Handbook of Perceptual Dialectology ~ Volume 1

HANDBOOK OF PERCEPTUAL DIALECTOLOGY

HANDBOOK OF

PERCEPTUAL DIALECTOLOGY VOLUME

1

Edited by

DENNIS R. PRESTON Michigan State University

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM / PHILADELPHIA

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences — Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984.

Library of Congress Cataloging-in-Publication Data Handbook of perceptual dialectology / edited by Dennis R. Preston. p. cm. Includes bibliographical references (p. ) and index. 1. Linguistic geography. 2. Speech perception. 3. Language and languages--Variation. 4. Language awareness. I. Preston, Dennis Richard. P375.H28 1999 417’.2--dc21 98-25334 ISBN 90 272 2180 4 (Eur.) / 1 55619 534 6 (US) (Vol. 1; alk. paper) CIP © 1999 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Company • P.O.Box 75577 • 1070 AN Amsterdam • The Netherlands John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA

This book is respectfully dedicated to Willem Grootaers, Takesi Sibata, and Antonius Weijnen — the pioneers of perceptual dialectology

Table of Contents

List of Figures

ix

List of Tables

xv

Preface William A. Kretzschmar, Jr.

xvii

Acknowledgments

xix

Introduction Dennis R. Preston

xxiii

P I The Dutch Contribution: ‘Little Arrows’ 1.

Informant Classification of Dialects W. G. Rensink

3

2.

Dialects Jo C. Daan

9

3.

The Netherlands-German National Border as a Subjective Dialect Boundary Ludger Kremer

31

P II The Japanese Controversy: ‘Subjective’ and ‘Objective’ 4.

Consciousness of Dialect Boundaries Takesi Sibata

39

5.

Consciousness of Linguistic Boundaries and Actual Linguistic Boundaries Kikuo Nomoto

63

6.

Dialect Consciousness and Dialect Divisions: Examples in the Nagano-Gifu Boundary Region Yoshio Mase

71

7.

On Dialect Consciousness: Dialect Characteristics Given by Speakers Yoshio Mase

101

8.

The Discussion Surrounding the Subjective Boundaries of Dialects Willem A. Grootaers

115

viii 9.

TABLE OF CONTENTS

On the Value of Subjective Dialect Boundaries Antonius A. Weijnen

10. Dialects and the Subjective Judgments of Speakers: Remarks on Controversial Methods A. C. M. Goeman

131

135

P III Images, Perceptions and Attitudes 11. Classification of Dialects by Image: English and Japanese Fumio Inoue

147

12. Subjective Dialect Division in Great Britain Fumio Inoue

161

13. Geographical Perceptions of Japanese Dialect Regions Daniel Long

177

14. Mapping Nonlinguists’ Evaluations of Japanese Language Variation Daniel Long

199

15. The Perception of Post-Unification German Regional Speech Jennifer Dailey-O’Cain

227

16. Variation and the Norm: Parisian Perceptions of Regional French Lawrence Kuiper

243

17. The Perception of Turkish Dialects Mahide Demirci and Brian Kleiner

263

18. Regional Variation in Subjective Dialect Divisions in the United States Donald M. Lance

283

19. A View from the West: Perceptions of U. S. Dialects by Oregon Residents Laura C. Hartley

315

20. “Welshness” and “Englishness” as Attitudinal Dimensions of English Language Varieties in Wales Nikolas Coupland, Angie Williams and Peter Garrett

333

21. Dialect Recognition Angie Williams, Peter Garrett and Nikolas Coupland

345

22. A Language Attitude Approach to the Perception of Regional Variety Dennis R. Preston

359

References

375

A Perceptual Dialectology Bibliography

393

About the Contributors and Translators

403

Index

409

List of Figures

Introduction Figure 0.1. Three approaches to language data Figure 0.2. Westernmost section of the North Brabant, showing traditional isogloss boundaries (thick lines) and the “little arrows” of respondent similarity perceptions Figure 0.3. Westernmost section of the North Brabant, showing traditional isogloss boundaries (solid thick lines) and respondent similarity perception boundaries (shaded thick lines) C 1 Figure 1.1. Respondent classifications of Dutch dialects Figure 1.2. Sites in The Netherlands referred to in Chapter 1 C 2 Figure 2.1. Sites in The Netherlands referred to in Chapter 2 Figure 2.2. Dutch dialect areas, perceptual and production data combined, with increasingly darker areas showing greater divergence from “standard” Dutch C 3 Figure 3.1. Subjective dialect boundaries in the western part of Westphalia C 4 Figure 4.1. How individual responses and boundaries were plotted Figure 4.2. Subjective dialect groups 1, 10, 11, 14, and 18 Figure 4.3. Subjective dialect groups 2, 5, 12, 15, 16, and 19 Figure 4.4. Subjective dialect groups 3, 6, 9, and 17 Figure 4.5. Subjective dialect groups 4, 7, 8, and 13 Figure 4.6. The nineteen subjective dialect areas Figure 4.7. Respondent identification of areas that are “difficult to understand” Figure 4.8. Boundaries of the areas that are “difficult to understand” Figure 4.9. Post-1901 administrative divisions Figure 4.10. Some traditional dialect isoglosses

x

LIST OF FIGURES

C 5 Figure 5.1. Perceptions of linguistic similarity (Category 1) Figure 5.2. Actual linguistic isoglosses Figure 5.3. Perceptions of specific linguistic boundaries Figure 5.4. Birthplace of spouse and years of education Figure 5.5. Perceptions of linguistic differences Figure 5.6. Perception of dialect boundaries C 6 Figure 6.1. The subjective dialect boundaries indicated by a respondent at Hiwada (57) Figure 6.2. The subjective dialect boundaries indicated by a respondent at Kohiwada (58) Figure 6.3. The subjective dialect boundaries indicated by a respondent at Nashinokidaira (7) Figure 6.4. The subjective dialect boundaries indicated by a respondent at Site #6 (Inekoki) Figure 6.5. Names (in Japanese) and numbers of the fifteen localities that make up the village of Nagawa; the subjective limits indicated by the respondents (between Sites #14 and #5) divide Nagawa into regions “d” and “e” (as shown in Figure 6.6) Figure 6.6. The sixteen dialect perception areas (based on responses to questions A and B) Figure 6.7. Nagano respondents’ responses to question D Figure 6.8. Gifu respondents’ responses to question D Figure 6.9. Nagano respondents’ responses to question C Figure 6.10. Gifu respondents’ responses to question C Figure 6.11. Integrated perception map Figure 6.12. Correspondence of the dialect perception areas with linguistic features Figure 6.13. Placement of isoglosses Figure 6.14. Vocabulary isoglosses. Figure 6.15. Grammatical isoglosses Figure 6.16. Pitch accent isoglosses Figure 6.17. Summary of the relationships among regional varieties C 7 Figure 7.1. Survey locations Chapter 8 Figure 8.1. Subjective dialect boundaries (Shimo-kita Peninsula) (left insert: subjective dialect studies in Japan) Figure 8.2. School districts (with administrative and selected subjective boundaries) C 10 Figure 10.1. Perception of similarity among dialects (enlarged portion of Goeman 1989, page 543 [from Willems 1886]; compare Figures 0. 2 and 0. 3)

LIST OF FIGURES

xi

C 11 Figure 11.1. Dialect classification by image using Hayashi 3 — Japan Figure 11.2. Dialect classification using cluster analysis of standard Japanese forms — Japan Figure 11.3. Distribution of evaluative words as a result of Hayashi’s Quantificational Theory Type 3 for British students (first and second axes) Figure 11.4. Distribution of dialects as a result of Hayashi’s Quantificational Theory Type 3 for British students (first and second axes) Figure 11.5. Average values of standardness and ruralness for British and Australian students (including their own dialects) Figure 11.6. Geographical distribution of students’ own dialect image (average value of standardness projected on dialect division map in Brook 1963) Figure 11.7. Dialect image and its environments C 12 Figure 12.1. Example of student data (student number 041) Figure 12.2. Example of data input (student number 041) Figure 12.3. Subjective dialect division of Great Britain Figure 12.4. Distribution table of Great Britain dialects as a result of Hayashi 3 Figure 12.5. Distribution of dialects as a result of Hayashi 3 (England only) Figure 12.6. Dialect division by Viereck (1986) Figure 12.7. Dialect division by Trudgill (1990) C 13 Figure 13.1. Hand-drawn map Figure 13.2. Tally map for Tôhoku-ben Figure 13.3. Gradient perceptual dialect map of “Tôhoku dialect,” Aichi informants (n = 60) Figure 13.4. Gradient perceptual dialect map of “Tokyo dialect,” Aichi informants (n = 52) Figure 13.5. Gradient perceptual dialect map of “Kansai dialect,” Aichi informants (n = 57) Figure 13.6. Composite map of perceptual dialect regions, Aichi informants (n = 70) Figure 13.7. Composite map of perceptual dialect regions, Okayama informants (n = 20) Figure 13.8. Gradient map of the perceptual area of “standard,” Kansai informants (n = 74) Figure 13.9. Gradient map of the perceptual area of “Kansai dialect,” Kansai informants (n = 195) Figure 13.10. Gradient map of the perceptual area of “Kansai dialect,” Kanto informants (n = 62) Figure 13.11. Production dialect divisions (Misao Tôjô) Figure 13.12. Gradient map of the perceptual area of “Kawachi dialect,” Kinki survey informants (n = 117) C 14 Figure 14.1. Kanto informants’ “Most Pleasant” Figure 14.2. Aichi informants’ “Most Pleasant”

xii Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure

LIST OF FIGURES

14.3. Eight-region average “Most Pleasant” 14.4. Kanto informants’ “Most Pleasant” (standardized scores) 14.5. Aichi informants’ “Most Pleasant” (standardized scores) 14.6. Gifu informants’ “Most Pleasant” (standardized scores) 14.7. Kanazawa informants’ “Most Pleasant” (standardized scores) 14.8. Kansai informants’ “Most Pleasant” (standardized scores) 14.9. Hiroshima informants’ “Most Pleasant” (standardized scores) 14.10. Fukuoka informants’ “Most Pleasant” (standardized scores) 14.11. Kagoshima informants’ “Most Pleasant” (standardized scores) 14.12. Kansai informants’ “positive characteristics” 14.13. Kansai informants’ “negative characteristics” 14.14. Kanto informants’ “positive characteristics” 14.15. Kanto informants’ “negative characteristics” 14.16. Kanto and Kansai informants’ “positive characteristics” (t test) 14.17. Kanto and Kansai informants’ “negative characteristics” (t test)

C 15 Figure 15.1. Map of thirty-six German regions rated for “correctness” and “pleasantness” Figure 15.2. Northwesterners’ mean “correctness” ratings for each individual region Figure 15.3. Northeasterners’ mean “correctness” ratings for each individual region Figure 15.4. Central Westerners’ mean “correctness” ratings for each individual region Figure 15.5. Central Easterners’ mean “correctness” ratings for each individual region Figure 15.6. Southwesterners’ mean “correctness” ratings for each individual region Figure 15.7. Westerners’ and Easterners’ perceptions of the “correctness” of Western and Eastern varieties Figure 15.8. Northwesterners’ mean “pleasantness” ratings for each individual region Figure 15.9. Northeasterners’ mean “pleasantness” ratings for each individual region Figure 15.10. Central Westerners’ mean “pleasantness” ratings for each individual region Figure 15.11. Central Easterners’ mean “pleasantness” ratings for each individual region Figure 15.12. Southwesterners’ mean “pleasantness” ratings for each individual region Figure 15.13. Westerners’ and Easterners’ perceptions of the “pleasantness” of Western and Eastern varieties Figure 15.14. Map of smaller regions included in the general region of “Bavaria” Figure 15.15. Map of smaller regions included in the general region of “Central East” Figure 15.16. Boundaries of all regions drawn by informants C 16 Figure 16.1. The administrative divisions of France Figure 16.2. Major dialect areas of France Figure 16.3. Composite map of respondents’ hand-drawn perceptual maps Figure 16.4. Multidimensional scaling of correctness ratings with five K-means clusters Figure 16.5. Multidimensional scaling of correctness ratings with four K-means clusters

LIST OF FIGURES

xiii

Figure 16.6. Multidimensional scaling of degree-of-difference ratings with six K-means clusters Figure 16.7. Multidimensional scaling of pleasantness ratings with five K-means clusters Figure 16.8. Multidimensional scaling of pleasantness ratings with four K-means clusters C 17 Figure 17.1. Correctness task Figure 17.2. Pleasantness task Figure 17.3. Difference task Figure 17.4. Mean score, by gender, for correctness and pleasantness tasks Figure 17.5. Mean score, by age, for correctness and pleasantness tasks Figure 17.6. Mean score, by gender and age, for correctness task Figure 17.7. Mean score, by gender and age, for pleasantness task Figure 17.8. Mean scores for eastern cities, by group, for correctness task C 18 Figure 18.1. New England and Northeastern dialect areas Figure 18.2. Northern dialect area Figure 18.3. Midwest dialect area Figure 18.4. Southern dialect area Figure 18.5. Southeastern dialect area Figure 18.6. Appalachian dialect area Figure 18.7. Upper South dialect area (all students) Figure 18.8. Mid-Southern dialect area (all students) Figure 18.9. Southwestern dialect area (all students) Figure 18.10. Texas dialect area (all students) Figure 18.11. Western dialect area Figure 18.12. Northwestern dialect area Figure 18.13. West Coast dialect area (all students) Figure 18.14. Examples of individual maps Figure 18.15. Major dialectal areas from Linguistic Atlas studies Figure 18.16. Dialect layers based on data from the files of the Dictionary of American Regional English Figure 18.17. Core areas of regional dialects (all students) C 19 Figure 19.1. Results of hand-drawn maps of a “South” Figure 19.2. Results of hand-drawn maps of a “Northeast” Figure 19.3. Results of hand-drawn maps of a “Midwest” Figure 19.4. Results of hand-drawn maps of a “West” Figure 19.5. Multidimensional scaling analysis of degree-of-difference ratings with K-means clusters Figure 19.6. Multidimensional scaling analysis of correctness ratings with K-means clusters

xiv

LIST OF FIGURES

Figure 19.7. Multidimensional scaling analysis of pleasantness ratings with K-means clusters Figure 19.8. Oregonian perceptual speech regions summarized C 20 Figure 20.1. Language and identity in Wales, based on “The Three Wales Model” (Balsam 1985) Figure 20.2. Williams’s (1985) perceptual culture regions from a sample of schools all over Wales, with isolines showing percentile levels of agreement about whether towns were Welsh-speaking Figure 20.3. Wales with the six focal towns/cities and neighboring English urban areas C 21 Figure 21.1. Map of Wales showing places mentioned in this chapter Figure 21.2. Variable independent continua of the modes of nonlinguists’ language awareness, with hypothetical settings C 22 Figure 22.1. A Michigan hand-drawn map Figure 22.2. Computer-assisted generalizations of hand-drawn maps showing where Southeastern Michigan respondents believe speech regions exist in the United States Figure 22.3. Means of ratings for language “correctness” by Michigan respondents for U. S. English Figure 22.4. Means scores for “pleasant” English by Michigan respondents

List of Tables

Chapter 5 Table 5.1. Distribution of survey items C 7 Table 7.1. Tabulation of informant comments divided by linguistic category and location to which they refer C 13 Table 13.1. List of all dialect names used Table 13.2. Percentage of informants who gave dialect names for a region C 14 Table 14.1. Number of informants who gave “characteristics” for speech regions C 16 Table 16.1. Means, by region, of the three rating tasks C 19 Table 19.1. Frequency of identification of dialect regions C 20 Table 20.1. Generalized summary of judgments of the accent/dialect communities (scales data) Table 20.2. Frequencies of categorized characterizations, by labeled dialect areas C 21 Table 21.1. Percentages of correct recognition of the community origins of the speakers Table 21.2. Percentage of young adults who achieved correct recognition of where speaker was from Table 21.3. Mean scores of young adult and teacher judges on the item, “How Welsh do you think this speaker sounds?”

xvi

LIST OF TABLES

C 22 Table 22.1. The two factor groups from the ratings of all areas Table 22.2. Means scores of attributes Table 22.3. Factor group means (all attributes combined)

Preface

The general view would have it that the linguistics of speech should be concerned with what people actually say. Speech production surely has had the majority of the attention in empirical linguistics, whether under quantitative or qualitative analysis. However, there is not only room for study of the reception of speech, there is the necessity that we study it. Constraints on what we say are not only determined by accident of birth but to some degree are a matter of choice. We choose our words according to how we perceive them or how we believe that others will perceive them. Every conversation is to some extent an exercise in such psychological brinksmanship. And to understand more fully the words that people actually produce, we therefore need to understand how people perceive those words. Empirical linguistics here meets psychology, whether the social psychology of groups or the individual psychologies of the participants in a conversation, in that what people actually say is bound up with how people perceive and understand their choices in what to say. Dennis Preston has done the field of empirical linguistics great service in his earlier work on perceptual dialectology, both to raise our consciousness of the phenomenon and to document some facts about the perception of English varieties. Now he has done it again in the Handbook of Perceptual Dialectology, to expose the foundations of the study of perceptual dialectology and to extend our knowledge of it around the world. Especially in the Netherlands and in Japan, early work in the field looked for correspondences between speech production, dialect boundaries based on evidence of what people actually said, and speech perception, where people thought the dialect boundaries might be. The two were seldom found to correspond, and thus the authors of the early papers and their successors found it necessary to measure the degree of correspondence between evidence of production and perception with empirical methods parallel to those used for investigation of speech production and to consider possible reasons for the mismatch of production and perception. This mismatch, I would argue, is one of the most important basic facts about language, and its discovery is one of the most important findings of modern empirical linguistics. Where should we locate the basic agreement of speakers about what belongs in their language and what does not? Perceptual dialectology reveals that we each have less a contract with other speakers and more a set of cultural (or other kinds of) assumptions about our interlocutors, less an agreement among ourselves and

xviii

PREFACE

more a shared experience of incompletely fulfilled expectations. The mismatch between perception and production, which the Handbook documents in many of the world’s languages, might be seen as subversive to our intuitive sense of linguistic systems. It does more to throw us back on the solipsism of idiolect than any other single aspect of language — and yet it also requires each speaker to escape from merely personal idiosyncrasy to form impressions of the speech of others and, presumably, to act on the most salient judgments in everyday conversation. The notion of perceptual dialectology greatly deepens our understanding of the dynamics of speech communities. The Handbook of Perceptual Dialectology offers a detailed record of the development of the field and its modern extension. Of particular interest for empirical linguistics are the different methods that have been employed to document perceptions, for subtle differences in the questions asked of speakers can apparently have far-reaching effects in the results. And some of the earliest articles turn out still to be the most impressive in their grasp of the issues of speech perception in relation to speech production. The Handbook is a reference book, but its chapters also read well seriatim. It will stand as a basic text in empirical linguistics for years to come.

William A. Kretzschmar, Jr. University of Georgia

Acknowledgments

The following are gratefully acknowledged for their permission to reprint the previously published material included here: De Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam and the author for Figures 0.2 and 0.3 (Antonius A. Weijnen, 1946. “De Grenzen Tussen de Oost-Noordbrabantse Dialecten Onderling.” In Antonius A. Weijnen, J. M. Renders, and Jac. van Ginneken (eds), “Oost-Nordbrabantse Dialectproblemen.” Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wettenschappen te Amsterdam 8:1–15. Mededelingen der Centrale Commissie voor Onderzoek van het Nederlandse Volkseigen, for Chapter 1 (W. G. Rensink, 1955, “Dialectindeling Naar Opgaven van Medewerkers” in Amsterdam Dialectbureau Bulletin 7:20–23). Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wettenschappen te Amsterdam and the author, for Chapter 2 (Jo Daan, 1970, Dialekten; Jo Daan and D. P. Blok, Van Randstad tot Landrand, #37:9–43). Nedersaksische Studies and the author, for Chapter 3 (Ludger Kremer, 1984, “Die Niederländisch-Deutsche Staatgrenze als Subjektive Dialektgrenze.” In Grenzen en Grensproblemen. Een bundel studies uitgegeven door het Nedersaksisch Instituut van de R. U. Groningen ter gelegenheid van zijn 30–jarig bestaan (essentially the same as in Nedersaksische Studies 7; zugleich: Driemaandelikse Bladen 36). Gengo Kenkyû and the author, for Chapter 4 (Takesi Sibata, 1959, “Hôgen Kyôkai no Ishiki,” 36:1–30). Jinruikagaku and the author, for Chapter 5 (Kikuo Nomoto, 1963. “Kotoba no Ishiki no Kyôkai to Jissai no Kyôkai, 15:271–81). Tokyodô and the author, for Chapter 6 (Yoshio Mase, 1964, “Hôgen Ishiki to Hôgen Kukaku“ (Studies presented to Professor Misao Tôjô on his 80th birthday). Tokyo, 270–302. Nagano-ken Tanki Daigaku Kiyô and the author, for Chapter 7 (Yoshio Mase, 1964, “Hôgen Ishiki ni Tsuite: Washa no Genkyûshita Hôgenteki Tokuchô,” 18:1–12). Orbis, for Chapter 8 (Willem Grootaers, 1964, “La Discussion Autor des Frontières Dialectales Subjectives” in Orbis 13:380–98).

xx

ACKNOWLEDGMENTS

Elsevier Science BV, for Chapter 9 (Antonius A. Weijnen, 1968, “Zum Wert Subjektiver Dialektgrenzen” in Lingua 21:594–96). Espaces Romans and the author for Chapter 10 (A. Goeman, 1989, “Dialectes et Jugements Subjectifs des Locuteurs. Quelques Remarques de Méthode a Propos D’une Controverse” in Espaces Romans (études de dialectologie et de géolinguistique offertes à Gaston Tuaillon, Vol. II). Université Stendhal — Grenoble 3: Ellug, 532–44). Zeitschrift für Dialektologie und Linguistik and the author, for Chapter 11 (Fumio Inoue, 1995, Classification of dialects by image — English and Japanese. In Wolfgang Viereck (ed.), Verhandlungen des Internationalen Dialektologenkongresses, Bamberg, 29.7. — 4.8.1990; 4 vols. Stuttgart: Steiner, 75–77:355–68). The University of Alabama Press, for Chapter 12 (Fumio Inoue, 1996, “Subjective Dialect Division in Great Britain” in American Speech 71(2):142–61). Dialectological Circle of Japan and the author, for parts of Chapter 13 (Danny Long, 1990, “Hôgen Ninchi Chizu no Kakikata to Yomikata” in Proceedings of the 50th meeting of the Dialectological Circle of Japan 50:7–16. Les Presses de L’Université Laval, for parts of Chapter 13 (Daniel Long, 1993, “The Role of Linguistic Features in Perceptual Dialect Regions” in Proceedings of the XVth International Congress of Linguists 371–74). In addition, I am indebted to a very large number of people for their help in the completion and even the inspiration for this volume. Jan Berns, then Associate Director, Ton Goeman, Harrie Scholtmeijer, and other members of the P. J. Meertens Institute in Amsterdam answered some of the dumbest questions ever asked about Dutch. I apologize for my ignorance and am very grateful for their patience and expert guidance (which extended even to the “administrative” concerns of this volume). Jan and Jaap van Marle, then Director of the Institute, invited me to give one of the plenary lectures at the Second International Congress of Dialectologists and Geolinguists in Amsterdam in July-August, 1997, where I was lucky enough to be able to discuss his important early work in variety perception with Professor Antonius Weijnen personally. I will be forever grateful to them for that opportunity. A note of thanks to the skilled and careful translators for this work — Karen Bouwer (Dutch), Betsy Evans and Larry Kuiper (French), Jennifer Dailey-O’Cain (German), and Midori Yonezawa (Japanese); a special expression of gratitude to my good friend Danny Long, who translated two of the chapters here from Japanese, kept me from making foolish mistakes about that tongue (one in which even my own family will attest I have only the barest of competences), and assisted in the preparation of this volume in numerous ways. Finally, even my German (which I am supposed to know!) got an occasional nudge from my colleague Tom Lovik at Michigan State University, to whom I am also grateful. While I am on this personal note, I will point out that the studies presented in Part III of this volume nearly all come from scholars who have held my work in enough regard to carry it on, developing new approaches and methods and/or applying older ones to new

ACKNOWLEDGMENTS

xxi

areas. I am honored by their confidence; I hope that this area of investigation has added in at least some small way to a sense of accomplishment in their linguistic research careers. I am very fortunate to be able to say that several joint enterprises in addition to the preparation of this volume have turned this group of scholars into friends as well as colleagues — Nik Coupland, Jennie Dailey-O’Cain, Mahide Demirci, Peter Garrett, Ton Goeman, Laura Hartley, Fumio Inoue, Brian Kleiner, Larry Kuiper, Don Lance, Danny Long, and Angie Williams. We’ve never all been in one room, but when subsets of us have, it ain’t all been work. Although I am flattered by the acknowledgment of my work in Part III, as can be seen in the first parts of this volume, it is, in fact, a continuation of a long and productive tradition (much of which, sadly, I was ignorant of when I began this research initiative). I hope the inclusion (and translation) here of some of this previously published work will expose it to a more general public and serve to strengthen what I believe can be an important contribution to an ever-expanding interdisciplinary approach to language structure, use, and regard. I have been lucky enough to meet Professors Grootaers, Mase, Sibata, and Weijnen and correspond with Professor Daan from that earlier group of distinguished scholars, and my own work has been inestimably enriched by that contact. Finally, as readers of nearly everything else I have written will already know, who else but Carol Preston read every word of this work, straightened out the prose, and as always, gave the best advice and comfort possible.

Introduction Dennis R. Preston

Why a “perceptual dialectology”?1 There are many, many ways to skin a linguistic cat, but let’s assume here that there are basically three: a, b, and c (see Figure 0.1). At the top of this triangle sits the vast majority of research on language. What people actually say not only provides ethnographic, conversational, and other studies of language performance with their raw data but also, by exposing how different groups say things differently, feeds (at least) historical linguistics, linguistic geography, and sociolinguistics. Lurking behind a, however, is a′; linguists seek not only to classify language use but also to account for it by determining the cognitive, social interactive, geographical, and other forces and conditions that explain its acquisition, shape, distribution, change, and employment. That ought to be enough.

Figure 0.1. Three approaches to language data

xxiv

INTRODUCTION

But sociolinguists, ethnographers, social psychologists, and anthropologists (among others) notice that some of the underlying explanations at a′ are related to the social regard that groups and individuals have for one another and that such regard seems to be strongly related to language use. They have fashioned, therefore, a separate approach to language study that has come to be known generally as the investigation of “language attitudes,” situated in the triangle shown in Figure 0.1 at b. Like a, lurking behind such attitudes is an explanatory b′, which touches on facts such as historical relations among groups, psychosociological associations (and the mechanisms that account for them), and a host of other values, beliefs, and cultural stereotypes. Finally, although the tradition may be as old as Polle (1889), the c corner of the triangle was more recently defined for and recommended to modern linguists by Hoenigswald (1966): “[W]e should be interested not only in (a) what goes on (language), but also in (b) how people react to what goes on (they are persuaded, they are put off, etc.) and in (c) what people say goes on (talk concerning language)” (20). What people say about what goes on (and what lies behind their statements), the c and c′ of the above triangle, is the stuff of folk linguistics, and perceptual dialectology is a subbranch of that general area of investigation.2 I provide here only minimum justification for studying folk concepts, and it surely requires little justification for social scientists in general. What nonspecialists believe about any technical field has at least the following general values: 1.

2.

3.

Such beliefs are a part of the folklore, ethnography, and cultural anthropology of groups; studying them is justified along this dimension quite independent of the separate scientific field to which the beliefs are related. There may be interaction as well as simply contrast between folk belief (and practice) on the one hand and scientific or specialist knowledge on the other. For example, folk medicines have been found in the laboratory to be extremely effective. When folk knowledge relates to social facts, however, the interaction may be defining as well. The social-psychological truth of such interaction is so profound that it has become a part of our own folk belief: For example, “If children believe that they will not succeed, they won’t.” In such cases, knowledge of the folk belief is explanatory, scientific knowledge about the phenomenon under investigation. It seems obvious that instances of language change and so-called language attitudes, to take examples from the a and b corners of the triangle respectively, might be profoundly influenced by folk beliefs about language, particularly beliefs about the status of language varieties and the speakers of them. Finally, although not at all least consequential, even when there is considerable contrast between scientific and folk information and when some scientists may find little of value in the folk facts, those who labor in applied fields will want to know what nonspecialists believe if they plan to intervene successfully. This is perhaps most important in language teaching — native as well as second or foreign — especially when different languages and/or varieties are held in different regard by

INTRODUCTION

xxv

members of speech communities. To lack this information is surely as debilitating as a physician’s not knowing the folk terms for and/or beliefs about diseases. Perceptual dialectology, then, represents the dialectologist’s-sociolinguist’s-variationist’s interest in folk linguistics. What do nonspecialists have to say about variation? Where do they believe it comes from? Where do they believe it exists? What do they believe is its function? Once linguists are in possession of these facts (satisfying Condition 1 above), they may go on to compare scientific and folk characterizations of dialect areas (or social class varieties or others), discovering, for example, that there may be perceived folk dialect areas where there are none scientifically and vice versa (satisfying Condition 2 above). Finally, a more intelligent approach to instruction, materials, teacher education, language and law, language and medicine, and a number of other applied matters that touch on language diversity may be taken once the folk as well as the scientific facts are known (satisfying Condition 3 above). Once it has been decided that the folk facts of dialectology are worth knowing, it remains only to visit the folk and find them out. This book provides detailed accounts of how that has been done and the results of doing it. It, therefore, aims to provide the following for perceptual dialectology: – – – – –

A historical survey A regional survey, adding to the earlier preponderance of studies in Japan, the Netherlands, and the United States A methodological survey, showing, in detail, how data have been acquired and processed An interpretive survey, showing how these data have been related to both linguistic and nonlinguistic facts A comprehensive bibliography

In fulfilling these major aims, articles that have appeared in journals not accessible to a majority of scholars or in languages that may not be a part of every scholar’s competence are made available. Although the major organization of the book is historical, there is some skipping around to keep regional and/or methodological concerns together.

1.

Part I

In Part I, the earliest systematic technique for determining folk perceptions of dialect boundaries is surveyed.3 The following two questions concerning respondents’ beliefs about the degree to which surrounding communities were linguistically similar to or different from their own were included in a 1939 Dutch dialect survey: 1.

In which place(s) in your area does one speak the same or about the same dialect as you do?

xxvi 2.

INTRODUCTION

In which place(s) in your area does one speak a definitely different dialect than you do? Can you mention any specific differences? (Rensink 1955: 20; see Chapter 1, this volume)

Antonius Weijnen (see Chapter 9, this volume) devised a method to represent the information uncovered by the first question. This Pfeilchenmethode (little-arrow method or pijltjesmethode in Dutch) connects a respondent’s home area to another that the respondent asserts to be linguistically similar. Groupings of these connected areas, representing the response of a single respondent at each location, are then identified as “unities” based on the dialect consciousness or “awareness” of the respondents. The earliest of these maps (for the North Brabant) was first published in Weijnen 1946. Here, however, I will show (in Figure 0.2) only the westernmost portion of that map since it is not so intricately detailed as some other parts but will nevertheless allow an illustration of the method. The thick lines are “traditional” dialect divisions (“bundles of isoglosses”) in this area. Perceptual areas can be determined by encircling those community labels (letters) connected by arrows. For example, in the northwest of this section of the map, the respondent from W (Willemstad) has indicated that no nearby community sounds like W, and, therefore, no arrow is drawn from that site. Similarly, no surrounding community has identified W as sounding like it, so no arrow is drawn toward it. In contrast, the respondent from D (Dinteloord) believes that the variety in F (Fijnaart) is the same as the local one, and the respondent from F returns the favor; hence, an arrow from D to F and one from F to D. (Double-headed arrows replace this reciprocal identification in, for example, Chapter 3.) The F respondent also identifies K (Klundert) as the same, but unlike D, this perception is not reciprocal. If there were a perfect match between perception (the arrows) and production (the area enclosed by the thick line), every site within the production boundary (W, D, F, and K) would be connected to every other one with two arrows (W to D, D to W, W to F, F to W, D to F, F to D, etc.). That is obviously not the case. On the other hand, one must be impressed with the perceptual-production match here, for, although not all the sites are connected to one another, none identifies as similar a site outside the production boundary, nor is any identified as similar by a site outside the production boundary. A more complex relationship exists in the area just to the east of this section. There Z (Zewenbergen) identifies M (Moerdijk, just to the north) as being the same (although reciprocal identification is not given), and Z itself is identified as the same by a respondent from one site rather far to its southwest. In both these cases, however, the thick line just to the east is not crossed. The respondent from Z, however, also asserts the similarity to Z of both ZH (Zevenbergschen Hoek) and L (Langeweg), both clearly across the production boundary, although the respondent at neither ZH nor L identifies Z as similar. In general, however, there are a relatively small number of “production boundary crossings” in Weijnen’s research on folk perception in Dutch-speaking areas.

INTRODUCTION

xxvii

Figure 0.2. Westernmost section of the North Brabant, showing isogloss boundaries (thick lines) and the “little arrows” of respondent similarity perceptions (enlarged from Weijnen 1946)

In Figure 0.3, I show my attempt to combine the results of the little-arrow method with traditional production boundaries. Where the shaded and solid thick lines run together, there is agreement between the two; where there is only one line or another, of course, there is only a perceptual or production boundary. As Figure 0.3 shows, the major mismatches between the two methods are perceptual “isolates,” all but one of which are single communities, and all of which, interestingly enough, with perhaps only one exception, lie on or near dialect or other natural or political boundaries.

xxviii

INTRODUCTION

Figure 0.3. Westernmost section of the North Brabant, showing traditional isogloss boundaries (solid thick lines) and respondent similarity perception boundaries (shaded thick lines) (enlarged and modified from Weijnen 1946)

Evaluations of Weijnen’s and his followers’ approach have not all been favorable. Peeters (1952) criticizes the enterprise by noting that since it relies on subjective factors, it is difficult to control the “psychological” elements that enter into the respondents’ judgments. As Goeman (Chapter 10, this volume) points out, however, this criticism seems a bit odd since Weijnen admits from the outset the subjective nature of his respondents’ judgments and hopes only to correlate such responses with the objective results of

INTRODUCTION

xxix

traditional dialect study. Goossens (1970) is critical that respondent judgments do not always correspond to production boundaries and worries that such consideration might impede the progress of dialect study in general. Again, however, as Goeman (Chapter 10) notes, this rejects the validity of the study of the psychological reality of the respondents’ regard for language variation and ignores Weijnen’s insistence on the necessary consideration of traditional boundaries in any study, as he asserts, for example, in the final sentence of Chapter 9. In fact, once such data are recorded, other interpretations than those that correlate folk perception and actual production boundaries may be offered. On the one hand, one might ask what linguistic facts (or even what sort of fact — i.e., phonological, lexical, grammatical) of those that are “bundled” into the production isoglosses are most salient to the folk. Weijnen, in fact, believes they are phonological ones, since, according to him, they are “sharper” than syntactic and morphological boundaries and less specific than those that arise as the result of the difference of a single lexical item; they are therefore more “locally noticeable” (e.g., 1961: 5–6; 1966: 194–95). As a corollary to that search, of course, one might ask what linguistic facts not uncovered in production dialect studies might play a role in folk perception, and Daan reflects on this question in Chapter 2, pointing out that intonation, for example, might play an important role in folk awareness but is rarely studied in the construction of production maps. (One might add, of course, vocal quality, speech rate, and a number of other factors.) One might also ask what sorts of sociohistorical (or other “nonlinguistic”) facts influence perception. For example, in Figures 0.2 and 0.3, what makes the respondent from Z look north and east rather than south and west in identifying similar areas, causing the most significant production “boundary breaking” in the comparison between results based on the two procedures. As Daan shows in Chapter 2, a religious boundary may account for respondents’ strong feelings that there is also a linguistic one there (when, in fact, none exists). As we shall see, these questions, the very ones suggested in these earliest investigations, will be the concern, in one way or another, in all the studies included here, in spite of the fact that different methodologies have been employed. Chapter 1 (Rensink), although it does not directly show the little arrows, provides a generalized map of Dutch-speaking areas based on those perceptions gathered in the 1939 survey (in which the thick lines are boundaries drawn around the bundles of interconnected little arrows), and Chapter 2 (Daan) is an ambitious study of all contiguous Dutchspeaking areas, one that combines the perceptual information with that of the usual production dialect data in representing dialect divisions. Daan provides a detailed account of both the justification for the use of subjective material and the explanation of divergences between subjective and objective results.4 In Chapter 3 (Kremer), a map that actually shows the little arrows supports a study that asks to what degree perceptual dialect areas may or may not cross national boundaries, where dialect similarity from a purely linguistic point of view across such a boundary may be great. In general, although in this research program respondents were also asked to mention areas that were different and linguistic features that divided them from their neighbors,

xxx

INTRODUCTION

the principal motivation seemed to be a desire to give dialect boundaries greater (or lesser) weight by establishing their folk validity on the basis of perceived similarity at very local levels. This is most directly and thoroughly discussed in a great deal of the work of Weijnen (e.g., 1966 and Chapter 9, this volume). On the other hand, as Chapter 3 (Kremer) shows, use has also been made of this method to simply characterize folk information and see how it patterns with nonlinguistic facts (here, the Netherlands-German political border).

2.

Part II

In Part II, a series of articles that includes responses and counterresponses introduces both the study of and controversy over subjective boundaries of dialects in Japan and in general. In Chapter 4, Sibata acknowledges that the survey of the Itoigawa region (in western Japan) included perceptual questions that were partially inspired by the work in the Netherlands, specifically by a summary of Rensink 1955 (see Chapter 1, this volume). It is also noted, however, that the Japanese interest in where boundaries are to be drawn (and if folk information should be included) was reflected in the earlier work of Misao Tôjô, whose reaction to the work of Sibata and others is included in Chapter 8. Sibata explains both how his team conducted the perceptual part of the research and how they came up with the generalized maps that reveal the findings. Respondents were asked to indicate which nearby villages were (1) not different, (2) a little different, (3) quite different, or (4) mostly incomprehensible. The question from which maps for Dutch perceptual dialects were derived (which asked where dialects were similar, presumably the equivalent to Question 1 just above) was found to be of little or no value in the Japanese research. In fact, Grootaers, a coworker on this project, notes that for the Japanese research “the first one ‘no difference’ and the second ‘slight difference’ proved to be superfluous” (1959c: 356). Therefore, the results of Question 1 above were ignored, the results of Questions 2 and 3 were combined for one mapping effort, and those for Question 4 were treated separately. At first glance, therefore, one might suggest that the Dutch area maps are ones of similarity and that the Japanese maps are ones of difference (but see Chapter 9 and comments on it later in this introduction). The Japanese team did not know the little-arrow method developed by Weijnen, and instead, they indicated, by increasingly thick lines, those areas that formed the “difference boundaries” for groups of respondents. When respondents performed similarly in stating where differences began, they were grouped into a subjective “speech community” (e.g., Figure 4.1). Sibata (in Chapter 4) and Grootaers (in Chapter 8) claim that the resulting subjective dialect boundaries are of little or no interest to linguists since they do not generally correspond to traditional dialect boundaries. In further work in Japan, Nomoto (Chapter 5) and Mase (Chapters 6 and 7) carry out studies on the role that dialect boundaries based on linguistic features, not just political and other boundaries, play in subjective boundaries, but Grootaers (Chapter 8) continues

INTRODUCTION

xxxi

his and Sibata’s claim that, except for incidental overlap, the value of folk perceptions is minimal to the scholar interested in drawing dialect maps. In this controversy in general, there is little or no reflection on the value of the folk knowledge itself, a theme that was to play a much larger role in later, more sociolinguistically oriented studies. Nomoto (Chapter 5) is able to calculate perceptual and production divisions very straightforwardly since he deals with a “straight line” of flat seacoast villages and simply tallies the number of distinctions (whether of perception, production, awareness of production, or accuracy of awareness of production) made. He finds an interesting greater awareness of features that support actual dialect boundaries among better-educated respondents, although his general findings do not show such good matches between perception and production as those of Weijnen’s. Mase, in Chapter 6, provides the first “calculus” for drawing perceptual maps. Although the little-arrow method provides graphically clear evidence for such boundaries, one feels somewhat uneasy that even one respondent may link two large, otherwise fully differentiated regions by simply naming one village in the other region as similar. (Note, for example, how the one respondent from Z in Figure 0.2 “ruins” the perceptual unity of the largest, western dialect area of the North Brabant.) Nomoto’s boundaries do away with this possibility in a straightforward numeric way, and Mase’s is even more quantitatively sophisticated. He not only requires a numerical standard (two-thirds) for isolating a group of sites as “perceptually similar” but employs a second (one-third) to identify subregions of perceptual similarity within major ones. On the nonlinguistic side, both Mase and Nomoto find school districts rather than feudal and other political administrative zones to be very similar to perceptual boundaries. Grootaers, however (Chapter 8), finds that school boundaries are not a factor in another area and concludes that dialect consciousness plays the same role as other nonlinguistic factors in accounting for the distribution of dialect features, but he is adamant that only dialect features themselves be used in constructing dialect maps. In Chapter 8, Grootaers also translates and quotes liberally from Hammarström, an advocate of perceptual considerations in dialect study. Weijnen (Chapter 9), who devised the little-arrow method for the Dutch perception data (Part I), criticizes the Japanese approach to perceptual studies in Itoigawa by noting what he considers to be a fatal flaw: The Sibata-Grootaers team asked people if there were differences (which, according to Weijnen, always exist) rather than asking people where others spoke the same. Not surprisingly, Weijnen praises Mase, who found a greater parallel between perception and production, for his use of the “more appropriate” question.5 It is clear that the two questions may lead to radically different maps. For example, Figure 4.1 shows (inside “toothed” boundaries) two different subjective dialect areas in Japan (on the left and right sides of the map, which duplicate the same geographical area). These boundaries, however, are based exclusively on the similarity of responses given within them at a number of sites concerning other sites that are thought of as different. In Figure 4.1, areas that have been mentioned as different have had lines (of varying thicknesses, representing frequency of mention) drawn in front of them (with thin lines

xxxii

INTRODUCTION

pointing back to the individual sites that mentioned them, which, in turn, have small dark squares on the circles around them pointing to the area mentioned as different). If we look only at the sites enclosed within the two toothed boundaries on the left and right of Figure 4.1, it should be immediately apparent that only Sites 20, 92, 24, 48, and 15 were mentioned by any other sites within either zone’s boundary as different (and only 20 and 48 by more than two respondents). If, on the other hand, we were to assume a sort of negative evidence — that is, that nearby areas not mentioned as different might be regarded as similar — we could, in effect, draw little arrows and connect those (local) sites that did not mention one another as different. If we apply this technique, all the sites in both of these areas (seen as distinct by the Japanese research team) would be interconnected by a massive pattern of little arrows and would constitute a single perceptual area. In Chapter 6, as noted earlier, Mase’s more quantitatively oriented technique considers both similarity and difference ratings in calculating subjective boundaries. Interestingly, Sibata’s work on Shimo-kita peninsula (which Grootaers summarizes in Chapter 8) asks where villages are similar (as well as different), and the map that results from this work (Figure 8.1) is based on similarity judgments rather than on the more complex consideration of “difference rating similarities” shown in Chapter 4. In spite of having used the method that Weijnen would have approved of, Grootaers finds that this more recent work of Sibata’s still does not show the close correspondence between subjective boundaries and objective boundaries that he would like to find in establishing the value of the former. The last word in this controversy may have to do with ends rather than means. If one seeks to supplement the details of production dialect maps with “awareness weights” from local ratings (telling the reader which dialect boundaries have and do not have greater folk significance), then subjective maps that result in boundaries that do not generally correspond to production boundaries will be of little help. Although that appears to be Weijnen’s opinion, Daan (Chapter 2) seems to be more interested in what one might learn from the mismatches (as well as from the matches) in the two methods. If one seeks a general approach to dialect mapping, apparently the general Japanese motivation, then, like Sibata and Grootaers, one may be disappointed that subjective boundaries do not provide a ready-made way to offer a succinct (perhaps even guiding) picture of language distribution. If, however, one seeks corroborating and explanatory evidence for dialect distribution (as Grootaers himself concludes in Chapter 8), one should not ignore the voice of the folk in this matter. That this voice would have independent value (and applications) does not seem to be a conclusion reached in this early work. Finally, in Chapter 10 Goeman reviews much of the Dutch and Japanese work presented in Parts I and II, identifies (and illustrates) some interesting precursors, and provides both dialectological and general theoretical criticism of much of this early enterprise.

INTRODUCTION

3.

xxxiii

Part III

Part III is devoted to what one might call the sociolinguistic (or “modern”) trend in perceptual dialectology. It begins with a sample of the work of Fumio Inoue, a scholar whose main concern has not been with the correspondence between perceptual and production boundaries (although he carried out such studies in northern Japan, the results of which are very briefly summarized in Chapter 12). Inoue was convinced of the value of folk knowledge, and he has devised a technique that identifies the attributes of speech regions on the basis of what he calls “dialect image.” His method is inventive, sensitive to language attitude concerns, and statistically sophisticated. Chapter 11 illustrates the foundation for much of Inoue’s subsequent work. He identifies (through evaluative words associated with predetermined regions) the two principal characteristics associated with dialect image in Japan as “intellectual” and “emotional.” These correspond closely to the “status” versus “solidarity” factor groups that emerge from most quantitative work done on language attitudes by social psychologists of language (e.g., Ryan, Giles, and Sebastian 1982) and to the “correct” versus “pleasant” characteristics of varieties assessed in much of the work reported in Part III. In plotting these dialect images, Inoue relies on a Japanese version of multidimensional scaling known as Hayashi 3. It allows a researcher to group together both the evaluative labels assigned to varieties and, later, the varieties themselves. Inoue goes on to illustrate an application of this technique in Great Britain with a resulting dialect image map. Interestingly, the components selected in the statistical treatment of labels used in Great Britain were not emotional and intellectual but “rural” and “standard.” In Chapter 12, further information about the treatment of variety perception in Great Britain is detailed, and Inoue shows more directly how the multidimensional scaling approach is applied both to the evaluative labels and to the regions evaluated. He also displays a technique for determining the perceptual dialect areas of the country based on hand-drawn maps, a theme that will be returned to several times in Part III, although like Sibata and Grootaers, he believes that nonlinguistic factors determine folk perceptions of the extent of regional varieties. In Inoue’s work and in the rest of Part III, the reader will see the more recent concern with multiple responses from one area. In short, in the majority of these remaining studies, the researchers are more concerned with averages and other generalizations that can be drawn from a number of respondents from the same area. In some of these studies, therefore, internal demographics among the respondents will play an important role in the analysis as they could not in single-respondent studies, which are typical of much older work in traditional dialectology (in which the respondents were referred to as “NORMs” — nonmobile, older, rural, males — in Chambers and Trudgill 1980: 33). The remainder of Part III contains a number of articles that use and elaborate on techniques that I devised and reported on in a series of publications in the 1980s and 1990s (listed under Preston in the bibliography). None of that work is included here since much of the early stages of it was collected and summarized in a monograph (Preston 1989), and most of the newer items are readily available in English-language journals and

xxxiv

INTRODUCTION

anthologies. In addition, it is important here to give ample space to new approaches (and approaches to new areas) being carried out by other scholars. The principal techniques that I developed for perceptual dialectology in the 1980s were the following: 1.

2.

3.

4.

5.

Draw-a-map. Respondents draw boundaries on a blank (or minimally detailed) map around areas where they believe regional speech zones exist; a technique developed by Preston and Howe (1987) allows computerized generalizations to be compiled from individual responses to this task. Although respondent hand-drawn maps were well-known in cultural geography (e.g., Gould and White 1986), there does not appear to be a long-standing tradition for the use of this technique in the study of dialect perceptions. Degree of difference. Respondents rank regions on a scale of one to four (1 = same, 2 = a little different, 3 = different, 4 = unintelligibly different) for the perceived degree of dialect difference from the home area. Versions of this technique are longstanding in perceptual dialect study and are reported in Parts I and II of this volume (e.g., Chapters 1 and 4). “Correct” and “pleasant.” Respondents rank regions for correct and pleasant speech; such ratings are common in other areas of cultural geography (e.g., Gould and White 1986) and reflect principal findings from language attitude studies (e.g., Ryan, Giles, and Sebastian 1982), although in the latter, respondents judge actual voice samples rather than their internal representation of speech differences when confronted simply with a regional label. Dialect identification. Respondents listen to voices on a “dialect continuum,” although the voices are presented in a scrambled order. The respondents are instructed to assign each voice to the site where they think it belongs.6 Qualitative data. Respondents are questioned about the tasks they have carried out and are engaged in open-ended conversations about language varieties, speakers of them, and related topics.

The principal generalizations that emerged from those earlier works include the following: 1. 2.

3.

Draw-a-map. Respondents first draw stigmatized and then local areas most frequently. “Correct” and “Pleasant.” Respondents from areas with a great deal of linguistic security rate the local area as uniquely correct, but they include a larger region in the area they consider most pleasant; respondents from areas of linguistic insecurity rate the local area as most pleasant, but they rate a number of areas as most correct. Respondents from areas of high linguistic security often find the same areas to be least correct and least pleasant; respondents from areas of high linguistic insecurity often find different areas least pleasant and least correct. Degree of difference. Respondents from areas of linguistic security find regions of least correctness and pleasantness also most different (often ranking them as even

INTRODUCTION

4.

xxxv

unintelligible). Respondents from areas of linguistic insecurity may find areas of either high or low correctness and pleasantness maximally different from the local area. Dialect identification. Nonlinguists are surprisingly good at distinguishing voices along a dialect continuum, even at sites where traditional dialect geography has not posited an isogloss. Respondents from different areas, however, “hear” boundaries at different places and with different degrees of intensity, more often making more distinctions closer to the local area and fewer in areas farther from the local area.

Conversational data are, of course, not so easy to summarize. One might, however, note the following trends: 1. 2.

3.

Face-to-face contacts are much more frequently mentioned than popular culture vehicles in accounting for familiarity with other varieties. Overt identification of details of other varieties is very weak (perhaps particularly at phonological levels), but imitations of the varieties are often convincing (at least to nonlinguists) and contain accurate as well as inaccurate representations of linguistic elements of the varieties (Preston 1992, 1996a). Concerns with correctness are more frequently mentioned than any others in overt discussions of language and variety.

Preston (1993a, 1993e, 1996a, 1997) and Niedzielski and Preston (forthcoming) present a great deal of such conversational data concerning language, much of it focused on variety. The remaining articles in Part III expand on these methods, introduce others, apply newer statistical techniques to the results, offer alternative interpretations of findings, and extend the program of research to new areas. In Chapter 13, Long fine-tunes the use of computerized generalizations of hand-drawn maps in a study of the perception of dialect areas in Japan. He shows how respondent hand-drawn maps are converted into numeric data and used to produce computer generalizations of a large number of responses. Long shows both the value of studying the names given to dialect regions by the respondents (including, of course, evaluative names such as “standard”) and the value of looking at the intensity with which respondents identify parts of a region as belonging to it or not. Although Long shows that some perceptual boundaries correspond to administrative districts (Sibata and Grootaers’s complaint, Chapter 4), he finds other boundaries that do not and concludes that a variety of influences are at work, some difficult to distinguish from one another. In Chapter 14, Long continues to apply perceptual techniques to Japanese varieties but this time studies the idea of “most pleasant” spoken Japanese. Rather than applying typical factor-analytic techniques to the results of a questionnaire that asked respondents to rank areas along this dimension, Long standardizes scores and directly represents respondents’ judgments of the most pleasant speaking areas of Japan on a map. In a series of such maps, he isolates those characteristics associated with most pleasant (both linguistic and nonlinguistic). Finally he compares his findings with those of Sibata and

xxxvi

INTRODUCTION

Grootaers (e.g., Chapter 4) and Inoue’s “dialect image” (e.g., Chapter 11), and with the bulk of my work (outlined above). Jennifer Dailey-O’Cain’s work in Chapter 15 is a part of her larger study of perceptual dialectology with a special focus on the former east-west political division of Germany. In some ways, her work is a combination of older and newer approaches since she has representative data from a large number of sites all over the country. She finds echoes of the earlier political division in her respondents’ assessments of both pleasant and correct spoken German and in their hand-drawn maps. Her technique of combining responses for hand-drawn maps is the one more commonly used in these studies. That is, predetermined areas are referred to and represented as part (or not part) of a respondent groups’ perceptual dialect area identification. In Chapter 16, Larry Kuiper focuses on an area of France with “high linguistic security” (Ile-de-France) and asks for the “full range” of perceptual reactions — degree of difference, hand-drawn maps, and pleasant and correct assessments. Like some other recent work (and like Inoue’s use of the Hayashi 3 technique), Kuiper uses multidimensional scaling combined with cluster analysis to assess both degree of difference and evaluative ranking tasks, showing that the linguistically secure indeed think very well of themselves for correctness but, as in earlier work, not exclusively well of themselves for pleasantness. In fact, for these respondents, Provence wins over the local area, even though only slightly, for pleasantness. Kuiper’s work is also enhanced by ample quotation from and analysis of respondent interviews. In Chapter 17, Demirci and Kleiner also conduct a one-site (Bursa) high-linguistic security survey of an entire country (Turkey). Again using sophisticated statistical treatments of ranking data (multidimensional scaling with cluster analysis and analysis of variance [ANOVA]), they show a relatively even (declining) west-to-east pattern in the ratings of spoken Turkish (by these western respondents), although there are interesting “islands” or “pockets” of disruption of this flow. Most interesting in this report, however, are the significant differences among respondents in their ratings. For example, men rate numerous areas higher for both pleasantness and correctness than women do; older respondents also gave higher ratings in general, but middle-aged (not youngest) respondents gave the lowest ratings (perhaps a parallel to the sociolinguistic phenomenon of “age-grading”). Donald Lance, in Chapter 18, provides a multi-area picture of perceptual maps from numerous regions of the United States. Like Inoue and others, Lance takes “pre-set” regional areas as the background from which to figure the intensity of perceptual regions. Although the most striking feature of Lance’s study is clearly the variation in regional identification offered by his regionally various respondents, he also offers interesting comparisons of his respondents’ representations with “traditional” production studies. Chapter 19, Laura Hartley’s contribution, again focuses on the responses from a smaller area (Oregon) that also has a reputation for high linguistic security. Unlike earlier findings in Michigan (and France), for example, Hartley shows that Oregon respondents rate themselves highest for both pleasantness and correctness but are, in many cases, unwilling to negatively rank other areas. Again, sophisticated statistical techniques

INTRODUCTION

xxxvii

(multidimensional scaling with cluster analysis and chi-square treatments to avoid parametric analyses of data that do not fit normal distributions) are used to show the interesting patterns of area rankings that emerge in this study, although unlike the Turkish study, internal patterns (e.g., between men and women or age groups) were not found. In Chapter 20 Coupland, Williams, and Garrett study language perception and attitude in Wales. Asking teachers all over the country to draw in variety differences and evaluate them, they show that the labels given to regions correspond quite directly with a dialect assessment task (in which “described” varieties rather than actual samples are presented). The rankings show a typical language attitude grouping of the ranking tasks into “pleasant,” “dynamic,” and “prestigious” factor groups, but a factor identified as “Welshness” did not correlate highly with any of these groups. The principal finding of this study is that varieties of Welsh English that are viewed as more or less Welsh cannot be linked to any other specific affective or prestige dimension of dialect. In a further study (Chapter 21), Williams, Garrett, and Coupland asked Welsh schoolage children to tell them where dialects (actual samples) were from, a response to a criticism leveled against language attitude surveys in my earliest work in perceptual dialectology (e.g., Preston 1989). Interestingly, although dialect identification was not particularly good, evaluations of speakers (all telling short narratives) were quite varied and appeared to have as much to do with content and narrative style as with dialect itself. In Chapter 22, I try to come full circle in language attitude studies by submitting the perceptual areas that resulted from an earlier study of hand-drawn Michigan maps for a semantic differential rating. The labels used were elicited from similar respondents, and university-age respondents used them to evaluate several perceptual dialect areas. I report here only on the rankings for South and North Central (the local area) and show corroborating evidence for earlier findings concerning pleasant and correct varieties from this part of the United States.

4.

What Next?

I believe the many scholarly perspectives presented in this volume that have been taken on the folk identity of and regard for language varieties should attract the attention of dialectologists, sociolinguists, and students of the social psychology of language, and perhaps that is a foregone conclusion. I also believe, however, that there is a more general interest among social and cognitive scientists (including those who would like to apply their knowledge to public venues such as law, medicine, and education) in knowing what the folk believe about this most human of enterprises. If folk linguistics is of any value to such scholars, then this volume surely shows that one of the dominating folk concerns in language is variety and pre- (and pro-) scription. Much of this work might continue as it has above, refining the methodologies and applying them to new situations. I also believe, however, that these findings will engage the interests of those who more centrally locate their practice in linguistics. To take only

xxxviii

INTRODUCTION

one example, I believe that future work in the perception of variety might focus more specifically on the exact linguistic elements that give rise to perception rather than on the global presentation of varieties (or variety or area labels) in eliciting responses. Although folk imitations are one way of approaching this problem (e.g., Preston 1992), the presentation of specific elements (by name, by actual sample, or by computer-modified samples) for identification, placement, and evaluation by respondents is surely one way to grasp even greater details of the triggering mechanisms of language regard among the folk and, through such study, the potential influence of such regard on the more general processes of variation and change. Of course, one might argue that the presentation of actual speech samples makes these techniques a part of the language attitude tradition, and of course, the boundaries between that tradition and perceptual dialectology are difficult to draw. In many cases, it is hard to determine when the principal concern of a piece of research has been with the determination of folk sensitivity to regional speech boundaries or with an assessment of the respondents’ attitudes toward regional speech. Admittedly, when attitudes to regional labels rather than actual samples are given, I have given studies using this technique “full status” in the perceptual dialectology enterprise. From one point of view, of course, the presentation of labels rather than actual speech samples for judgment is simply an alternative technique in the study of language attitudes. From another, however, since attitudinal factors have been shown to be strong determiners of the salience of areas themselves, any study of responses to regional speech is an integral part of the perceptual dialectology enterprise. If we include social dialectology as a part of our regular definition of dialects (and I obviously believe we should, e.g., Preston 1993d: 2–3), there is an even greater risk of perceptual dialectology’s growth. It is clear, for example, that studies such as Labov (1966), which asked respondents in New York City to evaluate the social status of speakers on the basis of the frequency of nonprevocalic /r/ deletion or stop substitution for the interdental fricatives, were, according to the above definition, early examples of “perceptual social dialectology.” When Graff, Labov, and Harris (1986) instrumentally manipulated the onset of the /aw/ diphthong of an African American speaker from Philadelphia (in which the /a/ portion was “fronted” to a position nearer /æ/), they succeeded in showing that this fronting alone (with no other alteration of the speaker’s performance) was enough to signal “white” ethnicity to both African and European American respondents. This was, as well, a case of perceptual social dialectology, with ethnicity as its principal target. In the long run, I doubt if we will be successful in showing in any ironclad way what linguistics is (at least a linguistics that includes the study of language variation and change) and what the “social psychology of language” is. Perhaps linguists will be more interested in isolating the specific language features that trigger attitudinal responses and identifications, and social psychologists will be more interested in isolating the sociocultural forces that form and maintain the set of predispositions responsible for attitudes, but I am not at all surprised (nor disheartened) by the prospect of a great deal of interdisciplinary poaching.

INTRODUCTION

xxxix

Whatever the boundaries of this enterprise (and whatever directions it takes), I am delighted and honored to be able to present this feast of both past and present studies of the folk perceptions of language varieties.

5.

How to Read This Book

In the new as well as older material, I and the various authors have provided extensive cross-referencing (which ought to allow the reader considerable opportunity to compare specific techniques and interpretations). Although I have tried to be faithful to the originals (whether translations or not), I have “normalized” references, no longer leaving as footnotes, for example, items in earlier works that simply referred to bibliographical matters. In some cases, the translators and I have inserted editorial and/or translation comment, directly in the text when brief, in notes when more extensive. For that reason, the note or footnote numbers of the originals will not be the same as those given here. Such material is either given in square brackets or clearly identified as editorial or translator comment. In reprinting previously published work, I have left many of the authors’ typographical conventions intact (e.g., small caps, italics, bold), but I have standardized the heading and subheading style. Finally, I have gathered the references into a final section, since there would have been considerable overlap if this had not been done. I have also tried to include a comprehensive list of references having to do with perceptual dialectology. That is, in a separate bibliography, I have included everything I know of on perceptual dialectology, whether the item was referred to in any of the chapters of this work or not. I would appreciate hearing from anyone who has any additional information about publications in this area that I have inadvertently omitted.

Notes 1. The term perceptual dialectology in relation to these studies was first used, so far as I know, in Preston (1981). If it were not for the common and unfortunate misunderstanding of folk as “false,” I would now prefer folk dialectology, and that use would make it clear that this initiative is but one of any number of subareas of investigation in “folk linguistics.” 2. A more detailed characterization of folk linguistics is given in Niedzielski and Preston (forthcoming) and Preston (forthcoming), and the former contains a number of such investigations outside the area of dialectology. 3. According to Goeman (Chapter 10), the first systematic investigation of dialect perception by nonlinguists was carried out by Willems (1886); although the data are unpublished, Goeman represents Willems’s findings in a little-arrows map (Figure 10.1). 4. Unfortunately, the color-coded map that results from Daan’s work is too detailed to include in this volume. I have reworked it (Figure 2.2) to show the main results, but those interested in the details of the characteristics on which it is based should, of course, consult the original. 5. Weijnen (1961) also evaluated a 1939 work by Büld that attempted to determine dialect boundaries on the basis of humorous folk caricatures of speech. After attempting to carry out Büld’s technique in a Dutch

xl

INTRODUCTION dialect setting, Weijnen found it unsatisfactory. There is a brief summary of Weijnen’s criticism of this approach in Kremer (Chapter 3).

6. A similar technique is suggested in the most outspoken proponent of the need for perceptual data in drawing dialect boundaries. Jernudd (1968) simply contends that folk knowledge is an integral part of the scholarly representation of dialect divisions. He outlines a “program” for such research in which he recommends eliciting folk responses to actual dialect features, a procedure used in very little of the previous work in Dutch-speaking areas or in Japan (but see Nomoto, Chapter 5, and Mase, Chapters 6 and 7). A similar strategy is used in the Welsh studies reported in Chapters 20 and 21, and Diercks (1988) uses actual voice samples in determining regional speech awareness in a small area of northern Germany.

P I The Dutch Contribution ‘Little Arrows’

C 1 Informant Classification of Dialects W. G. Rensink Translated by Dennis R. Preston

Among the questions asked in “Questionnaire #8” in 1939 were the following two: “(1) In which place(s) in your area does one speak the same or about the same dialect as you do? (2) In which place(s) in your area does one speak a definitely different dialect than you do? Can you mention any specific differences?”1 We tried to construct a map [i.e., Figure 1.1] of the classification of dialects using the answers given to these two questions. In spite of the map presented here, the attempt was only partially successful due to the following reasons: (1) In many sites there are no informants, so there is missing data. (2) The data gathered are often contradictory. (3) There are often no clear boundaries. One should see this map as a tentative sketch, a sample of what the informants know. If there are errors or omissions on the map in your own area or if you have suggestions or additions, please write us or draw a map in which you show your ideas of the boundaries and explain why. We would appreciate this. Following are some examples of districts in Holland which provide examples of answers to the questions and the difficulty in interpreting them. An informant from Groningen Province (Wagenborgen) answered Question 1 (above) as follows: “Nieuwolda, Termunten, Termunterzijl, ’t Waar, Noordbroek, Siddeburen, Oterdum, Heveskes, Woldendorp and all other villages in far surroundings.” The answer to Question 2 (above) was “None.” Another informant from Woldendorp noted that a variety of the local dialect was spoken in Oterdum and Heveskes. There they say “tweide” [Tuesday] and “vrijdag” [Friday] for the local (i.e., Woldendorp) “twaide” and “vraidag.” It is clear that both informants did their best in filling out the questionnaires but used very different standards. The first noticed large differences, the second, small. Both ways involve certain risks. If one looks too broadly over a wide area, important differences may be missed. If one is too detailed, it will turn out that no two villages speak the same dialect. In actual practice, however, most informants find a good middle ground for responding. In Friesland there are actually more dialects than indicated on the map. Some, for example the dialect of Hindeloopen, are restricted to one site; for others, we were unable

4

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

to define boundaries based on our data. There will always be borderline cases. For example, informants from Oldeboorn indicate that they speak the same dialect as in Akkrum but that Beets and Tijnje have the same dialect as in Eastern Friesland (“Woudfries”). Grouw is said to have such differences as “waansdei” [Wednesday] for Oldeboorn’s “wèènsdei.” Of course, Frisian spoken in the city should not even be referred to this way. An informant from Leeuwarden, with a guilty conscience, answered the question about different speech by saying, “Around Leeuwarden — they speak Frisian.” The dialect in Drenthe is sharply separated from that of Groningen by the peat colonies. According to answers given in Drenthe, “corruption” of speech was a large concern. The informants there provided a greater number of boundaries than are actually given on the map. “’t Hoogeveen” stands apart, and around Dwingeloo there is a special pronunciation of “aa” as “èè,” so that one “met de wèègn deur ’t wèèter jèègt dat zo klèètert teegn de glèèzn” [“drives his wagon so fast through the puddles that the water splashes on the windows,” a sentence constructed to illustrate multiple substitutions of “èè” for “aa”]. A respondent from Zuidwolde concludes: “In every village certain words have a different pronunciation, but to write down the differences is out of the question.” From the data there appears to be a clear boundary between Drenthe and Overijssel, but the boundary was drawn with some misgivings since there was contradictory evidence. Informants in Nijverdal indicate that the dialects of Hellendoorn, Wierden, and Rijssen are the same as theirs (that is, in Nijverdal). A Hellendoorn informant writes, however, that “The dialects spoken in Vriezenveen, Rijssen, and Wierden are so exceptional that they cannot be written about in a few lines.” This is true for Vriezenveen, but for the other two as well? The job of drawing boundaries in Gelderland has been especially difficult due to a lack of data, specifically from Veluwe. The area between the big rivers creates problems due to such responses as these: “The Upper Betuwe has different sounds, but I wouldn’t say that it is a different dialect” (from Kesteren) and “The further away from a place, the bigger the difference in language gets. I’m not able to draw a fixed boundary or line” (from Opheusden). This latter comment is, of course, often true, and there is no fixed boundary. The map, with its fixed, sharp boundaries creates the wrong impression. The reader should have a map with all the responses separated and be allowed to see the overlapping of the areas. Such a map seems, however, impractical. There is a gradual transition from North Holland to South Holland, from West Friesland to the islands of South Holland. Only the “Zaans” [a highly industrialized area just north of Amsterdam] seems to deviate strongly. According to the informants, most southwesterly islands are seen as separate dialect areas. Goeree-Overflakkee and Schouwen-Duiveland fall into two separate areas as do such cities as Bruinisse and Zierikzee. The villagers in the area know to say that in Bruinisse the “zunne in de tunne” shines, not the “zonne in de tonne” (sun in the barrel). On Walcheren, the towns of Westkapelle and Arnemuiden are in exceptional positions compared to the usually homogeneous Walcheren. Several informants said that in Arnemuiden “they sing as they speak their dialect.”

INFORMANT CLASSIFICATION OF DIALECTS

5

From North Brabant Dr. A. Weijnen provided the material and had already prepared a map which showed lines connecting any two places with the same dialect [see Figure 0.2]. Lines showing major areas were also placed on his map as on the one presented here. We copied several of these, but he provided even greater detail due to his knowledge of the area. We left these out since they would have given much more detail than in the rest of our map.2 The dialects of Limburg differ considerably from place to place, but we were unable to distinguish major areas. For example, we got a map from Vaals which had some lines helpful to us, and we hope to get more like it.3 On a map, not shown here, the sites which were mentioned as having the same (or about the same) dialect were connected to the informant’s local area. Lines were drawn between the groups of connected sites to prepare the map below.4

Figure 1.1. Respondent classifications of Dutch dialects

6

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

Figure 1.2. Sites in The Netherlands referred to in Chapter 1

INFORMANT CLASSIFICATION OF DIALECTS

7

Key to Figure 1.2: The provinces mentioned in the text are identified by outline type (e.g., FRIESLAND); two areas of western Gelderland (Veluwe and Betuwe) and three areas in the southwest (Zeeland) are underlined (e.g., VELUWE). The communities are identified with the following two-letter codes: In Groningen (Northeast) HV Heveskes NW Nieuwolda NB Noordbroek OT Oterdum SB Siddeburen TM Termunten TW ’t Waar TZ Termunterzijl WG Wagenborgen WL Woldendorp In Friesland (Northwest) AK Akkrum BT Beets GR Grouw HD Hindeloopen LW Leeuwarden OL Oldeboorn TJ Tijnje In Drenthe DW Dwingeloo HG ’t Hoogeveen ZW Zuidwolde

In Overijssel HL Hellendoorn NV Nijverdal RJ Rijssen VZ Vriezenveen WR Wierden In Gelderland KS Kesteren OP Opheusden In North Holland ZN Zaans On Schouwen-Duiveland BR Bruinisse ZR Zierikzee On Walcheren AR Arnemuiden WK Westkapelle In Limburg VL Vaals

Notes

1. Editor’s note: Rensink refers to the questionnaire from which the data for this perceptual map (and the one in Chapter 2) were (in part) derived. A description of the development of the questionnaire itself is given by Weijnen (Chapter 9). A more general characterization of the history of the concern for dialect perceptions in Dutch is given by Goeman (Chapter 10). 2. Editor’s note: Rensink’s characterization of Weijnen’s maps is not correct. Rensink’s map shows lines where no little arrows connected; that is, the lines in Figure 1.1 are “perception” lines, grouping together areas that are identified as “similar.” On Weijnen’s maps (e.g., Figure 0.2), the lines indicate the traditional “production” dialect boundaries (for comparison with the “spaces” left where there are no connecting arrows). 3. Editor’s note: It is odd that Rensink uses Weijnen’s data from his study of North Brabant (1946) but apparently does not know of his perceptual map of Limburg (1947). 4. Editor’s note: Figure 1.1, the original, does not identify the sites referred to in the text. Figure 1.2 provides a key to the regions and communities mentioned.

C 2 Dialects Jo C. Daan Translated by Karen Bouwer

1.

Introduction

A contemporary “Hollander” who strolls from The Hague [see Figure 2.1 for the location of sites mentioned in the text] and stops to chat from time to time will hear the rural dialects gradually change, although sometimes such changes will be very noticeable indeed. Upon seeing the towers of Woerden, he or she will no longer hear hand and gras but something like haand and graas, reminding our imaginary traveler of the radio program Baartels (Bartels) which typified Utrecht. After crossing this Bartels/Baartels border, another familiar linguistic phenomenon will change: when Utrecht’s cathedral is in sight, heb ie becomes heb je. (The second person singular pronoun, when inverted, is pronounced ie in most positions in the west, where the traveler started, while in the rest of Holland and in the Netherlands, it is je.) Between Utrecht and Amersfoort our traveler will hear the t of ik loopt less often. But in the region of the eastern border of the province of Utrecht, he or she can’t help but notice that the n at the end of words is pronounced more often and more clearly. From this final-n isogloss, this phenomenon increases towards the north and the east. In fact, the zettn® and loopm® of the inhabitants of Groningen is often the characteristic for which a “farmer” (boer) is ridiculed. The above imaginary journey is but one example; one can, fanning out in all directions from the outskirts of urban areas, have similar experiences. For every speaker of Dutch, the northeast area is that of the final-n, the south that of the soft-g. Besides these important differences, there are small isoglosses1 that run everywhere but are not evenly distributed; there are differences in their density, direction, and importance. The linguist can determine all these isoglosses and apply statistical methods to compare their density and direction. But because of the objectivity of the approach, it will not be possible to distinguish which isoglosses are important and which are not; the linguist will be unable to evaluate the relevance of the isoglosses with respect to the delimitation of dialectal regions. All too often, dialects have been studied mainly in written or printed material in which the phenomena are inevitably oversimplified. But

10

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

dialects are spoken languages, and in speech a role is played by elements that cannot be expressed even in a phonetically written text. Cultural and social factors give rise to exceptions which may be more important than the actual structures. In 1939 the Department of Dialects of the Royal Netherlands Academy of Sciences and Letters at Amsterdam, which has some 1,500 correspondents in all parts of the country, sent a questionnaire to these respondents asking them to indicate the places in their vicinity in which virtually the same spoken language was used as in their own place of residence. For various areas the dialect boundaries resulting from this survey were found to differ in some respects from those of older maps. The discrepancies appear to be the result of various factors. The title [of the book from which this chapter is taken, i.e., Van Randstad tot Landrand], borrowed from the imaginary experience described above of the contemporary “Hollander” in the Netherlands, is not, however, automatically applicable to Dutchspeaking Belgium. There the situation, because of differences in historical and contemporary development, strongly diverges from that in the Netherlands. That part of the research could not be carried out in the same way because a large group of respondents (such as the central office in Amsterdam has always had) was not available. In contrast with the Netherlands research, the work on Dutch-speaking Belgium is based on data obtained from linguists who are almost all speakers of dialects, which gives sufficient certainty that in this region, too, the experience of the dialect speaker is properly expressed.

2.

How We Got This Far

The Dialect Bureau in Amsterdam receives many requests for information. Questions concerning the number of existing dialects and the meaning of the concept dialect itself far outnumber other requests. From such letters it often appears that the prevailing misunderstandings in this field arise from the transmission of faulty interpretations and incomplete information. One example of a persisting legend is that the best Dutch in the Netherlands is spoken in Haarlem. It is possible that a statement by Johan Winkler has popularized this misconception. Anyone who reads the following quote carefully will see that he is saying something quite different: The present dialect of the city Haarlem is undoubtedly closer to the current Dutch language than any other dialect of Holland and therefore of all of the Netherlands. The spoken language closest to Haarlem is most like the current written language. Moreover, the true dialect of Haarlem, insofar as it still exists, is probably spoken by half of the inhabitants; the other half, among whom there are many foreigners, speak modern Dutch. (Winkler, Part 2, 1874, 77)

Winkler was thus saying that in Haarlem there are many people who speak Dutch and that the dialect spoken in Haarlem differs less from Dutch than the other regional varieties. But he does not say that the Dutch of the inhabitants of Haarlem is better than that of the

DIALECTS

11

people of Amsterdam, The Hague, Utrecht, etc. The origins of this annoying and perhaps indestructible misunderstanding, that dialects are a corruption of Dutch, lie much deeper; they are to be found in the social conflicts of the 19th century. These oppositions are also responsible for the troublesome myth that speaking a dialect is uncivilized and speaking Dutch is civilized. Social relations, especially in the previous century, formed a rich source for these misconceptions, which would also feed the Haarlem myth. The accepted public opinion of the previous century about uncivilized dialects is supported by, among others, Terwey’s grammar, which was used to educate many generations of teachers. In the introduction to the sixth edition he writes: The language written and spoken by civilized Netherlanders is called the Dutch language (Nederlandsche taal). The language is also used in part of the Kingdom of Belgium; there it is called Flemish. The languages spoken — seldom written — by uncivilized Netherlanders are called dialects. The number of dialects can be considered larger or smaller depending on whether one concentrates on similarities or differences among dialects. Three major dialects are however to be distinguished: Frankish, Saxon, and Frisian. From the first, under the influence of many others, over time, the general spoken and written Dutch language originated. (Terwey 1885, 2)

The second part of this quote makes it clear that Terwey was aware of the linguistic theories of his time; the first part corresponds to the popular accepted notion of the time. It was the idea of the first part which lodged itself in the beliefs of the layperson; that of the second part, on the whole, made no impression. The evil is so deeply rooted that the words of C. H. den Hertog, in the introduction to De Nederlandsche taal (Den Hertog, Part 1, Comments, 1897, 10ff), appear to have had very little or no influence at all. Probably his judgment, although linguistically sound, was too far from public opinion, and he was too far ahead of his time. His definition of dialects is as follows: To local language usage one adds the dialects, which can be distinguished from one another by, among other things, singular sound differences. These are to the general language as wild plants are to cultivated ones. They are only rarely written and are less susceptible to the striving for unity and regularity and to the changes manifested by the continuing development in the life of a language in general. When one pays attention to small differences, a good 100 dialects are spoken in our country. If one pays attention to important things, then we distinguish three groups: Frisian, Saxon and Low Franconian dialects, the second in the eastern parts, the last in the south, center and west of the country. The general language usually evolves out of one or more dialects, is used as a written language, and, for this reason, also becomes more or less native in the regions where other dialects are spoken, so that it is elevated above these and gradually becomes recognized as the general spoken and written language. Middle Dutch is principally the written form of the languages spoken in Flanders, Brabant, and Holland. Because Holland was the most strongly identified by later political and literary developments, the elements from this area take precedence over the others. The State translation of the Bible also had a great influence. The Frisians rebelled the most vehemently against the superior power and have a small literature of their own.

12

THE DUTCH CONTRIBUTION: “LITTLE ARROWS” The general language is actually an abstraction, because almost all language has something personal. Where the personal characteristic is almost completely absent — in the language of the authorities, of the courts and the church, commerce and industry, science and school, newspapers and magazines — there the general language can be most clearly observed. From the preceding it follows that the term “civilized language” for “general language” includes an arrogation. There are dialects which are ugly and crude, but there are also some that are quite sweet and pleasant sounding. The differences reflect the fact that the common language must fulfill many needs and is therefore more strongly developed. It is as with clothing. Local clothing is sometimes prettier and more picturesque than the general.

The opinion that dialects are uncivilized is found especially among inhabitants of the “Randstad” [the urban agglomerate of the west of the country; see the editor’s note] even among the so-called lower classes as well as among those of more developed or so-called higher classes. In the last fifteen or twenty years, there has been an improvement, which may correspond more or less with the regional language movements in a few areas of the country. But the reaction to the most recent intense phase of the Frisian language struggle, during the last month of 1951, was proof of how much lack of insight still exists and to what extent feelings can be aroused (Meertens 1952). In recent years especially the relationship between randstad [middle] and landrand [countryside, periphery, or nonurban areas] in the Netherlands has much improved through a growing awareness of the inhabitants of the northern, eastern and southern provinces because they were no longer so isolated and also because of social improvements and industrialization in these areas. The conclusions of the linguist will have had little influence on these facts, although it is perhaps possible to have a very slight, indirect impact. In Belgium, however, the relationship is obviously different. The dialect-speaking teachers-to-be, who during their studies hear about dialects and the science of dialects, will no longer feel like uncivilized dialect-speakers (as their older colleagues who were educated in the spirit of Terwey’s book). In this way a more objective, less emotional judgment of dialect and dialect-speakers can gradually become common knowledge. The above-mentioned social factors, and democratization in general, have prepared the way for this more objective view. How these various factors contributed is not clear, but a gradual change is obvious and can, for example, be seen in the fact that Dutch with a “regional color” is dismissed less vehemently and by fewer people than was the case before World War II. City-country changes are also gradually taking place, changes that are being promoted by altering social structures. The judgments of urban and rural varieties are independent of relationships in areas outside of language. Many nuances that appear in this domain would have to be studied jointly by linguists, sociologists and psychologists; this is not the place to consider such an enterprise in great detail.

