above were given to just one question that was not even specifically designed for studying multi-word verbs.6 To gauge what speakers have done in the past and what they do now, we need more wide ranging evidence, like the kind that usually comes from historical and synchronic corpora. Unfortunately, nominalizations of multi-word verbs rarely show up in any of the ICAME corpora, whether historical or contemporary, as the following lists demonstrate (the lists are meant to be exhaustive for each corpus7): Nominalizations from Historical ICAME Corpora – CEECS: brynger up, lokers on – Helsinki: driver away, drivers forth, loker on, standers about – Innsbruck letters: comers between – Innsbruck sampler: brynger forth, makeres up – Lampeter: stander-by – Newdigate: standers-by – Old Scots: uplifter, upraiser, indweller Nominalizations from Contemporary ICAME Corpora – ACE: diner out, onlooker, passer-by, runner-up – Brown: bystander, hanger-on, onlooker, passer-by, runner-up – Kolhapur: bystander, hanger-on, onlooker, passer-by, runner-up – LOB: passer-by, runner-up, taker-up, washer-up

Fixer-uppers and Passers-by

269

Only a few nominalizations show up in these corpora, and many of those are simply highly conventionalized terms like runner-up, passer-by, and hanger-on. These occurrences are too few to draw any conclusions, except perhaps that the construction has been rare in English and that the only patterns that have occurred have been the picker-up and bystander patterns. Part of the problem is simply the size of these corpora. For investigations into rare word-formation patterns, the corpus has to be large enough for the construction to have a chance to occur. The ICAME corpora are probably too small for such investigations. Yet the British National Corpus (BNC) also turns up relatively few instances of such nominalizations (again the list is meant to be exhaustive): Nominalizations from the BNC – picker-up pattern: holder up, passer-by, runner-up, sweeper up, teacher out, washer up, waker up – by-stander pattern: backslider, bystander, downtoner, incomer, inlier, off-breaker, onlooker, onsetter, outlier, outfitter, overlooker, overruler, upholder – picker-upper pattern: washer upper A few of these words, like holder up and sweeper up, go beyond the conventionalized forms and look like they have been produced from a productive process, but only a few; most have long been a conventional part of the English lexicon. The explanation for so few terms cannot be the size of the corpus, for the BNC is fairly large at 100 million words. Instead the explanation probably owes more to the status of the nominalizations of multiword verbs. Those nominalizations probably are felt to be too colloquial, too non-literary to be included in the writing that makes up the bulk of all the corpora. Indeed, the multi-word verbs themselves are often felt to be informal or colloquial (Claridge 2000: 103). The nominalizations have the additional obstacle of being derivations, and odd-looking ones at that. They frequently have the feel of being ad hoc or ludic in some way; they feel less acceptable, more a nonce form, more self-consciously creative than un-selfconsciously productive. We would need to search for these nominalizations where writers feel less inhibited in using them. They will have to be found in less conventional writings. Such records are hard to come by for historical stages of English. Perhaps the best source will be the Oxford English Dictionary (OED), since citations have been gathered from a variety of sources specifically to illus-

270

Don Chapman

trate a wide range of words. In principle, the OED includes citations from many sub-literary texts, though admittedly the citations for earlier periods tend to be more literary than the later periods (Brewer 2000: 49–56). Still, the OED is a better source than the readily available historical corpora, if for no other reason than it already represents the compilers’ searches for words. It helps that the citations can be searched, so the entire OED can be used as a corpus. In contrast to the historical sources, informal writing is not hard to come by in contemporary sources. We have an abundance of unconventional, informal writing on the internet. The world-wide web is much larger and less edited than the texts used for the BNC, so we should expect more colloquial or nonce forms to make their way into the writing. But the internet comes with its own difficulties, not least in creating and controlling searches. The next two sections will present some of the difficulties and results from the OED and internet respectively.

3.

Multi-word verb nominalizations in the OED

Indeed, many multi-word verb nominalizations are found in the OED, as shown in Table 2 below. This at least gives us an idea of the repertoire of constructions available to speakers. But dictionaries have their limitations for measuring productivity. Performing token or type frequency analysis is practically impossible, even when treating the citations as a corpus, because the citations have been selected specifically to illustrate particular words. We cannot necessarily assume that the frequencies in the OED are indicative of the use in general. This is a restriction, when trying to measure productivity, since there is little attempt to differentiate frequent words from rare words in the OED and the occurrence of a new type in a dictionary does not necessarily reflect the operation of a productive word-formation rule (cf. Baayen 1992). But since nominalizations of multi-word verbs occur too rarely in corpora to measure their productivity by the number of hapax legomena or appearance of new types, we are mainly left with simply capturing the types that have been created in the past and deducing from those which patterns were more-or-less productive. The OED is at least useful for capturing such types, and in the absence of better, it can be pressed into service.

Fixer-uppers and Passers-by

271

The distribution of the four patterns in the OED is given in the following table, where the numbers represent the first appearance of new words (types) for each century. Table 2. by-stander pattern picker-up pattern picker-upper pattern pick-upper pattern

1400 5 7 0 0

1500 8 44 0 0

1600 14 84 0 0

1700 7 53 0 0

1800 23 142 0 0

1900 8 54 13 0

Total 65 384 13 0

When normalized per million words, the distribution looks like this:8 Table 3. by-stander pattern picker-up pattern picker-upper pattern pick-upper pattern

1400 1500 3.6 2.5 5.0 13.8 0.0 0.0 0.0 0.0

1600 2.7 16.4 0.0 0.0

1700 1.9 14.4 0.0 0.0

1800 2.2 13.6 0.0 0.0

1900 0.7 4.9 1.2 0.0

The normalized distribution is seen in Figure 1: Nominalization Patterns 18 per million words

16 14 12

by-stander pattern picker-up pattern picker-upper pattern pick-upper pattern

10 8 6 4 2 0 1400

1500

1600

1700

Centuries

Figure 1.

1800

1900

Total 13.7 68.1 1.2 0

272

Don Chapman

The distribution is skewed to some degree for the early periods, especially the 1400s, since my searches incorporated few variant spellings. The effect of this skewing is mainly in the starting point of the picker-up pattern. Even with more variant spellings, we would probably still see growing use in the fifteenth and sixteenth centuries, but perhaps not as stark as the curve here would suggest. The variant spellings will make practically no difference in the later periods, however, so apart from not being able to deduce with confidence just how rapidly these constructions took hold in English, little is lost by their absence. What becomes immediately apparent in Table 3 is that the picker-up pattern has been the most productive pattern throughout the history of English. The second most occurring pattern is bystander, which started out as a relatively productive pattern, but has declined in the past two centuries. These are precisely the two patterns that most word-formation handbooks have noticed. The pattern that escapes notice, picker-upper, is a newcomer: most of the occurrences occur in the 1940s and later, which would explain why Jespersen does not include this pattern in his list of possibilities for nominalizing multi-word verbs. The construction apparently was too new for him to notice. Jespersen had noticed the fourth pattern, pick-upper, however, even though there are no occurrences in the OED. The sharp decline of the picker-up pattern in the twentieth century and the continual decline of the bystander pattern also stand out in this distribution. This decline in the twentieth century occurs at the same time as the increase of the picker-upper pattern. It could be that the two trends are related and that picker-upper is supplanting the two older patterns. The distribution of these patterns changes slightly when they are examined by the types of multi-word verbs that serve as inputs (Table 4 and Figure 2). In other words, the nominalizations of transitive and intransitive phrasal verbs have a different distribution from those derived from prepositional verbs. In this distribution, the prepositional verbs stand out as being more stable than the transitive phrasal verbs for centuries, but then they drop off more precipitously. Both the intransitive and transitive phrasal verbs show a couple of peaks, then a sudden drop in the 1900s. The peak in the 1600s could well owe to the linguistic creativity of the period. In other word-formation patterns, the 1500s and 1600s proved to be times of much variability (Chapman and Skousen 2005; Kwon 1997). The peak in the 1800s may also reflect a genuine increase in the use of multi-word verb nominalizations, especially in technical vocabulary, but it may also be an

Fixer-uppers and Passers-by

273

Bystander + Picker-up Normalized 10.0 9.0

per million words

8.0 7.0 6.0

intransitive phrasal

5.0

transitive phrasal

4.0

prepositional

3.0 2.0 1.0 0.0 1400

1500

1600

1700

1800

1900

1600 1.4 9.2 8.8

1700 0.8 6.8 8.7

Centuries

Figure 2. Table 4. Intransitive Phrasal Transitive Phrasal Prepositional

1400 0.7 7.9 1.4

1500 0.6 6.9 9.1

1800 1.4 9.1 5.8

1900 0.5 3.5 1.6

artifact of the OED. In other words, it may be that more nominalizations show up in that century because that is when the editors were amassing their citations and word-lists, so they noticed more sub-literary texts and words (Brewer 2000: 46). The data from the OED, then, indicate that the picker-up pattern is the most common pattern in the history of English. The by-stander pattern is common enough to be noticeable, but it shows a steady decrease throughout the centuries. Both these patterns decline significantly in the twentieth century. The few occurrences of the picker-upper pattern in the OED are all in the twentieth century, suggesting that it is a fairly new construction. The pick-upper pattern does not occur at all, suggesting that if it has occurred in the history of English, it has been too unconventional to be noticed by the OED compilers.

274

4.

Don Chapman

Multi-word verb nominalizations on the internet as measured with Google

How are multi-word constructions nominalized today? As noted, these nominalizations do not show up readily in edited, printed work. The few forms that occur in the contemporary ICAME corpora and the BNC are mostly of the picker-up pattern, but most of these are limited to the highly lexicalized types like runner-up, passer-by, and washer-up. The twenty-one types occurring in the BNC are simply too few to draw any conclusions about today’s usage. So corpora, even large corpora like the BNC, are of limited use for investigating the use of such nominalizations. The internet would be a much more useful tool since the writing reflects less conventionalized choices. Unfortunately the limitations of the popular search engines make internet searches difficult. Most search engines do not allow substring searches, so looking for any word ending with -er followed by another word ending in -er was not an option the way it was for the OED searches. Most search engines also limit the number of total hits and hits per search that will be returned. Some even limit the number of searches that can be performed within a given time period. Given these limitations, it was not possible to search widely for all possible nominalizations. It was possible, however, to search for particular nominalizations, like finder out, finder outer, find outer, and out-finder. The strategy used in this paper, then, was to search for all four possible nominalization patterns for a number of specific multi-word verbs. Rough notions of productivity for each pattern can then be formulated by comparing the tallies of each pattern for all the multi-word verb inputs. This is an analysis of types, not tokens, so that if a nominalization occurs even just one time, it is still counted in the tally. The list of multi-word verbs that served as inputs for the search terms came from the BNC, by taking every combination of a verb followed immediately by a particle.9 This resulted in 384 multi-word verbs, each of which were given four forms, like wearer out, wearer outer, wear outer, and out-wearer. These 1536 terms were then used as search terms for Google. Ideally, the total number of hits for each term could have been used to analyze its token frequency. For example, if the search for wearer out produces 292 hits, and the search for wearer outer yields 175, we would like to say that wearer out is a slightly more popular nominalization than wearer outer. But there are two problems with such an approach. First, the numbers reported by the search engines are not necessarily reliable

Fixer-uppers and Passers-by

275

(Andersen 2002: 44; Véronis 2005). Second, the hits contain numerous false positives, like “far and away the most generous giver in history” where giver and in do not represent a nominalization of a multi-word verb, and the percentage of false positives varies greatly with each pattern. False positives are relatively infrequent for the bystander and picker-upper patterns, since nominalizations on these patterns usually do not resemble other syntactic constructions. Nominalizations of the picker-up and pick-upper patterns, on the other hand, frequently resemble other syntactic constructions, as illustrated by the “best giver in history.” The combination of unreliable reports of total hits and varying rates of false positives makes it hard to do any token analysis of these nominalizations. That is a pity. A token analysis would be more informative than a type count for telling which patterns speakers find more comfortable. Counting two patterns as equal when a type of each occurs is misleading, if one type has 500 tokens in the first 1000 hits and the other has only 2. But again, something is better than nothing, and type counts will still give us some basis for telling which patterns are favored by speakers. The method for obtaining the type counts for each search term was to scrutinize the search results for any true positives. If a true positive occurred, the tally for that pattern was increased. Since Google returns only 100 hits at a time, the search proceeded in rounds of 100 hits each. The results of the searches are given in Table 5 below. The numbers below each column represent the number of additional types that show up as a true positive in that round that had not occurred in previous rounds. As the table shows, few new types were added after the first three rounds, so only five rounds of the possible ten were examined.10 Table 5. Pattern

Round 1

Round 2

Round 3

Round 4

Round 10

Total

picker-up picker-upper pick-upper

115 266 120

27 7 23

13 1 14

4 0 6

3 0 1

162 274 164

The results, then, show that the picker-upper pattern is the most popular, as the survey in Table 1 had suggested. Whereas only a handful of types of the picker-upper pattern could be found in the BNC, on the internet, 274 (71%) of the 384 multi-word verbs occur with a suffix on both the verb and parti-

276

Don Chapman

cle. The most popular pattern in the OED, the picker-up pattern, still remains productive and forms new nominalizations, but it is not as popular as the picker-upper pattern. The pattern that produced no types in the OED or the BNC, pick-upper, turns out to be popular on the internet – just as popular as the picker-up pattern, in fact. Finally the pattern bystander is practically unproductive today. Two of the three that show up are self-conscious creations in poetry and fiction: 11a. downsinker sun in a sheen of paraffin and time that forges its 11b. the Angel of love, his next ontaker, standing strong he decides to engage These internet results reveal that the patterns picker-upper and pick-upper are much more robust today than corpora of published writing would suggest. How long such constructions have been in the language is harder to determine. The data from the OED suggest at least some diachronic development – a few occurrences of the picker-upper pattern show up late in the historical record, at about the same time that the dominant picker-up pattern declines. Probably part of the reason that we see more instances of the picker-upper and pick-upper patterns on the internet is their increasing popularity today. But that is only part of the explanation. Another large part must be the register differences between the OED and the internet. For roughly the same time period, namely the late twentieth and early twentyfirst centuries, the OED and internet have vastly differing distributions of the picker-upper and pick-upper patterns. Most likely, the OED still remains more conservative and less colloquial than the internet. What this likely means is that the picker-upper and pick-upper patterns probably have been relatively recent innovations, but those innovations have led out in more colloquial language. They have been around at least long enough to have become the most popular patterns in colloquial language today. How these patterns could have come about probably has to do with developments both with the -er suffix and with multi-word verbs. Ryder (2000: 293–96) has shown that the -er suffix has become attached to an increasingly wider set of bases, until today it can be attached to such constructions as prepositional phrases (out-of-towner, by-the-roader), conjoined prepositions (down-and-outer, up-and-downer), verb phrases (do-ityourselfer, get-rich-quicker), and so on. The attachment to a particle in a multi-word verb is another instance of the wider set of bases that -er has taken. Whereas in earlier stages of English, -er was mainly limited to verbs,

Fixer-uppers and Passers-by

277

it has now become generally available for many types of bases, so attaching it to the particle instead of or in addition to the verb in multi-word verbs should not be difficult for speakers. At the same time, the constituency of the verb and particle may well have become stronger over the years for multi-word verbs. As speakers identify the verb and particle as a single term, perhaps the less objectionable they would find attaching the suffix to the end of the term, even though a particle is in that end position. If indeed the suffix is showing up on the particle because the multi-word verb is regarded as a single term, this construction would be further evidence of grammaticalization in multiword verbs. Brinton and Traugott (2005: 123) argue that the particles of phrasal verbs are grammaticalized, in that they usually show aspectual or situational meaning, instead of their original spatio-temporal meaning. Typical particles in phrasal verbs are up, out, over, down, and out. In contrast, the prepositions in prepositional verbs, like in and on, usually retain a spatio-temporal linking meaning. Within the internet data, the particles that most take the -er suffix, whether in the picker-upper or the pick-upper pattern, are nearly all particles typical of phrasal verbs, like back, down, off, out, and up, as shown in Table 6.11 In contrast, the prepositions in prepositional verbs, like about, along, around, round, and through all have low numbers of instances with the -er suffix. Two prepositions, in and on, can be part of a prepositional verb or phrasal verb. Both have a fair number of Table 6. Particle about along around back down in off on out over round through up

Picker-up pattern 4 1 2 7 18 16 12 13 31 8 4 2 33

Picker-upper pattern 0 0 0 22 37 19 20 11 74 1 2 0 84

Pick-upper pattern 0 0 0 6 21 10 10 3 23 0 2 0 67

Bystander pattern 0 0 0 0 1 0 0 1 0 1 0 0 0

278

Don Chapman

-er instances, yet even their number of types is proportionately lower with respect to the picker-up pattern than that of the particles mainly limited to phrasal verbs. It appears that the -er suffix can most readily attach to particles that are more likely to undergo grammaticalization and that the attachment of a suffix on these particles would be a further step in their grammaticalization. In the case of picker-upper, there remains the question of why speakers would double mark a construction. Why are two -er suffixes used? According to Ackema and Neeleman (2000), the nominalization of phrasal verbs is similar to agentive suffixes that children attach to compounds as they acquire the -er suffix. A crucial stage in the acquisition of compounds, according to Clark et al. (1986) is acquiring the N-V compound order (ballkick-er). Before that stage, children keep the verb before its noun object in compounds, just like the order of syntax. When they attach the -er suffix to this type of compound, they end up making the same constructions as seen with nominalizations of verb-particle constructions: (12) a. a giver-present (someone who gives a present) b. a dry-hairer (someone who dries hair) c. mover-boxer (someone who moves boxes) The advantage of (12a) is that the -er is attached to the kind of word it usually attaches to, namely a verb. The disadvantage is that it interrupts the structure of the verb-noun compound and it places the wrong noun in the head position (the person is a giver, not a present). The advantage of (12b) is that it comes at the end of the construction as expected, but the disadvantage is that it is attached to the wrong kind of word. The advantage of (12c) is that it is attached to the right kind of word (verb) and it occurs in the right place (the end). The disadvantage is that it is repeated. This same analysis holds for the suffixation of multi-word verbs. Speakers have to put the suffix somewhere. In the oldest forms, speakers had available to them the bystander pattern, which resembles the adult form of compounds (hair-drier, box-mover). There is no problem attaching the suffix in these constructions, because the word that the suffix should attach to (the verb) is also in the position where the suffix should attach (the end). But as the on-look type of multi-word verb has declined (Claridge 2001: 86–89; Brinton 1988: 204–14), so too has the pattern of nominalization. The remaining three patterns (picker-up, picker-upper, and pick-upper) all have the problem of where to attach the suffix. Originally, it appears that

Fixer-uppers and Passers-by

279

speakers preferred attaching it to verbs, presumably because that is where suffixes go on simple verbs. At some point – by the middle of the twentieth century at the latest – speakers began to double mark the picker-upper nominalizations, perhaps to get the suffix both in the right place and on the right word. It could be that the picker-upper pattern represents the middle stage of a transition from picker-up > picker-upper > pick-upper. As a multi-word verb is more strongly regarded as a single term, the less need there would be to use a suffix on the verb part of the term at all, giving us the pickupper pattern that appears to be productive today. Perhaps one day pickupper will supplant picker-upper, meaning that the suffix gradually will have moved from the verb to which such suffixes have historically belonged, to the end of the term, where suffixes usually go, with an intermediate stage in which the suffix occurs in both positions. Whether this will happen remains to be seen. But for now it appears that both picker-upper and pick-upper are productive patterns. So even though it means putting a suffix in two places, or even though it means putting a deverbal suffix on a particle, speakers nowadays do not seem to mind – at least in speech. It will be interesting to see how far this development will go.

5.

Conclusion

So the -er nominalization of multi-word verbs has been around for a long time and continues to be used today. Nevertheless it is hard to find in published writing, and examining it has required using unorthodox corpora, namely the OED and the internet. Using these corpora requires care in characterizing the diachronic and synchronic use of these nominalizations, but we can tentatively note several trends. Insofar as the citations from the OED accurately represent historical stages of English, we can first note that the picker-up pattern has historically been the most common nominalization of multi-word verbs, but that its use dropped off in the twentieth century. Second, the by-stander pattern occurred fairly commonly in past stages of English, but declined steadily through the centuries. Third, the picker-upper pattern appears to be a recent innovation, probably of the twentieth century. It is hard to say much about the pick-upper pattern, since it does not occur in the OED; its exclusion could mean that the construction is recent or that it is too unconventional to have shown up in the OED citations. Insofar as the internet type counts accurately represent present-day

280

Don Chapman

usage, we can note that the picker-up pattern continues to be used today, but not as much as the picker-upper pattern, which is the most popular pattern. The pick-upper pattern is robust today, as it occurs as much as the picker-up pattern. Perhaps usage is moving from historically favoring the picker-up pattern to eventually favoring the pick-upper pattern, with the picker-upper pattern occurring as a middle stage.

Appendix 1 Nominalizations from the OED, given by first occurrences of each type.12 1400 picker-up pattern setter-forth bringer-out putter-out rattler-out finder-up holder-up raiser-up 1400 on-looker pattern forth-bringer in-holder up-holder out-putter out-rider 1500 picker-up pattern bearer-about carrier-about groper-after looker-after pothunter-after evidencer-against inveigher-against player-against rebeller-against rebutter-against

biter-at player-at pleader-at shifter-away stealer-away passer-by stander-by writer-down labourer-for mourner-for provider-for pourer-forth departer-from flitter-from dealer-in dweller-in hanger-on looker-on finder-out layer-out rooter-out searcher-out seeker-out shredder-out thruster-out weeder-out bringer-up

lifter-up setter-up taker-up encroacher-upon goer-upon partaker-with player-with 1500 on-looker pattern by-hanger by-walker in-dweller out-cryer out-dweller out-stander up-setter up-taker 1600 picker-up pattern discontinuer searcher-after thirster-after contriver-against disputer-against practicer-against trespasser-against writer-against

Fixer-uppers and Passers-by aimer-at carper-at driver-at nibler-at plodder-at practicer-at scoffer-at snatcher-at waiter-at caller-away seller-away slipper-away drawer-back goer-between interceder-between intervener-between batterer-down knocker-down angler-for fisher-for stander-for suiter-for holder-forth setter-forward putter-forwards dissenter-from distracter-from raiser-from bringer-in delighter-in fetcher-in usherer-in diver-into breaker-off cutter-off shaker-off carrier-on depender-on doter-on

drawer-on fastener-on leader-on player-on puller-on putter-on inquirer-out letter-out pourer-out sifter-out squarer-out stretcher-out striker-out sweater-out thrower-out gatherer-together layer-together worshipper-towards binder-up layer-up puffer-up rooter-up screwer-up shorer-up shutter-up spinner-up stirrer-up treasurer-up worker-up wriggler-up essayer-upon runner-upon speaker-upon complier-with consulter-with converser-with sider-with

281

1600 on-looker pattern by-stander forth-putter in-bringer in-comer in-creeper on-looker on-setter out-comer out-lier over-looker over-ruler up-bearer up-giver up-lifter 1700 picker-up pattern enquirer-after poacher-after trotter-after pursuer-against repiner-against attender-at enterer-at frowner-at gambler-at grasper-at knocker-at stayer-at putter-down thrower-down arguer-for runner-forth eloper-from wanderer-from believer-in whipper-in equirerer-into bearer-off

282

Don Chapman

clawer-off putter-off rubber-off borderer-on creeper-on decider-on layer-on spurrer-on treader-on looker-out layer-over adviser-to setter-to writer-to putter-together filler-up gatherer-up locker-up picker-up stringer-up tider-up tucker-up winder-up intruder-upon invader-upon performer-upon trampler-upon usurper-upon sharer-with trifler-with stayer-within 1700 on-looker pattern down-hauler in-hauler off-putter out-hauler out-setter over-runer

up-striker 1800 picker-up pattern hanger-about idler-about rusher-across burrower-after hunter-after inquirer-after luster-after seeker-after striver-after conspirer-against insurer-against offender-against striver-against journeyer-along trudger-along beginner-at boarder-at flincher-at needler-at refuser-at winner-at worker-at scarer-away taker-away wiper-away bringer-back caster-back snapper-back tapper-back saunterer-by boiler-down cougher-down hander-down hewer-down pusher-down screwer-down

smoother-down softener-down talker-down atoner-for carer-for clamourer-for pleader-for struggler-for sufferer-for putter-forth deliverer-from escaper-from ouster-from sufferer-from breaker-in cutter-in dabbler-in dropper-in feeder-in giver-in licker-in lier-in looker-in mover-in putter-in roper-in shutter-in stroker-in swearer-in taker-in twister-in unbeliever-in leader-into researcher-into caller-off knocker-off leader-off marker-off taker-off

Fixer-uppers and Passers-by warder-off wetter-off bringer-on commenter-on getter-on helper-on holder-on hooker-on hounder-on lingerer-on runner-on stander-on tryer-on walker-on blotter-out caller-out chucker-out cutter-out diner-out piecer-out pusher-out sitter-out wiper-out turner-over listener-to speaker-to teacher-to gluer-together screwer-together builder-up caller-up caster-up chipper-up crumpler-up digger-up drawer-up getter-up helper-up knocker-up

licker-up maker-up patcher-up penner-up plodder-up preacher-up putter-up roller-up runner-up scratcher-up seller-up shover-up springer-up sticker-up stubber-up swallower-up tier-up tosser-up vamper-up wakener-up washer-up writer-up meditater-upon performer-with prevailer-with sporter-with dweller-within lodger-within 1800 on-looker pattern down-comer down-looker down-puller forth-comer forth-speaker in-breather in-gatherer in-layer in-lier

283

in-player off-bearer off-comer off-scourer on-bearer on-hanger out-fitter out-goer out-pourer out-speaker out-sticker over-lier up-shutter up-turner 1900 picker-up pattern gawper-after safener-against flier-at guesser-at hanger-back slower-down quitter-from chipper-in drawer-in listener-in mucker-in sitter-in weigher-in listener-into jacker-off seer-off washer-off coupler-on passer-on researcher-on setter-on worker-on doler-out

284

Don Chapman

hander-out noser-out opter-out stamper-out tailer-out boiler-over looker-round slinker-round muddler-through pointer-towards backer-up cleaner-up drier-up jacker-up lighter-up louser-up mopper-up opener-up pepper-up

ringer-up scuffer-up shaper-up snapper-up sparker-up speeder-up sucker-up tidier-up toucher-up warmer-up whooper-up thinker-upon 1900 picker-upper pattern tearer-downer helper-outer builder-upper looker-upper

maker-upper mucker-upper opener-upper pepper-upper picker-upper tidier-upper waker-upper warmer-upper washer-upper 1900 on-looker pattern forth-teller forward-looker in-setter in-taker on-goer over-setter over-cutter up-converter

Fixer-uppers and Passers-by

Appendix 2 Nominalizations found in Google Nominalization Bring about Come about Go about Hang about Mess about Set about Worry about Come along Go along Get around Go around Hang around Lie around Look around Move around Run around Turn around Walk around Arrive back Bring back Come back Cut back Date back Draw back Drive back Fall back Fight back Fly back Get back Glance back Go back Hit back

Picker-up pattern X

Picker-upper pattern

Pick-upper pattern

X X X

X

X X X X X X

X

X X

X X

X

Bystander pattern

285

286

Don Chapman

Nominalization Hold back Hurry back Lay back Lean back Look back Move back Pay back Pull back Push back Put back Report back Run back Send back Sit back Smile back Stand back Stare back Step back Take back Throw back Trace back Turn back Walk back Welcome back Bend down Bog down Break down Bring down Burn down Calm down Climb down Close down Come down Cut down Drop down Face down

Picker-up pattern

X

X

X

X

X

X X

Picker-upper pattern X X X X X X X X X X X

Pick-upper pattern

X

X

X

X X X X X

X

X X X X X X X X X X

X X X

X X X X

Bystander pattern

Fixer-uppers and Passers-by Nominalization Fall down Get down Glance down Go down Hand down Jump down Kneel down Knock down Lay down Let down Lie down Look down Play down Pull down Put down Run down Set down Settle down Shoot down Shut down Sink down Sit down Slide down Slow down Stand down Stare down Step down Take down Track down Turn down Walk down Write down Break in Bring in Build in Call in

Picker-up pattern X

X

X

X X X X X X X X X

X

Picker-upper pattern X X

Bystander pattern

X X X X X X X X X X X X X X X

X X X X X

X X X X

X X X

X X X

X X X X X X X X X

X

X

X X

Pick-upper pattern X

X X X X X

287

288

Don Chapman

Nominalization Cash in Climb in Close in Come in Fall in Fill in Fit in Get in Give in Go in Join in Live in Move in Put in Send in Step in Stir in Swear in Take in Throw in Walk in Break off Call off Come off Cut off Drive off Fall off Finish off Get off Go off Kill off Make off Move off Pay off Pull off Put off

Picker-up pattern X

X X X X X X X X

Picker-upper pattern X X X X X X X X X X X X X

X X X X

Pick-upper pattern X

X X X

X X X

X X X X X

X X

X

X

X

X

X

X X X X X X X

X X X

X

Bystander pattern

Fixer-uppers and Passers-by Nominalization Run off Sell off Send off Set off Shake off Show off Spark off Start off Switch off Take off Trail off Turn off Walk off Ward off Write off Bring on Carry on Come on Follow on Get on Go on Hang on Hold on Keep on Live on Move on Pass on Press on Put on Stay on Switch on Take on Walk on Wear on Work on Bear out

Picker-up pattern X

X

Picker-upper pattern X X

Pick-upper pattern X

X X X

X

X

X X X

X X X

X X X

X

X

X X

Bystander pattern

X

X X

X X X

X X X

X X X X X X X

X X X

X X X X X X X X X

X

X

289

290

Don Chapman

Nominalization Blow out Blurt out Break out Bring out Burn out Burst out Call out Carry out Catch out Check out Climb out Come out Cry out Cut out Draw out Dry out Fall out Figure out Fill out Find out Fly out Get out Give out Go out Hand out Hang out Help out Hold out Jump out Keep out Knock out Lash out Lay out Left out Let out Look out

Picker-up pattern

X

X

Picker-upper pattern X X X X X X X X X X X

Pick-upper pattern

X X X

X X X

X X X X X X X

X X X

X X

X

X X

X

X X X X X X X X X X X X

X X

X X

Bystander pattern

Fixer-uppers and Passers-by Nominalization Lose out Make out Map out Mark out Meet out Miss out Move out Opt out Pay out Phase out Pick out Play out Point out Pour out Pull out Put out Reach out Read out Roll out Rule out Run out Seek out Sell out Send out Set out Share out Shoot out Single out Slip out Sort out Speak out Spell out Spread out Stand out Stare out Start out

Picker-up pattern X X

Picker-upper pattern X X

X

X

X

X X X X

Pick-upper pattern X X

X

X X

X

X X

X X

X X X X X X X X X X X X X

X X

X X

X X X X X

X X

X X X X X X X

X

X

Bystander pattern

291

292

Don Chapman

Nominalization Stay out Step out Stick out Stretch out Strike out Take out Throw out Try out Turn out Walk out Watch out Wear out Wipe out Work out Write out Come over Fall over Give over Go over Hand over Lean over Leave over Look over Pass over Roll over Run over Take over Turn over Voice over Walk over Come round Get round Glance round Go round Look round Spin round

Picker-up pattern

X X X X

X

Picker-upper pattern X X X X X X X X X X X X X X

Pick-upper pattern X

Bystander pattern

X X

X X X X

X

X X

X X X X

X X X X

X

Fixer-uppers and Passers-by Nominalization Swing round Turn round Come through Get through Go through Add up Back up Beat up Bind up Blow up Break up Bring up Build up Call up Catch up Clean up Clear up Climb up Come up Conjure up Cover up Curl up Divide up Draw up Dress up Dry up End up Fill up Finish up Follow up Gaze up Get up Give up Glance up Go up Grow up

Picker-up pattern

Picker-upper pattern

X X X

X

X

X X X X X X X X X X X X X X X X X X X X X X X X X

X

X

X

X X X X X

Pick-upper pattern X

X X X X

X X X X X X X X X X X X X

X X

X X

X X

X

Bystander pattern

293

294

Don Chapman

Nominalization Hang up Hold up Hurry up Join up Jump up Keep up Leap up Light up Line up Link up Lock up Look up Make up Meet up Mix up Move up Open up Pack up Pay up Pick up Pile up Prop up Pull up Put up Reach up Ring up Rise up Roll up Run up Screw up Set up Show up Shut up Sign up Sit up Snap up

Picker-up pattern X

X X

X X

X

X X X

X X

Picker-upper pattern X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X

Pick-upper pattern X X X X

X X

X X X

X

X X X X X X X X X X X X X X X

Bystander pattern

Fixer-uppers and Passers-by Nominalization Speak up Speed up Split up Spring up Stand up Stare up Start up Stay up Step up Stir up Straighten up Sum up Swallow up Take up Team up Throw up Tidy up Tie up Total up Turn up Use up Wake up Walk up Warm up Wash up Weigh up Wind up Wrap up

Picker-up pattern X X X

X

X

Picker-upper pattern X X X X X X X X X X

X

X

X X X X X X

X X X X X X X X X X

Pick-upper pattern X X X X X X X X X X X X X X X X X X X X

X X

X X

Bystander pattern

295

296

Don Chapman

Notes 1. Though sometimes -er has created nominalizations that are patients or fulfill other roles, such as keeper, meaning a fish that is big enough to keep when fishing. Cf. Ryder 2000: 295. 2. This construction has been given many names, such as phrasal verbs, multiword verbs, verb-particle combinations, etc. This paper will use the terminology from Biber et al. 1999. 3. Besides, these patterns do not represent the same kind of nominalizations of the verb-particle construction. For picker (1: 4), the original verb-particle construction will seldom be recoverable. Perhaps listeners can relate listener to listen to or waiter to wait on, but can turn on be recovered from turner or get out from getter? Such relations would have to become entirely conventionalized, as waiter < wait on has. For drop-in (1: 6), the range of meanings is much broader than forms with the -er suffix. A drop-in can be personal (e.g. someone who drops in), but it is more likely to be impersonal (e.g. the action of dropping in, an occurrence of someone dropping in) (Sørensen 1986: 278–79). 4. Finding a good cover term for these different kinds of nominalizations is a problem. Word-formation studies often speak of these four different nominalizations as being different word-formation types. But since this study analyzes the occurrence of these nominalizations in corpora, it will be necessary to use the sense of type (as opposed to token) from corpus studies. To avoid confusing these two senses of type, I refer to the word-formation types as wordformation patterns, of course realizing that pattern can also mean different things in different studies. 5. Štekauer et al. 2005: 50. 6. It was, in fact, this question that led me to this topic. Pavol Štekauer, a native Slovakian speaker, had designed the first draft of this question without any of the double suffixes. As I collaborated on the draft, I told him that we needed to include butter-inner and cutter-inner. He expressed some surprise and incredulity at such a double-marked construction, and I could shed little light on the construction, since that was the first time that I had really noticed it. But at least the respondents agreed with my intuitions. 7. I used regular expression searches to find the various -er patterns for each corpus, and then inspected the over-generated list to find instances of multiword verb nominalizations. It is possible I could have inadvertently excluded some nominalizations, but not many. 8. Normalizing the data is not completely straightforward. For this study, the normalized figures represent the number of new types that occur per million words in each century. The total size of the corpus for each century represents the number of words in the citations, not the number of head-word entries. Normalizing per X (e.g. million) words is typical in corpus studies, but is used

Fixer-uppers and Passers-by

9.

10.

11. 12.

297

here for less typical reasons. In a typical corpus study, we assume that the instances that we are studying occur at a constant rate. The rate, however, that new words will appear for the first time in a corpus is not constant, but instead logarithmic (cf. Baayen 2001). For a typical corpus, a normalized distribution would be misleading for analyzing the occurrence of new types. But the OED is not a typical corpus, and we are not measuring the likelihood that a new word (type) will appear in the corpus as the number of words increases. Instead we are measuring the likelihood that the OED editors will record a new word, as the number of words increases. The rate of recording new words will be constant if the occurrences of the nominalizations are largely limited to citations illustrating the nominalization as a head-word (instead of occurring in citations illustrating some other word), and if the number and size of citations is the same from entry to entry. The first condition is so, and if we assume the second, we are justified in using a words-per-million normalization for measuring the rate at which new words show up in the OED. In short, it is because the OED adds citations at about the same rate as it adds new head-words that we can assume that the number of new words will increase as the number of total words increases. Of course many other multi-word verbs undoubtedly occur with some kind of separation between the verb and the particle. Such have not been included in the internet searches. Admittedly, this procedure works better for the picker-upper and bystander patterns, where the percentage of true positives is relatively high in all rounds, but even for the picker-up and pick-upper patterns the low number of additional types in rounds four and ten suggests that these patterns probably would not produce many new types in the other rounds available, but unexamined (rounds 5–9). The phrasal particle over does not, but that may owe to the phonological shape of over-speakers resist putting two identical sounds together. The nominalizations are spelled with hyphens in this appendix simply for the sake of uniformity in presentation. They certainly did not all occur in the OED spelled with a hyphen.