DIALECTS

13

2.1 What Is a Dialect? By altering Den Hertog’s statement that all language contains something personal, one could say that each dialect contains something regional. But not all varieties that contain something regional are dialects, because Dutch can also have something regional without being called a dialect. In such cases the regional origin is most clearly discernible in the pronunciation; the words, word formation, declension, conjugation and sentence structure will less often diverge from Dutch. A clear border between dialect and regionally colored Dutch cannot be drawn. In the regions where the dialect is still spoken by many, their “language will” is the best criterion; in other words, the name given by ordinary speakers coincides with the variety they intend to speak. Divergence from one’s intention can then be considered a derailing, caused by one’s knowledge of a second variety. However, one can seldom if ever use this criterion in relation to the dialects of the large cities in the west and center of the country, not in Amsterdam, The Hague, Haarlem, Leiden and Rotterdam, and the surrounding countryside, nor in cities that are even farther removed from the urban agglomeration. In the heart of each population of any size, including towns, linguistic forms originate that are judged according to the social position of the groups that use them, based on the relationship among the groups. The less valued forms are considered vernacular (volkstaal, “folk speech”), a variety which appears to be a corruption of Dutch. The different forms of “folk speech” are, however, also regionally recognizable; the difference between the vernaculars from Amsterdam and Rotterdam is clear. But the characteristics of these varieties are to be found less on the level of the “language will” and more in the formation of habits. The vernacular speaker can very well avoid using a certain number of words that are considered less civilized, but it is not easy to wipe out peculiarities of pronunciation. The tendency to do so will in any case be checked by group consciousness. Because each dialect is bound to its own region and has characteristics that are also bound to its region, one can replace the word dialect with the less ambiguous term regional variety (streektaal) which is more comprehensive and less burdened with any contemptuous connotation. The above mentioned “vernaculars” can also be thus designated and included here: Frisian can also be called a regional variety, and even the so-called hypercivilized Dutch, which at least in its origins can be located in the Leiden-Hague region, can be included here. Dialects are therefore regional colloquial varieties in the broadest sense of the word, but with the exclusion of the different nuances of regionally colored Dutch. It is not possible to indicate in general what one means by regionally colored Dutch; in each part of the country it is dependent upon the characteristics of the dialect spoken there which resist the influence of standard Dutch. However, every Dutch speaker knows what is meant by this. It is a way of speaking which brings to mind the following thought: “One can still very well hear that he is from Limburg (or Groningen, Friesland etc.).” For

14

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

example, the soft pronunciation of the g and the fact that the n is pronounced so emphatically at the end of the verb forms are both very noticeable; in the case of other regions the intonation is an indication, etc. Instead of regionally colored Dutch, the term regionally colored A. B. N. [Algemeen Beschaafd Nederlands, “General Cultured Netherlandic”] has also been used. Although A. B. N. will be used as little as possible in what follows, it is nevertheless applicable here. If one simply pays attention to the features of regionally colored Dutch, one can hear that a speaker comes from one place and not another. But if one can very easily hear that a speaker’s father was of lower social status, it is more than simply regionally colored A. B. N which is involved. There are no fixed rules to be given; reactions are more or less subjective and dependent upon milieu, personal orientation, etc.2 2.2 How Many Regional Varieties Are There? When the question is asked, the person asking the question has to start with the assumption that regional varieties can be clearly distinguished from one another. But two regional varieties can in one instance differ from each other and in other instances coincide. For example, the variety spoken in Waterland to the north of Amsterdam coincides with Dutch in its phonemic structure,3 with the varieties spoken in Zaan and West Friesland in its vocabulary, and with those of Amsterdam and Amstelland in its pronunciation; the variety spoken in Weststellingwerf coincides in phonemic structure, at least according to Heeroma and Fokkema 1961, with that spoken in South Drenthe, but differs from it in a morphological phenomenon such as the plural form of the present tense verbs. Which criteria must then be taken into account in order to determine whether we are dealing with one or more regional varieties? Only if one knows the answer to this question will it be possible to identify the borders of the regional varieties, draw them on a map, and count them. We talk more extensively about the possibilities for division below after having provided an overview of what has been done in this regard, i.e., the history of the older dialect maps.

3.

A Little History

One can place the beginning of this history in 1852, when the Association of Dutch Literature in Leiden wrote out a prize question for a “speech map” with explanatory text. The appearance of the second edition of K. Benhardi’s Sprachkarte von Deutschland in 1849 perhaps led to this. Not a single answer was given to that prize question for the simple reason that no material was gathered to create the map. After the failure of the first attempts, several others failed, of which one can find a more detailed report in J. te Winkel’s Northern Dutch Dialects (Te Winkel, Part 1, 1899, 2ff). Only with the appearance in 1874 of Johan Winkler’s Dialecticon, containing the translation of the parable of the prodigal son in the dialects of Northern and Southern Dutch and of the Dutch-speaking

DIALECTS

15

part of Germany and of French Flanders, was there any material available, and this publication stimulated further attempts. In 1879, on the suggestion of its chair H. Kern, the Geography Association sent out a questionnaire to which 284 answers were given for 212 regions or places. But the processing of the material did not take place, due to a variety of circumstances. When in 1892 Te Winkel was asked, he stated that he was willing to work on these data, but while doing so he came to the conclusion that the material at his disposal was insufficient. This is not surprising when one understands that the Royal Dialect Bureau had access to 1,500 to 2,000 correspondents for approximately 1,000 towns and cities and that for many regions only incomplete data were available. Therefore a second list was sent out in 1895, differing from the previous one, which resulted in 209 lists for 194 regions or places, among which 171 were new. Te Winkel therefore now had data for 383 regions and places and with this had sufficient data for his dialect map (1898) as well as for the maps of â and î (1899–1901). When one compares his dialect map with both detailed maps one sees that the first is to a great extent composed of data from the latter two: a number of isophones4 of the â and î maps are also to be found on the dialect map, sometimes with small discrepancies. We must admire Te Winkel’s accomplishments. Not only because he did his work very accurately and thereby laid down one of the most important foundations for dialect geography of the Dutch language area, but especially because in later years he was seldom contradicted as far as important points were concerned, even though much material was added in order to refine and improve his divisions. In 1913 J. van Ginneken published a map which only diverges from Te Winkel’s as far as subdivisions are concerned; he chose other names for a number of dialects and established different links. The only objection one can make to these maps, when one takes into account the time at which they were made, is the fact that the designers do not justify their divisions. One would have liked to know how both scholars arrived at their choices. We know that we cannot draw lines as far as borders for dialects are concerned, but that the borders are made up of clusters of isoglosses and isophones, which sometimes run close together, but sometimes farther apart and thus criss-cross through each other so often that one can hardly speak of a border area. In 1895 Te Winkel writes: On account of the often (but not always) gradual transition from one dialect to another, it is extremely difficult to determine the salient characteristics that distinguish one dialect from another. Until now one is used to speaking of three main dialects, Frankish, Saxon and Frisian, and to acknowledge a mixed dialect in the regions where dialects no longer reign purely. (Te Winkel 1895, 51–70)

When he wrote this he was busy preparing the linguistic map, but in the process he came to the conclusion that more and better material was needed and that more thorough preparation was indispensable. His plan was to make detailed maps of vowel and consonant phenomena and a number of morphological and word maps. Of these only two vowel maps would appear, the â and î maps, with a detailed text entitled De Noordnederlandsche tongvallen (Te Winkel 1899–1901). This linguistic map, which appeared in his

16

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

Geschichte der niederländischen Sprache (Te Winkel 1898) and in the Dutch translation thereof (Te Winkel 1901), was but a temporary division. Te Winkel had probably chosen certain isophones and isoglosses as borders. However, he did not name them. Nor did Van Ginneken justify his divisions. On page 18 of his textbook he writes that he borrowed details for the borders of the Northern Dutch dialects from Winkler and Van Wijk (Van Wijk 1911) and that he had been able, for Southern Dutch, to draw on P. Willems’s material, which can be found at the Royal Flemish Academy. He does mention Te Winkel’s map, but not as a source, although he no doubt used it, because the similarities are often great.

4.

Method of Dividing

4.1 Isoglosses The many maps of word and sound phenomena, appearing hereafter in word atlases, periodical articles, and more extensive publications, have made us even more careful than Te Winkel was. In review of various efforts of less importance, I mention as the latest Weijnen’s dialect map (Weijnen 1966), in which he indicates eighteen isophones and isomorphs.5 He supplements this map with sixteen smaller maps, each of which details individual phenomena. The division of the Southern dialects — Western Flemish, Eastern Flemish, those from Brabant and Limburg — is more or less satisfactory. In the northern area, however, a troubled and therefore unclear image emerges. The creator of this map also did not justify his choice of phenomena, although he surely did not make them arbitrarily. 4.2 Phoneme Structures In early 1961 Structuurgeografie appeared, two lectures concerning structural phenomena given by K. Heeroma and K. Fokkema. Heeroma published with them a map of a dialect division of the northeast of the country, based on the system of the long vowels from the aa and ie series — also in Heeroma (1963). Several borders are found on it, including all accepted borders, such as those between Friesland, Kollumerland and Groningen. This map is only a first attempt; structural linguistic geography is still in its infancy, and Heeroma deals with only one part of the vowel structure. With this map he has indicated, however, that one must do further research in this direction. Further research was done by Goossens, who would provide the phonetic notation of dialect sound material for a phonemic division (1965). He came to the same conclusion even though, given the material at his disposal (the Dutch dialect atlases for East Flanders and Brabant and the Eastern part of the Zeeland region), his task was more difficult than Moulton’s had been because the latter had worked with the Swiss dictionary for his phonological geography of Swiss dialects. Does the division according to phonemic systems give satisfactory results? Before

DIALECTS

17

expressing a preference for a regional variety division according to phonological principles, one would have to show that the phonological characteristics are predominant compared to others. One would also have to take into account articulatory, phonetic, morphological, syntactic, lexical, and even semantic characteristics. I am afraid that it is not possible to furnish such proof and that it will be equally impossible to explain the interaction of these different factors. One can make a regional division according to phonological characteristics, as well as ones on the basis of phonetic and morphological phenomena, but these divisions will not coincide. 4.3 Dialect Unities In Section 40 of Weijnen 1966, he distinguishes different methods of division. The older ones, Van Ginneken’s and Te Winkel’s, he calls the intuitive. It seems to me that he probably does not do Te Winkel justice. Thereafter there is a division on the basis of demarcated isoglosses whereby the frequency of all close isoglosses represents a weightier criterion for recording, even if they are not of great importance from a linguistic point of view (192).6 As a third method he names the structural (193) and for a fourth reminds us of a statement by Heeroma that one must start with dialect unities and not with a number of loose characteristics (Heeroma 1942: 41, discussion of “De Nederlandse dialecten”). How must one determine the unities? Do we not then again end up with the intuitive approach? The division by non-specialists and linguistic scholars is very similar, not only if they do their work according to the intuitive method, but also when they use isoglosses and structures as characteristic features. Linguists try to take characteristics from a corresponding group of linguistic phenomena; non-specialists make divisions in terms of what they find most striking. In its least subtle form it is the pronouncement that Groningen speech is characterized by the pronunciation of the final n and the aspiration of plosives; the latter concept is unknown to the laypersons, but they demonstrate this phenomenon with the imitation of the word “Martsinitoorn.” In the same way the speaker with the soft g is identified as the Limburger. More balanced judgments exist among many Dutch people, and we need to ask ourselves what these are based upon. The concepts of regional variety and dialect existed long before historical and structural linguistics. In Kiliaan [Cornelius Kilianus, a Dutch lexicographer who published the first Dutch dictionary on scientific principles in 1574] it finds expression e.g. in lexical differences, in Lambert ten Kate [1723, an early Dutch linguist, 1664–1731, said to be well ahead of his times] in phonetic phenomena. There must be something which functions as dividing phenomena without bringing linguistic or philological insights to bear. 4.4 The Judgment of the Language User In order to discover how this unknown functions, the following question was included in Questionnaire 8 of the Dialect Commission of Amsterdam, sent out in 1939: 1.

In which location(s) in your region is the same or almost the same dialect spoken as in your own?

18 2.

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

In which location(s) in your area is a dialect spoken which is clearly different from the one spoken in your own? Can you name a few differences? (Vragenlijsten met register [Questionnaires with index] (1931–1958) 1960: 14)

The data for this list for North Brabant were processed by Weijnen, and a map of these data was published in 1946 [in Weijnen 1946: 14; reprinted in Weijnen 1966: 198; see Figure 0.2]. The dialect bureau drew a map of the whole Netherlands which was published in provisional form: 25 jaar Dialectbureau (In: Mededelingen der Centrale Commissie voor onderzoek van het Nederlandse Volkseigen 7, 1955: 22 [see Chapter 1, Figure 1.1, this volume]). One can learn about this approach in general in Weijnen’s Nederlandse Dialectkunde (1966: 195ff) and in his article in Lingua (1968 [see Chapter 9, this volume]). Both maps show that there was an important similarity between the division based on the dialect consciousness of the speakers, derived with the help of the material of Questionnaire 8, and that based on the isogloss and isophone strands, but there are also important discrepancies. When judging the differences one must not forget that, to determine the isoglosses and isophones, material was used that came from different sources and was collected many years apart. In some parts of the Dutch language area, the dialects have been subject to rapid changes in recent decades, changes which are not reflected in the material from earlier dates. A large proportion of the respondents of the Dialectbureau are older dialect speakers. They are unaware of these changes or have simply shut themselves off from them, because for them the dialect of their youth is the dialect. The changes have not been noticed, have been denied, or have been rejected as wrong. But even then the dialect was no unified entity because then, as now, regional varieties were not realized in the same way at different social levels. The material either omits these differences or presents them in a very inadequate or incomplete way. Dialectologists have known this and pointed it out but have, until now, collected very little material on the subject. In this regard it is informative to read what Kuitert wrote in his article Romantiek en realiteit (1965), from which I quote a few passages: Until a few years ago, I accepted the official opinion that in the province of Groningen, with slight variations, different local varieties have been spoken, which can be easily distinguished from each other. This unquestioning credence was probably supported by memories of pronunciation overheard during my childhood, such as “ze proaten doar hailendal aans as bie ons, zie zeggen mot en wie zeggen zeug.” [They talk entirely different there than we do; they say “mot” and we say “sow”; N. B.: The spelling is modified to reflect local pronunciation; Standard Dutch spelling would be “ze praten daar heel anders als wij, zij zeggen mot en wij zeggen zeug”; other lexical, morphological, and syntactic facts also reflect local usage — e.g., “hailendal aans bie ons” = “heel anders als wij”; “mot” is a local word for Standard Dutch “sow,” i.e., “adult female pig.”] (3) From the spoken language of Groningen I can’t conclude anything more than from the written: the seven dialects according to Ter Laan, weakly upheld by Schuringa, are not (any longer) to be found, but the differences between the dialects of Westerkwartier and Groningen are as clear as daylight; personally I tend to speak of the varieties of Westerkwartier and Groningen as of two separate ones. (13)

For further information I refer the reader to the article itself. But on p. 14 he continues.

DIALECTS

19

And yet anyone who listens with a sensitive, willing and practiced ear, can hear more or less where in the province someone speaking this denaturalized variety comes from. One can often hear it even when Groningers speak Dutch. What makes this possible? Is it rhythm or intonation (in whatever variation) or tempo? Is it a remnant of a phenomenon that one has not been able to observe, because one has been concentrating too much on that which can be observed by eye and ear, or has it developed over the last few years?

When the same Kuitert reviewed what remained of the Westerwold dialect, the area which before 1940 was principally oriented towards the bordering Eemsland, but between 1940 and 1945, the years of World War II, of necessity towards the city of Groningen, he came to the conclusion that the young people were hardly using it at all anymore. This is obvious because a speaker from Westerwold, coming to the city, will be treated as a “farmer” (a “boer”) and perhaps in the same way as the agrarian worker from the tip of North Holland visiting Alkmaar or Amsterdam, or as … you can complete the rest yourself. This is the result of the differences in social status. But many inhabitants of Groningen believe, according to Kuitert, that the Westerwold dialect still exists and on a few dialect maps these forms still appear (probably greatly outdated or used only by the oldest people), such as griggen on the map for names of greaves or cracklings and toffel on the potato map (Taalatlas, 1939–1987, part 6, map 10; part 8, map 2). Another example of a rejuvenation, which probably took place more gradually, is to be found in the province of South Holland. The æ-area, which Te Winkel drew on his â map, was in the 1960s, with the exception of a few relics, no longer to be found. In discussions about dialect similarities and differences, dialectologists, because of their need to divide and determine borders, also have the tendencies of the language user to make exact pronouncements or to interpret vague explanations in accordance with their own inclinations. Most dialectologists still come out of the Neogrammarian school, and they want to draw isophone and isomorph borders based on phenomena taken from historical grammar. However, these phenomena often have far less importance for language users and are complicated by “fanning phenomena.” A good example of this is given by Kloeke in his article dealing with the apocope line in the north-east of the country (1919). In Groningen, one could, when carrying out research, speak of a border, but farther south it became a zone because the area where the final e was retained in substantives was greater for some words than for others. The phenomenon of the disappearance of this e functions as a dialect border in the north but not in the south.7 Language users form their judgments in far less precise ways than the dialectologist. From the answers to question 2 cited above, nothing could be determined because on this basis the number of dialects in the Netherlands would be more or less equal to the number of densely populated areas, i.e. cities, towns and villages. Additionally, the answers to the question about the rationale or basis for decisions could not be used at all. And that is not so strange. I have been trying for years, going around questioning like Socrates during his time in Athens, to find out what allows an inhabitant of Amsterdam to hear that someone is from Kattenburg, Jordaan or another part of Amsterdam. By listening to the recordings made by J. Mittelmeijer, I can now hear the difference between the Amsterdammers of Kattenburg and Jordaan, but only through comparison. But I can, without any trouble,

20

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

distinguish from one another the inhabitants of Rotterdam and Leiden, Twente and Salland, West Friesland and Zaankant. Sometimes one word is enough to identify someone’s origin with certainty. Like the language user who is not a dialectologist, I cannot always put into words that which concerns me. The pronunciation of a single sound is sometimes sufficient, such as the way in which the Frisians articulate the s (Daan 1957), but more often it is a speech habit in general, e.g., the articulation, intonation, or stress. And even if I can provide a justification after the fact, I have the impression that my understanding and linguistic development lag far behind something else which has been refined through habit.

5.

Reconciliation of Opposites

These experiences have strengthened my conviction that the division of the regional varieties must start with the vague, yet real consciousness of the language users, but at the same time must try to provide a justification of the consciousness with the help of isophones and isomorphs that more often coincide than do word borders, despite the fact that a dialect difference can sometimes be characterized with one word. In many cases the justification is quite satisfactory, in other cases less. The divergence will exist because we have not yet brought sufficient phenomena to the task, especially phonetic ones such as the aspiration of voiceless plosives, the diphthongization tendency of the long vowels, the extent of voicing of the “weak” g, and assimilation (Leenen 1954; Van Haeringen 1955; Kloeke 1956). These phenomena may play a more important role in dialect division than the isophones and isomorphs established thus far. Dialect geographers have collected data for the larger regions primarily via the indirect, written method and have published these data with the help of written symbols, which for the purposes outlined here are very inadequate. As a result, distinctions have been exaggerated. Here are a few examples to illustrate what I mean. In order to represent the West Germanic au (Dutch ô) before gutturals and labials, Karsten spells it ou (Karsten 1931: 40). When transcribing recordings for the Atlas of North Holland, I only used the notation [fu] a few times, but mostly I used [o] or [œo], followed by a weaker or stronger nasal which sounded like an unclear vowel, sometimes u-like (Daan 1969). In the meantime I found a second example in J. Naarding (1947). He provided me with a boundary for Groningen diphthongization, the different from those on the dialect maps with written investigation material, and he explains this with the statement that this diphthongization in the south has not only been recent but is moreover so slight that the speakers are not aware of it. By the way, what western so-called A.B.N. speakers know that they are pronouncing the ee and oo of keek and kook as diphthongs? When designing dialect maps of the â in the words schaap and laten, which is realized differently in some parts of the Dutch language area, it appears that the written material could not be interpreted unless it was phonetically transcribed.

DIALECTS

6.

21

The Map

6.1 The First Design A map of dialect divisions according to the awareness of language users, which served as a basis for further research, was based on the same methods Weijnen applied to the Northern Brabantine area (Weijnen 1946, 1966: 196 [see Figure 0.2, this volume]). When a correspondent responds to the first question on Questionnaire 8, that the dialects of B and C are the same as that of their own area A, an arrow is drawn from A to B and C. When in D the same dialect is spoken as in B and E, these places are linked. In this way larger units emerge where the same dialects are thought to be spoken; the dialects of Bildt (in Friesland), Zeeland, and Limburg emerge without any complications. Sometimes borders from the emerging dialect areas coincide with well-known isophones and isomorphs, e.g. the Uerdinger and the mich-line, the Benrather line, a part of the ui/oeborder in the south-east, and the border between the area with or without umlaut in the diminutives ending in -ien to the north of Overijssel. But Kloeke’s diphthong line (ui/uu) of Spakenburg from the south-east up to the river Waal does not seem to be a dialect border according to the judgment of the language users.8 6.2 Relationship of Linguistic and Other Criteria Starting with this map [i.e., a map derived from the “little-arrow” method; see the Introduction, Figure 0.2 and Chapter 1, Figure 1.1, this volume], I have tried to find or determine isophones and isomorphs based on this division or, in the case of divergence, to see what the causes could be. An example of a clear and sharp border is that between the varieties of Friesland and Stellingwerf on the one hand, and Kollumerland and Groningen on the other. The border of Friesland is based on old linguistic opposites, but it still exists. The differences between Frisian and the bordering regional varieties spoken in Stellingwerf and Kollumerland (both also spoken in part of the province of Friesland) are so large that the separation works; the linguistic difference is indeed the most important criterion here. A pure linguistic border is the one between Bildt and the Frisian-speaking part of Friesland. In Bildt they speak a Dutch which in articulation and also in other ways has undergone the influence of Friesland. The linguistic difference is great, but there are no important differences in other areas. Kloeke sees the emergence of Bildts in the cultural dominance of the Hollanders (Kloeke 1927: 61–75). The existing border between Zeeland and North Brabant is an equally clear linguistic border, backed up by isophones, etc., but this one coincides with a separation between Protestants and Catholics. More than 80% of North Brabant’s population is Catholic; in the Zeeland islands more than 70% of the population is reformed or Calvinist. Here then, in addition to a dialect opposition, there is also a religious difference. Also in Zeeland Flanders and between the north-west corner of North Brabant and the remaining part of this province, the border between jij and gij is the same as that between the great

22

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

concentrations of Protestants and Catholics. This linguistic border must coincide with the historical events of the 17th century. Between Arnhem and the German border the jij/gij border also coincides with a religious border, but, in the region in between, this religious border corresponds to a part of the border between the doeget/heget border. This area is never crossed by an arrow; i.e., the judgment of the language users wholly coincides with Hol’s (1965), who established the doeget/heget border. When the dialect differences are smaller, the doubt as to whether other criteria should be given more weight than dialect differences is justified. Examples of this are to be found in North Holland. The linguistic differences between the varieties spoken in the Zaan region and West Friesland are not great; at least they were not great when Questionnaire 8 was sent out. And the similarities in vocabulary are great. Both inhabitants of the Zaan region and the West Frisian region emphasize the differences more than is justified from a linguistic viewpoint. One of the differences mentioned is the prefix of past participles: Zaan has e- as prefix, West Frisian has no prefix (in Zaan: elope, in West Frisian: lopen, Dutch: gelopen). I imagine that the linguistic differences are accorded more weight due to the differences in economic and social structure. The industrial Zaan area does not identify with agrarian West Friesland and vice versa. Also for the varieties of Zaan and Waterland the non-linguistic differences seem to be more important; until recently Waterland’s principal means of subsistence have been cattle raising and fishery in the coastal towns (Daan 1955). After 1648, the area where Dutch-Brabantine (Westhoeks) is spoken belonged not to the States General but to Holland. In Beekman’s work (Geschiedkundige atlas van Nederland [1915–1929], De Republiek der Vereenigde Nederlanden in 1648, pp. 1 and 2 [1919]) this possible seat for Dutch (Hollands) is larger and does not lie in the same spot as calculated for our map. These displacements could be the result of later circumstances, among which migration (including that resulting from marriage) had an important influence. Another example is the dialect of Overflakkee. It is considered a Zeeland dialect, but there are differences from the southerly situated Zeeland islands and simularities with the northerly situated islands Voorne and Putten. The political ties of Overflakkee with Voorne and Putten were strong whereas with Zeeland they were weak. The differences among the dialects of Voorne and Putten may also result from an earlier condition when Voorne belonged to Zeeland and Putten to Holland (Landheer 1955: xviii). This is not to say that the political bond exerted a greater influence. But social contacts and cultural influences resulting from the political situation could have had a great influence. Weijnen wrote in detail about the possibilities and difficulties of a comparison between dialect borders and other borders (Weijnen 1966: 14–23a). He draws our attention to the fact that dialect borders seldom, if ever, completely coincide with other borders. But I cannot get rid of the impression that the similarity is very significant, even though there is not complete overlap.

DIALECTS

23

6.3 Controls and Supplements Great similarity between linguistic boundaries based on data from Questionnaire 8 and those determined by scholarly means can also raise doubts about the criteria for other reasons. Take, for example, the division of Friesland and Limburg. The division of the Frisian area corresponds to Hof’s map, except where these isoglosses come together. In this area, where the inhabitants of Stellingwerf live under Frisians and the dialect of Heerenveen influences that of the schoolchildren, the borders are less sharp. This becomes apparent because the information provided by the correspondents does not always coincide. Hof writes about this uncertainty on pp. 21–24. We can ask ourselves whether the correspondents who responded to Questionnaire 8 were aware of Hof’s division and kept track of that when completing the questionnaire. Some of them probably did know the book; it appeared in 1933 and approximately half the correspondents in 1939 (when the list was sent out) were teachers. The Frisians among them had surely read Hof’s work or owned the book itself. In addition, interest in Frisian was increasing due to the establishment of the Fryske Akademy in 1938. However, it is no longer possible to distinguish whether these data were dependent on Hof’s or not. One can anticipate such doubts if one studies the data for Limburg more closely. The isoglosses of Ramisch, already described by Schrijnen in 1902 and which he drew up for Dutch and Belgian Limburg (Schrijnen 1902) and printed with bright red lines on Van Ginneken’s (1913) dialect map (for years the prime example demonstrating that sound laws were not always without exception), would have been known to many Limburger correspondents. But the resemblance was striking also in areas for which in 1939 no or only less generally known studies were published. The border between Drenthe and GeldersOverijssel, drawn up in the area of Regge, completely coincides with the border between the diminutives ending in -ien with or without umlaut (Kloeke 1919). Small maps with dialect borders from the eastern area of Overijssel had appeared shortly before in 1948 (Bezoen 1948: 194, 198); the doeget/heget-isogloss was published in 1965 (Hol 1965). These examples make it probable that division based on material obtained from correspondents can be trusted. Both the coinciding and non-coinciding borders were controlled in our research in a variety of ways over time: conversations were held with dialectologists who knew the area in question well or who spoke the dialect themselves; during the making of recordings differences and similarities in regional varieties were discussed; literature was used, both from interested laymen as well as from specialists; and where such sources were not available, personal research was undertaken, sometimes with a shibboleth as guide. Where the division indicated by correspondents from Drenthe was not usable (only a few had responded) I, in conjunction with J. Naarding (1947), took the pronunciation of the a in water as a criterion. The border for the Groningen diphthong is also based on his findings, which he shared with me in a letter in 1958. The Southern Holland-Utrecht area provided the greatest difficulties. “Intuitively” there should be a border between the two regional varieties, but there was no clear

24

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

criterion to be found. Finally the radio provided a solution with the name Bartels, which in order to typify Utrecht was pronounced Baartels. From a phonetic examination it appeared, in addition, that this pronunciation of a belonged to the resistant characteristics of Utrecht. It was, however, not possible to determine the border between the South Holland and the Utrecht a by means of the questionnaire material. This was finally done with the help of sound recordings. This material in “South Hollands” was accounted for in 1965 (Daan and Heeroma 1965: 18). Later it could be supplemented, especially in the northern area with many words. The most contestable datum is the course of an isogloss which first runs east-west about halfway between Amsterdam and Rotterdam, to the north of which a phonetic distinction can be made between the lengthened and original long oo before dentals on the one hand, before labials and velars on the other; to the south of which a difference in pronunciation is made between lengthened and original long oo, thus an etymological difference. In the east the border is still to be established; in the west these isoglosses only provide the farthest northern border of the etymological distinction. It was still clear in Katwijk and Rijnsburg, but further east, near the Oude Rijn (to the north of the isophone), only in relics such as brood and dood were traces of differences as to phonetic position still to be found. Around 1900 Gouda still belonged to the region where the etymological differences were still made (Lafeber and Korstanje 1967), and in Alblasserwaard, the eastern area of this phenomenon, speakers were still clearly aware of it. It is the only isophone in our work which cannot be explained along its entire length, but it is important enough from a historical viewpoint not to be omitted; it is also the only one which runs through the area in which one can hear a clear distinction between the regional varieties to the north and south of the Oude Rijn, apart from much more recent phenomena such as the addition of t in the first person singular of the present and ie as pronoun for the second person singular in inversions. For these phenomena, which also appear to the north of the Oude Rijn, it is equally difficult to establish precise borders (Daan and Heeroma 1965: 4–24). 6.4 The Final Stage Based on considerations of these multifaceted and careful controls, the borders of the regional varieties are still in place in our present work. They are formed by isophones and isomorphs or parts thereof, inasmuch as these are relevant as dialect distinctions. Only a few isophones and isomorphs were continued when a subdivision was desirable. For more detailed data, and further divisions, I refer you to the monographs and the geographical dialectology literature, especially to Weijnen 1966. 6.4.1 The Regions In the division of the regions the starting point was the above mentioned principal characteristics: the presence or absence of the final -n and the jij/gij-border. The southern area is that of gij (approximately that of the soft g), the northeastern that of the final-n. The Frisian-speaking area is distinct; the remaining area, that of the western dialects, is only slightly different from the urban agglomerate (or Randstad).

DIALECTS

25

The largest percentage of the Dutch-speaking population, or those whose spoken variety seldom diverges in many phenomena from what one generally considers to be Dutch, lives in the Randstad. The rural dialects have practically disappeared or continue to manifest themselves only in a few pronunciation relics and in articulation and intonation in general. This urban area gradually shades into slightly different varieties, which in the south are bordered by Haringvliet, in the north by the IJ [the canal from Amsterdam to the west, to the North Sea], and in the East by the a/aa-isophone. Towards the south and the north these differences become more intense. The differences from the Dutch of the Randstad become greater, and the dialects become less understandable for the monolingual, Dutch-speaking inhabitant of the Randstad. Towards the east the differences from the Randstad increase. Utrecht forms more of a transitional area between west and east rather than having its own clear-cut characteristics, because important isophones run through here. Nevertheless Van Veen did find (1964) a few phenomena to characterize the dialect. To the east the final-n isogloss, bordering on Utrecht, begins the so-called Lower Saxon area. The isoglosses in our research have been extended over the national borders if the phenomenon is also to be detected on the other side of the border, in order to make it clear that national borders are not dialect borders. The material east of Overijssel and Gelderland was provided by Dr. F. Wortmann of Münster; that for Limburg was taken from Schrijnen (1907).9

26

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

EM BL FRIESLAND

KL GRONINGEN WK GR

HV ST WS

NORTH

DRENTHE

HOLLAND WF

AL ZA HA

WT

FLEVOLAND

OVERIJSSEL

AS AM

RG SL

TW

SP KT AF GELDERLAND LE OR HOLLAND RJ WR UT AR TH GO UTRECHT HV VR RO WA AB PT OV WH NORTH BRABANT SOUTH

ZEELAND

GERMANY LIMBURG

BELGIUM

Figure 2.1. Sites in The Netherlands referred to in Chapter 2

WW

DIALECTS

27

Key to Figure 2.1: Provinces are given in outline type (e.g., FRIESLAND); other regions, communities, and features are given in two-letter abbreviations: In Groningen EM Eemsland (in Germany) GR Groningen (city) WK Westerkwartier WW Westerwolde In Friesland BL Bildt HV Heerenveen KL Kollumerland ST Stellingwerf WS Weststellingwerf In Overijssel RG Regge (river) SL Salland TW Twente In Gelderland AR Arnhem In North Brabant WA Waal (river) WH Westhoeks In North Holland AL Alkmaar AS Amstelland AM Amsterdam HA Haarlem WF West Friesland WT Waterland ZA the Zaan region

In South Holland AB Alblasserwaard GO Gouda HV Haringvliet (estuary) KT Katwijk LE Leiden OR Oude Rijn OV Overflakkee (island) PT Putten RJ Rijnsburg RO Rotterdam TH The Hague VR Voorne WR Woerden In Utrecht AF Amersfoort SP Spakenburg UT Utrecht (city)

28

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

Figure 2.2. Dutch dialect areas, perceptual and production data combined, with increasingly darker areas showing greater divergence from “standard” Dutch (adapted from Daan 1970)

NOTE: To make the black-and-white shadings of the areas as distinctive as possible, I have divided the map into six ‘zones’ as follows: 1 REFERENCE 1 South Holland dialect 2 NORTH (cross-hatch) 2 Kennemerland dialect 3 Waterland dialect 4 Zaan dialect

DIALECTS

3

4

5

6

29

5 Northern North Holland diaelct 28 Dialects of Het Bildt, Frisian cities, Midsland, and Ameland Island (N.B.: The several lines pointing east from the 28 on the map indicate the several cities [white spots] within the dark 27 area (Friesland) where ‘Frisian cities’ dialect is spoken — Harlingen, Franeker, Leeuwarden, Dukkum, Drachten, Bolsward, Sneek, Staveren, and Heerenveen.) 27 Frisian NORTHEAST (left slant lines) 23 Southern Drenthe dialect 25 Kollumerland dialect 22 Stellingwerf dialect 24 Central Drenthe dialect 26 Groningen and Northern Drenthe dialect EAST (horizontal lines) 6 Utrecht and Alblasserwaard dialect 18 Veluwe dialect 19 Gelderland and Western Overijssel dialect 20 Western Twente and Eastern Graafschap dialect 21 Twente dialect SOUTHWEST (right slant lines) 7 Zeeland dialect 9 West Flanders and Zeeland Flanders dialect 10 Dialect between West and East Flanders 11 East Flanders dialect SOUTHEAST (dots) 8 Dialect of the region between Holland and Brabant 12 Dialect of the region between Flanders and Brabant 13 Dialect of the river region 14 North Brabant and Northern Limburg dialect 15 Brabant dialect 16 Dialect between Brabant and Limburg 17 Limburg dialect

Notes 1. Borders of the same word, as well as of a linguistic phenomenon in general. 2. Editor’s note: A phonograph recording (included with the original) is briefly described here (see the remarks below). 3. Vowel and consonant structures. 4. Borders of the same sound phenomenon. 5. Borders of the same morphological phenomenon. 6. The statistical method is used here and there. A recent publication is that of Houck (1967). Reference to Pavle Ivic is made in this work. 7. One gets an impression of the young, cycling Kloeke’s field work when reading his diary, published by M. J. Kloeke — van Lessen in 1968. 8. Kloeke (1927) (with the muis/huis map). The same border of î diphthongization is to be found in Te Winkel (1901). 9. Editor’s and translator’s note: This selection is taken from the book Van Randstad tot Landrand (Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen

30

THE DUTCH CONTRIBUTION: “LITTLE ARROWS” te Amsterdam XXXVII) by Jo Daan and D. P. Blok, pp. 7–43. In a few sentences, the English summary of the original (42–43) was relied on. Figure 2.2 does not show areas outside the Netherlands. A phonograph recording illustrating Dutch dialects and a full-color map (showing the perceptual boundaries combined with important elements of the production boundaries and many other dialect features outlined in the text) were included with the original, but for technical reasons they could not be reproduced here. I have prepared a version of the map in black-and-white (Figure 2.2). In it, the increasing darkness of the areas indicates the degree to which they differ from “standard” Dutch (the area of South Holland, #1). In the original color map, the relationship among the areas was also indicated by shading, but that, unfortunately, cannot be shown here. (Entjes 1974 produced a numbered version of this map but, unfortunately, misunderstood the numbers to refer to the linguistic distance of each area from South Holland [“De oplopende getallen geven ook een relatief grotere afstand tot het Nederlands aan”], but as Figure 2.2 and the original show, that is not the case. The numbers simply identify the different areas.)

C 3 The Netherlands-German National Border as a Subjective Dialect Boundary Ludger Kremer Translated by Jennifer Dailey-O’Cain

1.

The National Border as a Language Border

Our insights into the structure of the continental West Germanic languages come mainly from studies that were carried out around the end of the nineteenth century or in the first half of the twentieth century. These usually focused on understanding the basic dialect structure as it was spoken in its “purest” form in the mouths of the oldest living inhabitants — that is, a language as uninfluenced as possible by the Dutch or German standard language. As long as it was a matter of taking stock of geographically based lexical differences, researchers in this tradition usually asked about old-fashioned words having to do with animals, plants, farming, or rural crafts which even then were dying out. This is still the case in studies being carried out today, for example in the case of the Dutch Taalatlas — if the occasion arises that linguistic changes under the influence of the standard language do occur, they “are not noted [and are] ignored or discarded as incorrect” (Daan 1970: 22 [see Chapter 2, this volume]). A result of these older dialect studies was a picture of the “unshifted” continental West Germanic language region as a dialect continuum between Dünkirchen and Königsberg, which, as a horizontal linguistic system,1 exhibited more or less meaningful differences between regional dialects or between the individual language regions. However, it never revealed a border between the Netherlands and Germany which it would have been necessary to recognize as a language boundary. The Netherlands-German border appeared as a dialect boundary only along short stretches, and this was often the result of recent leveling procedures (Kremer 1979: 109). This picture changes if we widen our scope and include in the study dialect speakers of the younger generation, the language usage of less dialectally fixed and more geographically mobile speakers, or lexical items of more recent origin. In all of these cases, the German or Dutch standard language enters our field of vision as an interfering factor.

32

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

The observation of interference from the standard language and transference into the dialects is certainly not new. Even one of the founders of traditional dialectology, Philipp Wegener (1880, 464), called for the examination of the process of dialect change as a result of standard language interference, and later dialectologists such as Maurer (1930), Bach (1950: 250), and Henzen (1954) addressed these problems quite thoroughly. Yet research into these issues has made notable progress only since the emergence of sociolinguistics (Mattheier 1980: 9). While the entanglement of dialects with the corresponding standard language became part of the definition of the term “national language” in the west Slavic area (Belic 1965; Chloupek 1980), this usage is only valid with respect to the demarcation of Dutch and German dialects in very recent times (Kremer 1983). The first studies of language change based on linguistic material from the Netherlands-German border region, for which older dialect-geographic works form a foundation as well, have found the national border to be a (nearly completed) language border. A stronger incorporation of the expanded methodology mentioned earlier (younger generation, linguistic registers above the basic one, modern cultural words) can only confirm this result and strengthen this tendency which can be recognized based on studies in the older traditions. The linguistic reality shows the Netherlands-German border to be a language border at which two (vertical)1 linguistic systems meet. These are certainly related, but they have become distinctly different from each other. However, linguistic borders are not simply a concern of academic knowledge — that is, linguistics. They are also a part of general knowledge — that is, the consciousness of the speakers, if only in rudimentary form. It is interesting in such a situation to explore whether and in what ways the NetherlandsGerman border appears as a linguistic border in the minds of the speakers.

2.

Subjective Dialect Boundaries

We refer to language varieties as they are perceived by dialect speakers on the basis of these speakers’ own experience or on the basis of general knowledge as subjective dialect boundaries. Weijnen (1946, 1961, 1968 [see Chapter 9, this volume]) presented seminal work in the research of such boundaries, in which he mainly concerned himself with methodological implications by grappling with the methods of other such studies. Many methods have been used in the determination of subjective dialect boundaries. In a folkloristic and dialect geographical monograph written as early as 1939, Büld attempted to apply the popular knowledge about linguistic differences found in caricatured verses about language to a classification of the north Westphalian dialects. Other scholars have since criticized this work: first of all, according to Büld’s own observation, not every important dialect border can be found in caricatures; second, Büld used an unjustified selection of his material in order to establish this classification; and third, it is often the case in caricature that linguistic occurrences are addressed which no longer exist (Weijnen 1961, 1968).

A SUBJECTIVE DIALECT BOUNDARY

33

Weijnen (1946) introduces the “little arrow method” as an appropriate method with which to determine subjective dialect boundaries. This method asks dialect speakers questions about which local dialects near the place where they live are approximately the same as the ones which they speak. All of the corresponding dialects are then connected with the informant’s town by means of an arrow on a map, so that a picture emerges which looks much like a net or a chain. Dialect borders are then formed by the remaining white spaces. By emphasizing similarity, Weijnen (1961, 1968) wards off Grootaers’s (1959c, 1963, 1964 [see Chapter 8, this volume]) criticism of the little arrow method. In order to establish the subjective dialect boundaries in Japan, Grootaers did not ask about knowledge of similarities, but instead asked about how the neighboring dialects were different from each other [see Chapter 4, this volume]. As a result of this, Grootaers reached unsatisfying conclusions. Because minimal differences exist almost from village to village, and because it would demand too much of the informants to expect a weighting of these differences’ relative importance, a thick, unusable honeycomb pattern emerges when using this method, a pattern which is completely different from the isoglosses dialectologists have established by means of a corpus. Weijnen’s method, developed through attempts to classify regional varieties, was also used by Rensink (1955 [see Chapter 1, this volume]) and Daan (1970) in the language region of the northern Netherlands. The latter of these two used the map obtained by the little arrow method as a basis for a map classifying Dutch dialects which also took isophones and isomorphs into account.

3.

The Map

The map which appears here [see Figure 3.1] results from the answers to the question I asked my informants: “In which town or towns which are close to yours do they speak the same or approximately the same dialect as they do in your town?” This research was carried out as a part of a lexical dialectological study for the map segments represented by p to z from 1974–1977 and for the segments A’ and B’ in 1980. One informant from each larger village was interviewed, and the majority of the informants ranged from 30 to 50 years of age. Often neighborhoods or smaller settlements were mentioned in the answers which were not in turn represented by informants. In these cases the answer was evaluated, but the relevant town or part of a town was represented exclusively by a dot. The informants’ towns are represented by symbols which, for the German part of the region in question, are in agreement with the main map of the Westphalian Dictionary. If, according to the opinion of an informant in town A, his town’s own dialect is (approximately) the same as the one in the neighboring town B, this relationship was expressed by means of an arrow leading from A to B. If the informant from B was also convinced of the similarity of these two local dialects, an arrow was drawn in both directions. Possible uncertainties about the degree of similarity can then be found in oneheaded arrows (which, of course, is not true for the smaller settlements which are represented only by dots). If the dialects are only similar when used by the older

34

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

Figure 3.1. Subjective dialect boundaries in the western part of Westphalia

A SUBJECTIVE DIALECT BOUNDARY

35

generations (approximately older than 50 years of age), the arrow was drawn with a dotted line. If an informant felt that his dialect was not similar to any of the surrounding dialects, the corresponding symbol for the town was circled.

4.

Interpretation of the Map

A first glance at the map shows two things. First, the informants in the German portion of this area recognize subjective dialect boundaries more clearly than the informants in the Dutch portion. Second, the national border is crossed by an arrow in only five places, and each of these is only a one-headed arrow. On first observation the impression I developed during conversations with the informants is confirmed: the Netherlands part appears to be organized into larger groupings than the German part. It contains more one-headed arrows, an occurrence which points to a greater uncertainty of the informants or to smaller differences between the dialects. The independence of the relic dialects (Kremer 1979, 1:89) of Wintersijk (Ww, x 7) and of Vriezenveen (Vv, s-r 6); of the dialect of Vroomshoop (Vh, r 6), which is a combination of the Hollandic, Drentic, and Twentic dialects; and of the dialect of Coevorden (Cv, p 7), which is a Drentic offshoot, are easily recognizable. Less recognizable is the group of Vechte dialects in the northwest corner of the map (p-q 6), the border between Twentic and Achterhoekic dialects along the province border (u-v 6–7), and the separation of the Regge region near Rijssen (Rn) and Enter (Et) at the western edge of the map (s-t 6). The differentiation of eastern and central Twentic dialects (s-u 7–8) can only be established on the basis of a very small number of twoheaded arrows. This relatively small differentiation of the map’s picture may be explained by the emergence of regionally leveled dialects, or at least by a corresponding regional consciousness, the existence of which is also recognizable by the fact that the regional dialects all have different names (Achterhoeks, Twents, Sallands, and Drents), in contrast with the undifferentiated German term “Platt.” Unlike the German part of this area, the Dutch part exhibits a higher level of transference from the standard language, for which the greater structural similarity between these dialects and the Dutch standard language is likely responsible (Kremer 1979, 1:194). The German part of the map mirrors the familiar divisions of the Westphalian dialects of this area quite exactly, although the uniqueness of the city dialects of Coesfeld (Kf, x 9–10) and Burgsteinfurt (Bu, v 10) mentioned elsewhere (Kremer 1979, vol. 2, map 19; 1983b, map 5) is not expressed here. Uncertainty arises with respect to the differentiation among west Vestic (A’-B’ 7–8) and east Vestic dialects (A’-B’ 9–10) in the area just north of the Ruhr region. This is most likely explained by the fact that my informants in that area did not have very much contact with other villages’ dialects, since the use of regional dialects is generally reduced to a minimum there. It is obvious from the fact that few arrows cross the German-Netherlands national border that this border is also a language border in the minds of the dialect speakers. However, its significance for Dutch and Low German dialects as a whole cannot be

36

THE DUTCH CONTRIBUTION: “LITTLE ARROWS”

discerned from this fact, just as it is also not possible to discern the relative weights of these subjective dialect boundaries in relation to each other. A change in these border residents’ view is observably connected to the cognizance of the national border as a subjective dialect boundary, because the few remaining arrows suggest an earlier feeling of solidarity or togetherness, at least in parts of this area. The dotted arrows in the cases of Aalten (Aa, x-y 6) and Oud-Schoonebeek (Os, p 8) provide evidence for the explicit explanations of the informants that the similarity between the two dialects exists only for the older generations. In the case of the doubly identified town of Wielen (q 7) along the western edge of Bentheim, the Dutch residents (Wi1) emigrated approximately three generations ago from a nearby Dutch area and still have much contact with that area, some of which takes place in standard Dutch. The German minority, on the other hand (Wi2), feel more connected with their neighbors to the east.

5.

Results

The map of subjective dialect boundaries shown here is meant to indicate to what degree transference from the standard language has become apparent to the speakers. It has been determined that in light of the national border, harmony between the linguistic-geographical reality and speaker knowledge exists. The experience of the difference of the dialects on each side of the border is well-known. Many speakers seek to use the dialect as a means of communication when they cross the border, which they do not do during contact between towns in their respective countries (Kremer 1979, 1:138), although it is increasingly the case that speakers use the standard language (usually German) when crossing the border. In their subjective evaluation of the national border as a language and dialect border, the people who live in this area are well ahead of linguists, who can only produce partial results in the evaluation of insights into objective dialect boundaries.

Notes 1. Editor’s note: Kremer apparently means by “horizontal” and “vertical” systems those viewed as continuous (i.e., without a “sharp” break) by the former and those viewed as autonomous by the latter.

P II The Japanese Controversy ‘Subjective’ and ‘Objective’

C 4 Consciousness of Dialect Boundaries Takesi Sibata Translated by Daniel Long

1.

Dialect Boundaries and Their Perception

Misao Tôjô, one of the pioneers of Japanese dialectology, in his Hôgen to Hôgengaku [Dialects and dialectology] states the following: The feeling that a dialect is different is something which everyone can relate to. Even if this is not a clear perception based on specific criteria, one may still have a vague feeling. In our country’s language, no one can deny the perception of a difference between the Kansai and Kanto dialects. (1938: 29)

If we refer to this “feeling” as “dialect consciousness,” we find Tôjô maintaining that this dialect consciousness should form the first basis for the establishment of dialect regions. He writes the following in Nihon Hôgengaku [Japanese Dialectology]: The only way to establish dialect regions is to begin with the dialect consciousness of ordinary people and make additions and corrections on the basis of surveys of those special linguistic characteristics which are the basis for these perceptions. (1954: 11)

From this passage, it appears that while Tôjô expects that the regions created by dialect consciousness and those created by objective isoglosses will basically match up, he also anticipates that this overlapping will not be perfect. But even without a perfect match between the two, wouldn’t we expect to find a degree of similarity between them? If so, then in order to determine dialect regions (hôgen ryôiki), we would need only to investigate dialect consciousness, and investigating (the distribution of) individual words would be unnecessary. What if we surveyed the dialect consciousness in every hamlet of an area by a single method? This type of survey has never been carried out before in Japan. It had not been conducted in Europe either until recently. I have obtained information that a survey of this type has been conducted recently in the Netherlands. In Pauwels (1956: 113–114), it is reported that someone named W. G. Rensink (1955) performed such a survey [see Chapter 1, this volume]. In a postal questionnaire, he asked “What dialects are the same

40

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

as your own?” “What dialects are completely different from your own?” and reportedly produced the results in map form. Regrettably, at the time of this writing, we have been unable to obtain any information other than this.

2.

The Survey

In the fall of 1957, we conducted a dialect survey of Itoigawa City and Ômi Town which are separated by the Oyashirazu Promontory.1 Along with about sixty linguistic (grammatical, phonetic, accent, and lexical) questions and about ten questions on society and religion, we also investigated dialect consciousness. The survey locations consisted of every hamlet in the survey area — all the hamlets shown on a 1:50,000 map of the Geographical Survey Institute. With the inclusion of a few hamlets from neighboring Nagano and Toyama prefectures, the total number of survey locations (with one informant per hamlet) came to 164. It goes without saying that this was not a random sample, but an “on the spot” survey. Having no previous studies to consult, we had to come up with the wording of the questions ourselves. We put the following question to each informant and wrote down the names of hamlets or areas which were given for each of the four choices.2 Is the speech here from the speech of the neighboring hamlet? (with the following choices in order of difference) 1. 2. 3. 4.

not different a little different quite different mostly incomprehensible

As a rule, our informants were native males over sixty. (Natives are defined here as those who spent all of their language formative period years in the hamlet and after that had not lived away from the hamlet for more than a total of six years.) These people were not dialectologists, but in their sixty years in this hamlet, we can assume they have had numerous opportunities to notice dialect differences either in their own experiences or through hearsay. However, our queries about “dialect consciousness” did not always seem to register with the informants, and some were not able to respond to the question at all. Even among those who did respond, we had answers such as “everywhere is the same” or “the way I talk can be understood everywhere.” This was probably due to the fact that two of the three researchers (Sibata and Munemasa Tokugawa) belonged to the National Language Research Institute and people had the impression that we had come from the Ministry of Education to do a survey on Standard Japanese. The third member of the survey team was a foreigner (Father Willem Grootaers), and it is possible that people gave this sort of answer out of consideration towards him.

CONSCIOUSNESS OF DIALECT BOUNDARIES

41

These ambiguous responses were, however, in the minority; all but a few respondents answered our question clearly. Nevertheless, we felt that dividing the answers into three distinct groups was problematic, and when we created the maps for this question we decided to use a distinction between “somewhat different” (answers 2 and 3), and “mostly incomprehensible” (4). In this way, we posited that “dialect consciousness” was dichotomous in structure with a distinction between the world of things close versus the world of things far away. The former we called “Class I dialect consciousness” and the latter “Class II dialect consciousness.”

3.

Class I Dialect Consciousness

3.1 The Consciousness Wall Here we will explain how we dealt with the informants’ responses and how we illustrated these on the maps. We should point out here that because of the very nature of a “dialect consciousness” survey, we had to be very flexible in our approach to the data analysis. First, look at the left half of Figure 4.1. In the upper left, the number 92 marks the hamlet of Niôdô.3 Henceforth, we will refer to this as location 92. The informant at location 92 answered as follows: Gozenyama is somewhat different from us, and they say the way we talk is a little different from Kotaki as well. Their speech is bad, and they have some of the Nagano dialect mixed in. When you go to Ômi, they are quite different and their talk reeks of Toyama.

We treated this answer as follows. First, we concluded that there was no location which fit into the “mostly incomprehensible” category. The Gozenyama mentioned here is 3.5 kilometers east of Niôdô at point 37. To show that the informant from 92 perceived a dialect boundary between himself and point 37, we encircled 92 and put one peg pointing east in the direction of 37.4 To correspond to this, we drew a bar in front of 37 with one prong (pointing towards 92). This bar shows that there is one hamlet which perceives a dialect boundary at that place. Next, we looked at the consciousness of Niôdô (92) towards Kotaki. Kotaki is the name of a former village,5 and corresponds to locations 70, 52, 60, 29, 21, 35, and 72 on Figure 4.1 as well as to locations 08 and 63 not shown on the map.6 As above, we added a peg to our circle around 92, this time in the direction of Kotaki (to the southwest). To correspond to this, we drew a bar in front of Kotaki with one prong pointing towards Niôdô. On the map, there is a wall in front of Kotaki with ten prongs sticking out from it, indicating that nine locations besides Niôdô also perceived a linguistic boundary between themselves and Kotaki. The wall is formed by stacking up bars, and so the thickness of the wall corresponds to the number of hamlets (ten) perceiving the boundary.

42

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 4.1. How individual responses and boundaries were plotted

Figure 4.2. Subjective dialect groups 1, 10, 11, 14, and 18

CONSCIOUSNESS OF DIALECT BOUNDARIES

Figure 4.3. Subjective dialect groups 2, 5, 12, 15, 16, and 19

Figure 4.4. Subjective dialect groups 3, 6, 9, and 17

43

44

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 4.5. Subjective dialect groups 4, 7, 8, and 13

We show responses regarding the community of Ômi (in other words, the central hamlet of Ômi Town, not shown on Figure 4.1) in the same way. The third peg on 92 is pointed north-northwest towards Ômi and a seven-pronged wall at the top of the map corresponds to this. We moved south performing this same procedure as we went. However, we noticed, as we moved south in the upstream direction of the river, that like answers decreased and an interesting phenomenon occurred along the way. This phenomenon can be seen by comparing (the areas encircled on) the left and right sides of Figure 4.1 (which illustrates two different views of the same area). The encircled area on the right map shows the upstream area (which shared a similar dialect consciousness and one quite different from that exhibited by the hamlet encircled on the left-hand map). First, on the left map the bars stack up to form walls. Two hamlets perceive dialect boundaries in front of location 24, three in front of location 48, and one in front of location 15. These walls are situated deep within the Nechi Valley. In contrast, on the right map, we see walls formed at the mouth of the valley. There is a three-bar wall in front of location 20, and a one-bar wall in front of 92. Furthermore, on the left map, there is a seven-pronged wall in the northern part, while on the right map, a seven-pronged wall appears at the south. Also, on the left map, a boundary is perceived at the river (the Himekawa River that runs from south to north the length of the map) by four informants, but no such boundary appears in the right map. Rather, as stated above, a boundary appears at the south of the valley. In addition, the left map shows the perceptions of a boundary in the northeast direction (one bar in front of 37, three in front of 36, and one in front of 01). On the other

CONSCIOUSNESS OF DIALECT BOUNDARIES

45

hand, on the right map, the only bar in this direction is in front of 85. Also, whereas on the left map we clearly see a wall perceived in front of Kotaki, only a two-bar wall appears here on the right map. Thus, with the right and left maps, we see an odd but clear opposition between the two. The spiked ring encircling the survey locations with similar responses shows that, insofar as dialect consciousness goes, these two areas form different worlds. Naturally these two divisions have an area of gradual transition. For example, if we look at the contact area of these two blocks, we see that within the spiked ring area on the left map, locations 28 and 46 perceive the wall west of the river along with the other informants here, but at the same time they responded that location 45 located far to the south was “not that different from here,” thus showing similarities to the hamlets encircled on the right map. Similarly, location 59 perceives the wall at the mouth of the valley (a trait of the group on the right), but at the same time does not perceive the southern wall which is a hallmark of this group. In order to find a perception of the southern wall, we must go a bit further south to location 82. Thus, in deciding which division a certain hamlet should be placed in, we had to redraw the maps many times, in search of a balanced division scheme.7 3.2 Isolated Communities We divided the entire survey area into eighteen parts according to the method described above. We found seven hamlets which were isolated (forming language islands) which we combined as division number 7 in Figure 4.6. This gave us a total of nineteen dialect divisions. (The divisions numbered 8 and 10 in Figure 4.6 correspond to the left and right sides of Figure 4.1, respectively.) Figure 4.6 shows the contour lines for the 1:50,000 topographical map of the survey area. The areas colored solid black are altitudes over 1500 meters; the thicker bold lines show the 800–meter mark, and the thinner lines outside of this illustrate the 400–meter mark. In the map, we see higher areas at the left and right and a narrow strip of flat land along the seacoast with five rivers running through it. The river valleys and the corresponding divisions on the map are as follows, from east to west: Hayakawa, 17, 18, 19; Umikawa, 14, 15; Himekawa, 5, 6, 9, 11, 12; Tômigawa, 4; Ômigawa, part of 3, part of 7. In addition to these, we have the Nechi Valley, 8, 10, seen in Figure 4.1, and the seacoast areas consisting of 16, 13, 2, 1, and part of 3. Let us return to Figures 4.2 through 4.5 to see how Figure 4.6 was constructed. In each of the divisions, there are large areas which contain smaller areas within them. Look at the far left side of Figure 4.2 at the bold line which encircles the two hamlets of division 1. Surrounding this is a larger area encircled with a dotted line. This outer line reflects the subjective consciousness of the informants; it shows how far the realm of the informants’ dialect consciousness extends.8 The bold line forming the inner circle encloses those hamlets (in this case two) which share this perception of how far their dialects extend. The outer circles overlap a great deal; in order to make these easier to see, we have illustrated them separately as Figures 4.2, 4.3, 4.4, and 4.5.

46

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 4.6. The nineteen subjective dialect areas

Division 1 (shown on Figure 4.2) and Division 2 (Figure 4.3) both include only two hamlets each. These are isolated societies. They are separated by the almost impregnable pass of Oyashirazu (indicated by the thick vertical arrow), and until recently transportation along the coast was difficult. Until 1884, there was no road here. People simply had to wait for the waves to subside and run along the coast while they could. In 1885, a onemeter wide road was cut through, which subsequently fell into disuse after the railroad came through in 1912. The road in use now was constructed in 1938. The four hamlets in division 1 and division 2, although technically part of Niigata prefecture, are economically part of Toyama Prefecture. All of the other hamlets included in the survey area go to Itoigawa’s town center (locations 52, 53, 54, and 64 on the map) to do their shopping, but these four hamlets in division 1 and division 2 shop in Tomari Town in Toyama Prefecture, and their rice comes from there also. Since their dealings with the rest of the prefecture are limited, their dialect consciousness is restricted to a narrow realm. On the east side of Oyashirazu, other coastal areas reveal these effects of isolation. Only four hamlets are included in division 3 (Figure 4.4) and only 3 hamlets in division 4 (Figure 4.5). The outer limits (comprehensibility region) of division 4 were found to match up with the inner circle (shared consciousness region), so only the inner circle is indicated here. Now let us shift our attention to the language isolates grouped as division 7. These are the five hamlets indicated by the arrows in Figure 4.5. As shown on Figure 4.6, these division 7 hamlets are found along the seacoast and the Himekawa River. At all of these

CONSCIOUSNESS OF DIALECT BOUNDARIES

47

hamlets, people said of even the nearest hamlets, “their speech is quite different (from ours).” Of these five hamlets the three along the Himekawa River were isolated topographically and transport-wise. First, location 30 (Suganuma) is all the way at the end of a Himekawa River tributary (the Mushikawa River), a hamlet deep in the mountains with only nine households. Location 12 (Mushikawa) was at one time the site of a domestic customs gate (sekisho), but now it is a completely isolated hamlet of nineteen households. Location 18 (Yamamoto) is located along the Himekawa River, at a place where there is no bridge; residents used to go all the way to location 22 to cross. Of the remaining two hamlets, location 77 (Ôsawa) lies sandwiched between a mountainside and the employee housing for a nearby factory. The company employees (numbering almost ten thousand) have moved here from all over the country and, to the indigenous population of the village, constitute a dialectally different group. The remaining hamlet is number 79 (Ômi Town center), which had a consciousness barrier between not only it and location 77, but with 64 as well. This is probably because until 1871 there was a han (administrative) boundary between 79 and 64. From east of this was the Itoigawa han, west of this was the Takada han. Also, this is where the pitch accent boundary runs. For example, to the east they say ishi ga “stone” or ami ga “net” (nouns + ga, a subject particle) with a low-highlow pitch accent. To the west, the pitch drops after the initial mora in these words, resulting in ishi ga and ami ga being pronounced high-low-low. On Figure 4.3, division 12 is an isolated community. Of the three hamlets in this division, location 72 (the numbers are more legible on Figure 4.5) is the site of an elementary school. Children from location 64 (also in division 12) attend this school as well. (There is a road going to location 45, but the children there cross the mountain to location 15 to attend school. When the mountain pass is impassable because of snow between November and March, a temporary classroom is set up within their hamlet. Location 18 in division 7 has a similar situation.) The other hamlet in division 12, location 7, is a newly established one. There were only three households here in the Meiji era (from 1868 to 1912), and all three of these were migrants from other areas. Our informants came in 1896 at the age of five from a place sixteen kilometers south. Thus the informant did not really fit our criteria for “native,” and we could just as well have removed this location from the map, but we decided to include it because the dialect consciousness was the same as the other two hamlets. Division 10 on Figure 4.2 is contained in location 45. Whereas this location forms a division with the hamlets in Nechi across the mountains, division 12 is pressed up against the bank of the Himekawa River. It seems this area is isolated by the river. It was 1882 when a bridge was first built across the Himekawa River, but the summer floods washed it away almost yearly. Thus transportation across the river was extremely difficult. The first train to reach here (Ôito line) wasn’t completed until three months before our survey in 1957. Above, we showed how geographical and historical divisions formed the boundary consciousness of divisions 1, 2, 3, 4, 7, and 12. Division 16 (Figure 4.3) is also an isolated society, but in this case, it is the result of social isolation. The three hamlets which are in this division are fishing villages, in which the men go out to sea and the

48

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

women peddle the catch around the area. At many of the survey locations, these hamlets (collectively known as Uramoto) were commonly known and their speech regarded as being “rough.” 3.3 Large Divisions Let us look at the Hayakawa Valley (divisions 17, 18, and 19 in Figure 4.6). Division 17 can be seen in greater detail in Figure 4.4. We see a double circle in the center of this division indicating Aramachi, where the surrounding hamlets come to do their marketing. There is a strong consciousness of a boundary between these hamlets and the coastal hamlets of Uramoto (locations 20, 03, 86, division 16) as shown by the very thick wall along the coast where nineteen out of the twenty-four hamlets perceived a boundary. Six other hamlets perceived boundaries even further back into the valley. Next, we turn to division 18 (Figure 4.2) where a double circle marks the central location of Otosaka (location 65). Of the nine hamlets in this division, seven perceive a downstream boundary at location 3 (Aramachi) and eight perceive a boundary upstream in the source of the Hayakawa River. Look at division 19 (Figure 4.3) with a double circle marking location 67 (Yunokôchi). This valley is cut off by the 400–meter Yakeyama Mountain, and there is no road to the southeast. The people here perceive a boundary at Uramoto (shown by a thick line at the coast) and Aramachi (shown by a thick line parallel to this further inland). There are also hamlets which perceive a boundary at places even closer to the core area (shown by the three thin lines). This is the first place we have seen that perceives a boundary towards the Umikawa Valley (the area to the south-west including division 15). At division 13 on Figure 4.5, hamlets at the mouths of the Hayakawa River and Umikawa River and along the seacoast have combined to form one division. The informants here emphasize that the dialect at the source of the Hayakawa Valley is different. Let us compare it to division 6 (Figure 4.4). Whereas division 13 does not display a strong consciousness boundary to the west until the Ômi city-center, division 6 senses a western boundary much earlier than this at the Himekawa River. This fact we find interesting. Division 6 (Figure 4.4) constitutes the former boundaries of Itoigawa City, and so displays a strong sense of cohesion and a sensitivity to language which differs from their own. For example, whereas division 13 perceives no boundary at the Nechi Valley (which contains divisions 8 and 10), division 6 displays a strong boundary consciousness towards this area. Division 14 (Figure 4.2) perceives a boundary at some of the seacoast hamlets. Unlike division 13, we find boundary consciousness here not only towards the Nechi Valley, but towards the upstream areas of their own Umikawa Valley as well. Division 15 (Figure 4.3) is the first we have seen which displays a boundary consciousness towards both coastal areas and towards the “west of the (Himekawa) river” area. Location 53 is excluded from the core area just as it is from division 8 (seen to the north-east of the pronged ring in Figure 4.1). Division 5 (Figure 4.3) constitutes the former Imai Village, and perceives a strong consciousness boundary towards the areas which constituted the former Ômi Town (the

CONSCIOUSNESS OF DIALECT BOUNDARIES

49

center of the present Ômi Town). What is more, two of the hamlets here perceive a boundary along the coast between themselves and the old Itoigawa City areas as well. The group of isolated villages in division 7 (locations 12, 18, and 30) are strongly disliked by six of the ten hamlets here. Of course, it may be said that these isolated villages also tend to hold the hamlets of division 5 at a distance as well. We have already examined divisions 8 and 10. (The small print on Figures 4.5 and 4.2 is more legible on Figure 4.1.) Division 9 on Figure 4.4, and division 11 on Figure 4.2 constitute the Kotaki area (formerly Kotaki Village). Interestingly, divisions 9, 10, and 11 all perceive a boundary between themselves and location 72 (Ômi, marked by the double circle on division 12, Figure 4.3). A 651–meter mountain peak cuts divisions 9 and 11 off from each other. Although people in division 9 can easily go to the Himekawa River, the Nechi Valley does not form part their region. With division 11, the Nechi Valley does not even enter into the informants’ consciousness. 3.4 The Origin of Class I Dialect Consciousness What is the relationship between the nineteen divisions based upon dialect consciousness and the areas formed by objective isoglosses? In our survey, we obtained dialect forms for about sixty items. Then we drew about one hundred maps from these. We examined all of our maps which showed a distribution similar to our dialect consciousness divisions. Of course, we had not expected to find one item where there were nineteen different areas, but we did find maps where one of the forms showed a distribution corresponding to one or two of the consciousness divisions. Below we list some of these correspondences. 1.

2. 3. 4. 5. 6.

7.

Standard Japanese kamakiri “praying mantis” – [haFætorimYwi] — divisions 1, 9, 12 – [sentaro:mYwi] — divisions 18, 19 – [ibottsY ¨ rimYwi] — divisions 8, 10 (Tokugawa 1959a) Standard Japanese jagaimo “potato” – [dŠo:woimo], [dŠo:wYimo], [dŠY:woimo], [dŠo:wo], [dŠo:wY] — divisions 18, 19 Standard Japanese tômorokoshi “corn” – [tsY ¨ makibi] — division 2 Standard Japanese matsukasa “pine cone” – [dŠY]?Yri], [zY]]Yri] — division 16 Standard Japanese otamajakushi “tadpole” – [memento], [memetto] — division 19 Standard Japanese kataguruma “riding on a person’s shoulders” – [te]]YrYma] — divisions 19, 1 – [kakkarakatsY ¨ ] — divisions 17, 18 Standard Japanese tsumuji “crown of the head” – [dŠine] — divisions 9Y 11; – [gidŠi] — division 2

50 8.

9. 10. 11. 12. 13.

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Standard Japanese itadori “knotweed” – [itadzY ¨ ra] — division 2 – [tantaro:] — division 4 Standard Japanese usugôri “thin ice” – [dzaFæ], [dzaeæ] — division 16 Standard Japanese medaka “killifish” – [isado] — division 16 Standard Japanese tokage “lizard” – [kanahebi] — division 16 Standard Japanese kakato “heel of foot” – [kibiso] — division 19 Standard Japanese initial morpheme of kusu-guttai “ticklish” – [kotsY ¨ -] — divisions 9, 10 – [kotso-] — division 10 – [koso-] — division 1 – [kotwo-] — division 2

As far as our survey could reveal, there were similarities between the dialect distribution and the dialect consciousness for divisions 1, 2, 16, and 19. However, as we mentioned previously, 1, 2, and 16 are isolated societies, and the number of locations in these areas was very low. Thus, the basis for stating that there are areas which show an overlapping of language regions and consciousness divisions is very weak. With divisions such as 5, 6, 8, 12, 13, 14, 15, and 17 we did not find a single linguistic item whose distribution corresponded. In the future, if we increase the number of survey items, we might be able to find word forms whose distributions corresponded to these divisions. However, even so, I do not believe there is any possibility that these divisions will match up with a reasonable number of isoglosses, and certainly not a bundle of isoglosses which are of any importance (from the standpoint of the history of the Japanese language).9 Even if these divisions should match up with some isolated form, it would be impossible to equate this with the “dialect territory” (hôgen ryôiki) which Tôjô was aiming for. But, we have to examine one exception to this. Division 19 is an area in which wordmedial [g] is not nasalized. With the exception of this division, the entire survey region has a nasalized []] in this position. Divisions 17 and 14 are very conscious of this, and more than a few informants reported this. According to them, “In Kami-hayakawa, they say ga ga strongly.” This is an important phonetic isogloss, so we must acknowledge that this is an example where subjective dialect consciousness and objective linguistic reality matched up. 3.5 Non-linguistic Factors We must look for non-linguistic factors which match up overall with the nineteen divisions. First, let us look at the “shopping spheres.” We found these by processing the

CONSCIOUSNESS OF DIALECT BOUNDARIES

51

results of the question “where do you go when you have to do some large-scale shopping?” According to these results, there is an overall match up for divisions 17, 18, and 19 in the Hayakawa Valley. In division 17, all the hamlets go to Aramachi (shown by the black-centered circle on division 17, Figure 4.4) to do their shopping. With division 18, all hamlets do their shopping in Otosaka (shown by the black-centered circle on division 18, Figure 4.2). With division 19, everyone goes to Yunokôchi (shown by the blackcentered circle on division 19, Figure 4.3). Are there any other relevant factors, such as administrative divisions, besides shopping spheres? The administrative divisions at the time of the survey were newly established ones which had only been in existence for three years. Since the entire survey region consisted of only Itoigawa City and Ômi Town, this difference is not of significance here. We must instead look into the history of the administrative divisions prior to 1954. First, under the han (feudal domain) system which lasted until 1861, this area was mostly the domain of the Itoigawa han and the Takada han. Scattered around were patches of land administered by temples, shrines, and the shogunal government. One of the borders between the Itoigawa han and the Takada han was, as we mentioned previously, between the former Ômi Town and Teraji (location 64) just to the east of it. As we said before, this border corresponds to the accent isogloss. The Takada han administered lands in the Nechi, Umikawa, and Hayakawa Valleys surrounding the Itoigawa han, but we see no relationship between the consciousness divisions and the old feudal domains. In 1872, the prefecture of Kashiwazaki was established, and divided into thirteen subsections. This lasted until 1889 when (with the establishment of the “city-town-village system” reforms) this entire area was more finely divided into twenty-six towns and villages (excluding division 12, which at the time was part of Nagano Prefecture). It was during this time that our informants were born. The system was again reformed in 1901, and the twenty-six administrative districts were combined into fourteen towns and villages. It was under this system, which lasted until 1954, that our informants went through their language formation period and reached adulthood.10 In this paper we compared the nineteen perceptual divisions which we obtained with the administrative divisions in place from 1901 (to 1954). These post-1901 administrative divisions are shown on Figure 4.9. (The dotted lines show parts of the 1889 divisions.) We can thus compare these administrative divisions with the perceptual ones we illustrated in Figure 4.6. We find a surprising degree of similarity between these two maps. Of course, because of the nature of this study we would not expect an exact match, but as a whole the two match up very well. Division 1 was Ichiburi Village, division 2 was Agero Village, division 16 was Uramoto. All three of these overlap perfectly with our divisions. However, these divisions also overlapped well with the distribution areas of specific lexical items as well, so this itself cannot be taken as evidence of a relationship to administrative divisions.

52

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 4.9. Post-1901 administrative divisions

However, division 13 corresponds to all of Yamatogawa Village plus one hamlet from Itoigawa Town (location 23, Oshiage). Division 5, as well, corresponds to Imai Village (excluding the southernmost hamlet of Yamamoto, location 18, and with the addition of Tômi, location 47, from Ômi Town). Division 3 matches Ura-Tonami Village (with the addition of Hashitate, location 75, from Ômi Town). Division 17 is Shimo-hayakawa Village (excluding Higashizuka, location 70, and with the addition of locations 35, 36, and 37 from neighboring Kami-hayakawa Village). Divisions 18 and 19 correspond to a single administrative district (Kami-hayakawa Village), but the dividing line between them basically corresponds to the location of an 1889 village border. We see the same situation with divisions 8 and 10. They correspond to a single village (Nechi Village). The 1889 boundaries divided this into three villages, two of which combined correspond perfectly to division 8. The third corresponds to division 10 (except for locations 6 and 7, for which we did not obtain answers to this question). Divisions 14 and 15 once again combine to form one village (Nishiumi), although in this case the 1889 boundary runs further south than our consciousness division line. Divisions 9 and 11 also correspond to the village of Kotaki. This was a single village even by the 1889 boundary lines, but the village is in reality divided into two communities by the existence of the 651–meter Ômine Pass which forms a barrier, as shown with our consciousness divisions.

CONSCIOUSNESS OF DIALECT BOUNDARIES

53

We found only one case of a division larger than a village. The area corresponding to division 6 includes both Itoigawa City (excluding Oshiage) and Ôno Village. We assume that the area of community life for these people is larger than just a single administrative area. We see then that, although there is a great deal of correspondence between the dialect consciousness divisions and administrative divisions, the correspondence is not perfect. There are cases of our divisions being both larger and smaller than the political units. Thus it seems best to conclude that out dialect consciousness divisions reflect “community life areas” (kyôdôseikatsu ken) which developed on the basis on these administrative divisions. I wish to stress here that during the time we were drawing Figures 4.1 through 4.6, we never noticed at all the correspondence to the village boundaries, and thus were genuinely surprised at the discovery of these correspondences. This was particularly true in the case of division 13 (Figure 4.5) for which we could find no natural boundaries to account for the division, but for which the former village boundaries corresponded to our divisions. 3.6 Consciousness of German Dialect Boundaries Recently, Professor P. Zinsli of Bern has published a dialect survey of Bern German. In part of the conclusion, he writes the following (Zinsli 1957: 113):11 I think it was appropriate that we clarified several phonemic boundaries, and further, established dialect divisions from the phonemic, morphemic and lexical item areas within the city of Bern. But, although these divisions one by one have dialectological significance, according to the consciousness (Bewußtsein) of the people there, this is all one dialect area, and they do not perceive even one division within it. The Emmentaler of the Emmen River valley say Boom rather than Boum [“tree”], and Leetere rather than Leitere [“ladder”]. Likewise the people of Oberland say leng and dengle [sic dengele] for läng [“long”] and dängle [sic dängele, “to remove dirt and grass from a scythe while mowing”]. These differences are well-known. But, when we actually perform surveys, Boom and Leetere actually show distribution beyond the Emmen River. Moreover, while the distribution of leng extends across a broader area than just Oberland, that for dengle [sic dengele] is actually smaller than the area of Oberland. According to the late Professor H. Baumgartner, to the average nonlinguist speaker, these phonological characteristics are markers of “class” more than of region. In other words, to avoid these characteristics is to show that one can speak the refined, urban, educated “regional common language.” This means, that the phonology of the “regional common language” used throughout the city of Bern is vivacious and has begun to spread into the lower strata in towns. But, if you ask these people to subdivide their region and to describe what the characteristics of a neighboring area are, phonology or morphology are not mentioned. It seems that these traits do not provide definite proof (of a person’s area). For people here, the divisions formed by old village borders, current political boundaries, or new transportation systems are more definitive factors. Moreover phonological-level characteristics like Boom and dengle which are learned from everyday experiences, are perceived only as being part of the regional common language of Bern German. People often speak of the Thun dialect or that of Emmental or Oberland or

54

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” Seeland, but when asked to give specific markers for any of those dialects, they are not able to do so. This situation is no different for any other parts of the distribution area of Swiss German. Professor Baumgartner searched for the boundary between the Seeland dialect and the Mittelbern dialect (which are generally regarded as being a single dialect). But in order to make the dialect consciousness boundaries match up with objective dialect boundaries, he had to piece together the isoglosses of four different linguistic variables (Baumgartner 1922: 167–68).

What we should note here is that, although the state of Bern is perceived as being a single dialect area, Professor Zinsli found that multiple dialect isoglosses cut through it. We began from a standpoint different from his, but still reached a similar conclusion.

4.

Class II Dialect Consciousness

Up to this point, we have processed the information based on the hamlets which were identified as 2 “a little different” and 3 “quite different” in our survey. Next, we turn our attention to those hamlets identified as 4 “almost incomprehensible.” We will examine the results in 3 directions. 4.1 To the West Look at Figure 4.7. Out of 164 hamlets, forty-five placed a Class II boundary somewhere in the area between the (former) Ômi Town and the Himekawa River. The forty-five informants who held this perception were located in the area highlighted by the thick diagonal lines in Figure 4.7. This is the hinterlands of the (former) Itoigawa City. Out of 164 hamlets, fifty gave the name Oyashirazu, or a point further west as a place with an “almost incomprehensible” dialect. The place which came up the most was Tomari Town in Toyama Prefecture, fifteen kilometers past the western edge of the map. The fifty hamlets which answered in this way were in the areas blackened out on Figure 4.7. Looking at Figure 4.7, we notice that the blacked-out areas are mainly west of the Himekawa River. The location of the forty-five hamlets marked by the diagonal lines are in clear contrast to this. East of the Himekawa River informants perceive the boundaries at Ômi Town, but when you cross the Himekawa River, they perceive the boundary at Oyashirazu or Tomari, both farther west. All the same, we must not overlook the fact that there are also twenty-six hamlets colored black in the valleys east of Itoigawa City as well. Furthermore, eleven of the 164 hamlets perceived a Class II boundary at “Toyama,” even farther away than Oyashirazu or Tomari. Whether they meant “Toyama Prefecture” or “Toyama City” (more than 60 kilometers from the left edge of our map), is something which we could not always determine. In some cases, informants specified “Toyama City.” At any rate, the eleven locations which answered “Toyama” are indicated on Figure 4.7 by the double circles. All of these locations are remote mountainous areas which are among the most distant in our survey from Toyama Prefecture and Toyama City.