298

Don Chapman

References Ackema, Peter, and Ad Neeleman 2000 M-selection and phrasal affixation. In UCL Working Papers in Linguistics 12. UCL. C. Iten and A. Neeleman (eds.), 307–342. Adams, Valerie 1973 An Introduction to Modern English Word-formation. London: Longman. Andersen, Gisle 2002 Corpora and the double copula. In From the COLT’s Mouth … and others’. Leiv Egil Breivik and Angela Hasselgren (eds.), 43–58. Amsterdam, Rodopi. Bauer, Laurie 1983 English Word-Formation. Cambridge: Cambridge University Press. Baayen, R. Harald 1992 On frequency, transparency and productivity. Yearbook of Morphology: 181–208. 2001 Word Frequency Distributions. Dordrecht: Kluwer. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward Finegan 1999 The Longman Grammar of Spoken and Written English. London: Longman. Brewer, Charlotte 2000 OED Sources. In Lexicography and the OED. Lynda Mugglestone (ed.), 40–58. Oxford: Oxford University Press. Brinton, Laurel J. 1988 The Development of English Aspectual Systems: Aspectualizers and post-verbal particles. Cambridge: Cambridge University Press. Brinton, Laurel J. and Elizabeth Closs Traugott 2005 Lexicalization and Language Change. Cambridge: Cambridge University Press. Claridge, Claudia 2000 Multi-word Verbs in Early Modern English: A Corpus-based Study. Amsterdam: Rodopi. Clark, Eve V., Barbara Frant Hecht, and Randa C. Mulford 1986 Coining complex compounds in English: Affixes and word order in acquisition. Linguistics 24: 7–27. Chapman, Don and Royal Skousen 2005 Analogical modeling and morphological change: The case of the English adjective negative prefix. English Language and Linguistics. 9: 333–357.

Fixer-uppers and Passers-by

299

Fraser, Bruce 1976 The Verb-Particle Combination in English. New York: Academic Press. Jespersen, Otto 1942 A Modern English grammar on Historical Principles. Part VI: Morphology. Heidelberg: Carl Winter. Kwon, H.-S. 1997 Negative prefixation from 1300 to 1800: A case study in in-/unvariation. ICAME Journal 21: 21–42. Lindelöf, Uno 1938 English verb-adverb groups converted into nouns. Commentationes Humanarum Litterarum 9. 1–41. Marchand, Hans 1969 Categories and Types of Present-day English Word-Formation. 2nd ed. Munich: C. H. Beck’sche. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Ryder, Mary Ellen 2000 Complex -er nominals: Where grammaticalization and lexicalization meet? In Between Grammar and Lexicon. Ellen Contini-Morava and Yishai Tobin (eds.), 291–331. Amsterdam: Benjamins. Sørensen, Knud 1986 Phrasal verb into noun. Neuphilologische Mitteilungen 1986 (87): 272–283. Štekauer, Pavol, Don Chapman, Slavka Tomascikova, and Stefan Franko 2005 Word-formation as creativity within productivity constraints: Sociolinguistic evidence. Onomasiology Online 6: 1–55. Véronis, Jean 2005 Web: Google’s counts faked? Technologies du Langage. Weblog. http://aixtal.blogspot.com/2005/01/web-googles-counts-faked.html.

Words and constructions in grammaticalization: The end of the English impersonal construction*

Graeme Trousdale 1.

Introduction

This article provides a new analysis of the loss of a construction which has been the topic of frequent discussion in work on English historical syntax (for instance Elmer 1981, Fischer and van der Leek 1983, 1987, Denison 1990, 1993, Allen 1995, and Anderson 1997a), namely the impersonal construction. The example under (1) below is often cited to provide evidence of the impersonal construction in Old English (OE): (1) him ofhreow þæs mannes 3SM-dat pity-3SPast the-gen man-gen to-him was-pity because-of-the-man ‘He pitied the man’ Or: ‘The man caused pity in him’ (ÆCHom I XIII.281.12) In the development of English, the impersonal construction disappears; processes which historically were coded in such constructions come to be construed personally. Examples such as (1) are described in the literature as ‘impersonal’ since they have the following morphosyntactic properties: within the clause there is no nominative argument, and the verbal inflection is third person singular, irrespective of the number of either argument. The present aim is to outline aspects of the history of the impersonal construction in English to show how the demise of the construction is associated with the emergence, via a process of grammaticalization, of the transitive construction (TrnCxn, on which see further section 2 below), and to show how constructional taxonomies are implicated in such grammaticalization processes. Grammaticalization has been defined as “the change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions and, once grammaticalized, continue to

302

Graeme Trousdale

develop new grammatical functions” (Hopper and Traugott 2003: 18). It is this second stage of grammaticalization on which I intend to focus here: specifically, I argue that (grammatical) constructions may be subject to further grammaticalization as they are used in different ways by speakers to code events in particular ways. Such grammaticalization may not result in new grammatical functions, but may instead involve the emergence of more schematic constructions, which in turn sanction instances which deviate yet further from a prototype instance of that construction. The place of constructions in grammaticalization is considered with reference both to early work on grammaticalization by Meillet (1958 [1912]), and later work more explicitly concerned with Radical Construction Grammar (e.g. Croft 2001, Traugott 2006, forthcoming a, b). The main thesis developed in this article is that constructions do not necessarily solely constitute the context within which lexemes can be grammaticalized; they may also be subject to grammaticalization themselves. The development of the TrnCxn is an instance of category strengthening, a phenomenon which is well attested in the history of English syntax, in the development of the tense auxiliary do, and the modals (as argued by Hudson 1997), and this claim is supported by corpus work on transitivity in present-day American English (Thompson and Hopper 2001). The structure of the remainder of this paper is as follow. Section 2 provides the general framework for a constructional account of grammaticalization, while section 3 provides a discussion of relevant aspects of the history of the impersonal construction, up to and including the early Modern period. Section 4 considers aspects of transitivity and subjecthood in Modern English. The central thesis, developed in section 5, is that the loss of the impersonal construction is tied in with the increased productivity and schematicity of the transitive construction, and that this constitutes an instance of grammaticalization.

2.

Constructions and grammaticalization

In this paper, I adopt the model of a linguistic construction as presented in Croft (2001). Constructions are symbolic form-meaning pairings (see figure 1); constructions, and parts of constructions, form a network, in which elements are related by taxonomies. These taxonomic links connect both constructions and parts of constructions (see figure 2).

Words and constructions in grammaticalization

syntactic properties

303

CONSTRUCTION

morphological properties FORM

phonological properties Symbolic correspondence (link)

semantic properties pragmatic properties

(CONVENTIONAL) MEANING

discourse-functional properties

Figure 1. The symbolic structure of a construction (Croft 2001: 18)

SbjArg

IntrSbj

Kick1

Pred

IntrVerb

kick

TrSbj

Kick2

TrVerb

Kiss1

kiss

TrObj

Kiss2

Figure 2. A partial constructional taxonomy (Croft 2001: 56)

This constructional taxonomy is highly relevant to grammaticalization. Grammaticalization has traditionally been seen to affect lexical items within constructions, but not constructions themselves. In fact, ‘construction’, in traditional accounts of grammaticalization, is often simply another

304

Graeme Trousdale

term for the context for grammaticalizing lexemes. The word ‘construction’ in such accounts does not mean a form-meaning pairing; it means the other members of the string of words in which the grammaticalizing form appears. However, as has been reported elsewhere in the constructional literature, lexical items are constructions, in the technical Radical Construction Grammar sense, since constructions operate on, or are defined at, all levels of grammatical structure; this is echoed in another variant of Construction Grammar when Goldberg (2006: 18) writes “it’s constructions all the way down” (emphasis original). What distinguishes a lexical construction from a syntactic construction is that a lexical construction is both substantive and atomic (a syntactic unit, fully specified in both form and meaning) while a syntactic construction is both schematic and complex (a combination of syntactic units in which form and meaning are less specific, see Croft 2001: 16–17). Given the parallel between lexical and syntactic constructions, therefore, we might expect syntactic constructions also to undergo grammaticalization (particularly secondary grammaticalization, as defined by Hopper and Traugott 2003: 18, as noted in the introduction). As lexical constructions grammaticalize, they become more schematic, in both form and meaning, and develop new grammatical functions. We would therefore expect syntactic constructions to behave in a similar way. Traugott (2006, forthcoming a, b) has shown that NP of NP constructions (such as a bit/lot of) have grammaticalized as they developed from Partitive Constructions to Degree Modifier Constructions. The question to be addressed here is the extent to which argument structure constructions can also legitimately be said to grammaticalize. In what follows, the notion of constructional prototype is also relevant, and this too ties in with the idea of a constructional taxonomy outlined above. Each level in the constructional taxonomy is sanctioned by a more schematic construction at a higher level. Thus in Figure 2, the [SubjArg]-[Pred] construction sanctions the Intransitive Construction; similarly, at a lesser degree of schematicity, the Transitive Construction sanctions the [Kiss1][kiss]-[Kiss 2] construction. At all levels in the taxonomy, within each category, not all instances will be of the same status; some instances come closer to the conceptual core of the category, while others represent extensions from that prototype. The more an instance deviates from the prototype, the more likely it is that the instance may be recategorised (Labov 1973). Since constructions display prototype effects, we might expect examples whereby non-prototypical constructions (Cx) of category A share

Words and constructions in grammaticalization

305

features with non-prototypical constructions (Cy) of category B; and Cx and Cy may indeed be ambiguous in their category membership. Having outlined features of constructional taxonomies, I now review aspects of the history of the impersonal construction in English.

3.

Aspects1 of the history of the English impersonal construction

3.1. The situation in Old English Example (1) in the introduction – repeated here as (3) for convenience – represents a prototypical instance of the OE impersonal construction: (3) Him ofhreow þæs mannes 3SM-dat pity-3SPast the-gen man-gen to-him was-pity because-of-the-man ‘He pitied the man’ Or: ’The man caused pity in him’ (ÆCHom I XIII.281.12) Many impersonal constructions of this type form a subgroup of a more general category of experiencer constructions (see Allen 1995) in which one of the arguments is the experiencer of some sort of psychological state or process. An instance of an experiencer construction (hereafter ExpCxn) which is not an instance of the impersonal construction is given as (4): (4) hu him se sige gelicade how 3SM-dat the-nom victory-nom please-3Spast how to-him the victory pleased ‘how the victory pleased him’ Or: ’how the victory caused pleasure in him’ (Orosius 156. 25) Here although there is an experiencer (him), the presence of a nominativemarked nominal encoding the Source of the pleasure, means this is not an instance of the impersonal construction. Nonetheless, it is common to relate the impersonal construction to other ExpCxns which have nominative arguments. Such constructions appear under a number of different classificatory labels in, for example, Allen (1995), Denison (1993), Elmer (1981),

306

Graeme Trousdale

and Fischer and van der Leek (1983). I follow Elmer and Allen in their classification, which can be summarised as follows under (5): (5) 2-NP Experiencer constructions in Old English Type N Case of Experiencer: Dative or Accusative Case of Source: Genitive Example: him ofhreow þæs mannes 3SM-dat pity-3SPast the-gen man-gen to-him pitied because-of-the-man ‘He pitied the man’ Or: ‘The man caused pity in him’ (ÆCHom I XIII.281.12) Type I Case of Experiencer: Dative Case of Source: Nominative Example: him ne ofhreow na þæs deofles hryre 3SM-dat not pity-3SPast not the-gen devil-gen fall to-him not pitied not the devil’s fall ‘He did not pity the devil’s fall’ Or: ‘The devil’s fall did not bring about pity in him’ (ÆCHom I XIII.281.14) Type II Case of Experiencer: Nominative Case of Source: Genitive ofhreow Example: se mæssepreost þæs mannes the-nom priest-nom the-gen man-gen pity-3SPast the priest because-of-the-man pitied The priest pitied the man Or: ‘The priest felt pity because of the man’ (COE) ÆLS (Oswald 262) The same verb can appear in more than one ‘type’ of ExpCxn in Old English. We can see this in (3), in that ofhreowan appears in Type N, Type I and Type II constructions. Allen (1995: 85) has helpfully collated these variations in one table, which I give in modified form as Table 1 under (6):

Words and constructions in grammaticalization

307

(6) Table 1. Variation of the 2-NP constructions with the Old English experiencer verbs (Allen 1995: 85)

I only

N only

II only

I and N

II and N

All

losian

lystan

behofian

ofþyncan

sceamian

ofhreowan

gelician

langian

wilnian

þyncan

tweonian

mislician

giernian

hreowan

oflician

reccan

lician eglian gehreowan laþian Another type of construction to be considered is exemplified by the clause in (7): (7) he acwealde þone dracan 3MS-nom kill-1/3SPast the-acc dragon-acc ‘he killed the dragon’ (Ælfric Homilies (Supp.), XXI, 455) Here the NP functioning as agent is case-marked as nominative, and the NP functioning as patient is case-marked as accusative. The verb denotes an action involving the transfer of physical energy. Since case marking on nominals in any given clause in OE “is correlated at least in part with the perspective taken on the state of affairs described” (Traugott 1992: 211), I consider such case marking to be an instance of grammatical organization symbolizing aspects of conceptual content. I classify the example in (7) as a prototypical instance of another type of construction, Type T; prototype instances of Type T involve action verbs. We are now able to construct a partial taxonomy for these 2-NP constructions in OE. This is given in Figure 32. I take Type N constructions to be the prototypical instances of the more schematic ExpCxn category (illustrated in Figure 3 by the emboldened box around the Type N construction), while Type I and Type II are extensions sanctioned by the ExpCxn schema. Type N is maximally distinct from

308

Graeme Trousdale NP1

ExpNP

ExpV erb

Pred

NP2

SouNP

AgNP

AcVerb

PatNP

TYPE T Exp Dat

Exp Verb

TYPE N

Sou Gen

Exp Dat

Exp Verb

Sou Nom

TYPE I

Exp Nom

Exp Verb

Sou Gen

TYPE II

Figure 3. A partial constructional taxonomy for 2-NP construction in OE

Type T in terms of verbal morphology (lack of agreement) and case marking. The extensions are non-prototypical members of the category, since the NPs have different case markings; such case marking shows an overlap with members of the Type T category3. The highest construction in this partial taxonomy is what I will call the Transitive Construction (TrnCxn). It is this superordinate level construction which ultimately sanctions all of the others, and is schematic in both form and meaning. We have seen above how variation in case marking indicates the formal variation in subtypes of the TrnCxn; I turn now to other ways in which the various subtypes of the TrnCxn vary, using some of the parameters for transitivity as defined by Hopper and Thompson (1980), given in (8) below. (8) Table 2. Some parameters for transitivity (adapted from Hopper and Thompson 1980: 252)

KINESIS ASPECT PUNCTUALITY VOLITIONALITY AGENCY AFFECTEDNESS OF O

HIGH action telic punctual volitional A4 high in potency O totally affected

LOW non-action atelic non-punctual non-volitional A low in potency O not affected

Words and constructions in grammaticalization

309

Such parameters allow us to identify significant differences between instances of Type T and Type N in OE. For example, prototypical instances of Type T are telic, volitional and denote action; prototypical instances of Type N are atelic, non-volitional and often denote states rather than actions. 3.2. The later history of the impersonals Patterns relating to the late history of the impersonal construction have been established by the account provided by Allen (1995), in her detailed history of the experiencer verbs as a whole. Allen provides evidence of the increasing subject assignment for the Experiencer, an analysis based on the syntactic criteria that Preposed Dative Experiencers (PDEs)5, and Experiencers which have nominative inflection if pronominal, should be classified as subjects, though the number of examples in the corpora used in standard accounts are small. Thus rewen6, which never appears in Type II constructions in OE, could have an unambiguous nominative subject experiencer in Middle English, as in (9): (9) We schold rew þat sore (?1325 Swet Iesus v2, l6) The general pattern then is for Experiencers rather than Sources to become coded as subject. While this pattern is true generally, the development was a very gradual one, and we can notice what seem to be different trajectories for individual lexemes, such as ail and behoove. Allen (1995: 271) notes that both ail and behoove resist nominative experiencers: “we find examples of PDEs in the late fifteenth century in texts which contain no examples with nominative experiencers.” The construction in which ail appears is largely fixed, a reported question of the type what DAT ail. Such fixed expressions are not characteristic of behove, witness (10): (10) Hym behoued non other lighte (Caxton, 77.13) which, by the seventeenth century, had evolved into the non-impersonal (11): (11) those Persons that rob me, are not fit to be Evidences against me, because it behoves them that I be convicted, to prevent their being indicted for Felony. (Lisle)

310

Graeme Trousdale

Similarly, in the sixteenth century, there is evidence of the continuation of Type I constructions with PDEs, in the writings of Sir Thomas More, exemplified by examples (12) through (14): (12) Thou shalte labour to thy power by juste and trewe busynesse, to geate that thee and thyne behoueth (More 1557 [c.1522] 90) (13) And sone after one hower betwene .x. & .xi. he returned into ye chamber among them, al changed with a wonderful soure angrye countenaunce, knitting the browes, frowning and froting and knawing on hys lippes, and so sat him downe, in hys place: al the lordes much dismaied & sore merueiling of this maner of sodain chaunge, and what thing should him aile (More 1513, 47) (14) And for asmuch to her seemed the Cardinall more redy to depart (More 1513, 40) Finally, the examples in (15) show that variation with like continued well into the seventeenth century. (15) a. these two, trauelinge into east kent, resorted vnto an ale house there, being weried with traueling, saluting with short curtisey, when they came into the house, such as thei sawe sitting there, in whiche company was the parson of the parish; and callinge for a pot of the best ale, sat down at the tables ende: the lykor liked them so well, that they had pot vpon pot (Harman 1567) b. her name is Buckle, a Sharpsheare woman: if you like of it, I would thinke of haueing of her; for I haue no body aboute me (Harley 1633) c. That in the mean tyme yt wyll lyke ye to wryte to my lord how carefull and myndfull I have byn of him, shall doe me a great pleassur (Dudley 1585) d. yf my cosin like it, I will send him more. (Cornwallis 1664) What we see, therefore, is an interplay between lexical variation and constructional variation – some of the early Modern data show specific behaviour of particular tokens (e.g. ail, behoove), and some show behaviour of

Words and constructions in grammaticalization

311

particular types (e.g. Type I constructions). This issue has been noted by Allen: The reader might reasonably object that in saying that PDEs lastest [sic] longest in particular constructions because they had always (or at least, for some time) been most common in those constructions is begging the question of why they were more popular in some constructions than in others. (Allen 1995: 282–283, emphasis added)

The question, then, is why some constructional variants of the ExpCxn are more resistant to change than others. The development, and the subsequent loss of PDEs, is usually explained as a change from lexical to structural case marking: “[t]he disappearance of the PDE as a structural possibility is best seen as a gradual disfavouring of the option of case-marking subjects lexically” (Allen 1995: 347). However, I propose an alternative account, based on grammaticalization and constructional taxonomies as outlined in section 27. Following Bybee, Perkins and Pagliuca (1994), Langacker (2005) and Traugott (forthcoming b) on constructional change generally, I argue that the TrnCxn in the history of English has become more productive (sanctioning a wider range of subject types, for instance), more general (encoding a wider range of thematic relations between the verb and its arguments), and less compositional (since some variants of the TrnCxn have become less clearly telic, for instance). Increased productivity, increased generalisation, and decreased compositionality, taken together, are evidence of the TrnCxn having grammaticalized over time. This structuralization (on which see further Thompson and Hopper 2001, and section 4 below) is related to the strengthening of the TrnCxn as a superordinate category in English. Evidence from this comes in part from an analysis of the TrnCxn in present-day English, which is discussed in the next section.

4.

Transitivity and subjecthood in Modern English

Like the impersonal construction, transitivity in English has also been the subject of a number of analyses. In an article cited earlier (see (8) above), Hopper and Thompson (1980) attempted to provide a set of attributes which characterised clauses in terms of high and low transitivity. In a more recent paper, Thompson and Hopper (2001) reconsider the concept of transitivity in light of evidence from spoken corpora. They suggest that clauses which

312

Graeme Trousdale

are canonically ‘transitive’ – i.e. telic, punctual, volitional on the part of the agent, and so on – are surprisingly infrequent in naturally-occurring conversation, and also that the findings of the study support the claim, made by Hopper (1987: 150), that “the more useful a construction is, the more it will tend to become structuralized, in the sense of achieving cross-textual consistency, and serving as a basis for variation and extension”. Their data suggest that clauses with two participants were not only rare in themselves (most of the clauses had only one participant), but also very low in transitivity: for instance, the clauses were typically atelic (86 %); non-punctual (98%); and had a non-affected object (84 %). This phenomenon appears to be true not only of English: Thompson and Hopper also cite other studies which suggest that this occurs cross-linguistically. Thompson and Hopper also discuss predicate meanings with reference to Roland and Jurafsky (2002), where they state that: “the sense of a verb or predicate is related to the grammatical schemas that it can occur in” (Thompson and Hopper 2001: 44), suggesting that such meanings: can only be understood as including a vast range of semantic and pragmatic associations regarding the sorts of activities, states, and participants that can invoke their use … these ‘meanings’ are actually generalizations from many repetitions of hearing predicates used in association with certain types of human events and situations over the course of a person’s lifetime. What appears to be a fixed ‘structure’ is actually a set of schemas, some more ‘entrenched’ (Bybee 1985, 1998; Langacker 1987) than others, arising out of many repetitions in daily conversational interactions. (Thompson and Hopper 2001: 47).

This view also accords with that proposed in construction grammars of various kinds (e.g. Goldberg 1995, 2006), where the claim is that usagebased grammar is essentially organised around meaning. The generalisations or extensions that are witnessed typically relate high type frequency verbs (such as verba dicendi) to high type frequency constructions (such as the ditransitive). Thompson and Hopper also suggest that verbs of high frequency (get, think, see, want and mean, for instance) have argument structures which are difficult to categorize – they tend to appear more frequently in fixed (lexicalised) expressions – while verbs of low frequency (such as spray, swarm, or load) have a relatively fixed argument structure. Tomasello (1998: xviii) has explicitly linked this to a prototype analysis of transitivity in English, as follows:

Words and constructions in grammaticalization

313

All constructions, whether composed of one word, or of many words in specific orders with specific markers and intonations, and at whatever level of abstraction, derive from recurrent events or types of events, with respect to which the people of a culture have recurrent communicative goals. Because they are formed this way historically, they exhibit prototypical structure just like other cognitive categories and schemas. Thus in English, the basic transitive construction has as its prototype utterances such as He broke the vase in which an animate actor does something to cause a change of state in an undergoer (Hopper and Thompon 1980). But the construction over historical time has been extended to other, less prototypical situations in which the ‘force dynamics’ are not so clear or are only metaphorical, as in, for example, John entered the room and The car cost $400. In German … the transitive construction has stayed much closer to the prototype.

And this is illustrated by Taylor (1998: 188) with examples such as those in (16): (16) a. The hotel forbids dogs b. The tent sleeps six c. *Das Hotel verbietet Hunde d. *Das Zelt schläft sechs Constructions which do not instantiate the prototype of the category are not necessarily any ‘less grammatical’ than those which do; rather, the transitive construction is organised as a network (cf. the earlier discussion of constructional taxonomies), with certain properties of the more central members of the network displaying grammatical properties which do not characterise those at the periphery (i.e. those extensions which, while sanctioned by the schema, may display properties of other categories). Such a network analysis is applicable too to the category of subject. Taylor (1998) gives examples of the kinds of constituents which can function as subjects, in addition to typical noun phrases, as in (17): (17) PP subject: Over there is fine Clausal subject: That Horace should eat meat is astonishing Existential subject: There are six people in that car All are clearly morphosyntactic subjects, since they can undergo raising, as in (18):

314

Graeme Trousdale

(18) Over there seems to be fine and have a regular tag question formation, as in (19): (19) That Horace should eat meat is astonishing, isn’t it? But the use of such subjects is constrained (see Anderson 1997a, b). Locative PPs as subjects of passives can be problematic, witness the dubious grammaticality of (20): (20) ?Over there was sniffed by the dog Clausal subjects do not appear in interrogatives, as illustrated by the ungrammaticality of (21): (21) *Is that Horace should eat meat astonishing? And the syntax of there as a subject (as Lakoff 1987 has demonstrated) shows various odd patterns of behaviour as exemplified by the clauses in (22): (22) There came a dense fog over Dalkeith *There left a delegation for Dalkeith Such patterns can also be accounted for in terms of constructional taxonomies. This is illustrated in Figure 4. Just as we witnessed extensions from the prototype of the ExpCxn in OE (see Figure 3), so in Modern English we see extensions sanctioned by the TrnCxn, to allow for instances such as This tent sleeps six. These must be considered as extensions because they do not display grammatical properties associated with central transitives (e.g. no passive variant *Six are slept by this tent, see Taylor 1998). The next section therefore considers two related issues: how did the ExpCxn come to be coded ‘personally’ rather than impersonally; and how are we to understand the parallel between extensions to the ExpCxn which are attested in OE on the one hand, and extensions to the TrnCxn which are attested in Modern English on the other? I argue that a grammaticalization account explains both these issues.

Words and constructions in grammaticalization Subj

TrnSubj

TrnVerb

Pred

315

Comp

TrnObj

C op s ub j

be

Predicative

e.g. Over there is fine Exp Subj

Trn Verb

Sou Obj

e.g. Bob likes Sue

Agent Subj

Trn Verb

e.g. He killed the dragon

Pat Obj

Locative Subj

Trn Verb

Theme Obj

e.g. This tent sleeps six

Figure 4. A partial constructional taxonomy illustrating subjects and transitivity in Modern English

5.

A constructional account of grammaticalization

There is a particular phrase in Meillet’s 1912 article which is very frequently cited as a definition of grammaticalization, namely that grammaticalization is: (23) l’attribution du caractère grammatical à un mot jadis autonome (Meillet 1958 [1912]: 131) [the attribution of grammatical character to a word which once was autonomous (my translation)] But as Traugott (2006, forthcoming a, b) has noted, Meillet allowed for the possibility that grammaticalization affected not only lexical items per se, but groups of lexical items: (24) Les mots ne sont du reste pas seuls à être sujets à devenir des éléments grammaticaux; la façon de grouper les mots peut aussi devenir un procédé d’expression grammaticale (Meillet 1958 [1912]: 147) [Words are not the only things which can become grammatical elements. The manner of grouping words may also become a pathway for grammatical expressions (my translation)]

316

Graeme Trousdale

His topic was specifically the development of certain word order patterns in the evolution of Late Latin into some of the Romance languages. But the critical issue is that of “la façon de grouper les mots” – the development of groups of words into grammatical items lends itself to a constructional account of linguistic structure. Indeed, Meillet suggests that, in terms of the actual processes which characterise grammaticalization, the behaviour of lexical items on the one hand, and groups of such items on the other, is the same: (25) “Le phénomène est de même ordre que la ‘grammaticalisation’ de tel ou tel mot; au lieu que ce soit un mot employé en groupe avec d’autres qui prenne le caractère de ‘morphème’ par un effet d’habitude, c’est une manière de grouper les mots” (Meillet 1958 [1912]: 148) [The phenomenon is of the same order as the grammaticalization of this or that word; rather than a word, working in concord with others, which takes one the character of a morpheme through habitual use, it is a way of grouping words (my translation)] Such a view predates, but accords with, cognitive and constructional accounts of grammaticalization. For instance, Bybee (2003: 153) defines one aspect of grammaticalization as the “process of automatization of frequently occurring sequences of linguistic elements”, or in Cognitive Grammar terms, grammaticalization is increased entrenchment (cf. Langacker 1987, 1991). But what, precisely, is entrenched? It is not the instances of the schema (which we expect in low-type, high-token cases), but rather the schema itself. Given that such schemas typically form superordinate categories, we might expect to find, following Croft and Cruse (2004) in their discussion of features that distinguish basic level from superordinate categories, that the transitive construction will show a high degree of differentiation in comparison with other schematic, superordinate categories (such as the copula construction illustrated in Figure 4 above) but a low degree of mutual similarity between members of that category, which is indeed the case in Modern English. For instance, while the copula allows non-NP subjects (Over there is fine), few8 verbs in the TrnCxn do (mutual distinctiveness with respect to other superordinate categories); on the other hand, not all instances of the TrnCxn allow passives (low degree of mutual similarity between members of the category). How has this come about? I suggest that this is one of the consequences of grammaticalization – that as a construction type grammaticalizes, it becomes more schematic, and thereby

Words and constructions in grammaticalization

317

automated. On the relationship between linguistic constructions and other cognitive processes, Tomasello (1998: xvii) argues that the schematic constructions are typical of cognitive processes in general; they are “relatively automatized procedures that operate on a categorical level”. Moreover, these constructions, schematic though they may be, are not ‘meaningless’: “each of these abstract linguistic schemas has a meaning of its own, in relative independence of the lexical items involved” (ibid). This links into language as a creative process, where speakers extend their use of constructions by inserting different words into that frame. They form part of the “inventory of symbolic resources that language users control” (Tomasello 1998: xviii). This too is expected if we assume that constructions which are abstract or schematic are not unique to language, but share properties with schemas in other cognitive domains. It is then necessary to ask ‘how schematic is schematic’? Schematicity, just like entrenchment, is a gradient phenomenon: Langacker (1987, 1991) has repeatedly suggested that the meaning of grammatical elements is usually quite schematic, but given the gradient nature of schematicity, it would not be unwarranted to assume that some grammatical elements are more schematic than others, and that the process by which grammatical elements become more schematic is known more generally in linguistic theory as grammaticalization. This can be illustrated concretely by examining the relationship between impersonal and transitive constructions. Note that I am not discussing the grammaticalization of particular impersonal phrases (as in methinks), although this has been proposed by Palander-Collin (1997). Specifically, the development of the Types N, I and II constructions between Old English and Modern English illustrates a kind of grammaticalization. The development of the transitive and impersonal constructions in the history of English is an example of specialization (in the sense of Hopper 1991), whereby a narrower range of construction types license a larger range of specific tokens, though the grammatical meaning indicated by such construction types becomes more general: specifically, the transitive construction has emerged as a superordinate category (on the criteria listed, for example, by Croft and Cruse 2004). Thus in OE, the basic level constructions were: (i) the ExpCxn (which sanctioned the prototypical Type N, and its extensions, Type I and Type II) and (ii) Type T. In ME, structural case assignment becomes the only option, following the demise of lexical case assignment, but the gradual loss of the various subtypes of the ExpCxn is more consistent with a grammaticalization account which involves the gradual disfavouring of a set of constructional subtypes (cf. Fischer 1992 for a slightly

318

Graeme Trousdale

different argument regarding the gradualness of the change) and an increase in productivity for the remaining types. By Modern English, the range of distinctive constructional types has reduced, and the TrnCxn licenses an increasingly larger range of tokens. This change is facilitated by the ambiguous – and sometimes nonexistent – force dynamics of the processes associated with impersonal constructions. On stative mental relations, Croft (1998) notices significant cross-linguistic variation regarding the assignment of experiencer and stimulus (or source) to the grammatical roles of subject and object. In these mental states, we cannot appeal to force dynamics, since the participants do not enter into such a relationship: Hence, in some languages (such as English), the experiencer is normally made subject; in other languages (such as Russian and languages of southcentral Asia), the experiencer is normally made (indirect) object; in still other languages, the experiencer and stimulus are both encoded as either subjects (Japanese ‘double-ga’ constructions) or as objects (Eastern Pomo) (Croft 1998: 85).

The selection of either the experiencer or the source as subject (as in the OE Type II and Type I respectively) is warranted by the fact that both roles share features associated with prototypical subjects: experiencers are typically animate and human; sources are associated with causation, like agents.

6.

Future prospects and concluding remarks

Denison (this volume) has observed that the present paper does not provide any information regarding the relative frequency of the various constructions involved in the change. It is my intention to develop the framework established in this piece of research by carrying out a collostructional analysis (see Stefanowitsch and Gries 2003; Gries and Stefanowitsch 2004) of the standard parsed corpora that Denison lists, in order to establish the frequency with which certain verbs appear in the various subconstructions identified here. Such an analysis will also shed light on the process of constructional blending mentioned in footnote 3.

Words and constructions in grammaticalization

319

In summary, we can see how the demise of the impersonal construction relates to grammaticalization by considering the three steps for grammaticalization suggested by Croft (2001): 1. Extension of function of construction type A to construction type B This stage involves “alternative conceptualisations of the same experience”, that is, variation in construal, but specifically whereby “the new construction at least partially imposes the conceptualization of its original structure and function” (Croft 2001: 127). In our case, this is where experiencers become coded as more agentive (and have grammatical properties associated with agents, such as subject marking.) 2. Marginalization of the old construction The prototypical impersonal construction (Type N) is already rare in Old English. Many experiencer verbs already have either the source or the experiencer coded as nominative, and increasingly such constructions involve NPs with distinctive subjectlike properties (e.g. raising, verbal concord.) 3. The new construction undergoes shifts in grammatical behaviour. At this stage, we expect to see ‘transitives’ with an increasingly wide scope of ‘subject’ types, and fossilisation of the remnants of the old construction (as in methinks, and possibly if you like), which, while they are a feature of the language of modern English speakers, could not be considered as evidence of a productive grammatical process. Given the prominence he has been accorded elsewhere in this article, it is fitting that the final quotation should come from Meillet: (26) A chaque fois qu’un élément linguistique est employé, sa valeur expressive diminue et la répétition en devient plus aisée (Meillet 1958 [1912]: 135) [Each time a linguistic element is used, its expressive value diminishes, and the repetition becomes easier (my translation)] Meillet’s work on grammaticalization has significant resonances for those working on constructional change in English. His views on lexical grammaticalization are well known; his views on the place of constructions in grammaticalization, less so. But the development of impersonal and transitive

320

Graeme Trousdale

clauses in the history of English provides a compelling argument for a constructional account of grammaticalization9. Grammaticalization can be seen both as increased entrenchment and as increased schematization (Bybee 2003), or perhaps more succinctly, as the entrenchment of schemas.

Notes *

This article is based on a paper presented at the Studies in the History of the English Language 4 conference held at Northern Arizona University in Flagstaff in September 2005, and on a presentation at Roehampton University’s English Language and Linguistics research seminar series in December 2005. I am grateful to the audiences at both venues for their helpful and insightful comments, and to Nik Gisborne, Dirk Noël, Lene Schøsler and especially to Elizabeth Traugott, for a number of particularly useful discussions on grammaticalization and constructions, and to David Denison, for very helpful discussions based on his reactions to the paper (see Denison, this volume). I would also like to thank the two anonymous reviewers of an earlier version of this piece; their comments were extremely useful. All remaining shortcomings are my own. I am also very grateful to the British Academy and to the University of Edinburgh Development Trust for grants (British Academy Overseas Conference Grant 41410 and University of Edinburgh Small Project Grant 2081) which paid for my travel costs to attend SHEL 4 and to present the original paper. 1. As Denison (this volume) rightly notes, this paper presents only a partial history of some variants of the impersonal and transitive constructions (but see also footnote 3). A planned larger study will consider not only the kinds of complexities raised by the kinds of constructional blends discussed in footnote 3 and by Denison, but also the crucial issue of word order as discussed in detail by Allen (1995), which is also not addressed here. 2. In Figure 3, for the central constructional variants under discussion, I give the thematic role played by each NP, and the case marking on the NP, but do not ascribe a (grammatical) function such as subject or object; in Figure 4, however, I do not give the case marking, but do give the grammatical function. This is meant to reflect what I see as a change in the organization of the grammar of English. The grammaticalization of the transitive construction involves the grammaticalization of the category of subject in English (on which see further Anderson 1997b). It is not clear to me that we can be confident on subjecthood in relation to many instances of the OE ImpCxn (cf. Denison 1993: 61–2), but we can be clear(er) on the roles played by participants in any

Words and constructions in grammaticalization

3.

4.

5.

6.

7.

8.

9.

321

given clause. The category of ‘subject’, as a result of its grammaticalization, has strengthened as it has evolved in English. Some verbs (e.g. OE lufian) show yet a different kind of constructional blending, which is not covered by Types N, I, II or T. With lufian, the experiencer verb has a nominative experiencer argument and an accusative source argument. Denison (this volume) provides further examples of clauses which show yet further kinds of constructional blending. I would classify his examples (22)–(24) as non-prototypical extensions of the Type T construction in OE. His suggestion that such examples might be analysed using a multiple inheritance model of constructional taxonomies (see Goldberg 2006) is clearly worth pursuing, and I have argued elsewhere (see Trousdale 2008) that such multiple inheritance models can provide a principled account of the development of some light verb constructions in the late Modern English period. In Hopper and Thompson (1980), the authors follow Dixon (1979) in using A and O to denote the two participants in a two-participant clause, Agent and Object respectively; but the authors “make no claims about the grammatical relations that the NP arguments referring to these participants might bear to the verb” (Hopper and Thompson 1980: 252, footnote 1). Preposed Dative Experiencers occur in clauses in which “a non-nominative Experiencer is fronted, without regard to the case of the Theme” (Allen 1995: 21–2) It is possible that the example in (9) is a development of OE ofhreowan, which did occur in Type II constructions, though the form arewen was also common in the fourteenth century. The change from lexical case to structural case is an important factor in the loss of the PDEs, but I see this development as part of the grammaticalization process; this view is consistent with that of Meillet (1958 [1912]) on grammaticalization. In an earlier version, I suggested that no verbs in the TrnCxn appear with nonNP subjects, but the observations made by Denison (this volume) show that this was too strong a claim. However, note that the example that he gives (Over there suits me fine) is a marginal (i.e. non-prototypical member) of the transitive category: there is, for instance, no passive variant (*I am suited fine by over there). See Noël (2007) for a discussion of the problems in bringing together grammaticalization and constructional accounts of language.

322

Graeme Trousdale

References Primary sources ÆCHom I = Clemoes, P. (ed.) 1997 Ælfric’s Catholic Homilies: the first series. Oxford: Early English Texts Society. ÆHom (Supp.) = Pope, J. (ed.) 1968 Homilies of Ælfric: a supplementary collection, volume II. Oxford: Early English Texts Society. Caxton = Blake, N.F. (ed.) 1970 The History of Reynard the Fox Translated from the Dutch Original by William Caxton. Oxford: Early English Texts Society. Cornwallis = Griffin, R. (ed.) 1842 The Private Correspondence of Jane, Lady Cornwallis, 1613–44. London: S. & J. Benson, Wilson and Fley, 1842. Dudley = Bruce J. (ed.) 1844 Correspondence of Robert Dudley, Earl of Leycester, during his government of the Low Countries, in the years 1585–1586. Camden First Series. Harley = Lewis, T.T (ed.) 1854 Letters of the Lady Brilliana Harley, wife of Sir Robert Harley, of Brampton Bryan, Knight of the Bath. Camden First Series. Harman = Viles, E. and F.J. Furnivall (eds.) 1937 Thomas Harman [1567, 3rd ed.] A Caveat or Warening for Commen Cursetors Vulgarely called Vagabones. London: Early English Texts Society. Lisle = The Trial of Lady Alice Lisle In R. Hargreave (ed.), A Complete Collection of State Trials and Proceedings for High Treason and other Crimes and Misdemeanours from the reign of King Richard II to the end of the reign of King George I. London, 1730. ME Sermons = Woodburn O. Ross (ed.) 1940 Middle English Sermons. London: Early English Texts Society. More 1513 = Sylvester, Richard S. (ed.) The Complete Works of Sir Thomas More volume 2: The History of King Richard III. New Haven and London: Yale University Press. More 1557 [c.1522] = Facsimile text of The Four Last Things by Sir Thomas More in Campbell, W.E. and A.W. Reed (eds.) 1931 The English Works of Sir Thomas More, volume 1. London: Eyre and Spottiswoode.

Words and constructions in grammaticalization

323

Orosius = Sweet, H. (ed.) 1883 King Alfred’s Orosius. Oxford: Early English Texts Society. Swet Iesus = ‘Swet Iesus, hend and fre’ [The Poem of Michael Kildare] in Heuser, W. (ed.) 1904 Die Kildare-gedichte. Bonner Beträge zur Anglistik 14. I have also made use of Healey and Venezky (1980)’s Microfiche Concordance to Old English, listed in the secondary sources below.

Secondary sources Allen, Cynthia 1995 Case Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Oxford University Press. Anderson, John M. 1997a Subjecthood and the English impersonal. In Language History and Linguistic Modelling, Volume 1. Raymond Hickey and S. Puppel (eds.), 251–263. Berlin/New York: Mouton de Gruyter. 1997b A Notional Theory of Syntactic Categories. Cambridge: Cambridge University Press. Bybee, Joan 1985 Morphology: A Study of the Relation between Meaning and Form. Amsterdam/Philadelphia: Benjamins. 1998 The emergent lexicon. CLS 34: The Panels, 421–435. 2003 Cognitive processes in grammaticalization. In The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, volume 2. Michael Tomasello (ed.), 145–167. Mahwah, New Jersey: Lawrence Erlbaum. Bybee, Joan L., Revere Perkins and William Pagliuca 1994 The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago: University of Chicago Press. Croft, William 1998 The structure of events and the structure of language. In The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, volume 1. Michael Tomasello (ed.), 67–92. Mahwah, New Jersey: Lawrence Erlbaum. 2001 Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge: Cambridge University Press.

324

Graeme Trousdale

Denison, David 1990 The Old English impersonals revived. In Papers from the Fifth International Conference on English Historical Linguistics. Sylvia Adamson, Vivien Law, Nigel Vincent and Susan Wright (eds.), 111– 140. Amsterdam/Philadelphia: Benjamins. 1993 English Historical Syntax. London: Longman. Dixon, Robert M.W. 1979 Ergativity. Language 55: 59–138. Elmer, Willy 1981 Diachronic Grammar: the History of Old and Middle English Subjectless Constructions. Tübingen: Niemeyer. Fischer, Olga 1992 Syntax. In The Cambridge History of the English Language volume II, 1066-1476 Norman Blake (ed.), 207–408. Cambridge: Cambridge University Press. Fischer, Olga and Frederike van der Leek 1983 The demise of the Old English impersonal construction. Journal of Linguistics 19: 337–368. 1987 A “case” for the Old English impersonals. In Explanation and Linguistic Change (Current Issues in Linguistic Theory 45). Willem Koopman, Olga Fischer, and Frederike van der Leek (eds.), 79–120. Amsterdam/Philadelphia: Benjamins. Goldberg, Adele 1995 Constructions: a Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. 2006 Constructions at Work: The Nature of Generalizations in Language. Oxford: Oxford University Press. Gries, Stefan Th. and Anatol Stefanowitsch 2004 Extending collostructional analysis: a corpus based approach to ‘alternations’. International Journal of Corpus Linguistics 9.1: 97–129. Healey, A. and R. Venezky 1980 A Microfiche Concordance to Old English. Dictionary of Old English Project: University of Toronto. Hopper, Paul, J. 1987 Emergent grammar. BLS 13: 139–157. 1991 On some principles of grammaticalization. In Approaches to Grammaticalization volume 1. Elizabeth Closs Traugott and Bernd Heine (eds.), 1–35. Amsterdam/Philadelphia: Benjamins. Hopper, Paul and Sandra Thompson 1980 Transitivity in grammar and discourse. Language 56: 251–299. Hopper, Paul J. and Elizabeth Closs Traugott 2003 Grammaticalization, 2nd ed. Cambridge: Cambridge University Press.

Words and constructions in grammaticalization

325

Hudson, Richard A. 1997 The rise of auxiliary DO: verb-non-raising or category strengthening? Transactions of the Philological Society 95: 41–72. Labov, William 1973 The boundaries of words and their meanings. In New Ways of Analyzing Variation in English. C. J. N. Bailey and R. Shuy (eds.), 340– 373. Washington, DC: Georgetown University Press. Lakoff, George 1987 Women, Fire and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Langacker, Ronald 1987 Foundations of Cognitive Grammar volume 1: Theoretical Prerequisites. Stanford: Stanford University Press. 1991 Foundations of Cognitive Grammar volume 2: Descriptive Application. Stanford: Stanford University Press. 2005 Construction grammars: cognitive, radical, and less so. In Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction. Francisco J. Ruiz de Mendoza Ibáñez and M. Sandra Peña Cervel (eds.), 101–159. Berlin/New York : Walter de Gruyter. Meillet, Antoine 1958 L’évolution des formes grammaticales. In Linguistique historique et [1912] linguistique générale. Antoine Meillet, 130–148. Paris: Champion. Noël, Dirk 2007 Diachronic construction grammar and grammaticalization theory. Functions of Language 14.2: 177–202. Palander-Collin, Minna 1997 A medieval case of grammaticalization, methinks. In Grammaticalization at Work: Studies of Long-term Developments in English. Matti Rissanen, Merja Kytö and Kirsi Heikkonen (eds.), 371–403. Berlin/ New York: Mouton de Gruyter. Roland, Douglas and Daniel Jurafsky 2001 Verb sense and verb subcategorization probablilities. In The Lexical Basis of Sentence Processing: Formal, Computational, Experimental Issues. Suzanne Stevenson and Paolo Merlo (eds.), 325–346. Amsterdam/Philadelphia: Benjamins. Stefanowitsch, Anatol and Stefan Th. Gries 2003 Collostructions: investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8.2: 209–243. Taylor, John 1998 Syntactic constructions as prototype categories. In The New Psychology of Language: Cognitive and Functional Approaches to

326

Graeme Trousdale

Language Structure, volume 1. Michael Tomasello (ed.), 177–202. Mahwah, New Jersey: Lawrence Erlbaum Thompson, Sandra and Paul Hopper 2001 Transitivity, clause structure, and argument structure: Evidence from conversation. In Frequency and the Emergence of Linguistic Structure. Joan Bybee and Paul Hopper (eds.), 27–60. Amsterdam/Philadelphia: Benjamins. Tomasello, Michael 1998 Introduction. In The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, volume 1. Michael Tomasello (ed.), vii–xxiii. Mahwah, New Jersey: Lawrence Erlbaum. Traugott, Elizabeth Closs 1992 Syntax. In The Cambridge History of the English Language volume 1: the beginnings to 1066. Richard M. Hogg (ed.), 168–289. Cambridge: Cambridge University Press. 2006 Constructions and language change revisited: Constructional emergence from the perspective of grammaticalization. Paper presented at Directions in English Language Studies. University of Manchester, April 6–8 2006. Forthcom. The grammaticalization of NP of NP constructions. In Constructions a. and Language Change. Alexander Bergs and Gabriele Diewald (eds.). Berlin/New York: Mouton de Gruyter. Forthcom. Grammaticalization, emergent patterns, and the notion of ‘newb. ness’. Plenary paper presented at the High Desert Linguistics Society. University of New Mexico, November 9–11 2006. German translation to appear as ‘Grammatikalisierung und emergente Konstruktionen’ in Konstruktionsgrammatik und grammatische Konstruktionen. Anatol Stefanowitsch and Kerstin Fischer (eds.), Tübingen: Stauffenburg. Trousdale, Graeme 2008 Constructions in grammaticalization and lexicalization: evidence from the history of a composite predicate in English. In Constructional explanations in English grammar. Graeme Trousdale and Nikolas Gisborne (eds.), 33–67. Berlin/New York: Mouton de Gruyter.

Variation in Late Modern English: Making the best use of ‘bad data’

Joan C. Beal 1.

Overview

The three papers in this section all deal in different ways with what Labov (1994) identified as the key challenge for historical linguists: that of “making the best use of bad data”. The researcher engaged in a study of contemporary English can design a balanced, representative sample and has access to information about the social factors and external influences that may impinge on linguistic variation, but, as Stefan Dollinger points out in his paper “Taking permissible shortcuts? Limited evidence, heuristic reasoning and the modal auxiliaries in early Canadian English”, those dealing with historical data have a much more difficult task. They have to rely on evidence that “survive[s] by chance not design” (Labov 1994: 11), which is often in what Labov terms a “normative dialect”, and for which, as Nevalainen (1999) points out, the metadata is lacking. For scholars researching the Late Modern period, these problems are less acute than for medievalists, since more data survives, including documents from lesseducated, lower-class writers, and the social and historical background is easier to reconstruct. On the other hand, as Dollinger points out here, the “colonial” varieties which emerge during this period present their own challenge to the researcher: the impossibility of finding sufficient data from the large number of input varieties which influence the formation of e.g. Canadian English. Dollinger here proposes a heuristic approach to the “bad data” problem and tests this out in a study of the semantic development of modal auxiliaries in early Canadian English. Connie Eble’s paper “English/French bilingualism in nineteenth-century Louisiana: a Social network analysis” deals with a different “bad data” problem: that of accessing information about, and documentation of, language use in a bilingual community of the Late Modern period. Working within the social network framework first applied to sociolinguistics by James and Lesley Milroy in the 1970s (Mil-

328

Joan C. Beal

roy 1987), and applied to historical linguistics in the collection of papers edited by Ingrid Tieken-Boon van Ostade, Terttu Nevalainen and Luisella Caon (2000), Eble analyses an archive of family papers to discover the reasons for the shift from French to English between the late eighteenthand the early twentieth century. Ray Hickey’s paper “‘What strikes the ear’: Thomas Sheridan and regional pronunciation” tackles what is perhaps the most intransigent “bad data” issue of all: the lack of information concerning the pronunciation of any variety of English before the invention of the phonograph. Taking what we might call a historiographical or even philological approach, and taking as his primary source texts which might easily have been dismissed as “normative”, the works of the Irish-born elocutionist Thomas Sheridan, Hickey discusses the evidence for IrishEnglish pronunciation in the eighteenth century.

2.

Eble on bilingualism in nineteenth-century Louisiana

Social network studies of Late Modern English have hitherto tended to be based on the analysis of letters, since these provide both samples of language and evidence of network connections in writers and recipients themselves and third parties referred to in the letters. There are now several corpora of letters, such as the Corpus of Early English Correspondence (Raumolin-Brunberg and Nevalainen (2007)); the Corpus of Scottish Correspondence (Meurman-Solin 2007); the Corpus of Late Eighteenth-century Prose (van Bergen and Denison (2007)); and the corpus of nineteenthcentury Scottish correspondence compiled by Marina Dossena (2004). Other corpora of Late Modern texts, such as the Network of Eighteenthcentury Texts (Fitzmaurice 2007) contain substantial subcorpora of letters. The material analysed by Eble is slightly different in that it constitutes the whole of a family archive, Prudhomme Family Papers (#613), held in the Southern Historical Collection of the University of North Carolina at Chapel Hill. Whilst personal and business correspondence makes up part of this archive, there is also a diversity of other materials including “baptismal, marriage and death certificates, diaries, inspirational verse, personal correspondence, school lessons and composition books, scrap books and autograph books, greeting cards and religious cards, invitations, newspaper clippings about social and civic activities, and family genealogies … plantation journals, ledger books, inventories, promissory notes, acts of sale, and bills” (p. 338). Whilst much of this will have provided little in the way

Variation in Late Modern English: Making the best use of ‘bad data’

329

of linguistic data, and Eble herself admits that it was at first “disappointing”, it is invaluable for piecing together the details of family life: socioeconomic circumstances, kinship, friendship and business ties, etc. which are vital for a social network analysis. Whilst this data may be “bad” in the sense that it provides no direct evidence for linguistic variation and/ or change, nor any overt comment on the use of French or English, it does allow Eble to gain insight into the position of the Prudhomme family and other white creoles in plantation society and to demonstrate how the changes in their social networks led to the shift from French to English. The descendants of early French settlers such as the Prudhommes formed an elite whose bilingualism separated them from both the later (Acadian) French-speaking settlers and the English-speaking incomers, plantation managers and slaves. In the early nineteenth century, their position in Louisiana society was quite isolated, so their networks of kinship, marriage and business were dense and multiplex. Certain members of the family would have weak ties to other groups and individuals with whom business was conducted. Eble notes that, at this stage, French was used within the dense, multiplex family network, and English outside it: Phanor Prudhomme (1807–1865) kept his journal and conducted family correspondence in French, but his business and legal correspondence was in English of nativespeaker standard. Eble comments that he “functioned in the Englishspeaking world of the Americans” (p. 344). The next generation, exemplified by Phanor’s son Alphonse (1838–1919) used English for most business purposes, including the plantation journal, with French now relegated to being used for “personal and family reasons” (p. 345). By the next generation, French had become a foreign language learnt at school. Eble argues convincingly that the shift from French to English in the generation of Alphonse Prudhomme was facilitated by the weak network ties created as a result of the young people of this generation being sent out of Louisiana for their education, which was conducted in English. This paper provides a good example of the insights to be gained from applying the social network model to historical material. It also demonstrates the value of collections of family papers as providing, not only linguistic corpora to be mined for evidence, but also rich evidence of the social and historical background against which these texts were produced. There must be many more collections like the Prudhomme Family Papers yet to be discovered by historical linguists: Eble’s paper provides a useful model for their exploitation.

330

3.

Joan C. Beal

Dollinger on modals in Early Canadian English

Stefan Dollinger’s paper directly confronts the issue of “bad data” and puts forward the idea of using a heuristic method to overcome this. Although Dollinger deals with variation and change in modal verbs in early Canadian English, this is presented as a case study and the main thrust of the paper is methodological. Dollinger cites the definition of heuristic as: A process, such as trial and error, for solving a problem for which no algorithm exists. A heuristic for a problem is a rule or method for approaching a solution. (Blackburn 2005: s.v. heuristic, cited p. 358).

He points out that heuristic principles have been used extensively in other disciplines, such as engineering, computer sciences and social sciences, but, perhaps surprisingly not in historical linguistics, where they might usefully be applied to the analysis of “bad data”. Historical studies of Canadian English are few, and Dollinger, as a pioneer in this field, rightly argues that, if we were to wait until corpora of a size large enough to provide statistically sound evidence for influence from all the varieties which could have provided input to early Canadian English, progress would be very slow indeed. Instead, he suggests that we take what data we have and compare results from this with those from the richer and more extensive corpora of Present-day English, in order to measure the relative conservativeness or progressiveness of Canadian English from different genres and at various points in history. Dollinger’s analysis of modal verb usage is based on the comparison of data from the Corpus of Early Ontario English, preConfederation section (CONTE-pC); A Representative Corpus of Historical English Registers, version 1 (ARCHER-1), which provides American and British English data, and the Corpus of Late Eighteenth-Century Prose (CL18P), which he uses for data from the North-west of England. The Canadian, American and British data is subdivided into three genres: letters, diaries and newspapers. The data are also divided into four periods: 1776– 1799, 1800–1824, 1825–1849 and 1850–1899, but here the “bad data” problem becomes apparent, as only the British English component of the Archer corpus provides data for all periods. However, such comparisons as can be made from this incomplete dataset do suggest trends in the development of “root possibility” and “permission” senses of CAN in early Canadian English. For root possibility uses of CAN, Canadian English is closer to American English in those periods and genres for which comparable data is available. Compared with British English usage, Canadian Eng-

Variation in Late Modern English: Making the best use of ‘bad data’

331

lish appears more progressive in its use of “root possibility” CAN in diaries and more conservative in letters and newspapers. In the later periods, Canadian and British usage in newspapers converges, perhaps suggesting “a stylistic change towards more BrE norms in Canadian newspapers” (p. 367). With regard to “permission” uses of CAN, Dollinger’s analysis shows an increase in all genres in both British and Canadian English, suggesting a parallel development. Whilst the data suggests these interpretations, Dollinger admits that “no change of CAN is statistically significant” because of the low number of tokens involved. However, using heuristic reasoning, he is able to use the knowledge of CAN’s status as the unmarked variant, gained from more statistically robust studies of 20th-century English such as Coates (1983), to extrapolate an interpretation of his incomplete data as “a LModE parallel development in both CanE and BrE leading up to the PDE distribution” (p. 368). Dollinger then goes on to apply the same methodology to a comparison of the other “core” modals across the same genres, varieties and periods. His conclusion is that, whilst in the first period Canadian English usage is close to American English usage for six out of the nine modals, confirming the Loyalist base of early Anglophone Canadian society, overall, Canadian English can be seen as developing its own patterns rather than simply following British or American usage, which tallies with previous findings in the area of vocabulary. Readers accustomed to the statistical rigour of corpus linguistics might find Dollinger’s account of developments in the modal system of early Canadian English unconvincing, but, as he points out, the task of creating a corpus of sufficient size representing all the input varieties would be considerable. In the meantime, Dollinger’s “best guess” approach at least provides us with a set of working hypotheses. In fact, accustomed as we are to “making the best use of bad data”, historical linguists have always employed heuristic methods, but in recent times we have increasingly felt guilty about it. Dollinger’s paper provides a rationale for the heuristic approach, and at the same time a spur to the collection of “better” data which would allow us to test out his hypotheses more rigorously.

4.

Hickey on Thomas Sheridan and regional pronunciation

For those of us who are interested in the pronunciation of Late Modern English, the data is not so much bad as non-existent. For any period before the invention of the phonograph (1877), there is no possibility of finding

332

Joan C. Beal

“data” as such, and even for the first half of the 20th century surviving recordings are sparse and their preservation and subsequent availability often a matter of luck. The ONZE corpus (Gordon et.al. 2007) is a rare example of systematically collected data from this period (collected between 1946 and 1948, but including speakers born as early as 1851) being preserved intact and made available to linguists. Given the paucity of “real” data, historical phonologists have to reconstruct the pronunciation of earlier periods from the accounts of orthoepists, rhymes, puns and “incorrect” spellings in letters and diaries. I have discussed the relative merits of these sources elsewhere (Beal 1999: 38–47) and pointed out that, although Dobson (1957: I, 311) dismissed the eighteenth century as producing no writers to compare with the sixteenth- and seventeenth-century orthoepists who provided the sources for his magnum opus, the elocutionists and authors of pronouncing dictionaries whose works began to appear towards the end of the eighteenth century constitute a very rich source of evidence for variation and change in Late Modern English pronunciation. The relative neglect of this period which caused Charles Jones to refer to the eighteenth and nineteenth centuries as “the Cinderellas of English historical linguistic study” (1989: 279), has to a great extent been rectified, not least by Jones’s more recent (2005) work, which is to some extent a “Dobson” for the Late Modern period. Nevertheless, there is still a great deal to be said about the influence of individual authors – elocutionists rather than orthoepists in this period – on the development of varieties of English in the Late Modern period. Raymond Hickey focuses on one of the most influential elocutionists of the late eighteenth century, Thomas Sheridan. Like his ultimately more successful rival John Walker, Sheridan was an actor who turned his hand to elocution at a time when there was a demand, particularly from the upwardly-mobile and especially those living in Scotland, Ireland and parts of England remote from the capital (“the provinces”) for clear and explicit guides to “proper and polite” pronunciation. As I have pointed out (Beal 2004) the normative pronouncements of eighteenth-century authors can provide evidence for regional pronunciations, especially those which were most salient and hence most likely to be condemned. Hickey here takes up this suggestion and examines Sheridan’s comments both as a source of evidence for Irish-English pronunciation in the eighteenth-century and in order to determine the extent of Sheridan’s influence on the subsequent development of this variety.

Variation in Late Modern English: Making the best use of ‘bad data’

333

Hickey tells us that Sheridan was born and educated in Ireland and was the godson of Jonathan Swift. However, whilst Swift was “held up as a paragon of English style” by authors such as Lowth, Sheridan, as an Irish Catholic, albeit with impeccable connections, was subject to criticism. Dr Johnson questioned his credentials for teaching correct pronunciation to the English on the grounds that he was Irish. As Hickey points out here, the disparity between Swift’s reception and Sheridan’s could be due to the fact that the former was an Anglo-Irish Protestant clergyman and therefore respectable, whereas Sheridan was “a Catholic actor” with the temerity to be “a self-appointed arbiter of correct English usage” (p. 389). Hickey bases his account on Sheridan’s Rhetorical Grammar of the English Language (1781) in which he inserted a series of “Rules to be observed by the natives of Ireland in order to attain a just pronunciation of English” which had first appeared in his General Dictionary of the English Language (1780) and was later to be taken over wholesale by John Walker in his Critical Pronouncing Dictionary (1791). Hickey notes that Sheridan only mentions the features of Irish English that “struck his ear” (p. 394), and that he has nothing to say about the consonantal features which are so salient in present-day Dublin English. Nevertheless, Hickey’s comparison of Sheridan’s observations with later studies of Irish English does provide evidence for the earlier history of these features. Perhaps the most familiar of these to scholars in the field of English historical phonology is the use of /e:/ rather than RP /i:/ in words such as least, deal and beet. Sheridan condemns this as the chief “mistake” of the Irish, but goes on to discuss what is clearly a hypercorrection, where the “well-educated natives of Ireland” whose speech he describes, pronounce words such as prey, convey and bear, with /i:/, even though these words would not have this vowel in educated London English. This suggests that the stereotypical Irish-English /e:/ in, for instance Jaysis!, was already heavily stigmatised in the eighteenth century. This is corroborated by Maria Edgeworth who writes of Irish ladies who are “ashamed of their country” and “betray themselves by mincing out their abjuration, by calling tables teebles and chairs cheers” (1802, cited in Crowley 2000: 136). A similar instance of hypercorrection is suggested by Sheridan’s remark that the Irish used the vowel which he represented as , corresponding to /¡/ in IPA, in words such as pulpit, pudding, bush, bull, push and pull, where both present-day RP and Irish English would have /7/. We know from other sources such as Kenrick (1773) and Walker (1791) that the use of /7/ in e.g. cup, blood was already stigmatised as a Northern English feature, and that there was evidence of a similar

334

Joan C. Beal

hypercorrection amongst northerners (Beal 1999: 135), so Hickey’s inference is well supported. Hickey concedes that Sheridan’s evidence is piecemeal, and that he “appears to have been selective in his censure” (p. 405): the centralised diphthong in PRICE which was to become an “Oirish” stereotype in the nineteenth century is not mentioned, and, of course, where Irish pronunciation coincided with the “correct” usage of London, as was the case with the lengthened pronunciation of the vowel in cloth, off, etc. Sheridan makes no comment. Even here, though, negative evidence, the “dog that didn’t bark in the night” is useful in pointing to the social evaluation of variants in eighteenth century Britain. Hickey also discusses the extent to which Sheridan’s “rules” actually had any effect on the pronunciation of the “natives of Ireland”. Whilst his strictures appear to have had no immediate effect, some of the pronunciations which he condemned, such as /Ε/ in catch and /Ο:/ in psalm have since disappeared from Irish English. Whether they became stigmatized because of Sheridan’s censure, or were noticed by Sheridan because they were already shibboleths of Irish speech, is impossible to tell. Hickey’s paper is a good example of the ways in which eighteenthcentury normative texts can be mined for evidence of non-standard pronunciation, making accounts of the histories of these varieties possible, and providing time-depth to synchronic studies of variation and change. There are many more such texts available electronically in Eighteenth-century Collections Online, so I look forward to seeing much more work in this vein.

5.

Conclusion

The three papers in this collection provide different solutions and methodologies for dealing with “bad data”, and give pointers to future work on variation in Late Modern English. Eble demonstrates how family archives can be useful for social network studies, even when the texts themselves are largely written in standard varieties; Dollinger challenges the naysayers who might be disheartened by a lack of sufficient data for statistically significant results, advocating a heuristic approach which can extract workable hypotheses from incomplete datasets; and Hickey makes use of a normative text to provide evidence for the pronunciations which its author sought to eradicate. Trudgill (2002: 44) asserts that there is “no excuse” for confining

Variation in Late Modern English: Making the best use of ‘bad data’

335

accounts of Later Modern English to studies of British and US English: the papers in this section demonstrate that the problem of “bad data” is likewise no excuse for neglecting the study of variation in this period.

References Beal, Joan C. 1999 English Pronunciation in the Eighteenth Century: Thomas Spence’s “Grand Repository of the English Language”. Oxford: Clarendon Press. Beal, Joan C. 2004 “Marks of disgrace”: Attitudes to non-standard pronunciation in 18th-century pronouncing dictionaries. In Methods and Data in English Historical Dialectology. Marina Dossena and Roger Lass (eds.), 329–49. Bern: Lang. Beal, Joan C., Karen P. Corrigan and Hermann L. Moisl (eds.) 2007 Creating and Digitizing Linguistic Corpora. Basingstoke: Macmillan. Blackburn, Simon 2005 The Oxford dictionary of philosophy. 2nd ed. Online version. Oxford: Oxford University Press. Coates, Jennifer. 1983 The semantics of the modal auxiliaries. London: Croom Helm. Crowley, Tony 2000 The Politics of Language in Ireland, 1366–1922. A Sourcebook. London: Routledge. Dobson, E. J. 1957 English Pronunciation 1500–1700. Oxford: Clarendon Press. Dossena, Marina. 2004 Towards a corpus of nineteenth-century Scottish correspondence, Linguistica e Filologia 18: 195–214. Fitzmaurice, Susan. 2007 Questions of standardization and representativeness in the development of social networks based corpora: the story of the Network of Eighteenth-century English Texts. In Beal, Corrigan and Moisl (eds.), 49–81. Gordon, Elizabeth, Margaret Maclagan and Jennifer Hay 2007 The ONZE corpus. In Beal, Corrigan and Moisl (eds.), 82–104. Jones, Charles 1989 A History of English Phonology. London: Longman.

336

Joan C. Beal

Jones, Charles 2005 English Pronunciation in the Eighteenth and Nineteenth Centuries. Basingstoke: Palgrave Macmillan. Kenrick, William 1773 A new dictionary of the English language. London: John and Francis Rivington. Labov, William 1994 Principles of linguistic change. Volume 1: Internal factors. Oxford: Blackwell. Meurman-Solin, Anneli. 2007 The manuscript-based diachronic corpus of Scottish correspondence. In Beal, Corrigan and Moisl (eds.), 127–147. Milroy, Lesley 1987 Language and Social Networks. Oxford: Blackwell. Nevalainen, Terttu. 1999 Making the best use of “bad” data. Neuphilologische Mitteilungen. 100 (4): 499–533. Raumolin-Brunberg, Helena and Terttu Nevalainen. 2007 Historical sociolinguistics: The corpus of early English correspondence. In Beal, Corrigan and Moisl (eds.) 148–171. Sheridan, Thomas. 1780 A General Dictionary of the English Language. London: R & J. Dodsley, C. Dilly and J. Wilkie. Sheridan, Thomas. 1781 A rhetorical grammar of the English language calculated solely for the purpose of teaching propriety of pronunciation and justness of delivery, in that tongue. Dublin: Price. Tieken-Boon van Ostade, Ingrid, Terttu Nevalainen and Luisella Caon (eds.) 2000 Social Network Analysis and the History of English. Special issue of the European Journal of English Studies (EJES) 4.3. Trudgill, Peter. 2002 The history of the lesser-known varieties of English. In Alternative Histories of English. Richard Watts and Peter Trudgill (eds.), 29–44. London: Routledge. Van Bergen, Linda and David Denison 2007 A corpus of late eighteenth-century prose. In Beal, Corrigan and Moisl (eds.), 228–246. Walker, John 1791 A Critical Pronouncing Dictionary and Expositor of the English Language. London: G.G.J. and J. Robinson and T. Cadell.