CONSCIOUSNESS OF DIALECT BOUNDARIES

55

Figure 4.7. Respondent identification of areas that are ‘difficult to understand’

Perhaps because people in distant areas have less opportunity to hear the far-away Toyama Prefecture dialect, their judgments were dulled and they were even more likely to come up with the names of far-away places. In these remote areas, “Toyama (City)” may be a symbol of another world. These answers for “Kawanishi” (the area west of the Himekawa River), Oyashirazu, and Toyama (City) may be rooted in knowledge gained from hearsay. The informants in our survey refer to the speech of this region as “Ettchû kotoba” (or “Ettchû ben”) “the Ettchû dialect.” Since our survey area was in the Echigo region, these answers may derive from a perception (on the part of our informants) that their dialect belongs to the Echigo dialect and not the Ettchû dialect. In the furthest west of our survey locations just across the river from Toyama Prefecture (in division 1 on Figure 4.2), informants told us, “we don’t like going to Itoigawa because people mistake us for Ettchû people on account of our accents.” We took the answers above and processed them differently than we had done for Class I dialect consciousness answers, arriving at the three thick black lines drawn at the Himekawa River, at Oyashirazu, and at the Toyama prefectural border. The thickness of the lines on the map corresponds to the numbers of locations represented.

56

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

From these we can see that our entire survey region perceives a marked contrast between their own dialect and that of Ettchû. For the people living east of the Himekawa River, this Ettchû dialect is perceived to begin on the other side of the river. 4.2 To the East Compared to the 106 hamlets which responded with a western boundary, only twenty-four indicated one in the east. Of those twenty-four hamlets, fifteen perceived a boundary in front of Uramoto (the three eastern-most coastal hamlets). On Figure 4.8, we show these hamlets with a sharp sign (#). The remaining nine hamlets responded with Nadachi (twenty kilometers east of our map) or Naoetsu (thirty-five kilometers east of our map) and are designated by a black square. These nine informants also added that the speech east of Nadachi or Naoetsu was all the same. The thick black line at the north-east of Figure 4.8 represents the consciousness of these nine hamlets.

Figure 4.8. Boundaries of the areas that are ‘difficult to understand’

We have also drawn a large black triangle near the left edge of Figure 4.8. This marks the location (Ichiburi) where an informant responded that the speech was different east of Oyashirazu (indicated by the vertical arrow). This marks the first time that we have seen the reaction of an Ettchû dialect speaker to the Echigo dialect to the east. If we had extended our survey into Toyama Prefecture, we might have been able to identify this boundary more clearly.

CONSCIOUSNESS OF DIALECT BOUNDARIES

57

4.3 To the South As Figure 4.8 indicates, the consciousness boundary with Nagano Prefecture is very thin. Only seven locations perceived a speech boundary with Nagano. On Figure 4.8, the black circles show the location of these hamlets. Almost all of these hamlets are situated deep within their respective valleys. The Himekawa and Nechi River valleys in the center of the map are areas of traffic especially to and from Nagano. 4.4 Overall Tendencies Thus we can make the following generalizations about Figure 4.8. In the southwest part of Niigata Prefecture there is a consciousness of the differences with the Toyama (Ettchû) dialect, but these perceptions begin to appear at locations before we actually reach the prefectural border. Moreover, many of the informants in our survey perceived Oyashirazu as the beginning of another world, and roughly the same number perceive the Himekawa River in this way. As we saw earlier, the divisions created by Class I dialect consciousness responses were not language divisions, but rather corresponded very closely to old administrative divisions. Let us then examine the Class II dialect consciousness divisions west of the Himekawa River to determine whether they correspond with language divisions. Look at Figure 4.10, showing the actual isoglosses of several linguistic features which are located between the Himekawa River and the Toyama prefectural border. Most of our survey locations can be characterized by whether they fall to the west or the east. Some of these isoglossess are very important from the standpoint of Japanese language history, namely at (0) the word “sty” (meboro, meborai, contrasting with other forms); (1) the copula (da, contrasting with dŠa); at (3) the imperative verb form (e.g., okiro “wake up!”, contrasting with okiyo); at (5), (6), and (7), three categories of pitch accent variables; at (8) the lack of a semantic distinction between toge “splinter” and bara “thorn” (contrasting with areas which support the distinction); at (9) “two days after tomorrow” (wiasatte, contrasting with wiwasatte). These east-west isoglosses divide not only the survey area, but all of Japan as well into Eastern and Western varieties. Looking at these facts, how could we possibly conclude that, unlike the Class I dialect consciousness divisions, Class II divisions correspond to linguistic boundaries? I mentioned previously that the old han feudal clan boundaries run through this area as well. The border between the old Takada han and the Itoigawa han (shown as F on Figure 4.10) ran between the former Ômi Town (location 79) and Teraji (location 84). We may assume that it was as the result of the han boundaries which were here for so long that language differences developed. Furthermore, if the Class I and Class II dialect boundary consciousness are indeed the same characteristic, then we must conclude that Class II dialect boundary consciousness is also the result of non-linguistic factors. In the future, we hope that similar surveys may be carried out in different areas to shed some further light on this subject.

58

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 4.10. Some traditional dialect isoglosses

5.

Class I Dialect Consciousness and Class II Dialect Consciousness

Does the difference between the places identified as “a little different” and “incomprehensibly different” correspond to something in the consciousness of the informants? Did the manner of our question allow the informants to sufficiently grasp the difference in the two categories? There may be some questions about this. Let us examine the issue. Figures 4.2 through 4.5 give us some partial answers to these questions. For example, in division 14 on Figure 4.2, eleven out of the fifteen hamlets perceived a boundary at Ômi Town (location 79), as indicated by the eleven-unit line. What is more, we find division 5 (Figure 4.3) which also has a strong perception of a boundary at Ômi. This accounts for six out of ten hamlets. Next, we see with division 6 (Figure 4.4) that there are three lines (counting both narrow and broad) between the Himekawa River and Ômi Town. Totaled these represent the consciousness of ten hamlets. Furthermore, division 13 (Figure 4.5) indicates a combined total of six lines between itself and Ômi. These four divisions (14, 5, 6, and 13) thus indicate the perception of a boundary between themselves and the areas west of the Himekawa River. When totaled these boundaries between the Himekawa River and Ômi Town number thirty-seven lines (15+6+10+6). Seeing how so many boundary lines clump up here reminds us of the

CONSCIOUSNESS OF DIALECT BOUNDARIES

59

boundary in Figure 4.8, where we have found boundaries for forty-five hamlets. This similarity seems to hold some degree of significance. Let us look again at Figures 4.2 through 4.5. As outlined in the following, the opposition between Uramoto [i.e., division 16] and other areas is very clear: – – – – – –

eastern boundaries of division 14 (Figure 4.2) 5 hamlets eastern boundaries of division 5 (Figure 4.3) 4 hamlets northern boundaries of division 19 (Figure 4.3) 4 hamlets eastern boundaries of division 6 (Figure 4.4) 6 hamlets northern boundaries of division 17 (Figure 4.4) 19 hamlets eastern boundaries of division 13 (Figure 4.5) 7 hamlets

The boundaries with Uramoto here were perceived by a total of forty-five hamlets, so it is clear that many informants perceived Uramoto as being different from themselves. However, when the same people are asked to name a place where the speech was “incomprehensibly different,” only fourteen informants placed a boundary here (as shown in Figure 4.8). From this it would appear that our two ways of asking this question have tapped into different perceptions. Looking once again at Figures 4.2 through 4.5, we see from the boundary lines listed below that the Nechi Valley appears to be a separate division. – – – – – – –

southern boundaries of division 14 (Figure 4.2) 4 hamlets northern boundaries of division 10 (Figure 4.2) 3 hamlets southern boundaries of division 15 (Figure 4.3) 7 hamlets southern boundaries of division 5 (Figure 4.3) 3 hamlets northern boundaries of division 9 (Figure 4.4) 4 hamlets southern boundaries of division 6 (Figure 4.4) 8 hamlets northern boundaries of division 8 (Figure 4.5) 5 hamlets

The Nechi Valley is defined by the (Class I) boundaries of thirty-four hamlets, but with the Class II boundaries, only five hamlets responded in this manner. It was because this number was so small that we did not even indicate these responses on Figure 4.8. From these results we conclude that our two questions did indeed elicit different types of responses, and it appears we were correct in combining the results from our three questions into two categories rather than into only one.

6.

Japanese Dialectology and Dialect Divisions

When Georg Wenker planned a dialect survey of the entire northern and central portion of Germany in 1876, one of his aims was the establishment of German dialect divisions (Pop 1950: 741). However, the results were remarkably different from what he had

60

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

expected. In other words, it was clear that Wenker’s own personal “dialect consciousness” did not match up with the objective facts of the language distribution. With Professor Zinsli’s survey of the state of Bern as well, it became clear that the dialect consciousness divisions of the people of Bern did not match up with actual language divisions. Likewise in our survey, the dialect consciousness of the area residents did not match up with linguistic divisions. Indeed, it was administrative divisions, or more accurately “community life areas” with which they matched up. Thus, we must conclude that it is not appropriate to use dialect consciousness as a basis for the establishment of dialect divisions.12 We must instead use objective linguistic boundaries in order to determine dialect divisions.13 Does this mean that dialect consciousness is of no use to linguistic geography? In linguistic geography, it is necessary to bring in non-linguistic factors in order to explain dialect distribution. These non-linguistic factors include natural factors (mountains, rivers) as well as cultural factors like shopping spheres, transportation spheres, marriage spheres, administrative divisions, folklore characteristics, and so on. We saw that “dialect consciousness” matches up well with administrative divisions (current ones as well as feudal domains), although not perfectly. Indeed, as we stated previously, we should actually regard these areas as “community living areas.” This being the case, we can position “dialect consciousness” alongside other non-linguistic factors as an important explanatory tool. But is this the appropriate positioning of dialect consciousness? Our survey area was one which had many nationally and (from the standpoint of the history of Japanese) important isoglosses transversing it. What is more, it was an area that is undeniably unique in terms of dialect consciousness. For these reasons, we eagerly await the repetition of this type of survey in other regions in the future.

Notes 1. At the time of this writing, there have been numerous other papers written based on this survey. They include: (1) Sibata (1957), (2) Grootaers (1958a) on mono-morai “sty (in the eye),” (3) Grootaers (1958b) giving an overview of the survey, (4) Sibata (1958a) on kinoko “mushroom,” (5) Sibata (1958b) on jagaimo “potato,” katta, “bought,” (6) Grootaers (1958c) on shi-asatte, yano-asatte “day after tomorrow,” (7) Tokugawa (1959a) on kamakiri “praying mantis,” (8) Tokugawa (1958a) on kamakiri “praying mantis,” (9) Grootaers (1958d) overview of research, (10) Tokugawa (1958b) on methodology, (11) Grootaers (1959a) on mono-morai “sty,” (12) Sibata (1960) on toge “briar,” (13) Grootaers (1962) on kusuri-yubi, “ring-finger,” (14) Grootaers (1959c), (15) Sibata (1959), (16) Tokugawa (1959b) on hiki-gaeru “toad,” (17) Tokugawa (1959c) on hiki-gaeru “toad,” (18) Grootaers (1959b). [Translator’s note: Sibata and Grootaers have recently produced a comprehensive 12–volume report on their work in Itoigawa. See Sibata and Grootaers 1988, 1990, 1995.] 2. We also asked the following three questions for reference. (1) Are there any places around here that are wellknown for speaking strangely? (2) When you hear someone speak, can you readily tell what hamlet they are from? (3) When you were a child, did you ever feel that the speech of another hamlet was very odd? 3. These numbers correspond to the seventh and eighth digits of the map coordinates. The full location number for Niôdô would thus be 5621.1592. For more information about this, see Tokugawa (1958b).

CONSCIOUSNESS OF DIALECT BOUNDARIES

61

[Editor’s note: The numbering system is also explained in more detail in Grootaers (1959c: 360, footnote 1) as follows: The numbers used here are affixed according to the system of the Linguistic Atlas of Japan. Each square on [Figure 4.1] is divided into 100 smaller ones, and the locality receives the number according to its position in the grid. See for instance the locality 00 (upper left corner of a square) in the middle of [Figure 4.1], left; see then the locality of 02 (two positions to the right of 00 in the square) in the upper right corner of the map. Further, each square has also its own number, as can be seen on [Figures 4.2] to [4.5]. For instance, the locality 02 of the Nechi valley can be found on [Figure 4.2] in the square 16, with the vertical line numbered 76. This gives as the full number of the above hamlet the number 1602. [Figures 4.2] to [4.5] give also in the upper right hand margin another number 5611 (or 5612), which indicates one sheet of the 50,000 scale map. Such a system expresses by way of numbers the exact geographical location of every hamlet.] 4. The locations not encircled are those which did not respond to this question. Locations 10 and 80 enclosed within the spiked ring on the left side of Figure 4.1 are examples of this. 5. Translator’s note: The administrative boundaries have changed and such a village no longer exists as an official administrative entity. 6. Translator’s note: In Japan, the words shi “city”, machi “town” and mura “village” are not merely general words for communities, but terms for specific kinds of administrative units. Villages are usually comprised of clumps of dwellings called shûraku, translated here as “hamlet.” 7. After constructing and printing this map, we discovered a problematic location. This is location 00, the westernmost of the locations encircled by the spiked ring in the right-hand map of Figure 4.1. The pegs point in the direction of Kotaki (locations 70, 29, 52, 60, etc.) and to the south. If we had placed this location within the ring on the left-hand side of this map instead, this would have changed the thicknesses of the surrounding walls. There seem to be no clear criteria to tell us in which ring 00 should be included. The fact that a road connects 00 to 66 would seem to indicate, however, that it might have been more appropriate for us to include this location within the ring on the left-hand map instead. 8. Translator’s note: These dotted lines are drawn by connecting the perceptual boundary “bars” discussed earlier. 9. In German, the term for this “dialect territory” is Kernlandschaft. Bach ([1934] 1950: 52–55) defines this as an area with a core surrounded by bundles of important isoglosses. The term was first used by Haag (1898, 60) to denote a “unified dialect region” (einheitliches Mundartgebiet). In Germanic and Romance linguistics, researchers began by trying to find the “boundaries” between two dialects, but they realized that the isoglosses for each word were different, and that even if there were a region in which bundles of these isoglosses were concentrated, this could not be justified as the existence of an actual boundary. Hence, the focus shifted to the search for regions in which a unified dialect was used. Naturally, this region must possess boundaries of sorts, but the emphasis had shifted from the search for “borders,” to one for a “distribution area.” In the present study we found that the nineteen perceived divisions are NOT the type of area surrounded by isogloss bundles. 10. Translator’s note: The “language formation period” is the concept that a speaker’s language system is formed between the ages of five and fourteen. It is based on empirical evidence from a study of the dialect acquisition of children who were relocated from urban to rural areas during World War II (e.g., Kitamura 1952). 11. Translator’s note: This section is translated from the Japanese text and not from the original German. 12. Misao Tôjô does not deal with small-scale areas as we have here. This leads one to question whether dialect consciousness is applicable only in the cases of large-scale dialects such as the Kantô dialect, the Kansai dialect, or the Kyûshû dialect. But in a passage closely following the one we quoted earlier Tôjô (1938) writes, “Naturally if we narrow our observation, we sense differences in the language within a single region. For example, people who live in Aomori Prefecture easily notice the difference in the Tsugaru and Nambu dialects there. This consciousness goes all the way down to the village level. The opposition between the Yamanote ‘uptown’ and Shitamachi ‘downtown’ dialects in Tokyo is another

62

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” example of this” (29). From this passage, we can see that Tôjô does indeed see his dialect division theories as applicable on the village level.

13. I agree with Okumura (1958) when he states that we can rely on factors such as natural-geographical boundaries, administrative divisions, transport routes, and the “dialect consciousness of the local people” to construct dialect divisions, maintaining that we must use the “distribution map layering” (bumpuzu kasane-awase) method.

C 5 Consciousness of Linguistic Boundaries and Actual Linguistic Boundaries Kikuo Nomoto Translated by Daniel Long

1.

Linguistic Boundaries

It is well known that there are linguistic boundaries and that linguistic geography is based on these boundaries. As part of the same research project discussed here,1 Takesi Sibata has dealt with this theme from a linguistic standpoint, utilizing the common framework of “regional characteristics” (Sibata 1963). The methodology ordinarily employed in linguistic geography involves constructing maps of individual linguistic items. But while language is a systemic thing, it is not always easy to construct a map of that system. Distribution maps of, for example, different systems of word pitch accents or of the phonological systems of the dialects on the Sea of Japan side versus those on the Pacific side of Japan are maps of language systems. In the report on the Noto Peninsula (Sibata and Nomoto 1955), we also attempted a distribution map of the phonological systems in use there, but from the standpoint of an overall language system, this was only the examination of two individual phenomena. While in some cases the consciousness of speakers does focus on an individual feature of language (as evidenced in comments such as, “the people that live over there in X call a Y a Z”), more often than not, people’s consciousness is based on general, overall impressions. Of course, it is only natural that language consciousness differs from actual linguistic boundaries based on the linguistic geography of individual items. But at the same time, we would not expect to find that language consciousness was completely different from actual linguistic boundaries either. In this paper, we want to determine how much of a difference exists between actual and perceived language boundaries.

64 2.

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Survey Methodology

The reader may wonder why there is little information in this paper concerning the island of Sado itself, but we have used Sado only as a testing ground for our methodological techniques. In the methodology employed here, we followed the typical practice of going to a survey location and initially asking the typical questions from dialect surveys such as “what is this called?” or “what do you say for this thing?” but then we went on to ask “what do they call this in the neighboring village or the one past that?” To my knowledge, there are no surveys which have systematically surveyed informants’ knowledge of what linguistic forms were used in neighboring communities. The survey was based on questions concerning individual linguistic forms, but we also asked “how far away is the language (dialect) of this hamlet used?” For a concrete, detailed description of the survey techniques, see Sibata 1959 [see Chapter 4, this volume].

3.

Survey Area

The areas of the survey were Sotokaifu and the flatland area of Niibo. But in this paper, we will deal only with those results for Sotokaifu. Although Niibo is a flatland area, the transportation routes are complicated, whereas the routes in Sotokaifu all go (with a few insignificant exceptions) in the same direction along the seacoast. Thus we chose to analyze Sotokaifu because it was more suitable to analyze by a linear graphic model. There were twelve survey locations located in a straight line from Kitatanoura in the south to Negai in the north, a total distance of seventeen or eighteen kilometers. The survey sites, in order from the northernmost, were the following: Negai, Kitaushima, Masaragawa, Iwayaguchi, Ikaura, Seki, Yagara, Ôkura, Koda, Ishina, Onomi, and Kitatanoura. In each location, we chose one elderly male informant who was born and raised in that hamlet and had never lived elsewhere. Of course, the use of only a single informant is not without its problems, but we chose to take this risk. This method does have its advantages, namely that it avoids having to deal with variation among informants in the same location. (For more information concerning this problem, see Nomoto 1953, 1954, 1957.)

4.

Mapping Perceptions of Similarity

When informants were asked the abstract question “how many of the survey locations do you perceive as being the same as your own speech?” the responses of the twelve locations revealed a division into four neat groups of three hamlets each (Figure 5.1). The perceived divisions are shown by the dividing lines or “slashes.”

65

CONSCIOUSNESS OF LINGUISTIC BOUNDARIES North

South

(Ne)gai (Ki)taushima (Ma)saragawa / (Iw)ayaguchi (Ik)aura (Se)ki / (Ya)gara (Ôk)ura (Ko)da / (Is)hina (On)omi (Ki)tatanoura

Figure 5.1. Perceptions of linguistic similarity (category 1) (N.B. Parenthesized abbreviations are used in the following two figures)

These lines are easily explainable because they correspond to natural or political boundaries. We will consider the boundaries in order from the south (at the right of the graph) to the north. 4.1 The Boundary between Koda and Ishina Ishina and areas south made up the administrative district known (after 1889) as Takachi village. Koda and areas to the north as far as Negai were Sotokaifu village. Thus these boundaries here correspond to village borders, indicating that the administrative borders are at least one of the relevant factors in the construction of consciousness boundaries, a fact which has been emphasized in Sibata 1959 [see Chapter 4, this volume] as well. 4.2 The Boundary between Seki and Yagara Nowadays there is a bus which runs through Seki and Yagara, but in days past this was a difficult pass, and the bus from Aikawa only ran as far as Yagara. 4.3 The Boundary between Masaragawa and Iwayaguchi Although Masaragawa and Iwayaguchi were once part of the same Sotokaifu village, everything up to Iwayaguchi now lies in Aikawa Town, and the locations from Masaragawa northward lie in Ryôtsu City. Thus this boundary corresponds to a political boundary. Moreover, even now the bus from Aikawa goes only as far as Iwayaguchi. To get from there to Masaragawa, one has to cross the peak on foot. 4.4 The Boundary between Moura and Negai There is a boundary perceived between Negai and a place called Moura which lies to the north of it (outside of the survey region). From long ago, Negai was the northernmost hamlet of the administrative district called Hada-gumi. Areas to the north of this belonged to Ebisu-gumi, which became Uchikaifu Village with the administrative restructuring of 1889. After this, new settlers came in and started the new hamlet of Moura in the southernmost part of Uchikaifu, and predictably this hamlet is regarded as being linguistically closer to the other hamlets of its village than it is to Negai. We have outlined above the natural and political factors which appear to have had an effect on the consciousness of the residents. In addition to these, the perceived “blocks” described here also correspond to modern school districts, a factor which must have contributed to the stronger perception of boundaries.

66 5.

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Mapping Linguistic Boundaries

Next, we drew maps based on the twenty linguistic survey items. In Table 5.1 below, we have outlined the results of the fourteen items which revealed a coherent geographical distribution. Related linguistic forms have been categorized and the slashes separate these different categories. Table 5.1. Distribution of Survey Items Standard

English

Distribution of Dialect Terms

otedama bunke

‘beanbag’ ‘offshoot branch of family’ ‘rake’ ‘icicle’ ‘day before yesterday’ ‘ten days’ ‘radish’

omotsu, omutsu Yagara / Seki tantan inkyo Yagara / Seki inkyo, bunke Masaragawa / Kitaushima inkyo various Seki / Ikaura kumade amembô Ôkura / Yagara various ototoi Ishina / Koda ototsui

kumade tsurara ototoi tôka daikon

‘white’ shiroi ‘interesting’ omoshiroi otamajakushi ‘tadpole’ eda

‘tree branch’

rôsoku shika nonda

‘candle’ ‘only’ ‘drank’

tûka Onomi / Ishina tôka rêkon Kitatanoura / Onomi dêkon Ôkura / Yagara rêkon Ikaura / Iwayaguchi dêkon shirê Yagara / Seki shiroi omoshii, omosê, omoshê Ôkura / Yagara omoshirê kêriko, kêruko Yagara / Seki otamajakushi Iwayaguchi / Masaragawa kaerunoko eda Kitatanoura / Onomi era Yagara / Seki eda (but with era in Ikaura) rôsoku Ishina / Koda dôsoku Okura / Yagara rôsoku shika Onomi / Ishina hika nonda Kitatanoura / Onomi nonra Ishina / Koda nonda

For the 20 linguistic maps, we systematically tabulated the numbers and locations of the isoglosses and illustrated these in Figure 5.2. Looking at Figure 5.2, we find the following: 1. 2.

3.

We definitely see isoglosses coinciding with the consciousness boundaries of Figure 5.1, but certainly not to the extent that we would have expected. On a positive note, however, we find few isoglosses at the divisions of NegaiKitaushima, Ikaura-Seki, Ôkura-Koda, and Ishina-Onomi. This is the result we would have expected because these are all hamlets which reported belonging to a single dialect division in Figure 5.1. At the same time, we are faced with another problem. In spite of the positive results we saw above in (2), we are still left with one relatively thick bundle of isoglosses inside three of the four perceived “blocks” found in Figure 5.1, and these are not accounted for by the Category 1 consciousness ( see explanation below) we saw

CONSCIOUSNESS OF LINGUISTIC BOUNDARIES

67

there. Specifically, these isoglosses lie between Kitaushima-Masaragawa, YagaraÔkura and Onomi-Kitatanoura. From this standpoint, we can say that the second block (Iwayaguchi to Seki) is a strongly bonded group. Ne Ki////Ma////Iw///Ik//Se//////Ya/////Ôk/Ko//////Is//On//////Kt Figure 5.2. Actual linguistic isoglosses

6.

Mapping Perceived Isoglosses

Next, we will look at the results for the questions about what linguistic forms informants thought were being used in neighboring hamlets. In Figure 5.3, we drew a line each time an informant reported a geographical difference in one of the survey items. We can think of Figure 5.3 (as well as Figure 5.1) as showing perceptual boundaries. But, compared to the boundaries for “similarity” (referred to as “First Category consciousness” lines in Figure 5.1), the lines here (“Second Category consciousness”) represent differences in specific, individual linguistic items. Ne/Ki/////Ma(?)Iw//Ik///Se(?)Ya//////Ôk//////Ko//////Is///////On/////Kt Figure 5.3. Perceptions of specific linguistic boundaries

At the Masaragawa-Iwayaguchi and Seki-Yagara junctions, informants on both sides replied that they did not know what people in the neighboring hamlet said, so we could not draw a line. We can make the following observations about the Second Category consciousness lines in Figure 5.3. 1.

2. 3.

The number of lines between Koda and Ishina (six) is no greater than the number of lines to the north (six) or to the south (seven), even though in Figure 5.1 we found a perceived dialect boundary running between these two hamlets. This indicates that Category 1 and Category 2 consciousness questions do not necessarily elicit the same responses. There is little correlation between this map and the actual isoglosses (Figure 5.2). Although it is difficult to interpret, there is a tendency to have an abundance of lines in the south. Is this due to a large degree of fluctuation among the hamlets in the south, or is this due to individual differences among the informants? But even if we attribute this to individual informant differences, the fact that these individual differences are concentrated in a single area leads us to think they are a reflection of something in the language. In one way or another, this may well be a regional trait.

68

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

In Figure 5.4, we look at the educational level of the informants and the places of origin of their spouses. The information above the survey locations shows the place of origin of the informant’s spouse, with the “s” indicating they come from the same hamlet. The numerals at the bottom indicate the number of years of education the informants had. s s s Hiroshima s s s s s s Kitatanoura Iwayaguchi Negai Kitaushima Masaragawa Iwayaguchi Ikaura Seki Yagara Ôkura Koda Ishina Onomi Kitatanoura 0 6 3 4 3 0.5 4 4 9 6 8 8

Figure 5.4. Birthplace of spouse and years of education

We see a tendency for informants in the south to have more education. This may be one factor contributing to the abundance of perceived isoglosses in the south which we saw in Figure 5.3. We also looked at the place of origin of the informants’ parents and found that, with the exceptions of Kitatanoura (mother from Aikawa town) and Negai (both parents from Hokkaido), all of the informants’ parents were both from that same hamlet. Another group in this Sado research project has been conducting a survey of marriage spheres (the areas in which men seek their marriage partners), but at least with our informants, it seems that, although there appears to be a little movement in the south, people basically marry within their hamlet. We also found it interesting that all of the hamlets think that the next hamlet uses the word inkyo for bunke (see Table 5.1). Perhaps this word has become a famous dialectal term within the area.

7.

Mapping Perceptions of Language Differences

In Figure 5.3 above, we plotted informants’ perceptions of isoglosses. In Figure 5.5 we plotted on the map the percentage of words (surveyed) which informants think differ in neighboring hamlets. These answers are probably strongly influenced by whether the individual informant is a keen observer of language or not. In Figure 5.5, the numbers at the top show these percentages, and again we see a tendency as we saw before for larger numbers in the south. These numbers were a bit larger than we had expected. 0 8 20 5 5 10 18 13 10 25 5 25 Negai Kitaushima Masaragawa Iwayaguchi Ikaura Seki Yagara Ôkura Koda Ishina Onomi Kitatanoura 0 100 50 50 50 80 90 80 50 80 50 80

Figure 5.5. Perceptions of linguistic differences (Numbers on top indicate the percentage of words which informants thought differed in neighboring hamlets; percentages on the bottom indicate the accuracy of these perceptions)

The numbers at the bottom of Figure 5.5 indicate the approximate percentage of correct guesses. In other words, these are indicators of how well informants actually knew the

CONSCIOUSNESS OF LINGUISTIC BOUNDARIES

69

linguistic forms used in neighboring areas. Again, we saw a tendency for numbers on the right to be large and numbers on the left small (although there are exceptions, such as Kitaushima). But on the whole, we can say that the informants give correct answers if they have a keen sense for language differences. Finally, we can combine the values for two neighboring hamlets on Figure 5.5 and arrive at Figure 5.6. We see a resemblance between this diagram and the numbers of language consciousness boundaries (the Second Category boundaries) in Figure 5.3. In particular, the numbers here are about five times the number of boundaries which we saw in Figure 5.3, with a 1:5.1 ratio between the two means. 8 28 25 10 15 28 31 23 35 30 30 Negai Kitaushima Masaragawa Iwayaguchi Ikaura Seki Yagara Ôkura Koda Ishina Onomi Kitatanoura 5 25 (?) 10 15 (?) 30 30 30 35 25

Figure 5.6. Perception of dialect boundaries (numbers on top indicate the sum of neighboring values from Figure 5.5; those on the bottom indicate values from Figure 5.3 multiplied by 5)

Speakers at the junctions of Masaragawa-Iwayaguchi and Seki-Yagara seem to have no knowledge of each other, so we must be careful about combining the numbers, but if we do, we come up with twenty-five and twenty-eight, which would correspond to about five lines in Figure 5.3. In summary, we can say that consciousness boundaries are not as strong as they appeared from the First Category consciousness data, nor are the actual isoglosses as strong as we would have expected from the First Category consciousness data.

Notes 1. The project referred to here was an interdisciplinary survey of the island of Sado in Niigata prefecture conducted by the members of the group known as the Kyûgakkai Rengô [Association of Nine Academic Societies] which also conducted surveys on the Noto Peninsula reported in Sibata and Nomoto (1955). [Editor’s note: The site of this research is shown in Figure 8.1.]

C 6 Dialect Consciousness and Dialect Divisions Examples in the Nagano-Gifu Boundary Region Yoshio Mase Translated by Midori Yonezawa

1.

Introduction1

The determination of dialect divisions is one of the topics actively discussed recently in dialectology, and I am particularly interested in the relationship between the perception of dialects and such areas. Needless to say, the perception of dialects is the starting point in establishing regions in the discussion of dialect areas in Misao Tôjô’s work [see Grootaers’s translation of Tôjô’s comments in Chapter 8]. It is also needless to say that Hôgen kyôkai no ishiki [The perception of dialect boundaries] (1959) by Takesi Sibata [see Chapter 4, this volume] is the first and the strongest counter-argument to Tôjô’s theory of dialect areas. Space is too limited for me to review these earlier works and to discuss several interesting points and questions that occurred to me after reading them, ones which drove me to write this paper. As described in detail in later sections, I conducted a research program on the perception of dialects in a region2 which was similar to but slightly different from the Itoigawa research introduced in Sibata’s paper. Here I will examine what perception of dialects people in that region have and what relationship these perceptions have to the dialect areas of this region based on the findings of a more traditional inquiry. The region of this research is approximately forty-five kilometers from east to west and approximately fifty kilometers from north to south; it includes sixty-nine hamlets in Nagano Prefecture (Shinshû) on its east side and Gifu Prefecture (Hida) on the west side of the Japanese Alps. It can be called the Shin-pi boundary region (see Figure 6.6). By present governmental districts, the Nagano Prefecture side includes all of Nagawa Village and Azumi Village in Minamiazumi County and a part of Hata Village in Higashichikuma County, and the Gifu Prefecture side includes the halves of Kamitakara Village in Yoshiki County and Nyûkawa Village in Ôno County that are closer to Nagano Prefecture and all of Takane Village in Ôno County. The sites of the research were chosen to cover all the

72

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

hamlets that appear on the map of the research region of 1/50000 published by Kokudo Chiri-in [the Geographical Survey Institute], but some were not covered because a few hamlets had disappeared and subjects were not available in others3. The subjects were male speakers of the variety of the region, fifty-five years old or older. When no person with these characteristics was available, female speakers with the same characteristics were substituted. The researchers were Iku Shimizu4 and Shigeko Kobayashi5, and I. Kiyoko Hokari6 was in charge of a part of the research region. The date of the research was July 22 to July 31, 1963. In addition to this research, a smaller research program described in 3 below was conducted in Nagano Prefecture from September 26 to September 28, 1963. I was the principal researcher, and Hiroshi Minemura7 assisted me.

2.

Perception of Dialects by the Subjects

2.1 Procedures We asked the subjects the following questions in order to examine what natural perceptions the people in this region have: A. B. C. D.

E.

Where do people speak the same as here (giving the name of the hamlet)? Where do people speak a little differently from here? How is it different? Where do people speak quite differently from here? How is it different? Where do people speak totally differently from here? How is it different? Is it so different that you can’t understand it? (If the reply was yes, it was classified as a response to question E.) Where do people speak so differently that you can’t understand? How is it different?

We asked these questions of the subjects, one after another, while showing them a map in which all the names of the villages and hamlets in the research region were written large enough for them to read and in which there was some additional geographic information; we then asked them to reply to each question and to describe how the speech was different. This procedure follows that of the Itoigawa research by Sibata, but it differs in the following ways. The primary difference is that we asked the subjects to describe the differences of the languages by giving actual examples. This is because they probably cannot reply easily, based on linguistic facts, to the question “How is it different?” if there are non-linguistic facts which make the subjects perceive language differences. Even if they could reply, it would probably be based on their imagination. In our research we added a level between “quite different” and “so different that you can’t understand” because we wondered if there might be a dialect which people feel is

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

73

“totally different” but still intelligible. At this point, therefore, we are asking about more levels of dialect perception than in the Itoigawa research. Another difference is that we asked them to reply while showing a map. The reason we did this is because I hoped I could obtain the perception of dialects by the subjects more realistically, more easily, and more accurately. How did the subject actually reply? An example is given below, from the reply by a subject from Ryûshima (2), which is located at the east end of the research region. The identification numbers are supplied after the names of hamlets. The utterances of the subjects are not recorded word for word, but recorded as an appropriate summary. Utterances in non-standard are rewritten to some extent. The descriptions the subjects gave as actual characteristics of dialects were recorded using the symbols of the International Phonetic Alphabet, but katakana [shown in bold face type] has been used in this paper due to the convenience of printing, except for a few cases. To question A. Hashiba (4), Shimashima (3), and Ônota (1). Ryûshima (2) belongs to Hata Village, but it is far from the village government office, and it has more relations with a part of Azumi Village closer to here. Now it is different, but when my generation was young, we went to Azumi Elementary School in Shimashima (3). Marriages are frequent between here and Azumi Village. So, the villages we belong to are different, but I use the same speech as Hashiba (4), Shimashima (3), and Ônota (1). To question B. The speech in Inekoki (the Inekoki district in Azumi Village, which consists of the three hamlets Myôgadaira [5], Inekoki [6], and Nashinokidaira [7]) is a little different with different accents. They shorten omaedachi “you (pl.)” (Standard Japanese “omaetachi”) and pronounce it as omedachi. They use funny words like namaborene for akiramerarenai “can’t stand to give up on something.” The language in Hata is somewhat different. We pronounce yakuba “village government office” with HLL pitch accent but they pronounce it with LHH pitch accent. Nagawa (Nagawa Village). The speech there is more different from mine than that in Inekoki and Hata, but less different than that in Ônogawa (the Ônogawa district in Azumi Village, which consists of five hamlets: Sawando [8], Ônogawa [9], Bandokoro [10], Hokoratôge [11], and Hinokitôge). The speech there has a Kiso accent. The speech in Matsumono (its center is eighteen kilometers away from Ryûshima to the east) is also a little different, and it sounds elegant. To question C. The speech in Ônogawa is quite different. They say sôdancha for sôda “it is so, right.” The ending of the sentence in the speech there flips up, and it somewhat resembles the speech in Iwate (Iwate Prefecture). To question D. The speech in Hida is totally different. They use ja as in nanja kanja “It is this and that.” (To the question “Is it so different that you can’t understand it?”) Long time ago when Nomugi (the Nomugi road) was open, many people came over here from Hida, but there was no case in which I don’t understand their speech. To question E. The speech I don’t understand is that in Ettchû (currently Toyama Prefecture). They say namo all the time, and pronounce su for shi, and tsu for chi, so I don’t understand what they are saying. The speech in Ettchû is out of the question.

74

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

In this way, we recorded the utterances of the subjects, noted the information on the map, and determined what perceptions they have about the dialects of the areas around them. 2.2 The Regions Where People Speak the Same Variety Let us summarize the dialect perceptions of the subjects at different levels, beginning with “the regions where people speak the same” — that is, the replies to question A, with some reference to the replies to questions B and C. Let us take Hiwada (57) of Takane Village in Gifu Prefecture, close to the boundary with Nagano Prefecture, as an arbitrary point. A subject in this area answered questions A and B as follows. Question A. The speech is the same within Hiwada (the Hiwada district, which consists of three hamlets Hiwada [57], Kohiwada [58], and Tomenohara [59]). Sometimes the melodies of songs for the festival do not match depending on hamlets, but the speech is the same. Question B. Kaida (Kaida Village, Nishichikuma County, Nagano Prefecture, the nearest hamlet of which is reached over Sekiya Pass from Hiwada in 3 hours). The people in Kaida pronounce words differently, and prolong words. We pronounce them short. They pronounce the same words in a funny manner, such as warê for ware “you” (Standard Japanese “omae”). Nomugi (55). We call people only by their names [without using terms of respect], but the people there do not do that. Until just recently, that area had been lively because it is on the main route. Since it is close to Shinshû, they often say nê, nê, as in sorede nê, korede nê “and so, therefore.” The speech in Kamigahora (62) is also a little different. Going to the west from there (areas such as Ôburui [63] and Hikage [65]), the speech becomes better. The speech in Adanogô (56) is also a little different. They have a Shinshû [i.e., Nagano] accent, but not so much as in Nomugi.

The summary of the perception above is shown in Figure 6.1. The perception by the subjects in Kohiwada (58) is shown in Figure 6.2. These two figures resemble one another surprisingly, except for the presence of a perception boundary for Kibyû (60), Shôzô (61), and Ikegahora (64). The perception from Tomenohara (59), which is the last hamlet among the three hamlets in Hiwada, is almost the same as the one in Hiwada (57), so the map is not shown here. The surrounding hamlets all have a perception boundary for these three hamlets in Hiwada. The following is just one example. A subject in Ikegahora (64) states: The language in Hiwada (the Hiwada district in Takane Village) is quite different, even though that area also belongs to Takane Village. It is just recently that we started to have contact with those people. For a long time, they had contact with Kiso. Their temple is still in Kaida Village. I guess their language is influenced by Kiso language, and it is rough and they use da “to be” and zura “auxiliary of conjecture.”

Other hamlets also have a clear perception boundary for the three hamlets of Hiwada, just as Ikegahora (64) does, but the examples are not given here.

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

75

61

65

Figure 6.1. The subjective dialect boundaries indicated by a respondent at Hiwada (57)

Figure 6.2. The subjective dialect boundaries indicated by a respondent at Kohiwada (58) Thus, the perception by the subjects in the hamlets surrounding Hiwada (57), Kohiwada (58), and Tomenohara (59) coincides in asserting that the people in the three hamlets speak the same dialect. Therefore, I decided to recognize that a region such as these three hamlets forms a dialect perception region.8 This case can be generalized in the following rule.

76

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” Rule 1: In arbitrary hamlets c, d, … of a, b, c, d, e, f, …, when the subjects of c, d, … perceive each other as the hamlets of the identical dialect and the subjects of the rest of the hamlets perceive c, d, … as the hamlets of a different dialect from theirs, it is recognized that c, d, … form a dialect perception region.

Let us take an example from Nashinokidaira (7) in Azumi Village. A subject in this area stated the following. Question A. Nashinokidaira (7), Inekoki (6), and Myôgadaira (5). But the speech in Myôgadaira (5) is just a little different. People here and in Inekoki (6) say dôi koto da ne for dô iu koto desu ka “What do you mean?” but people in Myôgadaira (5) say dôi koto daî for that. Question B. The speech in Shimashima (3), Hashiba (4), Ônota (1), and Ryûshima (2) is a little different. Their speech is like that of the town (which refers to the plain area around Matsumoto City). The speech in Nagawa (Nagawa Village) is somewhat arrogant. They say ô nani shitoru yo even to a stranger for nê nani o shite imasu ka “Oh, what are you doing?” But within Nagawa, they speak somewhat differently. The speech in Kiso is also a little different. Question C. The speech in Ônogawa (the Ônogawa district in Azumi Village) is quite different. It seems that the differences come from the ones in accent (not pitch accent, but more like intonation). Sometimes names of things are different. Here, we call an inferior person ware “you,” but they call him or her wara.

The map of dialect perception based on this subject’s responses is shown in Figure 6.3.

Figure 6.3. The subjective dialect boundaries indicated by a respondent at Nashinokidaira (7)

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

77

Figure 6.4. The subjective dialect boundaries indicated by a respondent at Inekoki (6)

The map shown in Figure 6.4 is based on the dialect perception of a subject from Inekoki (6), where, according to the subject from Nashinokidaira cited just above, people speak the same. Inekoki (6) is located next to Nashinokidaira. These two figures [i.e., 6.3 and 6.4] resemble one another, but the statements about the areas where people speak the same are not exactly identical. The subject from Inekoki (6) states as a reply to question A: Nashinokidaira (7) and Myôgadaira (5). Originally, this area is Inekoki (6), and Nashinokidaira (7) and Myôgadaira (5) are simply derived from it. So, the speech is the same in the three hamlets.

This statement shows a slight difference from the statement by the subject from Nashinokidaira (7). How, then, does a subject from Myôgadaira (5) perceive this? The speech in Myôgadaira (5) is the same as the one in Inekoki (6) and Nashinokidaira (7), but it might partially be close to the speech in town. This is because marriages have taken place between here and town recently.

This perception is understood as identical to the one by the subject from Nashinokidaira (7), who perceives that people in Nashinokidaira (7), Inekoki (6), and Myôgadaira (5) speak the same language but that Myôgadaira (5) is subdivided as a region where they speak slightly different. In this case, judging from the fact that all the hamlets surrounding these three hamlets recognize them as a single variety as well, it is appropriate to recognize the three hamlets as a dialect perception region and then to subdivide it into Myôgadaira (5) and the other two areas with a dotted line. This process can be generalized as Rule 11 as follows.

78

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” Rule 11: When a subregion where people speak a slightly different variety is recognized within a dialect perception region already recognized, two thirds or more of the hamlets that form the dialect perception region must agree.

There was very little difficulty in recognizing the dialect perception regions in the two cases described above because the subjects’ perceptions were uniformly very stable. However, this is not always the case. Let us take an example from Nagawa Village. A map was obtained based on the responses of the subjects in Tanokaya (14), located in the north of the village. Similarly, maps from Kanabara (19) and Kamiya (24), located in the middle and the south of the village, respectively, were obtained. The three maps are not shown here, but they have similarities and differences. One similarity is that all of them have perception boundaries between themselves and Nagawa Village. On the other hand, not all of them have the same perception boundaries inside the village itself. To examine where and how the fifteen hamlets place these perception boundaries, they are expressed quantitatively in the following way. A single line is given to the reply “a little different,” and a half line is given to the reply “slightly different” or “just a little different.” Then the lines are drawn automatically next to the hamlet where the perception boundaries are mentioned. Two half lines are counted as one full line. The result is shown in Figure 6.5. In this figure, some boundaries are made up of many lines, and others are not. The boundary with the most lines is between Tanokaya (14) and Furuyado (15). It has 11.5 lines, which means most hamlets have indicated a perception boundary there. The next most frequent boundary is the seven lines between Kamiya (24) and Hodaira (25), then six lines between Hodaira (25) and Kawaura (26), then 5.5 lines between Nyûyama (12) and Tsunogadaira (13) and between Kanabara (19) and Oidaira (20). There is probably no objection to recognizing Tanokaya (14) and areas to the north on the one hand and Furuyado (15) and areas to the south on the other as two different dialect perception regions. The number of the lines (11.5) exceeds two-thirds of all fifteen hamlets. The 11.5 lines consist of ten for “a little different” and three for “slightly different.” How should we judge the seven lines between Kamiya (24) and Hodaira (25)? The seven lines consist of four for “a little different” and six [counted as half lines and, therefore, totaling three] for “slightly different.” Judging from the number of the lines and from the content, it does not seem appropriate to recognize this boundary as dividing Kamiya (24) and Hodaira (25) into two separate dialect regions with the perception status of “a little different,” but it seems appropriate to recognize it as subdividing them into two dialect regions with the perception status of “slightly different” within a larger dialect region. The six lines between Hodaira (25) and Kawaura (26), and the 5.5 lines between Nyûyama (12) and Tsunogadaira (13) and between Kanabara (19) and Oidaira (20) should be recognized in the same manner. The number of the lines exceeds one-third of the total number of the hamlets, but does not reach two-thirds. The boundaries where the number of the lines does not reach one third are not meaningful as boundaries separating different dialect perception regions, and they may be ignored. The process described above can be generalized as the second rule, as in the following.

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

79

Figure 6.5. Names (in Japanese) and numbers of the fifteen localities that make up the village of Nagawa; the subjective limits indicated by the respondents (between sites #14 and #5) divide Nagawa into regions ‘d’ and ‘e’ (as shown in Figure 6.6) Rule 2: In arbitrary hamlets c, d, e, f, … out of the hamlets a, b, c, d, e, f, g, h, …, when the dialect perception by the subjects is not unified and varies depending on the subjects, a value of a full line is given to each response “a little different” and a value of a half line is given to each response “slightly different” or “just a little different.” The lines are drawn next to the hamlet where the subjects perceive that the speech is different. If the

80

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

NAGANO

KAMITAKARA

AZUMI

NYÛKAWA

GIFU TAKANE

NAGAWA

Figure 6.6. The sixteen dialect perception areas (based on responses to Questions A and B) total of the lines is or exceeds two thirds of the number of the hamlets, a boundary that separates two dialect perception regions is recognized, and if it is or exceeds one third but does not reach two thirds, a boundary that separates two perception regions of slightly different dialects within a dialect perception region is recognized. In this case, the hamlets c, d, e, f, … and the rest of the hamlets do not perceive each other as a whole or even partially as belonging to the same dialect perception region.

The dialect perception regions of the rest of the hamlets are determined using the two rules above. Figure 6.6 shows the dialect perception regions based on the responses to questions A and B in the research region. We call these regions small dialect perception regions. The sixteen small dialect perception regions obtained are identified by the letters a through p, as shown in Figure 6.6. The rules used to determine the small dialect perception regions were the following. By Rule 1 a, b, c, g, m, n, o By Rule 2 d, e, f, h, i, j, k, l, p

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

81

Before moving on, let us compare the results of this research with the ones found in Itoigawa [see Chapter 4, this volume]. The sixteen small dialect perception regions are roughly like the nineteen districts obtained from the primary dialect perceptions in the Itoigawa research [Figure 4.6]. In the Itoigawa research, the nineteen districts coincide with the villages of the subjects’ childhood. What about our results? Do the small dialect perception regions coincide with the villages of the subjects’ childhood? No. Taking the Nagano Prefecture side as an example, Azumi Village has had identical limits since Meiji 7 (1874)9, and Nagawa Village has had identical limits since the Tensho era (1573–92).10 Similarly on the Gifu Prefecture side, Takane Village and Nyûkawa Village have had identical limits since 1874.11 Thus, the sixteen small dialect perception regions do not coincide with the villages of the subjects’ childhood.12 In addition to Nagawa Village, which has had the same limits since the Edo period (1603-1868)13, as stated above, Azumi Village used to be four villages, namely Ônota, Shimashima, Inekoki, and Ônogawa.14 Takane Village used to be twelve villages, namely Inohana, Nakanoshuku, Nakabora, Shimonomukai, Hikage, Ôburui, Kamigahora, Ikegahora, Adanogô, Nomugi, Hiwada, and Kohiwada.15 Nyûkawa Village used to be seven villages, if restricted to the region of this research, namely Kute, Ikenomata, Iwaidani, Hatahoko, Dayoshi, Shioya, and Hiomo.16 All these are irrelevant to the small dialect perception regions. Do the dialect perception regions ever coincide with any non-linguistic facts? Yes, there is one non-linguistic fact with which they coincide very precisely. It is school districts. Needless to say, however, it is the school districts which were in place when the subjects grew up. The dialect perception regions that coincide perfectly with the school districts are a, b, g, i, j, k, l, m, n, o, and p, and many of the rest of the regions partially coincide with the school districts.17 Thus it is obvious that the small dialect perception regions are closely related to the school districts in this research region. Actually, the subjects of nine out of sixty-nine research sites replied to the questions by mentioning school districts and stated that speech is the same within the same school district. Then, when asked where people speak the same, did the subjects, consciously or unconsciously, give these responses mentioning school districts regardless of their relation to linguistic facts? I will discuss this in a later section. 2.3 The Regions Where People Speak So Differently That the Subjects Cannot Understand Let us move on to the summary of the responses to questions C, D, and E. First, let us look at the responses to question E. E. Where do people speak so differently that you can’t understand? How is it different?

Before starting this research, I expected that the respondents would cite places such as Kansai, Kyûshû, or Tôhoku, but the vast majority of them named Ettchû (or, referring to the same area, Toyama). Other responses were very few. Look at Table 6.1.18 Why, in both Nagano and Gifu Prefectures, do the vast majority of respondents cite Ettchû (Toyama)? Although we came across this response very frequently, I neglected to

82

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Table 6.1. Places named as so different that they could not be understood Nagano Prefecture

Gifu Prefecture

Village

Hata

Azumi

Nagawa

Kamitakara

Nyûkawa

Takane

Total

Locations surveyed

(1)

(10)

(15)

(17)

(10)

(15)

(69)

Place

Hida

0

2

0

0

0

0

2

name

Ettchû

1

2

12

6

6

11

38

given by

Tôhoku

0

1

3

0

0

0

4

respondents

Kyûshû

0

0

1

1

1

0

3

ask during the summer research what experiences of the subjects caused it. While we were doing this research, I also conducted research on shopping areas, and one of the questions was “Where do you buy medicine?” Almost all the hamlets answered this question by saying they bought medicine from Ettchû. In fact, it is traveling medicine sellers who come from Toyama once a year who supply medicine to this region. Judging from this response, I expected that the speech of the medicine sellers gave the impression of being so different that the subjects could not understand it. However, when I conducted research in Nagano Prefecture recently to confirm this, I realized that my judgment was wrong, at least in Nagano. The most frequent response to my new question was that people came from Toyama Prefecture a long time ago to make charcoal in the mountains and that their speech was almost incomprehensible. Additionally, groups of young people came over Tokugô Pass to be employed as silkworm raisers after planting rice in their home region during the rainy season, and their speech was incomprehensible. The latter response was given in Shimashina (3) and its surrounding areas. The subjects uniformly stated that there was no case in which they could not understand the speech of the medicine sellers. Some of the subjects could describe the characteristics of the Toyama dialect quite accurately. (I will discuss these points in the future due to the limited space in this paper.)19 This is probably because they have direct and continual experiences of speaking with people from Toyama. In addition, the vast majority of them responded “Ettchû” (or “Toyama”) in general, without specifying areas within. This is probably because they had the experience of speaking with the people, called “Ettchû-san” (Mr./Ms. Ettchû), who came over from Toyama as charcoal makers or silkworm raisers, rather than the experiences of traveling there or going there to work. On the other hand, the subjects have seldom had the experience of speaking with people from Kansai, Kyûshû, and Tôhoku, and this is, probably, the main reason why those areas are cited so seldom that they are not meaningful compared to the frequency of the response “Ettchû” (or “Toyama”).20 It should be noted that the Toyama dialect is perceived by more than half of the subjects as a variety so different that they cannot understand it — in other words, a totally different type of speech with a system which they are unaccustomed to, compared to their own. This can be seen in their comments about being “Completely lost” with the

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

83

Toyama dialect or its comprehension being “Out of the question,” or “I feel as if I were in a foreign country,” and “It is so different I can’t imitate it,” although some of them grasp its linguistic characteristics accurately. Looking at the region of this research, there are only two responses of “Hida” in Nagano Prefecture and only one response of “Hiwada in Takane Village” in Gifu Prefecture as regions where people speak so differently that they cannot be understood. This means that there is no region where people speak so differently that they cannot be understood within the region of this research. In other words, most subjects feel that they can communicate with each other in their own variety wherever they go in the region of this research, regardless of whether they are in Nagano or in Gifu. It also means that the relationship between Nagano and Gifu is not one of no interest or no relations, as the relationship is between this region and Tôhoku, for example, but one in which the regions are close to each other even though the Japanese Alps intervene. 2.4 The Regions Where People Speak Totally Differently, and the Regions Where People Speak Quite Differently Next is the summary of the perception of dialects based on the responses to question D, repeated here: D. Where do people speak totally differently from here? How is it different? Is it so different that you can’t understand it? (If the reply is yes, it is classified as a reply to question E.)

Figures 6.7 and 6.8 are the maps obtained by summarizing the responses of Nagano Prefecture and Gifu Prefecture separately and determining the thickness of the boundaries depending on the frequencies of the responses.21 In both figures, the prefecture limits are the primary boundaries of perception. In Gifu Prefecture, the perception boundaries around the three hamlets of Hiwada are very thick near the prefecture limits. The perception of dialects at this level divides the research region into two large parts, so it might be called a large dialect perception, and the region determined by a large dialect perception could be called a large dialect perception region. Question C is repeated here: C. Where do people speak quite differently from here? How is it different?

First, let us look at Figure 6.9, obtained by summarizing the responses in Nagano Prefecture. Unlike Figure 6.7, which is based on responses to question D, the boundary perception at the prefecture limits seen in Figure 6.9 is significantly small, and, instead, the boundary perception in the interior of Nagano Prefecture is conspicuous. Following the way of classifying small dialect perceptions as shown in Figure 6.6, this contrasts c against d and e, b against c, and b against d and e, due to the thick perception boundaries.

84

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 6.7. Nagano respondents’ responses to question D

Figure 6.8. Gifu respondents’ responses to question D

Next, let us look at Figure 6.10, obtained by summarizing the responses from Gifu Prefecture. It differs from Figure 6.8 in that the perception boundaries are found within the prefecture around the three hamlets of Hiwada, as also seen in Figure 6.9. A major

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

85

difference from Figure 6.9, however, is the thickness of the boundary perceptions at the prefecture limits. The boundaries are very thin in Figure 6.9, while they are very thick in Figure 6.10 and almost as thick as the ones in Figure 6.8. Now the question is why is the difference between the Nagano dialect and the Gifu dialect rarely perceived as “quite different” on the west side of the Japanese Alps while it is perceived as “quite different” by approximately half of the respondents from the east side. This is probably because the dialect of Nagano, located east of the Japanese Alps, is closer to that of Tokyo in grammar, vocabulary, and pitch accent than is the dialect of Gifu, and because the dialect of Gifu is always under the influence of Tokyo speech, which is considered to be superior.

Figure 6.9. Nagano respondents’ responses to question C

Perception of dialects at this level may be called an intermediate dialect perception, and the dialect region determined by such a perception is called an intermediate dialect perception region. Before moving on to the next section, I need to mention one other matter. There might be an objection that responses such as “so different that I can’t understand it,” “totally different,” “quite different,” and “a little different” are based on very unstable subjective judgments and that such perceptions cannot be reliable. I conducted another small program of research at twenty-six points in Nagano Prefecture to confirm this

86

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

matter. I personally carried out this research, and the researcher identity was therefore changed in all sites except for seven, and the respondents were different at two sites. Sometimes the utterances by the respondents were different from the ones given in the summer research. Nevertheless, after summarizing the perceptions of the subjects and creating maps following the same procedures, I obtained almost the same tendencies and almost the same results as the ones found in the summer research.

Figure 6.10. Gifu respondents’ responses to question C

2.5 The Integrated Perception Map Up to this point, dialect perception by the subjects has been discussed separately for each level; now let us incorporate the dialect perceptions from each level into a single map. Two points are given to the response “so different that I can’t understand it,” 1.5 points to the response “totally different,” 1 point to the response “quite different,” and the thickness of the boundaries is determined depending on the percentage. Figure 6.6 is overlaid onto the map obtained as described above. In this case, Figure 6.6 is not simply overlaid, but thin straight lines are drawn to separate the dialect perception regions on the map.22 The regions where people speak “slightly differently” within the same dialect perception region are separated by dotted lines. The map created by these procedures is the integrated map of perception as shown in Figure 6.11. In this figure, it is obvious that the primary boundaries which separate the dialect in this region into two are the prefecture limits. On the left side of the prefecture limits, Gifu Prefecture, the dialect of the three hamlets in Hiwada is perceived as a distinctive one surrounded by thick boundaries, and the rest is divided into three dialects, namely Takane Village (excluding the three hamlets of Hiwada), Nyûkawa Village, and Kamitakara Village, and each of the three is further divided into several small dialect perception

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

87

Figure 6.11. Integrated perception map

regions. In Nagano Prefecture, on the other hand, the primary region is divided into three dialects, namely Nagawa Village, the Ônogawa district of Azumi Village, and the rest of Azumi Village,23 and some of these three are further divided into several small dialect perception regions.

3.

The Characteristics of Dialects That the Subjects Described

Using a slightly different procedure from the Itoigawa research program, I asked the subjects to describe how the dialects are different by giving concrete examples so that I could make sure if linguistic facts made them perceive the difference or non-linguistic facts did, when, in fact, they perceived that dialects were different from theirs. The result was that the subjects from most research sites could describe several characteristics of the dialects which they perceived as different. Naturally, the next question is to what extent the characteristics they described are reliable. As mentioned earlier, I conducted a “confirming research” program on just this point at twenty-six research sites in Nagano Prefecture, selected from the sites studied in the summer and two months later. I obtained results which show that as high as 91.0% of all the occurrences and 87.9% of the items characteristic of the dialects are described accurately by these subjects in Nagano Prefecture.

88

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Although this confirmation research was conducted at only twenty-six sites, many of the characteristics described by the subjects from other research points in the summer and from other references were similar. Therefore, the descriptions given by subjects from our research sites were taken to be highly reliable. The details of this research can be found in Mase 1964b [see Chapter 7, this volume].

4.

Perception of Dialects and Dialect Districts

What is the relationship between the dialect perception regions, based on the perception of dialects by the speakers, and the dialect regions created by “objective” isoglosses? I used the results of the research conducted in the Nagano-Gifu boundary areas in the summer of 1962,24 to examine the relationship. First, let us look at the sixteen small dialect perception regions. A distribution of vocabulary that exactly coincides with the sixteen small dialect perception regions cannot be expected, but there are several words which coincide with them quite well, with some differences. One of them is amenbô “water strider.” Look at Figure 6.12. The circles and ovals indicate small dialect regions, and the dotted lines divide them into regions of slightly different dialects. It can be seen that some of the small dialect perception regions coincide with the distribution of the variants of amenbô quite precisely. As well as amenbô’s, those of otamajakushi “tadpole” and arijigoku “ant lion” reflect the small dialect perception regions quite faithfully. These words are closely tied with the daily life of the small regions, especially with children’s daily lives. It is no surprise that the distribution of the variants of these words coincides with the small dialect perception regions because the small dialect perception regions are closely related to the school districts. In addition to the words children use, there are other distributions of words which coincide partially with the small dialect perception regions. Next, let us look at the large dialect perception regions. There are many items that reflect them. For example, among grammar items, there is the elision of /s/ in verbs whose roots end in “s” [e.g., yielding otoita “dropped” for Standard Japanese otoshita or saita “pierced” for Standard Japanese sashita], the verb endings of the imperative form (the differentiation of -ro and -yo), the presence or absence of the sound change to -u of the adverbial form of adjectives, and the auxiliaries of conjecture (differentiation of -zura and -rô); among pitch accents, there are the group of words such as su “nest,” hito “person,” suna “sand,” komugi “wheat,” shigatsu “April,” keru “to kick,” and deta “got out” and the group of words such as nakatta “was not there”; in vocabulary, there are hôki “broom,” shimoyake “chilblains,” tosaka “cockscomb,” inaka “the bars on which to hang rice plants”, and hô “cheek.” The distribution sometimes reflects the prefecture limits, sometimes the three units of Hiwada is a region of mixed use, and sometimes those units show that they belong to Nagano rather than Gifu. The research region is divided, as shown in Figures 6.9 and 6.10 and explained in a previous section, into four intermediate dialect perception regions on the Gifu Prefecture side: Kamitakara Village, Nyûkawa Village, Takane Village (excluding Hiwada), and

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

89

WATER STRIDER

Figure 6.12. Correspondence of the dialect perception areas with linguistic features

Hiwada. In Nagano Prefecture it is divided into three, two in Azumi Village and the third all of Nagawa Village. There is no distribution of variants that reflects the intermediate dialect perception regions perfectly faithfully, but several reflect it quite faithfully. Such items are yodare “saliva,” otedama “beanbags,” ayatori “cat’s cradle,” and fukinotô “butterburr flower.” So far, I have described the relationship between isoglosses and the dialect perception regions. Next, I will explain how the dialect districts are connected with the dialect perception regions. In order to establish the dialect districts in this region, I adopted the method of imposing isoglosses. I again used the results of the language research at the Nagano-Gifu boundary area conducted in summer 1962. I selected seventy-one items of vocabulary, nine of grammar, two of phonology (in the narrow sense), and twenty-seven of pitch accent, a total of 109 items. The total number of items in the Nagano-Gifu research is 180, but I used 109 by eliminating the ones for which maps had not been completed, the ones which had too many variants to determine the isoglosses, the ones which had variants depending on stylistic rather than regional characteristics, and the ones which had many “cannot answer” responses.

90

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

There are many problems in determining isoglosses, but one important point should be noticed here. It might be self-evident, but I drew an isogloss outside the area of a distributed word. Let me explain using Figure 6.13. The words A, B, and C are distributed in the three hamlets X, Y, and Z, respectively. The isoglosses are drawn as in II instead of I. Therefore, when two units have a distribution of different words, two isoglosses are drawn between the two units. Similarly, when hamlets X and Y have A and B, respectively, and Z has both A and B, the isoglosses are drawn as in IV instead of III. The isoglosses are drawn taking transportation routes, geography, and government limits into account as much as possible, but sometimes they could not help but be drawn mechanically. There are many problems left in solving the procedure of determining isoglosses.

Figure 6.13. Placement of isoglosses

When two isoglosses are drawn, their weight must be considered. Let us take vocabulary as an example. Comparing amenbô “water strider” and seou “carry something on one’s back,” it is obviously the latter that is important in terms of frequency of use and

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

91

its centrality in the vocabulary. Therefore, when two isoglosses are imposed, weight is added to the isogloss of the more frequently used item, in contrast with the one less frequently used. It is easy to understand this methodologically, but it is quite hard to determine it with actual data. One method is to refer to previous research on basic vocabulary. However, this method evokes one question after another because the basis of our daily life is not the same as that of our subjects. For example, items such as shujn no suwaru seki “the place where the husband sits” and jizaikagi “a hook to hang a cooking pot above the fireplace” do not appear in any previous publication about basic vocabulary, but they might be critically important words for language use in farm life. There is another problem. An item that shows a small number of variants distributed in a wide area within the research region can be considered to be more important than an item that shows considerable variation distributed in tiny cell-like regions. In other words, an item that occupies a large space can be more important than one which occupies a small space. For example, nanoka “7th” (date) shows two variants, nanoka and nanuka, in the Nagano-Gifu boundary region, while arijigoku “ant lion” shows more than thirty variants, so the isogloss of the former has more weight. On the other hand, tosaka “cockscomb” shows two variants, while mabushii “too bright” shows seven variants, such as hidorokkoi, mamakkoi, and babakkoi. In this case, is it really appropriate to add more weight to the isogloss of tosaka rather than to the one of mabushii, adopting the same procedure? Also, from the viewpoint of system and from the history of the Japanese language, there are many problems associated with the weight of isoglosses. No solution to them could be obtained. I took examples from vocabulary only, but examples from grammar have the same problem. The number of items is smaller, but the difficulty of determining their weight increases. For example, which must be considered to be more important, whether the verb suru “to do” is conjugated as a Kami-ichidan verb or as a Shimo-ichidan verb, or whether or not otoshita “dropped” and sashita “pierced” have a deleted /s/ [see above]. Whether or not the variety in this region has the realis form is a critical characteristic, but which is more important, this fact or the distribution of the declarative auxiliary da “to be”? I thought about these kinds of questions but could not find any decisive procedure at this stage, so I leave behind the problems of determining the weight of isoglosses of grammar items as well. As for pitch accent, I determined the weight of isoglosses for independent items [i.e., items whose pitch accent configuration is not predicted by rule] and for items determined by the system because such determination is comparatively easy. I did not deal with the relative weights within either of these types of isoglosses. When isoglosses were drawn, the ones for vocabulary, grammar, phonology, and pitch accent were done separately. For vocabulary, I selected the following seventy-one words: Ayatori “cat’s cradle,” menko “pasteboard card,” kataashitobi “hopping,” takoage “kite flying,” takeuma “stilts,” otedama “beanbags,” kataguruma “carrying a child on one’s shoulders,” kakekko “running race,” kakurenbo “hide-and-seek,” onigokko “tag (game),”

92

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” dengurigaeshi “rolling forward,” kaminari “thunder,” yûdachi “evening shower,” amadare “raindrops,” geta ni hasamatta yuki “snow stuck between two supports of wooden clogs,” shimoyake “chilblains,” nanoka “7th (date),” myôgonichi no tsugi no hi “the day after the day after tomorrow,” hashi “edge,” shujin no suwaru seki “the place where the husband sits,” irori no fuchi no ki “wood at the edge of the fireplace,” jizaikagi “a hook to hang a cooking pot above the fireplace,” irori no ue no mono o kawakasu tana “a shelf to dry things above the fireplace,” kaikonchi “land that is being newly cultivated,” haza “the bars on which to hang rice plants,” kakashi “scarecrow,” wara o utsu toki no dai-ishi “a stone table to pat dried rice plants on,” kaiba-kiri no dôgu “a tool to cut fodder,” yuki-kaki “snow shoveling,” hôki “broom,” manaita “cutting board for cooking,” kinoko “mushroom,” tômorokoshi “corn,” sugina “field horsetail,” kuwa no mi “fruit of the mulberry,” hotarubukuro “bellflower,” fuki “butterbur,” fukinotô “butterburr flower,” tsukushi “spore of field horsetail,” toge “thorn,” kyûri no shichû “a stick to support cucumber plants,” tentômushi “ladybird,” arijigoku “ant lion,” amenbô “water strider,” otamajakushi “tadpole,” kaeru “frog,” hikigaeru “toad,” tonbo “dragonfly,” oniyanma “a type of dragonfly,” namekuji “slug,” sanagi “pupa,” tosaka “cockscomb,” o “tail,” mayuge “eyebrow,” shita “tongue,” tsubaki “slobber,” yodare “saliva,” hô “cheek,” kurubushi “ankle,” fukurahagi “calf,” ibiki “snoring,” hidako “blotched or mottled skin,” tabi no kohaze “closures for Japanese socks,” akanbô “baby,” akanbô o ireru kago “a basket to put a baby in” shaburu “to suck,” seou “to carry something on one’s back,” suteru “to discard,” katamuku “to incline,” mabushii “too bright,” kinakusai “to smell like something burning,” onbusuru “to hold a baby on one’s back”

I selected the following nine items from grammar: The verb suru “to do,” the verb iru “to be,” the presence or absence of the deletion of /s/ (examples with otoshita “dropped” and sashita “pierced” [see above]); the imperative ending of the verb (examples with miro “Look!” and shiro “Do it!”); the presence or absence of the realis form (examples with kuru “to come”); the presence or absence of sound change into -u with the adverbial form of adjectives (examples with shiroku naru “become white,” akaku naru “become red,” and kuroku naru “become black”); the auxiliary of conjecture (examples with iku darô “will (probably) go”); the declarative auxiliary da “to be” (examples with sôda “That’s right” and watashi no mono da “It’s mine”); the negative auxiliary (examples with ikanai “not go” and shinai “not do”)

I selected the following two items from phonology: Double vowels [ai] (examples with hai “ash”) and /kwa/ (examples with kashi “sweets” and kannon “Buddhist goddess of mercy”)

The following is the list of twenty items regarding pitch accent (independent of the system): Su “nest,” shita “below,” nashi “pear,” hito “person,” koe “voice,” oku “back (place),” suna “sand,” kita “north,” zô “elephant,” kimono “kimono, clothes,” komugi “wheat,” hashira “pillar,” hibashi “chopsticks with which to pick up burning coals,” makura “pillow,” megane “eye glasses,” hotaru “firefly,” shigatsu “April,” iru “to be,” keru “to kick,” and naderu “to stroke”

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

93

Besides these, I selected the following seven items determined by the system: The type with hi “sun,” ha “leaf,” and na “name”; the type with deta “got out” and mita “saw”; the type with ageta “raised” and karita “borrowed”; the type with atsumeru “to collect” and nagareru “to flow”; the type with hajimeta “started” and hirogeta “spread”; the type with akai “red” and asai “shallow”; and the type with nakatta “was not there” and yokatta “was good.”

When isoglosses of pitch accent are drawn, the items determined by the system were tentatively given a weight five times greater than those which are not determined by the system. There may be weight differences among the items determined by the system, but they are ignored. Adjacent hamlets never show completely the same vocabulary variants (Figure 6.14). At least five isoglosses are drawn between adjacent hamlets. On the maps, however, any isogloss that consists of fewer than ten isoglosses is omitted due to the complexity of the maps. The numbers in the lower right corners indicate the numbers of isoglosses by range.

Figure 6.14. Vocabulary isoglosses

94

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Not many items of grammar were examined, so the exact numbers of the grammatical isoglosses are represented directly by the thicknesses of the lines on the map (Figure 6.15), instead of by setting ranges depending on the numbers of isoglosses, as was done for the vocabulary and pitch accent items. The phonology map is not presented because the number of items is small and space is limited. The results with so-called double vowels /ai/ (examples with hai “ash”) are the following: In Nagano, [e:] is distributed everywhere except for the small dialect perception region d. In Gifu, [æ] is distributed in regions f and i, [ja:] in n (three hamlets in Hiwada), and [ai] in the rest. The results with /kwa/ are the following: /kwa/ is distributed in quite a lot of hamlets in Nagano, while it does not occur at all except for the three hamlets in Hiwada in Gifu.25 The map representing pitch accent is created by drawing boundaries based on ranges depending on the numbers of isoglosses (Figure 6.16).

Figure 6.15. Grammatical isoglosses

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

95

Figure 6.16. Pitch accent isoglosses

The procedure by which the maps of all the categories shown above were combined was not carried out. The maps show that the primary boundaries which separate the research region into two are the prefecture limits, in other words, the Japanese Alps. In Gifu Prefecture, at the second level, the three hamlets of Hiwada are recognized as a specific dialect region, and the rest is divided into dialect regions along with villages, which are further divided into regions at the third level. If one uses the identities of the small dialect perception regions and examines the regions of the third level village by village in all three maps, the three hamlets f, g, and j in Kamitakara Village are clearly separated, and the boundaries between h and i are conspicuous due to the distribution of the so-called double vowel [ai]. In Nyûkawa Village, k and l are clearly separated due to the distribution of vocabulary and pitch accent, and boundaries of slightly different dialects within k are also clearly indicated due to the distribution of vocabulary. In Takane

96

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Village, excluding Hiwada, m and o are separated due to the distribution of vocabulary and pitch accent. In p, the three hamlets Kibyû (60), Shôzô (61), and Ikegahora (64) have thick boundaries in relation to the other hamlets, and it might be possible to treat these three as an independent small dialect region, taking all the maps into account. In Nagano Prefecture, the research region is divided into three regions: a, b/c/d, and e. At the third level, a and b, and d and e form dialect regions, respectively. One peculiar point is that Hokoratôge (11) in Azumi Village is surrounded by quite conspicuously thick boundaries on each map, and it appears to be a transitional area between two dialect regions, which can be called a peripheral dialect region.26 Or it might be an independent small dialect region.27 Finally, Figure 6.17 summarizes the dialects in the region of this research, covering both the large and small dialect regions (The dialect districts are indicated by the names of prefectures, villages, districts, or hamlets, and the dialect regions at other levels are indicated by the name of the small dialect perception regions when they coincide. p1 refers to Kibyû (60), Shôzô (61), and Ikegahora (64), and p2 refers to the rest within p.). Finally, then, no lengthy explanation is necessary; as shown above, what made subjects perceive dialect differences is based on the actual characteristics of dialects, and the subjects grasp and describe the characteristics of these dialects very accurately. This fact leads us to the conclusion that the dialect perception regions coincide fairly well with the regions surrounded by the bundles of isoglosses of actual dialects — in other words, with the actual dialect areas in the regions of this research.28 Examples of no correspondence between the dialect perception regions and the actual dialect regions are also found, as in the dialect of Hokoratôge in Azumi Village, Nagano Prefecture and of the small dialect regions p1 and p2 in Takane Village, Gifu Prefecture. I do not consider that the results I have obtained may always apply to the other areas of the Nagano-Gifu boundary region. The results of this research are different from the conclusions reached in the Itoigawa research by Sibata, even after taking some slight differences of methodology into account. These differences may come from the differences of the internal organization or character of the hamlets as realized by political, social, cultural, and economic conditions along with underlying geographical conditions. Or they may come from the qualitative differences of the “community of interchange.” A “community of interchange” refers to a group of people who are connected by political, social, cultural, economic, and geographical conditions and by frequent internal communication. The relationship between the perception of dialects and actual dialect areas will be more clearly revealed if research is carried out on dialects in regions which have different conditions (in the sense described above) but using procedures similar to the ones used in this research.

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

97

a dialect Shimashima / Inekoki dialect b dialect Nagano dialect

Ônogawa dialect (excluding Hokoratôge dialect) Nagawa dialect

Hokoratôge dialect d dialect e dialect

Hiwada dialect

Kamitakara dialect Gifu dialect

f dialect g dialect h dialect i dialect j dialect k dialect

Nyûkawa dialect l dialect Takane dialect (excluding Hiwada dialect)

m dialect o dialect p1 dialect p2 dialect

Figure 6.17. Summary of the relationships among regional varieties

Notes 1. I would like to express my gratitude to Munemasa Tokugawa, who gave me a great deal of frequent advice and to the researchers who joined in this research and helped lead me to the conclusions presented here for their cooperation and friendship. I also am indebted for general direction to Hôgen kyôkai no ishiki [The perception of dialect boundaries] by Professor Takesi Sibata [see Chapter 4, this volume], but I worry that the examination of my data may be based on my misunderstanding of some of his points. I have also referred to Deutsche Mundartforschung by A. Bach (1950), for general ideas about dialects.