English/French bilingualism in nineteenth century Lousiana: A social network analysis

Connie Eble Dialects of French remained the ordinary spoken language of southern Louisiana west of the Mississippi River until World War II – almost a century and a half after the area had become part of the United States by the Louisiana Purchase of 1803. However, French had been the ancestral language in other parts of the state as well, including the farmlands drained by the Red River, whose center of population was Natchitoches, the oldest European town in the Louisiana Purchase territory, about 250 miles northwest of New Orleans. In that part of the state, French disappeared as the language of public life by the time of the Civil War. This essay applies the notion of social network to an archive of family papers to explore why. It describes the changing use of written French and English during the nineteenth century by a cohesive group of white creoles1 living in rural Natchitoches Parish in northwestern Louisiana, between what are now the cities of Shreveport and Alexandria. The social network itself and the language practices of its members are inferred from the letters, journals, scrapbooks, and other writings preserved in the Prudhomme Family Papers (#613), held in the Southern Historical Collection of the University of North Carolina at Chapel Hill.2 The Prudhomme Family Papers (#613) occupy 41 linear feet of shelf space and comprise more than 16,000 items dating 1765–1997. Preserving on paper the life stories of a single family living in the same parish for more than two centuries and in the same house for 170 years, they include baptismal, marriage and death certificates, diaries, inspirational verse, personal correspondence, school lessons and composition books, scrap books and autograph books, greeting cards and religious cards, invitations, newspaper clippings about social and civic activities, and family genealogies. The family’s financial condition for seven generations is shown by business correspondence, plantation journals, ledger books, inventories, promissory notes, acts of sale, and bills. The bulk of the surviving papers of the de-

338

Connie Eble

scendents of Jean Baptiste Prudhomme, the first family member born in Louisiana, in 1736, form the Prudhomme Family Papers (#613) in North Carolina. Smaller collections are held at Watson Memorial Library of Northwestern Louisiana State University in Natchitoches and the Historic New Orleans Collection. For a linguist, this immense archive is at first disappointing. It reveals nothing directly about language change or variation or contact between French and English, or even opinions or perceptions about language. I have not yet found one comment on the replacement of French by English. Yet the shift in public life from the language of the founding population to the language of the newcomers must have been meaningful to this family. This essay aims to extract information about language shift by the people whose lives are represented in this archive by thinking of them as a social network. A social network approach to language analysis takes as its premise that social connections create networks that serve as mechanisms for transmitting linguistic influence. Observing who speaks to whom, how often, and about what can be particularly useful for examining linguistic variation that cannot be due to major sociolinguistic factors like socio-economic class, ethnicity, educational level, or regional identity, i.e., variation among speakers who do not differ by these measures. For example, John Gumperz (1982) appealed to the notion of social network to explain the shift from Slovenian/German bilingualism to German in a small, rural Austrian village with little social stratification. Social network analysis was established as a tool of current synchronic sociolinguistics by the work of James and Lesley Milroy in three urban communities in Belfast, Northern Ireland, in the 1970s (Milroy 1980, 1987).3 Briefly and broadly, here are the main components of the Milroys’ network approach. A social network is all of the relationships that an individual has contracted with others, taken together. The point of entry is the individual. The patterns of interactions among individuals constitute systems that can be characterized and measured by type and strength. Networks can differ in scope: an individual’s relationships can be confined to a community (closed) or extended outside the community (open). The number of people to whom an individual is connected and who are connected to each other determines the density of the network. In the most dense network, everyone knows everyone else. The number of different ways in which members of a network are linked is the measure of multiplexity. Discovering the social features that constitute meaningful multiplex links

English/French bilingualism in nineteenth century Lousiana

339

for a network is a challenge to investigators, for properties that are important may differ from one community to another and may change. Other characteristics of linkages that contribute to the strength of ties and might elucidate the social dynamics are frequency, duration, and reciprocity. Dense and multiplex linkages can form sectors or clusters, e.g., a workplace cluster or a kinship cluster. An individual can be central to a network or peripheral. In the urban populations that the Milroys were studying, dense, multiplex networks – in which most of the speakers were linked to each other by several measures – tend to be norm-enforcing and to conserve language features. In other words, strong network ties resist change. Loose ties, on the other hand, provide the route for innovation and change. Mobility and migration often place speakers in situations in which their ties to a cohesive, well established group are loose (or weak), and their position is marginal. The weak ties of these individuals tend to be the conduits to localized networks of new information, ideas, and linguistic forms (Milroy 2000: 217–219). Lippi-Green (1989) showed that social network analysis could help explain a phonological change in progress in an entirely different setting – in a small, rural German-speaking community in western Austria. A workshop in conjunction with the 1998 International Conference on English Historical Linguistics took as its aim the application of social network analysis to older stages of the English language. Seven papers presented at the workshop were published together in the European Journal of English Studies in 2000, along with an introductory essay by Lesley Milroy. The papers span a millennium, from the monastic community of the Benedictine Reform of late 10th century England to the contemporary elite of Dublin. They reveal a range of problems in extending a tool developed for contemporary urban sociolinguistics to imperfect linguistic and sociological data from the past. A major problem is determining which ties were important to the individuals and groups in their lifetime. In general, the studies do support the generalizations that strong network ties favor conservatism, whereas loose ties serve as the agents for innovation. Deducing social networks from surviving written records also requires a good understanding of the historical context in which the documents were produced. It is against the following historical backdrop that the language practices preserved in the Prudhomme Family Papers must be examined. French-speaking settlers came to what is now the state of Louisiana in the early decades of the eighteenth century. The first colonial town, Natchi-

340

Connie Eble

toches, grew up around Fort Jean Baptiste, the military outpost established in 1714 on the banks of the Red River on the route to Texas. It marked the western extent of France’s claim. Just fifteen miles to the west, the mission, and later fort, Los Adaes marked the eastern boundary of Spanish Texas. The first Prudhomme came as a French soldier to Natchitoches in 1716 and stayed as a merchant and trader, marrying a Parisienne who came with a shipload of young French women to provide wives for the colonists. Their son Jean Baptiste Prudhomme prospered and became the King’s Physician under the French and a landowner. In 1763 as the result of the French and Indian Wars, France ceded Louisiana to Spain. Although Spanish officials took over the administration of the colony, few Spanish speaking colonists migrated to Louisiana, and the culture and language remained essentially French. Jean Baptiste’s son, Jean Pierre Emanuel Prudhomme, in 1789 secured a land grant from the Spanish governor for a tract on the banks of the Red River thirteen miles south of Natchitoches, called Ile Brevelle (now Bermuda, Louisiana). There, relying on the labor of slaves, he successfully cultivated tobacco and indigo and was perhaps the first planter in Louisiana to grow cotton. In 1821 he moved his family into a newly-constructed, fiveroom plantation house that is now called Oakland. For six generations his direct descendants lived in the house – his grandson, great-grandson, and great-great-grandson each celebrating fifty years of marriage there. Jean Pierre Emanuel’s plantation has been recognized by the United States government as one of two “bicentennial farms” west of the Mississippi River, having been owned and managed by the same family for 200 years. The Prudhomme Family Papers begin during the lifetime of Jean Pierre Emanuel, when Louisiana was a colony of Spain, and extend to 1997, when the house and its remaining outbuildings and land were acquired by the National Park Service as part of the Cane River Creole National Historic Park. Under both French and Spanish rule, the vast Louisiana territory was sparsely populated by European colonists and their descendants, except for the most southern portion around the port of New Orleans and the bayou and prairie regions to its south and west where Acadian immigrants displaced from Canada had settled. In addition, there were a few important enclaves, like St. Louis in Missouri and the Natchitoches area, which included an independent and prosperous community of free people of color who spoke French and were Catholic (Mills 1997). Regardless of their ethnic origins, the language of colonials in Louisiana and their creole descendants was mostly some variety of French. Understanding issues of language

English/French bilingualism in nineteenth century Lousiana

341

contact in Louisiana requires recognizing the continuum of varieties of French in use historically. A conventional description locates three points on the continuum: a variety close in vocabulary and grammar to standard continental French and used in writing; Acadian French based on provincial dialects brought to Canada in the seventeenth century and influenced in Louisiana by contact with other languages; and a creole developed by slaves and their descendants (Breton 1979: 22–23). Even before the Louisiana Purchase, English-speaking Americans saw opportunities afforded by the vast Louisiana territory and crossed the Mississippi River into the Spanish colony, exchanging nominal allegiance to the Spanish crown and the Catholic Church for land grants or administrative positions. The acquisition of Louisiana by the United States removed all impediments to their westward migration. During the nineteenth century, the influx of “Americans” overwhelmed the creoles established along the banks of the Red and Cane Rivers. The newcomers were quite different from the ancienne population: their forebears had come from the British Isles, they spoke varieties of English, and they were Protestants. One study of the settlers of the hill country of northern Louisiana maintains that 73.3% of them came from the five states of Georgia, Alabama, South Carolina, Tennessee, and Mississippi (Trout 1996: 473). In northern Louisiana, the development of an agricultural economy based on cotton also contributed to the region’s rapid Americanization, as it allied the commercial interests of the area to those of other southern cotton-producing states and increased the need for slave labor, a need that could be filled only by acquiring slaves from states east of the Mississippi. Linguist Michael Picone (2003) thinks that the large number of slaves required by cotton production forced the importation of monolingual English-speaking slaves from other parts of the South and was an important factor in the shift to English in the region. To this day, the cultural divide between northern and southern Louisiana is apparent, with northern Louisianans more similar to the people of Mississippi, Alabama, Georgia, and the Carolinas in folkways, religion, speech, politics and culture than they are to southern Louisianans. Beginning about 1830, the Prudhomme family became part of the linguistic and cultural minority of northern Louisiana. Many political leaders in the United States had opposed the Louisiana Purchase of 1803 because of the religious, cultural, and linguistic incompatibility of the people of the United States and the people of the Louisiana territory. To assure the rights of the non-francophone minority of the residents of Louisiana in 1812 when it became the eighteenth state of the Un-

342

Connie Eble

ion, the first state constitution was published simultaneously in both French and English. It provided that all laws passed by the legislature be promulgated in “the language in which the constitution of the United States is written” and that other legal documents be preserved and made available in English (Smith-Thibodeaux 1977: 32–33). The provision for public documents in French was not specified, as French was then the language of the majority of voters and elected officials. The state constitutions of 1845 and 1852 specifically required that the constitution and the laws of the state be published in English and in French (33). However, in 1847 twenty parishes in northern Louisiana were allowed to use English as the sole language of official publication, except in instances when residents of the parish required French (Trépanier 1988: 30). In the same year, public elementary schools were established in the state, with no specification of the language of instruction (29). The “Carpet Bag Constitution” of 1868 mandated English only, but the constitution of 1879 permitted laws to be published in French also, and in heavily French-speaking areas some provision for French in schools was possible (Smith-Thibodeaux 1988: 36–37). The constitution of 1921 is generally regarded as the deathblow to the French language in Louisiana, eliminating it as the medium of instruction in schools and reducing the numerous Louisiana-born monolingual speakers of French to the status of second-class citizens. The social network in which the members of the Prudhomme family led their lives and made their linguistic choices was embedded in the historic context sketched above and was influenced by it. The natives of Natchitoches Parish whose lives can be inferred from the Prudhomme Family Papers formed a dense, multiplex network. They identified themselves as creoles and, for the most part, distinguished themselves from Frenchspeaking Acadians, most of whom lived on subsistence farms or trapped and hunted for a living. Laura Locoul Gore (granddaughter of Phanor Prudhomme’s second wife), in her Memories of the Old Plantation Home, recalled, “The Acadians were the poor, uneducated whites. They seemed to have no ambition to learn …” (63). The white creoles of the Red River area constituted an elite. The names of the most prosperous of them show up in the Prudhomme Family Papers. They were related by birth or marriage or both to creoles in this area and to others along the Mississippi River and in New Orleans and St. Louis. They were Catholics. The basis of the family’s wealth was cotton. They owned slaves. In the most prosperous times, some owned townhouses in Natchitoches as well as the plantations on which their livelihood depended. Although they were wealthy, there were constant

English/French bilingualism in nineteenth century Lousiana

343

financial anxieties, as the entire local economy was subject to the vagaries of weather, flooding, and the price of cotton in Europe. The archive gives abundant evidence of lending, borrowing, buying, and selling among members of the network, both male and female. (Creole women were privy to financial matters and often controlled their own inheritances or managed the family corporation (Marmillion 2001: 135–137).) “In the isolation of the cotton fields around Natchitoches, the Creoles were content to live among themselves in year-long schedules of family dinners, outings, religious devotions and inter-marriage” (144). For most of the nineteenth century, they were bilingual in French and English. The relationships of an individual who figures prominently in the Prudhomme Family Papers can illustrate some of the characteristics of the Natchitoches Parish creole network in the period from the Louisiana Purchase to the Civil War. Pierre Phanor Prudhomme’s life (1807–1865) spans roughly the entire antebellum period. (Except for official documents, most of the time he signs his name as Phanor Prudhomme.) Most evident in the documents and difficult to untangle are the relationships of kinship. Phanor was the youngest of six children; his mother was a Lambre. His siblings married into the local St. Anne, Metoyer, and Lecomte families. Phanor himself married two granddaughters of his aunt Susanne, who were daughters of his cousin Marie Aurore Lambre and Benjamin Metoyer. Another of his aunts married Phillipe DuParc of St. James Parish on the Mississippi River. The daughter of Phanor’s second wife (and his niece) married the grandson of his aunt DuParc. Their infant daughter Laura Locoul Gore and her mother lived with Phanor during the Civil War. After the war, when grown, Laura participated in creole society in New Orleans. Phanor’s older son married the daughter of Ambroise Lecomte, whose plantation was near Phanor’s and who shared Phanor’s love of horse racing. Phanor’s younger son married a Buard, the cousin of his first daughter-in-law. Phanor’s daughters married young men of the Brazeale family. Phanor’s brother-inlaw Felix Metoyer shows up often in papers concerning inheritance, and he served for a while as executor of Phanor’s estate. Other connections of kinship give Phanor ties to the remainder of the landholding creole families of the area, who bear the surnames Bossier, Cloutier, Deblieux, Hertzog, Rachal, and Sompayrac. The network was clearly dense – not only did everyone know everyone else, they were all related by blood or marriage. Perhaps more important, many of the ties were multiplex. Phanor was related to Ambroise Lecomte, for instance, not only more than one way by kinship but also as a neighbor, fellow cotton planter, fellow thoroughbred horse

344

Connie Eble

breeder and member of the Jockey Club, fellow chef de famille expected to support financially the building of a new Catholic church, and fellow financial backer of the Confederate military unit from Natchitoches Parish. Phanor’s ties extended beyond this cluster, however – to cotton factors in New Orleans, to English-speaking overseers of his plantation, to his slaves and hired hands, to school masters and mistresses, to inventors of farm machinery, to government officials, and so forth. Although the family papers contain little on this topic, Phanor also served in the Louisiana legislature and was a delegate to the 1844 state constitutional convention. Within the kinship cluster, his language was almost entirely French. Outside that cluster, he used both French and English. Until the 1830s, the language of the Prudhomme Family Papers and, by inference, the language of their social network, was French. The first generation of Prudhommes after American acquisition preferred French in both personal and business affairs but used English beyond their creole circle in correspondence with government officials and other non-creoles with whom they did business. This is the generation of Pierre Phanor Prudhomme, 1807–1865. The plantation journals that he kept 1837–1841 are in French, mainly a list of dates followed by an event or observation.4 [1837] [1839]

22 mars

Commencé a planter le Coton 25 juillet Beau temps après beaucoup de pluie Novembre 10 donné les Souliers aux Domestiques de la Maison. Decembre 15 donné les Souliers aux Nègres du Champs (Folders 268–269)

Phanor Prudhomme’s Cotton Book for 1836, however, was kept by J. F. Culbertson, probably an overseer. It opens, “Began picking cotton Monday August the 29th 1836.” Although the book is mostly numbers tracking the amount each slave picked, the few notations are in English, for example, “on this side”, “over the river”, “total”, and “lbs” ‘pounds’ (Folder 267). The Cotton Book is signed and dated by both Culbertson and Phanor Prudhomme on November 15, 1836. Virtually all of Phanor Prudhomme’s personal correspondence is in French, and so is a large portion of his business correspondence, mostly with cotton factors in New Orleans. But in 1850 when he wanted to dispute a claim against his father’s estate and to inquire about installing a new kind of cotton press (Folder 143), he wrote in English with the usage and fluency of a native speaker. He functioned in the English-speaking world of the Americans: he read Southern Cultivator and Scientific American, took

English/French bilingualism in nineteenth century Lousiana

345

his daughters and nieces to New York City, and enrolled his sons in schools in Connecticut, Virginia, North Carolina, and Washington, DC. Although the writings of Phanor Prudhomme usually show no mixing of the languages, a letter to Ambroise Lecomte from New Orleans dated 28 Mai/60 reports a conversation with an American owner of racehorses. The letter is in French, but Phanor gives the words of the conversation in English (Folder 902). Although the documents cannot reveal the phonology of his French or English, Phanor Prudhomme appears to have been comfortably fluent in both languages. Phanor Prudhomme’s older son and heir, James Alphonse (1838–1919), left writings that show that he too knew both French and English well. English was his primary written language, and he seems to have written French only for personal and family reasons. When he was away at school, Alphonse wrote his diary entirely in English. As a student at the University of Virginia and then the University of North Carolina in the late 1850s, he would certainly have spoken English in all school courses and activities. Following his father’s death in 1865, Alphonse assumed the responsibility of running the family plantation. In contrast to his father, who ordinarily corresponded with longtime cotton factors in New Orleans in French, Alphonse conducted business with them in English. By 1870 both the plantation journals kept on site and communications with outsiders concerning plantation operations were in English. Alphonse is representative of the generation of creoles who grew up in the vicinity of Natchitoches during the 1840s and 1850s and came to maturity on the eve of the Civil War. When they were not away at school, where their lives were ordered and disciplined, the large group of sisters and brothers, cousins, and cousins-of-cousins led a merry life of visiting, attending dances and parties, and preparing for the next social event. The plantation area around Bermuda became known throughout French creole society as the Côte Joyeuse. These young adults all knew the standard variety of French taught in school, as shown by school notebooks of lectures on French literature and history and by school compositions written in French. And they probably understood and spoke some variety of French with older family members and with black creole household servants. However, the generation of J. Alphonse Prudhomme corresponded with each other in English and kept their diaries in English. Fulbert Cloutier’s brief Civil War diary, written in 1864, is in English except for a short French poem, “Le veritable Amour,” copied onto the first page (Folder 102). The notebook kept during the Civil War by Elise Lecomte, Alphonse’s future wife, is

346

Connie Eble

almost entirely in English, with petitions to God sometimes in both languages, for example, “Protect my loved Phonse from all dangers” and “Mon Dieu, ayez pitié de nous” (Folder 891). Strong, dense network ties usually favor conservatism in language. For this old, established network that found itself in the linguistic minority visà-vis newcomers, the close-knit ties did support the retention of French but retracted it into the private domains of family and religious practice. Further, the type of French they wrote was a standard variety instilled by education that gives no information about their spoken French and no hint of the local colloquial varieties used by monolingual black and white speakers that they came in contact with in Natchitoches Parish. Economic prosperity for members of this network required going outside the closed creole cluster and forming loose network links with monolingual English speakers. Weak links peripheral to a dense multiplex network provide the route for various kinds of innovation, including linguistic innovation, and the Prudhomme papers give evidence of this. The Red River creoles seem to have deliberately extended their network in an important way – in the education of their children. Although some among the wealthy in Louisiana employed tutors for their children, beginning early in the nineteenth century the Prudhommes appear to have sent their children – both boys and girls – to boarding schools. The earliest generations of the French creole network developing around Natchitoches in the eighteenth century were not all literate. As late as 1809 and 1810, two of them, Marie Louise Leconte and Louis Rachal, signed promissory notes with a “marque ordinaire d’une croix,” i.e., with an X (Folder 141). At the same time, others in the network were literate and perhaps even educated. By 1828, young Henri Herzog was enrolled at St. Joseph’s in Bardstown, Kentucky, and writing home to his parents.5 Mon cher Papa, J’ai recu votre derniere letter du 24 aout vous me dite croire que quelqu’un m’ede a ecrire mes lettres mais je vous assure que je les aicrit par ma propre main je suis etonné moi meme que j’aie apris si vite le stil de mes lettre je ne pouvait apeine ecrire une lettre quand je suis arivé au colege mas mon cher pere je vous assure par mon honneur que personne ne m’aide pas … . je ne fait qu’apprendre l’anglais et le français c’est parce qu’il faut que l’on aprenne bien l’anglais pour apprendre le latin et l’espagnole … (Folder 887)

English/French bilingualism in nineteenth century Lousiana

347

Henri’s native tongue is French, but, as the errors in his letter show, he is just beginning to learn how to write. It is also clear that he is expected to learn English before he will be allowed to study Latin, a compulsory subject at that time for entrance to higher education. Because of the schools established by two French orders of nuns, the Ursulines and the Religious of the Sacred Heart, many creole girls remained in Louisiana for their schooling, though the difficulties of travel usually meant that they returned home only once a year, and sometimes not that often. A beautiful handwritten arithmetic book in the collection attests that Lise Metoyer (the daughter of Benjamin Metoyer and first wife of Phanor Prudhomme) attended the Ursuline school in New Orleans. It is inscribed “Cahier d’Arithmétique fait par Lise Metoyer au couvent des Ursulines de la Nouvelle Orléans 14 septembre 1832” (Folder 111). At Ursuline schools in America, morning lessons were conducted in French and afternoon lessons in English (Aycock 1993: 213). Lise learned arithmetic in French. Although there is no record of how or where Phanor Prudhomme was educated, his “cahier D’Arithmétique[s]” also survives. This notebook contains a basic, structured course in arithmetic entirely in French. The handwritten book appears to have been either copied or written down from dictation from an established text, opening “On entend par mathématiques une science qui traite des grandeurs ou quantités … . L’Arithmétique est la science des nombres” (Folder 130). Both Lise and Phanor seem to have acquired their education in a Frenchspeaking context. In 1821 the Religious of the Sacred Heart, a French order of nuns, established a school for girls in Grand Coteau, Louisiana, followed by one sixty miles upstream from New Orleans on the Mississippi River in 1825, and one in Natchitoches itself in 1847 (Calan 1937: 775–780). By the time of the Civil War there were also Sacred Heart academies in St. Louis and in Manhattanville, New York, and at least one of the young women writers in the Prudhomme collection wrote a composition at Manhattanville. All of the Religious of the Sacred Heart were under one superior general, and all of the private academies that they established followed the same Plan of Studies and held many of the same religious and cultural activities. Phanor Prudhomme’s daughters and the daughters of his second wife attended the academy in Natchitoches in the 1850s. The 1851 report card of his stepdaughter, Désirée Archinard Locoul, listed as subjects Lecture française, Lecture anglaise, Grammaire française, Grammaire anglaise, Traduction anglaise, and Traduction française (Gore:144).6 Désirée’s daughter, Laura Locoul Gore, writing her memories for her children in 1935 and 1936, re-

348

Connie Eble

called about her mother, “She sang and played delightfully and had perfect command of both English and French” (24). During the nineteenth century, the Ursuline and Sacred Heart schools in Louisiana preserved and transmitted French language and culture and Catholic devotional practices while preparing the girls in their charge to live as English-speaking American women. Many of the young cousins of the Côte Joyeuse of James Alphonse Prudhomme’s generation were sent beyond Louisiana for their education. A number went to school in Bardstown, Kentucky, where the first Catholic diocese in United States territory west of the Alleghenies had been established and where boys could board at St. Joseph’s and girls at Nazareth Female Seminary, founded by a new American order of nuns, the Sisters of Charity of Nazareth. The 1850 census for Nelson County, Kentucky, lists 120 females nineteen years and younger living at the Nazareth Female Seminary. Of those, 29 (24%) show Louisiana as their birthplace (US GenWeb Census Project). At least three of the surnames – Bossier, Lambre, and Lecomte – are part of the Prudhomme family. The 1850 census reports 130 students in residence at St. Joseph’s for boys, fifty-five of them (42%) listed with Louisiana as their birthplace (US GenWeb Census Project). At both schools most of the other students had Irish, English, or German surnames, and most came from Kentucky, Mississippi, and Alabama. In Catholic schools outside of Louisiana, creole children learned English well and associated with the children of other prosperous Catholics, often from other parts of the South. Although items in the Prudhomme papers show that French grammar, literature, and history were subjects of study, in the Bardstown schools in the 1850s, the language of instruction was English, and English composition was emphasized. Pauline Bossier’s composition book of 1852 included essays on “Filial Affection,” “Who Approaches Nearest to Happiness on Earth?” and “Is Happiness Necessary to the Enjoyment of Life?”. Under the title “The Pleasures of Memory,” one of Pauline’s compositions uses President George Washington as an example of someone whose noble behavior must have yielded joyful memories: How much joy must not the noble, the illustrious Washington have experienced, when his memory took him back to the time, when he attained the height of his glory, when he promoted to his country’s good, by subjecting himself in every thing for so noble a cause. (Folder 96)

English/French bilingualism in nineteenth century Lousiana

349

Pauline’s last entry is a personal note: “We have read our last compositions for the year 1852. Great many of the sisters were there and Mother C. did something or at least said something I did not like at all. Oh! I am glad to think I will have six weeks to rest in, I mean I will not be bothered by those abominable compositions …” (Folder 96). At least two of the boys represented in the Prudhomme collection went to school in New England, Henry Herzog – the Henri who had to learn English before tackling Latin at Bardstown a decade earlier – and J. Alphonse Prudhomme. A year after his mother’s death, Alphonse, then fifteen years old, was attending the Commercial Institute of New Haven in Connecticut. A letter from Wm. H. Russell of New Haven to Phanor Prudhomme, dated October 18, 1853, acknowledges receipt of payment and adds, “Your son, I am happy to say, is in good health. We wish all our pupils were as amiable and correct as he” (Folder 144). In 1856 Alphonse went from New Haven to the University of Virginia in Charlottesville for two years, then to the University of North Carolina in Chapel Hill for two years, where he graduated with a Bachelor’s degree in Science in 1860. His course notebooks are entirely in English. One, filled cover to cover with chemical descriptions of candles, inks, gunpowder, pottery, and so forth, inside the back cover contains – as if a doodle – two instances of Sault Salt, elaborately written with flourishes (Folder 116). Perhaps in an idle moment his thoughts turned to Louisiana and brought the French word to mind. The year before the outbreak of the Civil War, Alphonse’s younger brother attended Georgetown College in Washington, D.C. When the girls finished school, they returned to Natchitoches Parish to await marriage, spending their time reading, playing the piano, doing needlework, making scrapbooks, and keeping in touch with each other for news of the latest parties or budding romances. They often spent several days at a time visiting each other’s plantations or meeting at their grandmother’s house in Natchitoches. In those days, a trip from one plantation to another or from a plantation in Bermuda to town thirteen miles away could not be undertaken on a whim. The young women kept in touch by mail or by the kindness of friends and family willing to deliver their notes and letters. The Prudhomme Family Papers contain a series of about thirty letters dated 1858 and1859 from Coralie Buard to her cousin Fanny Bossier. They are all in English, with just a handful of French sentences and phrases. The code-mixing in the September 12, 1858, letter from Coralie to Fanny is typical.

350

Connie Eble

There were no Creole gentlemen present, excepting our dear cousins (Herzog) and Felix. The party was gay and went off better than we had expected. John C.[loutier] was anxious to go, but on account of his brother’s death he was obliged to sacrifice his pleasure. He was seen at the door and window during the party. Ainsi, tu vois que son envie etait forte. As for the fishing party last Thursday, we also enjoyed it better than we had expected in the company of American girls and gentlemen. (Folder 1)

The content shows that despite their schooling away from their Frenchspeaking home and their use of English with each other, the young women still considered themselves culturally distinct from the monolingual English speakers that by then formed the majority of the population of Northern Louisiana where they lived – still in 1858 referring to them as American. For the young Red River creoles of Alphonse’s generation, their boarding school experiences made English their primary language. At the same time, at home in Natchitoches Parish, their parents – also bilingual – were becoming accustomed to an English-speaking environment. Many of their slaves spoke English, as did the overseers they hired to run their plantations and the people they dealt with producing and marketing their cotton. By March 1860, even the flier announcing the opening of the Natchitoches Races at the Prudhomme Course was in English only, despite the fact that the 114-member Natchitoches Jockey Club that sponsored the races was a bastion of old line creoles. Among the surnames on the subscription list were those that appear in the Prudhomme Family Papers: Bossier, Breazeale, Buard, Cloutier, Deblieux, Hertzog, Lambre, Lecomte, Metoyer, Prudhomme, Rachal, and Sompayrac. After 1870, the Prudhomme Family Papers contain almost no items written in French. The exceptions are a handful of letters and a few poems, prayers, and good wishes on the occasion of baptisms, weddings, and anniversaries. The generations following continued to study both French and English at school. Desirée Archinard Locoul’s daughter Laura attended Mrs. Cenas’ School for Young Ladies in New Orleans about twenty-five years after her mother had attended the Sacred Heart Academy in Natchitoches. Laura took almost the same subjects as her mother had, but Laura’s 1877 report card was in English and listed the subjects in English. Laura earned a perfect grade of 10 in Personality, in Manners and Conduct, and in Music. She scored 9 in English Recitation and in English Written Exercises and 8½ in English Elocution. Her grades in French were slightly lower, with 8 in

English/French bilingualism in nineteenth century Lousiana

351

French Grammar, in French Written Exercises, and in French Elocution and 7 in French Recitation (Gore 2001: 72). By 1920 when Alphonse and Elise’s granddaughter Adele Prudhomme was a student at St. Mary’s Academy conducted by the Sisters of Divine Providence in Natchitoches, French was still a required subject. But school notebooks show that the grammatical explanations of French were in English. In an eight-stanza poem that Adele wrote describing the school day, she does not mention French but gives English a complete stanza: And then of course there’s English A study all our own you know, Because it is our own language. Like French is for the French and so. (Folder 120)

In the twentieth century, English was the language of the white creoles of the Red River area. As shown by Adele’s verse, they learned French as a foreign language, as the language of a country across the Atlantic Ocean, just as other Americans did. The Prudhomme Family Papers tell the story of at least a half century of sustained English/French bilingualism only indirectly. In the documents, the two languages are kept separate, with almost no interspersing of the words of one language into the other. The collection contains few examples of grammatically faulty English or French, with the exception of stray miswritings and variable spellings. Through a large part of the nineteenth century, bilingualism in standard varieties of French and English was an important property of the network of white creoles of privilege in the Red River area. Knowledge of French differentiated them from the American immigrants who quickly outnumbered them after the Louisana Purchase. Knowledge of English differentiated them from the uneducated monolingual speakers of French like the Acadians and black creoles. Their bilingualism served their status as elites. Within a decade after the end of the Civil War, the Prudhomme Family Papers contain French only in devotional materials, phrases in memory books, and a very small number of communications between the generations in the creole network. In northern Louisiana in the public sphere, French had disappeared. After about 1870, whatever remained of either spoken or written French leaves no traces in the Prudhomme Family Papers.

352

Connie Eble

The documents left to posterity by families, like the Paston and Cely letters of fifteenth century England, have long been valued as evidence of past forms of the language. This study uses a collection of nineteenth century family papers from Natchitoches, Louisiana, not for analysis of the linguistic forms but for evidence of shift from French to English. The archive is treated as a network of the individuals whose lives are represented within its documents. The dominant cluster within the white creole network thus constructed, which is both dense and multiplex, is one of kinship. The original language of the network was French, which was retained within the network long after French had been displaced by English as the sole language of the majority of the population of northern Louisiana and the language of the public sphere. But by the end of the nineteenth century, members of the close-knit creole network also wrote only English. The strength of the ties within the kinship network had probably not diminished. However, weaker ties to individuals outside the cluster had facilitated the importation of English as an addition to French. The weaker ties were formed for economic reasons, i.e., to ensure the continued prosperity and privileged status of the members of the creole network, the founder population, after English-speaking newcomers became the majority in the area. In particular, sending creole children away to schools where they were taught to speak, read, and write English put the members of the younger generation in a position to serve as agents of innovation to their dense and strong kinship network in Louisiana. The English language became one of the properties of the linkages among the young members of the creole network and set them apart from their French-speaking contemporaries in Natchitoches Parish who were less privileged. By the beginning of the twentieth century, the generations of creoles for whom bilingualism had been a mark of class had passed away, and the creoles joined the monolingual English majority.

Notes 1. In this essay, creole means ‘native to the colony or descended from natives to the colony’, i.e., in Louisiana before 1804. 2. Because archives are organized according to the principle of provenance, all the documents saved by a particular family and entrusted to a library or museum are ordinarily kept together rather than dispersed by subject matter. This principle of organization does not facilitate research on a particular topic, e.g., the vocabulary of disease in 19th century America or thorough-bred racehorses

English/French bilingualism in nineteenth century Lousiana

3.

4.

5.

6.

353

in the antebellum South – though the writings preserved within the Prudhomme Family do hold such information. However, collections of papers saved within a family and transmitted to succeeding generations can be a good source for reconstructing past networks of who communicated with whom, how often, and about what. Furthermore, the documents saved by a family often include writings by individuals not linked by kinship, making the network represented by the papers larger than the biological family who preserved them. This is the case with the Prudhomme Family Papers. In the discussion period following the presentation of this paper at the SHEL4 conference, Ray Hickey suggested that local court documents such as depositions would likely give a better glimpse of the French used in the area. He is undoubtedly right. However, because this essay is in large part an attempt to explore whether or not family archives are amenable to a network approach, I have confined this study specifically to the network delineated by this one family’s papers. Lesley Milroy’s 1980 (2nd ed. 1987) Language and Networks is the foundational document of this approach. Chambers (1995) and a 20-page essay by Milroy (2002) survey the scholarly work in social network analysis of the past two decades in the context of the major issues in sociolinguistic theory. 22 March Began to plant cotton 25 July Good weather after much rain November 10 Gave shoes to the house servants December 15 Gave shoes to the field Negroes My dear Papa, I received your last letter of August 24. You told me that you believed that someone helped me write my letters. But I assure you that I wrote them with my own hand. I am surprised myself that I have so quickly learned the shape of my letters. I could scarcely write one letter when I arrived at college. But, my dear father, I assure you on my honor that no one helped me … . I am studying only English and French. That’s because one has to know English well to study Latin and Spanish. Reading in French, Reading in English, French Grammar, English Grammar, English Translation, and French Translation

References Aycock, Joan Marie 1993 The Ursuline School in New Orleans, 1727–1771. In Cross crozier and crucible: a volume celebrating the bicentennial of a Catholic diocese in Louisiana. Glenn R. Conrad (ed.), 203–218. New Orleans: the Archdiocese of New Orleans and the Center of Louisiana Studies.

354

Connie Eble

Breton, Roland J. L. 1979 Géographie du français et de la francité en Louisiane.Quebec: International Center for Research on Bilingualism. Calan, Louise 1937 The Society of the Sacred Heart in America. London: Longmans, Green, and Co. Chambers, J. K. 1995 Sociolinguistic Theory. Oxford: Blackwell. Gore, Laura Locoul 2001 Memories of the old plantation home. Vacherie, LA: The Z Company, Inc. Gumperz, John J. 1982 Social Network and Language Shift. In Discourse Structures. John J. Gumperz (ed.), 38–58. Cambridge: Cambridge U Press. Lippi-Green, Rosina 1989 Social network integration and language change in progress in a rural alpine village. Language in Society 18: 213–234. Marmillion, Norman and Sand 2001 A creole family album. In Memories of the old plantation home. Laura Locoul Gore, 115–166. Mills, Gary B. 1977 The Forgotten People: Cane River’s Creoles of Color. Baton Rouge: Louisiana State University Press. Milroy, Lesley 1987 Language and social networks. 2nd ed. Oxford: Blackwell. 2000 Social network analysis and language change: introduction. European Journal of English studies 4: 217–223. 2002 Social Networks. In The Handbook of Language Variation and Change. J. K. Chambers, Peter Trudgill, and Natalie Schilling-Estes, (eds.), 549–572. Oxford: Blackwell. Picone, Michael 2003 Anglophone slaves in francophone Louisiana. American Speech 78: 404–433. Smith-Thibodeaux, John 1977 Les francophones de louisiane. Paris: Editions Entente. Trépanier, Cécyle 1988 French Louisiana at the threshold of the 21st century. PhD. dissertation, Department of Geography, Pennsylvania State University. Trout, Robert O. 1996 The origin of the pioneer population of north central Louisiana hill country. In A refuge for all ages: immigration in Louisiana, The Louisiana Purchase Bicentennial Series in Louisiana History,

English/French bilingualism in nineteenth century Lousiana

355

Vol. 10. Carl A. Brasseaux (ed.), 470–477. Lafayette: Center for Louisiana Studies, University of Southwestern Louisiana. Prudhomme Family Papers (#613) 1765–1997. Southern Historical Collection, Wilson Library, University of North Carolina at Chapel Hill. [collection description: www.lib.unc.edu/mss/inv/htm.00613.html] USGenWebCensusProject. 1850. Federal Census, Nelson County, Kentucky, District 1. www.rootsweb.com/~census (Accessed on September 10, 2005).

Taking permissible shortcuts? Limited evidence, heuristic reasoning and the modal auxiliaries in early Canadian English*

Stefan Dollinger 1.

Introduction

The present study focusses on the ‘other’ North American variety of English, Canadian English (CanE) from a diachronic perspective. While CanE is a relative newcomer to the field of linguistics, diachronic studies in realtime, as Brinton and Fee (2001: 426) point out, have been almost entirely missing. Much of what is known about historical CanE is in the field of lexicology, as a direct result of the work on the Dictionary of Canadianisms on Historical Principles (Avis et al. 1967), which has become seriously out of date.1 In the present study, an attempt is made to illustrate a kind of “bad” data problem that is of particular relevance to the study of Late Modern English (LModE) varieties in colonial contexts. The modal auxiliaries in Ontario English (OntE) in the period from 1776 to 1849 shall serve as a test case here and are surveyed in relation to British English (BrE), with some comparisons to American English (AmE). One of the problems diachronic linguists are confronted with in many colonial contexts, such as in early Ontario, is a large array of input varieties and only exemplary real-time evidence that is available in machine-readable data formats. In this paper, an approach is suggested that aims to circumvent these data limitations in colonial contexts for the period between 1700 and 1900. The focus of the present paper is therefore not on the variables and their developments as such (see Dollinger 2008 in this respect), but on what is best called a heuristic method to overcome, at least in part and as an approximation, a) the lack of readily-available data, b) low token discourse frequencies and c) genre-specific variation that makes the characterization of linguistic variables within national varieties a particular challenge.

358

Stefan Dollinger

After a case study on the modals CAN and MAY that illustrates the approach adopted here, the results of this kind of reasoning will be presented for eleven modals and semi-modals in OntE in relation to BrE and an attempt will be made to assess their behaviour in the light of later developments on a cline from conservative to progressive behaviour when compared to BrE. It will become clear that we will inevitably have to rely on some kind of heuristic reasoning in order to arrive at generalizations. A heuristic process can be described as A process, such as trial and error, for solving a problem for which no algorithm exists. A heuristic for a problem is a rule or method for approaching a solution. (Blackburn 2005: s.v. heuristic).

For our purposes, we can equate the term “algorithm” with fully-fledged, empirical, corpus-based study, which is firmly based in the philological tradition that has traditionally been centred on a standard variety. Heuristic reasoning seeks to make the best of limited data and aims to allow statements on the development of colonial varieties when not all data are available. Heuristic principles, while used pervasively in other disciplines, such as in engineering, computer science, social sciences (see Michalewicz and Fogel 2004, Tversky and Kahneman 1982), have not yet been explicitly applied to English historical linguistics (while its principles may have played a role) because, perhaps, they do seem to be incompatible with the standard of description that is usually found in English historical linguistics. The test case presented here, however, will make it clear that the time may be ripe for embracing “good enough” solutions and approximations more directly even in this discipline, and thus follow the lead of the “harder” sciences. It will be shown that heuristic means are a kind of reasoning that may provide fairly reliable shortcuts to answers to research questions that would otherwise be years, if not decades, away. Two kinds of “bad data” scenarios are usually discussed in historical linguistics. While both points are not new, I will refer to two more recent contributions to this discussion. First, the problem that historical documents “survive by chance, not by design”, often containing “a normative dialect” and not the vernacular (Labov 1994: 11) is something one has to deal with. This problem becomes less pervasive in the LModE period, when data produced by minimally-schooled writers become available, which has proven to be a rich resource. Second, Nevalainen (1999) addresses the problem of acquiring social information on the informants and points out that in Early

Taking permissible shortcuts?

359

Modern English (EModE), and to an increased extent in LModE, the chances to reconstruct the social milieu of a writer increase dramatically. In the colonial context, however, I would like to suggest a third “bad” data problem which is revealed in scenarios of new-dialect formation (Trudgill 2004) and the complex nature of input varieties in colonial Englishes. Put in a nutshell, the problem is the sheer variety of input varieties that came to form new colonial Englishes. To exemplify the problem let us review briefly the external language history of the variety for the sociohistorically most important variety of Canadian English, Ontario English. 1.1. Late Modern English, colonial English, Ontario English The Late Modern English period (LModE) has only fairly recently become the object of linguistic enquiry. Although Jones (1989: 279) referred to the eighteenth and nineteenth centuries as “the Cinderellas of English historical linguistic study” not long ago, the late 1990s saw a considerable increase of research activity.2 The LModE period coincides with an accelerated spread of colonial Englishes and entails a focus on what Clyne (1992) refers to as “non-dominant” varieties of English, i.e. varieties other than British and American English. Indeed, Trudgill (2002: 44) proclaims that for the period after 1700 there is “no real excuse” to confine one’s focus to the two dominant varieties of English in more general works of English historical linguistics. Little pre-1900 historical work has been done on the historically most important variety of CanE, namely OntE, which provided the model for Standard Canadian English as an urban middle class dialect (cf. Chambers 1998: 252). Thomas (1991) is a notable exception. Based on linguistic atlas data from the 1950s and employing the apparent-time design, he traces the development of Canadian Raising back to the late nineteenth century. Chambers’ (1981, 1993, 2004) work on language attitudes in Victorian Ontario adds another important layer to the available documentation, as does Hultin (1967). M. Bloomfield (1948) and Scargill (1957), the two classic contributions on the origins of CanE, are, however, based on settlement history alone and are devoid of linguistic data. Only recently, a realtime study of pre-Confederation OntE on the modal auxiliaries from the viewpoint of new-dialect formation (Trudgill 2004) has been completed (Dollinger 2006b, 2008).

360

Stefan Dollinger

1.2. External history The external history of Canada is divided into four distinct settlement waves (Chambers 1991, 1998). Because of the temporal focus on preConfederation (pre-1867) OntE in the present paper, settlement waves I and II are of importance here. Wave I is comprised of early immigration from the United States into Ontario, from roughly around 1776 to 1793, with a small trickle continuing until 1812, when war with the United States broke out. This early immigration resulted in the first peopling of Ontario and the steady transformation of the “primeval forest west of Montreal” that had previously been inhabited by only “a few hundred English-speaking” people (Orkin 1970: 52f), most of whom were transient fur traders. This first wave, however, was a rather heterogeneous mix of “American loyalists”. Besides native English speakers, most of them of early AmE, there were significant elements of disbanded German soldiers (including some Swiss German speakers), large contingents of Scottish Gaelic speakers and sizeable Dutch elements, which were joined by some Scots speakers, Irish English speakers and French speakers (Dollinger 2008: 64–78). As those minorities came early, they are expected to have had an influence on early CanE via language contact phenomena, which included L2 varieties that waned in the late 1800s. Wave II is comprised of post-1815 immigration. After the War of 1812 had come to an end (in 1814) and the Napoleonic wars had ended in Europe, the European population surplus was dumped on the colonies. This wave is characterized by massive immigration from the British Isles (although sizeable immigration from the German states is reported prior to 1850, Bausenhart 1989). The second wave included Scottish Gaelic speakers and Scottish English speakers, speakers of Northern English dialects, large numbers of Southern Irish and Ulster Scots (from Northern Ireland and also from the USA) and only a minority of Welsh, southern English dialect speakers and AmE speakers. In 1812 the Ontarian (then Upper Canadian) population was somewhere in the vicinity of 83,000 inhabitants (Gourlay 1822: 612), the majority of whom were Americans (cf. Akenson 1999: fn 117). This demographic make-up was changed by wave II. Figure 1 shows the relative population input to Ontario for the years 1829–1859, giving an impression of the numerical swamping of the American base with immigrants from the British Isles. For instance, immigration in 1829–31 alone outnumbered the entire population of the province in the

Taking permissible shortcuts?

361

year 1812, with almost 16,000 coming in 1829, 28,000 in 1830, and 50,000 in 1831, and so it continued, with ups and down, in that fashion: Arrivals at Port of Quebec from overseas

Percentage of all immigrants

80 70 60 50 40 30 20 10

England %

Ireland %

Scotland %

Europe %

1859

1857

1855

1853

1851

1849

1847

1845

1843

1841

1839

1837

1835

1833

1831

1829

0

Maritime

Figure 1. Arrivals at the Port of Quebec 1829–1859 (raw data from Cowan 1961: 289, table II; classification based on ports of departure)3

We have some evidence from recent socio-historical studies that the overwhelming majority of these second wave immigrants did, contrary to one’s intuition, not settle in the established centres founded by the American loyalists. Akenson (1999: 36) reasons that for Irish immigrants, who represented the bulk in the later years in figure 1, “one has to conclude that the overwhelming majority of Irish migrants to Ontario settled in the countryside”. Indeed, there was plenty of unsettled land in Ontario at the time, just waiting to be cleared (Wood 2000: 29–31). This finding has important implications from a sociolinguistic point of view, as it contradicts the otherwise logical reasoning that most of these immigrants would have settled “in towns and villages founded by the Loyalists” (Chambers 1998: 252). It also implies that AmE varieties – both native and L2 varieties – would have tended to be used in Ontario’s centres

362

Stefan Dollinger

and that dialects from the British Isles would have been evident in the countryside, before they eventually would have merged (except in the case of Ontario’s linguistic enclaves such as the Ottawa Valley or Peterborough). Linguistically, however, the second wave has usually been considered from a twentieth-century perspective as having had only a highly limited influence on OntE (Avis 1978: 4, Chambers 1998: 263). Real-time data show some patterns that do not necessarily corroborate the limited influence of BrE variants for all areas of modal usage (Dollinger 2008: 235–243, data on SHALL and WILL). 1.3. Data and variables The data come from three corpora: the Corpus of Early Ontario English, pre-Confederation section (CONTE-pC) (see appendix 3), provides the CanE material, A Representative Corpus of Historical English Registers, version 1 (ARCHER-1)4, the AmE and BrE data, and the Corpus of Late Eighteenth-Century Prose (CL18P)5 is used for data from NW England. Table 1 summarizes the varieties, corpora, genres and periods for the available data: Table 1. Overview of corpora, periods and genres (shaded areas: nonexistent corpus data)

Genres Period 1 Period 2 Period 3 Period 4+5

AmE BrE CanE6 (CONTE-pC) (ARCHER-1) (ARCHER-1) letters, diaries, newspapers 1776–1799 1750–1799 1750–1799 1800–1824 1800–1849 1825–1849 1850–1899 1850-1899 1850–1899

NW-BrE (CL18P) letters 1761–1790

While most findings are based on periods 1–3, period 4+5 was used as an additional benchmark in AmE in one instance (item no. 6 in Table 3). Period 2+3, which is referred to later on, comprises the years 1800–1849. Grey shadings in Table 1 indicate data that are not available. While these gaps are apparent, the bigger gaps are not shown in Table 1 as they concern the input varieties for which we do not even have corpora. When aiming to characterize regional varieties (or national varieties),7 we would wish for corpus data from all input varieties. As we know from the external history

Taking permissible shortcuts?

363

in Section 1.3, this is a formidable task. Given the immensity of data needed to compare colonial varieties with their input, there is of course the question whether we may be in the position to produce findings at all. In order to fully address questions such as the influence of input varieties on newly-formed dialects or the conservatism or progressivism of a variety in relation to another one, we would need a complete set of input data corpora, which would include a number of varieties in the Ontarian context. For the pre-1800 period, besides AmE, (standard) southern BrE and north-western regional BrE, we would need to have Irish English data, Scottish English, L2 varieties of Scots Gaelic speakers, German, Dutch, French speakers and First Nations speakers, among others. For the post1800 period, we would require data from Northern British English, southern BrE, both Southern as well as Northern Irish English (Ulster Scots), AmE, Scottish and Scots Gaelic L2 speakers, AAVE and First Nations in Canada and L2 varieties from immigrants (differences in the genres in different varieties, e.g. personal letters, further complicate the picture and are almost beyond the control of the researcher). One way towards problem solving in heuristics is to consider the available data carefully, which is easily illustrated by examples from logic (see Michalewicz and Fogel 2004: 9). In historical linguistics the data may be – within the limitations of “bad” data outlined above – “available” in libraries and archives, but not readily accessible. Often this catch is interpreted as a deficit of the researcher, and not as a result of the vastness of the task of data mining. While it is always a good idea to increase one’s baseline data, I would argue that we should seriously consider heuristic approaches in historical linguistics and make use of “good enough” solutions that have proven useful in other disciplines. Related to the data question is the problem of quantitative vs. qualitative studies. Concerning the former, the cut-off point of what constitutes eligible frequencies (e.g. at least 10 tokens) is, of course, inspired by the available corpus sizes. The modal auxiliaries have usually tended to yield, on the whole, acceptable frequencies in the standard corpora. But what are we to do if conventional corpus sizes do not produce the expected token counts? I would argue, again thinking heuristically, that there may be an alternative to going straight back to the archives in search of more data. This line of reasoning, making the proverbial best of “bad” data, will be illustrated in the quantitative treatment – and statistical testing – of limited data that, combined with Present Day English (PDE) findings, may prove

364

Stefan Dollinger

just enough to characterize the LModE development of the modals in the Ontarian context. Even in the LModE period the modal auxiliary complex undergoes considerable changes in usage and most of those changes tend to be stylistic rather than categorical in nature (Denison 1998: 165, 1993). In Section 2, CAN and MAY serve as examples for the kind of heuristic reasoning used for an assessment of whether CanE showed progressive or conservative behaviour in the modal auxiliaries in relation to BrE. In Section 3, the overall picture for the modals in early OntE will be presented.

2.

A close-up: CAN and MAY in early OntE

The early development of CAN and MAY is well documented. In OE mæg, the formal ancestor of MAY, is prevalent in the sense ‘to be able to’. Following a pragmatic/semantic cline of grammaticalization, this ability use was extended to denote possibility and was subsequently developed into permissive uses. Permissive uses can be occasionally found in OE (Traugott 1972: 71f), but the “full performative use of may, as in You may go ‘I permit you to go’, did not gain wide currency until the sixteenth century” (Traugott 1972: 118). OE cann, on the other hand, originally meant ‘to know, be acquainted with, know how to’. Once CAN acquired the meaning of ‘to be able to’, MAY gave up the meaning of ‘to be able, be strong’ (Kytö 1991: 65). After this semantic change, it was only a question of time before the meanings of ‘possibility’ and finally ‘permission’ developed by implication, i.e. pragmatic factors in the context favouring these interpretations rather than ‘ability’ readings (Kytö 1991: 65). CAN and MAY, revolving around the notions of ability, permission and possibility, have been studied at the synchronic level and for older language stages up to EModE (e.g. Coates 1983; Facchinetti 2003, 2002, 2000; Kytö 1991: 81–258, Warner 1993: 176–178, Denison 1993: 292– 325), while in LModE general developments have been identified in the area of permission (e.g. Traugott 1972: 170f; Kytö 1991: 65; Denison 1993: 303). In comparison to the long-term development from OE to EModE, the study in LModE has not been pursued as rigorously. Simon-Vandenbergen (1984) seems to be the only quantitative study of CAN and MAY that includes semantic categories as well as a sizeable proportion of LModE texts (based on personal and official letters and drama data). She shows (1984: 364) that root uses of CAN, in the sense of permission, first occur in nine-

Taking permissible shortcuts?

365

teenth-century plays and encroach on the territory of root MAY. A number of examples are found in the three corpora that illustrate this shift of functions of CAN, as shown in example 1(a–c): (1) a. ability (usually with animate subject): I fancy Kitty can do nothing better … (ARCHER-1, BrE, 1750–99, letter section) b. neutral, root possibility (ambiguous case): I should like to by it if I can by it two Advantage (CONTE-pC, 1825–49 [a minimallyschooled letter writer]) c. permission: We have decided upon my sitting with Mamma every night during tea and Minnie during dinner as then I can read Mamma's prayers to her. (CONTE-pC, 1825–1849; a teenaged diary writer) [the girl is allowed to read to her sick mother then] As pointed out by Coates (1983: 139 for 1960s BrE), 1 (b, c) illustrate the core functions of MAY, which means that CAN was beginning to compete with MAY in the latter’s core domain. This long-term developmental scenario provides the background for the identification of progressive and conservative forms – CAN as progressive, and MAY as conservative – with CAN conveying more informal undertones that MAY (Coates 1983: 103). Table 2 shows the areas of competition for CAN and MAY in the late eighteenth and nineteenth centuries and gives examples of their uses. The grey shaded cells highlight the semantic areas in which CAN and MAY compete. The example in MAY/permission can be paraphrased in PDE by CAN, and it is possible to substitute CAN/permission with MAY without a significant change in meaning. MAY/root possibility can again be rephrased with CAN, while CAN/root possibility with MAY is more of a borderline case, but acceptable in some varieties. We are attempting to trace changes in these two areas, which expanded in LModE, in CanE in relation to BrE. As MAY had become obsolete as a marker of ability in ME and CAN was not used for epistemic possibility, the competition is limited to the areas of root possibility and permission. From a twentieth-century perspective, we know that CAN was to gain the upper hand in both permission and, to some extent, in root possibility meanings, as a result of influences from informal genres (see Coates 1983: 106f) and is now showing first uses in epistemic readings (Coates 1995 cites first examples of 1990s

366

Stefan Dollinger

AmE for epistemic CAN, Facchinetti 2000 finds first examples in early 1990s BrE). Table 2. CAN and MAY – semantic categories and examples Prototype FORM MAY

CAN

NOTIONAL FUNCTION Permission

EXAMPLE

If he show any Disposition to write me a penitential Letter, you may encourage it; not that I think it of any Consequence to me, but because it will ease his Mind and set him at rest. (BrE-1) Root possibility Any person wishing to purchase may depend upon getting a great bargain (CanE-2) Epistemic … and my ideas upon the several points which possibility may, between this and then, occur to me (AmE-1) Ability Permission

I fancy Kitty can do nothing better … (BrE-1) I have your Certificate that the land is not leased or vacant of course none [of the settlers] can be located without the sanction of the Lt. Gov. as in other cases. (CanE-2) Root possibility Nothing can be more satisfactory than the readiness and unanimity with which the Legislature have applied to meet the emergencies (CanE-3)

Figures 2 and 3 show the results for CanE and BrE in periods 1 and 2+3 and for AmE in period 1. All instances of CAN are shown in relation to MAY; the functions of ‘permission’ and ‘root possibility’ are shown by genre, as the choice of CAN and MAY seems to have been influenced by text type (See note8 for abbreviations used), but we will need to summarize the usage across different genres later on. First, we will focus on root possibility uses of CAN, as shown in Figure 2. By comparing the CanE with the BrE data, we can see a picture that is largely parallel in incidence in the diary and letter genres. Only in the newspaper genre the pattern diverges. The AmE data are closer to the CanE values across all genres, which indicates the loyalist base of CanE. Viewed in the larger diachronic context, CanE seems more progressive in its use of CAN in diaries (i.e. closer to 1960s BrE usage, where CAN is predominantly used in possibility readings, Coates 1983: 101), but more conservative in letters and newspapers than BrE. The apparent tendency for an increase of CAN in diaries and letters is, however, not confirmed by statisti-

Taking permissible shortcuts?

367

cal testing, which yields no significance for an increase in CAN for the changes in Figure 2 at the 95% level (Appendix 1). From our background knowledge, these data seem to suggest, by heuristic reasoning, a drift scenario in diaries and letters, even though statistical testing does not help us here. The different developmental patterns in CanE and BrE newspapers could either be a result of chance, or alternatively, as it shows the statistically strongest change (Appendix 1, n–1 and n–2+3), a stylistic change towards more BrE norms in Canadian newspapers. CanE uses MAY more often than CAN in letters, but not in diaries, which indicates a more formal style in CanE letters than in British ones. It is probably best to interpret these figures as reflecting instances of directional drift and stylistic variation in CanE, with more formal, possibly more conservative tendencies in the letter and newspaper genres. We can say that ‘root possibility’ in newspapers tended to be expressed by MAY in CanE, as opposed to BrE, but with both varieties converging in the 30% range in period 2+3 (Figure 2). Overall, this change at the time can be perceived as a change from above the level of consciousness, as the use of CAN is found be to documented – and castigated – by eighteenth-century grammarians (Sundby et al. 1991: 211). For the second function in which CAN and MAY overlap, permission uses, a change has been reported for LModE. Figure 3 shows the development for CAN denoting permission and we see an increase of CAN in both CanE and BrE in all genres. While differences in percentages appear in part to be considerable, again, no change of CAN is statistically significant (cf. appendix 2), which is partly a result of low token incidences. In 1960s BrE, informal texts tend to show more uses of permission CAN (Coates 1983: 101), but are generally used less frequently than possibility readings of CAN. Using a different classification system, Facchinetti (2002: 239) shows for early 1990s BrE that 5% of all uses of CAN are deontic readings, which are one core of our permission uses. What should we make of these results? It is clear that CAN is expanding its use in both CanE and BrE in the areas of permission and, for diaries and letters, root possibility uses. But neither the differences between CanE and BrE in each period, nor their increases from period 1 to 2+3 are significant, and statistical testing (neither chi-square nor Fisher’s Exact) does not help us much to confirm a long-term change corroborated in previous studies. The change appears to be too slow to reach the 95% significance level in 25-50 year periods in the data. We know by hindsight, however, that in twentieth-century BrE, CAN is usually the default variant and not MAY,

368

Stefan Dollinger

which “is marked for formality” (Coates 1983: 103 for BrE, Ehrman 1966: 12 notes the use of permission CAN in 1961 AmE dialogue data). With this knowledge, we would interpret the diachronic changes in Figures 2 and 3 as a LModE parallel development in both CanE and BrE leading up to the PDE distribution, despite their lack of statistical significance in our data. This ‘heuristic’ conclusion could easily be tested in a quantitative framework. As Appendices 1 and 2 show, the token frequencies are low. In BrE diaries (Appendix 1, first table), which shows a solid relative increase of root possibility CAN from period 1 to 2+3, the chi-square value is 1.04, and therefore short of the 3.84 required to reach significance at the 95% confidence interval. If we quadrupled the token frequencies, we would reach a value of 4.18, and this gives us an idea of how much data are needed. We would need at least four times as much as we have now, or c. 500,000 words for the late eighteenth and early nineteenth century to be able to base our study on firm statistical grounds. And here, even CONCE (Kytö, Rudanko, Smitterberg 2000: 89), the biggest corpus of nineteenthcentury BrE to date, would provide too little data for our three genres to meet the baseline data criterion (although the c. 250,000 words of letter data are a substantial body of evidence). Clearly, the data mining needed to test our heuristic reasoning on strict quantitative terms would be a project on its own. The examples show that changes which appear to be instances of slow moving drift or parallel development do not necessarily reach levels of significance in the data. We have also said, however, that based on our knowledge of the further development, it is justifiable to interpret the data as exhibiting a parallel development, since five out of six instances of a change point in the same direction (all except for the CanE and BrE newspaper data). This approach is best described as heuristic reasoning. However, we do not have AmE data from the early nineteenth century to confirm our reasoning of drift in the last instance, and we are at a loss for other input varieties of OntE (such as Scottish English, Ulster English or Irish English, or post-1800 Northern English). We do know, however, that in twentieth-century BrE, CAN has become the default form for root possibility meanings (cf. Coates 1983: 101) and more and more in permission uses, despite its condemnation by prescriptive grammarians. The method applied here is hardly new as it has been implicitly applied in many diachronic studies. What is new, though, is the claim that we cannot, and should not, expect studies to include data that are, strict sensu,

Taking permissible shortcuts?

369

'Root possibility' CAN (vs. MAY) in three varieties 100 80.0 80

60.0

percent

29.4 60

70.4

55.6

60.9 52.9

50.050.0

51.6 42.9

39.5 33.8

40 20.0

CanE BrE AmE

22.7

20

0 d1

d2+3

l1

l2+3

n1

n2+3

genres & periods

Figure 2. Diachronic development of CAN and MAY coding for ‘root possibility’ in CanE and BrE (AmE-1 for comparison) 'Permission' CAN (vs. MAY) in three varieties 100.0 100.0

100.0 100 80.0

75.0

80

percent

60.0 60

CanE

50.0

50.050.0

BrE 40

33.3

AmE

15.014.3

20 0.0

0.0

0.0

0 d1

d2+3

l1

l2+3

n1

n2+3

Figure 3. Diachronic development of CAN and MAY coding for ‘permission’ in CanE and BrE (AmE-1 for comparison)

370

Stefan Dollinger

needed to make any generalization about colonial varieties, but should embrace the educated guessing applied in heuristics. The method is an approximate comparison of the situation in early OntE with BrE (and AmE varieties) and it needs to be stressed that the method cannot rule out reverse developments. However, it would be somewhat unlikely for CAN to move backwards given clines of grammaticalization (e.g. Traugott 1989). As the method makes use of the data available and applies other sources and benchmarks, its heuristics provides us with insights into otherwise uncharted waters. Section 3 shows the results of this heuristic reasoning for eleven modals in a total of 19 contexts in early OntE.

3.

The bigger picture: eleven modals in early OntE

In this section, the nine core modals, CAN/MAY, COULD/MIGHT, SHALL/WILL, SHOULD/WOULD, MUST, plus OUGHT TO, and semimodal HAVE TO, will be presented in terms of progressive or conservative behaviour in CanE in relation to BrE. For this purpose, the results for each period (the diachronic development as well as the empirical base) have been considered, gauged and assigned to one category in the manner illustrated in the previous section. The assessment is carried out on a 5-tier grid, classifying each function into “conservative”, “neutral” and “progressive”, and “towards conservative” or “towards progressive” for intermediate cases. Clearly, this kind of attempt to classify the overall behaviour across the three genres also requires some form of heuristics, i.e. the consideration of the available data and a principle to synthesize the variables into one measure. As the empirical base for permission uses of CAN (cf. Appendix 2) is especially slim in CanE diaries and newspapers (CanE-d, CanE-n), preference is given to the letter data, which are slightly more conservative than BrE data. The overall assessment for permission uses of CAN in CanE is perhaps best described as “towards conservative” (item no. 1 in Table 3). Root possibility CAN, with diaries more progressive, letters more conservative and newspapers conservative (but approaching BrE values), is also best characterized as “towards conservative” (item no. 2). The remaining 17 contexts were assessed in a similar manner, resulting in Table 3. Table 3 shows that pre-Confederation OntE does not lean heavily in any direction. If anything, it appears to be slightly more progressive in its overall use of the modals than BrE. Interestingly, in six out of nine cases it patterns with AmE data from period 1, which linguistically confirms the loyalist base

Taking permissible shortcuts?

371

hypothesis for AmE input. The data from NW England (NW-BrE) in period 1, however, are rather distant from both CanE and AmE where available (items no. 6 and 15). Item no. 6, use of SHALL and WILL in the first person, is subject to massive change from period 1 to period 2+3, which is the reason for its occupying two slots (see note 9). Without more BrE data we cannot decide on the two categories, but it seems clear that the change was the result of regional BrE influence (Dollinger 2008). The overall pattern shown in Table 3 suggests that early CanE did not simply follow AmE or BrE usage, but was beginning to show idiosyncratic developments in the modal auxiliary complex and most likely elsewhere. For the area of vocabulary, the unique Canadian character has been long accepted (Lovell 1955: 5) and researched (Avis et al. 1967; de Wolf et al. 1997; Barber 2004). Table 3 provides some indicators for developments specific to CanE in the use of modal auxiliaries. For item no. 18, COULD in epistemic uses, for instance, the loyalist base theory of American input cannot account for this behaviour, as AmE and CanE are placed at opposite ends of the spectrum. In others areas, such as the use of first person WILL (item no.6), CanE may well have been more progressive than AmE, although this remains a hypothesis until post-1800 AmE data can be found. An overall assessment such as that represented in Table 3 allows new insights into the behaviour and genesis of colonial varieties and would warrant the element of “educated guesswork” inherent in heuristic approaches, of which the summarization in Table 3 is the result.

4.

Possible conclusions: evidence and reasoning

The overall results in Table 3 show that CanE modal use appears to have been slightly on the progressive side when compared with BrE. While largely matching AmE use, early OntE shows some developments that may have been specific to the variety. As LModE varieties are a very recent area of inquiry, and LModE dialectal variation largely remains to be studied (but see for instance Watts and Trudgill 2002), the present situation may be comparable to the state of knowledge of EModE variation in the 1980s (Görlach 1988). If one wishes to make generalizations along the lines suggested in Table 3 between a colonial variety and BrE – with some reasoning for AmE and regional BrE – some approximations and heuristic ways of reasoning are necessary

372

Stefan Dollinger

Table 3. Eleven modals in CanE in comparison to BrE (data for AmE-1 and NWBrE-1 is indicated in relation where applicable) (adapted from Dollinger 2008: 276) VARIABLES

CAN & MAY

OUGHT TO

SHALL & WILL

function / context 1)

permission

2)

root poss.

3) 4)

affirmative contexts negative contexts

5)

overall

6)

1st person

Conservative

Towards cons.

Neutral

Towards progr.

Progressive

(AmE-1) (AmE-1)

9

(NW-BrE)

(AmE-1)

nd

7)

2 person

8)

3rd person

9)

inanimate subjects

10) passive structures SHOULD & WOULD

11) hypotheticals 12) non-hypotheticals 13) root uses

MUST & HAVE TO

COULD & MIGHT

14) epistemic uses 15) ep. MUST NECESS. 16) affirmative contexts 17) negative contexts 18) epistemic uses 19) non-epistemic uses

(AmE-1) (AmE-1) (AmE-1)

(NW-BrE) AmE-1) (AmE-1)

(AmE-1)

Taking permissible shortcuts?

373

in the light of gaps in the historical corpus inventory. Given the usual production phase of a historical corpus of about three to four years and the multitude of input varieties for LModE colonial varieties, we are still some time off from more complete data sets of input varieties which were, as is well known, mainly regional dialects (Hickey 2004a: 1). We have seen in Section 1.3 that in the Ontarian context, at least ten groups in wave I and nine groups in wave II prior to 1850 would need to be considered. Of the required 19 corpora, we have just one regional corpus for period 1 (NWBrE), one AmE for period 1 and two BrE text collections (ARCHER-1 and CONCE), the latter two of which include texts that are closer to standard varieties and may not be the type of material one would ideally wish for.10 In some respects, we are 16 corpora short of being able to reach hard and fast conclusions. Moreover, given current copyright practices, we may be a couple of generations of researchers away from the necessary research tools. Circumventing this lack of resources, the approach laid out in Sections 2 and 3 outlines a method that synthesizes the OntE data with existing realtime resources in combination with studies of the variables in later periods. While this approach is somewhat compromised by the possibility of retrograde developments in the nineteenth and twentieth centuries, it offers a more precise scenario of language use in colonial varieties and their relations to dominant varieties of English.11 All that is needed is one reliable corpus of the variety to be surveyed, such as CONTE-pC, and access to the available corpora of input varieties. Findings gleaned from studies such as the present one are, given the lack of resources, likely to hold for many years to come. While the current approach employs both percentage changes in the distributional frequency of features and statistical testing, it does not necessarily rely on levels of significance in the interpretation of its findings. Given the short period intervals of 25 years, which are necessary to tap into the process of new-dialect formation (Trudgill 2004), the customary sample sizes in historical linguistics do not easily produce statistical significance at the 95% level even for relative high-frequency items like the modals. However, the percentage distributions appear to show a clear long-term trend from a present-day perspective. The biggest challenge for LModE variationist linguistics is the lack of specialized corpora for regional English dialects, and this is compounded by the lack of sufficiently large corpora for historical linguistic analysis using corpus linguistics techniques. We have seen that – in theory – at least

374

Stefan Dollinger

four times the customary sample sizes would be the base line needed to reach significance levels for the modal auxiliaries in semantic studies. However, as I have tried to show in the discussion of CAN and MAY, one can arrive at meaningful interpretations of even limited data sets by exploiting the relative temporal proximity of LModE to twentieth-century English, using Coates’, Ehrman’s and Facchinetti’s studies. As a result, given the limited resources for corpus compilation, the production of smaller, specialized corpora of more diverse LModE varieties, rather than the production of a bigger-sized corpus a single variety, should remain a priority. Logically, the “three main source regions of extraterritorial varieties of English” in the British Isles (Hickey 2004b: 33), English English (with a differentiation of north and south [Trudgill 2004]), Scottish English and Irish English (both Ulster Scots and Southern Irish English), from around 1700 to 1900, would be the first choices for new research tools for the spread of English in the LModE period. In the Ontarian context, 19th-century AmE data is another prime desideratum. While we would wish to have access to complete dialect lineages of all input dialects involved in the formation of a given variety (as called for by Wolfram and Schilling-Estes 2004: 182), corpora of LModE Scottish and Irish English would tremendously facilitate the line of heuristic reasoning proposed in Sections 2 and 3. Ultimately, only complete data sets can prevent us from drawing tentative conclusions about influences, independent developments and questions of colonial lag that may not quite stand the test of time. However, by taking a more explicit, exploratory heuristic point of view, we do not necessarily have to call it quits until these resources materialize.

Appendices: Statistical testing (95% level) For the following contingency tables, both the Chi-Square Test scores (the standard in much of corpus linguistics when statistical testing is applied) and Fisher’s Exact Test scores are provided. Please note that Fisher’s Exact is to be preferred in all cases where one of the cells contains less than five instances (Vogt 2005: 122; Oaks 1998: 25, cf. Woods et al. 1986: 144,) (which is the case in all but five tables). In all tables shown here, both tests arrive at the same conclusion about statistical significance, which supports Woods et al.’s assessment – given the “bad” data situation in historical linguistics – to “go ahead and carry out the chi-squared test even if some

Taking permissible shortcuts?

375

expected frequencies are rather too small” (1986: 145). Provided one is aware that the chi-square values tend to be rather larger than they ought to be and one would consider this in one’s interpretations, this seems to be acceptable practice. Chi-square calculations (Ȥ2) follow the test statistic used in Nelson, Wallis and Aarts (2002: 264–7) and are calculated to answer the question whether the choice of CAN is affected by the independent variable (periods 1 and 2+3, BrE and CanE respectively) and not whether the entire grammatical choice (table) is affected (see Nelson, Wallis and Aarts 2002 for an account). The latter method, usually offered by online tools (e.g. Georgetown Linguistics Web Chi Square Calculator), reaches significance more easily but says nothing about which of the two variables (CAN or MAY) is significant. Fisher’s Exact p-values are calculated with Preacher and Briggs’ (2001) online tool and show the two-tailed p-values (p), which produces values in between two more extreme values (one of which might be chosen according to the expected distribution of the variables, which was not applied in the following calculations). Appendix 1: Root possibility uses of CAN BrE-d 1 2+3

CAN

MAY

5 5

12 4

CanE-l 1 2+3

CAN

MAY

18 39

16 25

CanE-n 1 2+3

CAN

MAY

2 4

2 1

Ȥ2 = 1.04 < 3.84, not significant p = 0.23 > 0.05, not significant

Ȥ2 = 0.24 < 3.84, not significant p = 0.52 > 0.05, not significant

Ȥ2 = 0.3 < 3.84, not significant p = 0.52 > 0.05, not significant

CanE-d 1

BrE-n 1

n-1 CanE

CAN 2

MAY 8

2+3 25 49 Ȥ2 = 0.52 < 3.84, not significant p = 0.49 > 0.05, not significant

CAN 16

MAY 15

2+3 15 23 Ȥ2 = 0.56 < 3.84, not significant p = 0.34 > 0.05, not significant

CAN 2

MAY 8

BrE 16 15 Ȥ2 = 1.72 < 3.84, not significant p = 0.14 > 0.05, not significant

376 BrE-l 1

Stefan Dollinger CAN 24

MAY 16

n-2+3 CanE

2+3 19 8 Ȥ2 = 0.27 < 3.84, not significant p = 0.44 > 0.05, not significant

CAN 25

MAY 49

BrE 15 23 Ȥ2 = 0.2 < 3.84, not significant p = 0.68 > 0.05, not significant

Appendix 2: Permission uses of CAN CanE-d CAN MAY 1 1 1 2+3 1 0 2 Ȥ = 0.25 < 3.84, not significant p = 1.0 > 0.05, not significant

CanE-n CAN MAY 1 0 0 2+3 5 0 2 Ȥ cannot be calculated p = 1.0 > 0.05, not significant

BrE-l CAN MAY 1 1 6 2+3 3 2 2 Ȥ = 1.83 < 3.84, not significant p = 0.22 > 0.05, not significant

CanE-l CAN MAY 1 2 11 2+3 8 16 Ȥ2 = 1.01 < 3.84, not significant p = 0.44 > 0.05, not significant

BrE-d CAN MAY 1 2 2 2+3 4 2 Ȥ2 = 0.3 < 3.84, not significant p = 1.0 > 0.05, not significant

BrE-n CAN MAY 1 4 4 2+3 6 2 Ȥ2 = 0.4 < 3.84, not significant p = 0.61 > 0.05, not significant

l-1 CAN MAY CanE 2 11 BrE 1 6 Ȥ2 = 0.37 < 3.84, not significant p = 1.0 > 0.05, not significant

l-2+3 CAN MAY CanE 8 16 BrE 3 2 Ȥ2 = 0.78 < 3.84, not significant p = 0.34 > 0.05, not significant

Taking permissible shortcuts?

377

Appendix 3 Textual sources of the Corpus of Early Ontario English: periods, texts and sample sizes (number of words). CONTE-pC size: 125,000 words (CONTE size: ca. 225,000 words). period

newspapers

diaries

1 1776– Upper Canada Gazette, 1799 ca. 2,800 Canada Constellation, ca. 1,700

Benjamin Smith, ca. 1,800 Anne Powell, ca. 6,200

sum: 4,500 words 2 1800– Upper Canada Guard1824 ian, ca. 8,200 Upper Canada Gazette, ca. 5,000 Kingston Gazette, ca. 1,400 sum: 14,600 words 3 1825– Upper Canada Gazette, 1849 ca. 8,500 Niagara Argus, ca. 4,500 Gore Gazette, ca. 3,000

sum: 8,000 words Benjamin Smith, ca. 8,500 Ely Playter, ca. 8,500 (Eleanora Hallen, ca.3,700 not incl. in pC version)12 sum: 17,000 words Sophia MacNab, ca. 11,400 Charlotte Harris, ca. 9,200

sum: 16,000 words TOT genre total: 35,100 125,700

sum: 20,600 words genre total : 45,600

(semi-)official letters

61 letters (by 48 authors)

sum: 15,000 words 65 letters (by 48 authors)

sum: 15,000 words 77 letters (by 64 authors)

sum: 15,000 words genre total : 45,000

Period

SIN-speakers

Lower Class

Middle Class

1 2 3

2,200 1,700 1,000

3,200 1,000 3,100

11,700 14,000 12,000

Social (and regional stratification) within CONTE-pC: sample sizes for Scottish-Irish-Northern English speakers (SIN) and social class subsections.

378

Stefan Dollinger

Notes *

1. 2.

3.

4. 5.

6. 7.

The research for this paper was funded by the Austrian Academy of the Humanities and Sciences, Österreichische Akademie der Wissenschaften, DOC grant 21701. I would like to thank three anonymous reviewers for their feedbacks on an earlier version of this paper. Recently, a revision process has begun. See www.dchp.ca (21 Dec. 2006) for more details. The best indicators are probably the founding of the Late Modern English Conference series (LMEC) in 2001 and the appearance of special collections within bigger English historical linguistics conferences (Bueno Alonso et al. eds. 2007, Dalton-Puffer et al. eds. 2006), the appearance of textbooks and reference guides (Bailey 1996, Romaine 1998, Görlach 1999, 2001, Beal 2004) and major projects such as Tieken-Boon van Ostade’s ”The codifiers and the English language“ project, see http://www.ulcl.leidenuniv.nl/index.php3?m=9&c=122 (31 Jan. 2006). Inner-British migration is a factor to be considered here. While a departure port in England or Scotland does not necessarily mean that the immigrant was English or Scottish (Liverpool as a gateway for the Irish is a prime example, or Glasgow and Greenock in Scotland), “relatively few Britishers [English and Scottish] sailed from Irish ports” (Akenson 1999: 14). Cowan (1961: 287) stresses the fact the that Irish were using British ports, especially once steamboat transportation from Ireland to Britain had become available (probably around 1825) reducing the costs. This would cause the Irish element to be underrepresented in figure (1), further increasing their share of the total (later) immigration. I am indebted to Christian Mair at Freiburg University for granting me access to ARCHER-1 (compiled by Douglas Biber et al.) Compiled by David Denison, Linda van Bergen and Joana Soliva Proud. My thanks go to David Denison for granting me online access to the corpus (cf. http://lings.ln.man.ac.uk/info/staff/dd/papers/newcastle_late_18c.pdf, 24 Jan. 2006). OntE is referred to as CanE in the context of BrE and AmE, unless stated otherwise. In the past two decades, genre-specific analyses have made big strides (Biber 1988, Kytö and Rissanen 1983), and while it is self-evident that variables differ between genres, regional provenance will be foregrounded as the prime independent variable. We will therefore need to devise a means how to assess the behaviour of a linguistics variable across the genres used. In this respect, we aim to complement detailed genre analysis with statements of the overall behaviour of a variable in a given variety in relation to other varieties.