98

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

2. This research was conducted along with that of linguistic geography, and it contains some non-linguistic research such as that of regions where people shop and choose marriage partners. 3. The research sites are identified by Arabic numerals, and the numbers start at the eastern end of Nagano Prefecture and proceed westward. At first, I tried to follow the Itoigawa research and use the system that Kokuritsu Kokugo Kenkyûdjo [National Language Institute] adopted in creating the Japanese Language Map, but I adopted the method described above because the grid lines might be confused with isoglosses. In some maps, the numbers are not indicated. The following list gives the identification numbers and the names of the hamlets: 1 Ônota (Azumi Village), 2 Ryûshima (Hata Village), 3 Shimashima, 4 Hashiba, 5 Myôgadaira, 6 Inekoki, 7 Nashinokidaira, 8 Sawando, 9 Ônogawa, 10 Bandokoro, 11 Hokoratôge (3–11 Azumi Village), 12 Nyûyama, 13 Tsunogadaira, 14 Tanokaya, 15 Furuyado, 16 Kurokawado, 17 Yakatahara, 18 Komagahara, 19 Kanabara, 20 Oidaira, 21 Ôbira, 22 Sogura, 23 Yoriaido, 24 Kamiya, 25 Hodaira, 26 Kawaura (12–26 Nagawa Village), 27 Nakao, 28 Hirayu, 29 Fukuji, 30 Uejigane, 31 Hitoegane, 32 Tochio, 33 Murakami, 34 Kashiwate, 35 Imami, 36 Tadenomata, 37 Tagoroke, 38 Sasajima, 39 Akaoke, 40 Uwadaira, 41 Kanjikine, 42 Oimo, 43 Kuzuyama, 44 Hosogoe (27–44 Kamitakara Village), 45 Kamikute, 46 Shimokute, 47 Ikenomata, 48 Sôre, 49 Hatahoko, 50 Iwaidani, 51 Shioya, 52 Sode, 53 Dayoshi, 54 Myôgo (45–54 Nyûkawa Village), 55 Nomugi, 56 Adanogô, 57 Hiwada, 58 Kohiwada, 59 Tomenohara, 60 Kibyû, 61 Shôzô, 62 Kamigahora, 63 Ôburui, 64 Ikegahora, 65 Hikage, 66 Shimonomukai, 67 Nakanoshuku, 68 Nakabora, 69 Inohana. [Editor’s note: The location of this study area is shown in Figures 7.1, and 8.1.] 4. A teacher at Adachi Daiichi Junior High School, Adachi Ward, Tokyo. 5. An assistant at the International Christian University. 6. A graduate of the Nagano Prefecture Junior College. 7. A student of Japanese Linguistics, in the Department of Education of Shinshû University. 8. When several hamlets are surrounded by a line of dialect perception by many speakers, as in this case, we say that these hamlets “form a dialect perception region”; similarly, when a bundle of isoglosses of independent words surrounds a set of homogeneous and unified hamlets with critical dialect data, we say that these hamlets “form a dialect region.” 9. This is based on Meiji 7–nen kyû Chikuma-ken kanshô shirei no bu Azumi-gun gappei no bu [Volume of unification of Azumi County from the government commands of former Chikuma Prefecture in Meiji 7] in Kôbun hensan zen [Comprehensive edition of official documents]. 10. This is based on Kyûtaka kyûryô torishirabechô Shinano [Former region stipend, former territory examination records in Shinano] and others. 11. I am indebted to the mayor of Takane Village, Mr. Ueda, for the information on Takane Village, and the information on Nyûkawa Village is based on Nyûkawa sonshi [History of Nyûkawa Village] (1962). 12. The information of Kamitakara Village could not be confirmed, but it seems that it has had the same limits since the subjects’ childhood. 13. As stated above. In Shinano no kuni jû-gun takatsukechô Genroku no bu [Records of ten counties in Shinano Region, Volume for the Genroku era] and Shinano no kuni jû-gun takatsukechô Tenpô no bu [Records of ten counties in Shinano Region, Volume for the Tenpô era]; in Shinano shiryô [Shinano history records], Nagawa Village is written with a different kanji [i.e., character]. 14. This is based on the same reference as in 9. 15. I am indebted to the mayor of Takane Village, Mr. Ueda, for this. 16. This is based on Nyûkawa sonshi [The History of Nyûkawa Village]. 17. Region c consists of the hamlets in the Ônogawa district of Azumi Village, excluding Hokoratôge (11). The school district coincides with c excluding Hokoratôge (11), the area in which the languages are slightly different. Region d along with Hokoratôge (11) forms a school district. Region e consists of two

DIALECT CONSCIOUSNESS AND DIALECT DIVISIONS

99

school districts, and the boundaries of the two school districts coincide with the boundaries of two regions of slightly different dialects (the dotted line between Kanabara [19] and Oidaira [20]). Regions f and h form a school district. 18. This table shows the responses of items that occurred more than once. Responses such as “Akita” and “Kagoshima” are classified as “Tôhoku” and “Kyûshû,” respectively, and “Toyama” is classified as “Ettchû.” 19. Please refer to Mase (1964b) [see Chapter 7, this volume]. 20. When asked “How about the language in Tôhoku?” many of the subjects responded “I don’t know because I have never been there.” 21. The description of the dialects in sites or regions outside the area of this research is not indicated on the maps. 22. In drawing lines, geographic conditions, governmental districts, and transportation routes are taken into account as much as possible, but some lines are drawn mechanically. 23. The perception boundaries between Hashiba (4) and Myôgadaira (5) might be recognized, and the primary region may be considered to be divided into four, instead. 24. This research consists of 180 items of linguistic geographical research. The researchers were Iku Shimizu, and Shigeko Kobayashi, and I. Akikatsu Kagami (a PhD student at Tokyo Metropolitan University) did a part of the research region. Yasumichi Kobayashi, Yûko Uehara, and Hiroshi Minemura (Japanese Linguistics majors in the Education Department at Shinshû University) accompanied us. 25. Naturally, not all the research items result in the same distribution. The maps vary depending on the items. The research items were kashi “sweets,” kannon “Buddhist goddess of mercy,” and kaji “fire (accident)” for /kwa/, and gantan “New Year’s Day” and gaikoku “foreign country” for /gwa/. No /gwa/ is distributed at all, and among the three words for /kwa/, kashi “sweets” is distributed comparatively widely. /kwa/ is also found in other words, such as /tôkwa]i/ (tokage “lizard”) in hamlets 9, 10, and 11, and /kwanuki/ (inaka no tatebô “a vertical bar to hang rice plants on”) in hamlets 57, 58, and 59. Previously, this area had been considered not to have /kwa/ at all, and this is noteworthy. 26. A subject in Hokoratôge (11) stated that the language of this area was the same as Ônogawa (9), Bandokoro (10), and Sawando (8), but sometimes different. His daughter-in-law, who was accompanying him, objected by saying that his speech was somewhat between the varieties of Ônogawa (9) and Tsunogadaira (13). In this case, the daughter-in-law’s statement reflects the actual dialect uses. However, the elementary school for the children in this area used to be the Tsunogadaira Branch School of Nagawa Elementary School in Nagawa Village, but it changed to Ônogawa Elementary School in Azumi Village in the beginning of the Shôwa era (1926–1989), and the people in this area have relied on the route through Ônogawa (9) for food and other goods since World War II, so the general condition of speech in Hokoratôge (11) might be as this subject described. 27. If the dialect of Hokoratôge (11) is forced to belong to either one of the dialect regions, it will belong to dialect region d in Nagawa Village, taking the distribution of pitch accent into account. 28. Nevertheless, it is unnecessary to say that it is inappropriate to use dialect perception as a measurement in establishing dialect regions.

C 7 On Dialect Consciousness Dialect Characteristics Given by Speakers Yoshio Mase Translated by Daniel Long

1.

Dialect Boundaries and Speaker Consciousness of Them

When I go on dialect surveys, people often supply information about other dialects, such as “if you go to so-and-so place, they talk differently” or “over in so-and-so place, you can’t understand what they’re talking about.”1 It was Professor Misao Tôjô, the parent of Japanese dialectology, who positioned such dialect attitudes as the initial step in establishing dialect divisions. (See, for example, Tôjô 1938: 29–31; 1950; 1954: 11.) In 1959, Professor Takesi Sibata dealt with the topic of dialect consciousness and dialect boundaries, arriving at the following conclusions. a. The dialect consciousness of the residents of the Itoigawa region is divided into two levels. b. There are nineteen smaller divisions, but they have no relation to bundles of isoglosses. Even though informants were asked about dialect boundaries, they responded in terms of their consciousness of administrative boundaries (or more specifically, of community living regions). c. There were two larger divisions to the east and west of the Himekawa River, but these divisions also are thought to be based on non-linguistic characteristics. (Sibata 1959 [see Chapter 4], this volume) This study casts skepticism and doubt on the traditional dialect division theory of Misao Tôjô which held that dialect consciousness should be the starting point of dialect divisions. The fact that this study shook the very foundations of established theory made it extremely intriguing to me.2 I did, however, have one simple question in my mind as I read the Sibata paper. It contended that the nineteen smaller dialect divisions drawn on the basis of the language consciousness responses of the informants were actually based on a consciousness of

102

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

community living regions which were influenced by the administrative divisions in existence in the informants’ youth. But, in reality, do community living regions not have the power to shape dialect boundaries? The political, economic, and cultural, the public and the private dealings of life are quite often conducted within a single community living region. In many cases, these community living regions are the units by which issues are debated, decisions carried out, agricultural cooperatives formed, and elementary and junior high schools built. It is within these units that sports competitions are organized every autumn, that people bind together to maintain the roads, cooperate in the reaping of crops, organize festivals to celebrate the harvest, and oftentimes search for their marriage partners. These community living regions would have a very powerful influence in the uniformization (dôka) and leveling (heikinka) of dialects (and this influence would by no means be limited to language), which would lead to features which contrasted with those of neighboring dialects. If this is the situation in reality, then it would not be strange in the least to find that the community living regions from the days of the informants’ youths reflected dialect boundaries. Why then would we expect to find any discrepancies between the two? Moreover, when asked about dialect consciousness, did informants indeed give responses which were unrelated to linguistic factors? Is it not possible that the dialect consciousness that was triggered in them was indeed based on linguistic reality, but that this consciousness corresponded to the village boundaries existent in their youth? Furthermore, what results would we obtain if we attempted a similar survey in another location which differed geographically, politically, economically, and culturally? Spurred on by an interest in these questions, in July of 1963 I obtained the cooperation of three field researchers3 to conduct a survey similar but not identical to Sibata’s at another location.4 The questionnaire consisted of questions regarding dialect consciousness and linguistic features, along with a few non-linguistic questions about transportation and the extent of regions for shopping and finding marriage partners. We asked these questions while showing informants a map of the survey area with the names of the hamlets and some geographical features listed on it.5 I give the dialect consciousness questions below. A. B. C. D. E.

Where do they talk the same as they do here? (give specific names of hamlets) Where do they talk a little differently? What are some of these differences? Where do they talk quite differently? What are some of these differences? Where do they talk totally differently? What are some of these differences? Where do they talk so differently it is incomprehensible? What are some of these differences?

While these questions were based on those of the Itoigawa survey discussed in the Sibata paper, there were a few differences as well.6 The greatest innovation was in asking for specific examples of how dialects differed. If the dialect consciousness of informants was indeed based on non-linguistic facts, then it should be difficult for them to respond with concrete examples, and even if they could answer, their examples would not be realistic.

ON DIALECT CONSCIOUSNESS

103

On the other hand, if speakers’ consciousness of the difference with another dialect was indeed triggered by linguistic factors, then what are these factors? In this paper, I want to discuss in some detail the way in which speakers perceived dialectal differences, the patterns seen in their responses, and the reliability of the factors they gave.7 A discussion of the actual perceptual dialect divisions is found in Mase 1964a [see Chapter 6, this volume]. The type of research outlined here is one which, to my knowledge, has never been conducted before. It is one which, in spite of its importance to dialect research, has been left untouched until now.8

2.

Types of Speaker Responses Regarding Differences with Other Dialects

First, I will give specific examples of the ways in which speakers described the differences with other dialects, from one location each on the Nagano and Gifu sides of the prefectural border.9 (See Figures 6.6, 6.17, and 7.1.) At location 16 (Kurokawado, Nagawa Village, Minami-Azumi County, Nagano Prefecture), the informant answered question A in the following way: Locations 15, 17, 18, 19, 20, 21, 22, 23 talk the same as here. The only thing is that 21, 22, and 23 talk with a Kiso (especially that part known as Ogiso10) accent (namari),11 because they are part of that marriage sphere. At 23, they call girls by nicknames formed from the first syllable of their name plus -kkô, so that Motoko becomes Mokkô and Matsue becomes Makkô. This nicknaming is the same in Ogiso. With question B, the response was as follows: 24, 25, and 26 are somewhat different from here. There they answer ôi in response to the greeting konnichi wa, whereas we say ai here. And among these three, 25 is a bit different because they say sôgâ [with a low-low-high-low-low pitch accent pattern] (Standard Japanese sôka “is that so?”) where we would say sôga [with a high-low-low pitch accent pattern]. [See note 11 for an explanation of pitch accent assignment to morae.] 12, 13, and 14 are somewhat different. They say futtoru nyâ “it’s raining, isn’t it” for futteiru yo. There is even a song that goes “in Tsunogadaira and Nyûyama, it’s not just the cats that go nyâ,”12 so even people from other places know about them saying nyâ all the time. They are the only places around here that use nyâ or futtoru. Also, they say sôjagao or sôjarasugao (Standard Japanese sonna kotoga arumonodesuka “can such a thing be [the case]”), which surprises someone going there for the first time. Not even the old people at 14 use these expressions anymore. 12 only has one difference from 13 and 14. There the old people say ja (Standard Japanese da copula). The “Nyûyama ja talk” is famous around here. That area reeks of Hida talk. They say it’s true about those people coming over from Tsunokawa in Hida five or six hundred years ago.13

This speaker gave very specific information for questions C, D, and E as well, but since this gives the reader a good idea of the data, we will omit the rest. Next we will move on

104

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

to the speaker on the Gifu side (location 44, Hosogoe in Kamitakara Village, Yoshiko County). To question A, we received the following response. All the hamlets in the Nagakura school district have the same dialect.14

To question B, we received this response. The Tochio school district15 is a little bit different from here. Here we say oishii “delicious,” but there they say nmâi. For gyôsan aru ya nâ (Standard Japanese takusan aru nâ “there’s a lot, isn’t there”), they say jôni ya nâ. For nagai koto mienanda nâ (Standard Japanese nagai aida awanakatta “haven’t seen you for a long time”), they say hissa konanda nâ. The pronunciation there is different as well. Here we say akai “red,” but they say akyânâ. Locations 29, 30, and 31 have especially thick accents. 27 is part of the Tochio school district, but it is different from the other areas. There they are overly intimate and call people just by their names [i.e., omitting respect terms like -san]. Also, we refer to someone as omae “you” here, but there they say wari. The Kurabashira school district16 is a little different as well. Here we say warii ko ya nâ (Standard Japanese warui ko da nâ “such a bad child”), but there they say ozoi ko ya nâ. 28 deals with customers a lot, so their talk is more cultivated.

To question C, we received this response. Shinshû (Nagano Prefecture) is very different. The talk there resembles the way they talk in those historical drama movies. It is energetic. They say orâ igenê zo (Standard Japanese watashi wa ikenai zo “I can’t go”).

To question E, we received this response. Ettchû and here are as different as heaven and earth. We can’t understand a thing they say.

Of course, not all of the informants gave answers this systematic. Informants at two locations were not able to readily explain how other dialects differed from their own. Some informants asked the field workers why they didn’t just wait until they got to the other locations to find out for themselves how the people there talked. But almost every one of the informants in this region on the Nagano-Gifu border was able to give some specific linguistic examples of the way other people talked. It cannot be denied that this is an extremely important fact.

3.

Categories of Informant Responses

Next, we categorize the informants’ responses into “grammatical,” “phonological,” and “lexical” features, with unclassifiable responses grouped as “other.” We then divide these according to the area to which the comment referred (Table 7.1). In the analysis below, I divide the survey region into the sixteen “dialect consciousness” divisions (“a” through “p”) which I determined in another paper (Mase 1964a [Chapter 6, this volume]).17 For areas outside of the survey region, I use the village, town,

ON DIALECT CONSCIOUSNESS

105

Table 7.1. Tabulation of informant comments divided by linguistic category and location to which they refer Grammar Phonology Lexis Other Total Category Area N T N T N T N T N T Nagano all prefecture 12 04 07 03 02 02 14 10 35 19 Prefecture Azumi Village a* 03 02 00 00 02 01 01 01 06 04 b 05 03 04 03 08 06 03 02 20 14 c 15 04 22 16 06 05 08 06 51 31 Nagawa Village all village 05 05 03 03 00 00 03 02 11 10 d 36 09 03 03 04 03 03 03 46 18 e 06 05 14 08 03 03 11 08 34 24 Kiso region all region 00 00 00 00 03 03 01 01 04 04 Kiso Village 02 02 02 01 03 03 02 02 09 08 Kaida Village 00 00 04 01 13 10 08 07 25 18 Ôtaki Village 00 00 00 00 02 02 04 03 06 05 Fukushima Town 00 00 00 00 01 01 00 00 01 01 Hata-Azusagawa Village 04 02 01 01 00 00 00 00 05 03 Matsumoto City 01 01 00 00 00 00 04 04 05 05 Gifu all prefecture 19 05 01 01 05 05 00 00 25 11 Prefecture Kamitakara all village 00 00 01 01 03 03 01 01 05 05 Village f 06 04 06 04 03 03 08 05 23 16 g 00 00 05 04 00 00 09 06 14 10 h 07 03 04 03 08 06 02 02 21 14 i 04 03 14 07 08 07 06 06 32 23 j 00 00 00 00 04 04 05 03 09 07 other areas 02 02 02 02 02 02 04 03 10 09 Nyûkawa all village 00 00 00 00 01 01 04 03 05 04 Village k 01 01 02 02 10 07 03 03 16 13 l 00 00 03 02 02 02 04 04 09 08 other areas 00 00 00 00 00 00 02 01 02 01 Takane all village 01 01 02 02 02 02 04 04 09 09 Village m 05 04 03 03 03 03 08 08 19 18 n 00 00 02 01 00 00 04 04 06 05 o 06 03 10 07 09 08 11 09 36 27 p 03 03 02 01 01 01 06 06 12 11 Asahi Village 00 00 03 03 02 02 01 01 06 06 Shirakawa Village 00 00 00 00 00 00 01 01 01 01 Kamioka Town 01 01 02 02 00 00 04 03 07 06 Takayama City 00 00 02 02 01 01 02 02 05 05 Other places Ettchû 07 04 12 07 05 05 04 04 28 20 Kaga 01 01 00 00 00 00 00 00 01 01 Kyôto 00 00 00 00 01 01 00 00 01 01 Tôhoku 00 00 01 01 00 00 00 00 01 01 152 72 137 94 117 102 155 128 561 396 Total comments * Lower case letters a — p correspond to the locations on map 6.6. N is the total number of comments. T is the number of different types of comments.

city, or former feudal domain names which the informants themselves used. The reason that I use the feudal domain names Hida and Ettchû is because they appear so often in the informants’ comments, and because they do not correspond exactly to the modern-day prefectural boundaries. In the cases where informants commented only that Nagawa village used a certain form, without attributing it to a specific hamlet, I have categorized

106

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

these under “all village.” The “other areas” categories under Kamitakara Village and Nyûkawa Village indicate areas from those villages not included in the survey area. Looking at the overall results, we see that grammar and “other” responses are the most numerous, followed by phonology, with lexical items coming in last. On the other hand, because the same answer was often given for grammatical features, we see in the “T” column of Table 7.1 that this type of feature had the smallest variety of different answers. Let us look individually at the different categories to see what kind of answers were given. With grammar, the copula da and the verb iru [used as an auxiliary and as a verb of existence] were given forty and fourteen times, respectively, and account for over onethird of the total answers. Next are various sentence-final particles, with comments about nê being used at such-and-such a place,18 or about the use of châ (basically the same as Standard Japanese yo) in Ônogawa running neck and neck with twelve and eleven occurrences each. With phonology, the most numerous type of comments were the forty-three references to the short-long vowel distinction in such words as oi and ôî (Standard Japanese hai), accounting for almost one-third of the answers. There were numerous responses (twenty) regarding the monophthongization of the diphthong /ai/. Speakers in all but one part of Nagano showed no interest in this feature, with most of the references to it being from the Gifu side. This feature was identified with dialect divisions f, i, and n. The fact that the monophthongization occurs throughout almost all of Nagano and almost nowhere in Gifu is thought to have made it all the more salient for speakers there. Many of the phonological comments were very detailed. An informant at location 15 commented on the dialect of 12, 13, and 14 saying, “Around here we say [ye] (Standard Japanese /ie/ “house”), but they pronounce it [e] instead.”19 Over one fourth of the phonological comments (thirty-five) related to pitch, but most of these were about intonation20 with only five comments which could clearly be classified as accent.21 With lexical items, the most numerous occurrences (thirty) were of second person pronouns. Following this, there were eight comments on the term used towards the head of a household by his wife or by other villagers. It is only natural that these terms should be so salient to speakers, since they are features which have an extremely high frequency of usage and, being a type of treatment expression, are extremely conspicuous in the linguistic aspects of everyday life.22 Comments which could not be classified into any of the three categories discussed up to now were grouped as “other.” These include such comments as “their speech is good/bad,” “the way they talk’s not fancy,” or “their speech is four parts Shinshû and six parts Hida.” The largest subcategory here contained the thirty-two comments about the influences of other dialects, such as “the speech of so-and-so is mixed in,” or “the way they talk is influenced by so-and-so.” Following this were examples of “rough” (eleven), “good” (ten), “close to standard” (seven), “bad” (six) and “cultivated” (six). We see that many of the comments included here are evaluations. There were differences according to region in the type of comments given. For example, concerning the dialect of locations 12, 13, and 14, there was a very large number

ON DIALECT CONSCIOUSNESS

107

(thirty-six occurrences of nine different items) of grammatical features, while the number of other types of comments was conspicuously small. There were eight comments about the use of oru (including past tense otta, negative oran) for the Standard iru “to be,” seven comments about the use of the sentence-final particle nya (as in sô da nya, sô ja nya for the Standard Japanese sô da na, “that’s so, isn’t it”). Furthermore, a full ten informants mentioned the use of the copula ja in location 12. These examples show that the grammatical features of the dialects of these three hamlets stand out in comparison to the dialects around them. Regarding the dialect of the area called Ônogawa district in Azumi Village (locations 8, 9, 10, and 11), we received many responses regarding phonology,23 followed by those regarding grammar,24 and only a few lexical or “other” responses. A comment that typifies the responses is one by an informant from location 23 who said, “with the Ônogawa River dialect, it’s not the words themselves25 that are different, but the pronunciation.” Let us turn now to the “all prefecture” responses for the Nagano dialect. We see from the table that, while there were also many comments concerning grammar, Nagano (with fourteen occurrences of ten different responses) was the object of more “other” comments than all the other areas. The large number of different responses would seem to indicate a great degree of variety in the perceptions of Nagano dialect by Gifu speakers. However, a look at the actual individual responses shows that these answers can be grouped together into a few types and thus indicates that Gifu informants actually have a rather uniform view of the Nagano dialect. With the dialects on the Gifu side as well, we see trends, to some extent, in the responses for each of the dialects. 4.

The Reliability of Informant Responses: The Verification Survey

Up to this point, we have seen some of the responses from our informants, but how reliable are these responses? The ideal survey to check for this would have to be conducted at every location which informants mentioned in their comments, or, if this is impossible, to at least survey the sixty-nine locations covered in the initial survey. I made cards of the dialect characteristics which the informants had given in the summer survey, discarding those which were not surveyable; for example, comments such as “the way they word things is totally different,” “they talk fast and in places they draw things out,” “they put strange things on the end of their sentences,” and “they don’t talk fancy.” After omitting these from the data, we were left with fifty-eight different characteristics (from a total of one hundred responses) out of an initial 101 different characteristics (from a total of 168 responses). We set out to determine if these were indeed in use at the locations referred to. As an example, at location 12, we asked the following questions. 4.1 Grammatical Items 1. 2.

Do you say oru or otta for “(a person) is/was (in a place)”? When you say “look at that” (Standard Japanese are wo miro), do you say are mê?

108 3. 4. 5. 6. 7.

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Do you Do you Do you Do you etc.

say say say say

sô ja for sô da (“it is so”)? sô ja nya for sô desu yo (“it is so” [emphatic])? ame ga futtoru for ame ga futte iru (“it is raining”)? sô ja nô for sô da nâ (“that’s so, isn’t it”)?

4.2 Phonological Items “What do you call the indentations in the corner of the mouth that people get when they laugh?” We were trying to elicit ekubo “dimple” to determine if the initial syllable was pronounced [e] or not. 4.3 Lexical Items 1. 2. 3. 4.

Do you say wanda for omaetachi “you” (plural)? Do you call the head of a household gotê (shujin in Standard Japanese)? Do you call this farm implement a manno (Standard Japanese maguwa)? (Asked while showing a picture of the object.) In this picture, what are there three of? (Showing a picture of three houses) If the informant answered uchi, we asked whether or not he also used the word e.

Using questions such as those above, we surveyed the twenty-six Nagano locations asking which linguistic features were indeed in use at that location. The results showed that the reports of these linguistic features were correct in the case of 87.9% (fifty-one out of fiftyeight) of the different items reported, or 91.0% (ninety-one out of one hundred) of the total number of features reported. These numbers show that the reports had a very high level of accuracy.26 This verification survey was carried out only at the twenty-six locations on the Nagano side. However, among the linguistic features reported to be in use on the Gifu side were a large number of items verifiable because (a) they had already been included as survey items in the summer survey questionnaire, or (b) they appeared in other referenceable materials concerning these dialects. A look at these materials suggests that the informants’ reports concerning the linguistic features in use in these dialects have a high level of accuracy. Examples of items verifiable from the survey are: ja (Standard Japanese da copula), reported eleven times; oru (Standard Japanese iru verb of existence for animate agents), four times; shirônaru (Standard Japanese shirokunaru “to become white”), twice. Among the features verifiable from the available literature on the dialect were the use of the following in the Kaida village dialect: ware (Standard Japanese anata “you”), four times; waira (Standard Japanese anatatachi “you” plural), once; una (“you”), once; unrâ (“you” plural), once; tô (Standard Japanese otôsan “father”), once; teppo (Standard Japanese uso “a lie”), once. The existence of all these items was verified by their inclusion in the

ON DIALECT CONSCIOUSNESS

109

booklet called Research on the Kaida Dialect put together by Kaida Junior High School. In this way, we found that at least six out of the ten different lexical items (eight out of thirteen occurrences) reported for Kaida were actually in use there. Furthermore, for items reported to be used in the Ettchû dialect, we were able to verify the existence of grammatical features such as sakai (Standard Japanese kara “because”), once; namo, nanmo (Standard Japanese na sentence final particle), four times. For phonological features: [ittwu:] for Standard Japanese [ettwu:] “Ettchû,” name of a region, once; [su], [zu], [tsu] for Standard Japanese syllables [wi], [(d)Ši], [tsi], four times; [mizukai] for Standard Japanese [miŠikai] “short,” once; [kwawi] for Standard Japanese [kawi] “sweets,” once; [kwa] for Standard Japanese syllable [ka], once. These were verified by Iwai (1961), Ôta (1929) and Tôjô (1951). From all of the preceding examples, we can conclude that our informants had an accurate grasp of the linguistic features of the hamlets, villages, or broader areas which they were reporting on.

5.

Common Trends in the Answers of the Informants

Finally, let us look at two or three of the general tendencies found among the informant reports concerning the dialect usage in other areas. Informants’ comments tend to focus on the differences with other dialects, ignoring (with only a few exceptions) similarities. For example, with informants on the Gifu side, there were three occurrences of the comment that the hamlet of Hiwada (locations 57, 58, 59) used the copula da. This cannot apparently be taken as evidence that people in Hiwada do not use the other variant, however, because we have a report (although only a single report) of the usage of ja in Hiwada. In reality, ja has moved in only recently, and is usurping the territory of da. As a Hiwada informant (at location 57) told us, the old people a long time ago used da, but now most people use the more refined-sounding ja. Similarly, the informant at location 32 said about the dialect of location 27, “the speech in Nakao reeks of Nagano, what with them saying ikanê “not go.” This is not to say, however, that Nakao hasn’t begun using the other variant ikan. An informant in Nakao told us, “I got laughed at when I said ‘nomenê’ (‘can’t drink’). People don’t use the word -nê anymore.” Indeed, this informant used the other variant of the negative tsukawan (“don’t use”) in conveying this very information to us. From these two examples, we see that informants’ comments concerning other dialects are concentrated on the differences, ignoring the similarities. Furthermore, in these cases, the comments refer to old features of dialects, even to the extent that the feature may have already become a thing of the past. The informant at location 20 commented that at 12 and 13, they used ja for the copula. The informant at 13 negated this saying “people who would be 100 or 120 if they were alive today used to use ja, but it’s not used now.” It is entirely conceivable that informants are commenting on dialect characteristics which they heard as youngsters from the elderly people of other villages and which have stuck in their memories all this time. These characteristics may have long since disappeared in

110

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

these dialects. I mentioned that informants’ comments often concentrated on differences between dialects, ignoring similarities. There are, however, some interesting exceptions to this. These are cases in which the speaker’s own dialect is being influenced right before his eyes, and his attention is caught up in the homogeneity resulting from this influence. In these cases, informants tend not to point out differences between the influencing dialect and their own. Rather they report great differences with dialects which are either not influencing their dialect or which are actually being influenced by it. Let us look at the examples of the Ônogawa hamlets (8, 9, 10, 11) at the end of a dead-end valley on the Nagano side of the survey area, and the Hiwada hamlets (57, 58, 59) on the Gifu side. When asked about the characteristics of these two dialects, informants in hamlets surrounding them gave numerous concrete examples. In addition, regarding the Ônogawa dialect, informants replied that they “couldn’t imitate it” (three people), that it was “unique” (one), “generally not a very good dialect” (one), “sounds like Iwate or somewhere” (one). Concerning the Hiwada dialect as well, informants in other hamlets emphasized the differences with their own speech, describing the dialect as “bad” (two), “lacking in manners” (one), “similar to the speech of Kiso” (two), “like the speech of Kaida” (one), and “similar to the speech of Nagawa” (one). How then do the informants in Ônogawa and Hiwada view differences between their own dialect and those of surrounding areas or those which are influencing their dialect?27 While informants in Ônogawa did sense some differences between theirs and influencing dialects, the perceptions were much weaker than the differences reported by the speakers of those influencing dialects. What is more, Ônogawa informants gave almost no specific examples to illustrate what few differences they did report. They did, on the other hand, give numerous examples of differences between their dialect and that of Nagawa which is not exerting influence on them. Similarly, Hiwada speakers gave no specific examples of differences between their dialect and those influencing it, only a few vague comments such as “the speech in Kamigahora (location 62) is good” (from the informant at 59), or “the speech is better down below Kamigahora” (meaning locations 63 and 64; from the informant at 57). Furthermore, they emphasized the differences between their dialect and those of Kiso or Kaida (which informants in surrounding hamlets had told us were “similar” or “the same” as their own). Conversely, Hiwada informants gave specific examples of characteristics of the dialects of locations 55 (Nomugi) and 56 (Adanogô) which are not influencing their own. These tendencies are not peculiar to Ônogawa and Hiwada, but are seen to a greater or lesser extent in other dialects under the same conditions.

6.

In Summary

In this paper, I have examined the reports of speakers regarding linguistic features used in other dialect areas, using data from a survey of the Nagano-Gifu border region. We can sum up our findings from this study as follows:

ON DIALECT CONSCIOUSNESS

1. 2. 3. 4. 5.

6.

111

Almost all of our informants were able to give specific linguistic features for those dialect areas which they perceived as “different.” The types of linguistic features reported showed a clear tendency to differ according to the dialect being reported on. The linguistic features reported as being used in other dialects showed a very high level of accuracy. Informant perceptions tended to focus on dialect differences, leaving similarities largely ignored. However, when discussing influencing dialects, we saw the opposite trend. Speakers tended to focus on the similarities resulting from their dialect borrowing features from these influential dialects, and they seldom mentioned differences between the two. Informants’ reports of the linguistic features of other dialects tended to focus on older features, and some of the comments referred to features which have become obsolete in the dialects for which they were reported.

Notes 1. I would like to express my thanks to the numerous informants who cooperated with me for long hours in this survey and to the village officials and school teachers who assisted with the selection of the informants. I also wish to thank the people who, despite the transportation problems associated with this remote area, worked on the survey. 2. Before reading the Sibata paper, I had heard Father Willem Grootaers give a lecture at a 1959 spring semester course in dialectology at Tokyo Metropolitan University graduate school entitled “Hôgen kyôkaisen wo megutte” [Regarding Dialect Boundaries] and had become interested in the subject [Grootaers 1959]. 3. The field researchers who assisted in this survey were: Iku Shimizu and Shigeko Kobayashi. Kiyoko Hokari also surveyed some locations and Tomoka Yoshida and Reiko Nakagawa assisted with some surveys. 4. The survey location chosen was the rectangular area shown on Figure 7.1 which straddles the Hida Mountain Range known as “the roof of Japan.” The survey area on the Nagano prefecture side consisted of all of Nagawa Village and Azumi Village (both in Minami Azumi County), and one survey location in Hata Village (Higashi Chikuma County). On the Gifu prefecture side, we had all of Takane Village, and about half of Nyûkawa Village (both in Ôno County), and half of Kamitakara Village (Yoshiki County). We initially planned to survey every hamlet on the 1:50,000 map produced by the Kokudo Chiri-in [the Geographical Survey Institute], but some hamlets were no longer inhabited or could not for some reason be surveyed. The locations surveyed are the sixty-nine hamlets indicated by number in Figure 7.1 [and in Figure 6.6, this volume] and by name [in Note 3, Chapter 6, this volume]. 5. We surveyed one informant per hamlet, preferably male, over 55, born and raised in that hamlet, and selected by the village office or local school. 6. Sibata (1959: 2 [see Chapter 4, this volume]) explains the survey question as follows. “We asked: Is the speech here different from the speech of the neighboring hamlet? (with the following choices in order of difference) (1) not different, (2) a little different, (3) quite different, (4) not at all the same. We put this question to each informant and wrote down the names of hamlets or areas which were given for each choice.” 7. The survey for the verification of these characteristics was conducted in September 1963 after the results had been systematized to some extent. These surveys were conducted by Mase on the twenty-six locations in Nagano, and Hiroshi Minemura assisted with some surveys.

112

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 7.1. Survey locations 8. Kikuo Nomoto has just published a paper (Nomoto 1963 [see Chapter 5, this volume]) in which he discusses the comments of informants about the characteristics of neighboring dialects, but it should be noted that his methodology differs from mine. 9. Although during the survey the specific dialect examples which informants gave were written down in the International Phonetic Alphabet, in this paper we have rendered these in the regular script whenever possible for easy readability. Informants’ remarks were not written down verbatim, but were paraphrased in Standard Japanese to include the main gist of the comment. 10. This refers to the hamlet of Ogiso in Kiso Village, Nishi-chikuma County, Nagano Prefecture. 11. Translator’s note: In this paper, I have translated namari, referring to the pronunciation characteristics of a speech variety, as “accent.” To avoid confusion, I have translated akusento, referring to the high-low prosodic qualities which often distinguish lexical items as “pitch accent.” Note that high or low pitch accents are assigned to morae, not syllables, in Japanese; therefore, for example, long vowels contain two such moraic timing units. 12. Translator’s note: nyâ is the onomatopoeic term corresponding to the English “meow.” 13. The informant explained that Tusunogawa was in a place below Furukawa called Nishizanchû. This refers to the hamlet of Tsunokawa in Kawai Village (Yoshiki County), next to Furukawa Town. 14. The Nagakura school district is the one to which location 44 belongs.

ON DIALECT CONSCIOUSNESS

113

15. The Tochio school district lies to the east of the Nagakura district, and contains locations 27 through 41. In the past, 27, 29, 30, and 31 formed a separate district. 16. The Kurabashira school district is to the west of Nagakura. 17. The reader is referred to Mase (1964a) [Chapter 6, this volume] for details regarding the “dialect consciousness” divisions. These divisions a through p correspond to the following numbers in Figure 6.6: a (1–4), b (5–7), c (8–11), d (12–14), e (15–26), f (27), g (28), h (29–31), i (32–41), j (42–44), k (45–52), l (53–54), m (55), n (56), o (57–59), p (60–69) 18. Most of these comments attributed the use of ne very generally to the Nagano dialect. There was one comment each that it was used at Nomugi (location 55), and the three hamlets of Hiwada (locations 57, 58, 59) [see Figure 6.6, this volume]. 19. The use of this phonological feature was verified by the follow-up survey. 20. There were no less than fifteen informants here that used the word “accent” in their comments, but it was found that only one of these was using the word in the sense that Japanese linguists use it to mean the “pitch accent” of individual words. 21. These five comments were all regarding a differentiation in pitch accent forms. For example, the informant at Tadenomata (36) told us, “Here we say shimo no hô [low-high-low-low-low pitch accent pattern], but at Akaoke (39), they say shimo no hô [low-high-high-high-low pitch accent pattern].” [See note 11 for an explanation of the assignment of pitch accent to morae in Japanese.] We were surprised that an elderly speaker of this area would have noticed something as subtle as the pitch accent difference. 22. We received no less than ninety comments regarding the use of “treatment expressions” (honorifics and the like), revealing the extent to which such characteristics are noticed by the average speaker. 23. Of the phonologically related comments, most related to prosody. For example, at Hashiba (location 4), we were told “their words sag in the middle” which was explained as meaning that the beginning and end of an utterance had a higher pitch than the middle. We received comments such as, “They talk in singsong” (location 12), and “The end of their words shoot up” (location 2). There were even nine different comments (a total of twelve in all) which gave specific examples of prosodic differences as in the following: “In Ônogawa, they say wârya mâdoke iku kayo” (with the initial and penultimate syllables higher than the rest, Standard Japanese omae wa ittai doko e iku o ka, “where in the world are you going?”). 24. Of the grammar-related comments, no less than eleven were in reference to the sentence-final particle châ. 25. We took the expression “the words themselves” to mean lexical items. 26. Partially correct examples were treated as in the following example. The commonly heard comment “They say -shitoru a lot in Nagawa Village” was deemed correct even though in reality this form is only used in part of this village (locations 12, 13, and 14). If this type of comment had been treated as incorrect, the figures here would have been lower. 27. The dialect influencing Ônogawa is that of the large city of Matsumoto and the surrounding plain, as well as the dialects at the mouth of the valley. The dialect influencing Hiwada is survey location 62, Kamigahora, the center of Takane Village and the location of the village office.

C 8 The Discussion Surrounding the Subjective Boundaries of Dialects Willem A. Grootaers Translated by Lawrence Kuiper

1.

Introduction

My young colleague and friend, Mr. Mase, amiably permitted me to present, in a slightly shortened form, his important study of the subjective boundaries of the Gifu-Nagano region. While still retaining from his article the majority of the maps and the integral text of the methodological and critical section, it was possible, for an international reader, to omit the lists of villages and the linguistic and topographical details that are of interest only to the specialist in Japanese studies; such readers will naturally refer to the original text.1 As Mr. Mase says at the beginning of his work, it is the survey undertaken by the team Sibata-Grootaers-Tokugawa in 1957 in the Itoigawa region (Niigata Province) that introduced into Japanese linguistics the concept of “subjective dialect boundaries.” The part of the questionnaire relating to these subjective boundaries had in part been suggested by a reference to a written correspondence survey done by a Dutch scholar, W. G. Rensink, in 1955 [see Chapter 1, this volume]; I say “in part” because Japanese dialectology has had no more burning problem for the last half-century than that of dialect division and, therefore, that of the existence of dialect unities.2 The pioneer in this field, Professor Misao Tôjô, from the beginning of his studies, had taken as his starting point the respondents’ subjective awareness of the existence of their own dialect as opposed to different neighboring dialects. For us, therefore, consideration of such subjective classification has been very important. Since the survey of Itoigawa formed the basis for the first modern linguistic atlas ever attempted in Japan, it was necessary to put considerable emphasis on this problem which has long preoccupied traditionally oriented Japanese dialectologists. The results of our survey were published with identical maps but with a text adapted to different readers — on the one hand by Sibata 1959 [see Chapter 4, this volume] in Japanese, and on the other by Grootaers in English (1959c). The conclusion of our study can be stated as follows:

116

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” The subjective awareness that speakers have of the dialect unit is first based on an extralinguistic factor — in other words, on daily life in the village; thus subjective borders of the first degree (based on responses to the question: “Where do people talk like you?”) indicated by the inhabitants of the region of Itoigawa coincide with the administrative village boundaries which existed in their youth, without reflecting objective linguistic boundaries. But a second category with wider boundaries (based on responses to the question: “Where is the language different to the point of no longer understanding it?”) seems to combine the extra-linguistic elements of the medieval feudal political community with important isoglosses that still now separate the territories of ancient medieval feudal fiefs.

In general one can say that our conclusion was negative, throwing doubt on the objective value of dialect awareness as a basic criterion for determining dialect units. There were two interesting reactions to this study, one from Japan and one from Europe. I first present my translations of the originals.

2.

The Remarks of Tôjô (Tokyo) and Hammarström (Uppsala)

Professor Tôjô, who had been able to read the article of Sibata in manuscript [i.e., Chapter 4, this volume], added two pages of commentary to it that appeared in the same number 36 of the journal Studies in Japanese Linguistics, pp. 31–32. Here is the translation: Some remarks on the use I made of the term “dialect awareness.” 1. When I decide to study a dialect, I first choose a territory where it seems to me that a common language is in use, whether it be a large or a small dialect. Naturally the territory covered by the dialect in question will not be clearly outlined until after the survey, since before the survey one could not have had a clear conception of it. It is therefore precisely for this reason that I place at the starting point of a dialect study the awareness of the boundaries of their dialect that the inhabitants of the region have. What I call “awareness of their dialect” is the consciousness of their linguistic region that the inhabitants as a group possess, according to common sense and tradition. It is not necessary to question each one of the inhabitants to make sure of this awareness. Thus for the people of Kanto (the eastern half of the main island with the capital Tokyo) the language of Kyoto-Osaka (principal cities of the western half of the main island) is different in the consciousness of each inhabitant without exception. In the northeast of Japan, the speech of Tsugaru and Nambu (two feudal fiefs sharing the present day provinces of Aomori and Iwate) differ, and everyone knows it. Finally, in the city of Tokyo itself, during the Edo period until the beginning of the Meiji period (approximately 1600–1868), the language of Yamate (the “High City,” affluent neighborhoods) was considered to be different from that of Shita-machi (the “Low City,” working class neighborhoods). This is doubtless a very vague concept, according to which one would be very hard pressed to determine the exact boundaries. Here are a few more examples: the dialect of the region Chugoku (provinces of Okayama, Hiroshima and Yamaguchi) and that of Izumo (province of Shimane, to the north of the three preceding); the dialect of Kochi (south of the island of Shikoku) and that of Yawata (west of the same island); the dialect

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES of the province of Yamagata and that of Kasumi-cho. What I understand by the words “dialect awareness” is thus a kind of general opinion spread out over a region. The survey carried out in Itoigawa by Sibata and his colleagues was concerned with the detailed recording of the personal opinion of each respondent. By comparing the responses to extra-linguistic elements, they were able to determine the existence of nineteen small areas [see Figure 4.6, this volume]. This is an innovation in dialectology, and the results are extremely important. However, the concept of “dialect awareness” that I find in this study is quite far from the one I formerly meant by this term. I will thus endeavor to affirm that, although I may need to make an exception for the strip of land that neighbors the province of Toyama (the westernmost point of map 8 and the text in Grootaers 1959c: 378 [see Chapter 4, Figure 4.8, this volume]), the region covered by the survey of Itoigawa is totally reducible to one single dialect group in the sense that I described above. Even the valleys, such as the Hayakawa (the easternmost point of the same map) would not form a dialect territory apart from the others, since the common sense judgment of the inhabitants does not regard it as a distinct unity. 2. Dialect awareness is useful to me, therefore, as a first step in my dialect exploration, but when it becomes a question of definitively fixing dialect divisions, it is objective dialect areas that serve as criteria, and one is no longer tied to the “dialect awareness” that was used as a starting point. The division based on dialect awareness was thus no more than a touchstone, a sketch before the painting. After having traced the isoglosses, one can throw off the “dialect awareness.” Let us now look at a concrete example. A. The dialect boundary between the western and eastern Japanese dialect groups can be placed at the border of the civil provinces of Shizuoka and Aichi (approximately at the 137″ 30′ degree of longitude) if one follows linguistic common sense. But I myself place it rather at the border between the Shizuoka and Mie provinces (136″ 40′). If one considers only the most important bundles of isoglosses, one finds three by following the southern coast from east to west: the isoglosses of the Fuji River (longitude 138″ 40′), of Hamano Lake (longitude 137″ 40′) and of the Kiso river (longitude 136” 40′). I am led to prefer the choice of this last bundle, but this isn’t the place to give detailed reasons. I will point out that it is not an easy job to fix the boundaries of dialect units by superposing area maps because one must first determine the relative value of each isogloss. B. Naturally there are cases where dialect unities existing in the linguistic awareness of subjects correspond in no way to the areas determined by a dialect survey. Although we do not have numerous cases of this type, one could cite that of sixty or seventy villages scattered pretty much throughout the countryside (like Shirakawa in the Gifu province, etc.), where the descendants of the Heike family have settled (attached by legend to the feudal Taira family, deposed in the 12th century); or even those cases of islets of population whose origin is to be sought in the changes of feudal obedience; or, finally, even the distinction between the “High” and “Low” cities in Tokyo already referred to, which persists today in the linguistic awareness of the inhabitants, even though in reality the objective linguistic distinctions between the two are erased. C. It frequently happens that the boundaries of dialect units determined by linguistic awareness must be modified after the study of isogloss bundles. Thus what one calls the dialect of Izumo (province of Shimane) and the results of research that I undertook in Izumo-Hoki (the border between the Shimane and Tottori provinces) do not coincide,

117

118

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” especially for the eastern part of the territory, the western marches of Tottori, and the Island of Oki. D. There are also cases where, due to an important bundle of isoglosses, one may trace the objective boundaries of a dialect unit which, however, in no way correspond to the linguistic awareness of the inhabitants. Thus for a long time we have recognized the isoglosses of the Tone River (which, following the western boundaries of Tochigi and Ibaraki, divides the Kanto territory north of Tokyo into two dialects). However, on the Tochigi side, there is no common linguistic awareness of this boundary’s existence. We shall not be surprised if in the future similar cases are discovered in still greater number. 3. It is clear that linguistic awareness of dialect units does not always correspond to objective distinctions still existing at the present time. This awareness can reflect an older common sentiment. In any case, there is no doubt that as a starting point of dialect studies, or even as a contribution to historical dialectology, the use of dialect awareness is helpful, even though after having discovered bundles of isoglosses, one later needs to correct it.

If I have cited Professor Tôjô’s remarks extensively, it is to give the western dialectologist a chance to hear the voice of the founder of Japanese dialectology. The present study will provide a further opportunity for examining some of the assumptions that will explain our position towards Professor Tôjô’s “dialect awareness” and his numerous successors. However, before putting aside this text, it is necessary to note that dialect surveys done on location were unknown in Japan until after the war. Only a small number of monographs of questionable quality were available to dialectologists. The best dialectologists based their studies on surveys done somewhat narrowly limited to a small number of phenomena, or even based on their own in-depth knowledge of their native dialect. We were still far from having materials furnishing a large number of isoglosses which could have formed bundles running over great distances. On the other side of the globe, Professor Göran Hammarström, well known for his Auditory Phonetic Study on the dialects of Algarve (Portugal) and professor at Uppsala (Sweden) did me the honor of using the article that appeared in Orbis [Grootaers 1959c] as a text in his courses, as he amiably told me during the First Congress of General Dialectology (Leuven-Brussels, August 1960). A little later, a text by him on this subject — Inquéritos Linguisticos — appeared in the Revista de Portugal (Hammarström 1961), and here I give the translation of pages 16–17. On the subject of the problem of dialect boundaries W. A. Grootaers describes in the article “Origin …” [Grootaers 1959c; the English version of Sibata 1959, i.e., Chapter 4, this volume] an extremely important attempt to establish subjective dialect boundaries. Although he did draw boundaries on the map based on the opinions of the speakers, he concludes that “the dialect awareness of the average speaker has no linguistic basis” and consequently that a subjective boundary “is completely valueless as a starting point for the linguistic study of dialect units and dialect boundaries.” The observation of Grootaers implies that linguists are better off establishing with some certainty the “objective” dialect boundaries before comparing them to subjective boundaries. But this idea appears to me to be more of an illusion than a reality. In fact, to establish these objective boundaries, one resorts to criteria very often borrowed from

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES

119

the area of phonetics, the choice of which is very arbitrary. What can be the real value of studies establishing boundaries in this way? Do they count the number of dialects in a country or simply seek to rectify the boundaries established by another investigator? The method used by W. A. Grootaers appears to me more revolutionary than he himself believes. If one leaves aside the somewhat positivist concept of the linguist who seeks to establish dialect boundaries “objectively,” one could give the following fundamental definition: “dialect differences are the regional differences of a language as they are conceived by the speakers.” Regional differences are a conscious thing for speakers.3 Linguistic facts help a group of individuals to group themselves together and at the same time separate themselves from other groups. This is even one of the reasons which helps explain the existence of local varieties.4 As W. A. Grootaers has shown, if “subjective” boundaries do not coincide with “objective” boundaries, one can say: “too bad for the latter.” Even if one came to realize that the subjective boundaries change place without the objective boundaries moving, I can only repeat: too bad for the latter. In dialectology, the way to proceed which is the most proper to the subject of the study is to establish the traits of language upon which subjective dialect differences rest.5 One will do better only after that to speak of dialect differences, given the fact that the thing one wishes to express by the word “subjective” will have to be part of the very definition of dialect difference. There is no reason to despair if, for example, the same differences (objective) do not always play the same role from the point of view of the dialect speakers. This fact is analogous to that which is well known in phonology: different languages and dialects have phonological values that are determined in different ways according to their systems. The doubt we feel over our ability to express ourselves on the validity of the manner in which dialect boundaries are established must be extended to the establishment of each particular “isogloss,” each in itself a legitimate undertaking.

3.

Commentary about the Remarks of Tôjô and Hammarström

To begin, let us first notice that my article from 1959 in Orbis neglected to put the problem in its Japanese framework — one which would have emphasized the importance attached to the problem of dialect divisions; this element was therefore missing for Professor Hammarström. It is all the more interesting to note that Mr. Hammarström goes further than Professor Tôjô in justifying the use of subjective criteria to determine dialect units. Where Tôjô sees only a working hypothesis, Hammarström establishes a theoretical principle: since dialect differences constitute a human grouping and are born of them by a reciprocal interaction, it is legitimate to ask subjects to determine the line of dialect boundaries. I could add the following formulation: if one uses objective boundaries, their choice will be determined by the subjective choice of the investigator; it is better then to take the criterion, itself also subjective, of the choice of boundaries made by the speakers; here at least this “subjective” element plays a legitimate role, since it is a matter of the

120

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

judgment of those very people who use the language. Hammarström’s point of view helps advance the analysis of the problem. I now suggest going a little further, keeping in mind the following factors: A. There are a number of elements in all human language that escape the awareness of speakers; once a linguistic survey is carried out, it will reveal those elements and show the objective boundaries of which the subjects are unaware. Such elements, although unconscious, often influence linguistic mutations and are therefore a legitimate part of linguistic analysis. B. The choice of certain phonetic, grammatical and other criteria to define a dialect boundary is indeed often stained with subjectivism; but what will we say, if, thanks to a great number of surveys, the criteria amount to several hundreds, as is the case in the study of Grosse (1955) and of his colleague Protze (1957). One can then speak of sampling (as they say in statistics) that is complete enough to objectively reflect the totality of linguistic facts that constitute a dialect unit. C. This brings us to the following consideration: in my study in 1959, and in the remarks of Hammarström, the term “dialect boundary” is used in a way that perhaps hides a more important aspect of the problem: it is not so much a question of demarcating, subjectively or objectively, the dialects, but especially of determining dialect units; I mean that the studies of Grosse and Protze try to determine a center, a Sprachlandschaft whose characteristic is to combine at the same time a great number of linguistic elements possessed in common and a number of extra-linguistic elements. This last point is especially important in order to correct the impression of “positivism” that point B could have left. D. If we turn ourselves now toward the remarks of Professor Tôjô, we must note that he is a native of Tokyo, a city playing a centripetal role in Japan, parallel to the role of Paris in France; for the inhabitant of Tokyo, the subjective awareness of the division between eastern dialect (Tokyo and territory to the east of the Japanese Alps) and western dialect (Kyoto and territory to the west of the Alps) is very clear and living; it constitutes a fundamental element of every linguistic study.6 Now we have proven that the inhabitants of the island Kyushu, the southernmost of the four large Japanese islands, know nothing of this east-west division; they subjectively feel only the difference between “their” island and the rest of Japan. In the same way, speakers of Gallo-Roman dialects which have the items “mardi, mercredi …” [Tuesday, Wednesday …] are ignorant of the existence of dialects that have the items “dimars, dimyek …” except at the immediate boundary of the two varieties. In other words, to use the traditional east/west division of Japanese dialects in the guise of a working hypothesis is to expose oneself to the accusation of subjectivism in the choice of criteria. As Hammarström says so well, the function of criteria can differ, and in the case of the island of Kyushu, the point of view of the inhabitants indeed teaches us something about their linguistic mentality; but doubtless one cannot prohibit oneself from a little more analysis of this subjective element. It is indeed important to note that subjective differences are often non-existent when the speaker has not traveled; his/her limited

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES

121

human experience is reflected in a subjective linguistic judgment that doesn’t pass beyond a few kilometers from the native village. E. Again in the case of Tôjô, it is necessary to notice that Japanese dialectology since its beginnings (the “Wenker method,” adopted by Ueda, see Grootaers 1957: 346–49) and also during the career of Ueda’s best student Tôjô, has never attained the stage where linguistic maps could have been useful in great number for the interpretation of the past, present or the future of the language (linguistic factors) and even less for the interpretation of the past and present of the human society from which the language comes (extralinguistic factors). The problems of local surveys, maps, and interpretations, for example like those of Jaberg and Jud (1928–1940), are far from the thoughts of Tôjô. We now turn to a contribution to the study of the problem of the subjective boundaries of dialects.

4.

What the New Survey of Mr. Mase Teaches Us

At first an auditor of the dialectology seminar at the Metropolitan University of Tokyo, where the Sibata-Grootaers maps of subjective frontiers had been discussed (1959 [Grootaers 1959d]), then a collaborator on the survey of Itoigawa during its last period (1961 [the work culminating in Sibata and Grootaers 1988, 1990, and 1995]), and finally himself an initiator of a new dialectological survey, Mr. Mase was well placed to understand all the aspects of the problem when he decided not only to verify the results of the survey of subjective boundaries in a region of different character, but also to add certain refinements to it. I assume here that the reader of the present article has first attentively read the translation of the article of Mr. Mase that immediately precedes the present study [i.e., Mase 1964c, Grootaers’ French translation and summary of Mase 1964a and 1964b, Chapters 6 and 7, of this volume]. What will first strike the reader is the similarity existing between certain remarks by Mr. Hammarström and remarks made by Mr. Mase. Since there was no contact between these two researchers, it is appropriate to devote some very special attention to these points. The first problem can be formulated thus: Is it community life that gives birth to subjective dialect unity, or, on the other hand, did this community life first give birth to a linguistic differentiation which in turn is the basis of subjective dialect unity? The second problem: By asking subjects which linguistic elements separate them from their neighbors, one can try to seize upon one of the objective elements that form the basis of the subjective sentiment of linguistic differentiation. Mr. Mase, for this last point, resorted to precise questions, asking the subject to indicate concretely the dialectal difference that separates him from “foreign speech.” If he had had knowledge of this innovation, Mr. Hammarström would have heartily encouraged it, while adding that we mustn’t despair if the same objective differences don’t always play identical roles “in the awareness of the dialect speaker.” This will explain, therefore, why the subjects questioned by Mr. Mase did indeed indicate certain linguistic elements (sounds, accents, grammar) as the basis for a perceived difference, but from one corner

122

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

of the explored territory to another, it is often entirely different elements that are cited. I believe that here we have methodological progress in the examination of the problem that concerns us: it is extremely instructive to have a list of linguistic elements that play the role of catalyst in the speaker’s awareness of dialect differences. I may, however, add a consideration to which our two colleagues do not attach enough importance. As any investigator knows, and as Mr. Mase for his region and ourselves for the Itoigawa region have realized, each village, without exception, is separated from its neighbor by at least one isogloss, or sometimes by several. For a concrete example, see map 3 of Grootaers (1963: 367). We always find at least one dialect boundary that will coincide with the boundary of a school district or with the administrative boundary of a village, therefore with extra-linguistic elements. From this point of view, one can say that the coincidence of one single isogloss with any other cultural or subjective boundary does not mean much, but the fact remains that the subject questioned about an objective element which differentiates him from his neighbor must make a choice from among a great number of dialect differences which in fact exist, and, by his choice, he indicates which of these differences has penetrated all the way to the conscious level of his linguistic sense. Moreover, Mr. Mase showed that the territory of the villages of his region, vast because they cover the sparsely inhabited regions of the Japanese Alps, has not undergone an administrative change for centuries; it is therefore especially the school districts that form a community life, because they are the human dimension; this is an interesting contribution to the study of our problem. It is appropriate now to carefully examine the historical and geographical characteristics of each territory to be able to evaluate the role of extra-linguistic factors which influence the linguistic awareness of the inhabitants. We stated above that the statistical method that compiles isoglosses (as it is practiced by Protze and Grosse) is not necessarily mechanical or guilty of “positivism,” for which we occasionally reproach a certain American school. Indeed a certain number of linguistic phenomena will always escape the attention of the average subject and must be pointed out by the dialectologist. If these criteria are numerous, they escape the accusation of being nothing but a subjective choice (Hammarström). Here we could criticize Mr. Mase for having settled for too small a number of isogloss bundles to form objective boundaries. His choice can appear subjective, but the work of his survey is not yet advanced enough for a more complete synthesis, and it is by way of practical necessity that his number of choices is reduced. The two articles devoted to subjective boundaries (Sibata in Japanese [Chapter 4, this volume] and Grootaers [1959c] in English) end with an observation that neither Mase nor Hammarström seems to have brought forth. When it is a matter of subjective differences of the second degree (that is, those that make a dialect almost incomprehensible), it appeared to us that the subjective boundary splitting the territory into two regions seemed to coincide with a large bundle of isoglosses; these are even more important since they cut the main island of Japan in two. This bundle also coincides in our region of Itoigawa with the boundary of an ancient feudal fief (see Grootaers 1959c: 378). Mr. Mase, for his part, by tracing the second degree boundary in the middle of his

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES

123

territory, realizes that it follows the ancient feudal boundary; he thus brings an important confirmation to what we were saying in our text (Grootaers 1959c: 384) about feudal fiefs: “A social and economic unit that has lasted for such a long span of time cannot have been without influence on language evolution. The subjective boundary is therefore also a linguistic boundary; we certainly do not intend to spend our time here discussing which came first, the cultural unit or the linguistic one.” But to this we can now add an important observation. Indeed we had written in Grootaers (1959c) that subjective boundaries of the first degree (“one is conscious of some difference between dialects”) in the Itoigawa region have nothing in common with the objective isoglosses. This has just been confirmed in our study of the isoglosses of that region that we published in Grootaers (1963: 380); there we discovered that the subjective boundary splitting the Netchi valley into two parts, north and south, has no objective basis in the isoglosses. The subjective boundary is therefore quite influenced by the old administrative boundary, and it is legitimate to conclude: in the establishment of subjective and objective boundaries between dialects, priority must be given to the extralinguistic element, to the element that influences community life. Objective differences (linguistic elements) are normally a consequence of community life; they no doubt reinforce the latter, but in essence community life comes first.

5.

A Third Investigation into Subjective Boundaries

My colleague Mr. Sibata has just begun a new regional survey in the extreme north of the main island of Japan during the summer of 1963. His companions were three young researchers from the University of the Northeast (in the city of Sendai): Mr. Masonobu Kato, Mrs. Sadako Kato-Ooyama, and Mr. Eichiro Kawatomo. The left insert on Figure 8.1 gives the location on the general map of Japan of the four dialect surveys which have reserved a part of their questionnaires for the problem of subjective boundaries. Number 1 is the Itoigawa region (Sibata-Grootaers-TokugawaMase), number 2 the Nagano-Gifu region (Mase-Shimuzu-Kobayashi), number 3 the Shimo-Kita region (Sibata-Kato-Kawamoto), and finally number 4, Sado Island (Nomoto). The survey of the Shimo-Kita peninsula was pursued during the summer of 1964, but since the first part of the survey examined one out of two villages (sixty out of 115), the network of points is sufficiently dense to permit conclusions; we borrow the details that follow from the report made by Mr. Sibata to a conference held in May, 1964 in Tokyo. The questions asked attempted to determine two degrees of subjective dialect difference: First question: Does the speech in X (name of neighboring village) differ from yours? To where must one travel to notice a difference and up to which village is speech the same? Second question: To what point must one go so that the speech is so different that you don’t understand it?

124

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Figure 8.1. Subjective dialect boundaries (Shimo-kita Peninsula) (left insert: subjective dialect studies in Japan)

In order to begin a consideration of the second question, one must note that the territory explored is not near any ancient feudal fief’s border; this is noticed in the responses that reflect no perceived opposition in common toward a neighboring fief. On the contrary, the villages on the west coast of the peninsula, villages which incidentally have no overland communication with the rest of the territory, feel closer to the peninsula of Tsugaru (which is just across the bay to the west) than to the rest of the villages of their own peninsula. Certain very isolated villages in the mountains are known by all the other villages as having “very different speech.” Finally, when one moves southeast toward the pass where the peninsula is extremely narrowed, one discovers that the subjects are conscious of a great linguistic difference that begins approximately at the border of the canton. The responses to the first question (subjective difference of the first degree) are more interesting because they are easily adaptable to mapping. We will find therefore in Figure 8.1 the borders of villages that said they were “identical” to each other or were “very close in speech.” The dotted lines indicate the present administrative limits of the villages. Let us immediately note that the coincidence between administrative and

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES

125

subjective boundaries is quite frequent. Three-quarters of the civil boundaries coincide with a subjective boundary. But there are also a good number of subjective boundaries that bisect the territory of a village. In other words, the subjective areas are smaller than the areas of the villages, but the civil boundaries coincide with the exterior limits of the subjective areas. Thanks to the survey by Mr. Mase described in the preceding article [i.e., Mase 1964c], the attention of Mr. Sibata was attracted to the school districts; Figure 8.2 shows at the same time the administrative boundaries and the boundaries of the school districts, in other words, of the district inside of which the children of several hamlets go to the same primary school. It can be seen on the map that in only five cases do the subjective areas of Figure 8.1 coincide with a school district: this was indicated by a thick boundary on Figure 8.2. It appears that here, contrary to what is happening in the territory studied by Mr. Mase, the school districts are so reduced in size (because of communication difficulties) that they cover only one or two hamlets on the average. One thus often finds a subjective area which engulfs two school districts.

Figure 8.2. School districts (with administrative and selected subjective boundaries)

In different places in Figures 8.1 and 8.2, the reader will note the presence of a symbol in the form of a star with four points: it indicates the villages that form subjective

126

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

islets, which means that they declare themselves, each on their own, different from their neighbors, and that their neighbors agree, because none of the surrounding villages recognizes them as having identical speech. This type of islet had been discovered in our survey of Itoigawa (see the description of group 7, Grootaers 1959c: 366 [and Chapter 4, this volume]); these are communities that have almost no communication with the exterior. To be complete, it is necessary here to mention the study of Mr. Kikuo Nomoto, member of the National Language Research Institute (Tokyo); in his capacity as linguist he participated in a survey organized collectively by nine scholarly societies on the island of Sado (see no. 4, in the inset of Figure 8.1). The survey took place over two time periods, in 1960 and 1961, and the part of the report treating subjective dialect boundaries appeared in Nomoto 1963 [see Chapter 5, this volume]. After reading the study of Sibata-Grootaers (see above) Mr. Nomoto tried to determine from his subjects their degree of linguistic awareness in relation to neighboring villages. The territory chosen, by its very form, facilitated the survey: it is a string of twelve villages, running along eighteen kilometers of the northwest coast of the island. The first question was: “What village has speech identical to yours?” Responses divide the villages into four groups of three [see Figure 5.1, this volume]. The three subjective boundary lines correspond with (going from south to north): a. an administrative boundary dating from 1889; b. a mountain avoided by road traffic up until the use of busses; c. a mountain that still today prevents road traffic. This group of twelve villages also declares itself “different” from the north in relation to the villages at the island cape. At this point there had been an administrative boundary that dates from before 1889. Mr. Nomoto next compared fourteen isoglosses (six lexical and eight phonetic) obtained during the dialect survey of the subjective boundaries. This random choice of too few isoglosses does not tell us much; it simply confirms the fact that here also there is at least one isogloss between each village. But Mr. Nomoto’s third step is more interesting. After having noted the response of the subject for each of the sixteen preceding questions, he adds the question: “What does one say in the neighboring village for this word?” He was thus able to locate the places where the subjects believed the lexical or phonetic phenomena in question differed from their own. Mr. Nomoto here made a discovery about the subjective boundaries described above. Boundaries b. and c. disappeared in the responses to the last question; in other words the subjects who claimed in the beginning to have a different dialect from a given village (the first question) were incapable of giving one single concrete example of dialect difference among the sixteen linguistic phenomena. In the case of subjective boundary a., the subjects were able to indicate five concrete dialect differences, but one next sees that the following village, and the following, and again the following were also able to give the same number of dialect differences from their respective neighbors, even though no subjective boundary had been recorded at those points. Mr. Nomoto next tried to find some basis for this phenomenon in the local phonetics or in the degree to which respondents were educated (e.g., higher around subjective boundary a.), but the excessively reduced scale of the survey did not permit any solid

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES

127

conclusions. In short, this fourth survey leads us to a great deal of skepticism when considering the value of asking subjects not if they feel dialect differences from their neighbors but what these differences actually consist of. In conclusion we will note that each of the four surveys of subjective dialect areas encounters factors unknown to the other three. It is necessary, therefore, in this kind of study to keep in mind the characteristics of each region. On the other hand, the four surveys have numerous points in common: the way in which questions are asked, the way in which responses are categorized in order to group villages which have given identical responses, and the elements introduced for the comparison of different areas — feudal fiefs, old and new village limits, school districts. Let us here add an observation made by our third collaborator Tokugawa during the survey of Itoigawa. The questions asked by the investigators to determine the existence of subjective areas often used place names: “Does a linguistic difference separate you from this village?” “Starting with this village do you perceive a dialect difference?” Now one must notice that the first response given spontaneously by speakers often refers to geographical landmarks; they say “on the other side of the river” — “on this side of the mountain” — “starting with this promontory.” The names of ancient feudal fiefs are also used with great frequency. It was not until afterwards that they used the names of villages, doubtless under the influence of the question. It therefore seems that one can affirm that the spatial view of the rural informant, or at least the way in which this view is expressed, never resorts to linguistic indices, but rather reflects first and spontaneously a profound rooting in the countryside. There would therefore be room to refine interrogation methods; by trying different types of formulation, one could try to seize upon that which best reflects the local linguistic consciousness.

6.

Conclusion: Theoretical Bases and a Look at the Future

We wished to combine in this article new materials on the problem of subjective dialect boundaries with as loyal as possible a summary of an international discussion whose elements are sparse. I hope that the readers of Orbis will bring forth new elements that will help to move the study of this problem forward. To finish we wish to place the problem back in its theoretical framework, which for us is fundamental. Linguistic geography was first born in Germany (Wenker) in response to the needs of the neo-grammarian school which sought to understand the regularity of phonetic change. This school was at first amazed at the multiplicity of phonetic maps; then it brought these maps back into more general syntheses which culminate, as we have seen, in the works of Grosse and Protze, where five hundred maps are combined to outline two “Sprachlandschaft.” One can say that in all this the point of view of the regularity of phonetic changes has come out intact, although clothed in a good many correcting details. Linguistic geography was born again in France (Gilliéron) in reaction against the neogrammarian school, seeking to highlight the creativity and originality of dialect changes.

128

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

This French school very quickly directed itself toward the study of extra-linguistic elements which influenced the geographical distribution of dialect forms. Attempts at synthesis were more and more directed toward cultural and human geography. We roughly distinguish these two tendencies in order to simplify; because dialectologists very quickly engaged in study on both paths simultaneously, as is shown by the studies of Frings in the Rhenish country [e.g., Frings 1956 (1922), part I, 1–54], and of Jaberg and Jud in Italy [e.g., 1928–1940]; we also see the brilliant phonetico-historical syntheses of Kloeke on the origin of Dutch diphthongization and on the dialects at the beginnings of Afrikaans (1927, 1950). But in the meantime structural linguistics was born totally separately from linguistic geography. The structuralists did not carefully consider their ideas with regard to the atomization that language undergoes in studies of linguistic geography. In focusing on structures, they naturally highlighted the interior unity of linguistic systems. What then becomes of the problem that has occupied us in these pages? We could cite Heeroma: “Dialect division is based on one of two criteria: simple intuition or an arbitrary choice disguised by a statistical apparatus” (1961: 4). These two criteria summarize the point of view of Tôjô and Hammarström and that of Grosse and Protze (and Grootaers!). This allows us to introduce, in the guise of a perspective on the future, the works of two dialectologists who, independently, had the idea in 1960 of mapping the phonological structures of dialects. The first, Professor K. Heeroma from Groningen proposed a structural map presenting the system of long vowels in the dialects of the northeast of Holland [i.e., Heeroma 1961]. The second, Professor W. G. Moulton of Princeton devised five structural maps of short vowels in the dialects of German Switzerland (1960). The latter has continued his studies, which are promising enterprises in the renewal of the considerations of the problem of dialect division. He has just received a Guggenheim scholarship to pursue this work, and we can expect a synthesis of the new results.7 If these attempts at structural dialect geography succeed, they will provide us with a method of establishing dialect units that are at the same time fully “objective” and purely “linguistic.” We refer again to the new formula of Hammarström cited above: “Since dialect differences constitute a human grouping and are born of them by a reciprocal interaction, it is legitimate to ask subjects to determine the line of dialect boundaries.” It is quite necessary to notice that the three surveys described above allow the conclusion that it is the community life of human groups that plays the principal role in determining subjective dialect boundaries — community life, therefore, an extra-linguistic factor. Personally I am more inclined to attach the greatest importance to extra-linguistic factors in dialect geography because I deem it the only way to cover the entire territory. But if one desires an intrinsically linguistic criterion to determine dialect divisions, it seems that the new structural maps will lead us there.

THE DISCUSSION SURROUNDING SUBJECTIVE BOUNDARIES

129

Notes 1. Editor’s note: The two articles which Grootaers is referring to were translated and summarized by him as Mase (1964c); they are translated here from the Japanese originals, Mase (1964a) and (1964b), as Chapters 6 and 7. 2. We can judge the importance attached to this problem by noting that the publisher Tokyodô has just published a volume of 700 pages, in honor of the 80th birthday of Professor Tôjô, under the title Nihon no hôgenkukaku (Japanese Dialect Divisions), 1964, edited by the Dialect Society of Japan. 3. Compare on the other hand the differences that the linguist establishes for the different historical periods (therefore by diachronic means); in these cases, the differences are not consciously perceived by the folk speaker [Note 25 in the original]. 4. Insofar as today, because of technical progress, groups become in varying ways (culture, commerce, politics), parts of a more and more uniform world, local varieties have fewer and fewer functions to fulfill: they disappear [Note 26 in the original]. 5. This task is possible in the case of living dialects, since one can ask the necessary questions. As for vanished dialects, we will have to settle for “objective differences” since it is obvious that there is no longer anyone to interview. But even in this case, it will be possible to draw conclusions about subjective differences: in effect the latter can only exist if they are based on objective differences [Note 27 in the original]. 6. The reader who is not a Japanese studies expert can find a short description of these two dialects in the study by Sibata (1960). 7. Works by Moulton include the following: (1961a, 1961b, 1962, 1963a). For a fruitful reading on the use of precise phonetic notations for establishing structural maps, see the observations of Hammarström (1963), Moulton (1963b), and Kurath and McDavid (1961).

C 9 On the Value of Subjective Dialect Boundaries Antonius A. Weijnen Translated by Jennifer Dailey-O’Cain

At the symposium which was held by the Dutch Committee for Dialect Research on April 12th, 1944, I developed a method for the first time (by examining the answers to the first question of Questionnaire 8, which had been sent out by the committee in 1939) for using little arrows to classify dialects according to the perceptions of speakers. I was not aware then that Büld (1939) was also taking several steps in that direction at approximately the same time. I commented on the unsatisfactory results of Büld’s method in a monograph (Weijnen 1961). Later, I again used the method I developed — which for the sake of simplicity I would like to call the “little arrow method” — to study the perceptions of Limburgish (Weijnen 1966 [also Weijnen 1947]). Rensink (1955 [see Chapter 1, this volume]) also made use of this method. Briefly, the method is as follows: sites which informants perceive as being very similar are associated with each other by the use of little arrows. In this way groups of regions connected by chains are formed, and these stand out against the zones in which the connecting arrows are missing. These zones form dialect boundaries which are very much alive in the minds of ordinary speakers. Our Japanese colleague, W. A. Grootaers (1959c) [Sibata 1959, Chapter 4, this volume], is of a different opinion; he does not believe in the value of asking dialect speakers their perceptions of the geographical boundaries of dialects. I responded to this in the aforementioned journal [i.e., Weijnen 1961]. The reason for Grootaers’s rejection of this notion is not to be found in the diverse nature of this area of research. Grootaers, who in the far east was only indirectly informed about my method by reading the article by Rensink and was not aware of my original study at all, was forced to come to different conclusions than Rensink and I did as a result of an incorrect formulation of the question. He asked his informants whether they perceived stronger or weaker differences between their dialect and the neighboring dialects, which he indicated one by one. Of course there are small differences everywhere, in Japan as in the Netherlands. “I for my part was not surprised that each hamlet felt some difference with its neighbor,” Grootaers (1959c) writes. As he expected, his Japanese informants viewed every village border as an obvious

132

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

dialect border and perceived the old feudal borders as dialect borders for entire groups. Grootaers noted about this that neither the feudal borders nor the village borders corresponded to the linguistic reality of the dialect situation. However, as I have already stated, the formulation of his question was incorrect. One cannot ask ordinary people to say where faraway dialect borders exist; if that is done, errors are inevitable. And because differences within one’s immediate surroundings always do exist, the likelihood is very high that informants will answer the question of whether or not they perceive differences between their dialect and that of the neighboring village with an affirmative answer. This was the reason I, back in 1944, judged the kind of question formulation Grootaers used as misguided. Instead, one should ask informants where similar dialects are spoken.1 Recently Grootaers (1963, 1964 [Chapter 8, this volume]) again expressed his views against the subjective method. Even in this article he decidedly rejects the connection between subjective and objective borders as a result of the Japanese studies’ findings. This is made all the more noticeable by the fact that his colleague Mase (1964c [Chapters 6 and 7, this volume]), who also carried out a study in Japan, was able to establish a connection. In fact, Mase writes “that the objective dialect border coincides almost perfectly with the subjective impression of dialectal differences of which the subjects are aware.” The most likely reason why Mase found the connection and Grootaers did not, now becomes completely clear. Mase’s question was: “in which hamlet do people speak like you speak here?” It seems Mase inquired about the similarities, precisely in the way I did. Grootaers does not seem to be completely aware of the fact that this is a different method from his own. In Grootaers (1964: 381) he specifically writes about “the firstdegree subjective boundaries” as if they reflected a “response to the question”: “where do people speak like you do?” Yet he actually never asked this question, but instead inquired about possible differences. One must arrive at this conclusion, since he wrote in Grootaers (1959c: 356): “Does the language spoken in this hamlet differ in any way from that of the neighboring hamlet?” This question is always answered with a “yes.” Therefore his results were negative, while Mase, who used a better method, arrived at positive results.2 Of course it is still the case that the little arrow method is not sufficient in one way: it does not provide reliable data about the relationship to other dialect groups. Therefore, it is always to be incorporated into an objective method.

Notes 1. Editor’s note: Since Weijnen was working from French and English translations and summaries of the work in Japan, it will perhaps be important to confirm his characterization of these questions. Sibata (1959, Chapter 4) reports that the question under consideration in the Itoigawa research was “Is the speech here [_______] from the speech of the neighboring hamlet?” (with the following choices in order of difference) (1) not different, (2) a little different, (3) quite different (4) mostly incomprehensible.” Grootaers English version of this research (1959c: 356), which Weijnen cites, however, states that the following question was asked: “Does the language spoken in this hamlet differ in any way from that of the neighboring hamlet

ON THE VALUE OF SUBJECTIVE DIALECT BOUNDARIES

133

… 1. no difference, 2. a slight difference, 3. a noticeable difference, 4. almost not intelligible.” Mase (1964a [Chapter 6, this volume]) asked, on the other hand, “Where do people speak the same language as here?” 2. Editors note: Grootaers’s characterization of the question (“Does the language … differ in any way …”) appears to be an exaggeration of the question actually put to the respondents. Nevertheless, Weijnen appears to be correct in noting the degree to which the Sibata-Grootaers work in Itoigawa emphasized “little differences” for the respondents (seen in Sibata’s representation of the question) and the degree to which similarity was highlighted in his (and Mase’s) work.

C 10 Dialects and the Subjective Judgments of Speakers Remarks on Controversial Methods A. C. M. Goeman Translated by Betsy E. Evans

At the first Conference of Romanists in 1890 at Montpellier, Ch. de Tourtoulon took a position on the debate over the question of the existence of dialects. His hypothesis was that people from the lower classes can distinctly classify those vernaculars which they understand and that they are capable of making an “instinctive classification of such vernaculars.” Starting with this hypothesis, his research led him to the conclusion that “for the illiterate as well as for the literate … dialects exist.” De Tourtoulon obtained this result by soliciting from his respondents subjective judgments of the most striking characteristics of the dialects that they understood. De Tourtoulon should be credited for recognizing very early the role that subjective speaker judgments can play. As we will see in the following, this consideration of speaker judgments in determining the real existence of dialects, as justified as it was, complicated the basic problem of the relationship between subjective (global) speaker judgments and dialectological facts, that is, the linguistic features of these dialects. I want to illustrate this point starting with the application of de Tourtoulon’s theories to the dialectology of Germanic languages, more specifically, those of the Netherlands. The problem is all the more important to me because recently Mattheier (1985) has advocated dividing dialectology into two branches: one being that of dialectologists and the other being that of speakers. Certainly there is a relationship between these two domains, but we shouldn’t lose sight of this fundamental distinction.

1.

The Division of Dialects

Once the existence of dialects was recognized, de Tourtoulon’s program in the Netherlands became specialized and dissected. It became specialized through the use of subjective judgments to obtain a division of the dialects of the Netherlands. Speakers were asked for a subjective, global judgment concerning the degree of resemblance or similarity of the

136

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

dialects around the speaker’s home area.1 The dissection arose because the results of the judgments were separated — a subjective division of dialects — by a division based on “striking” dialectal features. Weijnen (1946) called the first of these divisions “the subjective awareness of cohesion between dialects” and the other “the objective division of dialectal features.”2 Before Weijnen, Büld (1939) made a similar distinction. He called the first “Sprachgefühl” (feeling of language) and the second “Sprachwirklichkeit” (reality of language). Büld’s research was conducted in northern Westphalia, but he did not seek judgments of general dialect similarity. Instead, he asked speakers for pejorative expressions used to characterize the striking and caricaturistic features of the dialects of their neighbors. Unlike the questionnaire used by Weijnen, Büld asked respondents about dissimilarity rather than similarity. Weijnen asserted that Büld’s procedure which relied on differences gave satisfactory results only for peripheral regions, and, even worse, that Büld made a totally arbitrary choice among the dialect features in these folk expressions (Weijnen 1961). Weijnen himself used Büld’s methods for his work in the Netherlands (1961) and concluded, with regard to the methodology, that judgments of dissimilitude did not lead to the desired results, that is, a division of dialects that corresponds to linguistic facts. Another conclusion that undermines the utility of this method relates to the dialectal facts that are contained in the expressions. In many of the cases there are features that no longer exist in that form in the dialect under consideration because they have changed over time. Therefore, regarding the methodology, the general conclusion of Weijnen is that only judgments of similarity give accurate results. The method that Weijnen (1946, 1947) used to create the maps of dialectal divisions of North Brabant and Limburg was the following: if A considers the speech of B similar to his, an arrow is placed pointing from A to B; this arrow will have two heads if B expresses the same judgment about A [see Figures 0.2, 0.3, and 3.1, this volume]. Thus chains and clusters of localities and groups of speakers are derived: the dialects and the empty spaces between the clusters form the boundaries according to the speakers’ awareness. Weijnen claimed that the correspondence between the subjective division and the objective division was established for the province North Brabant because there was a general parallel with the boundaries formed by the isogloss bundles. The first to criticize the method was the Limburger Peeters (1952). While admitting that the parallelism claimed for the North Brabant map gave the method some viability, he noticed that the Limburg map lacked the detail and validity of that for the Brabant area. It must be said regarding this subject that, at that time, the objective dialectal division of Limburg was not yet as well established as that of the Brabant and that Peeters gave not even a single concrete example of the divergences. At any rate, this objection should be kept in mind, and even Weijnen himself admitted that there were obvious sources of mistakes. He tried to correct them with maps of parallel isoglosses. Peeters also had objections of a more methodological nature. It was his opinion that there were psychological elements in these subjective judgments that were difficult to control. What is expressed by the speakers is not so objective because the concepts of cohesiveness and similarity are very complicated and above all rely more on feelings than on scientific reasoning. Peeters’s remarks on the non-rationality and non-objectivity of the

DIALECTS AND THE SUBJECTIVE JUDGMENTS OF SPEAKERS

137

data are certainly true, but they have a sort of tautological truth: the reactions were solicited because of their instinctive, naive global nature. They were presented honestly, reflecting the subjective knowledge of the speakers. Peeters’s remark on the non-objectivity of subjective judgments and the objectivity of dialect features, therefore, is not pertinent. The pertinent question is that of correlation. That presupposes, between the two domains, a more marked distinction than anyone has wanted to admit to up to now. The use of subjective judgments to establish the existence of dialects, as in de Tourtoulon, and to arrive at a division of dialects, as in Büld and Weijnen, risks considering the data only with regard to these two goals. Weijnen himself falls into the trap when he considers an equivalent relationship between these two domains desirable, although he perceived the distinction between them.

2.

Other Attempts

Weijnen was not the last to use judgments of similarity between dialects for the Netherlands. Rensink (1955) completed a map of the Netherlands in its entirety [see Figure 1.1, this volume]. Only the results indicating the dialect regions recognized by the speakers were published; the manuscript map held in the P. J. Meertens Institute archives was not. We can only guess the reasons why. I have the impression that the author was uncomfortable with some of the differences between the maps of dialect divisions and linguistic facts. Weijnen (1946) had already stated that in Brabant the speakers sometimes made a subdivision that linguistic facts did not confirm. At the end of the 1960s, Daan (1970) made a new map of dialect divisions based on the same data [see Figure 2.1, this volume]. She added the isophones and isomorphemes most characteristic of the significant borders between dialects. But this map had something different: it classified the resulting dialect regions on the basis of their relationship to the standard language. The darker the color, the more the dialect differs from the standard. It is possible to map this divergence because in the Netherlands the standard language developed from vernaculars in the province of South Holland in the seventeenth and eighteenth centuries. In an article dated 1963 [i.e., Daan 1963], the writer A. C. W. Staring was quoted as saying around 1800 that “Dutch is our Parisian dialect, the dialect of authors who want to be read and understood all over the republic.” Nowadays the dialects from the center of the province are still very close to the standard spoken language. A map of the provinces of South Holland and Utrecht was published in Goeman (1984) and analysis of the dialect continuum in the urban agglomerate that was assumed to be the “true” Dutch dialect was added. The responses were analyzed as structures of preference, seen as sociograms, because, it seemed to me, they had a variability of their own and because the question the respondents were asked does not give direct access to the complexity of dialect features. Thus it was necessary to take into account the socialpsychological properties of these data and exploit the property of transitivity contained in these judgments.

138

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Also in 1984, Kremer published a similar map of neighboring regions in the east of the Netherlands and of Westphalia in Germany [see Figure 3.1, this volume]. It dealt with data that he collected toward the end of the 1970s. In the interpretation of the map, he highlighted two principal results: a. the border of the two countries is transformed into a dialect boundary although these dialects form a continuum, with regard to their structure and their historical development, from one side of the border to the other. This border developed as a dialectal boundary as a result of the transfer of elements of the two standards — Dutch and High German. In a previous study (1979), Kremer showed this process of languages in contact at the lexical level of these dialects by comparing heteronyms of approximately one hundred words gathered during the pre-war period and again in the 1970s. The number of heteronyms had diminished in the Netherlands much more than in Germany. Under the influence of standard language, “regiolects” developed instead of dialects, although they continued to have a dialectal base; b. thus the differentiation in the subjective judgments is much less clear in the case of Dutch than in German Westphalia. Contrary to the results of Weijnen, who sometimes found overdifferentiation, Kremer claimed an underdifferentiation in relation to the division on the basis of known linguistic features. He concluded that the speakers were already ahead of dialectologists, who still scarcely understood the whys and hows, especially beyond the domain of lexicon.3 Kremer’s explanation, as promising as it might seem, nevertheless was at odds with the facts of development: the map of the 1939 data [i.e., Figures 1.1 and 2.1, this volume], which is also pre-war, does not show a differentiation for this eastern part of the country either.

What should be remembered about Kremer’s conclusions is that the only thing that seems to count for the speaker is the ideology of sharing the Twente dialect and being from Twente. This global ideology can also be explained by the relationship with the standard language, not, however, in the framework of linguistic facts, but by a mechanism that forces us to consider social psychology. The region of Twente has a very negative attitude with regard to anything Dutch, and thus towards the standard language, just as Daan showed (1963). If this negative attitude already existed in the period 1830–1840, its flame has been well fanned since. The eastern part of the country4 experienced uprisings because of the mobilization against the secession of Belgium. The central government severely punished the entire population of the region by imposing a prolonged billeting of troops there.

3.

An Overlooked Forerunner

The 1939 survey was not the first to interview speakers about the similarity of dialects. The question had already been posed in an enormous survey done in 1886 by P. Willems. He was a native of Maastricht, a city that had just been reunited with the Netherlands after a long occupation by the Belgians during the war of secession. This situation was not the result of a choice by the population but a decision made under the pressure of the great

DIALECTS AND THE SUBJECTIVE JUDGMENTS OF SPEAKERS

139

powers. Maastricht was, therefore, a city where certain social strata seemed very francophile. The press was still francophone in 1857, the year that young Willems took the grammar school exam, and the report in the Courier de la Meuse about the prizes that were awarded was in French.5 Willems studied classical philology at Louvain, and also studied at Leyden, Paris, and Berlin. He opted for Belgian nationality which he was able to claim, having been born in previously Belgian territory. He was chair of Latin classics and Latin literature, became a Flemish militant, and was co-founder and first president of the Flemish Royal Academy. Beginning in 1866, he also served as chair of the History of Dutch Literature for four years, and, in 1890, he taught courses in the history of the Dutch language. In 1886 he began a large study of more than 4,000 items in Dutch, French, and German because he wanted to study the Low Franconian dialects to their full extent from Dunkirk to the Rhine, including the Rhineland. He collected 347 questionnaires from 337 different localities amounting to 19,060 pages. He foresaw a comparative phonology and morphology of the Low Franconian dialects of Belgium, from the south of the Netherlands to the Rhineland. The preparatory work was, for the most part, completed, but the core work was never done because Willems died in 1898. The data and the tables are now in the archives of the Royal Academy of Dutch Language and Literature located in Ghent. Willems’s son bequeathed them, and the linguist Van Ginneken of the University of Nijmegen received them on loan. Van Ginneken, in his work, established a map of the dialectal division of the Netherlands and Dutch-speaking Belgium and used the Willems data for the southern part of the map. He does not specify which parts of the data he used, but it no doubt was the judgments of dialectal similarity because the dialect boundaries in Belgium according to Van Ginneken differ from those of Daan (1970) even though they changed very little in the intervening period.6 Above all, the boundaries drawn by Van Ginneken between east and west Flemish on the one hand and east Flemish and Brabant on the other conform more to Willems’s subjective data than to the boundaries of Daan’s maps. The revisions in the second printing of this book concerning the Meetjesland and the Belgian Brabant regions move the dialect boundaries eastward, over the empty regions left around the arrow clusters. In order to clarify the history of the Dutch-speaking dialect division, I reproduce here the map with arrows [see Figure 10.1, this volume]7 from data that is one hundred years old. It was devised by a scholar from a francophone environment who turned Flemish, and who was also a reputed classical philologist and became a dialectologist of a modern language.

4.

The Weijnen-Grootaers Controversy

Peeters’s remarks have already touched on questions of methodology. In the more general framework of materialist philosophy, Schirmunski (1962: 145) denied the value of any subjective-psychological treatment of questions concerning the formation of linguistic communities. He feared that such a treatment (the one recommended by Mitzka), if it had to replace one based on social and economic factors (which constituted the core data according to Wrede and Frings) would result in an inauspicious idealism that amounts to

140

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

Same dialect Slightly different

Very different Completely different

Figure 10.1. Perception of similarity among dialects (enlarged portion of Goeman 1989, page 543 [from Willems 1886]; compare figures 0.2 and 0.3)

the use of an unrestricted set of explanatory factors for the host of dialectal phenomena. Thus any socio-historical explanation of boundaries and the formation of dialect communities becomes impossible. Schirmunski simplified things a little bit. It is the idealism of Mitzka that he is attacking, but it is possible to take into account psychological and sociohistorical factors. Weijnen had already explained one aspect of the subjective boundaries found in North Brabant and in Limburg with economic factors, or better yet, he recognized the correlation of the two factors and envisioned the possibility that it could be produced by other social factors of long duration. Goossens (1970) criticized the maps by Daan, Rensink, and Weijnen on secondary points. All three claimed that they tried to correct their maps when their data seemed to contradict the objective division according to dialect features. Goossens is of the opinion that they betray their own method and that they are contradicting themselves. They are happy when there are similarities and worried when there are differences. But, the differences provide something new that, in general, results in deeper thinking about scientific problems. Correction hinders progress. Goossens has, in my opinion, best understood the specific nature of subjective judgments (his note 25); they reveal,

DIALECTS AND THE SUBJECTIVE JUDGMENTS OF SPEAKERS

141

according to him, something about linguistic psychology. He recognized that the data can vary for neighboring localities and also for speakers of the same locality, but that did not lead him to condemn this data; he simply had reservations about a direct correspondence with linguistic features. The most consistent criticism came from Grootaers (1959c [see Chapter 4, this volume], 1963, and 1964 [see Chapter 8, this volume]). He gave a report of the discussion in Japan on the problems of [subjective dialect] division. He applied the method to Japanese dialects, asking in every case for judgments about differences. He asked informants to choose among four degrees of dissimilarity, which was an innovation. The informants only used the highest degrees. He analyzed the judgments of difference between the dialects and clarified aspects of the bearing of these judgments dictated by the geographic situation. Geography has a strong influence on human communication. Nevertheless it appeared that administrative divisions are even more important; in effect the smallest unit was a village but only because the villages were considered as separate dialects. Grootaers concluded that the communal life of a village engenders a feeling of cohesion which is subjectively considered to have linguistic repercussions leading to differences among villages. In reality the objective dialect borders do not correspond to the subjective borders. The general conclusion is that the awareness of the naive speaker has no linguistic value. It is shaped in communal life and can be very transient. The opposition between a region and its neighboring village is psychological and linked to tradition. This opposition has no value for the linguistic study of regions and dialect borders. Weijnen responded (1961, 1968 [Chapter 9, this volume]) that he had emphasized since 1944 that the question of differences was the wrong one. Questions about similarity should be asked because everyone is in a position to cite some small difference which will surely produce a foreseeable result: every region researched will remain isolated from the others, and bigger regions will not emerge. Weijnen illustrated his point with his own experiences in North Brabant and also with the results of Mase (1964c), who used the method again but this time on the basis of judgments of similarity. Mase had to conclude that his subjective division of land corresponded almost perfectly with the objective division based on dialect features [see Chapters 5 and 6, this volume]. Weijnen did not add that Mase supposed that the difference between his results and those of Grootaers came from a total difference between the two regions concerned, a difference in the history and structure of the villages. Mase left as understood that the geographic structure of a region is a decisive factor. But Grootaers (1964 [Chapter 8, this volume]) is more explicit in noting that Mase’s sparsely populated region of the Japanese Alps had not experienced any administrative changes for centuries, and it is not surprising that here and there the school districts coincide with the subjective and the objective dialect regions at the same time. According to Grootaers, school districts also form a community. As for the priority of culture or linguistics, Grootaers finds it legitimate to give priority to extralinguistic factors, the factors that influence the community. Linguistic differences are a consequence of this community. What is essential then is the community. If the reasoning of Grootaers is pursued, it has to be concluded that among the influences, namely extralinguistic factors, community, and dialect differences, subjective

142

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE”

awareness has to be considered along with dialect differences because the subjective regions were also formed by the community. Thus it seemed that there was an impasse, all the more so because Grootaers claimed that the linguistic differences can reinforce the community. It would be interesting to know why subjective awareness does not do it on its own. Weijnen (1968 [Chapter 9, this volume]) is happy to point out the positive results for the flat country known as the Brabant and the region of the Japanese Alps. Contrary to Weijnen, Grootaers seems to imply that similarity and dissimilarity are almost the same thing. It is curious to find that they are both right. In theory the choice between similarity and dissimilarity has no importance. A measure of dissimilarity can be transformed mathematically into a coefficient of similarity by inversion, for example by subtraction from the maximum achieved or the theoretical maximum value (Fox 1982). The inverse is as valid for qualitative data as it is for quantitative data. But there are some problems in actual practice: there are many different measures, all with their own pros and cons, and this is a theoretical problem. With regard to the data, the problem is the following: can we consider the data so complete that we can fix their maximum values? The problem of the omnipresence of differences posed by Weijnen is typical for similarities among aggregates where the difference of one trait is considered as distinctive as a difference in all the traits. There are satisfactory measures for such cases (Fox 1982). The branch of social psychology that deals with linguistic attitudes8 has been restricted to searching for judgments about the personality of a speaker, but the relationship between such judgments and the specific linguistic traits that make up the speech is scarcely considered to this day. These judgments, from a linguistic point of view, are as global as those of similarity. Therefore, as in linguistic geography, the separation between subjective judgments and objective linguistic data seems to be complete. In any case, it must be recognized that the problem of the relationship between subjective judgment and objective dialectal phenomenon came up very early in dialectology.

5.

Other Interpretations

These subjective judgments can be used for taxonomic purposes as has always been the case. But we can go farther than that when we understand two characteristics of these data and their relationship to the standard language, provided that a standard language can be geographically localized. As we have seen, that is definitely the case with Dutch.9 First there is the orientation of similarity judgment and second the average orientation and dispersion. One could imagine a dialect that orients itself in another direction than that in which the standard language is situated and that the speakers of this language would behave differently. Then one could draw the shortest path between a given dialect and the position of the standard language. One could calculate a value for the estimated connections; for example, the double arrow would be given .05; one arrow 1.0; dialect C passively chosen by B and A in the middle of B and A would get 2; no connection between B and A would get 3. These two characteristics take into account certain aspects of the perception of the

DIALECTS AND THE SUBJECTIVE JUDGMENTS OF SPEAKERS

143

distance of dialects from the standard language. These measures of perception can be used as variables alongside geographical position and real distance in a causal model. Such a model would allow us to calculate perceptual influences on linguistic features which originate in the behavior of speakers. It appears that the perceived distance from the standard language is one of the causes of the process of apocope of word-final /t/ and the shortening of vowels before consonant clusters. The more distant one perceives his or her Dutch from the standard language, the more occurrences there will be of apocope of /t/ and vowel shortening. The role of mean distance in relation to the center of standard language is potentially a factor in the apocope of /t/.10 Recently Preston (1986) conducted research on subjective judgments of speakers and the perceived relationship with the standard language in the United States. His goal was to arrive at “mental” maps of speakers, and he provides examples of speakers from different regions of the country. He was openly inspired by Rensink (1955), and it seems that he was not aware of other endeavors. Dear Tuaillon,11 I present to you these observations, that took us from the flat lands to the Japanese Alps, from the French language to other languages, from taxonomies to speaker behavior, and from denied objectivity to a sought for subjectivity of subjective judgments because of their importance to methodology. This discourse regarding methodology has already lasted a century, and it is thanks to the debates that we sometimes arrive at the truth. These debates have sparked ingenious and fruitful reflection. That is not the least of things brought about by this aspect of the dialectology of speakers. The discourse on methodology will never cease.

Notes 1. “Do some dialects in your region more or less resemble yours? Which ones?” [See Chapters 1 and 2, this volume.] 2. The questionnaire of the Amsterdam Center for Dialectology dates from 1939; Weijnen used one part of the responses, notably those of the North Brabant and Limburg. [See Chapters 1 and 2, this volume.] 3. It must be said, nevertheless, that the relationship of the standard language to the dialects in the Netherlands is more direct than in Westphalia because there the dialects are part of the Low German patois and the relationship with High German is therefore less direct. 4. Personal communication with R. Rentenaar. 5. Cf. the biography by A. de Ceuleneer (1909) from which I took what follows. 6. There were not any data for the Belgian part from the (1939) data; Daan got her information from Belgian colleagues who based their data on dialect features. [Author’s note: Now my presumption has been confirmed by Weijnen (1989), who demonstrated that all of Van Ginneken’s dialectological work drew heavily on Willems’s survey.] 7. Editor’s note: I have not reproduced the entire map or even the enlarged portion of the western section of it from the original. The interested reader may find these there, but the “little arrows” on even the enlarged portion of the original map would have been impossible to see in any detail on a page the size of those in this volume. I have, therefore, enlarged that portion of Goeman’s representation of Willems’s map which corresponds to the enlarged portion of the western North Brabant shown in Figures 0.2 and 0.3. One may note, for example, in comparing these two figures, that Willemstad, an isolate in Weijnen’s map, is

144

THE JAPANESE CONTROVERSY: “SUBJECTIVE” AND “OBJECTIVE” indicated as the “same” by respondents from two nearby areas in Goeman’s representation of Willems’s study. There is sufficient detail in the portion enlarged here to show how Goeman has reworked Willems’s data in a “little arrows” representation.

8. One of the first was the group of studies done by Lambert. The “matched guise” technique was used: different accents are performed by the same man/woman and judged by the informant, cf. Lambert et al. (1960). Resynthesis, as in phonology, of dialect features seems to be a more direct method and certainly easier to manipulate experimentally. 9. In Belgium the Brabant dialect seems to constitute a sort of standard. 10. For these two results, cf. Goeman (1987). 11. Editor’s note: This paper appeared in a festschrift for Gaston Tuaillon.

P III Images, Perceptions and Attitudes

C 11 Classification of Dialects by Image English and Japanese Fumio Inoue

1.

Introduction

In this paper non-linguists’ images of various dialects and accents will be quantificationally treated. This study is based on Japanese and English data and has three purposes. a. It aims at opening up a new field of study which has been ignored, the field of “perceptual dialectology”; b. It is an attempt at interlingual dialectology, which can be characterized as a contrastive study of dialects across languages, the ultimate aim being a universal theory; c. It presents a model showing the usefulness of multivariate analysis in dialectological studies. 1.1 Definition of a Dialect Image The topic of this study, dialect image, can be defined as the socio-psychological image of a (geographical or social) dialect. The study of dialect images can be approached through a quantitative method which is similar to the Semantic Differential Method (Osgood, Suci, and Tannenbaum 1957). 1.2 Theoretical and Social Background Perceptual dialectology is a branch of social dialectology. As the study of dialect images is concerned with practical (social) problems, this study is also part of applied dialectology. In this paper we will show the usefulness of the perceptual study of dialects through concrete examples. A dialect is not only an abstract entity for intellectual communication; it has also an emotional aspect, and it should be treated as a social phenomenon. Social discrimination

148

IMAGES, PERCEPTIONS, AND ATTITUDES

because of the dialect one speaks is sometimes reported. Dialect (inferiority) complex is a term coined by T. Sibata (1958c). Several examples are given in his book, e.g., cases of people becoming neurotic because of their accents, and even an extreme case of a person committing suicide. These kinds of phenomena are not limited to Japan; similar cases have been reported in England (Crystal 1988; Honey 1989). Dialect images have been studied in Japan as a source of this kind of dialect discrimination. To be impartial, it should be pointed out that the situation of dialects in Japan today is changing owing to the rapid standardization of language. Changes in the sociopsychological status of dialects can be observed. One such change is the appearance of new dialect forms (Inoue 1986); the other is the disappearance of a dialect (inferiority) complex. Dialect realism in the form of the use of dialect in TV dramas seems to have accelerated the use of dialects. Competitions of performances in dialect held every year in a small town in northern Japan are a further example of this. 1.3 Multivariate Analysis A multivariate analysis which will be mentioned frequently in this report is Hayashi’s quantificational theory type 3 (hereafter referred to as Hayashi 3). This method is applied to non-numerical (categorical) data and analyzes the data just like factor analysis. The method is also called quantification for grouping, or a method for gathering similarities (Hayashi 1954; Inoue 1986). 2.

Previous Studies of Japanese Dialect Images

2.1 The Study of Dialect Images in Japan Before moving to an analysis of English dialect images, it may be useful to review the studies undertaken in Japan because in the present study we have employed a similar method to study English in Great Britain. Precursory studies were conducted for three main dialects at high schools in three corresponding areas using seventeen word-pairs. Three Japanese dialects were clearly differentiated, and the seventeen word-pairs were also classified into three groups through the application of multivariate analyses. In order to ascertain that the selected evaluative words were appropriate and that this method is applicable to other dialects of Japan, another study was also carried out. University students from various parts of Japan were asked whether two hundred evaluative words were applicable to their own accents. By applying multivariate analyses, sixteen representative evaluative words were selected, which served to classify Japanese dialects according to their image. The data of the university students was then utilized for a division of Japan by dialect image. The result of the application of Hayashi 3 to the data can be recapitulated as follows: Japanese dialect images were shown to be composed of two dimensions. The evaluative words can be divided into four groups by crossing the two dimensions. Meanings of the evaluative words suggest that the horizontal (first) axis is related to

CLASSIFICATION OF DIALECTS BY IMAGE

149

intellectual (mental) judgment, and that the vertical (second) axis is related to emotional reaction. Socio-psychological studies also show that the composition of these two dimensions is almost universally found in studies related to images. Figure 11.1 shows a dialect classification by image (Inoue 1988, 1989). In applying Hayashi 3, values can also be calculated for the informants. The average values according to the native prefectures of the informants suggest that at least four types of dialect image can be distinguished in Japan. The division of dialects into East/West seems essential, and the sub-division into North and South is also important.

Figure 11.1. Dialect classification by image using Hayashi 3 — Japan

This result of dialect division by image is plausible and persuasive because it matches quite well both with actual dialect distribution and with people’s common sense or intuition.

150

IMAGES, PERCEPTIONS, AND ATTITUDES

The borderline between Eastern and Western Japan can be explained by dialect boundaries. Linguistic evidence of the appropriateness of dialect images and especially of dividing Japan into East and West can be shown in Figure 11.2. Figure 11.2 shows a dialect classification which is based on cluster analysis. The application of factor analysis also presented similar results, and application of Hayashi 3 to grammatical and lexical data also provided similar results (Inoue and Kasai 1989). 2.2 Evaluation of the Methodology The above study of dialect images brought about very clear results. The applied methodology has both merits and demerits. The drawback is that the image investigated is not

Figure 11.2. Dialect classification using cluster analysis of standard Japanese forms — Japan

CLASSIFICATION OF DIALECTS BY IMAGE

151

based on linguistic data, but on a kind of common-sense knowledge, stereotypes, superstitions and traditions. The method only clarifies the image of the word naming the dialect; it does not directly treat the image of spoken dialects themselves. In order to observe people’s reactions when actually listening to the dialects, a comparison with studies based on recorded speech is necessary. Several pretests have been carried out. But there are also problems involved with recorded speech. It is difficult to obtain exactly comparable recorded speech, and the influence of the speakers’ personality is often inevitable. These are problems which arise in the studies of mutual intelligibility or matched guise tests (Giles 1971; Lambert, Giles, and Picard 1975; Chambers and Trudgill 1980). Our method of dialect image is free from this kind of influence.

3.

Methods for English Dialect Images

3.1 Starting Point: Difference from Japan The success of the analysis of Japanese dialect images inspired us to apply the same methodology to another language to see whether the basic structure of dialect images is similar. The application to another language would verify the validity of our methodology and would open up a new field of interlingual dialectology. Anecdotal descriptions of English dialect image are found in several books (Trudgill 1975, 1983; Shuy and Fasold 1973; Crystal 1988; and Honey 1989). The pattern of English dialect images seems a little different from Japan’s. At least the low status of urban dialects in England pointed out in the above books is a phenomenon we do not find in Japan. This indicates that different mechanisms may be at work in England. Surface phenomena certainly show differences but an investigation may show that the deeper mechanism is the same. But the information on English dialect images is fragmentary and scarce. Thus it seems necessary and worthwhile to investigate English dialect images. 3.2 Methods of Research The actual research was planned while the author was staying at Essex University in England in 1989. A revised English version was made with the adoption of new evaluative words which were found to describe dialects and accents in England. The collection of data was carried out by several university staff members at various places in Great Britain in 1989, and in Australia in 1990. 3.3 Procedure of Analysis First, in order to grasp the overall pattern of the answers, Hayashi 3 was applied to the whole data from both the British and the Australian students. The interpretation of the results was difficult because the resulting pattern was unclear. The same method was later applied to British and Australian students separately. A very clear pattern then appeared,

152

IMAGES, PERCEPTIONS, AND ATTITUDES

especially for British students. In this paper, the results of the British students will mainly be discussed. The Australian data showed a pattern which is a little different from that of Great Britain. The results will be briefly mentioned later.

4.

Results

4.1 Results of Hayashi 3 The application of Hayashi 3 offers two kinds of results: one for evaluative words, another for the dialects. 4.1.1 Classification of Evaluative Words Figure 11.3 renders the evaluative words into four clear groups by means of the first and the second axes. (The function of the third axis was found similar to the first axis.) The meanings of the words on the right-hand side are related to standardness. But interestingly enough, they do not always have positive nuances; for example, “snobbish” is included in this group. Meanings of the words on the left-hand side are related to accentedness; they are mostly adjectives with negative meanings, except one positive evaluation “animating.” The two words in the lower central area show urbanity, while those in the upper central area are related to pastoral (rural) meanings. Thus, the first axis (right-left direction) is related to a class-related distinction, and the second axis (upper-lower direction) is related to urbanization. Application of Hayashi 3 was thus successful in that the inherent structure of the data (answers) was made clear. The pattern shown in Figure 11.3 does not correspond with the Japanese pattern consisting of intellectual and emotional factors as noted above. Moreover, plus and minus (positive and negative) answers were not distinguished in the British analysis. This surface difference from Japanese data may be explained by a deeper mechanism working on the dialect image. 4.1.2 Classification of Dialects Evaluated dialects can also be characterized by Hayashi 3. Figure 11.4 shows that two main groups are formed on the first axis. They are further classified into four when the second axis is taken into consideration. Theoretically Figure 11.4 shows patterns corresponding with Figure 11.3 in relative distribution. Thus by comparing the two figures the main dialects can be characterized as follows: 1. 2. 3. 4. 5.

Cambridge University: standard but snobbish BBC (News announcer): standard Scotland and villages in Norfolk: pastoral and accented London, Liverpool, USA (western), and Australia: urban but accented Hawaii and Informants’ own: neutral

CLASSIFICATION OF DIALECTS BY IMAGE

153

Figure 11.3. Distribution of evaluative words as a result of Hayashi’s Quantificational Theory Type 3 for British students (first and second axes)

This result shows that, for our student sample of Great Britain, differences in images of dialects within British English are greater than among varieties of English in other countries. 4.2 Counting The faithful presentation of percentages of answers in numerous graphs would be easier to interpret than the above graphs of multivariate analysis, but the information is too

154

IMAGES, PERCEPTIONS, AND ATTITUDES

Villages in Norfolk

Scotland

PASTORAL (RURAL)

0.5

–0.5 ACCENTEDNESS

Cambridge University STANDARDNESS

0

0.5

Hawaii

Australia

USA (Western) London

BBC (News Announcer)

URBANNESS axis 2

Own

Liverpool

axis 1

Axis 1: Class Difference Axis 2: Urbanization

–0.5

Figure 11.4. Distribution of dialects as a result of Hayashi’s Quantificational Theory Type 3 for British students (first and second axes)

complicated to allow us to grasp the overall pattern at once. Thus, a method which allowed us to count the numbers of similar answers for each informant was applied to the data. The evaluative words were divided into four groups as shown in Figure 11.3. As two groups are in contrast on both the first and the second axes, the average values for “standardness — accentedness” (first axis) and “urbanity — ruralness” (second axis) were calculated by subtracting the values of the corresponding group. As the number of evaluative words included in each group is different, a simple formula was used to calculate the two main factors:

Standard sum = Standardness × 6/5 − Accentedness Urban sum = Urbanity − Ruralness × 2/3

CLASSIFICATION OF DIALECTS BY IMAGE

155

Figure 11.5 shows average values of standardness and ruralness. Answers from British, Australian, and other students are shown in the same graph by connecting lines. The overall distribution of Figure 11.5 approximately reproduces the resulting graphs of Hayashi 3 shown by Figure 11.4. Thus once a multivariate analysis was applied and the inherent structure made simple, the structure can be recapitulated by simple calculations. Similar successful results have been obtained in the analysis of Japanese dialect images.

Figure 11.5. Average values of standardness and ruralness for British and Australian students (including their own dialects)

4.3 Analysis of Students’ Own Dialects By adopting the simple formula given above, the informants’ replies for their own dialects can be calculated in the same way. In Figure 11.6, a classical dialect division by Brook (1963) is used to show average values. But the Eastern Dialect was subdivided into two, because the average values by counties showed that the eastern half and western half (near London) tend to have different images. East Anglia was thought to have a heavy accent. The border shows an approximate agreement with linguistic phenomena like the preservation of “h” or the distribution of certain phonological and grammatical items (Wells 1982; Trudgill 1974; Viereck 1985).

156

IMAGES, PERCEPTIONS, AND ATTITUDES

Scotland

–0.75 0.0

< 1.0

–2.0

< 0.0

–4.0

< –2.0

NORTHERN –3.16

MIDLAND

Liverpool Norfolk MIDLAND –1.9 –2.6 0.33 –4.0

EASTERN

WESTERN

Cambridge London

SOUTHERN 0.12

100 km 0

100 Miles

Figure 11.6. Geographical distribution of students’ own dialect image (average value of standardness projected on dialect division map in Brook 1963)

Evaluation of the Midland Dialect is intermediate. The value of the informants’ own image can be explained by its distance from London. It is interesting that the values for Central England were not as low as stated by Trudgill, Honey, or Crystal. However, cities like Liverpool, Manchester, and Birmingham are not rated as high as we expected. We expected these large cities to receive a higher evaluation because of their degree of urbanization.

CLASSIFICATION OF DIALECTS BY IMAGE

5.

157

Discussion

5.1 Social Differences The map in Figure 11.6 shows that non-linguists’ images of dialects are reliable and worth studying. Apparent differences between Japanese and English dialect images were found. These may be partly due to differences in our research method. Socially determined accents like “BBC (announcer)” or “Cambridge University” accents in the English version seem to have influenced our results. The Japanese version investigated only geographically determined dialects. But recent studies of images including “standard language” in Japan showed that the “standard language” is thought to have an image similar to the Tokyo dialect. These results suggest that dialect images are more socially determined in England than in Japan, but that the basic structure of dialect image is the same in both countries. There is occasionally some indication in the literature that the status of urban dialects in Great Britain is different from that in Japan, at least, if the speech of working class people is taken into consideration. Urban dialects tend to be evaluated higher in Japan than in Great Britain, partly because language standardization is rapidly going on among young cultivated speakers in large cities (Inoue 1988). It has often been pointed out that the function of social class in the two countries is different; the distinction between social classes is more vague in Japan, and social movement freer. Thus, although surface differences were found between Japan and Great Britain, the basic mechanism seems to be the same. The mechanism of the formation of dialect images depends on two kinds of sources: a. linguistic distance from the center of the standard language, b. extra-linguistic social prestige of (the residents of) the given area. Theoretically, the formation of a dialect image can be explained as illustrated in Figure 11.7. A dialect image is formed by the influences of dialects themselves and stereotypical areal images or residents’ images. A dialect image can then influence changes and the diffusion process of language and can also create a dialect inferiority complex. Dialect images should be studied in this wider perspective in the future (Inoue 1989). There are some other findings which deserve special mention. The evaluation of American English was unexpected. The influence of Americanisms on the English language is often reported in Great Britain. This is perhaps because of the frequency of contact with American English in the mass media and because of the high evaluation of American culture by young people. The low status of American English according to our samples is seemingly contradictory. The relation between prestige and image is not so simple. 5.2 Prospects for Future Research In order to interpret the results of our study of dialect images, it is necessary to measure the linguistic distance of dialects from Standard English. A simple method of counting standard forms is also helpful. Preliminary research with the Linguistic Atlas of Scotland

158

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 11.7. Dialect image and its environments

(Mather and Speitel 1975, 1977) showed that the Edinburgh area is lexically nearer to the standard than other areas of Scotland and northern England. As to phonological data, if the method of Funk (1991) is applied to more phenomena in more dialects, the result must be more fruitful and inspiring. The data of Viereck (1985) will also yield a much clearer picture of English dialects. Once computerized, various treatments are possible. Multivariate analysis of English dialects will be rewarding. It is effective because it allows us to consider all the possible information. Correlation of dialects and dialect images can be calculated in the future. The success of the characterization of English dialects invites us to progress further by gathering more data from all over Great Britain. The only information necessary is information about informants’ own dialects (as to the evaluation of other dialects, we seem to have acquired enough data.) This simplified method is perhaps applicable also to other countries, although starting anew from the investigation of the basic structure of an image is ideal. Contrastive studies with other languages, for example, German dialects, would be interesting. By this kind of contrastive study, it will become possible to compare the typology of dialect images in many languages, as was done in the study of The Measurement of Meaning (Osgood, Suci, and Tannenbaum 1957). In multilingual societies like Canada, the measurement of language images would also be interesting. The measurement of language images is possible even in monolingual societies like Japan, where discriminatory treatment of foreign languages is observed in the form of impersonal (institutional) bilingualism. The study of dialect image can then proceed to a study of language images all over the world.

CLASSIFICATION OF DIALECTS BY IMAGE

159

Acknowledgments The data for this study were gathered with the cooperation of scholars from many universities. In England, Peter Trudgill, Marcel Tatham, Doug Arnold, David Britain, and many other staff members of the University of Essex were very helpful. Jennie Cheshire of London University, Sandra Harris of Trent Polytechnic Nottingham, and David Clement of the University of Edinburgh were also kind enough to distribute and collect questionnaires. In Melbourne, Australia, Helen Marriott of Monash University diligently collected the data and kindly sent them to Japan. We are greatly indebted to all the people who cooperated to allow us to carry out this study.

C 12 Subjective Dialect Division in Great Britain Fumio Inoue

The main topic of this paper is the investigation of English students’ mental maps of English dialects, applying a multivariate analysis to the computerized data. Scotland, Wales, and Northern Ireland were characterized as dialectally very different from England. England was next divided into several areas stretching from north to south. Study of the results in comparison with various kinds of maps showed that mental maps of dialects seem to be heavily influenced by ordinary geographical division. We should also attempt to explain the process of the formation of dialect images. Research in dialect maps should not stop at the geographical level. A speculative model of the mechanism behind the formation of mental maps of dialect shows that subjects formed a dialect image without a specific map in mind. Thus “dialect images” may be nearer to people’s conceptions of dialects than mental maps are. The next step in perceptual dialectology will therefore be more fruitful if “dialect images” are taken into consideration. This kind of study will hopefully make a significant contribution to the sociopsychological study of dialect. The study of mental maps of dialects or subjective dialect division flourished in Japan decades ago, and it has now been established as a field of dialectology in the United States, termed perceptual dialectology (Preston 1989; see also Butters 1995). We can test the validity of the methodology by applying it to a different language — that is, English in Great Britain. Application to the English language is interesting because comparative or contrastive study with American English and Japanese can be easily executed.

1.

A Review of Studies in Japanese Dialect Image

Research in States now. Netherlands influence of

subjective dialect division seems to be achieving prominence in the United Preston (1989: 4–13) points out that the field of study originated in the and was later employed in Japanese dialectology. This was through the a Belgian linguistic geographer, Willem Grootaers, who has lived in Japan

162

IMAGES, PERCEPTIONS, AND ATTITUDES

since 1950. Grootaers and his coworkers have drawn up areal maps showing subjective dialect boundaries for several parts of Japan (Grootaers 1964; see also Mase 1992: 162–70). Long (1993) [included in Chapter 13, this volume], who has also been influenced by Preston, has recently attempted to draw mental maps of the whole country. Most of the initial Japanese studies were conducted over small areas. Sibata’s paper (1959) on subjective dialect division is based on the Itoigawa area, which is the area covered in the comprehensive volumes of the Linguistic Atlas of Itoigawa (Sibata and Grootaers 1988, 1990, 1995). The conclusion of the Itoigawa study was that people’s subjective dialect division is more influenced by old administrative areal divisions than by the dialect itself. The next major study was of the mountainous areas in central Japan, by Mase (1992 [Mase 1964a, 1964b, 1964c; see Chapters 6 and 7, this volume]). His results were different from Sibata’s in that the subjective dialect divisions coincided with actual dialectal differences. In the early 1970s, research had been conducted in northern Japan over a larger area (Inoue 1972). Correspondence between subjective dialect division and actual dialectal differences was found in mountainous areas, but in geographically flat areas administrative areas were more influential. In Inoue (1972) I concluded that people’s conception of dialect areas is influenced by a kind of mental map of the natural geography of the area. I have also tried to collect preliminary data on subjective dialect division for the whole of Japan, but I have found this study less rewarding than the study of dialect images, which will be explained next. Inoue (1988) is concerned with perceptual dialectology and includes dialectometrical analysis by means of multivariate analysis of dialect images. Some tentative tests were also made which will help us to understand the mechanism of the formation process of a dialect image. The DIALECTAL SPEECH-GUESSING TEST is a method in which students are asked to listen to recorded speech of various areas. This method is similar to the Intelligibility Test conducted for Native American languages by Voegelin and Harris (1951). The results in Inoue (1988) showed that the students’ conceptions of actual dialects are vague and incorrect. Students seem to be able to identify only the small number of dialects which are often heard in the mass media. The DIALECTAL WORDGUESSING TEST is a method of making students listen to artificially accented words. High school students in Tokyo were asked to guess the dialect for each word which was phonetically characterized. Several phonetic features were selected which are popularly known in Japan as markers of certain dialects, and the markers were correctly understood as symbolizing the dialects. Following Inoue (1988), I investigated images of dialects. University students in various parts of Japan were asked to characterize the image of their own dialects using a method which is similar to the Semantic Differential Test once popular in psycholinguistics (Osgood, Suci, and Tannenbaum 1957). A multivariate technique was applied, and the result showed clear geographical differences. The resultant map of dialect images in Japan appears in Inoue (1995 [Chapter 11, Figure 11.1, this volume]).

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

163

Some helpful insights into the formation of dialect images were acquired from the studies listed above. First, phonological characteristics of dialects were found not to be influential in the perception of dialects, while phonetic characteristics were. Moreover, the phonological perception of the evaluator was found to be essential. Japanese studies have shown no clear correspondence between dialect image and actual dialect. For example, two big neighboring cities, Osaka and Kyoto, are in sharp contrast as to dialect image (Inoue 1992). But this contrast does not appear on the national map of subjective dialect division. Nor can students actually identify the two dialects in recorded speech. Stereotypical judgments by students were often attested. It is also possible for students to give dialect images without listening to the speech itself. The methodologically useful result of the investigation of dialect image was that an image was found to be composed of two factors: an intellectual factor and an emotional factor (Inoue 1989). Thus Preston’s selection (1989: 51–52, 71–72) of evaluative words (“correct” and “pleasant”) is appropriate in its use of terms for intellectual judgments and emotional reactions.

2.

Dialect Image of English

The same methodology was applied to English dialect in Great Britain, with two aims: to understand better the dialectal structure of English in Great Britain and to test the validity of the methodology itself. The actual research was planned while I was staying at Essex University in England in 1989. Students were asked to fill in a questionnaire including a map. The collection of the student data was carried out by staff members of several universities in Great Britain. The average values of the students’ image of their own dialects were first calculated for each county (shire) in Great Britain, and one method of multivariate analysis was applied. The multivariate analysis indicated that dialect image was composed of two factors: ACCENTEDNESS/STANDARDNESS and URBANNESS/PASTORALNESS. The values for the accentedness factor are shown in Figure 11.6 [this volume] (projected on the base map of dialect division in Brook 1963).1 In Figure 11.6, the figures generally decrease with the distance from London, indicating a stronger perception of accentedness. The values are almost continuous from south to north. In Great Britain, the correlation between the degree of dialect image and the geographical distance from London can be confirmed, Scotland being an exception. A great difference in neighboring counties is shown eastward from London: here a clear break of values was observed along the border of Greater London and Essex. The dialect image of several popular dialects, including Cambridge English and American English, was also investigated. Figure 11.4 [this volume] shows the result of a multivariate analysis called Hayashi 3 (Hayashi 1952; see below) showing, for example, that a western American accent sounds rather coarse but urban to British students.

164 3.

IMAGES, PERCEPTIONS, AND ATTITUDES

Subjective Dialect Division of English

In Great Britain a second type of research was conducted. A question about subjective dialect division was added at the end of my 1989 questionnaire, and students who volunteered were asked to draw lines on a British map in order to divide the areas according to the accents or dialects they perceived.2 The data were all put into machine-readable form. The information about geographical distribution was entered in a simplified form. This is quite different from methods used by Long (1993) or Preston (1989), in which actual positions of lines are exactly recorded on computer and on maps. In my method, the only information coded is whether or not a particular county is included in a subjective dialect division. The data format can be understood by comparing one example of an answer sheet and its computerized form. Figure 12.1 shows the lines drawn by one student and is similar to the resultant subjective dialect division of this study as a whole as recorded on Figure 12.3. As an example of computer input, Figure 12.2 shows the data format for student number 041. One line of the data corresponds to one subjective dialect area drawn on a map by one student. Each column represents a county of Great Britain; cities are listed at the top as an index. As student 041 divided the whole area into 13 dialectal areas, 13 lines were prepared for data input. For each line of the data format, the counties which were included in the area are shown by the numeric “1.” When the students assigned labels to the dialect areas, each label was input at the ends of the lines; minute differences of distribution within a county were not considered. In principle, numeric “1” was input in a line when counties were fully included in an area, numeric “1” was input in two lines when one county was divided into two areas, and numeric “2” was input when a part of one county was included in a neighboring area apparently by chance, as a rough guess, or by a slip of the pen. Thus when one county is separated by a line, two “1”s are input in the two corresponding lines of data. This method simplifies the work in data input and makes calculation of multivariate analysis easier. The theoretical basis of this simplified format is the belief that lines drawn by the students may be only approximate and not precise. If students often make mistakes or rough guesses in recording where they perceive dialect boundaries, it is not worth the trouble of inputting the exact positions of lines. We must admit that people are often more sensitive to dialectal differences near their birthplaces. In this case, the lines are usually drawn meticulously and can be distinguished from rough guesses or slips of pens. 3.1 Calculation of Dividing Lines The patterns of lines drawn can then be analyzed by simple calculation. By counting the frequency of answers “1” for each county, the counties which are divided by lines (or within which dialect differences are noticed) can be pointed out. Where the number of “1”s is more than the average, dividing lines often run within the county. The map for student 041 shows (1) that the counties near the great dividing lines of dialect areas are

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

165

Figure 12.1. Example of student data (student number 041)

divided into two, and (2) that the counties with irregular boundaries are divided into two. These facts suggest that counties are divided mainly because of the difficulty of drawing exact lines along the county border and not because of students’ confidence that the county should be dialectally subdivided.

166

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 12.2. Example of data input (student number 041)

3.2 Resultant Map as a Conclusion We are now ready to analyze the entire pattern of subjective dialect division offered by English students in this study. The simple counting of labels given to dialect areas was helpful in forming a general idea. Next, one method of multivariate analysis was applied, and subjective distances among the counties were calculated. In order to make the conclusion clearer, see Figure 12.3. 3.3 Results: Simple Counting Seventy-seven students filled in the maps. There were two types of answers on the maps: one type with dividing lines only, and another with labels added. Sixty percent of the students who drew dividing lines on the map labeled the areas. A total of 876 areas were obtained. Although one student divided Great Britain into twenty-eight areas and three divided it into four areas, the average number of subjective dialect areas of Great Britain was 11.4 (876/77). The data with labels were analyzed first. About one hundred different labels were given in all. Seventeen of these were given by more than six students. The three most frequent of the seventeen labels were IRISH, SCOTTISH (or SCOT, SCOTS), and WELSH. But further study is necessary to ascertain the linguistic status of these areas. Some students may have given these labels to show that languages other than English are spoken there. But most of the students were probably intending to indicate that the English now spoken there is a little different from the English used in England. I listed the students’ dialect labels for each county and selected the label used most often to represent that county. I then attempted a division of the counties according to these dialect labels. The resultant dialect division by the names is shown in Figure 12.3.

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

167

Figure 12.3. Subjective dialect division of Great Britain

There is a problem of hierarchy of dialect division which was pointed out by Long (personal communication, August 1992). For example, when the area called GEORDIE is drawn on a map, does the student consider it a part (or subdivision) of the NORTHERN dialect, or is it an independent area which is on the same level as NORTHERN? This question is difficult to answer if the student does not clearly indicate the relationship. But the labels given by many students provide several hints. For example, the Liverpool area is labeled SCOUSE by some and NORTHERN by others. This shows that SCOUSE may

168

IMAGES, PERCEPTIONS, AND ATTITUDES

be a subdivision of the NORTHERN area, and that there are hierarchical systems in dialect areas. In Figure 12.3 the labels with uppercase letters show the large-scale divisions, and the labels with lowercase letters show subdivisions of the areas. However, Figure 12.3 does not show hierarchical relationships, and further explanation is necessary. As for England, the divisions which cover a wide area are NORTHERN, MIDLAND, and SOUTHERN. Further division of SOUTHERN is often given as EASTERN and WESTERN. Other dialectally conspicuous counties are labeled, such as YORK (or YORKSHIRE), LANCASHIRE, and NORFOLK, as well as LONDON and the HOME COUNTIES around the London metropolitan area. Sometimes nicknames of dialects for big cities are given: for example, COCKNEY for LONDON, SCOUSE (or LIVERPUDLIAN) for Liverpool, GEORDIE for Newcastle upon Tyne, BRUMMY or BRUMMIES for Birmingham. 3.4 Results: Multivariate Analysis It is relatively easy to analyze according to the simple counting method when labels are given as cues. However, the distribution pattern of the maps without labels needs to be analyzed using some quantificational method. For this kind of non-numerical, categorical data, one kind of multivariate analysis, called HAYASHI 3 — or more exactly, HAYASHI’S QUANTIFICATIONAL THEORY TYPE 3 (Hayashi 1952) — is applicable and provides further insight into the student subjects’ conceptions of areal division. The pattern of grouping the counties can be treated, and the results can be interpreted as Factor Analysis for numerical data. The results of HAYASHI 3 can be conveniently interpreted when the resultant values of the first and second axes are plotted on a graph. The first and second axes approximately correspond to the first and second factors of factor analysis. In Figures 12.4 and 12.5, counties which were often grouped together by the student subjects are plotted close to each other because of similar values of the first and second axes, and counties which were rarely grouped together exhibit scattered plotting. The numbers in Figures 12.4 and 12.5 are the county codes used for this analysis. (For reference, the numbers also appear in Figures 12.2 and 12.3.) Subjective distances between the counties are thus shown by application of the multidimensional analysis. Distance or difference among the counties would be difficult to calculate otherwise. Various applications of HAYASHI 3 were tried, each time with different conditions of calculation. In some cases the analyses were applied to the whole data, and in other cases to England only. Figure 12.4 shows the results for the whole of Great Britain and shows that subjects considered IRISH, SCOTTISH, and WELSH different in that order. As the values of these three areas are so different from that of England, another calculation is necessary to investigate the internal divisions of England. Figure 12.5 is the result of the analysis when only the counties of England were selected. This figure seems to reflect actual geographical distribution in the form of a map of England which is tilted 90 degrees. Counties in England are arranged into three groups by the value of the first (horizontal) axis, that is,

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

169

Figure 12.4. Distribution table of Great Britain dialects as a result of Hayashi 3

into NORTHERN, MIDLAND, and SOUTHERN. But by the second (vertical) axis, SOUTHERN dialects seem to be divided into WESTERN and EASTERN. In this figure the interrelationship of counties is shown diagrammatically. The same result can be shown geographically on a map by drawing thick lines between dialectally different counties, as in Figure 12.3. This result shows the usefulness of the multivariate analysis in classifying a large amount of non-numerical data. Cluster Analysis was further applied to the same data, but the result was not statistically significant. This is perhaps because the data were not numerical (continuous) but categorical (nominal). Further processing of these categorical data into numerical values with the technique utilized by Goebl (1982), for example, would be interesting.

170

IMAGES, PERCEPTIONS, AND ATTITUDES 3.5 18 17

3.0

16 15

2.5

Western

2.0 SOUTHERN

1.5

11

1.0

Geordie NORTHERN

12

0.5

Lancashire 31

0.0

14 33

13

–0.5 –1.0

61

03 04 41 * 42 45 Cockney 05 02 44 43 47 London 06 01 07 08

–1.5 –1.5

–1.0

32

Brummy 46

34

36 * 35 51

54 53

75 77

73 76 65

72 55

64

71

York

Liverpudlian Scouse

MIDLAND Home Counties

Eastern Norfolk

–0.5

62 37 52

63

74

0.0

0.5

1.0

1.5

2.0

Figure 12.5. Distribution of dialects as a result of Hayashi 3 (England only)

4.

Comparison with Other Maps

4.1 Comparison with Dialect Image The student subjects’ subjective dialect division was shown above in the form of maps. The next step is to investigate the origin of this division. Figure 12.3 will be compared with several kinds of distribution maps. First, the map will be compared with dialect image. Previous studies of dialect image in Japanese and English showed clear and reasonable results (Inoue 1993). Classification of Japanese and English dialects was made possible, reflecting perhaps a stereotypical reaction to the dialects. The English dialect image map (Figure 11.6), which was derived from the same data, was compared with Figure 12.3, but no correspondence was clear. The students’ conceptions of dialect division and dialect image seem to be independent. This issue will be discussed later.

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

171

4.2 Comparison with Dialect Division As dialectologists we would like to know whether this subjective dialect division reflects any dialectal distribution. If it does, the map will show what kind of dialectological phenomenon most influences the students’ subjective dialectal division. However, there are no maps showing a pattern corresponding to the subjective dialect division. Many attempts at dialect division and classification by scholars were compared with Figure 12.3, for example, Figure 11.6 from Brook (1963: 153), Figure 12.6 from Viereck (1986: 250), and a map by Trudgill (1990: 33), which is shown as Figure 12.7. Most of the classifications coincided, in that England could be divided into NORTHERN, MIDLAND, and SOUTHERN. The dividing line between NORTHERN and MIDLAND was sometimes similar. But no existing classification completely matched the students’ subjective dialect division when counties were compared one by one. Most of the dialect maps fail to show what kind of dialectal features were actually used to draw dividing lines. The results of recent research on dialectometry were referred to, especially the results of multivariate analysis; the recent works of Viereck (1986, 1992) were especially helpful because the maps in these studies are based on numerous concrete distribution data. The geographical distribution of individual dialect features was then considered. The next possible step would be to compare the subjective data with computerized distribution data of individual linguistic features. Multivariate analysis would be useful to find the features which show similar or identical distribution (Inoue and Fukushima, 1997). 4.3 Comparison with Ordinary Geographical Maps Maps students use in primary- and secondary-school geography classes in England seem to influence the students’ geographical conceptions. Various school educational maps and maps used for special purposes — for example, weather maps — were thus collected. Commonly used areal divisions showed an approximate correspondence with students’ subjective dialect division. Labels such as NORTHERN, SOUTHERN, and MIDLAND are used often, though the exact positions of dividing lines differ. This fact suggests the stereotypical influence of geographical labels. This study of perceptual dialectology or subjective dialect division of English was planned in order to look for linguistic clues of people’s perception of dialects. But the English data so far seem to show no clear dialectal background. Further study will be necessary to confirm this tentative and discouraging conclusion.

5.

Discussion

This study so far is only a first attempt at the study of perceptual dialectology in Great Britain. Our results have been compared with those in another country, Japan, in order to see if there are common features. If universal ideas and tendencies in the subjective dimension of dialects are found, such studies will become more fruitful. Contrastive study with results for American dialects found by Preston (e.g., 1989) should also give effective results.

172

IMAGES, PERCEPTIONS, AND ATTITUDES

3–5 heteromorphs 6–8 heteromorphs 9–13 heteromorphs

SCOTLAND

WALES

100 km 0

100 Miles

Figure 12.6. Dialect division by Viereck (1986)

Inoue (1989, 1995) presents Figure 11.7 to show the relation between dialect image and subjective dialect division; it poses subjective dialect division as one of the central terms in perceptual dialectology. This figure suggests that the underlying mechanism of a dialect image is twofold and that the same can be said for the subjective dialect division: (1) linguistic, influenced by dialect systems, and (2) nonlinguistic or sociopsychological (e.g., area images or mental maps).

173

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

SCOTS

NORTH

Nor thumberland NORTHERN

LOWER NOR TH Dur ham & North Yorkshire

Cumbria

Lancashire

East Yorkshire

South Yorkshire CENTRAL

EASTERN CENTRAL

WESTERN CENTRAL Staffordshire

Lincolnshire EASTERN COUNTIES

Leicestershire

EASTERN CENTRAL EAST Northamptonshire & Cambridgeshire

NOR THERN SOUTHWEST

EASTERN SOUTHWEST

Essex

Wiltshire & Hampshire

WESTERN SOUTHWEST WESTERN

SOUTHERN

East Cornwall & Devon

Figure 12.7. Dialect division by Trudgill (1990)

Berkshir e SOUTHEAST Sussex

Dorset & Somerset

SOUTH

Kent & Surrey

174

IMAGES, PERCEPTIONS, AND ATTITUDES

5.1 Mechanism of Formation of Dialect Image The basic mechanism of the formation of a subjective dialect division may distinguish two types of mechanisms, one dialectal and the other semantic. 5.1.1 Dialectal Two kinds of experiences may contribute to form a subjective dialect division, the INDIVIDUAL and the COLLECTIVE. The INDIVIDUAL source of a speaker’s dialect division is derived from actual first-hand linguistic experience; some linguistic features must have acted as a clue to indicate the dialectal difference. The COLLECTIVE source of dialect images of subjective dialect areas is found in the stereotypes perpetuated by mass media, independent of the personal experience of any individual. Images used in the media are often a conglomeration of people’s past experiences, and they induce as well as reinforce ordinary people’s images of the corresponding subjective dialect areas. 5.1.2 Semantic Problems of Dialectal Terms Of course, the mapping method is not perfect; we need to be cautious about drawing conclusions from maps. The relation between a dialect image and a subjective dialect map can be understood as a general semantic problem, like that of color terms or constellations in the night sky. Experts in the field may classify a whole semantic field with words, but laypersons usually know only one portion of the whole semantic field. Ordinary people’s ideas are often based on fragmentary experiences. People’s conception of dialect is the same. Obviously, no nonlinguist knows the dialects of a language completely. The layperson’s knowledge is sporadic and is in the form of points on an abstract surface, not in the form of actual two-dimensional geographical surfaces. When laypersons are asked to classify the whole surface of a country or a part of a country, they must try to do it on the basis of their partial knowledge. Separating and dividing the earth’s surface in the form of maps does not reflect the speaker’s consciousness of dialects. We should thus be cautious of answers in the form of maps. We may also doubt whether laypersons understand dialect differences in the form of geographical maps (Preston 1989: 14–22). When people actually travel around an area either by car or on foot, a kind of (somewhat skewed) mental map may exist in their heads. But as for dialect images, people can often hear dialect spoken without actually going to the place where it is spoken. Moreover, people often form dialect images even without listening to the actual dialect. Stereotypical images of a place or people living there help form dialect images. Thus, using maps to grasp people’s images of dialects is not always appropriate. The above result, that the subjective dialect division did not coincide with any concrete dialectal feature, must have derived from this mechanism. The same doubt will apply to the study of mental maps in general. But still we cannot deny that a subjective dialect division or dialectal mental map is convenient and important as an analytical tool for secular dialectology or perceptual dialectology.

SUBJECTIVE DIALECT DIVISION IN GREAT BRITAIN

6.

175

Conclusion: Two Directions for the Development of Mental Maps

The above results indicate that subjective dialect division is not based only on dialect itself. Thus further development of the study of subjective dialect division is possible in two directions: (1) the formation of combination maps — in looking for relationships with traditional dialectology, and (2) the inclusion of dialect images — in looking for a psycholinguistic foundation. This means that perceptual dialectology can develop in two main directions. 1.

2.

Mapping has long been the main concern in the field of linguistic geography. In my study, only subjective dialect boundaries were studied. Tests must be devised to look for individual linguistic features which influence perception of dialects. Such tests will also constitute a new field of dialectometry, as statistical analysis is useful and necessary. The purpose of such tests is twofold: (a) to see which linguistic features are effective in forming subjective dialect divisions — not all the linguistic features are equal psychologically (the concepts of stereotype, marker, and indicator advocated by Labov 1972 will be useful); and (b) to see which linguistic features are effective for perception in general. Perception tests in my study of Japanese dialects (Inoue 1973) showed that some phonetic features are effective in creating an image of some dialects. Whether any vowels or consonants are influential in perception is a basic problem of perceptual dialectology. These features are often reflected in “dialect souvenirs” for tourists or in popular books on dialects. This would be a development in the field of “secular dialectology.” A second possible development for perceptual dialectology is the inclusion of the study of dialect image. Dialect images should be studied if we want to know the actual form of people’s conceptions of dialects. People’s images of dialect are not flat or monotonous ones which a mere map can do complete justice to. The images are more complicated and dynamic, as shown in my previous study (Inoue 1989, 1995). Maps can only inadequately show the real structure of a dialect image, which should be understood multidimensionally. Research in the mental maps of dialects should now take off from the earth’s surface and fly into the human mind.

Acknowledgments The data for this study were gathered with the cooperation of scholars from many universities. In England, Professors Peter Trudgill, Marcel Tatham, Doug Arnold, David Britain, and many other staff members of the University of Essex were very helpful. Professors Jennie Cheshire of London University, Sandra Harris of Trent Polytechnic Nottingham, and David Clement of the University of Edinburgh were also kind enough to distribute and collect questionnaires. I am greatly indebted to all the people who cooperated to allow me to carry out this study. Data processing has been done by GLAPS (Generalized Linguistic Atlas Printing System), a packaged program developed by Professor Tsunao Ogino, and by SPSS2 developed by HITAC Computer Company. This paper is a revised version of a paper read at the session “In Honor of Takesi Sibata” at the Annual Meeting of the American Dialect Society, held in New York on December 29, 1992 (original title: “Subjective Dialect Division of English”).

176

IMAGES, PERCEPTIONS, AND ATTITUDES

Notes 1. The differences in degrees of urbanness were slight, partly because the evaluative words used were few. 2. The subjects were students from various parts of Great Britain, mainly from Essex University, northeast of London. The map portion of the questionnaire was used in Great Britain only, though the part concerned with dialect image was also used in Australia. The map (reproduced here as Figure 12.1 with the dialect boundary lines and dialect names inscribed by student 041) had the county names and boundaries printed on it.

C 13 Geographical Perceptions of Japanese Dialect Regions Daniel Long

1.

Background of This Study1

In 1954, Misao Tôjô (referred to as one of the two founders of modern Japanese dialectology) wrote, “in delimiting a dialect territory, there is no other way but to use the dialect consciousness of the ordinary people as the starting point, and make adjustments to this, surveying the distribution of characteristic features which will support this” (Tôjô 1954: 11). This statement prompted a wave of dialect division studies based on the dialect consciousness of the populace. In their Itoigawa survey of 1957, Takesi Sibata, Willem Grootaers, and Munemasa Tokugawa attempted dialect divisions based on the self-reports of informants regarding dialect differences (Sibata 1959; Mase 1964a, 1964b; [Chapters 4, 6, and 7 of this volume]). On the relationship between the dialect consciousness of the populace and dialect divisions, Tokugawa says the following: That which we now call dialect research originated in the interests of the masses, but we may state that it has proceeded based on its own methodologies. The dialect divisions above are a good example of this. Nonetheless, although we can say that this is the dialect division theory of scholars, … this research is not being advanced in isolation from the experiences and intuitions of the masses. To put it strongly, we might say that dialect division theory is an attempt to conceptualize the areal division consciousness of the masses. (Tokugawa 1981: 146–47)

Although this statement may seem somewhat negative, it acknowledges the deep relationship between dialect divisions and the dialect consciousness of the masses. In 1977, Tokugawa clearly indicated that the perceptions and the realities of dialect divisions are not the same thing but that both are worthy of scholarly research; he states, “We must attempt to clarify the relationship between the divisions which dialect speakers perceive and the actual state of dialects” (1977: 347). In this chapter, although I do not fully clarify the relationships Tokugawa mentioned, as an initial attempt, I investigate considerable quantitative data to construct perceptual

178

IMAGES, PERCEPTIONS, AND ATTITUDES

dialect maps that reflect the dialect consciousness of large numbers of people. The methodology that I used in this research is based on the ideas of Dennis Preston who adapted the mental map methodologies of cultural geographers for use in dialectology and proposed “perceptual dialect maps.” In his 1989 book, Preston reviews the results of Sibata and Grootaers’s Itoigawa survey [Chapter 4, this volume], as well as the work of Rensink published in the Netherlands in 1955 [Chapter 1, this volume], placing his own work within this historical context. Indeed, we may say that the research on dialect divisions presented here, based on the dialect consciousness of the masses, was born in the Netherlands, raised in Japan, reached adulthood in the United States, and has now returned to Japan.

2.

Methods

The informants in this survey were presented with blank maps of Japan (with only the prefectural boundaries drawn in) and asked to perform the following tasks: 1. Draw lines between the places where you think the speech is different. 2. On the map, indicate what these ways of speaking are called. In the first half of this chapter I examine data from surveys conducted in the spring of 1988 in Okayama (twenty informants) and in Aichi (seventy-four informants). In the second half, I use data collected in 1991 in Kantô (seventy-five informants, mainly from the Tokyo-Yokohama area) and Kansai (244 informants).2 Informants in all four groups were natives of their regions and had no specialized education regarding dialects. Here let us examine how the maps drawn by individual informants were tabulated for a quantitative analysis, taking the Tôhoku region as an example. Many of the informants included a Tôhoku-ben “dialect” region on their hand-drawn maps as did the informant in Figure 13.1, but the actual area covered by such a label differs from one informant to the next. Thus it is necessary to illustrate the data from all the informants on a single map. This need is answered by the “gradient map,” which allows us to take the results from a large number of hand-drawn maps and compress them into one visual expression of the data. Let us see how a gradient map is constructed. First, I chose one dialect such as Tôhoku-ben that I wanted to examine and selected all those informants who indicated that speech region on their hand drawn maps. Then, one by one, I traced each informant’s Tôhoku-ben boundary line(s) onto a blank working map. Figure 13.2 is a map onto which I traced the nineteen boundaries for Tôhoku-ben drawn by the twenty Okayama informants (one informant did not draw this region). The number “19” to the north of the top line indicates that this area was included in all nineteen of the informants’ Tôhoku-ben regions. In other words, the perceptions of nineteen people overlap in this area. As one heads south and crosses the perceptual dialect boundaries of various informants, the number of people whose perceptual regions overlap decreases, and hence the values on the map drop. Up to this point, I have been referring to the speech of this region as Tôhoku-ben, but I chose this as the dialect name for this region because it is the label used by the largest

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

179

Figure 13.1. Hand-drawn map

number of informants. Later I will examine the gradient map of Tokyo-ben from the Aichi survey. The area covered by this label is larger than Tokyo proper, corresponding more to the area commonly referred to as the Kantô region, but I still elected to use the dialect name preferred by the greatest number of informants themselves. I should point out here that some informants wrote labels (such as Tôhoku-ben or Kantô-ben) on their maps without drawing the boundaries for them. In these cases, I did not use those informants’ data for that particular area.

180

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 13.2. Tally map for Tôhoku-ben

3.

Dialect Names Used by Informants

Let us take a look at the labels informants in the Aichi and Okayama surveys used for the dialect regions they outlined on their hand-drawn maps. They are listed in Table 13.1. Place names account for the majority of these labels, but there are descriptive names as well: “zû-zû-ben” found in Hokkaidô and Tôhoku, “the accent of Tôhoku”3 used in reference to northern Kantô, “similar to Tôhoku-ben” or “has ke at the end of sentences” in Hokuriku, “similar to Tôhoku-ben” on the Japan Sea side of the Chûgoku region, and so on. Among the terms for language, -ben was the most common with -hôgen “dialect,” kotoba “talk” also appearing occasionally. In Table 13.1, I have divided the regions of

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

181

Japan in the customary way to give a rough idea of the frequency with which each region was identified by the informants. It must be pointed out, however, that the actual variety on the hand-drawn maps was much more complex than this indicates. (We will deal with those results below.) For example, not all informants identified the Hokuriku region as a separate one; some included it as a part of Tôhoku-ben or of Chûbu-ben. To give an example of an atypical dialect division, two informants drew a boundary around the Sanin region and the Inland Sea side of Shikoku, labeling this area as Setouchi-kotoba “Inland Sea talk,” or as Setouchi.4 Next, I made a very general distinction between those informants who gave a dialect name to a region (whether it be to the region as a whole as with the term Kansai-ben or to only a portion of it as with Mikawa-ben) and those who completely ignored a region by not labeling any part of it. Table 13.2 divides Japan into ten regions and shows the percentage of informants in the two groups who gave any sort of label to the regions. The number of Okayama informants was very small, and there was a general tendency for their maps to be drawn in greater detail than those from Aichi.5 Although a careful comparison is very difficult, one can nonetheless identify some broad characteristics. In both surveys, Tôhoku and Kansai are the most often labeled regions, and Hokuriku is the least recognized by both groups. Furthermore, there is a greater tendency for the Aichi informants (despite their maps’ being, on the whole, less detailed) to label their own Chûbu region; likewise a large percentage of the Okayama informants have labeled the Shikoku and Chûgoku regions that surround them. In other words, the informants have, as may be expected, labeled areas closer to home in more detail.

4.

Degrees of Dialect Perception

When the method of stacking up boundary lines on a working map (as described above) is used, one gets little “cells” between the boundary lines as shown in Figure 13.2. These “cells” were treated as follows: I grouped those with values between 1 and 9 together. I then continued this process for groups of 10 to 19, 20 to 29, 30 to 39, 40 to 49, and 50 to 59. The gradient maps produced in this way (as seen in Figures 13.3, 13.4, and 13.5) give us a dynamic view of the perceptual regions of dialects.

182

IMAGES, PERCEPTIONS, AND ATTITUDES

Table 13.1. List of all dialect names used (*The ‘A’ and ‘O’ columns indicate the number of responses from the Aichi and Okayama respondents, respectively) Hokkaidô Hokkaidô-ben Hokkaidô-kotoba Kyôtsû-go Hyôjungo Zûzû-ben Dosanko-kotoba Ainu-kotoba Ainu-go Ainu-ben Tôhoku Tôhoku-ben Tôhoku-hôgen Akita-ben Aomori-ben Iwate-ben Yamagata-ben Miyagi-ben Fukushima-ben Zûzû-ben Tsugaru-ben Kantô Kantôkei Kantô-hôgen Kantô-ben Kantô-go Ibaraki-ben Tokyo-ben Kanagawa-ben Chiba-ben Saitama-ben Yamanashi-ben Hama-kotoba Yokohama-ben Kyôtsûgo Hyôjunkei Hyôjungoppoi Hyôjungo Tôhoku-ben no akusento Kyôtsûgo ni chikai

A* O* Chûbu 9 5 Chûbuchihou no -kotoba 2 Chûbu-hôgen 2 Shizuoka-ben 2 Nagano-ben 1 Shinshû-ben 1 Gifu-ben 2 Mikawa-hôgen 110 Mikawa-ben 1 Nagoya-ben Owari-ben Owari-hôgen 530 160 Hokuriku 2 1 Hokuriku-namari 2 Hokuriku-ben 1 2 Toyama-ben 2 Nîgata-ben 2 Sado-ben 2 Kanazawa-ben 2 Tôhoku-ben ni chikai 1 3 Hanashi-kotoba no saigo ni 2 3 ke' ga tsuku

1 1 6 5 2 2 2 170 6 1 1 2 1 1 1 1 1 1 6 1 1 1

Kansai Kansai-kei Kansai-hôgen Kansai-ben Kyôto-ben Kyô-ben Kyô-kotoba Ôsaka-ben Wakayama-ben Nara-ben Banshû-ben Ise-ben Kawachi-ben Kishû-ben Settsu-ben Shikoku Shikoku-ben Kagawa-ben Tokushima-ben Ehime-ben Tosa-ben Tosa-kotoba

A

1

O 1 1 3 1 2

Shikoku (cont.) Kôchi-ben Awa-ben Iyo-ben Uwajima-ben Sanuki-ben Setouchi-kotoba Setouchi

2 1 350 1 460 7 Chûgoku 3 1 Chûgoku/Shikoku-ben 1 Hiroshima-ben Okayama-ben 1 Shimane-ben 4 Tottori-ben 1 1 San'yô-ben 3 3 San'in-go 1 Yamaguchi-ben 1 Iwami-ben 1 Izumo-ben 1 Honshûseibu no kotoba Tôhoku-ben ni nita hôgen Kyûshû 2 Kyûshû-ben 1 Kyûshû-hôgen 460 140 Kyûshû no hito no kotoba 140 9 Fukuoka-ben 1 Hakata-ben 1 1 Kagoshima-ben 170 110 Kumamoto-ben 1 1 Saga-ben 2 Satsuma-ben 5 Nagasaki-ben 1 Ôita-ben 4 Miyazaki-ben 3 1 Okinawa Okinawa (no) -kotoba 4 1 Okinawa-ben 2 Okinawa-hôgen 2 Ryûkyû-go 1 Ryûkyû-hôgen 4 Ryûkyû-ben 1 Ryûkyû (no) -kotoba

A 1 1

O 5 1 1 3

1 1

2 6

1 1

1 9 120 3 3 1 4 1 3 1 1

250 110 2 1 1 100 6 6 7 1 3 1 4 1 1 1

1 100 4 2 2 2 1

1 7 2 2

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

Figure 13.3. Gradient perceptual dialect map of “Tôhoku Dialect,” Aichi informants (n = 60)

183

184

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 13.4. Gradient perceptual dialect map of “Tokyo Dialect,” Aichi informants (n = 52)

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

Figure 13.5. Gradient perceptual dialect map of “Kansai Dialect,” Aichi informants (n = 60) Table 13.2. Percentage of informants who gave dialect names for a region Aichi

Okayama

Hokkaidô

36

45

Tôhoku

83

90

Kantô

43

80

Hokuriku

10

35

Chûbu

77

55

Kansai

84

90

Shikoku

13

85

Chûgoku

20

85

Kyûshû

54

80

Okinawa

31

50

185

186 5.

IMAGES, PERCEPTIONS, AND ATTITUDES

Analysis of Perceived Dialect Regions

In this section, I will examine the dialect perceptions of the Aichi informants based on their composite map (Figure 13.6) and will comment briefly on the Okayama map (Figure 13.7). On both maps, I have illustrated the areas where 50% of the informants’ data overlapped. The total number of informants who drew lines for an area differs from one perceptual region to the next. Of the seventy informants in the Aichi survey, only sixty indicated a Tôhoku-ben region on their hand-drawn map. Thus, on the composite map, the area identified as Tôhoku-ben is that where thirty people’s data overlapped. The area for Kantô-ben (combined here with Tokyo-ben and Hyojungo “standard”) was drawn by fifty-two people, so the composite map shows the area that overlapped for twenty-six informants. For Kansai-ben, the line was drawn where half of the fifty-seven informants’ data overlapped (twenty-nine). The dialect names shown on the composite maps are those given by the informants. The Aichi map includes all those given by at least five people, and the Okayama map those given by at least three. The size of the type used in the labels reflects the relative frequencies. 5.1 Home Prefecture It has been pointed out in studies of dialect perception up to now that people give more detailed responses for their native regions. In this survey, however, we find a great difference in the way in which the two groups of informants treated subdivisions of their home prefectures. First, in Okayama, there were only four people who perceived any dialect boundary within the prefecture; there were eleven who labeled but did not subdivide it and five who didn’t label it at all. Even among these four who subdivided it, two made only vague divisions and gave no names for them. It is safe to conclude that the Okayama informants do not sense any great dialectal differences within their own prefecture. In contrast, the overwhelming majority (forty-one) in Aichi subdivided their prefecture. (Fourteen did not subdivide, and the remaining fifteen did not indicate Aichi at all on the maps.) In other words, 75% of the Aichi informants who indicated Aichi on their maps subdivided it, compared with only 27% in Okayama. If the Aichi informants had drawn more detailed maps in general, then this subdividing of the home area would perhaps simply reflect their predilection for attention to detail. The fact is, however, that exactly the opposite is true (see note 5 and Table 13.2) — they provide less detail than, in general, Okayama informants do. Therefore, we may conclude that there are perceptions of the home area that are both detailed (e.g., Aichi) and not detailed (e.g., Okayama). 5.2 Hokkaidô Dialect To make the map given to informants as comprehensive as possible, the northern island of Hokkaidô and the southern islands of Okinawa were placed as insets (see Figure 13.1),

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

187

Figure 13.6. Composite map of perceptual dialect regions, Aichi informants (n = 70)

and we cannot rule out the possibility that some informants may have been influenced by this placement. However, in spite of the line on the survey map, a surprisingly high 66% (forty-six people) of the informants still drew their own line between Hokkaidô and Tôhoku. In this survey, I set two criteria for determining that Hokkaidô and Tôhoku had been identified as separate dialect regions by the informants. These are that: (1) a line was drawn between Hokkaidô and Tôhoku, or (2) different dialect names were given to BOTH of the areas (i.e., “Hokkaidô-ben,” “Tôhoku-ben”).

188

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 13.7. Composite map of perceptual dialect regions, Okayama informants (n = 20)

5.3 Tôhoku Dialect The perceptual dialect name that appeared most often in the data was “Tôhoku-ben.” Of the seventy Aichi informants, fully 75% used the label, and seven more drew a boundary around this region even though they placed no label on it. On the gradient map in Figure 13.3, one should note that both the northern half of Niigata Prefecture and the whole of Fukushima Prefecture are included in this region.

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

189

5.4 Kantô Dialect Figure 13.4 is the gradient map of the Kantô region, which includes Tokyo. It is labeled “Tokyo-ben” (“Kantô-ben”) because, even though the area is much larger than just Tokyo, this was the label used by most informants (eleven out of seventeen) who drew this area. Also, when one examines the 50% agreement areas for “Tôhoku-ben” and “Tokyo-ben” on the composite map, one notices an empty zone (referred to as a “trough” by Preston 1989: 120) between the two. This is not simply a result of the methodology of the map construction but, rather, of the fact that many informants left such an area unidentified on their hand-drawn maps. 5.5 Chûbu Dialect In the previous discussion of home prefecture dialects, it was shown how informants in the Aichi survey had subdivided their prefecture. This attention to detail extends only, however, to the home prefecture and not to the entire home region; there was not a single occurrence of a label such as “Chûbu-ben” or “Chûbu-hôgen.” In fact, there were only four informants who even recognized and labeled this region as a single dialect area, and even these four used the label “Nagoya-ben” (the principal city within this area) to describe the entire region. On the composite map (Figure 13.6) one gets the impression that the Chûbu region has been marked off as a separate area, but in fact this is not the case. The map simply shows the boundaries for “Kantô-ben” (to the east) and “Kansaiben” (to the west). The greater portion of the Chûbu region between these is (with the exception of the small “Nagoya-ben” and “Mikawa-ben” areas marked on the map) actually an empty “trough” area as mentioned above. 5.6 Kansai Dialect On the informants’ hand-drawn maps, the Kansai dialect was the second most commonly drawn region (after Tôhoku) with fifty-seven of the seventy informants (81%) delineating a Kansai dialect region in some manner. Of these, 66% (forty-six informants) used the label “Kansai-ben.” A fair number of informants used the labels “Ôsaka-ben” and “Kyôtoben” to refer to smaller subdialect areas within the “Kansai-ben” area they had drawn (Figure 13.5). 5.7 Shikoku-Chûgoku Dialect There were only fifteen Aichi survey informants (21%) who drew the Shikoku dialect as an independent area and only sixteen (23%) who did so for the Chûgoku dialect. The reason for these low figures is that the majority of the informants (forty-three, or 61%) saw Shikoku and Chûgoku as being included within other larger dialect areas. In particular, I found the following response patterns:

190 – – – – –

IMAGES, PERCEPTIONS, AND ATTITUDES

Shikoku and Chûgoku as a single region — sixteen people Shikoku, Chûgoku, and Kansai as a single region — ten people Shikoku, Chûgoku, and Kyûshû as a single region — nine people Shikoku, Chûgoku, Kansai, and Kyûshû as a single region — seven people The Inland Sea sides of Shikoku Chûgoku as the “Setouchi-kotoba” [Inland Sea talk] — one person

5.8 Kyûshû Dialect Fifty-two informants (74%) drew Kyûshû as a separate dialect region. In determining whether or not to recognize Kyûshû as a separate dialect, I use fundamentally the same criteria outlined for Hokkaidô above. The first criterion (drawing a boundary line between the main island and Kyûshû) was met by forty-four people. The second (not drawing a boundary but giving dialect labels to the regions on both sides of the channel) was met by the remaining six. Several informants also included the subdialect areas “Hakata-ben” or “Kagoshima-ben” within their “Kyûshû-ben.” 5.9 Okinawa Dialect As in the case of Hokkaidô, the survey map given to the informants had a line drawn between the main islands and the inset map of Okinawa. For this reason, the same criteria were used for determining whether or not Okinawa had been designated as a separate dialect area. As shown in Figure 13.6, ten informants labeled this dialect as “Okinawaben,” whereas a total of seven people used labels such as “Ryûkyû-go,” “Ryûkyû-hôgen,” and “Ryûkyû-ben.”

6.

Perceptual Dialect Regions and Their Names

In the remainder of this chapter, I will examine data from the Kantô and Kansai surveys. There are two main differences between the analysis employed here and that used with the Aichi and Okayama surveys above. First, the Kantô and Kansai surveys were carried out after the Aichi and Okayama surveys, and I added an additional task to the questionnaires. I asked the informants not only to label the speech regions they drew but also to list specific characteristics of the speech of those regions. Second, I tabulated the values for the composite maps and produced the gradient maps by computer. As with the Aichi informants above, the hand-drawn maps from Kansai and Kanto respondents show a great deal of variation in the number of dialect regions drawn. The simplest maps divide Japan into only two main eastern and western dialects, and the most complex ones attempt to label each of the forty-seven prefectures individually. The great majority, however, in the data I will discuss here from the Kansai (Ôsaka) and Kantô (Tokyo) areas, drew and named an average of five or six regions. Of these, the first four

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

191

are clearly defined and widely recognized cultural regions, and the areas contained in the perceptual dialect maps often correspond roughly with these areas. The labels applied to perceptual dialect regions mostly consisted of a place name followed by the word “ben” (dialect) — as in “Ôsaka-ben.” The two major exceptions to this were “standard” and “zû-zû ben” (a denigrating name for dialects in Northern Honshu that don’t distinguish between /i/and /u/ in some environments). Let us take a look at the “standard” region. Although the geographical region of “standard Japanese” is not officially defined, it is commonly held to be based on the speech of a small subregion of Tokyo Prefecture. In this survey, however, I found that it covered a much wider area.

7.