Taking permissible shortcuts?

379

8. Both the diagrams and the appendices use the following abbreviations: d = diaries, l = letters, n = newspapers, periods 1, 2+3 as defined in table (1). CanE-l would therefore mean “Canadian English letters” as used in the appendices. 9. 1st person WILL poses challenges for the 5-tier grid, without referring to the diachronic development, and shows the limitations of this, admittedly rough, assessment. 10. Note that Southern AmE corpora, SPOC and BLUR (see Schneider 2007: 355, 358) would not constitute input varieties of OntE. 11. However, given the steady, long-term development of the modals along clines (Traugott 1989, Abraham 2002), it is somewhat unlikely that reversals would have occurred. 12. CONTE-pC is the first part of the CONTE corpus (1776–1899, see Dollinger 2006a) and will be made accessible to reseachers as soon as copyright has been fully cleared and the manual has been completed (at the time being, please contact the author for more information). Cf. Dollinger (2006a: 25) for the provisional CONTE design (full version).

References Abraham, Werner 2002 Modal verbs: epistemics in German and English. In Modality and its interaction with the verbal system (Linguistik Aktuell, 47). Sjef Barbiers, Frits Beukema and Wim van der Wurff (eds.), 19–50. Amsterdam: Benjamins. Akenson, Donald H. 1999 The Irish in Ontario. A study in rural history. 2nd ed. Montreal: McGill-Queen’s University Press. Avis, Walter S. 1978 Problems in the study of Canadian English. In Walter S. Avis: essays and articles. Selected from a quarter century of scholarship at the Royal Military College of Canada, Kingston. Thomas Vincent, George Parker and Stephen Bonnycastle (eds.), 3–12. Kingston: Royal Military College of Canada. Avis, Walter S., Charles Crate, Patrick Drysdale, Douglas Leechman, Matthew H. Scharill, and Charles L. Lovell (eds.) 1967 A dictionary of Canadianisms on historical principles. Toronto: Gage. Barber, Katherine (ed.)2 2004 Canadian Oxford dictionary. 2nd ed. Toronto: Oxford University [11998] Press.

380

Stefan Dollinger

Bausenhart, Werner 1989 German immigration and assimilation in Ontario, 1793–1918. New York: Legas. Beal, Joan C. 2004 English in modern times 1700–1945. London: Arnold Hodder. Biber, Douglas 1988 Variation across speech and writing. Cambridge: Cambridge University Press. 2004 Modal use across registers and time. In Studies in the history of the English language II. Unfolding conversations. Anne Curzan and Kimberly Emmonds (eds.), 189–216. (Topics in English Linguistics, 45) Berlin: Mouton de Gruyter. Biber, Douglas, Susan Conrad and Randi Reppen 1998 Corpus linguistics. Investigating language structure and use. Cambridge: Cambridge University Press. Blackburn, Simon 2005 The Oxford dictionary of philosophy. 2nd ed. Online version. Oxford: Oxford University Press. Bloomfield, Morton W. 1948 Canadian English and its relation to eighteenth century American speech. Journal of English and Germanic Philology 47: 59–66 [reprinted in Chambers (1975), 3–11]. Brinton, Laurel J. and Margery Fee 2001 Canadian English. In The Cambridge History of the English Language. Vol. VI. English in North America. John Algeo (ed.), 422– 440. Cambridge: Cambridge University Press. Bueno Alonso, Jorge L., Dolores González Álvarez, Javier Pérez Guerra, and Esperanza Rama Martínez (eds.) 2007 ‘Of varying language and opposing creed’: New Insights into Late Modern English. Bern: Peter Lang (= Linguistic Insights). Chambers, J. K. (ed.) 1975 Canadian English. Origins and structures. Toronto: Methuen. Chambers, J. K. 1981 ‘Lawless and vulgar innovations’: Victorian views of Canadian English. Toronto Working Papers in Linguistics 2: 13–44. 1993 “Lawless and vulgar innovations”: Victorian views on Canadian English [rev. version] In Focus on Canada. Sandra Clarke (ed.), 1–26. (Varieties of English Around the World, G11) Amsterdam: Benjamins. 1998 English: Canadian varieties. In Language in Canada. John Edwards (ed.), 252–272. Cambridge: Cambridge University Press.

Taking permissible shortcuts? 2004

381

‘Canadian Dainty’: the rise and decline of Briticisms in Canada. In Legacies of colonial English. Studies in transported dialects. Raymond Hickey (ed.), 224-241. (Studies in English Language) Cambridge: Cambridge University Press. Clyne, Michael (ed.) 1992 Pluricentric languages. Differing norms in different nations. Berlin: Mouton de Gruyter. Coates, Jennifer 1983 The semantics of the modal auxiliaries. London: Croom Helm. 1995 The expression of root and epistemic possibility in English. In Modality in grammar and discourse (Typological Studies in Language 32). J. Bybee and S. Fleischman (eds.), 55–66. Amsterdam: Benjamins. Cowan, Helen 1961 British immigration to British North America. The first hundred years. Rev. and enl. ed. Toronto: University of Toronto Press. Dalton-Puffer, Christiane, Dieter Kastovsky, Nikolaus Ritt, and Herbert Schendl (eds.) 2006 Syntax, style and grammatical norms: English from 1500–2000 (Linguistic Insights, 39). Bern: Peter Lang. Denison, David 1993 English historical syntax: verbal constructions. (Longman Linguistics Library) London: Longman. 1998 Syntax. In The Cambridge history of the English language. Vol. IV: 1776–1997. Suzanne Romaine (ed.), 92–329. Cambridge: Cambridge University Press. Dollinger, Stefan 2006a Oh Canada! Towards the Corpus of Early Ontario English. In The changing face of corpus linguistics (Language and Computers, 55). Antoinette Renouf and Andrew Kehoe (eds.), 7–25. Amsterdam: Rodopi. 2006b The modal auxiliaries HAVE TO and MUST in the Corpus of Early Ontario English: gradience and colonial lag theory in Early Canadian English. Canadian Journal of Linguistics 51/2&3: 189–210. 2007 The importance of demography for the study of historical Canadian English: three examples from the Corpus of Early Ontario English. In J. L. Bueno Alonso et al. (eds.), 105–136. 2008 New-dialect formation in Canada: evidence from the modal auxiliaries (Studies in Language Companion Series 97). Amsterdam: Benjamins.

382

Stefan Dollinger

Ehrman, Madeline E. 1966 The meanings of the modals in present-day American English. The Hague: Mouton. Facchinetti, Roberta 2000 Can and could in contemporary British English: a study of the ICEGB corpus. In New frontiers of corpus research (Language and Computers. Studies in Practical Linguistics, 36). Pam Peters, Peter Collins and Adam Smith (eds.), 229–246. Amsterdam: Rodopi. 2002 Can. In Variation in central modals. A repertoire of forms and types of usage in Middle English and Early Modern English. Maurizio Gotti, Marina Dossena, Richard Dury, Roberta Facchinetti and Maria Lima (eds.), 45–65. Bern: Lang. 2003 Pragmatic and sociological constraints on the functions of may in contemporary British English. In Modality in contemporary English (Topics in English Linguistics, 44). Roberta Facchinetti, Manfred Krug and Frank Palmer (eds.), 301–327. Berlin: Mouton de Gruyter. Georgetown Linguistics Web Chi Square Calculator. Georgetown Linguistics Web Chi square Calculator. Programmed by Catherine N. Ball, Jeffrey Connor-Linton, Kathi Taylor. http://www.georgetown.edu/faculty/ballc/webtools/web_chi.html (28 Dec. 2006). Görlach, Manfred 1987 Colonial lag? The alleged conservative character of American English and other ‘colonial’ varieties. English World-Wide 8 (1): 41–60. 1988 The study of early Modern English variation - the Cinderella of English historical linguistics. In Historical dialectology. Regional and social. Jacek Fisiak (ed.), 211–228. (Trends in Linguistics. Studies and Monographs, 37) Berlin: Mouton de Gruyter. Gourlay, Robert 1822 Statistical account of Upper Canada. Compiled with a view to a grand system of emigration (CIHM 35937). London: Simkin & Marshall. Hickey, Raymond 2004a Introduction. In Legacies of colonial English. Studies in transported dialects. Raymond Hickey (ed.), 1–30. (Studies in English Language) Cambridge: Cambridge University Press. 2004b Dialects of English and their transportation. In Legacies of colonial English. Studies in transported dialects. Raymond Hickey (ed.), 33– 58. (Studies in English Language) Cambridge: Cambridge University Press.

Taking permissible shortcuts?

383

Hultin, Neil C. 1967 Canadian views of American English. American Speech 42: 243– 260. Jones, Charles 1989 A history of English phonology. London: Longman. Kytö, Merja 1991 Variation in diachrony, with early American English in focus. Studies on CAN/MAY ad SHALL/WILL. Bern: Lang. Kytö, Merja, Juhani Rudanko and Erik Smitterberg 2000 Building a bridge between the present and the past: a corpus of 19thcentury English. ICAME Journal 24: 85–97. Kytö, Merja and Matti Rissanen 1983 The syntactic study of Early American English. The variationist at the mercy of his corpus? Neuphilologische Mitteilungen 84: 470– 490. Labov, William 1994 Principles of linguistic change. Volume 1: Internal factors. Oxford: Blackwell. Lovell, Charles J. 1955 Lexicographic challenges of Canadian English. Journal of the Canadian Linguistic Association (Canadian Journal of Linguistics) 1/(1) (Mar.): 2–5. Marckwardt, Albert H. 1958 American English. New York: Oxford University Press. Michalewicz, Zbigniew and David B. Fogel 2004 How to solve it: modern heuristics. 2nd ed. Berlin: Springer. Nelson, Gerald, Sean Wallis and Bas Aarts 2002 Exploring natureal language. Working with the British component of the International Corpus of English. Amsterdam: Benjamins. Nevalainen, Terttu 1999 Making the best use of ‘bad’ data. Neuphilologische Mitteilungen. 100 (4): 499–533. Oaks, Michael P. 1998 Statistics for corpus linguistics. Edinburgh: Edinburgh University Press. Orkin, Mark M. 1970 Speaking Canadian English. An informal account of the English [1971] language in Canada. [Reprint]. New York: McKay Company. Preacher, Kristopher J. and Nancy E. Briggs 2001 Calculation of Fisher’s Exact Test: an interactive calculation tool for 2 x 2 tables. Avail. from http://www.quantpsy.org (20 Dec. 2006).

384

Stefan Dollinger

Scargill, Matthew H. 1957 Sources of Canadian English. Journal of English and Germanic Philology 56: 611-614 [reprinted in Chambers (1975), 12–15]. Schneider, Edgar W. 2007 MY BABY LOVES ME, SHE LOVE ME: verbal -s variability in the history of black and white dialects of the southern United States. In Tracing English through time: explorations in language variation (Austrian Studies in English, 95). Ute Smit, Stefan Dollinger, Julia Hüttner, Gunther Kaltenböck and Ursula Lutzky (eds.), 345–358. Vienna: Braumüller. Simon-Vandenbergen, A. M. 1984 Deontic possibility: a diachronic view. English Studies 65 (4): 362– 365. Sundby, Bertil, Anne Kari Bjørge and Kari E. Haugland 1991 A dictionary of English normative grammar 1700–1800. Amsterdam: Benjamins. Thomas, Eric R. 1991 The origin of Canadian Raising in Ontario. Canadian Journal of Linguistics 36: 147–170. Traugott, Elizabeth Closs 1972 A history of English syntax. A transformational approach to the history of English sentence structure. New York: Holt, Rinehart & Winston. 1989 On the rise of epistemic meanings in English: an example of subjectification in semantic change. Language 65 (1): 31–55. Trudgill, Peter 2002 The history of the lesser-known varieties of English. In Alternative histories of English. Richard Watts and Peter Trudgill (eds.), 29–44. London: Routledge. 2004 New dialect formation. The inevitability of colonial Englishes. Edinburgh: Edinburgh University Press. Tversky, Amos and Daniel Kahneman 1982 Judgment under uncertainty: heuristics and biases. In Judgment under uncertainty: heuristics and biases. Daniel Kahnemann, Paul Slovic and Amos Tversky (eds.). Cambridge: Cambridge University Press. Vogt, W. Paul 2005 Dictionary of statistics & methodology. Thousand Oaks, CA: Sage. Warner, Anthony R. 1993 English auxiliaries. Structure and history. Cambridge: Cambridge University Press.

Taking permissible shortcuts?

385

Watts, Richard and Peter Trudgill (eds.) 2002 Alternative histories of English. London: Routledge. de Wolf, Gaelan Dodds, Robert J. Gregg, Barbara P. Harris, and Matthew H. Scargill (eds.)5 1997 Gage Canadian Dictionary. 5th ed. Toronto: Gage. [11967] Wolfram, Walt and Natalie Schilling-Estes 2004 Remnant dialects in the coastal United States. In Legacies of colonial English. Studies in transported dialects. Raymond Hickey (ed.), 172–202. (Studies in English Language) Cambridge: Cambridge University Press. Wood, J. David 2000 Making Ontario. Agricultural colonization and landscape recreation before the railway. Montreal: McGill-Queen’s University Press. Woods, Anthony, Paul Fletcher and Arthur Hughes 1986 Statistics in language studies. Cambridge: Cambridge University Press.

‘What strikes the ear’ Thomas Sheridan and regional pronunciation*

Raymond Hickey Judging by the number of publications by different authors during the eighteenth century and by the success they achieved with their writings, this was the period in which public concern with correctness in language as a social qualification takes a firm foothold in English, Scottish and Irish society. The general concern with putative correctness has its roots in earlier debates about standard English, above all among the writers of the Augustan period. Prominent among these was Jonathan Swift, to whom later writers refer and to whose authority they appeal. Robert Lowth in the preface to his famous grammar of 1762 writes: “Swift must be allowed to have been a good judge of this matter (i.e. the imperfect state of our language – RH); to which he was himself very attentive, both in his own writings, and in his remarks upon those of his friends: he is one of the most correct, and perhaps the best of our prose writers” (Lowth 1995 [1762]: vi).1 But Augustan writers like Swift appear to have been concerned with changes in English because these separated the language from that of their predecessors and would render their own writings linguistically obsolete to later generations. Lowth is somewhat different in his stance. In his grammar he talks of expressing oneself “rightly” but he also stresses the notion of “propriety” which was definitely a social concept. The idea is foregrounded that incorrect, i.e. non-standard, grammar is offensive to educated, middle-class ears, especially those of the capital London and its environs. With that two issues become topical: (i) regional pronunciations of English and (ii) uneducated, or “vulgar” usage. In earlier centuries, we have indications of the regional origin and accents of public figures. For instance Sir Walter Ralegh (1554–1618) was from Devon and we know from contemporary remarks that he spoke with a southwest English accent. What is new in the eighteenth century is the social censure of regional accents which were seen as signs of poor breeding and education and generally to be avoided.

388

Raymond Hickey

There is an added issue here which concerns the attitude and reaction of people from the regions to centralist notions of linguistic propriety. The first major reaction comes from Ireland; others from Scotland and the north of England were to follow. It might at first sight seem strange that an Irishman should be among the first to adopt, in print, an openly prescriptive and corrective stance on the English language. But this is only an apparent paradox. Given that eighteenth-century Dublin was a major city of Great Britain but an outpost at the same time, it is understandable that some Dubliners may well have internalised centralist notions of correctness and indeed, in a rush of over-assimilation, sought to be more English than the English themselves.

1.

Thomas Sheridan

The Irish individual in question is one Thomas Sheridan who was born in 1719, grew up in Dublin and died in London in 1788. Sheridan enjoyed a diverse career as actor, lecturer and writer. Significantly, he was the godson of Swift and produced The Works of Swift with Life (18 volumes) in 1784. As a dramatist Sheridan is known for one play, Captain O’Blunder or The Brave Irishman (1754), which he wrote in 1740 as an undergraduate. He was also manager of the Smock Alley theatre in Dublin (Sheldon 1967) where he worked for some years. However, the plays of his son Richard Brinsley Sheridan (1751–1816) were better known in his day and were regularly produced in London at Covent Garden and at the Drury Lane theatre. The interest which Thomas Sheridan latterly showed in correctness of language goes back at least to his book British Education: Source of Disorders, which appeared in 1756 shortly after he had left Ireland for England (see remarks below), and in his 1761 book A Dissertation on the Causes of the Difficulties which Occur, in Learning the English Tongue. This combined interest in education and in correcting what he perceived as unacceptable usage may well be something which he inherited from this father. Thomas Sheridan senior (1687–1738) was a clergyman and educator who established his own school in Capel St., Dublin where pupils were trained in the classical languages. In the present context, Thomas Sheridan junior is important as the author of A Rhetorical Grammar of the English Language (1781) which contains a section on the Irish pronunciation of English. He is also the author of a successful General Dictionary of the English Language (1780, 2 vol-

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

389

umes) and an earlier A Course of Lectures on Elocution published in the same year as Lowth’s grammar – 1762 – and by the same publishers – R. and J. Dodsley in London. Probably on the grounds of these linguistic interests, Sheridan developed a close friendship with the lexicographer Samuel Johnson. The latter’s opinion of the work of Sheridan throws an interesting light on English-Irish attitudes of the time. Dr. Johnson saw it as a distinct disadvantage for Thomas Sheridan to be Irish and pronouncing on English (Beal 2004b: 331). This contrasts strongly with the English attitude towards Swift, who was so often held up as a paragon of English style. But Swift was a Protestant dean who had worked in England and only occasionally wrote on specifically Irish matters. Sheridan, on the other hand, was a Catholic actor, minor playwright and self-appointed arbiter of correct English usage.

2.

Elocution

Sheridan’s interest in language surfaces in his middle years with the publication of his lectures on elocution. There is a connection between this initial concern and his later pronouncements which will be the subject of discussion below. Let me show this by considering what was meant by elocution in late 18th century Britain and Ireland. To start with one can locate elocution in classical Latin writings as one of the five so-called “offices”, or sections, of rhetoric: (1) invention, analyzing a topic and collecting material for it, (2) disposition, arranging the material for a speech, (3) elocution, finding appropriate words for the topic, the speaker and the audience, (4) pronunciation, oral delivery of a speech and (5) memory, committing the contents of the speech to memory. Elocution seems to have been mainly the domain of oratory and concerned with good style and expression, but its meaning altered over time, as recorded in the definition given in the Oxford English Dictionary: elocution /elԥ/kju$ԥn/ n. Late Middle English [L elocutio(n-), f. elocut- past participle stem of eloqui: see ELOQUENT, -ION.] 1 Oratorical or literary expression; literary style as distinguished from matter; the art of appropriate and effective expression. Late Middle English. 2 Eloquence, oratory. Latin 16c to Latin 18c. 3 The art of (public) speaking, esp. of pronunciation, delivery, gesture, and voice production. English 17c. 4 Manner or style of speaking. English 17c. M19.

390

Raymond Hickey

There is a discernible shift in emphasis by the 17th century. Elocution as the third office of rhetoric came to refer to effective public speaking: pronunciation, gesture and voice quality are singled out here. It is not surprising then that elocution came to be associated with actors given their professional concern with speaking in public and the fact that Sheridan himself was originally an actor. Consider also cases like that of Sarah Siddons (1755–1831), a successful tragic actress and acquaintance of Sheridan’s dramatist son, Richard Brinsley, who engaged her in Drury Lane in 1782. The following year she was engaged to teach elocution to the children of the royal family.2 Returning to Sheridan, we can see that he was concerned with establishing serious academic underpinnings for elocution. He found in the work of the 17th century philosopher John Locke assumptions which improved the credentials of elocution. Locke believed that “words are the signs of ideas, tones, i.e. spoken language, the signs of passions”. As elocution was taught by reciting previously composed material in public, it was clearly associated with ideas and hence to be taken more seriously. The opinion was adopted that the way to teach individuals to speak well was to train them in reciting written material. This stance characterised elocution within educational systems in the English-speaking world and still does inasmuch as it is still a school subject.

3.

Sheridan’s influence

It is known that Sheridan travelled widely throughout the British Isles, lecturing on elocution and “correct” English, notably in Scotland. The question of language, specifically of the differences between Scottish and southern English usage, had become an increasingly sensitive issue there, something which is apparent in the works of Buchanan, such as his British Grammar (1762) and his Essay Towards Establishing a Standard for an Elegant and Uniform Pronunciation of the English Language, Throughout the British Dominions (1766), where the term “British” carries distinct political overtones, all the sharper given the union of the English and Scottish parliaments in 1707 (Beal 2004a: 96). Sheridan had a considerable influence on authors in Britain, notably on his main rival, London-born John Walker who in 1791 published A Critical Pronouncing Dictionary of the English Language much in the vein of Sheridan’s works and in which he compares his own dicta to those of his

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

391

Irish colleague (see the many discussions in this respect in Beal 2004a, e.g. p. 129). Sheridan’s influence on Walker should not be underestimated. Not only did the latter include Sheridan’s admonitions to the Irish, Scottish and Welsh (see his reprint of Sheridan’s original comments, Walker 1815: 13– 19) but he adopted wholesale the prescriptive remarks which the former made about regional pronunciation in general. Walker seems to have had the greater success3 and his dictionary was reprinted repeatedly until the beginning of the twentieth century (over 100 times between 1791 and 1904 to be precise, Beal 2004a: 171). Walker was very detailed in his discussion of pronunciation variants in English (far more so than Sheridan) and the 15th edition of his dictionary contains a preliminary section, ‘Principles of English Pronunciation’, which is nearly 70 pages long (Walker 1815: 21–90). Many authors quickly recognised that among the rising middle classes there was a market for works on elocution and they jumped on the bandwagon with alacrity. A good example is Stephen Jones who produced a work somewhat conceitedly entitled Sheridan Improved: A General Pronouncing and Explanatory Dictionary of the English Language which reached its 3rd edition by 1798. Sheridan had other rivals apart from John Walker who produced works of the same type as he did, indeed in one case with the same title, and almost at the same time. For instance, William Kenrick produced A New Dictionary of the English Language in 1773 in London and brought out A Rhetorical Grammar of the English Language in 1784 only three years after Sheridan’s work of the same name. There were not just imitators but detractors as well; for instance, there is an anonymous book entitled A Caution to Gentlemen who use Sheridan’s Dictionary which, given the title, scarcely needs comment. John Walker also had his imitators and improvers. In 1836 B. H. Smart brought out Walker Remodelled. A New Critical Pronouncing Dictionary of the English Language, and followed this with The Practice of Elocution in 1842.

4.

Correcting the natives

Elocution is not just about teaching what is “correct” pronunciation but also about censuring what is deemed by authority to be “base”, “vulgar”, “rude” or just “provincial”. The censures of the elocutionists can be used by linguists today to glean information about regional and/or colloquial pronunciation of their time. In his A Rhetorical Grammar of the English Language (1781)

392

Raymond Hickey

Sheridan inserted an appendix entitled “Rules to be observed by the natives of Ireland in order to attain a just pronunciation of English” (Sheridan 1781: 137–55). This also contains a number of remarks on the “mistakes” which the Irish make when speaking English. At one point he offers a list consisting of two columns with Irish and English pronunciations respectively. Here there can be no doubt about Sheridan’s prescriptive intentions. Sheridan does not hold back with criticism of regional accents in parts of Britain either: “With regard to the natives of Scotland as their dialect differs more, and in a greater number of points, from the English, than that of any others who speak their language, it will require a greater number of rules, and more pains to correct it” (Sheridan 1781: 146). He is unashamedly censorious, but kindly offers suggestions for curing the unacceptable speech habits of the Scots. Here the goal is quite clear: provincial pronunciations are to be abandoned and those of educated southeast England are to be adopted. The Welsh are taken to task as well, this time for the devoicing of consonants in initial and medial position, e.g. fice for vice, seal for zeal, ashur for azure, etc. and again the goal is to weed out such provincialisms from their speech and so make it acceptable in educated circles in and around the capital. Before looking at Sheridan’s comments in detail, we might well ask what basis for making such pronouncements he may have had, given that he was born and reared in Ireland, a place where exposure to educated southeastern English usage would have been minimal. Despite his Irish birth and upbringing, Sheridan can claim to have had exposure to educated southeastern English usage of the mid 18th century. After attending his father’s school in Dublin, he went to Westminster College in London before returning to Ireland to attend Trinity College Dublin. He later became actively involved in the management of Smock Alley Theatre in Dublin, but in 1744 he travelled to London and competed as an actor with David Garrick, whom he persuaded to come to Dublin for the season in 1745–6. In the following years he managed Smock Alley but following riots in 1747 and 1754 he retired from acting and management and went to England where he began to publish his works on education and elocution.

5.

Sheridan’s strictures and 18th century Irish pronunciation

In the appendix to his grammar Sheridan compares the pronunciation of Irish English with southern British English. He is fairly accurate in his de-

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

393

scription and representation of vowel quality. The Irish English forms which he quotes are those in which “well-educated natives of Ireland differ from those of England” (1781: 146). The only group in Ireland that could be described as “well-educated” at the end of the 18th century was the Protestant middle and upper class. At this time the middle class of the city was prospering (see Moody and Martin 1967: 232ff. and Johnston 1994 [1974]: 77ff., for descriptions of Dublin at this time). And more than ever before this social group enjoyed independence from England (up to the ‘Act of Union’ of 1801, Johnston 1994 [1974]: 164ff., Ó Tuathaigh 1972: 29ff.). At this time Dublin enjoyed a certain political freedom (Craig 1969) and this would have minimised the direct influence of southeastern British English on the Irish pronunciation of English. Although one can identify Sheridan’s group of speakers fairly easily, it is not equally apparent just what he regarded as the standard English that he refers to and who is supposed to have spoken it. In the preface to his grammar he talks of “our pronunciation” (1781: xxii) and refers to Johnson with regard to spelling (1781: xxiii). He further notes that the pronunciation of English by the people in Ireland, Scotland and Wales can deviate from a standard, without offering any more specific information on what he regards this standard to be. His praising remarks (1781: xix) on the correct pronunciation of the “Augustan Age” in England (Sheridan was after all the godson of Swift, Croghan 1990), are of little help. From this one can conclude only that Sheridan was a prescriptivist and assumed educated southeastern English, the variety of “polite” society in London and the Home Counties, as a yardstick with which to compare Irish English of his time. 5.1. Sheridan’s system of pronunciation In the following paragraphs the system which Sheridan used (Sheldon 1946) for indicating regional pronunciations of English is explained and the extent to which it can be viewed as a window on Dublin English in the 18th century is discussed. However, it would be misleading to interpret his references as relating solely to educated Dublin usage of his time. Many of his strictures concern pronunciations that were common in Britain at the time. For instance, the lowering of /e/ to /a/ before /r/, the raising of /æ/ to /ȏ/, especially after velars, and the realisation of short vowels before /r/ are all matters that are relevant to varieties of English in Britain during his time.

394

Raymond Hickey

Notwithstanding this general relevance, Sheridan’s motive in his description was clearly to show the Irish just how wrong their pronunciation of English was. In consequence, on several occasions the detail which one would like concerning local features is not given because it does not seem to have been pertinent to Sheridan’s goal of demonstrating to the Irish the wrongness of their phonetic ways. Furthermore, what Sheridan mentions is obviously what “struck his ear” which does not mean that he registered all non-standard features of late 18th century Dublin English. He is singularly silent about non-standard consonantal features, of which there are many in present-day conservative Dublin English (Hickey 2005: 34–45) and so probably in that of his time as well. In order to describe the special features of Dublin English Sheridan employs a system of notation in which the five vowel graphemes (and y) are used together with a number (from 1 to 3) as a diacritic to denote possible vowel values in Irish English (Sheridan 1781: 151). In the original the number is actually placed over the vowel. For technical reasons, it is placed here as a superscript digit to the right of the vowel letter. In his text, Sheridan organises the vowel values into three columns as follows: (1) a e i o u y

First ha1t be1t fi1t no1t bu1t love-ly1

Second ha2te be2ar fi2ght no2te bu2sh ly2e

Third ha3ll be3er fi3eld no3ose blu3e

Sheridan also offers some notes in which he explains his transcriptional system: All improper diphthongs, or, as I have called them digraphs, I mean where two vowels are joined in writing, to represent any of the simple sounds in the scheme, are changed in the second column into the single vowels which they stand for; as thus bear

be2re

hear

he3re

fourth

fo2rth

door

do2re

(Sheridan 1781: 152)

The ‘final mute e’ is added to certain words in Sheridan’s pronunciation system in order to disambiguate the transcriptions.

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

395

It needs to be stressed here that Sheridan was working within an existing tradition of describing and classifying vowels. For instance, Samuel Johnson in his dictionary of 1755 used the same system to describe the three vowels which he recognised, “the slender, open and broad”, i.e. the sounds as in face, father and all respectively (Johnson’s examples, see discussion in “A grammar of the English tongue”, Johnson 1755 [no page numbers]). Sheridan’s table of vowels is assumed to represent values from educated southeastern English usage of his time against which he then compared Irish (and Scottish) pronunciations. The phonetic values of the above vowels in educated southeastern English usage are not known in all cases. For instance, the vowel a3 cannot be decided upon precisely. It is true that /au/ before /l/ in the early modern English period (Dobson 1968: 603ff.) went to /o:/, but it is not certain just how close this vowel was, it may well have been closer to /<:/. The vowel a2 did not have the value which it now shows in non-regional varieties of present-day British English, i.e. /ei/. It is known that ME /a:/, the historical precursor of this vowel, was raised and that by the early 18th century it was probably in a range between /@:/ and /e:/ (this is the area in which the vowel can be located according to the rhymes of Swift, for instance). The vowel u1 implies a value as /v/ although this interpretation leads to unexpected values in Dublin English. The reasons are as follows: Sheridan gives two short vowels for u. As one of the two is /u/, the vowel in bush, the only other possibility of u1 is /v/, especially as Sheridan used u1 for transcribing unstressed vowels. na2shu1n ko2ru1s

(2) a. nation b. chorus

(Sheridan 1781: 154)

In this case he equates the unrounded and lowered vowel /v/ with the unstressed schwa /q/, a lack of distinction which is the rule with many commentators and lexicographers in the 18th century. Sheridan also recognised different short vowels before /r/ which is obvious from the following examples. (3) a. /v/: b. /e/

stir birth

stu1r, be1rth

bird

bu1rd (Sheridan 1781: 152)

This distinctiveness of vowels is typical of more conservative varieties of English and its presence in local Dublin English to this day is evidence of

396

Raymond Hickey

the conservative nature of the latter form of English, cf. [be:(r)t]. This distinction is not part of the supraregional standard of English in either presentday Ireland or England. 5.2. Vowel values Sheridan’s “Rules to be observed by the natives of Ireland” also contains a number of remarks on the “mistakes” which the Irish make when speaking English. At one point he offers a list consisting of two columns with Irish and English pronunciations of keywords respectively. Many forms are instances of hypercorrection. William Kenrick (1773: 34), writing a little later, mentions the hypercorrect use of /v/ by northern English (and Irish) speakers in words which show, and always did show, the high back short vowel, e.g. bull. He also comments extensively on the use of /e:, e:/ in words which had ME <ea> and chides others for recommending this pronunciation in some words (Kenrick 1773: 40f.). This vowel value showed considerable variation, cf. Johnson (1755 [no page numbers]) who mentions in the preface to his dictionary that “Ea sounds like e long, as mean; or like ee, as dear, clear, near”, a remark which shows that for him mean was still [me:n]. While such mid-front pronunciations were replaced by a high front value soon afterwards in England, the older pronunciation survived much longer in Ireland and still exists in rural varieties and as a vernacular option for urban speakers. Some of the words quoted by Sheridan seem to be lexicalised exceptions to general English sound shifts. For instance, Sheridan quotes ‘drought’ as having the pronunciation dro3th /dru:t/ in Ireland, but there is no further evidence for the non-diphthongisation of Middle English /u:/ to /au/ at this late stage of early modern Irish English. When discussing late 18th century pronunciation and when comparing Irish and English usage at this time it is necessary to be explicit about what varieties are being referred to. On the Irish side, the matter is relatively simple, as Sheridan clearly states that the strictures he articulates refer to the speech of “the gentlemen of Ireland”. This a fairly clear reference to educated Irish usage centred around Dublin as the main English-speaking city and centre of education in late 18th century Ireland. On the English side, however, the matter is not quite so simple. To determine just what Sheridan thinks is the English pronunciation worthy of emulation, one can best let him speak for himself. In the preface to his Rhetorical Grammar of

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

397

1781, Sheridan is at some pains to be explicit about what his yardstick of pronunciation actually is. It must be obvious, that in order to spread abroad the English language as a living tongue, and to facilitate the attainment of its speech, it is necessary in the first place that a standard of pronunciation should be established, and a method of acquiring a just one should be laid open. That the present state of the written language is not at all calculated to answer that end, is evident from this; that not only the natives of Ireland, Scotland, and Wales, who speak English, and are taught to read it, pronounce it differently, but each county in England has its peculiar dialect, which infects not only their speech, but their reading also. (Sheridan 1781: xvii)

After this lament for the lack of uniformity in spoken English across the regions of Britain and Ireland Sheridan continues to remark on the variation to be found in social groups which enjoy higher positions in society. It is well known, that there is a great diversity of pronunciation of the same words, not only in individuals, but in whole bodies of men. That there are some adopted by the universities; some prevail at the bar and some in the senate-house. (Sheridan 1781: xix)

The above statements are a preamble to his main objective, that of specifying the yardstick of pronunciation which he favours, namely that of the early 18th century, specifically that of the reign of Queen Anne (1702– 1714), the last of the Stuarts. There was a time, and that at no very distant period, which may be called the Augustan age of England, I mean during the reign of Queen Anne, when English was the language spoken at court; and when the same attention was paid to propriety of pronunciation, as that of French at the Court of Versailles. This produced a uniformity in that article in all the polite circles; (Sheridan 1781: xix–xx)

This provides him with a baseline from which to criticise the changes in English which had taken place between that time and his own. Many pronunciations, which thirty or forty years ago were confined to the vulgar, are gradually gaining ground; and if something be not done to stop this growing evil, and fix a general standard at present, the English is likely to become a mere jargon, which every one may pronounce as he pleases. (Sheridan 1781: xx)

398

Raymond Hickey

In keeping with the Augustan tradition, Sheridan is concerned with fixing the language and protecting it from the pernicious effects of variation and change. When we reflect, that no evil so great can befall any language, as a perpetual fluctuation both in point of spelling and pronouncing, it is surely a point to be wished, that a permanent and obvious standard of both should at some certain period be established; and if possible, that period should be fixed upon when probably they were in the greatest degree of perfection. (Sheridan 1781: xxiii)

Against this background it is now possible to examine some of the pronunciations listed by Sheridan which have changed since his time and which offer evidence for what he regarded as acceptable usage and what was subject to his disapproval. 18c IrE /u:/ In the late Middle English period there was some fluctuation between /o:/ and /u:/ (Dobson 1968: 681ff.). Some words which later appear with /o:/ show /u:/ in Sheridan’s list. These are instances of the failure of /u:/ to diphthongise before /r/ (Dobson 1968: 688ff.). /du:r/ ‘door’ (4) a. do3re /flu:r/ ‘floor’ b. flo3re c. co3urse /ku:rs/ 18c IrE /e:/ The lack of raising of ME <ea> is a well-known characteristic of vernacular Irish English, both north and south, and is still found in present-day vernaculars (Milroy and Harris 1980). This feature and the use of /a:/ in words like patron, matron (see below) is pointed out by Sheridan as the chief “mistake” in the Irish pronunciation of English. From the point of view of earlier English, i.e. late Middle English, one has a collapse of the distinction between /e:/ and /e:/ but no further shift to /i:/: least /le:st/, deal /de:l/, beet /be:t/. Sheridan’s group of speakers would seem to have been aware of this feature of Irish English and it is understandable that in attempts to avoid the local pronunciation they engaged in hypercorrection by shifting all instances of /e:/ to /i:/, even where this was not justified etymologically.