“Standard” as a Perceived Region

The Kansai area (including the historical trading center of Ôsaka and the former capital of Kyôto) is well-known for its rivalry with Tokyo and its dialect; therefore, it was expected that Kansai informants would be somewhat reluctant to label the Tokyo region as “standard” and favor terms such as “Tokyo-ben” or “Kantô-ben.” This was not the case, however; many Kansai respondents did label the Tokyo area as “standard.” Two factors must be considered here to understand these results. One is the extent of the area labeled as “standard.” Although “standard Japanese” is supposed to be a regionless language variety, it is generally held to be “based on” the speech of educated speakers in certain parts of Tokyo. The area labeled “standard” on many of the maps, however, often included a large area surrounding Tokyo, in several cases the north island of Hokkaidô, and in isolated cases even the southwestern Chûgoku region containing Hiroshima. In other words, the Kansai respondents have granted the status of “standard” to a language variety that, although often centering on Tokyo, is not necessarily restricted to it (Figure 13.8).6 The second is the frequent usage by Kansai respondents of terms such as cold to describe this speech area. This reaffirms the idea that many Kansai speakers perceive the Tokyo standard as sounding aloof and impersonal (Inoue 1989). The fact that in Figure 13.8, geographically distinct regions are labeled in the same way (i.e., as “standard”) lends weight to the claim that (at least in the case of drawing a region for “standard”) respondents are to some extent using linguistic features as opposed to cultural-political regions as the bases for their dialect perceptions. It is true that respondents may have little or no firsthand knowledge of the dialects in other regions and thus are actually making their judgments using stereotypes of linguistic features, but this is quite a different matter from drawing perceptual regions simply on the basis of stereotypical ideas of regional boundaries in general.

192

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 13.8. Gradient map of the perceptual area of “Standard,” Kansai informants (n = 74)

8.

Characteristics of the Perceptual Dialect Regions

Respondents in the survey were asked, in addition to identifying and naming dialect areas, to list the “special characteristics” of each region. These were divided into (1) nonlinguistic and (2) linguistic features. The former included a. b. c.

attributes (cold, crude, rough, country), comprehensibility (incomprehensible, hard to understand), classification/comparison (standard, similar to Kansai dialect).

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

193

And the latter were d. e. f. g.

paralinguistic (mumbling, fast-talking, nasal, loud), phonetic (“zû-zû-ben”), prosodic (pitch accent of words, phrase intonation), lexical/morphemic (specific lexical or morphemic examples).

9.

Perceptual Dialect Maps of the Kansai Dialect

Let us take a look now at the perceptual dialect regions for the area labeled “Kansai-ben,” focusing on some similarities and differences between the maps from the Kantô and Kansai informants (the latter drawing the area of their home dialect). First, the inset maps in the upper-left-hand corners of Figures 13.9 (Kansai informants) and 13.10 (Kantô informants) show the area where 50% of the respondents’ “Kansai-ben” maps overlapped. Using this as a convenient cutoff point here makes it appear as if the Ôsaka (Kansai) and Tokyo (Kantô) respondents share similar concepts of the area in which “Kansai-ben” is spoken. We will see below that this is not the case, but let us consider some explanations for the apparent similarities in the 50% maps. One possibility is that the areas included in the 50% inset maps in Figures 13.9 and 13.10 basically cover the cultural-administrative region called Kansai. But another possibility is that they actually very closely approximate the area that many dialectologists consider to be the Kansai dialect region. This fact can be verified by comparing the inset maps in Figures 13.9 and 13.10 with the traditional “production” map in Figure 13.11. Discovering the reason for these similarities will require further study, but one can notice some interesting differences between the two maps as well. The gradient map of the “Kansai-ben” region drawn by Tokyo respondents (13.10) is less “focused” than the map drawn by the Ôsakans (13.9). In other words, the Kantô respondents are less sure than their Kansai counterparts how far the dialect extends (both to the northeast and the southwest), as shown by the much greater size of the transitional zones from the region of highest density to the least. Preston’s work has used the 50% mark as a convenient cutoff point in such maps, but we can see here that relying on this standard exclusively could obscure interesting facets of the data.

10. Subdialects of the Kansai Region One phenomenon that appears quite often in these Japanese data but that was not dealt with much in Preston’s work on perceptual dialects in the United States is that of subdialect regions within larger dialects. On the “production” level, Japanese regional dialects have clearer divisions than those in the United States, and dialect subregions are

194

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 13.9. Gradient map of the perceptual area of “Kansai Dialect,” Kansai informants (n = 195)

recognized on a more localized level. Thus, maps drawn by nonlinguist Japanese often display more highly detailed subdivisions within the larger perceptual dialect regions, resulting in dialect layering. After conducting some initial surveys using an all-Japan map, I decided to focus on smaller areas of Japan, beginning with the Kansai region, which I chose for two reasons. First, Kansai was, in the past, the cultural and political center of Japan, and there is a strong pride in the dialect(s) there. Second, within the subdialects of Kansai, there is a great deal of variation that is highly salient to the average speaker. Figure 13.12 is a compilation of respondents’ perceptions of the “Kawachi-ben” region. This dialect is classified by linguists as one of the three subdialects of Ôsaka (which is in turn a subdialect of Kansai). Although “Kawachi-ben” appeared quite often, the other two Ôsakan subdialects hardly ever appeared on respondents’ maps. Furthermore,

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

195

Figure 13.10. Gradient map of the perceptual area of “Kansai Dialect,” Kantô informants (n = 62)

there are some cases in which it was unclear whether “Kawachi-ben” was considered to be part of the Ôsaka dialect or to be a separate dialect. Clarifications of this will require some methodological adjustments and further study.

11. In Conclusion In this chapter, I examined the preparation, quantification, and analysis of the perceptual dialect maps drawn by ordinary speakers. In the perceptual dialect maps from the Aichi survey above, one sees cases of informants making divisions that seem to be based on the

196

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 13.11. Production dialect divisions (Misao Tôjô)

traditional divisions of “region,” rather than on purely linguistic factors. The inclusion of Ibaraki and Tochigi Prefectures as part of “Kantô-ben” is one such example. Although usually included as part of the Kantô region in business and government divisions, from the standpoint of the linguistic realities (what Preston refers to as “production dialectology”), these are actually much closer to the Tôhoku dialects immediately north of them (Umegaki 1964).

GEOGRAPHICAL PERCEPTIONS OF JAPANESE DIALECT REGIONS

197

Figure 13.12. Gradient map of the perceptual area of “Kawachi Dialect,” Kinki survey informants (n = 117)

However, there are many cases in which the perceptual dialect maps do not reflect cultural or political divisions. For example, it is commonly held that both Hokuriku and Chûbu are distinct cultural regions, but in the Aichi survey results, Hokuriku was neither labeled “Hokuriku-ben” nor partitioned off from Chûbu (surprising, considering that the informants’ home prefecture of Aichi is itself located in Chûbu). Also, it is customary (in business, politics, weather reports, etc.) to treat Chûgoku and Shikoku as separate regions, but informants in the Aichi survey did not usually differentiate them. Thus one can safely say that the divisions made by informants do not always necessarily match up with cultural and political divisions. It is by no means easy to determine whether it is actually dialect regions that the respondents are identifying or simply administrative regions. However, dialect boundaries themselves often coincide with other cultural boundaries, and it was because these cultural boundaries preexisted that they became administrative boundaries. Thus it may prove impossible in the long run to satisfactorily separate these two factors.

198

IMAGES, PERCEPTIONS, AND ATTITUDES

Notes 1. This chapter is a revision of Long (1990a) and Long (1993). I have done extensive rewriting of the text, but the content, including survey data, methodology, terminology, and bibliographical references, has not been altered. I would like to thank Hiroyuki Kanazawa, Hiroshi Tahara, Jirô Nagase, Fumio Inoue, and Munemasa Tokugawa, who lent their cooperation in the collection of the data used here, and also their students who kindly completed the questionnaires. I would also like to repeat here my original thank-you to Dennis Preston for allowing me to test out his techniques in Japan and for his endless patience in answering my questions. 2. Informants were students at Okayama University, Aichi University of Education, Tokyo University of Foreign Studies, Senshû University, Osaka University, Osaka Shôin Women’s College and Momoyama University. 3. The Japanese word akusento “accent” here means the pitch accent of words. If the informant had been referring to the concept, such as the English term accent, of “all the phonological characteristics of a dialect,” she would have used a word such as namari. 4. In Tôjô’s (1927: 42) dialect divisions, he proposes an “Inland Sea dialect,” but in his (1933: 34) version, this is divided into Chûgoku and Shikoku. 5. In Table 13.2, it is apparent that the Okayama informants drew their maps in more detail than did the Aichi informants. Aichi informants gave a total of 381 language variety labels on their maps, averaging 5.4 per informant. Compared with this, the 20 Okayama informants gave a total of 260 labels, averaging 13.0 per informant. 6. The computer-drawn perceptual dialect maps used here were produced with software designed by the author and Masami Fujita for the NEC 98 series personal computer, based on the concepts outlined in Preston (1989).

C 14 Mapping Nonlinguists’ Evaluations of Japanese Language Variation Daniel Long

1.

The Study of Perceptual Speech Regions in Japan1

There is a long history of research on Japanese regional speech variation and the attitudes toward it (Inoue 1977/8, 1978/9, 1988, 1996 [see Chapter 11, this volume]). A great deal of work has also been done on the perceptions (“consciousness” is a more direct translation of the term ishiki used in Japanese) of dialect speakers toward the speech of their own and neighboring areas (see Part II, this volume). In fact, as this work attests, it is no exaggeration to say that the concept of mapping speakers’ consciousness of dialect boundaries was, if not “born” in Japan, at least “raised” there. This is a somewhat ironic historical development in light of the fact that not only did Sibata and Grootaers reach the negative conclusion that dialect consciousness boundaries were not determined by linguistic factors but that this is precisely the point they set out to prove in the first place (Chapter 4, this volume). In this chapter, I present some of the results from my own work on the spatial perceptions of dialects, examining gradient maps of perceived dialect usage areas and the attitudinal and linguistic factors that underlie them. Using the basic methodology outlined by Preston (1988a), as adapted for application to the Japanese linguistic situation in Long 1990a (incorporated into Chapter 13, this volume), I introduce procedural innovations, including the mapping of “pleasantness” rankings and nonlinguistic characteristics. I also demonstrate the usefulness of statistical significance tests and present these results in map form. In the analysis, I compare the perceptual areas for the “most pleasant” speech as drawn by informants in eight survey regions and contrast these with areas indicated as “standard.” I then analyze maps of those speech regions associated with positive and negative characteristics, comparing them with those rated most pleasant. The data were collected in eight regions of Japan from university students who were native to the region and had no specialized education in dialectology. In the survey, we gave informants a blank map of Japan with only the prefectural boundaries marked and

200

IMAGES, PERCEPTIONS, AND ATTITUDES

asked them to perform the following four tasks: (1) draw lines to show the areas where people speak differently, (2) label what the speech in these areas is called, (3) rank the speech areas you drew in order of their pleasantness, (4) give specific characteristics for the speech of the different regions you drew. I refer to these tasks as (1) perceptual regions, (2) labels, (3) pleasantness, and (4) characteristics.

2.

Regions Perceived as Most Pleasant

2.1 Percentage Maps With regard to the pleasantness question, I took those regions that informants numbered as “1” (i.e., most pleasant) and made gradient maps of them in the same way that I did for speech variety region maps (see Long 1993, incorporated into Chapter 13, this volume). Unlike the mapping of speech varieties (such as standard or “Kansai-dialect”) in which there is some fundamental level of agreement among informants concerning the location (see for example Figures 13.9 and 13.10, this volume), mapping for most pleasant means a great deal of variation among informants even of the same region. This results in maps with a weak but widely distributed area. The data for most pleasant regions showed the following results: Kanto informants highlight Tokyo and the surrounding Kanto region, as well as the Kyoto area of the Kansai region (Figure 14.1). Kansai informants highlight Osaka and the surrounding Kansai area. (These and 130 other maps produced from this data are published in Long 1998.) The maps by informants from the other six regions are very similar in that they almost all exhibit tendencies to include at least parts of (1) Kanto, (2) Kansai, and (3) the home regions. The Aichi informants’ map is shown here as an example (Figure 14.2). 2.2 National Average Maps Above we have seen the three clear trends found in the gradient percentage maps for the eight regions. To better understand the overall pattern of the data, however, I constructed a single map from the eight percentage maps.2 Since the number of informants within each group varied widely, I chose not to produce a percentage map for all 811 informants but instead to combine the values for the maps and divide the sums by eight to produce a map that equally reflected the results of the eight informant groups. I will refer to this as the “national average” map (Figure 14.3). Since none of the values were greater than 34%, I reduced the increments for the shading of the map, which shows that the Kansai region is ranked highest on the pleasantness scale, with the Kanto region ranking second. Thus, we see in the data two different tendencies: an agreement to include Kanto and Kansai, and a difference because informants tended to include their home region. The first tendency, characteristic of all the regions, forms what Gould and White (1986) refer to as the “national surface” (42–52). If one really wants to see what contours make the most pleasant maps of the eight regions unique, one needs to extract these

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

201

Figure 14.1. Kanto informants’ “Most Pleasant”

general tendencies from the data in the maps. In other words, one needs to remove the interference effects of this national surface to see what really makes the most pleasant maps of each of the eight regions unique. I do this by, in effect, comparing all eight of the informant region maps with the average, standardizing the values for each of the 27,700 dots on the maps and producing maps to show these standardized scores.

202

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.2. Aichi informants’ “Most Pleasant”

2.3 Standardized Score Maps As with the data for Standard Japanese discussed in another paper (Long 1997), the percentage maps for most pleasant are necessary to give one a basic grasp of the data, and so these maps have their place. However, in some cases, a more finely tuned idea of the unique qualities of each region is needed, and this is best accomplished by standardizing the scores and showing them in map form. The tendency to identify the home region as most pleasant, which was seen in the percentage maps, is greatly amplified on the standardized score maps. With the national

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

203

Figure 14.3. Eight-region average “Most Pleasant”

surface scraped away, it can be seen that the informants of all eight regions rate their own region’s speech very highly. In fact, the difference between the percentage for each group and the average for all eight groups is greater than 69.6 and thus statistically significant. There is (with a few interesting exceptions) a very rough general trend for pleasantness values to decrease as distance from the survey location increases. This trend is not as clear and orderly as the wave patterns (in which there is a unidirectional and steady decrease in shading as one moves away from a dark core area) seen in the maps for language varieties such as the “Kansai-dialect” (see Figures 13.5 and 13.8, this volume) or the “Kawachi-dialect” (see Figure 13.12, this volume). The shading patterns here are

204

IMAGES, PERCEPTIONS, AND ATTITUDES

more sporadic than this, and the maps do not show an exclusively unidirectional decrease in values as one moves away from the core area. Nevertheless, the general tendency is for pleasantness to decrease as one moves farther away from the informants’ home region. The clearest exceptions to this tendency, found in the Hiroshima and Fukuoka data, will be discussed below. In general, the standardized score maps reveal more clearly the same trends already apparent in the percentage maps, but in some cases, they reveal tendencies that were obscured in the percentage maps. I now examine the eight standardized score maps individually, starting with the Kanto informants. Keep in mind that their percentage scores (Figure 14.1) showed a high ranking for Tokyo (60–79%), with a slow, circular spread into the surrounding regions. The remainder of the island of Honshû (as well as all of Hokkaido and Shikoku) is blanketed with a very weak (0–19%) “carpet” of pleasantness, the consistency of which is broken only briefly by a weak (20–39%) bump around the Kyoto area. When one compares the standardized score map (Figure 14.4), there is a more focused picture of the Kanto informants’ unique rankings. First, although their percentage values for the Tohoku region were very weak, their standardized values for this region are greater than 69.6 (as indicated by the dark shading). Standardized values exceeding 69.6 are significantly higher than the national average at the p < .05 confidence level. In other words, the most darkly shaded areas on this and the next seven maps are statistically significant. Second, the Kanto and Chubu regions, with moderate percentage values (40–59), are still significantly higher than the national average for these areas. In addition, the slight “hill” (20–39) that was Kyoto on the percentage map (Figure 14.1), is now buried under an even higher hill that is the eight-region average. In other words, Kanto does not have as pleasant an image of Kyoto speech as the percentage map would have led us to believe (at least when viewed in the light of data from other regions).3 The Aichi informants’ percentage map (Figure 14.2) is very similar to the other percentage maps. That is, their most pleasant areas include some portion of the Kanto region, their own home region, and some portion of the Kansai region. Standardizing the scores, however, greatly clarifies the Aichi informants’ results. In particular, on the percentage map, their home region had no higher value than that of Kanto or Kansai; on the standardized score map (Figure 14.5), their home region is significantly higher than the average. The Gifu informants’ percentage map includes small parts of the Kanto region, and a broad area covering the home area and Kansai, with a somewhat stronger showing for Kyoto. The standardized map (Figure 14.6), however, shows a different picture. Although Kyoto remains comparatively higher than surrounding areas, the home region of Gifu itself is the only area on the map to show a rating significantly higher than the national average. Kanazawa informants also typically identify the three regions. On the percentage map, they include the Tokyo area, their own area (including the Hokuriku region surrounding it), and the Kansai region (particularly Kyoto and Osaka). Here again, however, even the highest percentages (for the home region) are in the 40% to 59% rank and are not very impressive. Looking at the standardized score map for Kanazawa (Figure 14.7), however,

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

205

Figure 14.4. Kanto informants’ “Most Pleasant” (standardized scores)

one sees that the home region is rated significantly higher than the national average. The percentage map of the Kansai informants differs greatly from those of the other seven groups; it is the only one that has a core area in the 80% to 100% range. This core is centered in Osaka and spreads out into the surrounding Kansai area. This impressive effect is not diminished when one looks at the standardized scores (Figure 14.8) and finds that the entire Kansai area is rated significantly higher than the national average. In the percentage map, it appeared that there were no more people who rated Chûgoku (the portion of Honshû island southwest of Kansai) as pleasant than there were for the Kanto area.

206

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.5. Aichi informants’ “Most Pleasant” (standardized scores)

Figure 14.8 shows, however, that this is not the whole story. When compared with the eight-region average, Kansai informants have a somewhat more pleasant image of Chûgoku and the island of Shikoku than they do of the Tokyo area. Furthermore, these relatively high pleasantness values for the rest of the Chûgoku region do not apply to Hiroshima. In light of the general trend for pleasantness to decrease gradually with distance, this Hiroshima sinkhole is somewhat intriguing.4 On the percentage map, the Hiroshima respondents focused on the home region, more weakly on the entire Kansai region, and very weakly throughout the Chûbu and Kanto regions.

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

207

Figure 14.6. Gifu informants’ “Most Pleasant” (standardized scores)

This tendency is amplified on the standardized scores map (Figure 14.9), but within the Kanto region, the Tokyo area stands out as being lower than the eight-region average. As mentioned earlier, all eight of the maps exhibit a rough but noticeable tendency for the values to decrease gradually with distance from the core area (in this case, the informants’ home regions). Hiroshima is one of the two conspicuous exceptions to this tendency. Going from Hiroshima to Fukuoka, there is an abrupt drop from an area whose pleasantness rating is so high as to be statistically significant, to an area that not one single informant designated as most pleasant. The percentage map for Fukuoka rates the home region with a moderate value

208

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.7. Kanazawa informants’ “Most Pleasant” (standardized scores)

(40–59%). Three areas — the remainder of Kyushu, the central Kansai region, and the central Kanto region — are somewhat more weakly rated. The standardized score map reveals the home region to be significantly higher than the average, with the Kansai and Kanto regions placing only in the average range (Figure 14.10). On this map it can be seen that the apparent animosity on the part of Hiroshima informants toward Kyushu is reciprocated by the Fukuoka informants. Going from Fukuoka across the channel to the main island, there is an extreme drop, since not one of the Fukuoka informants labeled the region including Hiroshima as most pleasant.

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

209

Figure 14.8. Kansai informants’ “Most Pleasant” (standardized scores)

Finally, on the percentage map for Kagoshima informants, the home region, along with small parts of the Kanto and Kansai regions, received the highest ratings (although these are by no means strong). The standardized score map reveals that the values for Kansai and Kanto are indeed only in the average range, but the home region is significantly higher than the eight-region average (Figure 14.11)

210

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.9. Hiroshima informants’ “Most Pleasant” (standardized scores)

2.4 Perceptual Maps for Most Pleasant and Standard Here, I will contrast the results from the present study with those of a related study, looking briefly at the relationship between those areas that the informants rated as most pleasant and those they designated as standard. The areas perceived as Standard Japanese are discussed in greater detail in Long (1997), but it is enough to say here that three tendencies emerged among the eight informant groups: (1) a primary tendency to indicate the area centered around Tokyo and radiating outward into Kanto and Chûbu; (2) a

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

211

Figure 14.10. Fukuoka informants’ “Most Pleasant” (standardized scores)

tendency (except among Kanto informants) to avoid including one’s own home region in the standard category; (3) a less general but nonetheless common tendency to indicate the northern island of Hokkaidô as standard. There were large differences between the areas labeled most pleasant and Standard Japanese among these informants. The results showed clearly that although the informants from all regions surveyed (excepting Kanto) recognize that their own speech areas are not standard, they nonetheless perceive the speech of their own regions as the most pleasantsounding variety of Japanese.

212

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.11. Kagoshima informants’ “Most Pleasant” (standardized scores)

By comparing the results of these two studies, I found that informants from all the regions saw Standard Japanese as a language variety located geographically in the Tokyo area. (It is important to keep in mind here that just because the informants think that standard is a language variety spoken in Tokyo, this in no way implies that they think all speakers in Tokyo always use Standard Japanese.) In contrast, for each of the eight informant groups, the most pleasant speech was firmly located in that group’s home region. Thus, one can conclude that the concepts of standard and pleasant are not only

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

213

independent of one another, but they stand (at least for speakers in the western regions of Japan where these surveys were concentrated) diametrically opposed to one another.

3.

Mapping Characteristics of Speech Varieties

3.1 Categorization of Characteristics In the survey, in addition to the pleasantness rankings analyzed above, the informants were also asked to list specific characteristics that were associated with the speech varieties they indicated on the maps. The responses to this open-ended question took the form of both linguistic and nonlinguistic characteristics. I proposed a categorization of such characteristics in Long 1993 (incorporated into Chapter 13, this volume): 1.

2.

Nonlinguistic characteristics a. Attributes (cold, crude, rough, country) b. Comprehensibility (incomprehensible, hard to understand) c. Classification/comparison (standard, similar to Kansai dialect) Linguistic characteristics: d. Paralinguistic (mumbling, fast-talking, nasal, loud) e. Phonetic (excluding prosodic characteristics; “zu-zu ben”) f. Prosodic (pitch accent of words, intonation of utterances) g. Lexical/morphemic (specific lexical or morphemic examples)

For this chapter, I took all characteristics from categories a, b, and d that had a positive tone and recategorized them as such. I then did the same for negative characteristics. There were, of course, some that could not be characterized as either positive or negative, and these will not be dealt with here. Listed below in Table 14.1 are the numbers of informants who gave the various types of answers. The total number of actual characteristics was much greater than this (total of 2,358) because one informant often gave responses for multiple speech regions. Table 14.1. Number of informants who gave ‘characteristics’ for speech regions (a,b,d) positive characteristic (a,b,d) negative characteristic (e) phonetic (f) prosodic (g) specific lexical or morphemic examples total no. of people giving characteristics

104 102 256 060 403 600

Note: some informants are included in more than one group

214

IMAGES, PERCEPTIONS, AND ATTITUDES

3.2 Differences between the Kanto and Kansai Characteristics Maps From the informants’ hand-drawn maps, I took those areas labeled with positive characteristics and made a gradient map of them, just as I did for the areas that they labeled as Standard Japanese or “Kansai dialect” (see Chapter 13, this volume). In the same way, I constructed a gradient map of those areas tagged with negative characteristics. I first examine the maps drawn by Kansai informants, who have identified their home region with positive characteristics (Figure 14.12). A look at the negative characteristics map, however, reveals that the Kansai area (particularly the Osaka-Nara-Kyoto area, where the majority of the informants are from) is strongly identified with negative characteristics as well (Figure 14.13). Although these results at first seemed paradoxical, speakers logically have more knowledge of and experience with nearby dialects and, one could hypothesize, are more likely to supply more detailed information about them. This hypothesis is short-lived as a general one, however, because in Figures 14.14 and 14.15 one sees that the overwhelming number of Kanto informants have not associated their home dialect with any positive or negative characteristics. It would appear that Kanto people do not see the speech variety of their region as being either good or bad; they perceive their dialect as being mushoku tômei “colorless and transparent,” like water. Perplexingly, the areas with which they associate positive and negative characteristics are those far from their home region — those with which one would expect them to have little contact, namely the Kansai region. A closer look at the two Kanto maps reveals that although Kanto informants locate both positive and negative speech areas in Kansai, they also draw a clear distinction between the positively viewed Kyoto and the negatively viewed Osaka. Although the cities of Osaka and Kyoto are located only forty-five minutes from one another (their prefectures share a border) and they use dialects that are linguistically very similar (see for example, Shibatani 1990, who repeatedly refers to the “Kyoto-Osaka” dialect), sociopsychological factors associated with the two dialects not only differ but often stand diametrically opposed. Inoue (1988: 21) has found, in his research on dialect image, similar tendencies for non-Kansai people, who hold opposing views of Kyoto and Osaka. Moreover, Long (1990b: 23) shows that in-migrants from other regions living in Kansai have more positive attitudes toward Kyoto than toward Osaka.5 Although my initial inclination was to combine the positive and negative values, subtracting the latter from the former, I elected not to do this after seeing the above results. I made this decision because combining the maps would actually disguise these contradictory perceptions of a single area. Equally high negative and positive values for the same region would cancel each other out giving a value of zero, and these highly controversial areas would then be indistinguishable from those that had a value of zero as the result of informants’ apathy toward or ignorance of them.

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

215

Figure 14.12. Kansai informants’ “Positive Characteristics”

4.

Statistical Tests

Here, as earlier in this chapter, conclusions drawn from visual comparisons can be tested statistically by comparing the differences between the values of the informant groups, using here the so-called proportion test (Davis 1990). I first performed this test on the values for areas associated with positive characteristics by the Kanto and Kansai informants. Figure 14.16 indicates the areas where there is a statistically significant difference between the percentages for the two groups. Kansai people have an even stronger positive

216

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.13. Kansai informants’ “Negative Characteristics”

image of the Kanto area than the Kanto informants themselves. However, this strong positive image does not extend to Tokyo, which is conspicuously excluded from this area. Figure 14.17 shows differences in the two groups’ perceptions of negative characteristics. Somewhat surprisingly, the Kansai informants rate the Kyoto area significantly worse than do the Kanto informants. This may be due to the large number of people from Osaka (which has a long-standing cultural and linguistic rivalry with Kyoto) among the Kansai informants. Conversely, there is an area in which the number of people giving negative characteristics is significantly higher among the Kanto informants. This area is located in Northern

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

217

Figure 14.14. Kanto informants’ “Positive Characteristics”

Kanto, a region that conjures up connotations of monotone speech (it has a so-called accentless dialect, meaning that it does not use pitch accent to differentiate lexemes) and is commonly held in derision by Tokyo speakers.6 Kansai speakers, on the other hand, do not appear to subscribe to the negative images that their Kanto counterparts attach to this dialect. 4.1 Differences between Pleasantness and Characteristics As seen thus far, maps based on rankings for pleasantness and those based on the freeresponse characteristics gave disparate results. These results seem to reveal that although

218

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.15. Kanto informants’ “Negative Characteristics”

speakers in Kanto regard their home dialect as the most pleasant, when it comes to supplying concrete traits, they appear to have nothing to say, good or bad, about their dialect. Rather the evaluative characteristics that they associate with speech varieties tend to relate to the dialects of the Kansai region. Conversely, speakers in the Kansai region (who also rate their home dialects as most pleasant) seem to have no problem coming up with both good and bad things to say about their own dialect. In fact, when they offer evaluative characteristics of speech varieties, positive or negative, they tend to center on dialects in their own region.

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

219

Figure 14.16. Kanto and Kansai informants’ “Positive Characteristics” (t test)

It is true that “Kansai-ben” (Kansai dialect) is the second most frequently drawn dialect region. In this study, over 76% (622 out of 811 informants) chose to represent this region on their hand-drawn maps. In fact, the Kansai dialect was second only to “Tohokuben” (81%). These top two speech varieties had a great lead over the other dialects, with the third most common response ranking below 50%. Less than a quarter of all informants indicated the Kanto dialect, and only one-fifth drew a region for the Tokyo dialect. Thus, it would appear that Kansai-ben is highly salient among the dialects, not only because a large number of speakers in various regions are conscious of it as the speech variety of a specific geographical location but also because they readily associate it with specific

220

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 14.17. Kanto and Kansai informants’ “Negative Characteristics” (t test)

characteristics (both good and bad).7 This leaves us with an intriguing difference between our results and those that Preston obtained in the United States. In his studies, the speech region that Midwest informants identified most often was “South.” For the most pleasant dialect, they designated their own area but rated the south as the least pleasant speech region (but see Chapter 22, this volume). This would seem to suggest that those speech regions most salient for speakers were those that they found unpleasant rather than those that they thought pleasant. But with our informants, “Kansai-ben” was an extremely salient dialect region (the second most commonly drawn speech region), although it was the number one

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

221

choice of our informants for pleasantness. From the results Preston obtained in the United States, one would have expected the informants of each area to choose their own home region as most pleasant, and this tendency was observed in the Japanese data after the interference of the national surface was removed. Nonetheless, one cannot completely ignore the existence of this national surface, and it will be interesting to see if such a phenomenon exists in U. S. dialects. Social psychologists have emphasized the difference in (1) the degree to which people’s attitudes are positive or negative, and (2) the strength of conviction that people attach to these attitudes (conceptualized as “latitudes of acceptance”).8 We may add to these another aspect of language attitudes, at least as they apply to the type of research outlined here: (3) the extent to which people are able to supply specific examples. 5.

Comparisons of Different Map Types

The methodology used here, based on the quantitative analysis of the perceptual language regions of multiple informants, differs somewhat from that of previous studies in its methodology and its application of statistical techniques. 5.1 Differences with the Grootaers-Sibata Dialect Boundary Consciousness Maps The present study (and the Prestonian paradigm in general) differs in several ways from the Sibata-Grootaers study outlined in Chapter 4 of this volume. Aside from the great differences in the research objectives of the two studies, the present work makes use of advances in computer technology made since the Itoigawa research (in the late 1950s and early 1960s), which facilitate the quantification of even greater amounts of data than were handled in that survey. One of the major methodological differences in the two studies is that SibataGrootaers were essentially looking for the perceived boundaries of dialects, whereas the paradigm used here focuses on finding the perceived areas of dialects. By shifting the emphasis from the identification of boundaries to the identification of areas (an innovation introduced in Preston 1981), I have avoided many of the procedural difficulties involved with attempting to stack and count boundaries. In quantifying the degree of “intensity” (overlap) of the perceived regions, I concentrate not on the outer extensions of the area but on its core and on the waves of decreasing intensity that radiate outward from it. This paradigmatic shift also allowed me to identify dialect areas with multiple cores, a phenomenon seen in, for example, the Standard Japanese map of the Kansai informants (Long 1997) in which the core area was divided between Hokkaido and Kanto. Moreover, the paradigm used here facilitates the quantitative analysis not only of areas regarded as “different” (from one’s own dialect) or as the “same” but also of the areas associated with aspects such as pleasantness (as analyzed in this chapter) or with differences in specific linguistic quality, such as pitch accent systems. Finally, the Sibata-Grootaers research differs fundamentally in the points that it set

222

IMAGES, PERCEPTIONS, AND ATTITUDES

out to investigate (prove or disprove) in the first place. Due to the similarities in concepts in the two studies (language perceptions, regional dialects, multiple informants’ data, etc.), it is easy to forget that Sibata and Grootaers did not set out to investigate speakers’ perceptions of dialects. Their objective was to determine whether or not people’s perceptions of dialect boundaries matched “real” dialect boundaries, and the cumulative conclusion of several studies was “not necessarily.” The present work, on the other hand, took their conclusion as its starting point, and asked, “What factors, linguistic or sociopsychological, are related to speakers’ geographical perceptions of dialects?” (This is basically the approach adopted in Preston 1986 and 1988c as well.) I contend that, in those cases where the two do not match up, one should be concerned not only with the question of why attitudes do not match the linguistic reality but also with why the boundaries drawn by linguists do not match the perceptions of great numbers of average speakers.9 5.2 Differences with the Inoue Dialect Image Maps The results reported here for most pleasant correlated quite well with those from a series of surveys that Inoue has conducted on the regional differences in “dialect images,” even though his methodology (Chapters 11 and 12, this volume) is fundamentally different from the one employed here. Inoue applied a multivariate analysis to the response patterns of informants and determined that various adjectives associated with dialects group into “emotional” traits and “intellectual” traits. These response patterns seem to have implications beyond just the Japanese situation because Inoue (1996; see Chapter 12, this volume) found very similar results for English dialects as well, where the adjectives formed themselves into groups of “urbanness” and “standardness.” In this survey, most of the eight regions showed an affinity for the Kansai dialect, as can be seen in the raw percentage values for most pleasant (Figure 14.1). Similarly, in Inoue’s (1988: 20–21; 1991: 1509–11) research, some (though not all) of the dialects of western Japan received high ratings on the emotional scale, even from Tokyo informants. Thus, the areas reported here as most pleasant are similar to Inoue’s emotional traits. Another similarity between the results of the two studies is that concerning Tokyo and the Kanto region, an area that is high in both Inoue’s intellectual traits and in my more recent determination of the area seen as standard (mentioned briefly above and discussed in detail in Long 1997). Inoue (1988: 21) found that although there were regional differences, there was an overall tendency for informants from various regions of Japan to rate dialects near Tokyo high on what he terms the intellectual scale. As a matter of fact, he regards the factor “close to Standard Japanese” as most representative of the sixteen intellectual factors (Inoue 1991: 1509). Returning to the comparison of my pleasantness and Inoue’s emotional factors, there are interesting dissimilarities between the results of the two studies in this domain as well. Inoue’s findings showed that only Kyoto was associated strongly with emotional factors; my survey revealed a positive evaluation (albeit slightly weaker) of the pleasantness of Kanto speech. Even more significantly, my study revealed a clear picture of speakers who highly evaluate the pleasantness of their home region dialects as well (an effect revealed

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

223

when the national surface, consisting of these general affinities for the Kansai and Kanto dialects, was removed from each region’s map). Of course, it is entirely plausible that the differences between the two studies are the result of language attitude changes over the past two decades. Indeed, there is empirical data to indicate that language attitudes toward nonstandard varieties have become steadily more favorable during the post-war period,10 and the differences in the results found in the two surveys may reflect the logical extension of these attitudinal changes. At any rate, the varying results obtained in my study for areas associated with pleasantness and for “positive characteristics” indicate that the problem is more complex than anticipated and that a more complete understanding of it requires further study. 5.3 Differences with Preston’s Methodologies Finally, although the current research builds on the early methodologies of Preston (1988a), there are significant differences in the methods used here and those employed in his studies. In Preston’s work, a predetermined number of geographical entities (e.g., the fifty states of the United States, Preston 1985, 1988d) were ranked by the informants for correctness and pleasantness. The present study also analyzed responses regarding standard speech and pleasant speech, but the decision whether or not to indicate a region for standard was left to the informants, and they gave pleasantness rankings only to regions that they had previously identified. Preston’s methodology has, of course, some advantages over that employed here. He obtains not just one area labeled as standard but also informants’ perceptions of the degree of correctness (in many ways comparable to the concept of standard in my survey) of all regions. It may be that Preston’s method is better suited to the United States, where the concept of standard has much less relevance to both linguists and nonlinguists than it does in Japan. My methodology seems quite appropriate for Japan, however, where the concept of standard, and its geographic location are (along with the topic of the pleasantness of the speech of various regions) very hot topics of discussion and debate even among average (i.e., nonlinguist) language users. Another difference between this research and that of Preston is his extensive use of multivariate analysis (as exemplified by the work of his colleagues in this volume). The application of this exploratory statistical technique to the Japanese situation holds a great potential for future research. Finally, although the methodology in this chapter is based directly on that of Preston, I have outlined several methodological innovations that, it is hoped, may be incorporated into the paradigm that has come to be known as “perceptual dialectology.” These include the mapping of traits such as pleasant according to the actual regions with which the trait was associated rather than according to pre-established administrative districts. Furthermore, the innovation of maps based on descriptive statistics (means, standard deviations) and inferential statistics (standardized values, significant differences) has made it possible to identify and verify differences in the geographical perceptions of two informant groups based on more than visual impressions.

224 6.

IMAGES, PERCEPTIONS, AND ATTITUDES

In Summary

In this chapter, I have used maps of speech areas designated as most pleasant as well as those associated with positive and negative characteristics to outline an alternative methodological approach to the mapping of nonlinguists’ evaluations of language variation. These findings may be summarized as follows: 1.

2.

3.

When identifying the most pleasant speech region, the overall tendencies were (roughly, in order of frequency) to indicate (a) the Kansai dialect region, (b) the Kanto dialect region, and (c) the home dialect region. The third tendency was most clearly evident when the values were standardized to highlight the uniqueness of each informant group’s map. In all eight regions (including those for Kanto and Kansai informants, whose home regions had high values in general), the values for the home region were higher than the eight-region average to a statistically significant degree. The results for pleasant were strikingly different from those found for standard in a previous study. Standard Japanese was associated with Kanto and Hokkaido and showed a tendency of the respondents to avoid inclusion of the home dialect, whereas pleasant ratings revealed a strong tendency to include (along with Kanto) the Kansai and home region. For positive characteristics, Kansai informants predictably indicated their own region, but Kanto informants unexpectedly indicated a strong affinity for the Kyoto area in Kansai. For negative characteristics, Kansai informants somewhat surprisingly indicated the central part of their own region, including areas that appear on their positive map as well. Kanto informants’ negative characteristics map indicated a broader area of western Japan but was similar to Kansai informants’ maps in its inclusion of Osaka in the most negative region. In statistical significance tests, it was found that more Kansai informants had positive things to say about Osaka and more negative things to say about Kyoto than their Kanto counterparts.

In the future, one of the most intriguing possibilities for research seems to be the use of perceptual dialectology methodologies to investigate aspects of speakers’ senses of linguistic identity. In particular, this would involve analyzing how extensively speakers perceive the area of “my dialect” to extend, as well as the regional patterns in language variety labels they use to refer to speech regions. After having pursued, in my own research up to now, topics such as the areal perceptions of standard and speakers’ evaluations of speech varieties, applying the methodologies of perceptual dialectology to the examination of speakers’ linguistic identities will, in many ways, bring the focus of perceptual dialectology research full circle, returning to the question that inspired its inception in Japan and Western Europe almost four decades ago. Notes 1. This research was partially supported by Grants-in-Aid for Scientific Research of the Ministry of Education, Science and Culture of Japan (1994, 1996) and by a Scientific Research Grant from the Japan

NONLINGUISTS’ EVALUATIONS OF JAPANESE LANGUAGE VARIATION

225

Science Society (1992). An earlier version of this chapter was presented at the AILA ’96 (the 11th World Congress of Applied Linguistics) in Jyväskylä, Finland on August 8, 1996. I would like to express my appreciation to those who offered their advice and opinions at the presentation. I would also like to thank Fumio Inoue, Jirô Nagase, Steve Horn, Hiroshi Tahara, Takashi Ushijima, Kenjirô Matsuda, Hitoshi Nikaidô, and Ichirô Ota for their assistance in the collection of the data presented here, as well as the informants themselves. 2. On this map, no area was rated over 34%, so the shadings have been readjusted accordingly. Thus, it is necessary to keep in mind that even the heavily shaded areas here correspond to the second shading level on the maps discussed earlier. 3. The results for the Kanto informants here raise the question of what information the informants are basing their responses about dialect image on? In other words, have the informants had actual contact with speakers from other dialects through friends, relatives, travel, and so on, or are these images merely based on factors such as media images, “common sense,” and popular folklore? This question is beyond the scope of the discussion here, but the reader is referred to Oki (1986) for a discussion of the formation processes of dialect images. Preston (1989: 127) also mentions the importance of such factors (“visited” vs. “not visited”). 4. Inoue’s studies differ greatly from the results reported here in that they focus on the images speakers hold of their own dialects. Nonetheless, it is interesting that just as my Kansai informants have a less positive impression of the Hiroshima dialect than those of neighboring regions, Inoue’s Hiroshima informants also have a less favorable image (on the “emotional image factors”) of their own dialect than informants in neighboring prefectures have of their home dialects (see Figure 11.1, this volume). If Inoue’s results for Hiroshima natives are not negative enough, Tomosada (1995) presents data from nonnative residents in Hiroshima whose image of the dialect is even more negative than their native-Hiroshiman counterparts. Similar negative attitudes by nonnative residents of Hiroshima toward the dialect are reported in Kandori and Takanaga (1988). 5. The study by Inoue (1988: 21) mentioned earlier, as well as a more detailed Japanese analysis of this data (Inoue 1992: 32), found that informants from Osaka and Kyoto as well as Kanto gave more positive ratings to the speech of Kyoto than they did to that of Osaka. 6. Inoue (Figure 11.1, this volume) shows that the dialect of Northern Kanto is viewed negatively (on both the intellectual and emotional scales) even by the speakers of the dialect themselves. 7. One thing we must keep in mind is that some of the information supplied by informants that might have been interpreted as positive or negative has not been included in the map tabulations here. Two major examples of this are the labels standard and “zu-zu ben,” both of which are analyzed in this research as indicators of “language varieties” and not as characteristics. It would have been equally logical to have included the label “standard” itself as a positive characteristic. (After all, as stated earlier, Inoue [1991: 1509] uses “close to the Standard” as his representative of positive “intellectual factors.”) If we had categorized standard in this way, the Kanto region would have been more predominant in our positive maps. Likewise, the label “zu-zu ben,” which shows up frequently in the data, could have been treated as a negative characteristic. This term is an unflattering one, describing the phonetic traits of dialects in the Tohoku region. The zu-zu accent is often the target of jokes and is considered one of the main factors in the “dialect (inferiority) complex” of Tohoku speakers (Inoue 1973). Its inclusion as a negative characteristic would definitely have given the northern end of Honshû Island a stronger showing on the negative maps. 8. Edwards (1979) refers to this concept of the strength of a respondent’s conviction as a “confidence rating,” expressing his results in terms of respondents’ being “less certain,” “more confident,” and showing a “willingness to commit” or “greater caution.” Of the same concept Giles (1979) says, “Furthermore, the relative ‘widths’ of the latitudes of acceptance, rejection and non-commitment of these evaluative items are thought to reflect the degree to which the individual is involved with the person or issue at hand” (63). 9. The notion that perceptual dialectology should ask not only why nonlinguists’ maps do not match production maps but also why the maps of professionals are so different from the consciousness of the masses is one implicit throughout much of Preston’s work. It is never voiced so explicitly, however, as in a review of Preston’s work by Butters (1995).

226

IMAGES, PERCEPTIONS, AND ATTITUDES

10. For example, in 1979 the social survey arm of the NHK public broadcast network conducted a nationwide random-sampling survey of over 2,300 people that showed that the younger the respondents the more positive their attitudes toward dialects were (NHK 1980a: 8; 1980b: 199).

C 15 The Perception of Post-Unification German Regional Speech Jennifer Dailey-O’Cain

1.

Background

Although Westerners and Easterners alike danced on the Berlin Wall the day it fell, the euphoria of the autumn of 1989 has long since subsided. Factory closings, high unemployment in both parts of Germany, the wage gap between Eastern and Western workers coupled with efforts to fill it — all of these things have had a powerful effect on the people. Many Westerners and Easterners have each come to see the other group as something foreign, not very much like themselves after all. In a study undertaken by the Friedrich-Ebert-Stiftung in connection with the polis and Sinus institutes for social research in 1991, 65% of Easterners and 68% of Westerners stated that they did not trust the other group of Germans (Becker, Becker, and Ruhland 1992: 24–27). In a more recent empirical study undertaken in eastern Germany by the Emnid Institut and reported in the newsweekly der Spiegel, 67% of those surveyed stated that they thought the proverbial “wall in the mind” was growing rather than shrinking (Emnid-Institut 1995). Based on these indications and our knowledge about the connection between language and social identity (Pool 1979; Chambers 1995: 250–53), it seems realistic to hypothesize that attitudes toward language variation will be strongly affected by the complex sociopolitical situation in Germany today. This study uses a combination of perceptual dialectology (Preston 1989) and qualitative methodologies to obtain data demonstrating these effects and analyzes these data in the hope of better understanding how geographic and sociopolitical factors can affect language attitudes.

2.

Methodology

The data for this study were collected during the summer of 1995, five years after the official unification of Germany. I traveled to forty-four different towns all over the country, north and south, east and west, interviewing a total of 218 informants. The

228

IMAGES, PERCEPTIONS, AND ATTITUDES

perceptual dialectology data was obtained by asking these informants what they thought of the language varieties spoken in thirty-four different regions, including the cities of Hamburg, Cologne, Frankfurt, Munich, Hanover, and Berlin, without a recorded stimulus. First I asked about how “correct” each informant believed each variety to be, on a scale from one (most correct) to six (least correct). Then I asked them to evaluate each region’s variety in terms of “pleasantness.” Although I also asked about “similarity,” I will not include these data in the paper. Then informants received a blank map of Germany depicting no political boundaries but including cities so that they could better orient themselves. On this map, they were asked to outline dialect boundaries where they believed them to be and to write on the map the names they use for the varieties they outlined. After the formal part of the study was over, I also recorded informants’ conversations about their language attitudes to obtain qualitative data with which to supplement the findings of the quantitative data (see Figure 15.1.)

Figure 15.1. Map of thirty-six German regions rated for ‘correctness’ and ‘pleasantness’

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

3.

229

Data and Analysis

I first examine the informants’ attitudes toward the “correctness” and “pleasantness” of the language of the thirty-four regions I inquired about. I do this by taking the mean rating of each region for Northwesterners, Northeasterners, Central Westerners, and Central Easterners, and displaying these means on maps.1 I then discuss the difference between Westerners’ and Easterners’ evaluation of the correctness and pleasantness of western and eastern varieties generally (using a mean of all western and a mean of all eastern varieties). 3.1 Correctness Figure 15.2 shows the evaluations of correctness according to Northwesterners. The lighter shadings correspond to the more positive ratings. It is immediately noticeable that Northwesterners perceive other northwestern varieties as generally more correct than they do eastern or southern varieties. There is a very obvious boundary that runs precisely along the former East-West border from north to south and a similar boundary that runs from west to east in the middle of central western Germany. Other immediately noticeable facts are that the eastern part of the southwest is evaluated as much less correct than the western part and that the eastern part of the central east is also evaluated as much less correct than the western part. Another noticeable characteristic is seen in the urban-rural distinctions made: Hanover is evaluated substantially more positively than other northwestern varieties, whereas the cities of Berlin and Cologne are evaluated less positively than surrounding varieties. The equivalent map for the Northeasterners (Figure 15.3) shows some marked differences. Although the southern part of Germany looks remarkably similar, Northeasterners do not agree with Northwesterners on how to evaluate the northern part of Germany. The central eastern varieties are also evaluated quite negatively, even more negatively than the Northwesterners rated them. However, the northeastern varieties are seen as just as correct as the northwestern varieties. Therefore, the former East-West border is only evident in the central part of Germany on this map, not in the northern part. Also, although Northwesterners and Northeasterners agree that people in the cities of Berlin and Cologne speak less correctly than people from other regions around them, they do not see the variety spoken in Hanover as significantly more correct than other regions around it. Finally, the ratings for the particular regions in central eastern Germany vary widely for the Northeasterners, but they are more uniform for the Northwesterners. This probably reflects local eastern tensions or actual dialect boundaries that the Northwesterners are unaware of, because there was little contact between Westerners and Easterners before 1989. Some of these differences between the perceptions of western Germans and the perceptions of eastern Germans are even more obvious when comparing the maps of the correctness ratings for the Central Westerners and the Central Easterners. The Central Westerners’ map in Figure 15.4 exhibits a pronounced “correctness border” between the

230

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.2. Northwesterners’ mean ‘correctness’ ratings for each individual region

northeast and the northwest, similar to the perceptions of the Northwesterners. The south and the central east are, as in previous maps, both regarded as less correct than other regions, and for Central Easterners, this extends to the western part of the south and not just the eastern part. On the other hand, the equivalent map for the Central Easterners in Figure 15.5 looks much more like that of the Northeasterners. The entire north is evaluated as quite correct, and the northwest is not seen as any different from the northeast. Even in the central part of Germany the former border is not as noticeable here, although it is the case that Central Easterners are more critical of certain highly stigmatized central eastern varieties than any of the other groups. The attitudes of the Southwesterners, as shown in Figure 15.6, are similar to the attitudes of the Central Westerners, apart from two noticeable differences: the Southwesterners are less critical of the south than the Central Westerners, and the Southwesterners see the strongest urban-rural distinctions. These maps suggest that Westerners, in general, think eastern varieties are less correct than western varieties but that Easterners don’t perceive much of a difference. In fact, if we take a mean of all correctness evaluations of western varieties and another mean of the equivalent evaluations of eastern varieties and compare them, we find exactly that to be the case. Westerners and Easterners agree precisely about the correctness of western varieties — both groups have a mean of 3.5 for all western varieties. However, although

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

231

Figure 15.3. Northeasterners’ mean ‘correctness’ ratings for each individual region

Westerners view eastern varieties as much less correct than western varieties, with a mean of 4.0, Easterners view them as exactly the same in terms of correctness. The difference between Westerners’ perceptions of western and eastern varieties is significant at .0005 in a paired-samples t test, whereas the difference between Easterners’ perceptions of the same two varieties is not significant. This evidence is also supported by the qualitative data. When asked about the language of eastern Germany, a thirty-two-year-old man from northwestern Germany compared it to other differences he perceived between the west and the east: “Today we see how different the years of separation have actually made everything — like twins who grow up in completely different situations.2 Similarly, a forty-year-old man from Hamburg expressed how foreign eastern varieties are for him by saying, “When I hear Saxon German, my first reaction is always: that doesn’t belong here.” On the other hand, a fortyyear-old northeastern man, in talking about how to tell a Westerner from an Easterner, said, “You can tell from the way they present themselves, from the way they behave, but not from the language. In the north over there they speak pretty much exactly like we do.” Although this informant makes important distinctions between Westerners and Easterners generally, language is not one of the things he views as different.

232

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.4. Central Westerners’ mean ‘correctness’ ratings for each individual region

3.2 Pleasantness The evaluations of pleasantness present a very different picture. Figure 15.8 shows that Northwesterners rate the far northwest as the most pleasant of all, although the central northwest and the far northeast are also seen as quite pleasant. This is different from the attitudes toward correctness, where Northwesterners perceived a big difference between northwestern and northeastern varieties. Northwesterners also see the language of the central east as just as unpleasant as it is incorrect, but the entire south and much of the central west is seen as neither pleasant nor unpleasant. The differences between the pleasantness perceptions of Northwesterners and those of Northeasterners are striking, however, as can be seen in Figure 15.9. It is the far northeast that Northeasterners find most pleasant of all, although the rest of the north, including the northwest, is also seen as quite pleasant. Northeasterners also find southern varieties generally less pleasant than Northwesterners do, but they also perceive more distinctions — the varieties in the far south along the French and Swiss borders are viewed as rather pleasant. Also, all of the central eastern varieties are seen as having varying degrees of unpleasantness. Equally interesting is that, in contrast to the correctness ratings, the former border is only slightly more visible for Northwesterners than it is for Northeasterners.

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

233

Figure 15.5. Central Easterners’ mean ‘correctness’ ratings for each individual region

This is not at all the case for the Central Westerners’ pleasantness ratings, however. Just as with their evaluations of correctness, the pleasantness map for the Central Westerners (Figure 15.10) shows a pronounced border between the north and central west. Most of the regions in the east and in the south are seen as neither particularly pleasant nor unpleasant, and the entire northwest, extending down into much of the central west, is seen as quite pleasant. For Central Easterners, however, as can be seen in Figure 15.11, the border is nearly invisible, since they perceive northern varieties to be generally pleasant and other regions’ varieties as less so. There is more agreement between the two groups about the central east, although, again, there is more variation in the Central Easterners’ map than in the Central Westerners’ map. However, the two groups agree quite a bit more about the southwest, whereas a far greater distinction can be drawn between the Northeasterners and the Central Easterners with respect to these varieties. The attitudes of the Southwesterners (Figure 15.12) are more similar to those of the Central Westerners than they are to the attitudes of the other groups, except that Southwesterners are more critical of central eastern varieties than Central Westerners and less critical of the varieties spoken in the far south of the southwest. This last point is curious; it seems that Southwesterners as a whole find the varieties of the far south, such as Baden and the Black Forest, the Lake Constance region, Upper Bavaria, and the city of Munich,

234

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.6. Southwesterners’ mean ‘correctness’ ratings for each individual region

Figure 15.7. Westerners’ and Easterners’ perceptions of the ‘correctness’ of western and eastern varieties

to be preferable in terms of pleasantness to the southwestern varieties spoken a bit further north, but this is not the case with correctness. So with the exception of the Northwesterners, most Westerners seem to perceive eastern varieties as generally less pleasant than western varieties. For Easterners, however, the general situation is less obvious from these maps. Generally, Easterners perceive

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

235

Figure 15.8. Northwesterners’ mean ‘pleasantness’ ratings for each individual region

differences in pleasantness to be more marked along the north-south axis than the westeast axis, although Northeasterners are more critical both of southwestern varieties and of central eastern varieties than Central Easterners themselves are. The mean scores of Westerners’ and Easterners’ pleasantness evaluations of western and eastern varieties (see Figure 15.13) show that Westerners definitely do perceive western varieties as more pleasant than eastern varieties, a difference significant at .0005 in a paired-samples t test. Interestingly, Easterners also perceive western varieties as more pleasant than eastern varieties, although the difference between these evaluations is smaller, significant at .001 in a paired-samples t test. Evidence that Westerners tend to find eastern varieties rather unpleasant can also be found in the qualitative data. The most commonly cited reason is that these varieties are simply unfamiliar to Westerners. One 32–year-old northwestern man, for example, stated, “There are aversions to many eastern German languages that we just weren’t used to. It all sounds so completely different.” A 26–year-old man from southwestern Germany expressed it like this: “There’s a tendency with me (to think) that eastern German sounds kind of stupid. I know that’s not really very nice, but I’ve noticed it.”

236

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.9. Northeasterners’ mean ‘pleasantness’ ratings for each individual region

3.3 Map Task In the map task, informants were given a map with only cities on it and were asked to draw Germany’s dialect boundaries as they themselves perceived them. The first step in analyzing these maps was to determine the relevant areas. I did this by looking at all of the maps and making a list of all of the different regions drawn. The relevant part of the data here is not what names informants use, but the regions themselves. This means that despite one informant’s calling the far northwest “Baltic-Sea-North-German” and another calling it “North German,” both of these were considered the same region and fall into the category of “far northwest.” I found a total of 46 regions that were drawn by at least 5% of the informants, which were generally easily divided into northern, central, and southern areas. These forty-six regions form the dependent variables for the analysis of the map task. To determine the extent of these regions, I used a transparent map of Germany the same size as the maps the informants used but with a crucial difference — this map included fifty-six small regions corresponding to a combination of present-day and historical political boundaries. I placed this map over the top of each map the informants drew and looked to see how many of the smaller regions were included in each of the dialect boundaries the informants drew. From that, I was able to draw the borders of each of

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

237

Figure 15.10. Central Westerners’ mean ‘pleasantness’ ratings for each individual region

these regions as my informants perceived them. Figure 15.14, for example, is a map of the regions included in the general region of “Bavaria.” The different shades correspond to the number of informants who included each region; the darker the shade, the more informants who included that region. The far eastern part of the south was outlined by more than eighty-one informants. Fewer and fewer informants included regions to the north and west. Figure 15.15 is a map of the regions included in the overall region of “central east.” Fewer informants drew this region, so not all of the shades are represented, but it can be seen that the center of the central east (the region of Saxony) is regarded as most prototypical, and fewer and fewer informants included regions to the north, west, and east. To produce a map of the general perceptual boundaries of all informants, I compared the differences in shades on all the maps for each of the forty-six regions drawn. In the map of the perceptual region of “central east” in Figure 15.15, for example, the perceived difference between the area shaded darkest and the surrounding regions is between Shades 3 and 2, resulting in a difference of 1. On the same map, at the top of the perceptual region where the most lightly shaded regions border on regions not included at all, I noted a difference between Shades 1 and 0, also a difference of 1. In one place on the same map, however, where the second-lightest shade borders on a region not included by anyone, I noted a difference between Shade 2 and Shade 0 — a difference of 2. I then drew on a map the differences between all the shades in all the regions perceived, totaling

238

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.11. Central Easterners’ mean ‘pleasantness’ ratings for each individual region

the differences from each of the maps. From that, I was able to produce the map in Figure 15.16, which shows a combination of all boundaries drawn by all informants. The thickest lines indicate boundaries perceived by nearly everyone, and the thinnest lines indicate rarely perceived boundaries. Boundaries where there was only a difference of one shade for the combination of all maps were not included. Although the northern part of Germany shows fewer pronounced boundaries, there is an obvious one along the former east-west border. In central Germany, however, the boundary drawn along the former border is much more obvious, falling into the category perceived by the maximum number of people, and is just as pronounced as the border separating the south from central western Germany.

4.

Summary and Interpretation

In summary, this study has indicated that Westerners find western varieties both significantly more correct and significantly more pleasant than eastern varieties. Easterners don’t perceive a difference in terms of correctness, but they do in terms of pleasantness — they also perceive western varieties as more pleasant than eastern varieties. And when asked to draw dialect boundaries on a map as they perceive them, Westerners and Easterners alike

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

239

Figure 15.12. Southwesterners’ mean ‘pleasantness’ ratings for each individual region

Figure 15.13. Westerners’ and Easterners’ perceptions of the ‘pleasantness’ of western and eastern varieties

tended to draw a pronounced boundary along the former east-west border, especially between the central west and the central east. It is obvious from this study that the muchdiscussed “wall in the mind” is still evident not only in terms of economic and social differences but also in terms of perceived language differences. More important, however, it is also clear that Easterners are at a clear disadvantage with respect to perceptions of

240

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.14. Map of smaller regions included in the general region of ‘Bavaria’

their language, to a rather large degree in the west but also to a lesser degree in eastern Germany itself. It is especially important that this sort of trend be noted during a period of social and political change, since language attitudes have been shown to have an impact on hiring practices, success in school, and whether people win or lose court cases. The qualitative data support fears that such prejudices may be in effect. A young man from southwestern Germany stated this problem quite bluntly when I asked him whether he thought someone could have trouble getting a job in a bank because of his or her accent. He stated, “A Bavarian (a region in southwestern Germany where the informant is from) would have no trouble getting the job.” Then I asked, “But what if it’s a Hessian (a region in central western Germany), and it’s really obvious where he’s from?” and he answered, “No, I don’t think so. I wouldn’t say that.” But when I asked, “And if it’s a Saxon?” (a region in central eastern Germany), he started laughing, and said, “Yeah, well, okay … but there I have to say that (it’s relevant that) we didn’t hear that dialect for forty years. It’s so different. I’m sure there are nice people, good people over there too, but from what I’ve seen so far, well … I think Saxon German is a language that sounds like they’re starting to puke every fifty seconds, like this: [vomiting noise].” It is important to emphasize here that these are only attitudes, and it would be incorrect to conclude, based only on such attitudes, that widespread accent discrimination exists in Germany. However, since many

THE PERCEPTION OF POST-UNIFICATION GERMAN REGIONAL SPEECH

241

Figure 15.15. Map of smaller regions included in the general region of ‘Central East’

other studies have demonstrated a strong connection between attitudes and language-trait focused discrimination (Rey 1977; Williams et al. 1976; Lippi-Green 1994), there certainly is cause for concern. There is one ray of hope, though. Over and over again, Westerners cited eastern varieties being so unfamiliar to them as the main reason that they found them less correct, less pleasant, and simply quite different from western varieties. Easterners never mentioned this. Indeed, it is the case that Easterners had lots of exposure to western German varieties through western television and radio, but the reverse was not the case at all. Also, Easterners had a strong motivation to visit the west after the Berlin Wall fell (since they’d just had a revolution that to a degree had been about being able to do just that), whereas motivations for Westerners to visit the east were not nearly as strong. Perhaps, if my informants’ descriptions of why they feel the way they do are true, these attitudes will change over time as Westerners become more familiar and more comfortable with the various eastern German varieties. In the meantime, however, it is important to make people aware of the political and social motivations behind differences in language attitudes so that people will be less likely to view them as simple facts of linguistic life. If people examine their own language attitudes critically, perhaps the consequences of them do not have to be so dire.

242

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 15.16. Boundaries of all regions drawn by informants

Notes 1. Other methods of analysis, such as a combination of multidimensional scaling and k-means cluster analysis, have been used to analyze such data elsewhere, often with more accurate results. Means are used here because of greater ease in comprehension, but these other methods were also used in a more detailed analysis (Dailey-O’Cain 1997). 2. The quotes in this chapter were originally in German. Translations are by the author.

C 16 Variation and the Norm Parisian Perceptions of Regional French Lawrence Kuiper

The French have long had a reputation as a speech community with a high degree of linguistic self-consciousness. This reputation can be linked to a tradition of prescriptivism arisen from a perceived need to standardize. Prescriptivist movements can be traced back to the Middle Ages (e.g., Lodge 1993). The great variety of dialects still present at the time of the French Revolution demonstrates that prescriptivism had been primarily aimed at administration and the literary elite. These earlier movements were incorporated into the ideology of the Revolution, whose egalitarian doctrine, rather than condemning elitist linguistic tendencies, universalized them. French1 was no longer enforced only as an administrative and judicial language; it was promoted as the primary tool of citizenship. The Abbé Grégoire’s plea for the “Nécessité [ … ] d’anéantir les patois” [Necessity of wiping out dialects] (in Gazier 1969: 290) was only marginally effective, so that while Grégoire reported that only three million (or about 10–12%) of the French population spoke French in 1790, city dwellers from as late as the outbreak of the Franco-Prussian war (1870) complain that their fellow citizens living in the country were incomprehensible. It is not until the First World War that the number of first-language French speakers in France definitively passed 50% of the total population (Weber 1976). During the twentieth century, aided by the economic interdependence of provinces, mass media, and an increasingly mobile population, French has taken hold, at least as a second language, even in the regions farthest from Paris. Monolingual speakers of languages other than French compose a small proportion of the population. Among these speakers, “la très large majorité [ … ] n’exprime aucune volonté politique à travers leur usage” [for the great majority this use is not an expression of political will] (Bonnemason 1993: 45). What remains today of most French regional languages are abundant stylistic, phonetic, and lexical traits incorporated into regional varieties of French. Beginning with the publication of the Atlas Linguistique de la France (1902–1910), much scholarly work has been done to classify and describe these regional varieties. More recently, sociolinguists have also studied speakers’ attitudes toward phonetic aspects of a perceived norm

244

IMAGES, PERCEPTIONS, AND ATTITUDES

(Guenier et al. 1978) and looked at French language attitudes from a historical perspective (Weber 1976, Lodge 1993). Little has been done, however, to assess contemporary European French speakers’ conscious perceptions of regional varieties of their language. The present study examines precisely these explicit views of French nonlinguists. Adapting techniques described in Preston (1989) to the French linguistic and cultural situation, this study reveals the complex views of its Parisian respondents, who attempt to understand an environment of language diversity and prescriptive severity.

1.

The Study

This section will describe the test sample and the tasks that respondents performed. 1.1 The Test Group The test group was composed of seventy-six respondents, all born and living in the Parisian Region (Ile-de-France). These speakers were divided into groups according to sex (forty-four female, thirty-two male), age (forty respondents were age 18–25, eighteen were 30–45, and eighteen were 55 or older), and socioeconomic status (twenty-seven respondents were working class; the rest were middle class). Although some comments in the taped interviews and on perceptual maps suggest that age may play a role in certain linguistic perceptions, none of the factors (age, sex, or socioeconomic status) proved to have a statistically significant impact on the present study. Results will therefore be discussed in terms of the group as a whole, without regard to these variables. 1.2 The Tasks 1.2.1 Perceptual Mapping Respondents were given a map of France, with only the following detail: the cities of Paris, Lille, Cherbourg, Rouen, Reims, Metz, Strasbourg, Dijon, Lyon, Brest, Bordeaux, Bayonne, Toulouse, Marseille, Nice, Orleans, Nantes, Tours, and Geneva; the Pyrenees, Alps, and Massif Central mountain ranges; the Rhône, Seine, Rhine, Loire, and Garonne rivers; the borders of France, and the names of the surrounding waters. Field-workers asked respondents to circle and identify in writing any regions “where people have a particular way of speaking.”2 1.2.2 The Rating Tasks Respondents were asked to rate twenty-four different regions, including Francophone Belgium and Switzerland, according to three different criteria: degree of difference, correctness, and pleasantness. The regions in all three rating tasks were chosen based on a number of criteria. It was impossible simply to name the twenty-two administrative regions, because these do not in all cases correspond with historical regions, nor do they always correspond to dialect boundaries.

VARIATION AND THE NORM

245

Figure 16.1 shows the current French administrative boundaries. There are ninety-six “départements” (not counting the four overseas départements) and twenty-two “régions administratives” in France. To complicate matters, there are thirty “anciennes régions” whose names often partially or fully correspond to names of départements (e.g., Maine, Savoie) and current administrative regions (Alsace, Bretagne, Picardie), although their geographical position may have changed substantially, with the new region covering more, less, or different territory than its older namesake (Bourgogne, Franche-Comté). To further confuse the situation, there are also distinct performance dialect boundaries in France. Some major dialect boundaries are revealed in Figure 16.2.

Figure 16.1. The administrative divisions of France

246

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 16.2. Major dialect areas of France (adapted from Carton et al., p. 12)

The names of these language varieties correspond in various ways to other regional names: Some correspond to the names of départements (e.g., Jura), some to old regions (e.g., Lyonnais), and still others to modern administrative regions (e.g., Auvergne). The task of choosing a set of regional names to include on the questionnaires was complicated: The regional list needed to meet three criteria: (1) include all the territory of France, (2) include names respondents would recognize and could associate with a language variety, and (3) not include regions respondents may be incapable of rating because they perceive the region as being home to two or more greatly differing varieties. To facilitate this task,

VARIATION AND THE NORM

247

prestudy interviews were conducted with nonlinguist native speakers of French in Bloomington, Indiana. Respondents were asked to give a list of regions where they felt a particular variety of French was spoken. They were then asked to react to various regional names based on the way people there speak French. From these interviews, a list was generated using traditional regional names that are still meaningful to respondents along with current administrative regions. The resulting list, which will be further refined in future studies, is as follows (in the order of appearance on the questionnaire): Nord (North), Picardie (Picardy), Normandie (Normandy), Ile de France, Champagne, Bretagne (Brittany), Touraine, Centre, Franche-Comté, Bourgogne (Burgundy), Lorraine, Alsace, Poitou-Charentes, Auvergne, Limousin, Massif Central, Lyonnais, Rhône-Alpes, Jura, Gascogne, Langue d’Oc, Provence, Belgique (Belgium), Suisse Romande (Francophone Switzerland). It will be noticed that in this list, some regions geographically overlap. One apparent overlap is Rhône-Alpes/Lyonnais. “Lyonnais” corresponds to a large metropolitan area, but also to an actual variety of French. “Rhône-Alpes” corresponds to an administrative division and to several varieties as well. It was hoped that this division might shed light on the force of large urban areas on the linguistic perceptions of the Parisian respondents. As will be seen, Lyonnais did score significantly closer to the Parisian region than did Rhône-Alpes in the degree-of-difference task, although in other tasks the two regions grouped together. The other pairing that overlaps significantly is Touraine/Centre, a region traditionally identified with the mythical norm. It was believed that this pairing would reveal whether identification was linked primarily to the geographical location of the mythical norm or, rather, to the name “Tours.” The statistical results show Touraine significantly differentiated from Centre on all tasks. 1.2.2.1 Degree-of-Difference Rating. Respondents were asked to rate the twenty-four regions, on a scale of 1 to 4 as follows: “1, if you think that the French in this region resembles your own; 2, if there is a resemblance, but not as strong; 3, if the French spoken in this region scarcely resembles your own; 4, if the French spoken in this region is incomprehensible to you.” 1.2.2.2 Correctness Rating. Respondents were asked to rate the correctness of the French spoken in the twenty-four regions on a scale from 1 to 7, translated as follows: “1 = they speak a French in this region that is not at all correct. 7 = They speak a French in this region that is completely correct.” 1.2.2.3 Pleasantness Rating. Respondents were asked to rate the pleasantness of the French spoken in each of the twenty-four regions on a scale from 1 to 7, with wording very similar to that on the correctness rating task, replacing the word “correct” with “pleasant to hear.” 1.2.3 Taped Interviews Each respondent was interviewed on cassette at two stages: immediately following the completion of the perceptual mapping stage and again after the three rating tasks.

248

IMAGES, PERCEPTIONS, AND ATTITUDES

Respondents were asked to give further detail about the regions they had mapped and to comment on the labels they had included. In the final interview, the field-workers attempted to elicit any possible linguistic detail about regional varieties, including commentary about phonetic and lexical traits, imitations, stereotypes, and caricatures of regional varieties.

2.

The Results

2.1 Perceptual Mapping 2.1.1 Some Extreme Cases The mapping task elicited a wide range of detail. Whereas some respondents left virtually no part of the map blank, others drew one simple line across the middle of the map circling two or three regions above and below. The most regions included by any one respondent was thirty-one. The least detailed map was drawn by a man, age 42, who began by drawing a line down the center of the Brittany Peninsula. He then drew a circle around an area north of the Massif Central, called it the “Limousin” and wrote “speak patois 5 to 6 dialects.” Finally, he abandoned the task altogether, writing at the bottom of the map, “I can’t commit myself, there are too many dialects in France (each region has its own dialect).” Although the results were quite different on these two maps, they may come from a perception that the two respondents share — one expressed by many others in the taped interviews — that France is home to an infinite variety of dialects, which vary greatly as one moves through the country. Although there were no significant statistical outcomes related to the age variable in any of the rating phases, it is probably not mere coincidence that the two respondents referred to above, as well as all others referring to the uncountable number of dialects in France, belong to one of the two older age groups. In fact, no one under the age of 40 expressed this view. Although the respondents in this study are Parisian, the demographic situation in Paris suggests that it is extremely unlikely that the parents of respondents from these age groups were also born in Paris, and it is almost certain that their grandparents were not. It is quite likely that the parents of these respondents spoke a local dialect with their own parents. In fact, they probably grew up in a linguistic environment where great variation from village to village was the norm. Such a perception could be passed on in the same way that other linguistic items are, maintaining its force in the second generation, due to the firsthand accounts and experiences of the first generation but losing its relevance to the third. 2.1.2 The Composite Map Extreme cases such as the very detailed and very empty maps discussed above, were the exception. In fact, of the remaining seventy-four respondents, only four demarcated a number of regions outside of the range of four to twelve. The average number of regions designated was ten. The median number was seven. Figure 16.3 is a composite map of all respondents’ demarcations.

VARIATION AND THE NORM

249

Figure 16.3. Composite map of respondents’ hand-drawn perceptual maps

It was decided that the map would be shaded on a gray scale with darker areas being those indicated by more respondents. A different shade was used to represent each 10% increment, based on all seventy-six respondents, with white for the two lowest and black for the highest density. Each area also contains a label of the exact number of respondents who named that area. Many respondents made only small circles around city names, stating that they identified regional varieties with the names of cities. Cities that were circled in this way by more than one respondent are also circled on the composite map, and the number of respondents who circled a given city is combined with the surrounding regional total. Certain regions may have been combined by some respondents, whereas others may have separated the same region into two areas. One clear example of this was the Alsace and Lorraine region(s). Thirty-five respondents drew Alsace as a region, whereas only fourteen drew Lorraine. Twenty others drew a border around both Alsace and Lorraine, naming it as one region. If we add the times that the regions were grouped together to the

250

IMAGES, PERCEPTIONS, AND ATTITUDES

times they were named separately, the total number of respondents indicating Alsace is fifty-five, and the total for Lorraine is thirty-four. Simply shading these regions would show the large gap between the number of respondents naming each region, but it would prevent an important detail from being revealed. Although many respondents named Alsace without naming Lorraine as well (an obvious fact given the different totals), no respondent indicated Lorraine without also drawing in Alsace. This means that on no respondent’s perceptual map is there a reference to Lorraine without one to Alsace as well, whether the two are marked separately or combined into one region. In the labeling of the two regions, Lorraine and its adjectival form are used a total of three times, whereas Alsace or its adjectival form occur sixteen times. When two regions were represented both together and separately, each separately named region is colored to represent the total frequency: The score of the “referent” region when it is the only region of the two delineated on the map (in this case thirty-five), added to the number of times when the “nonreferent region” was the only of the two delineated (in this case zero), added to the number of times the two regions were named together (in this case, twenty). A dotted, rather than solid, line separating the two regions indicates that they were often named together. Within each such region, the number of times the region was delineated separately from its “mate” is placed, with the exact number of these occurrences given. An initial glance at the composite map shows that the more influential perception among these respondents was that of border regions and regions where languages other than French are spoken. The only border regions that received relatively little attention in the mapping phase are those areas outside France where French is spoken. This omission is probably due more to the strong perceptual boundary that the international borders represent than to a neutral view of the way people in Belgium and Switzerland speak. In fact, later stages of the study will show that the attitude of these respondents toward Belgian and Swiss French speakers is among the strongest. We will return to this composite map and discuss how it reflects the statistical results of the other phases and the major perceptual groupings of regions revealed by this study. 2.2 Correctness Ratings A look at the rank order presented in Table 16.1, column one, reveals that the perception of correctness by these Parisian respondents is not a function simply of the geographical proximity to Paris of the region in question. The darkest regions on the composite map, those in general found nearest the border regions, are represented by the bottom six in the rank order: North, Provence, Lorraine, Alsace, Switzerland, and Belgium. Also near at hand are Auvergne and Brittany, rounding out the bottom third in the rank order. With the exception of Auvergne, all of these regions have large sections of the population that speak a first language other than French (Flemish, the Lorraine dialect, the Alsacian dialect, German, Provençal, and Breton). These are certainly not the only dialects spoken in France. The composite map and this rank order combined, however, seem to indicate that these dialects are the most powerful perceptually, at least in their relation to the norm. Figure 16.4 is the conversion of the

251

VARIATION AND THE NORM

Table 16.1. Means, by region, of the three rating tasks. Numbers in parentheses represent rank order of the region for the particular task.

Ile de France Touraine Lyonnais Champagne Normandie Centre Poitou-Charentes Rhône-Alpes Franche-Comté Bourgogne Picardie Limousin Jura Gascogne Massif Central Langue d’Oc Bretagne Auvergne Nord Provence Lorraine Alsace Suisse romande Belgique

1 (‘Correct’)

2 (Degree of Difference) 3 (‘Pleasant’)

6.00 5.78 5.53 5.46 5.27 5.25 5.08 4.93 4.92 4.92 4.92 4.73 4.63 4.53 4.49 4.49 4.36 4.25 4.20 4.17 4.14 3.73 3.49 3.44

1.04 (1) 1.34 (2) 1.62 (6) 1.36 (3) 1.57 (4) 1.64 (7) 1.59 (5) 1.89 (10) 1.91 (11) 1.79 (8) 1.80 (9) 1.99 (12) 2.07 (13) 2.29 (18) 2.13 (16) 2.33 (19) 2.09 (14) 2.24 (17) 2.11 (15) 2.57 (21) 2.53 (20) 2.78 (22) 2.89 (24) 2.87 (23)

(1) (2) (3) (4) (5) (6) (7) (8) (9–11) (9–11) (9–11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24)

5.75 (2) 5.73 (3) 5.15 (8/9) 5.49 (5) 4.97 (13) 5.19 (7) 5.12 (10) 5.15 (8/9) 4.88 (14) 5.08 (12) 4.41 (19) 4.81 (15) 4.75 (16) 5.37 (6) 4.44 (18) 5.56 (4) 5.11 (11) 4.49 (17) 3.68 (23) 6.05 (1) 3.85 (22) 3.56 (24) 3.99 (20) 3.93 (21)

regional correctness scores to a Euclidean plot, as performed by a multidimensional scaling statistical function. The regions are grouped here into five clusters by a K-means cluster analysis. This visual representation of the regional scores is not surprising in light of the rank order in Table 16.1: The regions with the highest scores, Ile de France and Touraine, are the farthest left along the X-axis, and the lowest-ranked eight regions are at the opposite end on the far right. The K-means groupings add support to our suspicion that a strong second language in a region has played a major role in the correctness ratings. The two groupings farthest right each contain three regions with well-known linguistic minorities: Brittany, Provence, Lorraine and Alsace, Switzerland, and Belgium. Notably, two of the regions from the bottom eight in the rankings have been grouped with regions that are their geographical neighbors, even though they are clearly not neighbors in the rank order: Auvergne is number 18, Limousin 12, North 19, and Picardy 11. This grouping by the cluster analysis is probably the result of a great many respondents’ rating the paired regions identically. Because the regions on the questionnaire were arranged somewhat geographically, the two pairs (North/Picardie

252

IMAGES, PERCEPTIONS, AND ATTITUDES 1.5

Nord

1.0

Picardie

Limousin

.5

Centre

FrancheComt‡ 0.0

Dimension 2

Touraine

Belgique

Lorraine

Suisse

Lyon PoitouNormandie Charentes Bourgogne Langue d'Oc Gascogne Champagne Rhône-Alpes

Ile de France

-.5

Auvergne

Massif Central

Alsace

Jura Provence Bretagne

-1.0 -3

-2

-1

0

1

2

3

Dimension 1

Figure 16.4. Multidimensional scaling of correctness ratings with five K-means clusters

and Auvergne/Limousin) appear consecutively. Many respondents may have instinctively rated these regions with one another, both because of their geographical proximity and because they perceived no linguistic difference between the regions. Nevertheless, if the regions were rated alike by so many, why did enough respondents choose to rate Auvergne and North so low that they would fall six and seven places, respectively, in the rank order? Several respondents, during the taped interviews, gave one possible reason for the lower rating of Auvergne. Many Parisians afford special status to Auvergnat speakers because of Auvergnat immigration to Paris in the middle of this century. These immigrants were particularly visible because many of them became proprietors of restaurants and cafés, and some respondents may have reflected this special status of Auvergnat immigrants in their correctness ratings. That this reflection was negative is not surprising, as this is often the reaction of a community toward perceived “outsiders.” The discrepancy between North and Picardy can be understood in spatial terms. The region “Nord” is one of the most marked on the perceptual maps (see Figure 16.3). Conspicuously absent from most perceptual maps, however, are delineations that include Picardy. This may be the result of a perceived need to create a wide boundary between Paris and the North. Indeed, one respondent from the suburbs remarked that she commuted from Picardy to work every morning! Surely a place so close to Paris could not resist the standardization process that has spread out from Paris for a millennium! Because the name Picardy contains the historical name generally given to northern dialects (Picard), the gap visible on the perceptual maps between North and Picardy is closed considerably in the correctness ratings, where there is no visual aid showing the proximity of Picardy to Paris, causing respondents to hesitate. Thus, the two are grouped together in the K-means cluster analysis. Nonetheless, enough respondents were conscious of the proximity of Picardy to have some effect on the final numbers.

253

VARIATION AND THE NORM

Finally, of note in the K-means cluster analysis, is the grouping of Ile de France and Touraine. Numbers one and two on the correctness rank order represent at least as strong a statistical group as any of the more marked regions from the bottom of the same rank order. In fact, as Figure 16.5 shows, when we force the K-means analysis to reduce the number of groups to four, only the Ile de France/Touraine cluster remains intact.

1.5 Nord

1.0

Picardie

Limousin

.5

Centre FrancheComt‚

0.0

Dimension 2

Touraine

Belgique

Lorraine

Suisse

Lyon PoitouNormandie Charentes Bourgogne Langue d'Oc Gascogne Champagne Rhône-Alpes

Ile de France

-.5

Auvergne Massif Central

Alsace

Jura Provence Bretagne

-1.0 -3

-2

-1

0

1

2

3

Dimension 1

Figure 16.5. Multidimensional scaling of correctness ratings with four K-means clusters

The tenacity of this two-region group suggests that perhaps the very strongest respondent perception is that of the norm. I shall return to this question as I present the results of the other tasks. 2.3 Degree-of-Difference Ratings The mean rankings in Table 16.1 show degree-of-difference ratings quite similar to the correctness ratings. Only five regions move more than two places in rank. Fourteen do not shift or shift by only one place, and none shifts more than four places. The five shifts of three and four places can be seen as primarily a north-south geographical alignment. This slight change in alignment from the correctness rankings coincides with a well-known dialect division stretching across France approximately at the level of the Dordogne river in the west — Clermont-Ferrand in the center and Grenoble in the east. We therefore see Lyonnais, commonly considered “the north of the south,” slipping three places away from Paris. Gascogne and Langue d’Oc slip four and three places, respectively, whereas Brittany and North climb three and four places, respectively. Even three of the four regions that move only two places relative to the correctness task follow this north-south

254

IMAGES, PERCEPTIONS, AND ATTITUDES

alignment: Rhône-Alpes loses two places; Burgundy and Picardy gain two. Figure 16.6 shows the multidimensional scaling and K-means cluster analysis degreeof-difference ratings. Although they appear to be quite similar to the correctness plot, these clusters show several differences as well. First, with the exception of the grouping of Touraine in the “home cluster” with Ile de France and Champagne and the grouping together of Switzerland and Belgium, all clusters contain geographically contiguous regions. 1.5

Bretagne 1.0

.5

Touraine Champagne

0 .0

Rhône -Alpes

Centre PoitouCharentes

FrancheComt‚

Bourgogne

Normandie Ile de France

Dimension 2

Provence

Jura Massif Langue d'Oc Central Gascogne

Picardie

- .5

Lorraine

Limousin

Suisse

Belgique Alsace

Auvergne

Lyon

- 1.0

Nord - 1.5 -3

-2

-1

0

1

2

3

Dimension 1

Figure 16.6. Multidimensional scaling of degree-of-difference ratings with six K-means clusters

This is not the case in the correctness groupings, where such a claim can be made for only one group in either Figure 16.4 or 16.5. Although this evidence alone is not sufficient to draw any conclusions, this difference in the groupings for the two ratings suggests that spatial organization of regions plays a greater role in the degree-ofdifference task than it does in the correctness task. The identical first-place rankings of the respondents’ own region (Ile de France) in the correctness and degree-of-difference tasks indicate a high degree of linguistic security. That is, these respondents are secure in their belief that the variety of French they speak is the most universally acceptable. This finding is reinforced by the cluster analysis of both tasks, which groups Ile de France with Touraine, the mythical center of standard French as taught from public school books at least since the nineteenth century. In fact, references to Touraine as the home of “pure French” date back as far as the middle ages (Lodge 1993). Finally, of note in the cluster analysis of the degree-of-difference ratings is that Brittany is a separate cluster. It is likely that this separation arises from the ambiguous linguistic position of Brittany in the perception of the respondents. During the interview stage, many respondents expressed the view that French speakers in Brittany spoke French perfectly well, but they were also cautious to add that “quand ils parlent leur patois, on ne comprend rien” [when they speak their dialect you can’t understand anything]. Because the view of French in Brittany and the view of Breton represent the two extremes in the rating for this task (1 = exactly like the respondents’ speech, 4 = totally incomprehensible), it

255

VARIATION AND THE NORM

is not surprising that Brittany was statistically set apart from other regions. It would have been difficult to recognize the statistical uniqueness of Brittany without the K-means cluster analysis, because the range of responses was so narrow for this task. A mere analysis of frequencies does not reveal this uniqueness. 2.4 Pleasantness Ratings Table 16.1 shows that the rank order, largely the same in the correctness and degree-ofdifference tasks, is considerably realigned in the pleasantness task. The average difference in rank order per region between correctness and degree-of-difference ratings is 1.58, which rises to 4.16 when comparing correctness and pleasantness rankings. The pleasantness rankings are most dramatically different from correctness scores for the southernmost regions: Gascogne (correctness rank 14, pleasantness 6), Langue d’Oc (correctness 16, pleasantness 4), and Provence (correctness 20, pleasantness 1). By the same token, five of the regions rated in the top twelve for correctness drop to the bottom twelve for pleasantness (Normandy, Poitou-Charentes, Franche-Comté, Picardy, and Limousin). What does remain constant for these respondents is the status of the Ile de France/Touraine grouping, which gives way to Provence for first rank in pleasantness but does so only barely as these regions slide to the second and third ranks. The K-means groupings of the pleasantness ratings in Figure 16.7, however, show that the “home group” is statistically separate from the southern regions. It is apparent from this separation that the pleasantness these respondents perceive in their own speech is of a different nature than that associated with the southern regions. 1.5

Suisse

Langue d'Oc Gascogne

1.0

Provence

Auvergne

.5

Champagne

0 .0

RhôneLimousin Alpes Jura Bourgogne Poitou-

BretagneCharentes

Lorraine

FrancheComt‚

Alsace

Lyon

Normandie

- .5

Dimension 2

Belgique

Massif Central

Touraine Nord

Centre Picardie

- 1.0

Ile de France - 1.5 -3

-2

-1

0

1

2

3

Dimension 1

Figure 16.7. Multidimensional scaling of pleasantness ratings with five K-means clusters

Whatever triggered the statistical division of these two groups, the qualitative data clearly support such a separation. North forms a cluster by itself for the pleasantness task, just as Brittany did for the degree-of difference-task (Figure 16.6). This may again be the

256

IMAGES, PERCEPTIONS, AND ATTITUDES

result of the ambiguity caused by proximity with Ile de France (where the speech is considered very pleasant) and the presence of a regional dialect, which, according to most of the qualitative data, is considered unpleasant. In Figure 16.8, where the regions are forced into four clusters, North groups with its geographical neighbor Picardy, and Massif Central clusters with its geographical neighbor Auvergne. The inclusion of these two regions with the perennial “foreign” regions of Alsace, Lorraine, Belgium, and Switzerland is perhaps understandable in terms of an urban/rural distinction. Massif Central and Auvergne are two regions typically referred to by Parisians (and by these respondents) as remote and backward. It seems that only in the pleasantness task was this perception visible. Many references were made in interviews to rural/ urban language differences, but this may be less visible in the quantitative treatments. 1.5

Suisse

Langue d'Oc Gascogne

1.0

Provence

Auvergne

.5

Champagne

0 .0

RhôneLimousin Alpes Bourgogne Jura PoitouCharentes

Bretagne

Dimension 2

- .5

Belgique

Massif Central

Lorraine

FrancheComt‚

Alsace

Lyon Normandie

Touraine Nord

Centre - 1.0

Picardie

Ile de France - 1.5 -3

-2

-1

0

1

2

3

Dimension 1

Figure 16.8. Multidimensional scaling of pleasantness ratings with four K-means clusters

This new configuration also pushes Touraine and Ile de France into the larger more central cluster suggesting that pleasantness is less salient than correctness in the perception of these regions. Finally, in the transition from five to four clusters, the southern group (Gascogne, Langue d’Oc, and Provence) remains intact, much like the home group in the correctness clusters. In summary, the statistical survey of the pleasantness ratings shows a favorable perception of the southern regions. The great shift in rank order and groupings between the pleasantness and correctness ratings (in contrast with the considerably smaller shift between the degree-of-difference and correctness ratings) demonstrates that perceptions of pleasantness are quite independent from those of correctness. Of course, the exception to this may be found in the regions on the extremely high and low ends of the correctness task, where movement in rank order was minimal. Four of the bottom five in the pleasantness task are also in the bottom five in the correctness task, and the top two in the correctness task are numbers two and three in pleasantness. It is safe to say that, for these respondents, extremely incorrect speech is by its nature unpleasant, whereas the most correct language cannot fail to be pleasant as well.

VARIATION AND THE NORM

257

2.5 Bringing It All Together: The Taped Interviews and Map Labels This section will use responses and imitations from the taped interviews, as well as written labels and commentary on the perceptual maps, to more fully explain the trends indicated on the composite map and in the statistical results of the rating tasks. Because of limitations of space, discussion will be limited in this section to the southwestern quadrant of the composite map, Alsace/Lorraine, Provence, and to perceptions of the norm. 2.5.1 The Southwest The southwestern corner of France on the composite map presents the most confusing picture of any section, a perspective echoed in the statistical results: Gascogne and Langue d’Oc, the regions roughly corresponding with this area of the map, cluster with the largest central K-means group in the correctness task; in the pleasantness task, however, they cluster with Provence to form a southern block of highly rated regions. These three also stay in the same K-means group in the degree-of-difference task. To understand the mixed statistics concerning this area of France and the muddled picture accompanying it on the composite map, it is perhaps pertinent to turn to the urban areas, whose reputations may in part account for this region not belonging more solidly to a “southern” perception. Although Bordeaux, Toulouse, and Bayonne all belong to southern France geographically, each has a specific quality that may cause many respondents to give it special status. Bordeaux possesses a long literary heritage, including Montaigne and Montesquieu, which may explain why respondents rated this region higher in correctness. Such a literary heritage connects Bordeaux to the literary capitol, Paris — that most correct of all regions. The status of Toulouse as a center of technology is another possible reason for the ambiguous linguistic definition of this region. The Toulouse that is home to the French aerospace industry stands against the traditional (and especially Parisian) view of the southern part of the hexagon as a backward agrarian society. Finally, Bayonne is identified with the nearby Basque population. Of the ninety-three different designations made in this region, sixty are exclusively placed around one of these latter two cities (Bayonne = 29, Toulouse = 31). Another nineteen designate the Bordeaux area separately. Twenty of the remaining designations are various combinations of the three cities. Eight of these respondents delineated all three city areas as one linguistic region, using widely varying labels. The other twelve outlined some combination of two of the three city areas. Finally, four respondents designated areas in the southwest of France that contain no cities, three circling the Pyrénées, and one more outlining an area between Bordeaux and Bayonne but including neither. Many remarks by respondents, both on maps and in interviews, indicate an ambiguity in how they should perceive this region: Should they view it as a part of the area where southern varieties are spoken, or should it be included in that area of France that has more or less succumbed to standardization? Of those who believed that southwesterners speak a southern variety, about half said that this region had “the same accent as in Marseille.” This comment was made especially in relation to Toulouse, although an equal number

258

IMAGES, PERCEPTIONS, AND ATTITUDES

believed that Toulouse has its own particular variety. Apart from the clear designation of the area surrounding Bayonne (twelve of the fifteen respondents labeled this area using the word “Basque”), it seems that the southwest is a region that, among Parisians, is currently undergoing a shift in perception. That shift is perhaps best summarized in the following respondent’s commentary on the region: The only real (accent) that is very recognized, that, that one can easily recognize [ … ] is the accent of the south, of … that go … one could almost say below the Loire, but not quite, in the west, it’s, since it’s more … um … how will I say? … it’s more “Parisianized,” um, more, um, Europeanized in a Parisian way.

Although a shift in perception may be taking place, the southwest is still a region that weighs strongly in the perception of these respondents. Despite a lack of consensus on how to delineate the boundaries in this area, the ninety-three designations are still the most in any corner of the map. 2.5.2 Alsace and Lorraine Certainly extralinguistic factors cannot be ignored in the perception of this historically charged region. Alsace and Lorraine as a territorial possession changed hands between France and Germany four times since 1871. Although the region was often named after Alsace (thirteen labels), and seldom after Lorraine (three labels), labels more numerous than either referred to its “German-ness.” A total of nineteen labels included either the adjective allemand(e) or germanique. These labels were often accompanied by other adjectives that seem more like a description of character than linguistic performance: strict, rigorous, hard. Adjectives describing phonological traits refer to choppiness, slowness, and again, to German-ness: cut, jerky, slow, German pronunciation, German consonance, Germanic pronunciation. No labels reflect a positive perception and some are clearly negative: hick, vulgar. Taped interviews echo these labels, many of the comments foretelling the statistical results of the other tasks. Clearly, the perceived proximity of Alsacian French to German, combined with the existence of a large group of speakers of the Alsacian dialect, has led to a high rating for degree of difference. The contention by many respondents that Alsacians are incomprehensible and use “their own expressions” led to analogously low correctness ratings. The follow-up interviews also found respondents forthcoming about their dislike of Alsacian speech. These comments were supported by statistical results in the pleasantness rating task. Comments made about the Alsace variety were in direct contrast to those about the southern varieties. Where the latter were “sing-songy,” “flowing,” and “warm,” the former were “choppy,” “jerky,” “cold,” and “hard.” An analysis of the phonological traits of this variety reveals why a speaker of standardized French may qualify it as jerky or choppy. The stress patterns of Alsacian French are word based, depending on the number of syllables in each word. In contrast, Standard French stress patterns are based on word groups. For example, a word of more than three syllables in Alsacian French would have alternating strong and weak stress,

VARIATION AND THE NORM

259

with a weak accent placed on the last syllable. Standard French places stress on only the final syllable of a word group, with relatively even stress on all other syllables. In addition, the rhythm and intonation systems of Alsacian French contain more elongated segments and have rising intonation contours on all prefinal (including initial) positions. It also has a slight rise and comparatively sharp drop in phrase-final position. The longer segments in particular may partly explain why some of the respondents described the variety as “slow.” The differences in numbers of designations on the perceptual maps between Alsace and Lorraine suggest that Alsace is a far more important referent than Lorraine in identifying this linguistic region. A great many extralinguistic factors may be responsible for this difference in status. The Alsacian capital, Strasbourg (also home to the European parliament), is larger than Nancy and Metz (the two major centers in Lorraine) combined. Strasbourg thus receives considerably more media and business attention. Alsace is one of the few regions of France whose modern administrative borders nearly match its traditional regional borders. It has, therefore, been able to maintain a certain regional ethnic identity that many administrative regions have not. Even though it is only about one-third the area of Lorraine and has considerably less population, Alsace has the ingredients necessary to be a strongly perceived linguistic and cultural entity: (1) a border with another powerful historical rival nation — Germany (Lorraine partially borders Germany, but about half of its northern border touches Belgium and Luxembourg, and its eastern border is all shared with Alsace), (2) a large and powerful urban center as its capital, and (3) long-standing traditional borders. 2.5.3 Provence Because Provence was rated at the very top for pleasantness, yet near the bottom for correctness, it should come as no surprise that it is the single area that received the most designations on the perceptual maps. In fact, labels seem to indicate that the region surrounding Marseille and Nice is for many the center of southern speech: Fourteen respondents labeled the region with a general term for the south (Midi, sud). Eleven respondents used a form of the word Provence, with two of these making reference to the Provençal dialect, one calling it a disappearing language and one citing it as the major influence for the regional variety of French. Two other respondents referred to Italian (notably less than referred to German for Alsace) as the external influence giving Provençal speech its distinctive traits. Seven respondents made comments about phonological attributes of the variety: Singing (two respondents); Singing accent of Marseille; Singing accent close to Italy; Dragging; They make words bounce and syllables drag; lively inflection.

The following words were also used to evaluate the speech of Provence: people’s accent; funny; warm; very strong; Pagnol (writer, playwright and director Marcel Pagnol who set his works in the region); “oh pochäre!” [imitation of an expression often associated with the region, rough translation: “poor baby!” (sarcastic)]

260

IMAGES, PERCEPTIONS, AND ATTITUDES

Strongly contributing to the region’s showing on the perceptual maps may be that Provence is well-known as a vacation spot. Many labels for the variety associate it with warmth and sun. In the follow-up interviews, this perception is reinforced. The following speaker’s comments are typical of those associating speech with weather: You tell yourself, “Ah the south of France, the people, um, the accent, it’s singing, it’s the sun, etc.” [ … ] But an accent or a language is always identified by the atmosphere that surrounds it, by the countryside there is, and it’s true that the south is much prettier than the north, and that explains the, the results.

These comments not only reflect a perceptual connection between the weather in Provence and the Provençal accent but also a general willingness to relate climate and landscape to language variety. Finally, the taped interviews indicated the strength of the linguistic caricature of this region: The most respondents overall readily imitated Provençal French. Respondents who did these imitations concentrated on transforming the nasal vowels of Parisian French into velar nasal palatalized consonants, on using a systematic pronunciation (rather than omission) of the “e instable” /e/, and on the changing of /r/ from a “back r” to a “front r.” Notable during these imitations was that nearly all respondents chose to use profanity in their imitations. One respondent almost refused to imitate, making the following comment: There is also an accent … and right away we come to dirty words as soon as we speak [… with that accent]. They say “putain” [‘goddam,’ ‘fucking,’ lit.‘whore’] every word, in every sentence they say “putain.”

Relating Provençal French to socially marked language implies a judgment on the speakers of the variety as well. Respondents were willing to use language they would otherwise find unacceptable with strangers (the interviewers) only when imitating this variety. This is the only evidence in the qualitative data that confirms Provence’s low correctness ratings. No explicit comments (apart from those quoted above) were made about the noncorrectness of the variety. In all rating tasks, Provence is at one extreme of the results, either highest or lowest. It is also the most often delineated region on the perceptual map and the most referred to and imitated variety in interviews. In other words, whether respondents were asked to discuss it specifically in relation to other regions (as in the rating tasks) or whether they brought it up on their own in a discussion of France’s linguistic environment, it is clearly the most marked region in this study (just as the U. S. “South” is in the work of Preston, e.g., 1997, and apparently, for many of the same reasons). 2.5.4 The Strength of the Norm Although Provence displayed a strong salience because of its perceived difference from Parisian French, two regions in the rating tasks are more solidly identified, at least statistically — Ile de France and Touraine. The statistical strength of the perception of these two regions is manifest in the consistency of the K-means analysis: In the correctness and pleasantness ratings tasks, these two cluster together consistently, whether the Kmeans function is asked for six, five, or four clusters. Champagne (sharing a boundary

VARIATION AND THE NORM

261

with the Ile-de-France administrative region) is the only region to be grouped with them and only in the degree-of-difference ratings. The common perception of Tours and the Loire valley as the cradle of standard French is confirmed by the labels of these areas on the perceptual map. The following is an inventory of all labels given by the eighteen respondents (eleven around Tours and seven others very nearby) who outlined a region on this part of the map: Touraine (two respondents); Touraine, the best French; pure French (two respondents); the purest French; normally: pure French; no accent (two respondents); the language at the origin of national French; little accent; tourangeot accent; Berri; Berry (two respondents); Loiret.

These last five respondents are the only ones who indicated the existence of a nonstandard variety in these regions. Four of these last five are represented in Figure 16.3 by a small circle to the northeast of Tours. The name “Berry” refers to a feudal Duchy whose sovereigns occasionally played important roles in royal politics during the middle ages and was also the name of one of the Provinces before the Revolution. Clearly, the majority of the remaining labels in this area are consistent with the notion of Tours as the center of standardness. Without the Touraine and Berry delineations, Figure 16.3 would have a wide section of blank space on all sides of the Loire River. Obviously, the great majority (more than 80%) of individual maps indeed show this wide empty area. In interviews, respondents were asked to comment on what these spaces represent linguistically. Their responses were quite similar to the labels given for the Touraine: no accent (five respondents); no specific accent; flat, Parisian, no accent because it’s my accent; neutral accent; the accent can’t be heard; they have an accent but, well, like us, you know; same as in Paris; they speak like me; the best French; It’s the center of France where they speak the best French; the blank spaces are like Paris; unitary unpronounced accent; slight accent; a less defined accent; no really marked accent; a less marked accent; I’m not very familiar with this region (six respondents); pretty much all the same; a continuum of accents; it’s the country; it’s country people, they roll their “r.”

Clearly the great majority of these respondents have a view of this region as the center for the French linguistic norm. This view may be more conscious (respondents saying the French is “best” there) or less so (those saying the speakers in the region have “no accent”), but it is beyond doubt among the strongest perceptions held by these respondents. The combined strength of the statistical results and the large number of respondents identifying Paris and the Loire region with the standard in interviews and on maps easily put it on an even footing with Provence as a force in the linguistic perception of these respondents. Other statements made during interviews about language correctness, and standardness, too numerous to detail here, indicate that perceptions of the norm may indeed outweigh any other linguistic view these Parisian respondents hold.

262 3.

IMAGES, PERCEPTIONS, AND ATTITUDES

Conclusion

I have attempted to show how the different phases of this study complement each other and create a clear picture of the linguistic perceptions of the respondents. The mapping phase allows us to look at how subjects react to geographical space with as little steering as possible. Although the map used in this phase had very little detail, the results suggest that the inclusion of so many cities may have led respondents to outline areas they may otherwise have not. Future perceptual dialect studies in France should strongly consider eliminating most or all cities from the perceptual map given to respondents. Despite this shortcoming, the perceptual mapping tasks did yield information that would not have been readable in the statistics of the rating tasks alone. Perhaps the best example of this is the southwestern corner of France. Conversely, the rating tasks gave results that may have been masked in the perceptual mapping. By naming specific regions on the questionnaire, the rating tasks directed the respondents’ attention to questions that they had perhaps not consciously considered before. The clearest example of results in the ratings that may have been unnoticed in the mapping is the extremely strong perception of the Parisian and Touraine norm. The rating tasks highlighted important results that would have been missed by simply assuming that blank or sparsely marked spaces on the perceptual map indicated a lack of clear linguistic perception. Finally, the taped interviews helped to steer the researcher toward answers to difficult questions raised in the statistical outlays and on the perceptual maps. Two examples of this are the comments by respondents about the North and Brittany, which reflected the ambiguous position of these regions in the minds of many respondents. The information provided in this chapter represents only a small part of the issues raised by the qualitative and quantitative data. Besides the omission of detailed treatment of strongly perceived regions (Brittany, the North, and Paris with its specific variety that many respondents claim exists but that none admits speaking), questions about folk definitions of linguistic terms (patois, dialecte, parler), and the refusal of certain respondents to perform certain tasks (most specifically the correctness task) were not even touched on in this chapter. Other issues, such as the rural/urban animosity displayed by many respondents, were touched on only briefly. The numerous questions raised and the answers (fully or partially) produced here attest not only to the richness of the data generated by studies such as this one but also to the necessity for further investigation into linguistic perception. It is evident that someone’s linguistic perception is not simply the linguistic component of an overall worldview but also the vehicle through which many other prides and prejudices are expressed. Notes 1. Editor’s note: In this article (and in many of the sources it cites), “French” is used to indicate “Standard” or “Parisian” French, and dialects are often referred to as “languages.” 2. Throughout this chapter, I have translated collection devices and commentary by respondents from the French. For clarity, I have translated as directly as possible, without concern for style.

C 17 The Perception of Turkish Dialects

Mahide Demirci

Brian Kleiner

The work of Preston, and work currently being done following his lead, has involved the examination of perceptions and evaluations of regional dialects in countries as diverse as the United States, Brazil, Japan, France, Spain, and Germany. However, no study has so far applied this methodology to examine the perceptions and evaluations of regional dialects in a Middle-Eastern country. Obviously, such additional studies are required to test whether the conclusions and generalizations drawn from earlier studies hold true across a wider range of cultures. The current study looks at how the people of a particular Turkish city perceive and evaluate the regional dialects of Turkey, a country that is a cultural and geographical bridge between the West and the East.1 Using the methods that Preston employed in his study of regional varieties in the United States, we hoped to find out whether similar conclusions would be obtained in our own study. The results show that many of the phenomena reported in findings for U. S. society are also found in Turkish society. On the other hand, we found some points of difference that may contribute to the growing body of knowledge within the field of perceptual dialectology. Specifically, we argue that the local social, political, economic, and linguistic realities of a particular nation contribute to the evaluation of regional speech varieties. Furthermore, in the evaluation of regional varieties, certain factors may be more salient in some cultures than in others. For example, for the Turkish respondents in this study, degree of economic development, identification with European culture, and the influence of neighboring countries all play a significant role in speech evaluation; these same factors, however, might not be so salient in other cultures. Although the results reported here have implications for the growing area of perceptual dialectology, in the second part of this chapter, we show that these results, when viewed in terms of the social attributes of the respondents, also have significant implications for what has emerged in recent years as “sociolinguistic dialectology.” Specifically, we establish that several significant, systematic, gender and age-based patterns emerge from the evaluation of the regional dialects of Turkey by Turkish

264

IMAGES, PERCEPTIONS, AND ATTITUDES

respondents. Based on these results, we argue that some of the same systematic patterns of social differentiation that emerge from the study of production data are also present in the conscious evaluation of language varieties. Thus, following Preston (1989), we conclude that studies in perceptual dialectology may provide illuminating and valuable supplementary material for inquiry in sociolinguistic dialectology.

1.

On Turkey

Due to Turkey’s location and rich history, its sociocultural life has been shaped by both European and Asian influences and is, therefore, very diverse. The tension between two largely different cultures is one of the defining features of modern Turkey. It is a developing country with most of its wealth and industry concentrated in the west, whereas the eastern part is relatively undeveloped and impoverished.

2.

Method

The Turkish respondents in our study were all natives of Bursa, the fifth largest city, which lies in the western part of Turkey on the Sea of Marmara about 280 kilometers southeast of Istanbul. The 142 respondents were fairly evenly divided into twelve subgroups according to gender, age, and class. They were divided into three age groups (17–25, 35–45, and 50+) and two social classes (middle and upper).2 Unlike Preston’s study, which involved groups from different regions, ours involved only one regional group. As will be shown later, this group was the equivalent of Preston’s linguistically secure southeastern Michigan group. Each respondent was given four tasks to carry out on a questionnaire: 1.

2.

3.

4.

Correctness task: assign a score on a scale of 1 to 7 to each of 73 Turkish cities for the “correctness” of the speech of that city (with 1 being the least correct and 7 the most correct). Pleasantness task: assign a score on a scale of 1 to 7 to each of 73 Turkish cities for the “pleasantness” of the speech of that city (with 1 being the least pleasant and 7 the most pleasant). Difference task: assign a score on a scale of 1 to 4 to each of the 73 cities for the degree to which the speech of that city is different from the home area (with 1 being least different and 4 being most different) Map drawing and comments: on a map of Turkey, draw circles around those areas of Turkey where people speak regional dialects and give comments and opinions about the speech of and the people living in those regions.

For each task, the respondents were allowed to refer to a map of Turkey on which the locations of all seventy-three cities were specified. This map was provided as an aid for

THE PERCEPTION OF TURKISH DIALECTS

265

subjects who might not have been familiar with the location of particular cities. The English word cities is somewhat misleading since its Turkish equivalent refers to officially recognized regions, which includes towns and villages.

3.

Results

The following section presents the results for the first three tasks, although some interpretation is given in a later section. Although the task four results (i.e., map drawing) will not be presented here, the written comments of respondents will be discussed where appropriate. It will become apparent that these comments are a rich source of clues for the interpretation of the ratings. For each of the first three tasks, the statistical procedures of multidimensional scaling and K-means cluster analysis were employed, but only the results of the K-means cluster analysis will be reported on here. 3.1 Correctness Task The results shown in Figure 17.1 represent the distribution of mean scores for eight groups selected by the K-means cluster analysis for the correctness task.3 The results illustrate, in general, a continuum of correctness ratings that is highest in the western part of Turkey and that decreases gradually toward the east. An exception to this pattern is that the cities to the west of Istanbul (such as Edirne, Tekirdag) in the “Balkan” region cluster along with cities in the western interior (mean score of 4.38) and that the city of Kirklareli was selected as a separate group by the K-means analysis (mean score of 4.58). This may have to do with the perception of these cities as the home of Turkish people who have immigrated from Greece, Bulgaria, and the former Yugoslavia and who speak an “accented” Turkish. In their written comments, respondents said that, although the people of this region are very cultured and sophisticated, they speak a variety that is very different from Standard Turkish. Some respondents remarked that these people invert their sentences and “drop the sounds” at the beginnings of words. The results show that respondents overall rate their own region highest for correctness. This region (mean score of 5.83) includes Bursa, Istanbul, Kocaeli, Adapazari, Canakkale, and Balikesir and skips to Antalya on the southern coast and the cities of Eskisehir and Ankara toward the interior of the country. The speech of this area, called “Marmara,” is said by respondents to be very clear, correct, “pure” Turkish and “without any accent.” The people of the Marmara region are said by respondents to be very “cultured,” “soft,” “liberal,” and “honest.” That Ankara, Turkey’s capital (and not part of Marmara), rates higher for correctness than its immediate neighbors (except for Eskisehir) is not surprising: it is a result very similar to Preston’s finding that Washington, D. C. in the United States scores higher for correctness than the states that surround it. On the other hand, although Ankara receives high scores for correctness, respondents wrote that the people there speak a variety that is more difficult to understand and has features of eastern speech. Antalya (also not part of Marmara) on the Mediterranean is probably rated

266

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 17.1. Correctness task

so favorably because it is popularly perceived as being a prestigious resort area for wealthy Turkish people and European tourists. The next most favorably rated group includes most of the western interior cities (a mean score of 4.38). This group extends from Mugla and Mersin on the Mediterranean all the way to the western Black Sea cities of Kastamonu and Samsun. Respondents say that the speech of this area is “rough” but not as bad as the speech to the east, and even though the speech is said to be “polite,” it is still considered very “rural” compared with cities to the west. The people living in this region are said to be “good,” “warm,” “sincere,” and “energetic.” In addition to the exceptions already mentioned (Ankara, Eskisehir, and Antalya), the two cities of Bartin and Sinop on the Black Sea do not cluster with the western but with the eastern interior group (mean score of 2.96). Although few respondents specifically discussed these two cities in their written comments, many did claim that the speech of the western Black Sea cities has some characteristics of the eastern Black Sea “Laz” dialect. The ratings dip lower for correctness farther east. The large eastern interior group stretches from Yozgat on the west side all the way to the border with Georgia and Armenia. It also reaches from Giresun and Rize on the eastern Black Sea down to Hatay and Urfa on the border with Syria. The speech of this area is considered very rough, and the entire region is thought of as undeveloped. Within this large territory, two cities were singled out by the K-means cluster analysis and, therefore, are regarded as distinctive by the respondents in terms of correctness: Trabzon on the eastern Black Sea and Gaziantep on the border with Syria. It is plausible that Gaziantep (mean score of 3.28) is distinctive because it is thought of as a very developed city; many respondents wrote that the people there are considered to be more liberal and literate than those in other eastern cities. In

THE PERCEPTION OF TURKISH DIALECTS

267

addition, this area is thought to be heavily influenced by Arabic culture and language. Trabzon (mean score of 3.46) is the home area of the dialect popularly referred to as “Laz.” It is a large, modern city and perhaps receives higher ratings for correctness because of this. The Laz dialect is thought by many respondents to result from the influence of Greek and Georgian. Cyprus was also singled out by the K-means cluster analysis as a separate group (mean score of 3.5). Respondents wrote that the speech of Cyprus is heavily influenced by Greek and English (Cyprus was a British colony), and therefore, the Turkish spoken there is less correct than in other regions. Many respondents claimed that the speech of Cyprus contains many lexical items not found in Turkish. Finally, the group selected by the Kmeans cluster analysis with the lowest ratings for correctness is found in the far southeastern part of Turkey. These are the cities of Hakkari, Sirnak, Van, and Bitlis; many Turkish Kurds live in this region (mean score of 2.16). In their written comments, many of the respondents expressed some antipathy toward these cities because of the social unrest there and recent terrorist activities. They commented that these cities are extremely undeveloped, that the people are “illiterate,” “backward,” “harsh,” and their speech is the most “degenerated” Turkish. Many noted that the rough climate has an influence on the language spoken there: For example, they say that the people “speak from the throat.” 3.2 Pleasantness Task Figure 17.2 represents the distribution of mean scores for eight groups selected by the Kmeans cluster analysis for the pleasantness task. The results for the pleasantness task show many parallels with those for correctness. In general, a continuum of pleasantness ratings is highest in the western part of Turkey and decreases gradually toward the east. However, it appears that the ratings were slightly less harsh for pleasantness than for correctness. For example, more cities cluster within the highest-rated group selected by the K-means analysis for pleasantness; almost all of the cities along the entire coastline of western Turkey fall into this highest-rated group. In addition, toward the east we find that the cities along the eastern Black Sea that clustered with the eastern interior cities for correctness now cluster along with the western interior cities for pleasantness. This may be because of favorable stereotypes concerning the people living in this region, referred to as Laz. They are regarded as very “hospitable,” “friendly,” “hardworking,” “honorable,” and “spirited.” Many of the respondents discussed specific linguistic features associated with the Laz region. For example, respondents said that the people use “thin” sounds (which might be interpreted as a folk description of the substitution of voiced stops for voiceless ones and/or the replacement of back vowels with their front counterparts). Although this region is commonly ridiculed in jokes because of its distinctive accent, one respondent claimed that the accent there is “sweet like candy,” a result similar to findings in the United States in which the U. S. South is ridiculed for its distinctive accent, but its speech is often characterized as especially “polite” or even “sweet.” Not surprisingly, the Bursa respondents also regarded their own speech as highly pleasant, with a mean score of 5.33, as did Preston’s linguistically secure Michigan

268

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 17.2. Pleasantness task

respondents. Bursa speech is said by respondents to be “polite” and the most pleasing to the ear. Figure 17.2 shows that again the capital city of Ankara, despite its location in the center of the country, belongs to the most pleasant sounding group. As with the results for correctness, two of the Balkan cities (Kirklareli and Tekirdag) cluster with the western interior cities for pleasantness. Also, the city of Bartin on the western Black Sea clusters not with its neighbors but with the eastern interior cities for pleasantness just as it did for correctness. Farther east, the city of Kayseri is viewed by respondents as distinctive in terms of pleasantness according to the K-means analysis (mean score of 4.08). This may be because Kayseri is famous for its businessmen, who are stereotyped on television, and respondents say that the speech of this region is “nice to hear” and understandable. For pleasantness, the city of Malatya clusters not with its neighbors but with the western interior cities, which could result from the popular knowledge (mentioned by many respondents) that several of Turkey’s presidents were from this city. Trabzon, on the eastern Black Sea coast, is viewed by respondents as distinctive for pleasantness (mean score of 4.65), just as it was for correctness. The city of Artvin, at the far northeast corner bordering Georgia, was also selected by the K-means analysis as a separate group (mean score of 3.57). The people of this region are generally viewed as “liberal,” “hospitable,” and “kind”; although the respondents say that there is an accent in this region, it is not a rough one. Once again, the speech of Cyprus was selected by the K-means cluster analysis as a separate group. Thus, it appears that the speech of this island is regarded as distinctive, not just in terms of correctness but also in terms of its pleasantness (mean score of 4.14). Finally, the group that receives the lowest mean score for pleasantness (3.03) is the large one that covers most of the eastern part of Turkey.

THE PERCEPTION OF TURKISH DIALECTS

269

3.3 Difference Task Eight distinct groups selected by the K-means cluster analysis for degree of difference are shown in Figure 17.3. Here again, there is a continuum from west to east: the farther east one goes, the more the speech is perceived as being different from the home site. The group selected by the K-means cluster with the lowest mean score (1.32) includes the cities of the Marmara region as well as Antalya, Eskisehir, and Ankara; the city of Mugla on the southwest coast was selected as a separate group by the K-means cluster analysis (mean score of 1.71). As they did for correctness and pleasantness, the Balkan cities of Tekirdag, Edirne, and Kirklareli clustered with the western interior group (mean score of 1.97), and as in the first two tasks, the western Black Sea cities of Bartin and Sinop did not cluster along with the western interior group. Bartin was selected by the K-means analysis as a separate group (mean score of 2.60), and Sinop clustered with the eastern interior group (mean score of 2.64), which included cities from Yozgat on the west side all the way to Artvin in the northeast. All of the eastern Black Sea cities fell into this group as well as the cities of Gaziantep and Hatay bordering Syria in the south.

Figure 17.3. Difference task

Cyprus was again selected as a separate group (mean score of 2.64), with many respondents describing the speech on this island as almost impossible to understand. Finally, the K-means analysis again selected a large eastern region (mean score of 3.19), whose speech was considered almost incomprehensible to the Bursa respondents.

270 4.

IMAGES, PERCEPTIONS, AND ATTITUDES

Summary and Discussion

The broadest generalization to fall out of the results of these three tasks is that speech in the cities in the west of Turkey are perceived by the Bursa respondents as being more correct, more pleasant, and less different than that of other cities in Turkey. Also, the farther east one goes, the lower the ratings fall for correctness and pleasantness and the higher the ratings rise for degree of difference. The Bursa respondents display a high degree of linguistic security, rating their own region’s speech highest for correctness and pleasantness, as did Preston’s linguistically secure Michigan respondents. In general, there appears to be a rough correlation between correctness, pleasantness, and difference, although some areas rate higher for pleasantness than for correctness, and vice-versa.4 There are “island” regions that, for one reason or another, are evaluated differently in the three tasks from their immediate neighbors. For example, the capital city of Ankara scores higher for correctness and pleasantness and lower for degree of difference than many of its neighbors. As Preston suggests for Washington, D. C., this result may be due to the relative prestige associated with the center of government. The “Balkan” cities west of Istanbul in general receive lower ratings for correctness and pleasantness and slightly higher scores for difference than one would expect for far western cities. As explained above, this may be due to the immigrant status of many of the residents of this region. In contrast, the city of Antalya on the southern coast is treated more generously than its neighbors by our respondents in all three tasks, and this may be due to its popular image as a resort area, as previously mentioned. The cities of Bartin and Sinop receive lower scores for correctness than their neighbors, although only Bartin scores lower than its neighbors in terms of pleasantness. The language spoken in these cities on the western Black Sea is, as noted earlier, heavily stereotyped. Along the Black Sea toward the east, the cities of Ordu, Giresun, Trabzon, and Rize all score low for correctness, but it appears that these coastal cities are perceived as more pleasant than other noncoastal cities in the east. Finally, the island of Cyprus is regarded as distinct in all three tasks, possibly, as mentioned above, the result of the perceived influence of Greek and English on the speech of Turks living there. The question of why the speech of the western cities of Turkey has so much prestige and why the speech of the eastern cities is viewed so negatively needs to be addressed. One account for the continuum of ratings from west to east is that western cities rate well in the three tasks because they are close to Istanbul, considered to be the source of the standard variety of Turkish. It is plausible that the Bursa respondents perceive their own speech and that of the other western cities as more or less close to the Istanbul standard and that increasing distance from the standard (regional) variety correlates with a decrease in the ratings. A second explanation for the west to east continuum is that these respondents orient toward and identify with European culture and, furthermore, associate western and eastern Turkey with Europe and Asia, respectively. It is clearly the case that the Bursa respondents identify with European culture: In their written comments, many said that the western cities are more “European” since the people are more “liberal,” “modern,”

THE PERCEPTION OF TURKISH DIALECTS

271

“educated,” “wealthy,” and “cultured.” If this is the case, it is not surprising that they gave higher ratings to the speech of western regions, lower ratings to interior regions, and the lowest ratings to eastern regions. A third account for the west-east continuum is that western Turkey is the most developed part of the country, supporting many universities, art centers, industries, and large cities. The farther east one goes, the more it is perceived by the Bursa respondents as being backward, undeveloped, and so on. Since social, economic, political, and linguistic realities can have an influence, all of these accounts might hold some validity. In the comparison of our results with those of Preston (1989), we have already shown that Bursa respondents behaved in much the same way as his Michigan respondents, by rating their home area highest for both correctness and pleasantness. As was the case in Preston’s study, for linguistically secure respondents, the high correctness ratings were more regionally limited (or “focused” on the home area) than were those for pleasantness. It is plausible that linguistically secure groups place a higher value on correctness than on pleasantness, since they appear to be less willing to share the former than the latter. In addition, both Turkish and U. S. groups rated neighboring regions fairly high, which suggests that it may be a cross-cultural truism that proximity to what is perceived as the best speech will also be viewed in a similar positive light. A case could be made, however, that this explanation is limited in certain respects and that “proximity” itself is not a salient feature for respondents but, rather, reflects bundles of other salient social, political, economic, and linguistic facts. For example, an important difference between our findings and those of Preston (1989) is that our results indicate a broad correlation between proximity to the home site and ratings for correctness and pleasantness, not the case in Preston’s study. Recall that his Michigan respondents rated southern states lowest for correctness and pleasantness but gave western U. S. states much better ratings for both even though these areas are located farther from the home site than the southern states. On the other hand, for the Bursa respondents, it is generally true that the farther away from home, the lower the ratings for correctness. In sum, the explanatory value of the notion of “proximity” to the home site of respondents is limited because ratings of regions farther away from the home site do not always correlate with distance (e.g., in Preston’s study). As shown above, the correlation between proximity to the home site and ratings for correctness, pleasantness, and difference in the Turkish results is generally good and reflects underlying linguistic, economic, and sociopolitical facts that are salient to the respondents. As suggested earlier, it is plausible that this correlation reflects there being in Turkey one well-established standard variety, whereas this is not the case in the United States, and the Turkish standard is based on a regional variety found in the western part of the country. On the other hand, we also suggested above that an even more important factor than proximity to the home site might be the degree of economic development of the city being rated, but it is difficult to isolate this factor since the range of development from west to east is similar to the proximity to the home area of the raters and to the source of the standard. Turkey is far more developed in the west, whereas the central part of the country

272

IMAGES, PERCEPTIONS, AND ATTITUDES

is less developed, and the eastern part of the country is extremely undeveloped. In their written comments, respondents displayed a keen awareness of the degree of development of the various regions of Turkey. In the United States, speech evaluation might be less dependent on perceptions of economic development, since all parts of the country are fairly equally developed.5 Perhaps in countries where there is greater disparity in terms of economic development from one region to the next, this factor might make a greater contribution to the perception and evaluation of regional speech varieties. There is some detailed evidence, however, for the argument that it is economic development that is salient to Turkish respondents, not proximity to the home site. For example, the city of Kutahya, adjacent to Bursa, clusters with the western interior group and not with the Marmara group along the dimension of correctness. This may be because this city is less developed economically and more rural than other cities to its west and it is perceived as such by the respondents. Another example is the city of Antalya, which consistently scored higher for correctness and pleasantness than its immediate neighbors. Many respondents commented that Antalya is the site of an active tourist industry that brings a lot of money to the area. One final example is the city of Gaziantep, which scored higher than neighboring eastern cities for correctness. It is a large developed city on the border with Syria that until recently prospered from trade with Arab countries. Recall, however, that the factor of economic development does not account for why the Balkan cities receive lower ratings than those cities in the Marmara region. In these cases, it appears that the salient factor in the low ratings was something like “degree of foreignness” rather than “degree of development.” In addition, of course, since the Bursa respondents appear to place a higher value on European culture, it is understandable that the speech of the “European,” western part of Turkey would be viewed in a more positive light and that the speech of the “Asian,” eastern part of the country would be viewed less favorably. Although it might first appear to be the case, the explanation of the continuum of ratings in terms of proximity to the home site is not adequate in accounting for the low rankings of the eastern cities. We believe that the very lowest mean scores for correctness and pleasantness, which go to the cities bordering Iraq, Syria, and Iran, indicate that more is involved than simply proximity. For a complete understanding of these low ratings, one must consider the political situation, since it is in these regions that there have been terrorist activities on the part of the Partiya Karkeren Kurdistan (PKK), an organization that aims to create a sovereign Kurdish state. Recall as well that respondents describe the south-central region, which includes Hatay, Gaziantep, and Urfa as heavily influenced by Arabic culture and language. Another important factor that might contribute to the low ratings is that some of the people in this area learn Kurdish and Arabic before they learn Turkish; therefore, their Turkish is thought to be influenced by these other languages. All these political, social, and linguistic facts may play a significant role in how people view the speech of the eastern cities. In fact, for many of our respondents the influence of neighboring countries and foreign cultures was a highly salient factor. This is true not only for the southeastern regions but in many other regions as well (e.g., the Balkan region in the west, Cyprus, and

THE PERCEPTION OF TURKISH DIALECTS

273

Artvin). Many of the cities and regions that were isolated from neighboring areas were perceived as distinctive because of this foreign influence. In other countries with fewer distinctive neighbors, it seems unlikely that such a factor would be a salient one (e.g., in the United States).6 In sum, it seems unlikely that proximity by itself is a salient feature for respondents in the evaluation of the speech of the various regions of Turkey. It is more likely that the continuum of ratings from west to east is a reflection of a host of linguistic, social, political, and economic realities that are salient to respondents. Rather than attempt to determine which of these factors is most salient, it seems more sensible to conclude that the combined perceptions of various social, political, economic, and linguistic realities may together have influenced the ratings of the regional varieties of Turkey. Clearly, for nonlinguists, it is difficult (and unnecessary) to untangle language from social, political, and economic realities, as the results of this study show. The results further indicate that there may be cross-cultural differences in what constitutes the salient factors influencing the respondents. For instance, for Americans, degree of economic development, identification with European culture, and the presence of neighboring countries may not be salient factors, whereas for Turkish people they are highly salient. Future research in perceptual dialectology should take into consideration that such important cross-cultural differences may contribute to the perceptions and evaluations of regional speech varieties.

5.

Social Differentiation in Dialect Evaluation

The remainder of this work will attempt to show that, because of the interplay of language use and evaluation, the study of the conscious awareness and evaluation of Turkish varieties reveals some of the same kinds of age- and gender-based patterns found in production data. Arguing for the viability of matched-guise language attitude tests, Labov (1972) claims that attitudes toward language varieties “do not emerge in a systematic form if respondents are questioned directly about dialects” (146). Contrary to this claim, we would like to demonstrate that the results of perceptual dialectology tests, which directly elicit conscious evaluations of language varieties, may emerge in systematic form. It should not be surprising that there may be significant age- and gender-based differences in the conscious evaluation of language varieties. The interaction between the evaluation of language and language use is complex and still not fully understood; nevertheless, it seems fair to say at least that these two phenomena are interwoven in the life of a speech community, and therefore one might expect that the same social forces that shape patterns of language use might also influence language evaluation, even at a conscious level. The interaction of language use and (subconscious) evaluation is well documented. For example, in his classic New York City study, Labov (1972) found that linguistic changes are accompanied by changes in subjective evaluation. He found that although their use of postvocalic “r” was stratified along class lines, the uniform positive evaluation of postvocalic “r” by the same group of speakers under 40 years old was indicative of a linguistic change near completion. Trudgill’s (1972) self-evaluation tests

274

IMAGES, PERCEPTIONS, AND ATTITUDES

demonstrated that just as women tend to use a higher degree of standard forms than men do, they also are more likely to exaggerate their use of standard features in the evaluation of their own performance, a result that may support the notion that women are more “status conscious” than men. Clearly, such evaluative stances taken by members of particular social subgroups play an important role in language use. The remainder of this work then will present the results of this study in view of the gender, age, and class of respondents and will follow with some discussion of these results. It will be shown that several typical sociolinguistic (gender- and age-based) patterns emerge and thus provide evidence that social differentiation manifests itself systematically in the conscious evaluation of language varieties as well as in language production. 5.1 Results 5.1.1 Gender A three-way multivariate analysis of variance (MANOVA) test was run on each of the eight groups selected by the K-means cluster analysis for both the correctness and pleasantness tasks. With respect to gender, there were significant differences between the ratings of male and female respondents of Groups 3 (Wilks’s lambda = .6965; df = 24; p < .05) and 7 (Wilks’s lambda = .6075; df = 28; p < .05) for the correctness task, and for Group 3 (Wilks’s lambda = .6705; df = 23; p < .05) for the pleasantness task. It is interesting to note that although only two of the eight groups showed a significant difference between men and women’s ratings for the correctness task, these two groups make up 71% of the cities (52 of 73). Overall, there were no interactions between the three independent factors of gender, age, and class for the correctness task and only a few for the pleasantness task (to be discussed below). The univariate analyses of variance (ANOVAs) run simultaneously with the MANOVAs for each city revealed significant differences between male and female respondents for thirty-one of the seventy-three cities for the correctness task and for sixteen of the seventy-three cities for the pleasantness task. Although these results do not appear to provide robust evidence for significant overall male and female differences in the ratings of the Turkish cities, a further look confirms that such gender differences are real and consistent: The results point to a strong tendency for the male respondents to give higher ratings for correctness and pleasantness than the female respondents. First, for each of the thirty-one cities for which there was a significant difference between male and female respondents for correctness, males gave the higher scores. The same was true for all sixteen cities for which there was a significant difference between men and women for the pleasantness task. Second, overall, the mean scores for the male respondents for both tasks were higher than the mean scores for the female respondents. Figure 17.4 illustrates the mean scores for the male and female respondents for the two tasks. If we compare the mean scores of men and women for each of the cities in the two tasks, the overall mean scores represented in Figure 17.4 are confirmed. For the correctness task, men had a higher mean score than women for sixty-three of the seventy-three

275

THE PERCEPTION OF TURKISH DIALECTS

cities. For the pleasantness task, men had a higher mean score for sixty-one of the seventy-three cities. 4.20

male female

4.10

4.00

3.90

3.80

3.70

3.60

correct

pleasant

Figure 17.4. Mean score, by gender, for correctness and pleasantness tasks

What is perhaps more interesting is that of the ten exceptional cities for which women had higher mean scores than men for correctness, eight were far eastern cities with low ratings from both groups (the other two were Istanbul and Cyprus). Similarly, of the twelve exceptional cities for which women had higher scores than men for pleasantness, nine were also cities in the east (three of the remaining four were Black Sea cities, and the fourth was Cyprus). For several of these eastern cities, the overall gender pattern is reversed for both the correctness and pleasantness tasks — for example, Bingol, Mardin, Siirt, and Tunceli. 5.1.2 Social Class There do not seem to be any class-related differences in the ratings of the speech of Turkish cities for correctness and pleasantness (although this study looked only at middleclass and upper-middle-class groups).

276

IMAGES, PERCEPTIONS, AND ATTITUDES

5.1.3 Age With respect to age (Group 1 = 17–25; Group 2 = 35–45; and Group 3 = 50+), the MANOVA revealed significant differences for Groups 2 (F = 4.42; df = 2; p < .05)7 and 3 (Wilks’s lambda =.5327; df = 48; p < .05) for the correctness task, and for group 5 (F = 3.92; df = 2; p < .05) for the pleasantness task. The simultaneous univariate ANOVAs revealed that for the correctness task there was a significant difference between the three age groups with regard to their ratings of sixteen of seventy-three cities, and for the pleasantness task, there was a significant difference for twenty-eight of the seventy-three cities. Although these results appear to indicate that the three age groups overall did not provide significantly different ratings of the Turkish cities for the three tasks, further comparison of the mean scores reveals several interesting patterns. First, for almost every city for which there was a significant difference between the age groups in either task, the oldest group scored highest and the middle-aged group scored lowest. For the correctness task, this was true for thirteen of the sixteen cities for which there was a significant difference; for the pleasantness task, this was true for twenty-three of the twenty-eight cities. (For the eight exceptional cases in the two tasks, the youngest group always scored highest.) Second, it appears that the older respondents (those in the 50+ age group) provided higher ratings overall for correctness and pleasantness than did the other two age groups. In the correctness task, the 50+ age group scored higher than the two other groups for sixty of the seventy-three cities. In the pleasantness task, the 50+ group scored higher than the other groups for fifty-four of the seventy-three cities. It is interesting to note that for all of the cities in both tasks, if the 50+ group did not score highest, the youngest group (17–25) gave the highest ratings. The middle-aged group (35–45) consistently gave the lowest scores for the correctness and pleasantness tasks of all three groups. There was only one city for either task for which the middleaged group scored higher than both other groups (Artvin, for correctness). In addition, there were only fourteen cities of the seventy-three for the correctness task for which the middle-aged group scored even second highest. For the pleasantness task, the results are even more striking in that there was only one city of seventy-three for which the middleaged group did not score lowest of all groups. These results indicate that overall the middle-aged group was the most conservative, giving lower scores than the other groups for both tasks but especially for the pleasantness task. Furthermore, the 50+ group tended to give higher ratings than the other groups for both tasks. The overall mean scores accentuate these points, and are illustrated in Figure 17.5. 5.1.4 Age and Gender Now consider several figures that illustrate the interaction of age and gender. Figure 17.6 shows age and gender results for the correctness task. Female respondents at each age level score lower than male respondents of the equivalent age group. In Figure 17.7, age and gender results for the pleasantness task are displayed, with the same result. In summary, examination of the mean scores of the various subgroupings of the Bursa respondents along gender, age, and class lines has revealed several patterns. First, the Turkish male respondents tended to give higher ratings than females in evaluating the

277

THE PERCEPTION OF TURKISH DIALECTS

pleasant

4.40

correct 4.30 4.20 4.10 4.00 3.90 3.80 3.70 3.60

17-25

35-45

50+

Figure 17.5. Mean score, by age, for correctness and pleasantness tasks

4.40

male

4.30

female

4.20 4.10 4.00 3.90 3.80 3.70 3.60 3.50

17-25

35-45

Figure 17.6. Mean score, by gender and age, for correctness task

50+

278

IMAGES, PERCEPTIONS, AND ATTITUDES

4.50

male female

4.40 4.30 4.20 4.10 4.00 3.90 3.80 3.70

17-25

35-45

50+

Figure 17.7. Mean score, by gender and age, for pleasantness task

correctness and pleasantness of the speech of the Turkish cities. Where these patterns are reversed, it is almost always the case that a city from eastern Turkey is being evaluated. Next, with regard to age, it appears that the oldest respondents (50+) tended to give the highest ratings for correctness and pleasantness, whereas the middle-aged group (35–45) consistently gave the lowest ratings. Finally, there were no class-related differences for either of the tasks. 5.2 Discussion 5.2.1 Gender Given the work in variationist sociolinguistics carried out in the past few decades, the differential distribution of responses according to gender is not surprising. In fact, the results of this study are consistent with many of the now well-established generalizations regarding the role of gender in language variation. First of all, researchers have claimed that across many cultures women tend to be more status-conscious and conservative than men with regard to language use (e.g., Trudgill 1974), and this leads female speakers to use a higher percentage of standard variety forms than male speakers of the same social class and age. On the other side of this claim is the assumption that men are more attracted to the “covert prestige” of working-class language and therefore are drawn more toward nonstandard forms in language use. An alternative (but not necessarily mutually exclusive) explanation of these gender-based differences in language use is that men

THE PERCEPTION OF TURKISH DIALECTS

279

usually belong to denser social networks than women and therefore feel more pressure to conform to group vernacular norms (Milroy 1980). In any event, these alternate explanations are meant to account for gender-based differences in actual language use. On the other hand, less research has been done comparing evaluations of language varieties by men and women. Trudgill’s study of selfreporting (1972) compared respondents’ actual language use with what they believed they did concerning particular linguistic features. He found that female respondents claimed that they used more of a standard variety feature than they actually did, whereas male respondents claimed less use of standard features than they actually produced. Trudgill then attributed these differences to overt and covert prestige. That is, since women tend to place a higher value on the use of more standard forms, they want to believe that they in fact use such forms, even when this is not really the case. Men tend to place a higher value on nonstandard working-class forms, and thus they tend to want to believe that they use more of these forms than they actually do. Moreover, very little sociolinguistic research has been carried out to compare the conscious attitudes and evaluations of men and women toward regional varieties. If the generalizations concerning gender differences in language use hold true, then one would expect that some of these underlying differences would also leave their mark on the evaluations of regional varieties. Our results confirm that this is indeed the case. For sixtythree of the seventy-three Turkish cities evaluated by respondents, men gave higher ratings overall for correctness than women, indicating that the female respondents were more critical of regional varieties along this dimension. The lower ratings given by women point to more stringent standards of acceptability, whereas the higher ratings given by men point to a greater acceptance of nonstandardness and a less critical stance in general. The same holds true for the results of the pleasantness task: For sixty-one of the seventy-three cities, male respondents gave overall higher ratings than female respondents. Again, this indicates a more critical perspective on the part of women toward regional varieties along the affective dimension of pleasantness and a more relaxed standpoint for men. Clearly, these gender differences are only tendencies, and the actual differences between men and women in their evaluations of regional varieties are slight. Nonetheless, it has been demonstrated here that the same underlying gender differences that lead to production differences in language use also appear to lead to differences in the evaluations of regional varieties. Still, the exceptional cases in which the gender patterns are reversed for correctness and pleasantness need explanation. Recall that female respondents tended to rate the southeastern cities higher than males for both correctness and pleasantness (whereas for all other cities this pattern is reversed). Figure 17.8 shows the results of the correctness task for each subgroup in our study for a block of ten southeastern cities.8 It is apparent from Figure 17.8 that the middle-class male respondents behaved in very striking ways compared with respondents from other groups. First, both young and old middle-class male respondents gave very high ratings to these eastern cities relative to the other subgroups. Second, the middle-aged middle-class male respondents gave very low scores relative to all other groups. We think that these middle-aged middleclass men are the culprits in reversing the gender-based pattern that predominates for all

280

IMAGES, PERCEPTIONS, AND ATTITUDES

other Turkish cities. It seems that this subgroup has very strong negative attitudes toward the speech of eastern cities, which might be a result of their political awareness of and opposition to the Kurdish people in this region. Finally, it appears to be the case that the Class × Gender interactions revealed for Groups 5 (F = 6.87; df = 1; p < .05) and 8 (Wilks’s lambda =.6869; df = 26; p < .05) by the MANOVA test for the pleasantness task can be explained in part by the performance of these middle-class males, whose ratings are rather extreme compared with other groups, at least with regard to the eastern cities. (Groups 5 and 8 are located in the eastern part of Turkey.)

Figure 17.8. Mean scores for eastern cities, by group, for correctness task

5.2.2 Age With regard to the differences found among the three age groups, recall that in the results for both the correctness and pleasantness tasks, the older group (50+) tended to give the highest ratings, followed by the youngest group. On the other hand, the middle-aged group lagged far behind the other two groups and consistently assigned the lowest ratings for correctness and pleasantness. The distribution of mean scores across age groups depicted in Figure 17.5 points to a well-documented age-related phenomenon called “age-grading.” Age-grading has to do with a recurrent pattern of orientation toward vernacular and standard norms, whereby young people orient toward the vernacular, middle-aged people orient more toward the standard variety because of the pressures of having to succeed in the working world, and older people, having left the workforce, once again orient more toward vernacular norms. That the youngest and oldest age groups perform similarly with regard to correctness and pleasantness and that the ratings for each city dipped much lower for the middle group is strong evidence that what is involved in our study is age-grading. The youngest and oldest Turkish respondents were less critical because, for them, less is at stake in upholding

THE PERCEPTION OF TURKISH DIALECTS

281

standard norms. On the other hand, the middle-aged group assumes a more critical stance, since for them more is at stake in maintaining standard norms. That age-related patterns may be involved in evaluations of regional varieties has been suggested by Preston (1988c), who found that evidence for attitudinal change may be gleaned from the comparison of age groups. In addition, he found instances of agegrading in the ratings of a few states; our results give evidence for very pervasive agegrading in the ratings of almost all Turkish cities.

6.

Conclusion

We hope to have shown in the second part of this chapter that Labov’s claim that attitudes toward language varieties “do not emerge in a systematic form if respondents are questioned directly about dialects” is not necessarily true. Specifically, we have identified several typical, very systematic, sociolinguistic patterns in our data, derived from tasks involving the direct questioning of respondents about dialects. It appears to be the case that gender and age, at least, are influential social factors, not just in language production but also in conscious language evaluation. These results confirm Preston’s claim that studies in perceptual dialectology may supplement sociolinguistic dialectology in illuminating and valuable ways since they may provide additional insight into the interaction of conscious attitudes toward language, social differentiation, and language production.

Notes 1. We would like to express our gratitude to Mahide Demirci’s family, especially her sister, Fahriye Osmanoglu, who collected most of the data. Thanks also to Dennis Preston for his guidance and helpful suggestions and to Gary Cook for advice on statistics. 2. The social class of each respondent was based in part on income, occupation, and neighborhood. 3. Although the K-means analysis does not select groups based on mean scores, the mean scores of each group are provided in Figures 17.1, 17.2, and 17.3 to illustrate the general progression of ratings from higher to lower as one goes from west to east. 4. In fact, it is always the case that the cities scoring higher for pleasantness than correctness are eastern cities, and the cities scoring higher for correctness than for pleasantness are western cities. 5. It is possible, however, that negative evaluations of Southern speech, for example, might be informed somewhat by stereotypes of the South as being poor, rural, and so on. 6. Even in the United States, however, Preston (1989) reports distinctive treatment of New Mexico, presumably because many believe that Spanish is extensively spoken there. 7. Group 2 in the correctness task is composed of only one city (Kirklareli), and therefore a univariate ANOVA test was run for this group. The same holds true for Group 5 (Kayseri) for the pleasantness task. 8. These cities were arbitrarily selected. However, extending this block of southeastern cities does not significantly alter the results. The labels along the x axis of Figure 17.8 should be read as follows: MC = middle class, UC = upper class, f = female, m = male, Y = young (17–25), M = mid (35–45), and O = old (50+), so that, for example, MCmO should be read “middle-class, male, 50+,” and UCfY should read “upper-class, female, 17–25.”

C 18 Regional Variation in Subjective Dialect Divisions in the United States Donald M. Lance

When professors or news sources refer to the Southwest, the South, the Northeast, or the Midwest, what kinds of maps appear in the heads of their listeners and readers? These mental maps are determined to some extent by where the listeners and readers grew up. This article discusses the results of an investigation of perceptual dialectology similar to research reported by Preston (1986, 1988b, 1993b).1 It is based on data from 268 students enrolled in university classes in nine states: Massachusetts (12), New York (26), Pennsylvania (19), Ohio (36), Georgia (21), Alabama (19), Missouri (66), South Dakota (42), and Washington (27).2 The students had not yet had formal instruction in American dialectology when they drew the maps on which this article is based. Each participant was provided a worksheet to use in the project; it had a map of the United States with only state lines on the top half and the following instructions on the lower half: Draw lines to show the locations of dialect regions in the United States. Also indicate the names that are commonly used for these areas, or the names that you would use. You may make up your own names. Please be serious in your answers. An exercise similar to this one has been done in Michigan and in Missouri. We want to get responses from people in various parts of the country so that a composite map may be compiled. Do not do any “research” as you draw your lines. We want to know what your notions are before you have done any formal study of American dialects.

Although the instructions refer to “dialect regions,” and the participants used terminology indicating that they were thinking about dialect or accent, much more than memory of others’ speech affected the placement of lines on the page: (1) personal knowledge about dialects and their locations, (2) knowledge of geography, (3) knowledge of cultural and ethnic diversity in the United States, and (4) experience and/or ability in dividing twodimensional space on a sheet of paper. The major portion of this chapter, focusing on (1)

284

IMAGES, PERCEPTIONS, AND ATTITUDES

and (2), will compare the locations and names of the putative dialect areas indicated by lines drawn by the nine sets of participants, beginning with the northeast corner of the country. The issues in (3) and (4) will be discussed at the end of the article, followed by a comparison of the students’ maps and the results of research conducted by professional dialectologists. In analyzing the maps, I scanned through each set of student maps several times to find recurring patterns (e.g., areas in the middle of the country with labels such as “midwestern”). In tabulating the states to be included in a given area, I counted the state if the lines demarcating the area included a major portion of the state or a historically or dialectally significant section of it. Some maps had no useful notations, some had one or two, and some had three or more. The mapping software did not allow me to select only a portion of a state in my tabulation. If a student placed a label on a map but did not clearly demarcate the entire perimeter of the area in question, I did not presume to know where the student would have drawn lines and then did not use that portion of the map. Although Alaska and Hawaii were included on the maps generated by the software, they are not included in the analysis. In some sets of maps, the respondents indicated their place of birth and where they had lived most of their lives. When this information was available, I did not use maps of those who had not lived a substantial amount of their school years in the state or area where they were studying. In the sets for which this biographical information was missing, I excluded maps that were substantially different from the majority of the set or that displayed patterns that analysis had shown to be typical of responses from other states.

1.

New England and the Northeastern Dialect Area

This section discusses the participants’ designation of the six-state area east of the Hudson River as New England and their selecting an area larger than New England that in the present study is called the Northeastern dialect area. Seven of the nineteen participants attending college in Springfield, Massachusetts, were not from New England and thus are not included in the analysis here. They were from various parts of the country and drew maps that were very different from those of the twelve natives. Six of the twelve whose maps are analyzed here clearly indicated a New England area east of the Hudson River consisting of Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, and Connecticut, and another respondent included New York State in New England. Unlike those from other states, the Springfield College students did not designate an area equivalent to what will be discussed below as the Northeastern dialect area. Eight of the twelve, in some way, indicated that dialects are different in northern (ME, NH, VT)3 and southern New England (MA, RI, CT), and one pointed out that there are different dialects in eastern and western Massachusetts. Several students marked other dialectal subdivisions in New England, but not consistently. Seven referred to “Boston accent” and five to the speech of one or more boroughs of New York City. As we shall see below, participants who do not live in New England have other

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

285

perceptions, with differences increasing as distance increases. Ten of the twenty-six participants from Potsdam College in Upstate New York, about seventy-five miles west of Lake Champlain, clearly labeled the six states east of the Hudson as New England, whereas eight others circled only the northern portion of these states (ME, NH, VT), calling this area New England but not labeling the other three states. Eleven participants designated a somewhat larger area that some of them labeled Northeast and others Northern, five of them including DE and MD (south of the historical MasonDixon Line). Figure 18.1a displays these data.4 Because there was no consistent pattern in the geographical distribution of areas called Northern and Northeastern in the data in this study, the label Northeastern is used here so that Northern may be used for the area that will be discussed next. Seven of the New York students referred to dialectal subdivisions within their state; although no two of them had the same lines and labels, most of them recognized that the speech in the part of the state where Potsdam is located is different from that in other areas. Fifteen of the twenty-six participants also made note of dialects in New York City and/or Long Island. Ten of the nineteen participants attending Bloomsburg University in east-central Pennsylvania labeled the six states east of the Hudson as New England, two of them including a small portion of New York State. Five others used this regional label for only the northern portion of New England. Eleven included NY, PA, and NJ and the six New England states in the same area, only one student using the term Northeastern. A map of these eleven responses would be very much like Figure 18.1a. Like their counterparts to the northeast, eleven Pennsylvanians were aware of subdivisions within their own state; although none of them marked all of the areas, several students recognized these divisions: Philadelphia (4), Northeast (1), Coal Area (4), Pennsylvania Dutch (3), Pittsburgh (3), and east versus west (5, without labeling the areas). Four of the thirty-six participants attending Ohio University in southeastern OH labeled the six states east of the Hudson as New England, and two others included NY in New England. Twenty marked an area from ME to PA as the Northeastern dialect area, although there was some variation in which states were included. Their names for this area were New England (8), Eastern (6), Northeastern (3), East Coast (1), New York (1), and Yankee (1); Figure 18.1b displays an analysis of these twenty maps. Proportionately more Ohioans than New Yorkers placed DE and MD in the Northeast, and one added WV and VA. Five of the twenty-one participants attending the University of Georgia designated the six states east of the Hudson as New England; three others considered New England to include NY. Sixteen marked an area that included the six states plus a larger area, with three regional labels: Northeastern (9), Northern (5), New England (2). See Figure 18.1c for an analysis of these sixteen maps. Four students noted that the dialect in ME is different from that in other parts of the Northeast. The participants made no reference to subdivisions within any of the Northeastern states — evidence of lack of familiarity with the area. They made relatively few references to specific cities (NYC 4, DC 1), one placing New York City in Delaware. Only two of the nineteen participants attending Auburn University in Alabama drew a line recognizing the six New England states as a dialect region, and only six designated

286

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 18.1. New England and Northeastern dialect areas

an area that could be considered Northeastern, with no patterns that would motivate preparing a map. Moving away from New England in the analysis, one sees an increase in the size of the area regarded as Northeastern. Of the sixty-six participants from Missouri, only five drew a line around the six states east of the Hudson, six others using the term New England for only a portion of the six-state area and five applying the name to the six states plus either NY or NJ. Like their counterparts in the Southeast, the Missourians used several names for the area being discussed here: Eastern (22), Northeast (13), New England (8), East Coast (3). Five labeled both a six-state New England area and a Northeastern or Eastern area. Figure 18.1d summarizes the data from forty-four Missourians’ maps with clear lines and labels indicating the existence of a Northeastern dialect area. Five of the forty-two participants attending the University of South Dakota drew a line indicating that New England consists of the six states east of the Hudson, and another added NY. Like the Missourians, the South Dakotans used a variety of names for the area from ME to IL: Eastern (13), Northeastern (6), New England (5), East Coast (5), Northern (1), or a less informative label, if any (6). The map for the South Dakotans’ Northeastern area would be similar to that of the Missourians, with Illinois added to the other states in Figure 18.1d. Only one of the twenty-seven participants attending Eastern Washington University,

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

287

near Spokane, indicated that New England consists of the six states east of the Hudson, labeling this dialect “proper English.” Several names were applied to the set of states treated here as Northeastern: New England (7), East Coast (6), Eastern (5), Northeastern (3), or less informative labels (4). Figure 18.1e is a tabulation of the twenty-two Washington maps with clear lines and labels. The student who had a complimentary name for the dialect of the New England states applied the label “New York/Jersey Drawl” to the seven states bounded by NY, DE, WV, and OH. One Washington student extended the area to include MN, but none of the Washington students included states south of Virginia, as did participants from GA, MO, and SD. One might wonder whether students from MO, SD, and WA truly had New England speech in mind as they drew lines including states such as Illinois and Michigan in the Northeast, but their use of labels such as “Brooklyn,” “Jersey Dialect,” and “Northeastern short dialect” for the area from ME to WI indicates that some of them believe the speech of Wisconsin to be similar in some ways to that employed in the metropolitan New York area and in New England as well. The Missourians who included North Carolina (5), South Carolina (1), Michigan (2), and Wisconsin (1) in the Northeast very carefully drew their lines, one of them cross-hatching the region, and they used either the name Eastern (7) or Northeastern (2) for the designated areas. Omitting these responses from the report of results would deny these students the seriousness of their efforts, whether they were actually remembering specific information about regional accent or were making a sincere effort to divide the map space into coherent geographic areas that manifest at least a few dialectal similarities — although perhaps without thinking about the names of the states they were choosing. Including the data is more informative than omitting it.

2.

Northern and Midwestern Dialect Areas

Although participants frequently used the term Northern to refer to the dialect area we are here calling Northeastern, the term was also frequently applied to the tier of states across the middle of the country next to the Canadian border. The putative existence and location of a Northern dialect area depends on how and where participants from a given state perceive Midwestern and Northeastern dialect areas. 2.1 Northern Students in six of the nine states perceived a Northern dialect region in the two or three tiers of states along the Canadian border, with students in one state (AL) extending it eastward through New England and three (NY, OH, MO) extending it through the Northwest to the Pacific Ocean. Students from three states did not mark a clear Northern dialect area (PA, SD, WA), and it is of interest that they would be perceived by other participants in this study as speaking a Northern dialect. Seven of the twelve Massachusetts students outlined a dialect area surrounding the Great Lakes to which they gave several names: Northern (3), Minnesota Accent (1),

288

IMAGES, PERCEPTIONS, AND ATTITUDES

Midwest-Northeast Mix (1), Northeast (1), Mid-Central Dialect (1). This area is shown in Figure 18.2a. Many of them called the speech of the states immediately to the west of MN and IA Midwestern. Three of the twenty-six participants from New York used the terms North or Northern to refer to New England and neighboring states, and ten used the term Northern to refer to various patterns in several groups of states extending across the country. Six applied the term to states extending westward from New York and Pennsylvania, but the western terminus of the area varied greatly. Three applied the term to an area bounded by MI-IN and ID-CO, and one to a smaller area bounded by MN, MT, UT, and MO. Figure 18.2b displays a combination of all states marked as Northern by these ten participants. Only two of the nineteen Pennsylvanians indicated a Northern area, one placing the word but no lines in the Northwest. When they circled an area that included states along the Canadian border, they called it Midwestern, Great Lakes, or something else — not Northern — but no clear pattern emerged in the analysis. Of the thirty-six Ohio participants, two used the term Northern for the area treated here as the Northeastern. Seven drew lines around an area in the middle of the country, applying the term Northern (Figure 18.2c). We see that the core of this area is the same as that of participants from Massachusetts and New York, although the expansions east and west are somewhat different in each of these three maps. Eleven of the nineteen Georgians labeled a Northern area, four of them designating the area treated here as Northeastern. The other seven Georgians, as shown in Figure 18.2d, had a rather clear notion of a Northern area along the Canadian border from Ohio to Montana, although one extended the area farther south than other students did. We will later see in the discussion of Figure 18.3h that there is some ambiguity in the Georgians’ perceptions of the locations of Northern and Midwestern dialect areas. The Alabama participants had a much clearer differentiation than the Georgians between the Northern and Midwestern areas. Eleven of the nineteen Alabama participants clearly marked a Northern area, from Minnesota through New England, and six marked a Midwest from AR and IA to MT, as we see in Figures 18.2e and 18.3g. Their Northern area extends through New England, only two Alabamians having noted the New England dialect area. Five Alabamians used the term Yankee to designate the speech of residents in the northeast quadrant of the country, and another indicated that residents in the Midwest speak “slow Yankee dialect.” One student, without drawing lines, wrote “Yankee, sharp” to the right of Cape Cod, and another designated the coastal area from NJ to Cape Cod as “Damn Yankee.” Twenty-nine of the sixty-six Missourians delineated an area along the Canadian border as Northern. As we see in Figure 18.2f, this area consists of the states from MI to ID, with the core being WI, MN, ND. Five designated a North Central dialect with the same core as the Northern area. Not one of the sixty-six Missourians labeled the Northeastern area as Northern. Only four of the forty-two South Dakotans marked a Northern area, one using this term for the Northeastern area and the other three marking an area consisting of nine states from WI to WA-OR. Only two of the twenty-seven Washington students labeled a

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

Figure 18.2. Northern dialect area

289

290

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 18.2. Northern dialect area (continued)

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

291

Northern area, neither with a convincing distribution; however, twenty Washington students and twenty-one South Dakota students designated a Midwestern area that overlaps areas that others labeled Northern. 2.2 Midwestern All Americans know where the Midwest is — right? It’s in the middle of our big country, where the regional accent is slight, an accent appropriate for television newscasters. One Georgia student drew a circle around NE, KS, CO and labeled the area “Newscasters.” The present study found far more variation in the location of the core, as well as periphery, of the Midwestern dialect area than of any other area. Either WI or MN is at the core of all participants’ Northern dialect areas in Figure 18.2, but no similar claim can be made for the Midwest. There appear to be four views on the location of the core of the Midwest dialect area: (1) the Old Northwest (OH, IN, IL), a view held by one set of students from WA (Figure 18.3c) and by the students from GA (Figure 18.3h); (2) the states settled in the second and third quarters of the nineteenth century to the west of the Old Northwest, the view of students from PA (Figure 18.3f), MO (Figure 18.3j), SD (Figure 18.3k), and a group of students from NY (Figure 18.3a); (3) “the middle of the West,” the view of groups of students from NY (Figure 18.3b), WA (Figure 18.3d), and MA (Figure 18.3e); (4) and an area encompassing all or major parts of the preceding three areas. The first four maps in Figure 18.3 demonstrate that in NY and WA, some students hold one of these four views on where Midwestern accents are used, whereas others hold a different one. Six of the New York students designated a set of states with the core in the middle of the general area under consideration, and six designated an area in the western part of the area (Figures 18.3a and 18.3b). Three Midwestern patterns are found in the Washington data: six participants marked an area centered on the Old Northwest (Figure 18.3c), ten marked an area centered on ND-NE (Figure 18.3d); four included both of these areas but are not included in these maps. A student from Kennewick carefully outlined an “Eastern Mid West” area in the Old Northwest and “PNW [Pacific Northwest] to Midwest” from the Mississippi River to the Pacific; the latter map was not included in Figure 18.3 because it admittedly included the Northwest area. Another student, from near Spokane, carefully drawing lines around an area from OH to MT and NM, made the notation “Midwest combinations area.” These and other notations on the maps indicate an awareness of considerable variation in what the students regard as Midwestern speech. Six of the participants from Massachusetts used the term Midwest, with five of them demarcating a “middle of the West” location of the Midwestern dialect, as we see in Figure 18.3e. One participant designated only three states (PA, OH, IN) as the Midwest. The patterns for Northern in Figure 18.2a and Midwestern in Figure 18.3e are complementary, with little overlap. Although one might wonder why a Massachusetts student would place WA and OR in the Midwestern area, one must remember that the student was responding to a question about dialects — where speakers live who share certain features of regional accent or share a lack of other features.