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

(5) a. prey b. convey c. bear

ĺ ĺ ĺ

pree convee beer

399

(Sheridan 1781: 143)

This hypercorrection was noted by other authors too. For instance, Maria Edgeworth in her Essay on Irish Bulls (1802) remarked that “There are Irish ladies, who, ashamed of their country, betray themselves by mincing out their abjuration, by calling tables teebles, and chairs cheers!” (Crowley 2000: 136). 18c IrE /v/ Sheridan indicates that the /v/ vowel was to be found in the Dublin of his time in words like the following. (6) a. pu1lpit /pvlpit/ b. bu1sh /bv$/ d. pu1sh /pv$/

b. pu1dding c. bu1ll e. pu1ll

/pvdin/ /bvl/ /pvl/

All but the first of these words have /u/ in (southern) Irish English and British English today. Bearing in mind that Sheridan’s speaker group consisted of “well-educated natives” one can safely assume that the above instances are cases of hypercorrection parallel to the unconditioned shifting of /e:/ to /i:/ mentioned in the previous paragraph. 18c IrE /a:/ Similar to the lack of shift for historical /e:/ to /i:/ one also finds that /a:/ is not always raised to /e:/: patron /pa:trqn/, matron /ma:trqn/. Sheridan remarks that forms such as those just quoted have the same vowel as in father. Unshifted ME /a:/ was obviously a salient feature of 18th century Irish English and appears abundantly in parody literature, e.g. in George Farquhar’s The Beaux’ Stratagem (1707) as seen in the following extract.

FOIGARD Ireland! No, joy. Fat sort of plaace (= [pla:s]) is dat saam (= [sa:m]) Ireland? Dey say de people are catcht dere when dey are young.

Swift’s rhymes also show that /a:/ was an educated Irish pronunciation in the early 18th century. The censure of this pronunciation probably followed on the raising of the /a:/ to /e:, e:/ in southern England some considerable time before Sheridan. Swift was likely to have maintained a pronunciation, at least in rhyming verse, which was conservative even in his time.

400

Raymond Hickey

How by heroes of old our chiefs are surpass’d In each useful science, true learning, and taste. Verses on the Battle of the Books Why should the first be ruin’d and laid waste, To mend dilapidations in the last? Ode to Doctor William Sancroft This realisation has been lost everywhere in Ireland as a result of the later superimposition of more standard pronunciations. Occasionally, one has spelling pronunciations like status [sta:tqs] or data [da:tq] which is not a remnant of the former situation, however. 18c IrE /i:/ References to an /i:/ pronunciation in cases where English has /ai/ are not numerous in Sheridan. For instance, he gives the following form: Mi3kil /mi:kԥl/, compare English Mi2kil /maikql/ which he also quotes. However, this may be a transfer of the vowel value in the Irish form of the name, Mícheál //mji:h<:l/, despite the medial stop in English. Alternation among high vowels A notable feature of vernacular Irish English from the 17th century onwards, is the interchange of short high vowels. In the dialect glossaries some words are attested with both /i/ and /u/. In Sheridan’s list there is some evidence for the use of /i/ for /u/, e.g. in inion /injqn/ ‘onion’. This form is confirmed by Joyce (1979 [1910]: 100). 5.3. Conditioned realisations /e/ after velars Sheridan’s list of words contains the following forms. (7) a. ke1tch b. ge1ther

/ket$/ /gedqr/

‘catch’ ‘gather’

(Sheridan 1781: 144)

These are indicative of a raising of short /æ/ (from Middle English /a/) after velars. This raising would seem to have been general in varieties of English and there is lexical evidence of this in words like ketch ‘double-masted yacht’ and keg, a variant of cag from Old Norse kaggi. This raising is also found in Ireland before the 19th century and is recorded in the glossaries for the archaic dialect of Forth and Bargy in the southeast corner of Ireland

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

401

which died out in the early 19th century. Indeed a general raising would help to account for the tendency in present-day supraregional Irish English to avoid the use of /e/ and lower the vowel to something near the cardinal vowel /a/. Nonetheless, a raised realisation of /æ/ is still common in southwestern varieties of Irish English and more generally in all vernacular varieties – including Dublin English – in the position before /-r/, e.g. part [pe:r8], car [ke:r]. Diphthongisation before /l/+C The diphthongisation of back vowels before /l/C is a typical phonological process of the Middle English period (Prins 1974: 147; We1na 1978: 193). It presupposes the velarisation of /l/ as the diphthong can only have arisen through the back migration of a [U] off-glide before velarised [1] into the nucleus of the syllable in which it occurred. The matter is not quite straightforward for Irish English as it now shows a generalised alveolar [l] in all syllable positions (except for newer varieties of Dublin English, Hickey 2005: 77). This is probably a more recent development, stemming ultimately from the use of a non-palatal /l/ for English /l/ in all positions by speakers of Irish during the period of greatest language shift between the mid 17th and the late 19th century (Hickey 1986a). The first indications of the velar diphthongisation of back vowels is to be found in the 15th century (Wełna 1978: 192ff.) so that it is quite possible that the diphthong in question was already present in Irish English in the first period (before the 17th century). Other English authors noted this; for instance, John Ray in A collection of English words not generally used (1674) mentions the vocalisation of velar [1] before /d/ in words like caud (cold) and aud (old) (Ihalainen 1994: 202). Sheridan gives two forms with the diphthong. These show that Middle English proceeded from /o:ld/ to /ould/ and further to /au/ in Ireland as it also did in Scotland: cowld /kauld/ ‘cold’, bowld /bauld/ ‘bold’. In Dublin English, the /l/ persists, that is, it is not vocalised as a result of diphthongisation. Lowering of /e/ before /r/ This is a phonological process which is widely attested in Irish English historically and is indicated by Sheridan in the word sa1rch /sa:rt$/ ‘search’. Given the relative brevity of Sheridan’s list one can assume that the non-standard forms that he quotes are intended to be merely representative of the pronunciation rather than exhaustive. If this is correct, one can further assume that this lowering before /r/ was a general feature of Irish English up to the 19th century (it is attested in many literary

402

Raymond Hickey

documents). It has been entirely removed by later superimposition of more standard pronunciations and cannot be used as a diagnostic for present-day Irish English. The lowering of /e/ before /r/ can be regarded as a feature transported from England to Ireland by English speakers. There is no tendency for such a phonetic shift in Irish so that transfer during language shift (Irish to English) can be excluded. Furthermore, the issue of just what words in English should have /ar/ rather than /er/ is one which was discussed widely by the composers of pronouncing dictionaries (see remarks below). Vowel lengthening before voiceless fricatives The long vowel in words like last, path or staff has its origin in a lengthening of inherited /a/ which occurred in the south of England and which, along with other isoglosses, divides the country in two (Wakelin 1977 [1972]: 86f.). The occurrence or lack of this lengthening is one of the major defining features for general southern Irish English vis à vis the dialects of the east coast (from north of Dublin down to Waterford) which stem from late medieval settlement by the English. Short vowels before voiceless fricatives have been remarked on by some authors, cf. Hogan (1927: 63). Sheridan’s material is not so clear on this matter. He quotes the word wrath and the adjective wroth ‘angry’ with long and short vowels respectively in English but with a short and long vowel in Irish English. (8) a. wrath b. wroth

IrE IrE

wra1th /ræt/ wra3th /r>:t/

Eng wra3th /ro:2/ Eng wro1th /r>2/

Here the difficulties with Sheridan’s English reference accent are most apparent. He would seem to favour a conservative southeastern variety, that of “polite society” as he specifies in his preface, but just what group in England had retraction, rounding and lengthening of /a:/ after /w/ in wrath along with a short vowel before /2/ in wroth is unclear. Lack of rounding of /a/ after /w/ The rounding and retraction of short /æ/ in a position after /w/ or /hw/, but not before velars, e.g. wander, what but wag, probably started in the 16th century and was adopted into more standard forms of English in the 17th century (Prins 1974: 149f.). Sheridan shows that this rounding had not occurred in Dublin English by the end of the 18th century: squa1dron /skwædrqn/ = squo1dron /skw>drqn/ and is still not found in popular Dublin English.

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

403

Cluster simplification In the relatively short word list Sheridan offers two words, already attested in earlier Irish English, which show a simplification of clusters: lenth /lent/ ‘length’ and strenth /strent/ ‘strength’. The typical instance of such simplification in present-day English, namely nasal and stop in word-final position, as in ground, band, is not given. It may well be that reduction of clusters in this position was too salient a feature of popular Dublin English and hence avoided by his speaker group. Final cluster reduction is, however, well-attested before the 18th century and is still a marked feature of popular Dublin English. Dentalisation before /r/ Of all the features of late 18th century Dublin English that of dentalisation before /r/ is explained most clearly by Sheridan. In his remarks on /t/ (1781: 35) he notes that the Irish “thicken t (and d) so that they say betther for better and utther for utter”. It is clear subsequently that he is talking of dentalisation: “this faulty manner arises from the same cause that was mentioned as affecting the sound of the d, I mean the protruding of the tongue so as to touch the teeth.” He furthermore recognises that there is a morphological alternation of alveolar and dental stops in the comparison of adjectives: “thus though they (the educated Dublin Irish, RH) sound the d right in the positives loud and broad, in the comparative degree they thicken it by an aspiration; and sound it as if it were written loudher, broadher” (1781: 29). This pronunciation has been and still is very widespread in vernacular forms of English throughout all of Ireland. 5.4. Word stress From Sheridan’s system of vowel quantity notation there would seem to have been a correlation between quantity and stress placement in early modern Irish English. Consider first the manner in which accent is noted by Sheridan: The accent is placed throughout over the letter on which it is laid in pronunciation; over the vowel, when the stress of the voice is on the vowel; over the consonant when the stress is on that as thus – Consonant subscript 1 stu r/ 1 lu v/ 1 bi z/zy 1 1 la f /tu r

Vowel subscript 2 be /re 3 he /re 2 gro /ne 2 so /shal

404

Raymond Hickey

The syllables of words are divided according to the mode of pronouncing them; that is, all letters which are united in utterance in the same syllable, are here kept together also in writing, and separated from the rest; ... (Sheridan 1781: 155)

Statements such as the above have to be interpreted in modern terms to make sense. Sheridan would appear to view all consonants which follow on a short vowel as belonging to the same syllable. His superscript stroke indicates the end of the syllable. This device is usually superfluous, however. With the following forms, for instance, the superscript digit indicates a long vowel and an interpretation via syllable structure is not necessary. (9) a. IrE: ze2alous /ze:lqs/ b. IrE: ze2alot /ze:lqt/

Eng: ze1llus /zelqs/ Eng: ze1llut /zelqt/

Long vowels occur in Sheridan’s list in final position where English has a short unstressed vowel, cf. cla1/mour /kla:mu:r/ – cla1/mur /klæmqr/. Furthermore, one also finds that words of three syllables with a short stressed middle vowel have a long vowel in Sheridan’s speaker group (Sheridan 1781: 146). (10) a. IrE: ende2avour b. IrE: mali3cious

/en/de:vu:r/ /mq/li:$qs/

Eng: ende1v/ur /en/devqr/ Eng: mali1sh/us /mq/li$qs/

This vowel length is a conservative trait as is the lack of accent shift to the front of words which is also attested by Sheridan. (11) IrE: mischi3/evous /mis/t$i:vu:s/

Eng: mi1s/chivous //mist$ivԥs/

Some of his transcriptions show deviations in the supposedly English forms. For instance, he contrasts long and short vowels in Irish and English pronunciation which in the latter only have long vowels today. (12) a. IrE: che3ar/ful b. IrE: fe3ar/ful

/t$e:rful/ /fe:rful/

Eng: che1r/ful Eng: fe1r/ful

/t$e(r)ful/ /fe(r)ful/

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

405

5.5. Features not proscribed by Sheridan In dictating modes of pronunciation, Sheridan may not have been very successful. We know that his famous playwright son, Richard Brinsley, had a distinct Dublin accent. In her reminiscences of famous individuals she knew, Fanny Burney (1752–1840) imitates the Irish accent of the playwright by referring to his pronunciation of kind as [kԥind], indicated orthographically as koind. But Sheridan appears to have been selective in his censure. The centralised diphthong of the PRICE lexical set would appear to have been a feature of educated middle-class usage in 18th century Dublin. But it apparently only became a stereotype in the following century. By the end of the 19th century, English authors, such as Kipling, when trying to indicate an Irish accent in writing, used the oi spelling to indicate the centralised diphthong onset as in woild Oirland. There is an important generalisation here: elocutionists like Sheridan seemed to have proscribed pronunciations which were salient for them (Hickey 2000). They ignored those that were not. Naturally they also ignored features where regional pronunciations coincided with educated southeastern English usage. What happened occasionally is that the latter changed but the regional pronunciation did not, leaving it distinct from later standard usage. A case in point is the length of the vowel in the SOFT lexical set. In southern forms of English both early modern /æ/ and />/ were lengthened before voiceless fricatives, thus giving familiar pronunciations today like path, staff, pass all with long /<:/. Lengthening before the back vowel resulted in pronunciations of words such as cost, cloth, off, soft with /o:/. This second lengthening was reversed in 19th century southern English, but it was retained in east coast varieties of Irish English, typically in Dublin where it is still to be found today. The stigmatisation of long /o:/ in the SOFT lexical set is something which was already registered by Walker, especially before /2/ (Walker 1815: 38) where he regards it as characteristic of “innacurate speakers, and chiefly those among the vulgar”. This view is strengthened throughout the 19th century, with a degree of vacillation at the beginning of the 20th century and the final assertion of the short vowel in southeastern English usage (Beal 2004a: 141f.). But here as in other instances, Dublin was left behind. Having adopted the long vowel during the 18th century, it has retained it to the present-day. Interestingly, varieties of English outside Dublin do not

406

Raymond Hickey

have this lengthening so that there is a split in Ireland between Dublin [s>:ft], [k>:st], [>:f] and non-Dublin [s>ft], [k>st], [>f]. There are a couple of other pronunciation features which Sheridan did not censure. One of these would appear to be traceable to the effect of the Irish language on English in Ireland. Sheridan favoured the assibilation of /tj/ to [t$] as in tune [t$u:n] whereas it is condemned by his English counterparts. In this, as Beal (2004: 147) rightly notes, Sheridan was favouring a particularly Irish pronunciation. 5.6. Summary By and large one can say that Sheridan’s prescriptive comments on late 18th century Dublin English provide a useful glimpse of what pronunciation must have been like at the time. Especially in the area of vowels, Sheridan provides information that corroborates findings for present-day local Dublin English. The area of consonants was not dealt with in any great detail by Sheridan – something which has been typical of the elocutionary tradition since then – so that confirmation or refutation of such typical present-day features as t-lenition is unfortunately not forthcoming from Sheridan’s rhetorical grammar.

6.

Developments in the 19th century

It is interesting to consider the long-term effects of the proscriptions embodied in Sheridan’s works.4 To assess these one must consider the situation in 19th century Ireland and see what has happened to the pronunciations of his time which he chose to censure. Here one can distinguish different situations. There are features which have disappeared entirely. An example of this is ‘CATCH raising’, i.e. the pronunciation of the word as /ket$/. The raised pronunciation, especially after velars, was common in both Britain and Ireland in the later 18th and the 19th century, indeed well into the 20th century when it disappeared after World War Two. (The vowel raising has, if anything, been reversed so that a pronunciation closer to /a/ is to be found in RP nowadays, Bauer 1994: 110–21.) There are many attestations of this raising up to the early 20th century when Sean O’Casey used it to indicate local Dublin pronunciation. Sheridan (1781: 144) illustrates this by writing gather as gether and catch as cetch. A similar raising

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

407

would seem to have applied to low back vowels in certain environments. Sheridan remarks that words like psalm, qualm, balm are pronounced as if written psawm, quawm, bawm (Sheridan 1781: 141), a feature attested elsewhere in historical documents of Irish English. Both types of raising have disappeared without trace or comment from Irish English. Another feature, destined to disappear in the course of the 19th century, is ‘SEARCH lowering’, indicated in Sheridan by sa1rch /sa:rt$/ which he explicitly contrasts with English se1rch /se:rt$/ (Sheridan 1781: 145). This lowering before /r/ was a general feature of Irish English up to the 19th century. It is also found in many dialects of English and can be seen in names like Derby, Berkshire, Hertfordshire and is responsible for pronunciations like dark and barn in present-day English. The lowering has been entirely removed from Irish English by the superimposition of later more standard pronunciations and cannot be used as a diagnostic for present-day Irish English. Sheridan showed a preference for some words with long /a:/ before /r/, e.g. merchant which Walker thinks should be pronounced with /e:/ (Walker 1815: 330). He remarked that /ar/ in words like service and servant is “still heard among the lower order of speakers” (Walker 1815: 30). However, these words survive with /ar/ into the late 19th century in Ireland as seen in many dramas such as those by Dion Boucicault (1820–1890) writing in the 1860s and 1870s, see Hickey (2007, Chapter 5). Beal (2004a: 131) discusses Walker’s attitude to Sheridan on this point. Walker rightly recognises that Sheridan was old-fashioned. His preference for the low vowel in such words is an instance of lag where Dublin pronunciation in the latter half of the 18th century was simply not keeping up with developments in England where a raised vowel, /e:/, was diffusing rapidly through the lexicon. In addition, as Beal (2004a: 132) rightly notes, Walker was likely to have been an innovator on the fringe of polite society and seeking to become part of it. If a feature was not removed from Irish English in the course of the 19th century, it may instead have been relegated to a vernacular mode with a more standard pronunciation representing default usage. Consider the case of what I call ‘BOLD diphthongisation’ (see 5.3 Conditioned realisations above). Sheridan examples – cowld /kauld/ ‘cold’ and bowld /bauld/ ‘bold’ as well as others – were still quite common well into the 19th century. However, what happened in supraregional forms of Irish English is that the standard English pronunciation with /o:/ was adopted and the forms with /au/ were confined to vernacular varieties. In addition a lexical split took

408

Raymond Hickey

place whereby a world like old with the diphthong, i.e. [aul], implies personal attachment, affection as in I’m afraid the aul car has finally cracked up.

7.

Conclusion

The aim of Thomas Sheridan in the late 18th century was to point out what he regarded as unacceptable features in provincial accents. How these were determined is not entirely clear, but certainly features which had become salient, the object of negative social comment and linked to vernacular varieties, were certainly among the preferred targets for criticism by Sheridan and other elocutionists who followed. A high degree of arbitrariness would seem to have been typical then and now. There are no clearly stated reasons why Sheridan should have denounced some features but failed to mention others. The writer Thomas Sheridan was in a way like John Walker, his British counterpart: he was an outsider to polite English society and just the sort of individual who would adopt emerging pronunciations in the language around him. Compared with Walker he was, however, more conservative given his Irish background and in some cases, as in the example of merchant with older [ar] rather than more recent [er], he recommended pronunciations which Walker did not. In common with Walker and other prescriptivists who followed him, Sheridan shared a strong conception of acceptable English, a socially preferred variety, based on non-local southeastern English usage with which vernacular forms were unfavourably compared (Mugglestone 2003). This is a specific development of the 18th century and can be seen by comparing the age of Swift, at the beginning of this century, with that of Sheridan towards the end of the century. In the intervening decades a sea change had taken place: concerns about the immutability of the English language had given way to concerns about what social groups spoke in what way. The avenue of sociolinguistic assessment and censure had been opened up and was never to close again.

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

409

Notes *

1.

2.

3.

4.

My thanks go to two anonymous reviewers who provided a number of useful and justified criticisms of an earlier version of this article, specifically regarding the need to view Sheridan and his pronouncements in a wider English context. Shortcomings in the final version are, as always, my own responsibility. Lowth also owes a debt to earlier grammarians, those of the previous century who were concerned with general principles of language and he states: “Grammar in general, or Universal Grammar, explains the Principles which are common to all languages” (Lowth 1995 [1762]: 1). Here we have an echo of 17th century authors such as Wilkins and their notions of a philosophical language (Barber 1975: 137–41) though Lowth is much more practical and concrete in his approach. Later a link between quality education and elocution became explicit and elocution became an established academic discpline. Again consider a case like the American Hallie Quinn Brown (1850–1949), who was an educator and elocutionist and instrumental in the setting up of women’s clubs for African Americans. In 1893 she was appointed professor of elocution at Wilberforce University, Ohio where she had studied as a young woman. One could speculate why Walker’s rather than Sheridan’s pronunciation guide was to survive into the 19th century (Sheldon 1947). One reason could well be that educated individuals in Victorian England (Phillipps 1984) were not inclined to use the work of an Irishman as a yardstick of correctness in their own language. See Mugglestone (2003: 44f.) for some comments on the possible influence of Walker and Sheridan on the pronunciation of English by the natives of Ireland.

References Barber, Charles 1997 Early Modern English. 2nd ed. Edinburgh: University Press. Bauer, Laurie 1994 Watching English Change. London: Longman. Beal, Joan C. 2004a English in Modern Times 1700–1945. London: Edward Arnold. 2004b “Marks of disgrace”: Attitudes to non-standard pronunciation in 18th-century pronouncing dictionaries. In Methods and Data in English Historical Dialectology. M. Dossena and R. Lass (eds), 329–49. Craig, Maurice 1969 Dublin 1660–1860. A Social and Architectural History. Dublin: Allen Figgis.

410

Raymond Hickey

Croghan, Martin J. 1990 Swift, Thomas Sheridan, Maria Edgeworth and the evolution of Hiberno-English. In The English of the Irish. Irish University Review 20:1. Terence P. Dolan (ed.), 19–34. Dublin: n.p. Crowley, Tony 2000 The Politics of Language in Ireland, 1366–1922. A Sourcebook. London: Routledge. Dobson, E. J. 1968 English Pronunciation 1500–1700. 2nd ed. Oxford: Clarendon Press. Hickey, Raymond 1986a Possible phonological parallels between Irish and Irish English, English World-Wide 7, 1: 1–21. 1986b Length and frontness with low vowels in Irish English, Studia Linguistica 39, 2: 143–156. 2000 Salience, stigma and standard. In The development of standard English 1300–1800. Theories, descriptions, conflicts. Laura Wright (ed.), 57–72. London: Cambridge University Press. 2001 The South-East of Ireland. A neglected region of dialect study. In Language Links: The Languages of Scotland and Ireland. Belfast Studies in Language, Culture and Politics, 2. John Kirk and Dónall Ó Baoill (eds.), 1–22. Belfast: Queen’s University. 2003 How and why supraregional varieties arise. In Insights into Late Modern English. Linguistic Insights, Studies in Language and Communication, Vol 7. Marina Dossena and Charles Jones (eds), 351– 373. Frankfurt: Peter Lang. 2005 Dublin English. Evolution and Change. Amsterdam: John Benjamins. 2007 Irish English. History and Present-Day Forms. Cambridge: University Press. Hogan, James Jeremiah 1927 The English Language in Ireland. Dublin: Educational Company of Ireland. Ihalainen, Ossi 1994 The dialects of England since 1776. In The Cambridge History of the English Language. Vol. 5: English in Britain and Overseas: Origins and Development. Robert W. Burchfield (ed.), 104–119. Cambridge: Cambridge University Press. Johnson, Samuel 1755 A dictionary of the English language. London: J. Strahan. Reprinted 1967 by The Ams Press, New York.

‘What strikes the ear’: Thomas Sheridan and regional pronunciation

411

Johnston, Edith Mary 1994 Eighteenth Century Ireland. The Long Peace. The New Gill History [1974] of Ireland, Vol. 4. Dublin: Gill and Macmillan. Johnston, William 1764 Pronouncing and spelling dictionary. London: The author. Jones, Stephen 1798 Sheridan improved: A general pronouncing and explanatory dictionary of the English language. 3rd ed. London: Vernor and Hood. Joyce, Patrick Weston 1979 English as We Speak it in Ireland. London: Longmans, Green & Co. [1910] Kenrick, William 1773 A new dictionary of the English language. London: John and Francis Rivington. 1784 A rhetorical grammar of the English language. London: Cadell and Longman. Kirk, John and Dónall Ó Baoill (eds) 2001 Language Links: The Languages of Scotland and Ireland. Belfast Studies in Language, Culture and Politics, 2. Belfast: Queen’s University. Lowth, Robert 1995 A Short Introduction to English Grammar. Reprint. London. [1762] Milroy, James and John Harris 1980 When is a merger not a merger? The MEAT/MATE problem in Belfast vernacular, English World-Wide 1: 199–210. Moody, Theodore and Francis X. Martin (eds) 1967 The Course of Irish History. Cork: Mercier. Mugglestone, Lynda 2003 ‘Talking Proper’. The rise of accent as social symbol. 2nd ed. Oxford: University Press. Ó Tuathaigh, Gearóid 1972 Ireland before the Famine 1798–1848. The Gill History of Ireland, Vol. 9. Dublin: Gill and Macmillan. Phillipps, K. C. 1984 Language and class in Victorian England. Oxford: Blackwell. Prins, Anton 1974 A History of English Phonemes. Leiden: University Press. Ray, John 1674 A collection of English words not generally used with their significations and original in two alphabetical catalogues. London: C. Wilkinson.

412

Raymond Hickey

Romaine, Suzanne 1998 The Cambridge History of the English Language. Volume IV 1776– 1997. Cambridge: University Press. Sheldon, Esther K. 1946 Pronouncing systems in eighteenth-century dictionaries, Language 22: 27–41. 1947 Walker’s influence on the pronunciation of English, Proceedings of the Modern Language Association of America 62: 130–46. 1967 Thomas Sheridan of Smock Alley. Princeton, NJ: University Press. Sheridan, Thomas 1761 A dissertation on the causes of the difficulties which occur, in learning the English tongue. London: R. and J. Dodsley. 1781 A rhetorical grammar of the English language calculated solely for the purpose of teaching propriety of pronunciation and justness of delivery, in that tongue. Dublin: Price. 1784 The life of the Rev. Dr. Jonathan Swift. London. Reproduced Garland 1974. 1967 A general dictionary of the English language. 2 vols. Menston: The [1780] Scolar Press. 1970 A course of lectures on elocution. Hildesheim: Georg Olms. [1762] Smart, B.H. 1836 Walker Remodelled. A New Critical Pronouncing Dictionary of the English Language. London: T. Cadell. 1842 The Practice of Elocution. 4th ed. Lond: Longman, Brown, Green and Longmans. Swift, Jonathan 1712 A proposal for correcting, improving and ascertaining the English language. Wakelin, Martyn F. 1977 English Dialects. An Introduction. 2nd ed. London: Athlone Press. [1972] Walker, John 1791 A Critical Pronouncing Dictionary and Expositor of the English Language. Reprinted by the The Scolar Press (Menston). 1815 A Critical Pronouncing Dictionary and Expositor of the English Language. 15th ed. London: A. Wilson. Wełna, Jerzy 1978 A Diachronic Grammar of English. Part One: Phonology. Warsaw: Panstwowe Wydawnictwo Naukowe.

Author index

A

Aarts, Bas, 375, 383 Abraham, Werner, 379 Ackema, Peter, 278, 298 Adams, Valerie, 266, 298 Adamson, Sylvia, 324 Akenson, Donald H., 360, 361, 378n, 379 Alexander, Michael, 123, 131, 132, 137, 141–143, 151 Allen, Cynthia L., 227, 301, 305, 306, 307, 309, 311, 320n, 321n, 323 Allen, Harold, 28, 50 Álvarez, Dolores González, 380 Andersen, Gisle, 275, 298 Anderson, Eric, 121 Anderson, John M., 301, 314, 320n, 323 Ash, Sharon, 48, 53 Attridge, Derek, 117 Avis, Walter S., 357, 362, 371, 379 Aycock, Joan Marie, 347, 353 B

Baayen, Harald R., 208, 210, 227, 236, 237, 270, 297n, 298 Bailey, Charles J.N., 325 Bailey, Guy, 19, 50, 51 Ball, Catherine N., 382 Baranowski, Maciej, 38, 48, 53 Barber, Katherine, 371, 379 Barbier, Sjef, 379

Barber, Charles, 409 Barton, Michael, 49, 51 Bately, Janet, 97n, 98 Bauer, Laurie, 214, 225n, 227, 235, 266, 298, 406, 409 Bausenhart, Werner, 360, 380 Beal, Joan C., 8, 229, 378, 389, 390, 391, 405, 406, 407, 409 Beckman, Mary E., 63, 76 Benskin, Michael, 58, 82, 97n, 98 Benson, Larry D., 156, 163, 175, 176, 177 Bergs, Alexander, 260n, 262, 326 Beths, Frank, 60, 78, 80 Beukema, Frits, 379 Biber, Douglas, 261n, 262, 267, 296n, 298, 378n, 380 Biewer, Carolin, 222, 229 Birner, Betty J., 2, 10, 65, 66, 71, 76 Bjørge, Anne Kari, 384 Bjork, Robert E., 151 Blackburn, Simon, 330, 335, 358, 380 Blake, Norman, 324 Bliss, A. J., 103, 118, 153 Bloomfield, Morton W., 359, 380 Boberg, Charles, 16, 46, 48, 51, 53 Bonnycastle, Stephen, 379 Borroff, Marie, 123, 134, 136, 139, 144–146, 151 Boucicault, Dion, 407 Bradley, David, 47, 51

414 Author index

Breton, Roland J. L., 341, 354 Brewer, Charlotte, 270, 273, 298 Briggs, Nancy E., 375, 384 Brinton, Laurel J., 277, 278, 298, 357, 380 Brogan, T. V. F. , 112, 118 Brunner, Karl, 57, 59 Bueno Alonso, Jorge L., 378 Bülbring, Karl, 55, 57, 81, 98 Burchfield, Robert W., 410 Burney, Fanny, 405 Burnley, David, 233, 234, 236, 247, 249, 259n, 262 Bybee, Joan, 255, 256, 259, 261n, 262, 311, 316, 320, 323, 381 C

Cable, Thomas, 3, 101–119, 125, 126, 133, 134, 151, 153, 159, 177, 205 Calan, Louise, 347, 354 Cameron, Angus, 56, 98 Campbell, Alistair, 55, 59, 82, 97n, 98, 115, 118 Caon, Luisella, 328, 336 Carver, Craig M., 44, 45, 51 Chambers, J. K., 18, 51, 353, 354, 359, 360, 361, 362, 380, 381, 384 Chapman, Don, 6, 7, 207, 210– 215, 222, 223, 227, 265–299 Chen, Matthew, 26, 51 Chickering, Howell D., 103, 118, 121, 123, 131, 132, 135, 137, 141–143, 151, 153 Chomsky, Noam, 104, 118 Claridge, Claudia, 267, 269, 278, 298 Clark, Eve V., 278, 298

Clarke, Sandra, 16, 51, 380 Clyne, Michael, 359, 381 Coates, Jennifer, 331, 335, 364, 365, 367, 368, 374, 381 Collins, Peter, 381 Connor-Linton, Jeffrey, 382 Conrad, Susan, 262, 298, 380 Contini-Morava, Ellen, 230 Corrigan, Karen P., 335 Cowan, Helen, 361, 378, 381 Cowie, Claire, 208, 227, 232, 235, 240, 241, 243, 244, 261n, 262 Craig, Maurice, 393, 410 Crate, Charles, 379 Croft, William, 220, 227, 302, 303, 304, 316, 317, 318, 319, 323 Croghan, Martin J., 393, 410 Crowley, Joseph Patrick, 96n, 98 Crowley, Tony, 333, 335, 399, 410 Cruse, D. Alan, 316, 317, 323 Curzan, Anne, 260n, 263, 380 D

Dalton-Puffer, Christiane, 232, 240, 241, 244, 246, 247, 262, 263, 378, 381 Delfs, Lauren, 72, 78 Denison, David, 5, 207–230, 259, 301, 305, 318, 320n, 320n, 324, 328, 336, 364, 378n, 381 Diewald, Gabriele, 326 Disner, Sandra, 14, 51 Dixon, Robert M.W., 321n, 324 Dobson, E. J., 332, 335, 395, 398, 410 Dolan, Terence P. 410

Author index 415

Dollinger, Stefan, 8, 327, 330– 331, 357–385 Dossena, Marina, 328, 335, 382 Drout, Michael, 122 Drysdale, Patrick, 379 Duggan, Hoyt, 105, 126, 133, 134, 151 Dury, Richard, 382 E

Eble, Connie, 8, 327, 328–329, Eckert, Penelope, 38, 49, 51 Edgeworth, Maria, 333, 399 Edwards, John, 380 Ehrman, Madeline E., 368, 374, 382 Elenbaas, Marion, 226n, 228 Elmer, Willy, 301, 305, 324 Elms, Ford, 51 Emmonds, Kimberly, 380 F

Facchinetti, Roberta, 364, 366, 367, 374, 382 Fasold, Ralph, 38, 51 Fastovsky, David E., 12, 51 Fee, Margery, 357, 380 Ferguson, Charles A., 23, 51 Finch, Casey, 123, 134, 136, 139, 144–146, 151 Finegan, Edward, 262, 298 Fischer, Kerstin, 326 Fischer, Olga, 76, 301, 306, 317, 324 Fisiak, Jacek, 382 Fitzgerald, Christina M., 3, 5, 101, 102, 113–116 Fitzmaurice, Susan M., 1, 201, 222, 228, 327, 335

Flasdieck, Hermann, 97n, 98 Fleischman, Suzanne, 232, 249, 259, 263, 381 Fletcher, Paul, 385 Fogel, David B., 358, 363, 383 Frank, Roberta, 82, 98 Franko, Stefan, 299 Frant Hecht, Barbara Frascarelli, Mara, 63, 77 Fraser, James B [Bruce], 212, 228, 266, 299 Fretheim, Thorstein, 64, 77 Fulk, Robert D., 3, 55, 56, 57, 58, 59, 81, 97n, 98, 103, 118, 153 G

Gadde, Frederik, 232, 263 Gisborne, Nik, 320, 326 Givón, Talmy, 64, 67, 77 Glass, Gene V., 158, 177 Glassie, Henry, 43, 44, 45, 52 Gneuss, Helmut, 97n, 98 Godden, Malcolm, 82, 98 Godfrey, J., 62, 77 Goldberg, Adele E., 220, 228, 304, 312, 321n, 324 González Álvarez, Dolores, 380 Gordon, Elizabeth, 332, 335 Gordon, Matthew J., 37, 52 Gore, Laura Locoul, 342, 343, 347, 351, 354 Görlach, Manfred, 371, 378, 382 Gotti, Maurizio, 382 Gourlay, Robert, 360, 382, Greenbaum, Sidney, 299 Gregg, Robert J., 18, 52, 381 Gregory, Michelle L., 62, 65, 77 Gries, Stefan Th., 226n, 228, 318, 324, 325

416 Author index

Guerra, Javier Pérez, 380 Gumperz, John J., 338, 354 Gundel, Jeannette, 62, 64, 65, 69, 77 H

Haeberli, Eric, 60, 78 Halle, Morris, 117, 118, 159, 161, 177 Hardy, Ashley K., 97n, 98 Harris, Barbara P., 381 Harris, John, 398, 411 Harrison, Keith, 123, 134, 136, 144–146, 151 Haudricourt, A. G., 27, 52 Haugland, Kari E., 384 Hay, Jennifer, 255, 259, 263, 335 Hayes, Bruce, 117, 118, 162, 177 Healey, Antonette diPaolo, 323, 324 Heaney, Seamus, 121, 123, 131, 132, 137, 138, 139, 141–143, 151 Hedberg, Johannes, 97n, 99 Hedberg, Nancy, 62, 64, 69, 77 Heikkonen, Kirsi, 325 Heine, Bernd, 324 Herold, Ruth, 22, 29, 42, 52 Herring, Susan C., 62, 77 Herzog, Marvin, 54 Hickey, Raymond, 8, 9, 328, 331– 334, 353, 373, 374, 381, 382, 387–413 Hilpert, Martin, 226n, 228 Hinterhölzl, Roland, 63, 77 Hogan, James Jeremiah, 402, 410 Hogg, Richard M., 2, 55, 58, 59, 76, 326 Holliman, E., 62, 77

Hopkins, B. R., 158, 177 Hopkins, Kenneth D., 158, 177 Hopper, Paul, J., 220, 302, 304, 308, 311, 312, 313, 317, 321n, 324, 325 Horn, Laurence J., 77 Hudson, Richard, 214, 215, 228, 302, 325 Hughes, Arthur, 385 Hultin, Neil C., 359, 383 Hundt, Marianne, 222, 228 Hutcheson, B.R., 131, 151 Hüttner, Julia, 384 I

Ihalainen, Ossi, 230, 401, 410 J

Jakobson, Roman, 15, 52 Jespersen, Otto, 31, 32, 52, 266, 272, 299 Johansson, Stig, 262, 298 Johnson, Samuel, 389, 393, 395, 396, 411 Johnston, Edith Mary, 393, 411 Johnston, William, 411 Jones, Charles, 332, 335, 336, 359, 383 Jones, Stephen, 391, 411 Jordan, Richard, 97n, 99 Joyce, Patrick Weston, 400, 411 Juilland, A. G., 27, 52 Jurafsky, Daniel, 312, 325 K

Kahneman, Daniel, 358, 384 Kaltenböck, Gunther, 384 Kaluza, Max, 155, 156, 178 Kastovsky, Dieter, 381

Author index 417

Kaufman, Terrence, 247, 264 Kehoe, Andrew, 381 Keller, Frank, 222, 229 Kemenade, Ans, van, 60, 61, 76, 77, 78 Kennedy, A. G., 212, 229 Kennedy, Charles W., 123, 125, 129, 131, 132, 136, 141–143, 151 Kenrick, William, 333, 336, 391, 396, 411 Kenyon, John, 32, 52 Ker, N. R., 88, 89, 90, 91, 92, 93, 94, 95, 99 Keyser, Samuel J., 117, 118, 159, 161, 177 Kiparsky, Paul, 159, 178 Kirk, John, 411 Kiss, Katalin É., 62, 63, 64, 77 Klaeber, F., 103, 118, 153 Kleinman, Scott, 97n, 99 Kniffen, Fred B., 43, 44, 45, 52 Knott, Thomas, 32, 52 Koch, John, 156 Kohonen, Viljo, 72, 73, 78 Koopman, Willem, 60, 76, 78, 324 Kroch, Anthony, 61, 72, 78 Krug, Manfred, 382 Kurath, Hans, 32, 52 Kwon, H.-S., 272, 299 Kytö, Merja, 325, 364, 368, 378, 383

Lakoff, George, 314, 325, Lambrecht, Knud, 62, 64, 65, 66, 78 Langacker, Ronald, 311, 316, 317, 325 Lapata, Maria, 222, 229 Lass, Roger, 8, 10, 189, 204, 335 Law, Vivien, 324 Lecoy, Félix, 155, 178 van der Leek, Frederike, 301, 306, 324 Leech, Geoffrey, 262, 298, 299 Leechman, Douglas, 379 Lehmann, Ruth P., 102, 104, 118, 123, 131, 137, 141–143, 151 Lennig, Matthew, 29, 53 Li, Xingzhong, 3, 5, 101, 106– 112, 155–179 Liberman, Mark, 159, 178, 225 Liljencrants, Johan, 14, 15, 53 Lima, Maria, 382 Lindblom, Bjorn, 14, 15, 53 Lindelöf, Uno, 55, 266, 299 Lippi-Green, Rosina, 339, 354 Lloyd, Cynthia, 225n, 229 Locke, John, 390 de Lorris, Guillaume, 155 Los, Bettelou, 61, 77 Lovell, Charles J., 371, 379, 383 Lowth, Robert, 333, 387, 389, 409n, 411 Luick, Karl, 27, 53, 55 Lutzky, Ursula, 384

L

M

Labov, William, 1, 2, 11, 17, 23, 27, 37, 38, 39, 42, 45, 47, 52, 53, 304, 325, 327, 336, 358, 383 Laing, Margaret, 58, 60

Maclagan, Margaret, 335 Mair, Christian, 378 Marchand, Hans, 225n, 226n, 229, 231, 232, 245, 263, 266, 299

418 Author index

Martin, Francis X., 393, 411 Marckwardt, Albert H., 383 Martinet, André, 11, 12, 14, 15, 27, 49, 53 Martínez, Esperanza Rama, 380 Masayuki, Ohkado, 73, 76, 78 McCone, Kim, 47, 53 McDaniel, J., 62, 77 McDavid, Jr., Raven I., 32, 52 McIntosh, Angus, 58, 60, 201n, 204 McIntyre, Andrew, 226n, 229 McKelvey, Blake, 41, 42, 53 Meillet, Antoine, 7, 10, 11, 53, 302, 315, 316, 319, 321n, 325 de Mendoza Ibáñez, Francisco J. Ruiz, 325 Menner, Robert J., 97n, 99 Merlo, Paolo, 325 Merwin, W.S., 105, 106, 123, 124, 134, 136, 139, 144–146, 150n, 151 Meurman-Solin, Anneli, 328, 336 de Meun, Jean, 155, 156, 157 Michaelis, Laura A., 62, 65, 77 Michalewicz, Zbigniew, 358, 363, 383 Miller, D. Gary, 232, 247, 263 Mills, Gary B., 340, 354 Milroy, James, 327, 338, 339, 354, 398, 411 Milroy, Lesley, 238, 263, 327, 336, 338, 339, 353, 354 Minkova, Donka, 1, 2, 10, 27, 54, 116, 118, 126, 152, 159, 178, 201n, 202n, 204, 205, 206 Mitchell, Bruce, 55, 56, 60, 99 Mondorf, Britta, 230 Moisl, Hermann L., 229

Moody, Theodore, 393, 411 Mossé, Fernand, 115, 118, 198, 204 Mugglestone, Lynda, 408, 411 N

Nearey, Terrance M., 48n, 54 Neeleman, Ad, 278, 298 Nelson, Gerald, 375, 383 Nesselhauf, Nadja, 222, 229 Nevalainen, Terttu, 79, 230, 327, 328, 336, 358, 383 Nightingale, Pamela, 239, 259, 260n, 263. Noël, Dirk, 320, 321n, 325 Nurmi, Arja, 79 O

Ó Baoill, Dónall, 411 O’Casey, Sean, 406 O’Dowd, Elizabeth M., 224, 229 Oakden, J. P., 119 Oaks, Michael P., 374, 383 Orkin, Mark M., 360, 383 Ó Tuathaigh, Gearóid, 393, 411 Ourioupina, Olga, 222 P

Pagliuca, William, 311, 323 Palander-Collin, Minna, 317, 325, Palmer, Chris C., 5, 207, 208–210, 222, 229 Palmer, Frank, 382 Parker, George, 379 Perkins, Revere, 311, 323 Peña Cervel, M. Sandra, 325 Pérez Guerra, Javier, 380 Peters, Pam, 382 Peterson, Paul W., 97n, 99

Author index 419

Phillipps, K. C., 409n, 411 Picone, Michael, 341, 354 Pierrehumbert, Janet, 63, 76 Pintzuk, Susan, 2, 55, 56, 57, 60, 61, 62, 76, 78, 79 Pope, John C., 97n, 99, 103, 119, 153 Pounder, Amanda, 250, 263 Pratt, Lynda, 221, 229 Preacher, Kristopher J., 375, 384 Prévost, Sophie, 72, 78 Prince, Alan, 159, 178 Prince, Ellen F., 2, 10, 62, 65, 66, 79 Prins, Anton, 401, 402, 411 Q

Quirk, Randolph, 217, 266, 299 R

Raffel, Burton, 104, 123, 131, 132, 135, 136, 140, 141–143, 150n, 152 Rama Martínez, Esperanza, 380 Raumolin-Brunberg, Helena, 328, 336 Ray, John, 401 Reddaway, T. F., 239, 264 van Reenen, Pieter, 62, 77 Reinhart, Tanya, 64, 79 Renouf, Antoinette, 381 Rissanen, Matti, 325, 378 Ritt, Nikolaus, 381 Roberts, Jane, 83, 97n, 99 Rohdenburg, Günter, 213, 229, 230 Roland, Douglas, 312, 325 Romaine, Suzanne, 228, 378, 381 Rosenbach, Anette, 222, 229

Rothwell, William, 238, 264 Rowling, J.K., 121 Rudanko, Juhani, 368, 383 Ryder, Mary Ellen, 211, 230, 276, 296n, 299 S

Saintsbury, George, 4, 10 Samuels, M. L., 58, 60 Sand, Lori, 50, 51 Santorini, Beatrice, 72, 78 Scargill, Matthew H., 18, 54, 359, 381, 384, Schabram, Hans, 97n, 99 Scharill, Matthew H., 379 Schendl, Herbert, 381 Schilling-Estes, Natalie, 374, 385 Schlüter, Julia, 62, 79, 226n, 230, Schneider, Edgar W., 378 Schøsler, Lene, 62, 77, 320 Schultink, Henk, 235, 264 Scragg, Donald G., 97n, 99, Sheehan, Peter M., 51 Sheldon, Esther K., 388, 393, 409n, 412 Sheridan, Thomas, 328, 331, 332– 334, 336 Shuy, Roger, 325 Siegel, Dorothy, 201 Simon-Vandenbergen, A. M., 364, 384 Sisam, Kenneth, 97n, 100 Skeat, Rev. Walter W., 155, 156, 165, 174, 175, 176, 177, 178 Skousen, Royal, 272, 299 Slovic, Paul, 384 Smart, B.H., 391, 412 Smit, Ute, 384 Smith, Adam, 382

420 Author index

Smith, Jeremy J., 82, 100 Smith, Jennifer Anh-Tran, 3, 4, 101, 102–106, 121–152, 153– 154 Smith, M. Estellie, 51 Smithers, G. V., 5, 10, 101, 102, 113, 114, 115, 119, 187, 188, 189, 198, 201, 202n, 204, 205 Smith-Thibodeaux, John, 342, 354 Smitterberg, Erik, 368, 383 Sørensen, Knud, 266, 296n, 299 Spevak, Olga, 72, 79 Speyer, Augustin, 61, 62, 79 Sprockel, C., 81, 100 Stefanowitsch, Anatol, 226n, 228, 318, 324, 325 Steiner, Richard, 37, 39, 50, 53 Štekauer, Pavol, 296n, 299 Stevenson, Suzanne, 325 Stockwell, Robert, 27, 54, 118, 126, 152 Stone, Brian, 123, 134, 136, 144– 146, 152 Sundby, Bertil, 367, 384 Suzuki, Seiichi, 103, 119, 153 Svartvik, Jan, 299 Swift, Jonathan, 333, 387, 388, 389, 393, 395, 399

U

T

Ukaji, Masatomo, 215, 230 Upchurch, Robert K., 97n, 100

Tarlinskaja, Marina, 107, 119, 159, 178 Taavitsainen, Irma, 230 Taylor, Ann, 60, 61, 72, 78, 79, 80 Taylor, John, 313, 314, 326 Taylor, Kathi, 382 Thomas, Eric R., 359, 384 Thomason, Sarah Grey, 247, 264

Thompson, Sandra, 220, 302, 308, 311, 312, 313, 321n, 324, 326 Tieken-Boon van Ostade, Ingrid, 222, 230, 328, 336, 378 Tillery, Jan, 51 Tobin, Yishai, 230 Tolkien, J.R.R., 105, 119, 122, 123, 133, 135, 136, 137, 144– 146, 152 Tomascikova, Slavka, 299 Tomasello, Michael, 312, 317, 326 Trager, George L., 42, 54 Traugott, Elizabeth Closs, 2, 55, 56, 57, 62, 76, 80, 226n, 277, 298, 302, 304, 307, 311, 315, 320, 325, 326, 364, 370, 379, 384 Trépanier, Cécyle, 342, 354 Trousdale, Graeme, 6, 7, 207, 215–221, 222, 230, 301–326 Trout, Robert O., 341, 355 Trudgill, Peter, 32, 43, 54, 334, 336, 359, 371, 373, 374, 384, Tuggy, David, 255, 264 Tversky, Amos, 358, 384

V

Vachek, Josef, 36, 49, 54 Vallduví, Enric, 62, 80 Van Bergen, Linda, 328, 336 Vantuono, William, 123, 134, 135, 136, 139, 144–146, 152 Venezky, R., 324 Véronis, Jean, 275, 299

Author index 421

Vincent, Nigel, 324 Vincent, Thomas, 379 Vleeskruyer, Rudolf, 81, 82, 83, 97n, 100 Vogt, W. Paul, 374, 384 W

Wakelin, Martyn F., 402, 412 Walker, John, 332, 333, 336, 390, 391, 405, 407, 408, 409n, 412 Walker, Lorna E. M., 239, 264 Wallis, Sean, 375, 383 Wang, William S.-Y., 26, 51 Ward, Gregory, 62, 65, 66, 71, 76, 77 Warkentyne, H.J., 18, 54 Warner, Anthony, 60, 79, 364, 385, Watts, Richard, 336, 371, 385 Weinreich, Uriel, 11, 54 Wells, John C., 15, 48, 54 Wełna, Jerzy, 401, 413 Wenisch, Franz, 97n, 100 Whitton, Laura, 61, 80

Wikle, Tom, 50, 51 Wilson, 82 Wimsatt, W. K., 117, 119 de Wolf, Gaelan Dodds, 371, 385 Wolfram, Walt, 374, 385 Wood, J. David, 361, 385 Woods, Anthony, 374, 385 Wright, Laura, 410 Wright, Susan, 324 van der Wurff, Wim, 61, 76, 80, 379 Y

Yaeger, Malcah, 37, 39, 50, 53 Yerkes, David, 86, 100 Youmans, Gilbert, 158, 159, 161, 179 Youssef, Amani, 51 Z

Zacharski, Ron, 62, 64, 69, 77 Zelinsky, Wilbur, 42, 54 Zink, Gaston, 198, 204 Zwicky, Arnold, 225n

Subject index

A

Ælfric, 57, 63, 74, 76n, 81, 82, 86, 91, 94, 96n Catholic Homilies I, 72, 73, 74, 75, 76n, 96 Catholic Homilies II, 74, 75, 76n Lives of Saints, 74, 75, 76n, 96 Æthelwold, 81, 96n actuation, 22, actuation problem, 11 affix, 208, 209, 225, 231–237, 240–258 detachability of affixes, 254, 258, 259 age, 6, 208–209, 231–259 age coefficient, 29, 30, 31, 34, 46, 50n aggregation, 207, 208, 240, 241, 244, 245, 246, 248, 254, 255 Alfred the Great, 82 algorithm, 40, 330, 358 Alliterative Revival, 127 alliterative verse, 123, 124, 126, 135 Middle English –, 122, 126, 129, 133, 139, 150n Old English –, 122, 127, 131, 132 alliteration, 4, 103, 105, 125, 127– 135, 140–145, 153 American English (AmE), 11–54, 220, 302, 330–331, 358, 362, 363, 366–370

Americanization, 341 Atlas of Northern American English (ANAE), 2, 12–18, 20, 22, 24, 26, 28, 31–32, 34, 37, 39, 41, 44, 46, 48n anapest anapestic feet, 105, 107, 171, 172, 173, 174, 176 anapestic substitution, 107, 171 Anelida and Arcite, 158 animacy, 67, 76n Anglian dialect features, 3, 58, 59, 81, 83, 84, 86, 88 antecedence, 64, 65, 67 apocopation, 188, 189, 190, 205 articulatory ease, 31 Augustan period, 387, 393, 397, 398 authorship, 5, 155, 174, 176 B

bad data, 8, 327, 328, 330, 331, 334, 357–359, 363, 374 Beowulf, 4, 101, 102, 117, 121, 122–125, 130–133, 136, 137 bilingualism, 8, 327, 328, 329, 337–352 Blickling homilies, 84, 89, 90, 91, 92, 93 Boethius, 63 bob and wheel, 124 Bodleian MS Laud Misc.108, 115, 187

424 Subject index

borrowing, 231–255 British English (BrE), 9, 330–331, 358, 359, 362–363, 366–370, 395, 399 British National Corpus (BNC), 211, 222, 267, 269, 270, 274, 275, 276 Brown Corpus, 56, 268 business writing, 328, 337, 344 Byrhtferth of Ramsey, 94 C

caesura, 138 CAN, 9, 358, 364–372, 375–376 root possibility sense, 330– 331, 365–369 permission sense, 330–331, 364–368 Canadian English (CanE), 16, 327, 330–331, 357–379, Standard Canadian English, 359 Canadian Raising, 359 Canadian Shift, 13, 16, 17, 18, 19, 21, 22, 26, 39 case marking, 217, 307, 308, 311, 320n category, basic level, 219, 223, 316, 317 category strengthening, 221, 222, 302, 311 Catholicism, 340, 341, 342, 344, 348, 389 causality, chain of, 11–13, 46 Cawdrey, Table Alphabetical, 235 chain shifts, 12–16, 26 Chaucer, Geoffrey, 5, 101, 106, 112, 155–176, 256 Canterbury Tales, The, 158

Book of the Duchess, The, 107, 158, 176 House of Fame, 107, 158, 176 Romaunt of the Rose, 5, 106, 117, 155–176 Fragment A (RomA), 111, 155, 156, 163–176 Fragment B (RomB), 111, 155, 156, 163–176 Fragment C (RomC), 111, 155, 156, 157, 163–176 Troilus and Criseyde, 158 children, 278, 347, 348, 352 Prudhomme children, 343, 346 coding system for corpora, 62, 64, 65, 66, 71, 72, 76n Cognitive Grammar (CG), 7, 255, 316 coinage, 209, 210, 234–236, 248 colloquial language, 269, 270, 276, 346 colloquial pronunciation, 391 colonialism, 8, 327, 357, language of colonials, 340, 358 colonial varieties, 358, 363, 370, 371, 373 composition, 140 oral, 134 visual, 134, 135, 138 compound words, 278 Construction Grammar, 207, 215, 218, 221, 222, 312 (Grammatical Construction Theory), 217 constructional taxonomy, 301– 305, 308, 311, 313–315

Subject index 425

corpus, corpus linguistics, 260n, 331, 358, 373, 374 corpora, 55, 56, 57, 61, 62, 208, 222, 236, 265, 268, 279, 297– 298n, 309, 328, 329, 330, 362, 373–374 annotated, 56, 66 historical, 55, 56, 57, 208, 222, 233, 237, 265, 270, 328, 362–365, 373–374 ICAME corpora, 268, 269, 274 multilingual, 237, 247 parsed, 56, 57, 61, 72, 318 spoken, 311 A Representative Corpus of Historical English Registers, version 1 (ARCHER-1), 236, 330, 362, 365, 369, 372–373 Corpus of Early English Correspondence (CEEC), 268, 328 Corpus of Early Ontario English, pre-Confederation section (CONTE-pC), 330, 362, 365, 373, 377, 379n Corpus of Late EighteenthCentury Prose (CL18P), 212, 328, 362 Corpus of Nineteenth-century English (CONCE), 368, 373 Corpus of Scottish Correspondence, 328 Helsinki Corpus of Historical English Texts, 56, 232, 236, 241, 244, 247, 268 Penn Parsed Corpora of OE, ME and eModE, 221,

Network of Eighteenth-century English Texts (NEET), 328, 335 York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE), 57, 61, 62, 66, 72, 76n counts, see frequency creativity, self-conscious, 209, 221, 225n, 235, 250, 269, 272 creole society, 8, 329, 343, 345 D

derivational morphology, 6, 231– 259 borrowed derivational morphology, 231, 232, 246, 249, 254, 257, 258 diachrony, diachronic, 232, 235, 240–241, 252, 258, 276, 357, 366, 368–370 dialect geography, 18 regional English dialects, 373 provincial dialects, 341 dialect mixture, 56, 57, 81, 82, 83 diaries, 328, 330, 331, 332, 337, 345, 362, 366, 367, 368, 370, 377 Dictionary of Old English, 56, 96n Dictionary of Old English Corpus in Electronic Form, 85 diffuseness test, 175, 176 directional drift, 367, 368 double marking, 211, 213, 214, 215, 226n, 278, 279, 296n

426 Subject index E

-e, final, 105, 107 see also Middle English -er, 6, 207, 211, 213–215, 222, 224, 225, 265–297 Early English Books Online (EEBO), 237 Early Modern English (EmodE), 26, 218, 221, 225n, 232, 235, 302, 310, 359, 364, 371, 395 educated south-eastern English usage, 9, 392, 393, 395, 402, 405, 408 education, 329, 338, 346, 347, 348, 388, 390, 392, 396, 409n Eighteenth-century Collections Online (ECCO), 334 elicitation tests, 210 elision, 107, 113, 114, 115, 188, 189, 190, 198, 205 elocution, 332, 389–392, 409n elocutionists, 9, 328, 332, 391, 405, 408, 409n English fourteenth century, 187, 188, 198, 201, 232, 237, 238 fifteenth century, 6, 232, 237, 238, 239, 258, 272, 309 sixteenth century, 215, 272, 310, 332 seventeenth century, 215, 235, 272, 309, 310, 332, 390, 400 eighteenth century, 8, 328, 332, 359, 365, 367, 368, 387, 389, 392, 395, 405–406, 407 nineteenth century, 8, 9, 235, 272, 329, 359, 364–365, 368, 373, 401, 406–407

twentieth century, 272, 276, 279, 331, 332, 365, 373, 391, 406 twenty-first century, 276 English usage compared with Irish usage, 333, 395, 408 entrenchment, 316, 317, 320 Erie Canal, 40, 41, 42, 46, 47, 50n experiencer, 305, 306, 309, 318, 319, 321n Experiencer Construction (ExpCxn), 216, 217, 219, 220, 221, 305, 306, 307 F

foot, 107, 114, footed, 201, 202n also see iambic tetrameter French, 5, 8, 101, 198, 231, 237, 238, 245, 249, 327, 328, 329, 340, 345–348, 397 standard continental French, 341, 345, 346, 351 Acadian French, 337–338, 341–342, 351 frequency, 208, 210, 220, 221, 232, 234, 238, 244–245, 255, 318, 363, 373 token, 208, 210, 211, 240, 242, 255, 266, 270, 274, 363, 368 type, 208, 210, 220, 240, 241, 243, 251, 257, 270, 275, 278, 279, 312 focus, 62, 63, 64, 65, 66, 69, 70, 71, 73, 75 functional load, 23

Subject index 427 G

givenness, 62, 64, 69, 73 genre, 62, 74, 238, 245, 252, 260n, 330–331, 362, 363, 365, 366–369, 370, 377, 378n Goldsmiths’ Company, London, 239 Goldsmiths’ Mistery of London, Wardens’ Accounts and Court Minute Books of, 232, 239 Google, 210, 211, 214, 224, 225, 226n, 274, 275, 285 grammaticalization, 7, 207, 217, 218, 221, 222, 277, 278, 301– 304, 311, 314, 315–320, 364, 370 Great Vowel Shift, 26, 27 Grocers’ Company, records of, (see also London Grocers), 237, 245 H

Halle-Keyser Theory, 162 hapax legomena, 208, 210, 236, 240, 257, 266, 270 haplology, 213 Havelok the Dane, 5, 101, 114, 115, 116, 117, 187–200, 205– 206 headless line, 167, 176, 177n, 188, 190, 197 hemistich, 103, 104, 105, 125, 126, 138, 159, 161, 164, 167, 169, 170, 171 headless hemistich, 167 heuristics, 1, 8, 9, 55, 222, 330, 334, 363, 370

heuristic reasoning, 327, 330– 331, 357–358, 364, 367, 368, 371, 374 homilies, homiletic prose, 81, 83 anonymous homilies, 83, 84 horror aequi, 213 hybrid, 208, 236, 246–259, hypercorrection, 333, 334, 396, 398, 399 I

iamb, iambic phrase, 168 iambic pentameter, 117, 158, 168, 173, 174 iambic tetrameter, 101, 157– 160, 170, 173, 174, 176, 189 footed, 159 unfooted, 159 ICAME, see corpora immigration, 239, 360, 378n, immigrants, 22, 42, 340, 351, 360–363 impersonal construction, 7, 207, 215, 216, 217, 218, 220, 221, 301–305, 309–319 information structure, 61, 62, 63, 64, 74, 75 Inland North (IN), 19, 21, 37, 38, 39, 40, 41, 45, 46, 50n internet, 5, 207, 211, 213, 222, 223, 226n, 270, 274–279, 297n Irish usage, compared with English, 392, 393 Irish English, developments in the 19th century, 9, 334, 401, 406–408 /iw/ vs. /uw/ contrast, 32, 33, 34, 35, 36, 49n

428 Subject index J

Dr. Johnson, 333, 389, 393, 395, 396 K

Kaluza’s Law, 104, 153 Kentish, 84, 85, 87, 90, 91, 92, 93, 94, 95, 96 koiné, 43, 46 /Kuw/ class, 33, 34 L

Lancaster-Oslo-Bergen Corpus (LOB), 56, 268 language contact, 338, 341, 346, 360 language shift, 328–329, 338, 341, 352 Late Modern English (LModE), 8, 321n, 327–336, 357, 358, 359, 364, 365, 367, 368, 371, 373, 374 pronunciation of, 331–334 Latin, 72, 231, 237, 238, 316, 347, 349, 353n, 389 Latinate bases, 208, 235, 243 Left-dislocation, 61, 62, 65, 67, 70, 72 letters, 328, 330, 331, 332, 337, 349, 350, 352, 362–364, 366, 367, 370, 377 lexical isogloss, 43, 44, 45 linguistic propriety, 387, 388, 397 London Grocers, 6, 231, 232, 238, 239, 242–259 London, Port of, 238 long and ingliding subsystem, 17, 22

Louisiana, 8, 327, 328–329, 337– 353 low back merger, 17, 18, 29, 36, 42, 47, 48n lowering of /e/ before /r/, 393, 401–402 M

margins of security, 13, 15, 27, 37, 38 maximal dispersion, 14, 15, 18, 27, 47 MAY, 9, 358, 364–372, 375–376 Mercian features, 81, 82, 84, 85, 88, 90, 95, 96n, 97n meter, 3, 101–104, 121, 187–189, 196, 201n accentual-syllabic, 101 Middle English, 101, 102, 105 Modern English, 104 Old English, 101, 102, 104 methodology, 124, 153, 157, 234, 240, 244, 258, 265–266, 275, 331–334, 369–370, 373, 375 metrical beat, 101, 108, 110–113, 117 off-beat, 189, 191, 194 on-beat, 189, 194 silent beat dip, strong, 126, 128, 129, 133, 134, 144–146, 150n dip, unstressed, 213, 214 deviation, 102, 158–161, 169, 176 elision, 107, 113, 114, 115 evidence, 110, 155–174 foot, 107, 159, 160, 163, 164, 170

Subject index 429

lift, 125, 126, 127–132, 134, 135, 141–146 parameters, 161 position, 103, 104, 126, 131 prototype, 158, 159–161, 163, 165, 171, 176 resolution, 128, 161, 170, 171, 172 style, 160, 162, 166, 170 saliency, 104, 160, 161, 164, 168, 169 metrics, English historical, 101, 205 Middle English (ME), 25, 26, 58, 122, 124, 207, 215, 222, 231– 264, 309 French loan words in, 161, 190, 198, 199, 203n, 231, 238 genitives, 187, 206 inflectional endings, 188, 189–201 plurals, 187, 198, 202n, 206 verse, 136–140, 153 weak past participles final –ed(e), 5, 113, 187, 189, 190, 191, 192, 194, 196–201, 205, 206 final –es, 5, 113, 115, 197, 205 final –en, 113, 187, 189, 190, 205 final –e, 105, 113, 115, 187, 188, 189, 197, 198, 201, 205 Middle English Compendium, 237 Middle English Dictionary, 260n minimal pairs, 23, 24, 32 Missprachen, 58, 59

modal auxiliary verbs, 9, 302, 327, 330–331, 357–385 Modern English (ModE), 218– 220, 302, 311–319 monophthongization of /aw/, 22, 25, 49n monophthongization of /ay/, 15 morphosyntax, 207, 217 multiple inheritance, 220, 221, 321n N

nativization, 235 naturalisation, 6, 207, 208, 232– 237, 246–249, 254, 255, 258 neologism, 243, 244, 257, 260n newspapers, 330, 331, 362, 366, 367, 370, 377 nonperipheral track, 18, 20, 39 nominalization, 6, 208, 210, 211, 222, 223, 224, 225, 226n, 235, 237, 253, 265–297 North American English, see American English Northern Breaking, 39 Northern Cities Shift (NCS), 13, 21, 37, 38, 44, 45, 46, 48n O

Old English corpus, 6, 55, 59 Oxford English Dictionary (OED), 7, 207, 209, 210, 213, 221, 222, 226n, 259n, 260n, 269–280, 389 Old English, 25, 55, 58, 60–64, 66, 69, 72, 301, 305–307, 309, 317 anonymous homilies, 81, 82, 83, 84, 85, 86, 97n, 207, 215

430 Subject index

composite homilies, 95, 97n smoothing, 84, 87, 91, 93 syntax, 2, 55, 61, 64, 75, 207, 215–220, 301, 305–307, 319 verse, 126, 130–132, 153 Old French, loan words, 25 Ontario English (OntE), 9, 330, 357, 358, 359, 360, 362, 364, 368, 370–371, 373 Optimality Theory, 226n, Orm, 110 Ormulum, 159 Orosius, 74, 75, 76n, 305 orthoepists, 332 /ow/, fronting of, 28, 31, 33 P

particle, 211, 212, 213, 224, 226n, 265–297 Paston Letters, 212, 260n, 352, pattern, 71, 207, 210, 212, 215, 217, 265–280, 309, 314, 316, 366–367 coining pattern, 210, 212, 236 information pattern, 62, 74, prosodic pattern, 63 peripheral track, 17, 39, 45, personal writing, 328–329, 337, 344, 349 Piers Plowman, 117 phonemic length, 104 phonology, 235, English historical phonology, 333 phonotactic constraints, 191, 194, 196 Pittsburgh Shift, 13, 20, 21, 22, 26, 38 playfulness, 208–209, 225n, 231, 235, 259n

poetry, 104, 105, 112, 113, 116 poset (partially ordered set), 65, 66, 69, 71 pragmatic features, 57, 66, 303, 364 preposed dative experiencer (PDE), 309–311, 321n Present Day English (PDE), 62, 63, 64, 69, 72, 73, 231, 232, 235, 311–315, 330, 331, 363, 365, 368, 403, 407 PRICE lexical set, 405 probability matching, 15 productivity, 5, 208, 210, 225n, 232, 234, 235–258, 265, 266, 270, 274, 302, 311, 318 lexically restricted, 235, 236 local, 6, 207, 208, 233, 240, 241, 246, 248, 254 measure, 208, 236, 240 morphological, 236, 237, 248, 252 pronunciation features not proscribed, 405–406 prosody, 62, 71, 101, 108, 123, 189 Prototype Theory, 217 Prudhomme Family Papers, 8, 328–329, 337–355 pyrrhic, 162 R

r-less dialects, 15 r-pronouncing dialects, 15 Radical Construction Grammar, 302, 304 raising of /æ/, 38, 39, 44, 45, 393, 400, 406 raising of ME <ea>, lack of, 398

Subject index 431

regional accents, criticism of, 332, 387, 392 regional variation, 81, 82, 84 register, 74, 82, 210, 231, 232, 235, 238, 276 religion, 341 Rhetorical Grammar, 333, 388, 391, 396, 406 rhyme, rime, 153, 156, 157, 175, 176, 190 imperfect rhyme, 175, 176 rhythm, 153, 214 rhythmic alternation, 226n rhythmic template, 212, 213 rising middle classes, 391, 393 Riverside Chaucer, The, 156, 175 Roman de la Rose, 5, 155, 156, 157, 175 S

sample, 327, 328 random, 158 size, 158, 210, 237, 373, 374, 377 systematic, 157, 158 scansion, 124, 125, 128, 130, 153 rules of, 5, 125, 126, 128–129, 130–134, 138, 154, 187, 189, 199 schema, 307, 312, 313, 316, 317, 320 schematicity, 302, 304, 317 Sensitivity Principle of Trochees, 166 Sheridan, Thomas, 8, 9, 328, 331– 334, 388–413, Sheridan’s influence, 332, 334, 390–391, 409n

Sheridan’s system of pronunciation, 393–395 shift Back Chain Shift before /r/, 13 Back Upglide Shift, the, 13 Sievers, 103, 125, 126 Type A, 125 Type A3, 128 Type D, 125 Type E, 125 Sir Gawain and the Green Knight (SGGK), 4, 102, 121, 122, 124, 126, 136–140 smoothing of /aw/, 25, 26, 49n social networks, 8, 221, 222, 238, 259n, 260n, 327, 329, 337– 339, 342, 344, 353n density, 238, 329, 338, 339, 342, 343, 346, 352 multiplexity, 238–239, 329, 338, 339, 342, 343, 346, 352 network ties, family, 329, 342, 346 kinship, 329, 339, 343–344, 352 economic, 239–240, 329, 346, 352 social status, 342, 351, 352 SOFT lexical set, 405 Southern Shift, the, 13, 14 standard English, 334, 387, 393, 396, 400, 402 stress, 101, 124, 132, 153 linguistic, 101, 108, 112, 159 metrical, 111, 126 primary, 103, 212 secondary, 103, 125, 161 patterns, 110, 160, 161, 167, 172, 173

432 Subject index

doublets, 107, 161 peak, 162 phrasal, 110 sentential, 212 style Chaucerian, 176, non-Chaucerian, 169, 176 subject, 309, 311, 313–316, 318– 319, 320n, 321n subjecthood, 302, 311, 320n subsystems, 14, 15, 16, 18, 22, 26, 47, 48n suffix, 108, 208–210, 214, 225, 231–258 noun-forming, 207 supraregional varieties, 396, 401, 407 Swift’s rhymes, 395, 399 syllabicity, 187, 194, 200, 201, 205 syllable, 107, 110, 135 end-stressed syllable, 165 extrametrical syllable, 105, 107, 161, 169, 170, 171, 173, 176, 189, 198 missing syllable stressed syllable, 162 unstressed syllable, 128, 133, 134, 135, 144–146, 172, syncopation, 5, 114, 116, 187– 201, 205 syncope, 84, 87, 114, 190, 196, 205 syntax-pragmatics interface, 61 T

text types, 81, 261n, 363 topic, 62, 63, 64, 65, 67, 70, 71, 72

pragmatic topic, 64, 65, 67, 70, 71, 75, 76n syntactic topic, 64, 72, 75 topic persistence, 64, 70, 71, 73, 76n topicalization, 61, 62, 64, 65, 66, 72, 73 transitive construction (TrnCxn), 7, 207, 216, 219–221, 301, 302, 304, 308, 311, 313, 314– 318, 320n, 321n Type N, 216–218, 220, 226n, 306–309, 317, 319 Type T, 216–220, 307–309, 317, 321n transitivity, 302, 308, 311–312, 315 translation, 81, 86, 92, 96n, 101, 103–104, 121–140, 153, literal, 139 poetic, 121, 139 personal, 139 translatorship, 157, 159, 161 triggering event, 2, 11–15, 19, 22, 23, 46, 47, 48n triple marking, 213, 222, 226 trochee, 108, 111, 162, distribution of trochees, 107, 163, trochaic makeup, 107, 161, 165, 176, trochaic substitution, 105, 107, 112, 161, 162, 163, 164, 176, trochees of disyllables, 166, trochees of monosyllables, 162, 166 /Tuw/ class, 33, 34

Subject index 433 U

/uw/, fronting of, 27, 28, 29, 47 V

verb, impersonal, 216, 220 intransitive, 267, 272, 273 multi-word, 265–270, 272, 274–279, 297n phrasal, 207, 211, 212, 223, 224, 226n, 267, 272, 273, 277, 278, 296n prepositional, 207, 211, 223, 224, 226n, 267, 272, 273, 277 transitive, 220, 267, 272, 273 weak, 114, 115 verse, 105 a-verse, 105, 133, 153 b-verse, 105, 133, 153 on-verse, 125–131, 133–135, 141–145 off-verse, 125, 127–131, 133, 135, 141–145 hypermetric verses, 128, 131 Vercelli homilies, 84, 88, 89, 90, 92, 93, 94, 96, Vespasian Psalter, 90, 96n violations, positional, 4, 127–129, 131, 132, 141–148, alliterating, 4, 127, 128, 131– 132, 134, 141–148

vowels, North American English vowels, 16, 24, 32, 40 short vowels, 15, 18, 19, 47, 393, 395, 402 upgliding vowels back upgliding, 15, 16, 34, 35 front upgliding, 14, 15, 16 vowel lengthening before voiceless fricatives, 402 W

weighted score, 127, 129, 134 West Saxon, Late, 3, 57, 58 West Saxon Gospels, 74, 75, 76n Whitbread Book of the Year, 121 word formation, 235, 236, 240, 250, 251, 253, 255, 261n, 266, 267, 269, 270, 272, 296n word order, 72, 75, 218, 316, 320n OSV, 72, 73, 74, 75 OVS, 72, 73, 74, 75 word stress, 208, 210, 226n Wulfstan, 90, 93, 94, 95 Y

Yiddish Movement, 66, 71, 73, 76n /yuw/ class, 31, 32, 36, 47