292

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 18.3. Midwest dialect area

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

Figure 18.3. Midwest dialect area (continued)

293

294

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 18.3. Midwest dialect area (continued)

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

295

Fourteen of the nineteen Pennsylvania students indicated a Midwest area, but two did not draw lines. Nine of the students (Figure 18.3f) outlined a Midwestern area with a core from IL to SD-NE-KS. Three others marked a “middle of the West” area with a core bounded by SD, KS, UT, WY. Thus, we see that students from our three states in the Northeast and the state in the Northwest have at least two views of where Midwestern accents are found. Only seven Auburn students clearly outlined a Midwestern dialect area, all but one of them using that name. As we see in Figure 18.3g, the core of the Midwestern area (NE, KS) is in the Plains States, with more than half of the seven extending the area into OK (5) and AR (4). In Figures 18.3g and 18.3h, we see that students from the two Southern universities in this study, merely 160 miles apart, have rather different views of where the Midwestern dialect is found. Georgians see it surrounding the states of the Old Northwest, and Alabamians associate it with the states surrounding NE-KS. Each group’s geographic placement of Midwest speech is influenced by where they perceive the Northern dialect to be spoken (Figures 18.2e and 18.2d), which in turn is affected by their views of the location of a Northeastern dialect (see Figure 18.1c). Few Alabamians reported a Northeastern dialect, and they used the term Northern to refer to the dialect of the Old Northwest; consequently, the Midwestern dialect for them would have to be spoken west of the Mississippi River, but it is not as far west as “the middle of the West.” Georgians perceive Midwestern speech as being located in the Old Northwest immediately below their Northern area, so one should not be surprised that they tend to use these two names interchangeably. Six of the eleven Georgians who marked a Midwestern area did not mark a Northern area, and four of the nine who marked a Northern area did not mark a Midwestern dialect. Of the thirty-eight instances of a Georgian placing a given state in a Northern area without also naming a Midwestern area, and the sixty-seven instances of placing a state in a Midwestern area without also specifying a Northern area, eighty-two times (77%) the student selected a state to which a classmate in the other group gave the opposite label. This overlap may be reflected in a note on one of the worksheets suggesting that Georgia students had debated the appropriate name for the area including the Old Northwest and the Great Lakes States: “Hey yous [sic] guys this is the North,” with an arrow pointing to an encircled area bounded by MN, MO, OH, MI. Twenty-nine of the thirty-six Ohioans indicated a Midwestern area, with three failing to include lines and one calling it “Michigan dialect”; Figure 18.3i is a tabulation of the twenty-five maps clearly labeled Midwest. There was a tendency for some Ohioans to divide into two parts the area that many others would regard as Midwestern (five instances), but the names and locations — “Corn Country” and “Plains” — suggest that these divisions are something different from the New Yorkers’ alternately perceiving a “Midwest” and a “middle of the West.” The core of the Ohioans’ Midwestern area is similar to that of their neighbors to the east, but a few Ohioans extended Midwestern speech into areas to which most others did not apply that name, notably east of Ohio and west to the Pacific Ocean. One may also note that fewer Ohioans placed their own state

296

IMAGES, PERCEPTIONS, AND ATTITUDES

(21) and IN (22) in the Midwestern dialect area than IA (23) and IL (25), although it is not possible to divine a reason from the students’ maps for why some Ohioans did not consider their own dialect to be Midwestern. Forty-seven Missourians drew lines around a Midwestern area. Four others used the term Midland and another the term Central. These fifty-two respondents clearly see five states as the core of the Midwestern speech area (Figure 18.3j): MO 46, IA 44, KS 43, IL 42, NE 41. In sixth place is IN (29), and next come OK (20), OH (18), WI and SD (17), MI and MN (16). Like students from other states, a few Missourians indicated that the speech of such seemingly disparate locations as ID, TX, and VA are not sufficiently different from Missouri speech to be placed in a different dialect area. Twenty-one South Dakotans designated a Midwestern area, one calling it Mid America; these responses are mapped in Figure 18.3k. As we have seen in Figures 18.3i and 18.3j for OH and for MO, participants in a state that considers itself to be part of the Midwest will have the core of the dialect area in or near the state where they live. All twenty-one participants from South Dakota placed IA and NE in the Midwest, with SD coming in third (19), followed by MN and ND (17). The South Dakotans did not expand this dialect as far beyond the core area as did respondents in OH, GA, and MO. When one looks at the states designated as Midwestern in Figure 18.3 with a count of less than 20% of the total, one is tempted to say, “Hold on! Those are not in the Midwest!” Nevertheless, the care that was taken, for instance, by the students who drew lines around TX or across a large portion of the state, by those who carefully included UT, and by those who drew bold lines around AR indicates that these students had something specific in mind as they included these states in the Midwest dialect area, and omitting their responses would suppress relevant data. As we have examined the maps of Ohioans, Georgians, Alabamians, and Missourians, we have seen a perceptible increase in the number of these “non-Midwestern” states circled by a few students. Because many participants wrote comments such as “not much accent,” “no dialects here,” “Normal,” and “neutral speech” in the areas designated by others as Midwestern, we might reasonably assume that what the participants were claiming by including the western two-thirds of TX or OR or AZ in the Midwestern dialect area is that one does not hear strong regional accents in these states.

3.

Southern Dialect Areas

Because the South as a geographical region has a complex history and most Americans are aware of dialect areas within the South, at least four categories are needed to present the findings of this study. A substantial number of participants from all states except MA designated a Southern dialect area, and large enough numbers to justify separate maps also outlined Southeastern and Mid-Southern areas. Some students also indicated an awareness of uniqueness in the speech of residents of Appalachia, but they were few, and the majority of their maps were geographically inaccurate.

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

297

3.1 Southern and Southeastern There is less variation in the location of the core of the Southern dialect area than the core of any other area in this study. Rather than include eight maps that look very much alike, in Figure 18.4, we have the map with the fewest states (AL), the one with the most (SD), and a map that combines the totals of all participants who marked a dialect area to which they gave the name South or Southern. At least 70% of the participants in each state placed the six-state area of NC-SC-GA-TN-AL-MS in the Southern dialect area, but there was considerable variation in the periphery. Thirty-seven percent (7/19) of the Alabama students included TX in the Southern area, whereas 60% (25/42) of the South Dakotans did so. Seventeen percent (7/42) of the South Dakotans included AZ in the Southern area, whereas only four percent (10/226) of the participants in the other eight states did. One student from SD placed NE in the Southern area, and two from SD and one from WA included CO. Ten of the fourteen students who considered KS to have Southern speech were from SD, two from WA; the others who thought Coloradans speak a Southern dialect were from MA and GA. The only group in which no one placed DE and MD in the Southern dialect area was the set from SD. Seventeen participants designated a Southeastern area, with its core in the same states as the core of the composite Southern map. Because of the ways in which the students drew lines and used the terms, we will here treat Southeastern as a subdialect within general Southern. Students from only five states used the term Southeast on their maps: PA (2), GA (4), MO (6), SD (1), WA (4). Figure 18.5 has four maps representing the responses from states in which at least five students indicated a Southeastern area; four students from GA, SD, and WA used no label or used a name other than Southeastern (e.g., “Hillbillies”) but were apparently designating a Southeastern dialect within the general Southern area and thus are included in the maps in Figure 18.5. The maps of the students from MA, MO, and SD have the same core area, with a fair amount of variation in the peripheral states, but the number of students represented in these maps is too low to justify much generalization. Participants from all nine states commented that dialects other than Southern are spoken in FL. Some made comments about “mixed dialects” or Northerners having settled in FL, but many simply drew the Southern or Southeastern boundary westward at the Florida state line. In Figure 18.4c, only 55% of those marking a Southern dialect area (154/232) included Florida, the same percentage as Kentucky. Those with the highest inclusion rate were WA (91%), PA (88%), and OH (72%); those with the lowest were MA (43%) and AL (38%). The state of Louisiana received special attention by fifty-four participants. Thirtyeight designated a portion or all of the state as speaking Cajun and nine as speaking Creole. The states with the highest number of participants referring to Cajun or Creole were AL (13), MO (12), and GA (8). Seven noted a New Orleans dialect, one of them (from MO) indicating that “New Orleans Cajun type accent” is spoken throughout the state. One NY student wrote “Cajun Creole” in the Missouri Bootheel.

298

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 18.4. Southern dialect area

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

299

Figure 18.5. Southeastern dialect area

Only four participants — two from MO and one each from OH and AL — referred to the speech of the Ozarks area of southern Missouri and northern Arkansas as a Southern dialect. 3.2 Appalachian Twenty-eight of the 268 participants designated an Appalachian area: MO, 11; OH, 10; GA, 5; AL, 1; NY, 1. No one from MA, PA, or WA marked an area that could be construed as representing the Appalachian speech area, but one student from SD drew a line around a portion of Appalachia and labeled it Hillbilly. Only the Ohio students manifested enough consistency and geographical accuracy to warrant the drawing of a map, but since the Appalachian area covers only parts of several states, it is more

300

IMAGES, PERCEPTIONS, AND ATTITUDES

informative to reproduce representative samples from individual maps. In Figures 18.6a-e the students have circled the mountainous area of the Southeastern states, with three of them extending their areas beyond the mountains and foothills — that is, beyond where one finds Appalachian culture and language. The students from whose maps Figures 18.6b and 18.6d were taken most accurately depicted this dialect area. Figures 18.6f and 18.6g are examples of instances in which the respondents knew that the speech of WV was Appalachian, but the remainder of their designated areas is substantially different from what we see in Figures 18.6a-d. The Missourian who drew the map in 18.6f was very likely thinking of the historical relationship between Appalachian and Ozark dialects.

Figure 18.6. Appalachian dialect area

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

301

3.3 Upper South As students drew lines around the Northeastern and Southern areas, the set of states that politicians call the Border States (KY to VA) received ambiguous treatment. Some participants included them in the Northeastern area, many placed them in the Southern area, and twenty-eight assigned them to their own dialect region, with a variety of names. Nine, six of them from OH, called the area Appalachian, four Mid(land) South, two Upper South, two “light Southern,” and one “Pseudo-South.” Seven Ohioans and nine Missourians marked the area, but no one from SD or MA did so. Figure 18.7 shows the distribution of data from the twenty-eight students who indicated an awareness of differences between the speech of residents of the Upper South and that of their neighbors. The core of the area is KY (27), VA (25), and WV (23), with TN (21) and NC (20) in fourth and fifth place. Two Missourians extended the area into southern MO and northern AR, and one from OH included all of MO and AR.

Figure 18.7. Upper South dialect area (all students)

3.4 Mid-Southern Is Texas drawl a Southern or Western dialect? In placing lines and choosing names, some participants in this project seemed to consider it Southern and others Western. This section will discuss maps in which the former label is used or implied, and maps with the latter label will be discussed in the section on the West. If a student marked a section in the central-southern part of the country and overtly indicated that the dialect is Southern, it was included in Figure 18.8. Also included are maps with a variety of names that designated a section including TX and the states of MO, AR, and/or LA. Twenty-one of the thirty-two participants who outlined the area analyzed in Figure 18.8 used the term South or Southern, only one using Southwestern. The students’ use of names like MidSouth(ern), Midland South, Midwest Southern, and West Southern motivated our use of

302

IMAGES, PERCEPTIONS, AND ATTITUDES

the term Mid-Southern here. Only the SD participants (with ten maps), and perhaps NY (6) and WA (5), marked the area often enough to justify a map, but it seemed preferable to prepare a composite of all thirty-two responses so that the states on the periphery would be included. Responses from all nine states were very much alike, with students from WA, SD, and NY providing the low-frequency periphery. Three students from WA, one from SD, and one from NY placed Colorado speech in the same area as TX, and the SD student also included the speech of NE with that of TX, labeling it “Harsh Southern.”

Figure 18.8. Mid-Southern dialect area (all students)

We see that TX and OK clearly occupy the core of this perceived area. These two states are also frequently associated with Western speech rather than Southern. Many students drew a line around TX alone or TX plus one or more surrounding states and labeled the area Texan or, in a few instances, Cowboy. These interpretations of Texas accent will be included in the following section of this article.

4.

Western Areas

Fewer students marked dialect areas in the Rocky Mountains and Pacific States than elsewhere, but there are some clear patterns. This section will discuss the Southwestern, Texas, Western, Northwestern, and West Coast areas.

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

303

4.1 Southwestern Forty-eight participants marked an area that they regarded as the Southwestern dialect area, shown in Figure 18.9. The core of the area is NM-AZ (42, 43), followed by TX (29). In the third level of frequency are states from OK to CA. In the peripheral area in this map, the states of MO, AR, and LA were all marked by participants from NY and SD, and an Ohio student included WY in the Southwest. One may note that the gaps between the levels of frequency in the legend of Figure 18.9 are larger than those seen in other figures in this article. Although there were enough respondents from WA, MO, and OH to justify separate maps, it seemed preferable to include all the responses in a single map, since the individual state maps would have looked very much alike.

Figure 18.9. Southwestern dialect area (all students)

4.2 Texas Eighty-one of the 268 participants (30%) gave special attention to the speech of Texas. Forty-seven circled the state and indicated that Texas speech was unique, and thirty-four drew a larger perimeter around an area that they labeled Texas or Texan (see Figure 18.10). Only one participant (from NY) stated explicitly that Texas speech is Southern; five others — from MA (1), GA (2), SD (2) — used the term Cowboy to refer to the speech of Texas and immediately surrounding states. These labels led to the placement of this section of the article within the Western area, along with Southwestern. The student who linked the

304

IMAGES, PERCEPTIONS, AND ATTITUDES

speech of TX and MS was from OH, and the one who extended Texas dialect to CO and KS was from WA.

Figure 18.10. Texas dialect area (all students)

4.3 Western An insufficient number of participants from MA, GA, and AL designated dialect areas in the Western States to justify attempting generalizations, but students from the other six states offered a variety of characterizations of the location of Western speech, as we see in Figure 18.11. In all six maps, we see that a substantial number of the respondents extended the area westward from MT-NM to the Pacific Coast, but there is considerable variation in where each group saw the core of Western speech. The respondents from NY and WA placed the core just to the west of their western Midwest (Figures 18.3a-d). The students who actually live in the West (WA) outlined a much larger area than did residents of the other states, with at least ten of the twelve respondents including nine of the eleven states in this dialect area. Missourians noted Northwestern and West Coast dialects more frequently than did respondents from other states, and as a consequence, fewer than half of the twenty students represented in Figure 18.11d included WA, OR, and MT in the Western dialect area. The students from PA seemed reluctant to assign specific regional dialects — not even Western — to the Pacific Coast states, but several commented on “California surfer talk.”

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

Figure 18.11. Western dialect area

305

306

IMAGES, PERCEPTIONS, AND ATTITUDES

4.4 Northwestern Forty-four participants marked a Northwestern dialect area. The greatest number noting the area were from MO (15), the fewest from AL (1), MA (1), and PA (2). Because the numbers from most states are low, we have combined the responses from all forty-four respondents in Figure 18.12. Not surprisingly, almost all the respondents placed WA and OR in the Northwestern area. Over half of them included ID and MT in the Northwest. A fourth of them included northern CA and the state of WY. All the students who extended the Northwestern area into UT, CO, ND, or SD were from MO or GA.

Figure 18.12. Northwestern dialect area

4.5 West Coast Thirty-four participants labeled a Pacific Coast or West Coast dialect area, as shown in Figure 18.13. Most of the students recognizing this area were from MO (19) and SD (8). No one from NY or WA designated a West Coast dialect area. Five students from different states included AZ. One from SD added ID, and one from AL added UT. The state of California received special attention in this project: 116 (42%) of the 268 participants either drew a line around all of the state and gave it a name, referred to different parts of the state by various names, or drew a line around a portion of the West and made reference to California in the name of the dialect of that area (11 individuals). The most common terms for California dialect were “Surfer Talk” (26) and “Valley Girl

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

307

Figure 18.13. West Coast dialect area (all students)

Talk” (15), the latter being a variety of speech associated with young females in suburban Los Angeles in the 1970s. Only seven participants overtly acknowledged that there was dialectal variety within the state.

5.

Awareness of Language Contact

Although speakers of English as a second language are numerous throughout the United States, only 75 (28%) of the participants acknowledged the presence of other languages that might have or definitely have influenced regional American dialects. Twenty-eight participants mentioned the presence of Spanish in Florida and the Southwest. It was not possible to determine whether the respondent was thinking of the presence of Spanish speakers or of the influence of Spanish on local English, but eighteen of the twenty-eight references to Spanish used phrases such as “Spanish accent” and “Mexican-American dialect,” suggesting that the student was referring to “foreign-accented English.” Of the

308

IMAGES, PERCEPTIONS, AND ATTITUDES

fifty who mentioned Cajun or Creole in Louisiana and neighboring parts of the South, only one was clearly referring to English (“french influenced Southern drawl”), but nine others used phrases that may be interpreted either way (e.g., “French Southerners”). Three of the twelve references to Native American languages in AK seemed clearly to refer to accented English (e.g., “Eskimo influences”). The references to Americans of Scandinavian (4) and German (1) ancestry were obviously referring to substratal influence in regional dialects of American English, because there are very few recent European immigrants in the Northern and Midwestern states where the notations were made. Although the topic of language contact is a tantalizing one for dialectologists, ambiguities in the brief entries on the student maps make it unwise to attempt broad generalizations regarding the students’ views of language-contact phenomena in American regional speech.

6.

Dividing the Map

When asked to place lines on a blank map as in the present exercise, students vary in their confidence in performing what is asked. Some of the maps received in this project had only a few lines and notations, whereas others were packed with information. Some of the variation in approaches to the task is shown in the maps in Figure 18.14. In 18.14a, the student elected to use straight lines to separate major divisions and then decided to make some revisions; because this student’s boundaries of Northeastern and Northern were so different from other Georgians’ maps, and the locations of lines seemed to be somewhat arbitrary, this one was not included in the tabulation. In Figure 18.14b, the student elected to comment on speech qualities (e.g., “string out their words — longer dialect”) as well as geographical regions, with areal quadrangles curiously leaning to the left (except one). Three of this student’s areas were very similar to the locations that other South Dakotans assigned to regional dialect areas and were included in the tabulations. “Jersey dialect” is approximately where other South Dakotans placed the Northeastern dialect, and several others used the term “Jersey.” Similarly, “Southern Drawl” and “Western dialect quickfast” approximate the locations of other South Dakotans’ Mid-South and Western. As an example of a student map that has more positive than negative qualities, Figure 18.14c has curved lines and regional labels similar to what the majority of the participants in the study submitted. Although “Mid-West” in this map may seem to have fortuitous southern boundaries, resulting from the student’s first choosing “Southern” and “South Western,” comments on numerous maps indicate that the students in this project tended to define Midwestern speech as not having qualities associated with neighboring dialect areas — that is, what remains after the more obvious regional dialects have been designated. In thinking about the dialect of her or his home area of Upstate New York and the surrounding region, this student commented, “These areas I don’t label as a dialect because nothing is different to me.” Similar notations were occasionally found on other maps. The most useful maps in this project had only three or four area distinctions that could be used in the tabulations. Apparently each participant was familiar with only a few

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

Figure 18.14. Examples of individual maps

309

310

IMAGES, PERCEPTIONS, AND ATTITUDES

of the dialect areas of the United States. Although the students’ maps in projects such as the present one pose some challenges for the analyst, in a set of a dozen or more maps, there will be enough similarities for reasonable interpretations and tentative conclusions.

7.

Comparison with Dialect Research Projects

How accurate were the dialect area designations reported in the preceding sections of this article? Rather than commenting on what the students did not know, one should compare their maps in a positive way with the findings of the many research projects conducted during this century on American regional dialects. Figure 18.15 contains a map based on research reported in seven studies that used procedures of the Linguistic Atlas of the United States,5 and Figure 18.16 contains a map based on an analysis of data from the files of the Dictionary of American Regional English.6 The areas in the top portion of Figure 18.15 and those in the bottom portion, although they appear to be compatible, should not be conflated because the informants in the project from which the latter derives represent a population that was born one or more generations later than the population represented in the former; as well, the research on which the terms “Interior Southern” and “Coastal Southern” are based included some modern vocabulary items that were not investigated by researchers who developed the other dialect area names. Also, one should not attempt exact correlations between the areas and terminology in Figures 18.15 and 18.16 because the latter used data and procedures that are substantially different from data and procedures represented in the former. With these caveats in mind, we will use the maps in Figures 18.15 and 18.16 to see the extent to which our students’ responses are similar to the findings of professional dialectologists. The location of New England as seen in Figure 18.1 matches Figures 18.15 and 18.16, and the core areas of the Northeastern and Northern dialect areas in Figures 18.1 and 18.2 are well within the Northern area of Figure 18.15 and the Upper North and Upper Midwest in Figure 18.16. The core areas of the Midwestern areas in Figure 18.3 are located in the North Midland area of Figure 18.15 and in the Upper North, Upper Midwest, and Central West in Figure 18.16. The core areas of the Southern and Southeastern areas are in the South and Coastal Southern areas of Figure 18.15 and in the Lower South in Figure 18.16. Except for Figures 18.4b and 18.5d, the frequencies reported for states in the South Midland area and the upper portion of the Interior Southern area of Figure 18.15 and the Upper South in Figure 18.16 are approximately 20% to 50% lower than the frequencies reported for the core Southern areas. These lower frequencies occurred either because these states were included in the Northeastern area or because they were set apart as the Upper South — an area equivalent to a major portion of the Upper South in Figure 18.16. The Mid-Southern area of Figure 18.8 is not reflected in either Figure 18.15 or 18.16. The Southwestern area in Figure 18.9 closely matches the West Texas and Southwest areas in Figure 18.16. (Because analysis of data from Linguistic Atlas studies in the Rocky Mountain and Pacific States is still under way, these areas are not included in Figure 18.15.) With the exception of Figure 18.11f, the core areas of the

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

311

Figure 18.15. Major dialect areas from Linguistic Atlas Studies5

students’ Western area are located in the Central West and the unidentified area to the north of it in Figure 18.16. The core of the students’ Northwestern area in Figure 18.12 is the same as the Northwest in Figure 18.16, but what the students claimed about the West Coast dialect (Figure 18.13) is not reflected in Figure 18.16. It would be interesting as a follow-up to add data from the states of CA, TX, CO, FL, VA, ME, WI, and possibly others to see if the additional data changed the locations of any of these core areas. How frequently did the students recognize dialect boundaries in the locations of the Northern/North Midland division in Figure 18.15 (and similar lines in Figure 18.16), the North Midland/South Midland division in Figure 18.15 (located farther south in Figure 18.16), and the Interior Southern/Coastal Southern division in Figure 18.15 (farther north in Figure 18.16)? In fewer than 7% of the maps. Participants tended to draw careful lines along state boundaries, somewhat vague circles or lines that did not coincide well with state boundaries, or boxes similar to those in Figure 18.14b, but about 40% seemed to have consciously drawn dialect boundaries across portions of one or more states. Three

312

IMAGES, PERCEPTIONS, AND ATTITUDES

NEW ENGLAND Western New England

NORTHWEST

NORTH

UPPER Central West

Northern California

ER LOW

Utah West

TH NOR

Hooner Apex

Southeastern Pennsylvania

Virginia Piedmont

Upper Atlantic South

Colorado West

H UT SO

Missouri Apex

Southern California

EASTERN NEW ENGLAND

ER UPP

Eastern North Carolina

Atlantic South

SOUTHWEST

LOWER West Texas

Northern Louisiana

TH SOU

Lower

Atlantic South

Alabama

Delta South Southeastern Louisiana

Southern Florida

Figure 18.16. Dialect layers based on data from the files of the Dictionary of American Regional English6

(in PA, OH, WA) drew a line across northern PA and westward through northern OH, IN, IL, IA (North/North Midland boundary); nine drew boundaries between Southern and Midwestern dialects above the Ohio River in OH, IN, IL, with eleven others placing the line through KY (North Midland/South Midland boundary); and six recognized that northern and southern sections of GA, AL, MS are in different dialect areas (Interior Southern/Coastal Southern boundary). Thirty-two participants drew a line across southern MO, twenty-one of them from that state. Seven of the thirty-six OH students and one in PA recognized that the dialect of southeastern OH is Appalachian. An interesting follow-up to the present study would be to ask participants to draw lines indicating the locations of dialect boundaries within their own states and in other states with which they are familiar. As we have seen in the preceding paragraphs, the core areas of the students’ maps are located in areas where dialectologists would expect to find them, except possibly the western portion of their Midwestern dialect. Figure 18.17 displays the locations of the core areas of all the areas discussed in this article. With the exception of parts of the West, the students’ composite map compares very favorably with Figure 18.16. Because core areas from disparate groups were used in setting it up, Figure 18.17 — like most generalizations — does not represent the actual views of any one individual or of the group as a whole. There was no way to know for sure whether the students could name all the states around which they drew lines and to which they applied labels, but the similarities between Figure 18.17 and the maps representing professional research suggest that their knowledge of geography was “close enough” for the subjective purposes of this study. If

SUBJECTIVE DIALECT DIVISIONS IN THE U.S.

313

Figure 18.17. Core areas of regional dialects (all students)

a professor or newsperson refers to Midwestern speech or to the Midwest as a region, we now know that some participants in this study will conjure up a map in which Indiana occupies a prominent position, whereas Missouri or Colorado will be prominent in others’ mental maps. Ohio may be in one person’s Northeastern map and in another’s Northern map. And Texas may or may not be in a participant’s map of the Southern dialect region. These mismatches may not cause problems in everyday conversations, but linguists, political scientists, and other academics would do well to become aware of how their own mental maps differ from those of their colleagues and their students.

Notes 1. This article is an expansion of a paper presented at the American Dialect Society in New York City in December, 1992. After analyzing data from a first-day exercise in my dialectology class at the University of Missouri in 1991, I decided to ask colleagues in a number of locations around the country to conduct similar exercises and send the maps to me for analysis. I received sets of maps from only four other states before writing the (1992) paper, but other maps arrived afterward, and a colleague and I collected more maps in our own classes. 2. I wish to express my appreciation to the following individuals: David R. Carlson, Springfield College (MA); Kelsie B. Harder, Potsdam College of the State University of New York; Riley B. Smith, Bloomsburg University (PA); David Bergdahl, Ohio University; William A. Kretzschmar, University of Georgia; Cynthia Bernstein, Auburn University (AL); Gilbert Youmans, University of Missouri; Thomas Gasque, University of South Dakota; Grant Smith, Eastern Washington University.

314

IMAGES, PERCEPTIONS, AND ATTITUDES

3. In the remainder of the chapter, the standard two-letter abbreviations will often be used to simplify references: Alabama (AL), Alaska (AK), Arizona (AZ), Arkansas (AR), California (CA), Colorado (CO), Connecticut (CT), Delaware (DE), Florida (FL), Georgia (GA), Hawaii (HI), Idaho (ID), Illinois (IL), Indiana (IN), Iowa (IA), Kansas (KS), Kentucky (KY), Louisiana (LA), Maine (ME), Maryland (MD), Massachusetts (MA), Michigan (MI), Minnesota (MN), Mississippi (MS), Missouri (MO), Montana (MT), Nebraska (NE), Nevada (NV), New Hampshire (NH), New Jersey (NJ), New Mexico (NM), New York (NY), North Carolina (NC), North Dakota (ND), Ohio (OH), Oklahoma (OK), Oregon (OR), Pennsylvania (PA), Rhode Island (RI), South Carolina (SC), South Dakota (SD), Tennessee (TN), Texas (TX), Utah (UT), Vermont (VT), Virginia (VA), Washington (WA), West Virginia (WV), Wisconsin (WI), Wyoming (WY). 4. The maps in this article were generated by Atlas Pro 1.5, c1991, 1993 (Strategic Mapping Inc., San Diego CA) on a Power Macintosh and exported in PICT format for placement in the published document. 5. Figure 18.15 is adapted from Lance (1994, Figure 32, p. 352). The names for the areas labeled New England, Northern, North Midland, South Midland, and South (in roman type) derive from Kurath (1949, Figure 3), one of the early publications based on data from the Linguistic Atlas of the United States. Later publications related to Kurath (1949) and used in constructing this map are Marckwardt (1957, maps 2 and 3); Allen (1964, map 6.1, p. 92); Dakin (1966, vol. 3, Figure 174, p. 104); Wood (1971, map 83, p. 358); and Faries and Lance (1993: 253). The names on the lower portion of the map (in italics) are from Pederson et al. (1992: ix-xv). 6. The map in Figure 18.16 is from Carver (1987, map 8.1, p. 248).

C 19 A View from the West Perceptions of U. S. Dialects by Oregon Residents Laura C. Hartley

Although much work has been done in this century to describe the actual linguistic situation within the United States, only within the past few decades has research focused on people’s attitudes and beliefs toward the varieties of speech that exist in the United States. In a series of studies in the early 1980s (summarized in the introduction to this volume), Dennis Preston examined the perceptual dialectology of residents in Hawaii, New York City, western New York, southern Indiana, and Michigan. This work has provided much helpful information in understanding American perceptions of U. S. regional speech; however, a comprehensive picture of the language attitude situation is not possible until data have been collected from each of the major geographical and dialectal areas of the United States. To date, no research has been done that examines the perceptions of West coast residents toward varieties of speech in the United States. This study begins to fill this gap in the literature by focusing on the perceptual dialectology of residents from one West coast state — namely, Oregon. The data in this study were collected primarily from residents living in the greater Eugene-Springfield area of Oregon, although a small number of residents of Portland also participated in the research. There were a total of sixty-six respondents, thirty-two males and thirty-four females, ranging in age from twenty to seventy-eight. All of the respondents were of European-American descent, with the exception of one respondent who identified himself as Native American. The respondents had all lived in Oregon for most of their lives, with twenty-seven of the sixty-six having lived only in Oregon.

1.

Analysis of Hand-Drawn Maps

Respondents were first given a map of the United States, which included state boundary lines but no state names. They were asked to draw circles around areas “where people talk the same” and to label them using whatever terms or phrases they felt best described the

316

IMAGES, PERCEPTIONS, AND ATTITUDES

speech in those areas. Table 19.1 shows the frequency with which regions were identified in this task as being distinct speech areas. Table 19.1. Frequency of identification of dialect regions Region Identified

# Respondents Identifying Region (n=65)

South Northeast Texas Midwest West New England California Pacific Northwest Hawaii Northern Southwest Alaska Plains & Mountains West Coast Louisiana / ‘Cajun’

60 49 36 31 29 24 17 16 15 14 12 09 09 08 08

% Respondents Identifying Region 92.3 75.4 55.4 47.7 44.6 36.9 26.2 24.6 23.1 21.5 18.5 13.8 13.8 12.3 12.3

The South was by far the most salient region, with 92.3% identifying some part of it as a distinct speech area. The Northeast followed, with 75.4% distinguishing it, and so on down the list. Although Table 19.1 provides an overall frequency with which different regions were identified in the hand-drawn maps, it does not indicate which states respondents believed composed each area. Since this varied from respondent to respondent, the best way to interpret overall perceptions of each region was to tally which states were included when that region was identified by a respondent as a distinct dialect area. By examining the overall tallies, it was then possible to differentiate “core” states that make up a perceptual region (i.e., those most commonly associated with that area) from “peripheral” states in a kind of layered map. Examples of such layered perceptual maps follow for several of the regions given in Table 19.1 (Figures 19.1, 19.2, 19.3, and 19.4).

2.

How Different Are They?

After completing the hand-drawn maps task, respondents were given an alphabetical listing of the fifty states, plus New York City and Washington, D. C., and asked to rate each area on a 4–point scale depending on how similar the speech in that area sounds to the respondents’ own speech. The scale was as follows:

A VIEW FROM THE WEST

Figure 19.1. Results of hand-drawn maps of a ‘South’

Figure 19.2. Results of hand-drawn maps of a ‘Northeast’

317

318

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 19.3. Results of hand-drawn maps of a ‘Midwest’

1. 2. 3. 4.

people people people people

there there there there

sound sound sound sound

like me a little different from me very different from me so different from me I can’t understand them

These “degree-of-difference” data were first subjected to chi-square tests of independence to determine whether gender, age group, or “visit status” (i.e., whether the respondent had ever visited the state) had any significant effect in the ratings. Since very few states (and none consistently) showed significant chi-squares for any of the independent variables listed above, subsequent statistical tests were run on the entire population as a whole. The next step in the analysis was to run chi-square tests that compared the ratings of each state to every other state. Results of these tests are not presented here, since the groupings of states produced by this procedure were too large and overlapping to give a meaningful picture of Oregonian perceptions of degrees of dialect difference. A clearer picture emerged once the data were subjected to multidimensional scaling (MDS) analysis, combined with K-means cluster analysis. Although these are two separate statistical operations, the results can be viewed together, as shown in Figure 19.5. In this figure, the plotted points are the output of the MDS analysis, and the circles represent the K-means cluster results.

A VIEW FROM THE WEST

319

Figure 19.4. Results of hand-drawn maps of a ‘West’

MDS analysis provides a graphic picture of how the states cluster together along two dimensions. Although the horizontal dimension is fairly easy to interpret, something like similarity to Oregon speech, the vertical dimension is less clear and apparently works primarily to differentiate Hawaii from all other states. It may be a “kind of accent” or even a “pleasantness of accent” dimension, since the southern states are clearly differentiated from the northeastern states, with western and midwestern states clustering around the zero point on this dimension. 3.

Whose Speech Is Most Correct and Most Pleasant?

Finally, respondents were given two more alphabetical lists (one at a time) and asked to rate each state or city on a 7–point scale according to how “correct” and how “pleasant” the speech in that area is (1 = least correct/pleasant, 7 = most correct/pleasant). The order in which these two tasks were presented to respondents was reversed for half of the respondents to eliminate any possible ordering effects for the two tasks. No definition or criteria for determining correctness or pleasantness were provided by the researcher. If the respondents asked questions about what was meant by these terms, they were told to use their own judgments.

320

IMAGES, PERCEPTIONS, AND ATTITUDES

HI

FL UT AK OR ID COAZ SD MI IA IL KS WA WY NM MT ND NV MN OH IN CA NE WI

OK

TX TN GA AL NC KY MS LA SC

AR

MO

WV VA MD NH RI CT MA VT ME WDC DE

PA

NY

NYC

NJ

Figure 19.5. Multidimensional scaling analysis of degree-of-difference ratings with K-means clusters

As with the degree-of-difference task, the correctness and pleasantness data were first subjected to chi-square tests to determine if gender, age group, or visit status influenced ratings. Again, no significant effects emerged overall for these independent variables in either task. Chi-square tests comparing each state with every other state were also run, followed by MDS and K-means cluster analysis.1 The results of these last two statistical tests are presented in Figures 19.6 and 19.7. By comparing the evaluations of states with one another for each of the three rating tasks (degree of difference, correctness, and pleasantness) in combination with the regions outlined and the labels given on the hand-drawn maps, it is possible to make some generalizations about this respondent group’s perceptions of U. S. regional dialects.

321

A VIEW FROM THE WEST

FL MO AR LA MS KY AL WV OK GA TN TX NC VA SC

NM

CA MT NE AK KS AZ SD IAND NV UT OH ID IN CO IL WI MN

NY

OR WA

MI PA DE MD CT VT RI NH

HI NYC

WY

NJ WDC MA

ME

Figure 19.6. Multidimensional scaling analysis of correctness ratings with K-means clusters

4.

Perceptions of Western States

To begin with, the hand-drawn maps and the degree-of-difference ratings show that Oregonians consider the states of Oregon, Washington, Idaho, Colorado, Utah, California, Nevada, Montana, Arizona, Alaska, and Wyoming to constitute a distinct dialect region. It is clear that New Mexico is a kind of border state between the perceptual regions of West and Midwest. The MDS analysis of the degree-of-difference data places it midway between Wyoming (a western state) and South Dakota (a midwestern state) (see Figure 19.5). Although thirty-two respondents rated it a “1” (i.e., people there sound like me) for degree of difference, the K-means cluster analysis for degree-of-difference groups it with the midwestern states. Finally, on the hand-drawn maps, the state is split, with the majority of respondents who drew a “West” including its northwest corner in the West (see Figure 19.4).

322

IMAGES, PERCEPTIONS, AND ATTITUDES

ME NH WDC VT RI PA CT OR DE MD NV MI CO UT ND OH IA WI VA WA ID NM IN MN IL MT AZ WY SDKS AK WV NE AR MO CA HI NC SC FL KY TN TX OK GA MS AL

MA

NY NJ

NYC

LA

Figure 19.7. Multidimensional scaling analysis of pleasantness ratings with K-means clusters

Although these western states are clearly distinct for Oregonians in terms of degree of difference, this distinction is less clear in the correctness and pleasantness ratings. The majority of western states are lumped together with the midwestern states in these categories according to a K-means cluster analysis (Figures 19.6 and 19.7). In terms of correctness, Oregon and Washington are distinguished from the rest of the western states as a separate cluster, and each makes up its own cluster in the pleasantness K-means cluster analysis. These results may reveal a desire on the part of the respondents to distinguish the local area (in this case, Oregon and Washington) from surrounding states (reflected in the degree-of-difference ratings), even when there is no noticeable difference in terms of accent. Such results conform with the general tendency for self-identification found in respondents from other areas (Preston 1986). This desire to identify Oregon and Washington as a distinct area is also illustrated by the sixteen respondents who drew a separate “Pacific Northwest” area on their hand-drawn maps (Table 19.1). In follow-up interviews conducted after the ratings tasks were

A VIEW FROM THE WEST

323

completed, several respondents commented that the speech in Oregon and Washington was very similar but could be distinguished from that of surrounding states. For example, one respondent said, “I can be in Washington and not feel like I’m not in Oregon, but I can be in Idaho and I can tell a difference.” Regarding the relationship between Oregon and Washington, in the MDS analysis for degree of difference, the two received identical scores (−1.46, +0.07). Although the MDS scores are not exactly the same for Oregon and Washington in terms of correctness, the K-means cluster analysis produced an Oregon/Washington cluster. When it comes to pleasantness, however, Oregon and Washington emerged as distinct clusters. This suggests that Oregonians want to maintain at least a small claim of uniqueness, but the interpretation of this result is not entirely straightforward. Preston (1993c) claims that “the preference for local norms along affective lines is stronger in areas where there is linguistic insecurity” (35). Since Oregonians view themselves as entirely unique only on the affective scale, this might indicate a slight degree of linguistic insecurity. This claim is somewhat tenuous, however, in that Oregonians also rate themselves highest in terms of correctness, which points to a great deal of linguistic security. It may be that the traditional categories of linguistic security/insecurity do not apply in the same way in western states, where a multiplicity and therefore awareness of distinctive dialects is not as prevalent as in eastern and southern states. A desire to distinguish California somewhat from other western states, and particularly from Oregon and Washington, is reflected most strongly in the pleasantness ratings. Although eighteen respondents rated California as a “7” in terms of pleasantness, almost as many (sixteen) rated it a “4.” It is unclear how to account for this discrepancy, since there were no significant effects on the ratings of California in terms of gender, age group, or visit status. What is clear is that many of the respondents wanted to differentiate California from other western states but did not choose to do so in terms of degree of difference or correctness. Instead, they opted to use the available affective dimension of pleasantness. Past research has revealed that status and in-group solidarity are the two primary evaluative dimensions along which language attitudes can be measured (Ryan, Giles, and Sebastian 1982). In this study, the category of correctness was used to measure status distinctions, and pleasantness was chosen to reflect degree of solidarity. In view of this, the ratings of California reveal that although these Oregonian respondents recognized the more or less equal social status/power of California residents, they feel a lack of solidarity with them. This lack of solidarity is underscored by California’s being singled out from other western and midwestern states in terms of pleasantness in the K-means cluster analysis even when Oregon and Washington are collapsed into this large cluster. Comments from the interviews following the quantitative data collection illustrate the somewhat negative view that some respondents have toward California (interviewer comments and questions appear in brackets): I lived in California for a short time when I was a teenager and wanted to be away from home, to start my own life, and I hated it … I was in L. A., well I lived in Whittier, which was outside L. A. … I only lived there eleven months and I was ready to come

324

IMAGES, PERCEPTIONS, AND ATTITUDES home. I remember the first time it rained, I looked up in the sky and I cried and said, “Thank you God.” People in California talk their own language which I don’t understand. [What do you mean by “talk their own language?”] Their whole terminology is a different task — it seems like their words are bigger and a lot of it seems to be, at least in the areas I’ve been in, technical terms and a lot it I guess I think of Silicon Valley in regards to computers and all the technical variety.

Although California is differentiated slightly from other western states in the respondents’ perceptions, Alaska is situated squarely within this cluster of states on all three scales. It is likely that this identification of Alaska with the West is a result of the historic connection between Alaska and the port cities of Seattle and San Francisco. The handdrawn maps support this conclusion as well. Although nine respondents (13.8%) indicated Alaska as a separate region, a greater number (twelve) included it in their “West” or “Pacific Northwest.” The remaining respondents simply did not include it in any specific region. Even when Alaska was circled as a separate speech area, several respondents labeled it as “frontier,” another indication that Alaska is perceived more in terms of the settlers who migrated there, embarking from the Northwest, rather than in terms of the native populations that inhabited the territory long before the arrival of the EuropeanAmericans. The inclusion of Alaska within the West is also interesting in that Hawaii was not considered part of this region on any scale. In fact, in the K-means cluster analysis for all three rating categories, Hawaii emerges as a distinct cluster. On the MDS analysis of the degree-of-difference data (Figure 19.5), it appears that the vertical dimension is used primarily to differentiate Hawaii (which received a value of +1.23) from all other states (which fall between −0.55 and +0.44 on the same dimension). Furthermore, fifteen respondents (23.1%) indicated on their hand-drawn maps that Hawaii composed a distinct dialect region, and only seven included it in another region (either West or California). Several respondents labeled the speech in Hawaii as “native-like,” which seems to indicate that the influence of the indigenous population of the islands on the speech in that state is significant in the respondents’ perceptions. Again, this makes Alaska’s ratings even more interesting because there does not seem to be an equally strong association of the effects of indigenous languages on the speech of Alaska. More substantial is the perception of Asian influence on the Hawaiian speech — one respondent labeled it “Japanese Golfland.”

5.

Perceptions of Midwestern States

According to the degree-of-difference MDS and K-means cluster analysis (Figure 19.5), the Oregonian conception of the Midwest consists of the states of New Mexico, North and South Dakota, Minnesota, Nebraska, Kansas, Wisconsin, Iowa, Illinois, Indiana, Michigan, and Ohio. The hand-drawn maps reveal that Iowa is the “heart” of this region (see

A VIEW FROM THE WEST

325

Figure 19.3). This cluster of midwestern states is the most closely related to the western states in several ways. First, in the K-means cluster analysis of the degree-of-difference data, the midwestern and the western clusters of states collapse together if only seven (rather than eight) groups are chosen. The midwestern and western states are also lumped together on the correctness (Figure 19.6) and pleasantness (Figure 19.7) tasks (with the exception of Oregon, Washington and, for pleasantness, California) in the K-means cluster analysis, although they are clearly two subgroups in terms of their MDS values. Michigan appears to be a kind of border state between the Midwest and the Northeast for these respondents. Although it falls clearly within the center of the Midwest cluster on the degree-of-difference ratings, it is part of the K-means cluster of northeastern states in terms of both correctness and pleasantness. On the hand-drawn maps, twenty respondents included it within their boundaries of a “Midwest,” whereas only six incorporated it into a “Northeast.” Ohio appears also to be a kind of border state between the Midwest and the Northeast but in a different way from Michigan. Ohio clusters with the midwestern states on all three scales, but the hand-drawn maps reveal some disagreement among respondents about which region it belongs to. Twenty-two respondents include the western half of the state in the Midwest, but only eleven incorporate the entire state in this region. On the “Northeast” map, nineteen respondents include only the eastern half of the state, whereas eleven respondents incorporate the whole state. Ohio thus appears to be a border between the Midwest and the Northeast in the same way that New Mexico is split between the West and the Midwest. It is important to note here that the first large group of settlers in the Willamette Valley in Oregon came from the Ohio Valley states and Tennessee. There were also large numbers of settlers from Missouri, Illinois, and Iowa (Carver 1987). Thus, although the geographical location of Ohio may have caused some respondents to incorporate it into the northeast, particularly on the hand-drawn maps, the historical connection between Ohio and other midwestern states as the major migration origination points for Oregon settlement was more influential in the rating tasks.

6.

Perceptions of Northeastern States

The next group of states that emerges from the MDS and K-means cluster analyses of the three tasks is a northeastern cluster. The primary states in this cluster are Maine, Vermont, New Hampshire, Rhode Island, Connecticut, Pennsylvania, Maryland, Delaware, with some variability among the three tasks for Washington, D. C., Massachusetts, Virginia, West Virginia, Michigan, and Wisconsin (as discussed previously). For the degree-of-difference task, Washington, D. C., is included in the northeastern cluster. It is isolated as its own group in terms of correctness, falling somewhere in the middle of the ratings for midwestern and northeastern states. For the pleasantness task, however, it is grouped with Massachusetts, New Jersey, New York, and New York City,

326

IMAGES, PERCEPTIONS, AND ATTITUDES

having received the third highest number of “1” (i.e., least pleasant) ratings (after New York City and New York). In terms of correctness, that Washington, D. C., emerges as a distinct cluster is likely a result of the specialized styles and jargon associated with the discourse of government functions. That it is grouped in terms of pleasantness with the cluster that includes New York City, clearly the cluster with the least favorable ratings, is perhaps the most interesting result. This may well be an indication of growing dissatisfaction with the current political scene, particularly in light of controversy at the time the data were collected surrounding an Oregon senator, which eventually led to his resignation from Congress. Massachusetts follows the same pattern as Washington, D. C., being clustered with the northeastern states in terms of degree of difference, emerging as a distinct cluster for correctness (also with ratings somewhere in the middle of the northeastern and midwestern states), and grouping with New York City, New York, New Jersey, and Washington D. C., on the pleasantness task (receiving the fourth highest number of “1” ratings). Since this grouping cannot be a result of the perception of government, however, an alternative explanation must be sought. The most likely reason for the differentiation of Massachusetts from other northeastern states is the stereotypical (and actual) Boston accent. On the hand-drawn maps, in fact, several respondents used the label “Bostonian” to refer to a general New England area. The historic connection between Boston and the Pacific Northwest, particularly in terms of the fur trade in the first half of the 19th century, may also be related to area recognition in this case. Virginia and West Virginia appear to be transitional states between the Northeast and the South for the respondents in this study. Although grouped with the northeastern states in the K-means cluster for degree of difference, they cluster with the southern states in terms of correctness. They are split in the pleasantness ratings, Virginia grouping with the Northeast and West Virginia clustering with the South. The hand-drawn maps also confirm these states as border states, since an almost equal number of respondents included them in the Northeast region as they did in the South (see Figures 19.1 and 19.2).

7.

Perceptions of New York City, New York State, and New Jersey

Figure 19.2 shows that New York is the center of the hand-drawn maps of a Northeast and that, in general, New York City and New Jersey are incorporated into this perceptual region. It is interesting then that New York City, New York State, and New Jersey are never included in the K-means clusters of the northeastern states. In fact, New York City and New York State are both included in the cluster of southern states on the degree-ofdifference scale. They form their own cluster in terms of correctness and, as stated already, combine with New Jersey, Massachusetts, and Washington, D. C., to form a cluster in terms of pleasantness. New Jersey emerges as a distinct cluster for degree of difference and correctness.

A VIEW FROM THE WEST

327

The results for New York City are not surprising given the stereotype of New York City inhabitants as fast-talking, cold, and rude. Consider these comments made in the follow-up interviews: When we used to go to the flea markets in, well, Greenfield and those we used to think New Yorkers were mad at each other — that’s just how they talk, they always sound like they’re arguing and that … yeah, and they’re just, that’s the way they speak, and we used to kind of, they’d yell and scream and you’d look back thinking they were fighting and they were just visiting. People seem to talk very fast up there from what I’ve experienced.

Although the results of the New York City ratings were expected, it is particularly interesting that New York State gets “dragged down” with New York City. Chi-square tests reveal no significant difference between the ratings given to New York State and New York City on any of the tasks. Furthermore, these two regions are always in the same K-means cluster. Thus, there appears to be little difference in Oregonian perceptions between New York City and the rest of the state. This result is strikingly different from the results of maps drawn by residents of both New York City and western New York State reported in Preston (1986). Both sets of respondents indicate that New York City and New York State are distinct dialect regions. The respondents in western New York even further differentiated western New York as a separate region. The ratings of New Jersey are also clearly influenced by the New York City “stigma,” although not in precisely the same way as New York state. For both degree of difference and correctness, New Jersey forms its own cluster, as mentioned previously. This may be due in part to New Jersey having its own stereotypical accent (i.e., “New Joisey”). It may also be the case that New Jersey is seen as a kind of “transitional” or “buffer” state between New York City and other northeastern states, such as Pennsylvania, Delaware, and Maryland. In general, New Jersey seems to occupy a location between New York City and the northeastern states on the MDS plots.

8.

Perceptions of Southern States

On the hand-drawn maps, the South is clearly the most salient dialect region, with 92.3% of respondents indicating at least some portion of it as a distinct region. The states unequivocally part of this region are Alabama, Louisiana, Mississippi, Georgia, Arkansas, North and South Carolina, Kentucky, Tennessee, Texas, and Oklahoma. Missouri is included in K-means clusters for correctness and pleasantness but forms its own cluster in terms of degree of difference. West Virginia joins the group in correctness and pleasantness, and Virginia is part of the cluster in terms of correctness, as discussed previously. Florida forms a distinct cluster on all three scales, although it joins the southern cluster on the correctness and pleasantness scales if fewer clusters are selected. In the degree-of-difference task, the southern states are rated the most dissimilar to Oregon. Between forty-three and fifty-eight respondents rated all the core southern states

328

IMAGES, PERCEPTIONS, AND ATTITUDES

except Oklahoma either a “3” or a “4” on this scale. Of nonsouthern states, only New York City received such a large number of high difference ratings. For correctness, the southern states fall at the bottom of the list, but the ratings for pleasantness rise dramatically for all of the southern states, with Georgia and Louisiana actually receiving the most “7” ratings after the western states. Missouri’s marginal status as a southern state (reflected in the fact that it forms its own K-means cluster) is interesting for several reasons. First, the largest number of settlers in Oregon in the latter half of the 19th century came from Missouri (Carver 1987). Thus, the recognition of Missouri as not as different from Oregon as the other southern states (in terms of degree-of-difference ratings) may be a result of this historic connection. That these Oregonian respondents group Missouri with southern states at all, however, is intriguing, since the respondents from the five areas that Preston (1986) examined tended to place Missouri in the Midwest rather than in the South. It seems that Missouri is another transitional state — in this case, between the South and the Midwest. This is supported by the hand-drawn maps in which more respondents include Missouri in the Midwest (Figure 19.3) than in the South (Figure 19.1), even though the respondents clearly associate it with the South in the ratings tasks. Texas is also an interesting state to examine because of the discrepancy between hand-drawn maps and the ratings tasks. Although 55.4% of respondents singled out Texas (sometimes including Oklahoma) as a distinct dialect area on their hand-drawn maps (Table 19.1), it did not appear as a distinct area in the K-means cluster analysis of any of the other three ratings tasks. It seems, then, that although there is some recognition of a distinct dialect in Texas, this distinctness is not great enough to overcome the general category “southern” in the minds of the Oregonian respondents. With Florida, there appears to be precisely the opposite discrepancy between the hand-drawn maps and the ratings tasks from that which occurred with Texas. In this case, Florida was generally incorporated into the maps of the South but emerged, at least potentially, depending on the number of cluster groups used, as a distinct K-means cluster on all three of the ratings tasks. There are two possible explanations for Florida’s being singled out. First, there may be the recognition of influences from northern dialects due to retiree migration from the North to the milder climate of Florida. This appears to be the explanation, for example, of Michigan raters’ perceptions of Florida, since Florida falls together with Michigan on factor analyses of both correctness and pleasantness data (Preston 1993c). The second explanation has to do with the influence of Spanish (primarily Cuban) on the language in Florida. Given that Florida would not be a likely retirement spot for Oregonians, this second explanation is more plausible; moreover, several respondents used labels such as “Cuban” on their hand-drawn maps. The southern states received the lowest ratings for correctness, especially the four states of the “deep south” — Georgia, Mississippi, Louisiana, and Alabama. The pleasantness ratings, however, are substantially higher for this area, with many southern states receiving ratings higher than many midwestern and northeastern states. This reflects

A VIEW FROM THE WEST

329

the stereotypes of “Southern hospitality” and a slower pace of life. Speaking about people from Kentucky, for example, one respondent says, They act like they’ve got all the time in the world, you know — “nothing wrong with me, I’m just taking my time,” I can’t do it, but it sounds great.

In general, then, Oregonians view the South as one large dialect region. Although they sometimes separate an “outer South” and Texas from the “deep South” on their handdrawn maps, this distinction is not great enough to create separate K-means clusters in any of the ratings categories. An examination of the MDS plots, however, does reveal that the Southern cluster is the “loosest” of all the clusters; that is, its points are the most spread out.

9.

Comments on Research Methodology

The most interesting difficulty that emerged in the course of this research was the reluctance on the parts of many respondents to provide correctness ratings. Fourteen respondents gave all states the same rating (anywhere from a “4” to a “7”), whereas an additional four respondents simply left this task completely blank. As they were filling out the correctness section of the questionnaire, many respondents made comments such as the following: People are correct for wherever they are from. I don’t consider speech in terms of correctness or incorrectness, but in terms of difference. Grammatically — as seen by a prescriptive linguist! It’s all so subjective!

In the follow-up interviews, people also offered explanations of their objections to this task. Consider the following comments: Well, of course, I think each individual thinks that they speak the proper way, and so somebody that differs real drastically, you’re thinking they really don’t know how to speak, you know, or they don’t know how to express themselves, but I’m sure that somebody who speaks drastically different than I do thinks their speech is perfect too. I honestly don’t believe that I have ever considered — I have never rated things when I was listening to them as being correct or incorrect, it was just that person’s way of doing it — living in the neighborhood where I did, why, we accepted everybody or we didn’t get along and I — I was aware that, well you can’t say that what was spoken in English was correct and what we spoke was not and it’s the same thing with what was spoken in the east coast and I never thought that what was spoken in Boston was any better than what I spoke out here. the correctness issue, I mean, well what is correct language anyways? You know, it’s all relative to who’s looking at it and the person you’re coming from.

330

IMAGES, PERCEPTIONS, AND ATTITUDES

That so many respondents objected to the correctness task is in sharp contrast with Preston’s (1993c) experience in giving this same task to respondents in Michigan and Indiana. He says of his raters: It should be noted that very few respondents complained about this task. … Although they complained that they did not have information about this or that state, the ranking of most areas for correctness was for them a reasonable task and represented opinions overtly held about the sites where better and worse English was spoken. (31)

I believe this difference between raters in Oregon and those in Michigan and Indiana is largely a result of the cultural and linguistic heterogeneity that has been a part of the West coast experience since early settlement days and continues to be a dominant force in the experience of westerners. Another possible explanation for the difference between Preston’s raters and the respondents in this study may also be the factor of time. Since Preston’s data were collected roughly ten years ago, the reluctance of the respondents in this study may be an indication that Americans are becoming more sensitive to the issue of what constitutes correct language. Interestingly enough, although there was so much objection to the correctness task, there was little complaint about any of the other tasks. Thus, to rate someone’s speech as less pleasant than one’s own didn’t appear to be as big an offense as labeling his or her speech incorrect. One respondent describes the difference between the two tasks in this way: [Why do you rate everyone the same on correctness but differentiate in terms of pleasantness?] Because it seems like it’s an aesthetic quality and something that brings you know, it’s kind of an artistic thing; it seems like it’s something you can appreciate, where correctness to me seems to me like there’s some scale and there’s some right and wrong to it, and I don’t see that as relevant in terms of speech.

Although Oregonians may be less dogmatic about what constitutes correct speech, it would not be fair to say that they are true linguistic relativists. Clearly, the results of the pleasantness ratings show that they have some definite ideas that not all regional speech is equally agreeable.

10. Conclusion The results of this study show that residents of Oregon tend to view the United States as divided into five major dialect regions: West, Midwest, Northeast, South, and Hawaii. Within each of these major regions, there are a few smaller subregions, such as the Pacific Northwest, (southern) California, metropolitan New York City, and Florida. Between the regions, there seemed to be one or two “transitional” or “border” states. These included New Mexico (between West and Midwest), Michigan (between Midwest and Northeast), Virginia (between Northeast and South), and Missouri (between South and Midwest). The overall findings of this study are summarized in Figure 19.8. The states of Oregon and Washington were rated the highest in terms of correctness and pleasantness, whereas New York City was rated the lowest in both categories. In

331

A VIEW FROM THE WEST

Northeast

Pacific Northwest

New England

New York City

Midwest West California

South Texas Florida

Major Dialect Region Boundary Dialect Subregion Boundary

Hawaii

Figure 19.8. Oregonian perceptual speech regions summarized

general, the other western states as well as the midwestern states were rated similarly, fairly high on both the correctness and pleasantness scales. The South, which was the most salient dialect region for the respondents in this study, was rated poorly for correctness but fairly well for pleasantness. Both Hawaii and Florida tended to fall into their own groups in terms of the ratings, primarily because of the influence of nonnative English speakers in those states. This study is the first to examine the perceptual dialectology of any West coast residents. It should be noted that the respondent group in this study represents only a small portion of Oregonians. They were almost all European-Americans and residents of two urban areas on the western side of the state. To make more comprehensive generalizations about Oregonian language attitudes, it would be necessary to survey a wider range of both geographical and ethnic groups. To gain a broader understanding of the language attitudes of western U. S. residents, it would be necessary to replicate this research throughout the region. At a minimum, it would be good to survey residents of Washington, northern California, and southern

332

IMAGES, PERCEPTIONS, AND ATTITUDES

California. Of course, each state has a unique settlement history, and the most comprehensive picture of western perceptual dialectology would require, therefore, research in every state. This study is thus merely a beginning to what could become a much more extensive research program.

Notes 1. It should be noted that the K-means cluster analysis makes no assumptions about how many groups there are in the data set. The researcher must specify the number of groups that should be calculated and will likely need to try different numbers of clusters before the best analysis is achieved. In Figures 19.6 and 19.7, the lines connecting two or more clusters represent the results of the K-means cluster analysis when fewer clusters were selected.

C 20 “Welshness” and “Englishness” as Attitudinal Dimensions of English Language Varieties in Wales Nikolas Coupland

Angie Williams

Peter Garrett

Research into language attitudes and perceptions is important to a whole range of questions.1 We can gain insights into stereotyping in terms of personal and social attributes, effects on communication outcomes, and language change. In territories where there is language contact, the study of attitudes and perceptions can help us achieve a better understanding of language maintenance and change, language decay or revival, cultural continuity, and issues of identity. “People react to a perceived environment” (Gould 1977: 111). Groups may become demarcated, and territories can take on a number of possible functions: – – – –

A defensive function, establishing “fortresses,” “no-go areas,” and so on An avoidance function, providing havens or sanctuaries for group members A preservation function, to defend distinctiveness in an assimilatory environment A resource base function, providing a means for attack, should it prove possible to alter the existing relationship with the dominant group (Boal 1976)

“Regional distinctiveness and territorial identification can become salient instruments in the defense of minority group cultures” (Williams 1985: 276). This question of regional distinctiveness has been largely neglected in the language attitudes investigations in Wales, and this partly accounts for the unclear picture that has so far emerged. For example, Bourhis, Giles, and Tajfel (1973) found Welsh English more favorably evaluated than received pronunciation (RP) and on a par with the Welsh language. Elsewhere, however, it has been found to be downgraded, along with RP, in relation to Welsh (Bourhis and Giles 1976) or downgraded compared with both Welsh and RP (Price, Fluck, and Giles 1983). The study we report here seeks to clarify this picture and also responds to Price et al.’s call for attitude studies of whole regions to build geolinguistic atlases of attitudinal variation.

334

IMAGES, PERCEPTIONS, AND ATTITUDES

We have therefore drawn our data from locations spread over the whole of Wales. Specifically, we have elicited the attitudes and perceptions of secondary schoolteachers in relation to a wide range of geographical varieties of Welsh English. This teacher focus was to allow us the advantage of comparison across regions, and also because, as gatekeepers, teachers are likely to feature prominently in discussions of standardness and nonstandardness in Welsh English. Space precludes a very detailed account of our research procedure, the coding and analysis of our data, and other aspects of our findings. Such details may be found in Coupland, Williams, and Garrett (1994); Garrett, Coupland, and Williams (1995); and Williams, Garrett, and Coupland (1996). We examine evaluations of these varieties not only along the dimensions typical of most such language attitudes work (e.g., the prestige, dynamism, and pleasantness that a variety is perceived to project) but also in terms of how Welsh or English these varieties are judged, to discover from this if some varieties of English in Wales are associated with a stronger Welsh identity than others. This dimension of authenticity is our focus in this chapter. It is generally neglected in language attitudes research, yet we feel it is of importance not just in Wales but in postcolonial contexts worldwide where colonial languages have taken hold or, indeed, anywhere where the survival of indigenous languages is in question. To some extent, our study challenges methodological convention. In recent years, language attitudes research has been dominated by the matched-guise technique, in which respondents judge audio-recorded speech samples that they believe are spoken by different people, when, in fact, they are hearing a single person producing a range of varieties. The idea is to minimize variation in features other than the one to be researched (e.g., accent). But there are well-documented drawbacks with this approach (e.g., see Giles and Coupland 1991a, 1991b). In addition, we were inspired by recent work in folklinguistics and perceptual dialectology (e.g., Preston 1989, 1993d), with whose approach we could better address the distributional questions of identity and authenticity that are of prime interest to us. Before outlining our methods and findings, it is useful to consider preexisting accounts of linguistic and cultural divisions in Wales as well as Welsh and English identity. As yet, there is no fully authoritative descriptive account of geolinguistic variation in Welsh English accents and dialects. However, Coupland and Thomas (1990), although emphasizing “many significant exceptions” (8), propose three broad categories. One is southwest Welsh English, with contemporary and substratal influence from the Welsh language in phonology, lexis, and grammar. Second is the urban Southeast, with far less substratal influence and almost free from contemporary influence from Welsh. Third is a northern type, comparatively dependent for its distinctive features on the Welsh language but modeled specifically on the Welsh language dialects of North Wales. See Coupland (1990) for descriptive summaries of some of the relevant dialectal norms. Edwards (1991) puts forward a two-dialects model of English in Wales: (1) a southern variety, related to the speech of the English West Midlands and the southwest of England and on which the influence of Welsh is mainly substratal, and (2) a northern variety, to some extent influenced by the speech of northwest England and on which the

“WELSHNESS” AND “ENGLISHNESS” AS ATTITUDINAL DIMENSIONS

335

Welsh language is very much a living influence because of the comparatively high proportion of Welsh/English bilinguals. Balsom (1985: 1) proposes a “three Wales” model to interpret Welsh cultural geography (see Figure 20.1), arguing that Wales has three distinct and identifiable sociolinguistic groups. The model is based on a combination of data from the 1981 Census and the Welsh Election Survey of 1979 (in Balsom’s view, a low point in Welsh selfassertion). Balsom’s (1985: 6ff) three main regions are (1) Y Fro Gymraeg: the Welsh heartland, Welsh speaking and Welsh identifying, almost totally Welsh born, mainly working class, and largely occupying the north and west of Wales; (2) Welsh Wales: nonWelsh speaking but Welsh identifying, almost totally Welsh born, predominantly working class, and largely occupying the traditional South Wales area; and (3) British Wales: nonWelsh speaking, relatively more British identifying, more middle class than working class, dominating the remainder of Wales. Williams’s (1985) mapping of culture regions was carried out by asking schoolchildren all over Wales to identify on a map which towns they felt were in predominantly Welsh-speaking areas. He is not concerned with showing firm borders between zones but instead produces sets of isolines on his maps, to reflect percentile rates of inclusion as Welsh-speaking. The map based on his whole sample (Figure 20.2) broadly reflects Balsom’s distinctions, but by reason of its isolines, shows finer gradations. Williams’s 50% threshold isoline threads a route across the map of Wales that largely coincides with the boundary of Balsom’s Y Fro Gymraeg. But there are also peaks of 70% and 80% isolines, which are only to be found in the Northwest. The area around Carmarthen (see Figure 20.3) forms the highest peak in the South, at 60%, with the area matching Balsom’s British Wales showing a decline to 20%. Because of the Welsh-speaking nature of Williams’s data, the isolines peter out from 40% to 20% across the Valleys, and this tends to lead to the Valleys’ simply being absorbed into Balsom’s British Wales. However, the paths of the contours do seem to coincide to some extent with the boundaries of Welsh Wales and suggest some similarity.

1.

Method

A questionnaire was prepared in which, following Preston (1989), we provided a map of Wales, blank, apart from some of the main towns. The questionnaire was sent out to teachers in secondary schools all over Wales. They were instructed to draw in what they felt to be the main English dialect regions of Wales (up to a maximum of eight), to supply labels they might typically use for these, and to supply short characterizations of each of the dialects in terms of what they find most striking about them. On the questionnaire, we referred to “accent/dialect” because we felt that it would be unhelpful to become involved with definitions of those terms in this study. Although important distinctions are drawn by linguists (with dialect subsuming accent), our subjects could be reasonably regarded as nonlinguists. Nonlinguists do not generally differentiate the two terms, and so we use these as lay terms in our report below.

336

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 20.1. Language and identity in Wales, based on ‘The Three Wales Model’ (Balsom 1985)

In the same questionnaire, after the drawing and labeling task, teachers were presented with six descriptively labeled geographical accents/dialects of Welsh English, along with Standard British English for comparison. The Welsh English varieties were selected so as to ensure a good geographical spread of communities and also to approximate the broad pattern of differentiation revealed by descriptive dialectological research (see Coupland 1990). The varieties were presented conceptually, a technique that has produced findings previously that have correlated very highly with findings from matched-

“WELSHNESS” AND “ENGLISHNESS” AS ATTITUDINAL DIMENSIONS

337

Figure 20.2. Williams’s (1985) perceptual culture regions from a sample of schools all over Wales, with isolines showing percentile levels of agreement about whether towns were Welsh-speaking

guise presentation (e.g., Giles 1970; Ball 1983). Thus, one of the Welsh varieties was written down as “the English associated with northwest Wales (e.g., Bangor).” The others, following the same rubric, were southeast Wales (e.g., Cardiff); southwest Wales (e.g.,

338

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 20.3. Wales with the six focal towns/cities and neighboring English urban areas

Carmarthen); South Wales valleys (e.g., Merthyr Tydfil); Mid-Wales (e.g., Newtown); northeast Wales (e.g., Wrexham). Standard British English was presented as southeast England (e.g., Cambridge). Below, we generally refer to these varieties by the town and city names. Figure 20.3 is a map of Wales showing these towns/cities and the main nearby English urban centers. (The map used with respondents included more towns and did not include any English urban centers.) Teachers were asked to fill in 7–point semantic differential scales (bounded by the poles definitely at one end, and not at all at the other) for each of the conceptually presented varieties. The scales related to the following judgments: pleasant sounding, dynamic sounding, prestigious sounding, truly Welsh sounding, lively sounding, likable sounding, well-spoken, and how well will this accent/dialect serve the overall interests of school leavers? At the end of the questionnaire, teachers were asked if they could identify

“WELSHNESS” AND “ENGLISHNESS” AS ATTITUDINAL DIMENSIONS

339

one Welsh accent/dialect of English that they thought carried more social acceptability than others and what its social advantages were. One hundred twenty-nine teachers all over Wales completed questionnaires, which were offered in Welsh or English. Among these teachers, there was a reasonable sex balance (51.9% male, 46.5% female, and 1.6% no answer), a good range of Welsh regional locations (e.g., 65.5% in the southern counties, 34.5% in the northern ones), a quite even split between Welsh and non-Welsh speakers (46.9% and 53.1%, respectively), and a fairly broad range of ages (mean of 40.8 years), lengths of residence in Wales (mean of 32.1 years), and teaching subject areas.

2.

Results

We will refer to the first part of the questionnaire (i.e., map, labeling, and characterization) as the “labels task,” and the second section as the “scales task.” A brief summary of the main results is given here, alongside our highlighting of judgments of Welshness and Englishness. We begin with the scales task. Reliability analysis employing Cronbach’s alpha showed that we could regard the scales as falling into three broad evaluative dimensions relating to those generally found in language attitudes research (e.g., Zahn and Hopper 1985): – – –

Pleasantness — likable, pleasant: alpha reliability of 0.84 Dynamism — lively, dynamic: alpha reliability of 0.82 Prestige — well-spoken, prestigious, “overall interests”: alpha reliability of 0.87

Most relevant for the present chapter, however, is that the item “truly Welsh sounding” did not correlate highly with the other scales, and we consequently singled it out as a distinct fourth dimension that we labeled “Welshness.” Subsequent multivariate analysis of variance showed a significant main effect only for accent/dialect community (Wilks’s lambda = 0.212; df = 24; p < 0.01). This means that the teachers as a whole were making evaluative distinctions among the accent/dialects. Univariate analysis of variance tests showed a significant effect for all four dependent measures (p < 0.01), meaning that the teachers were differentiating the varieties on all of the attitude dimensions. Ignoring precise mean scores here, Table 20.1 shows the evaluative profile of each variety. To summarize from Table 20.1, although teachers see Wrexham and Cardiff (both urban communities in eastern Wales) as carrying little prestige, pleasantness, dynamism, and Welshness, they give Carmarthen the “fullest” profile of all the varieties. They may demonstratively acknowledge Cambridge as carrying the most prestige, but they award it little else on the scales used in our study. Insofar as a standard variety requires subjective definition as such (Haugen 1966), Carmarthen emerges, on our scales, as the main contender for that title within Wales. Responses to the open-ended question on social advantages also reflected this. One respondent even called this variety “the Welsh version of RP.”

340

IMAGES, PERCEPTIONS, AND ATTITUDES

Table 20.1. Generalized summary of judgments of the accent/dialect communities (scales data)

Carmarthen Bangor Merthyr Tydfill Newtown Wrexham Cardiff Cambridge

Welshness

Prestige

Dynamism

Pleasantness

Very high High High Low Low Very low Very very low

Mid Low Very low Mid Low Low High

Mid Low Mid Low Low Low Low

High Low Mid Mid Low Low Low

Of more interest to us here, however, is that, apart from distinguishing the varieties in terms of prestige, teachers also clearly differentiate them in terms of their Welshness, and not necessarily in ways related to prestige. In the case of Merthyr Tydfil, the Valleys region traditionally associated with coal-mining communities, Welshness is high, but prestige very low. This is also the case with Bangor, in the northwestern area, traditionally associated with slate quarries, sheep farming, the mountains of Snowdonia, and the Welsh language. The scales data have a significant limitation (one largely overlooked in language attitudes research). We can see that, when we ask the teachers to use scales of prestige, Welshness, and so on, they do so. And they use the prestige and Welshness scales in particular to discriminate among the varieties. But although they clearly see these dialects as varying considerably in degrees of Welshness and prestige, we cannot be absolutely certain from these data that they actually see these two dimensions as being more important to them than other, less discriminating, dimensions. In this respect, the labels data are a useful complement, since they not only give us the possibility of confirming or refuting the patterns of discrimination that we see in the scales data but also, by revealing the labels that the teachers themselves have chosen to use, allow us to see if there are certain labels used less, which could indicate less important dimensions, and other labels used more, which could in turn indicate dimensions they regard as more salient. When analyzing the data from the labels task, our first job was to examine the labels to determine what dialect areas the teachers had identified. We obviously had to consider the labels in relation to the corresponding boundaries that the teachers had drawn on the maps. Hence, we were able to associate the label “towny” with Cardiff by referring to where the teacher had drawn the lines on the map. There was inevitable variation in the precise lines the individual teachers drew on their blank maps. When we grouped together labels relating to “Cardiff,” we found a broad consensus as to what this dialect region was, whereas when we grouped labels relating to “English” (as in “Little England,” or “Anglicized Welsh”), these connected with more diffuse regions. We arrived at a total of nine geographical groupings of labels: Valleys, Cardiff, North Wales, Mid-Wales and Borders, Southwest Wales, Liverpool, Pembrokeshire, along with the two relatively diffuse sets of labels referring to Englishness and Welshness. These last two excepted, these are the column headings in Table 20.2.

“WELSHNESS” AND “ENGLISHNESS” AS ATTITUDINAL DIMENSIONS

341

Table 20.2. Frequencies of categorized characterizations, by labeled dialect areas Valleys

Cardiff

North Wales

Mid-Wales/ Southwest Borders Wales

Liverpool

Pembrokeshire

Welshness

(24) 31

(1)

1

(15) 23

(7)

(14) 29

(1)

(1)

2

Non-Welshness

(3)

(5)

7

(8)

(14) 25

(4)

8

(35) 76

(31)

69

Linguistic form

(14) 18

(26) 34

(22) 34

(4)

(3)

6

(6)

13

(2)

4

Affective positive

(22) 29

(4)

(3)

(12) 21

(15) 32

(3)

7

(4)

9

Affective negative

(9)

(33) 43

(19) 30

(7)

13

(3)

6

(5)

11

(5)

11

Prestige

0

(2)

3

0

(3)

5

(1)

2

0

(2)

4

(2)

3

0

0

Negative prestige (4)

4

12

5

Rural

0

0

Urban

0

(4)

5

(1) 5

0

13

5

2

(3) 0

13

7

0 5

(2)

(1) 4

0

2

2

0

0

(5)

0

0

11

Note: Number of instances are mentioned in brackets, with rounded percentages following in italics.

The teachers’ own characterizations of these labeled areas fell into a number of categories: prestige, linguistic characteristics, affect, and the Welsh and English qualities of the variety. These are the row headings in Table 20.2. Many of these carried not only descriptive but also highly evaluative messages and constituted valuable attitudinal data. Table 20.2 again reveals quite separate profiles (this time expressed in the teachers’ own words rather than in words supplied by researchers on 7–point scales) for each of the (this time subjectively identified) Welsh English dialect regions. For example, the labels given to Cardiff most frequently (43%) concerned negative affective qualities (e.g., harsh, ugly, annoying) or (34%) linguistic qualities (e.g., flattened, stretched) and, occasionally, non-Welshness (7%) and lack of prestige (3%). The labels for North Wales were also frequently (34%) linguistic (e.g., nasal) or negative affective (30%) (e.g., painful, annoying, sounds like they have a bad cold), more frequently (23%) Welsh (e.g., Welsh 1st language) than those given to Cardiff (1%) but with less prestige (0%). North Wales labels varied, however, according to whether teachers had included northeast as well as northwest Wales in the accent/dialect area they marked on the map. Hence, although, for example, “strong/broad Welsh” was mentioned in relation to the Northwest, “Merseyside” and “Liverpool” are examples of labels that focused on the Northeast. Southwest Wales again had a strong profile: 32% of comments were affectively positive (e.g., melodious, soft), 29% Welsh (e.g., heartland accent, the true accent of Wales), and interestingly, only 6% based on linguistic forms.

342 3.

IMAGES, PERCEPTIONS, AND ATTITUDES

Discussion

The evaluative and perceptual importance of the Welsh/un-Welsh dimension is highlighted by its being used more than any other dimension in these characterizations (and far more, in these data, than the prestige dimension). It is highly salient in the un-Welsh characterizations of urban northeast Wales and Pembrokeshire in the extreme Southwest and also in the positively Welsh identification of the Valleys, southwest Wales (Dyfed), and northwest Wales. It is also striking that there is no inevitable association between Welshness and positive affect in our data. North Wales is viewed as relatively Welsh but relatively negative in affect. The dialect area labeled “Liverpool” is seen as very unWelsh and only mildly negative in affect. Cardiff is negatively evaluated but seen in the labels data, surprisingly, as only mildly un-Welsh. We now turn to closer consideration of the work by Balsom (1985) and Williams (1985) summarized earlier. First, the concept of the “Welsh heartland” (Y Fro Gymraeg) is supported by the teachers’ labeling of southwest Wales, North Wales, and perceptually Welsh zones as separate from others. Of additional interest here is that, conventionally, Y Fro Gymraeg is seen in terms of Welsh language use, whereas our teachers’ observations are based on English language use in Wales. Moreover, our data echo Williams’s indication of a subdivision of Y Fro Gymraeg into northern and southern communities, but our division is an evaluative one, showing more negative and more positive profiles respectively. Second, Balsom’s concept of a Welsh Wales is also endorsed in that the Valleys are seen as a distinctly labeled domain. But the Valleys and Southwest Wales are very similar in our data regarding Welshness yet dissimilar from the North Wales zone (which includes Williams’s Welsh-speaking peaks). This latter zone, although also viewed as very Welsh in both the scales and the labels tasks, is more negatively stereotyped. Third, Balsom’s British Wales is perceptually confirmed to the extent that Pembrokeshire in the far Southwest, Cardiff, Mid-Wales and Borders, and Liverpool are given boundaries similar to those in Figure 20.1, but again, there is striking differentiation within this area in our data. The Britishness of Pembrokeshire and the Liverpool-influenced area is marked by a strong perception of non-Welshness, but the Mid-Wales Border region is perceived as relatively mildly non-Welsh. Also, although teachers judged Cardiff as very low in Welshness on our prelabeled scales, its non-Welshness does not appear to be a particularly salient dimension, based on the low proportion of their comments in the labels task. (Rather, it is swamped by negative affect and negative linguistic stereotypes.) We are therefore inclined to be cautious about the notion of a sociolinguistic British Wales, at least as an English language zone. To conclude, we have shown how Welsh English can carry Welsh identity and that this occurs not only in areas identified with Welsh speakers but also in some relatively non-Welsh-speaking regions. Welsh identity is not necessarily lost where the colonial language has all but displaced Welsh itself. But although Welshness is a powerful perceptual discriminator of the English dialects, it may be better to think in terms of several Welshnesses. Prestige and affective properties, for example, are not uniformly

“WELSHNESS” AND “ENGLISHNESS” AS ATTITUDINAL DIMENSIONS

343

linked to Welshness but vary independently across the dialects. Welsh English dialects have their own individual profiles.

Notes 1. This research was funded by a University of Wales Intercollegiate Research Award granted to Nikolas Coupland and Peter Garrett. We are grateful to Jackie Guendouzi and Cathryn Williams for help with collecting, collating, and analyzing data and to the schoolteachers and administrators for their time.

C 21 Dialect Recognition Angie Williams

1.

Peter Garrett

Nikolas Coupland

Introduction1

1.1 The Significance of Dialect Recognition Dialectology and variationist sociolinguistics have been largely concerned with the distribution of dialect forms across geographical and social spaces (e.g., Labov 1972; Chambers and Trudgill 1980). The perceptual reality of dialect forms, on the other hand, and the distribution of beliefs about dialect variation have typically been researched in an independent tradition of “language attitudes” studies (for a review, see Giles and Coupland 1991a, chap. 2). For an adequate account of the social impact of dialect (or “dialect in use,” see Coupland 1988), data on both these dimensions are needed, preferably in integrated research designs. In both these traditions, there has been a tendency to presume that people regularly and fairly accurately recognize localized dialect varieties as part of their sociolinguistic competence as speech community members. In Labovian community surveys, for example, the question of how consistently community members, let alone outsiders, identify the varieties in question as regionally or socially localized forms is rarely asked. Yet one would expect that low levels of dialect recognition would necessarily limit people’s ability to position themselves psychologically within, or in opposition to, local community norms. A dialect is often presumed to confer a sense of social identity. But the extent to which people map social or geographical information onto dialect forms that they hear around them, or for that matter that they use themselves, would appear to be an important intervening variable. Furthermore, work on dialect recognition by Preston (1993d: 188ff) provides evidence that nonlinguists may be far more sensitive to dialect boundaries than dialect surveys are and that they may identify and be influenced by boundaries that linguists have not yet discovered. In relation to the language attitudes research tradition in which respondents are presented with voice samples that they are then asked to rate on a range of scales, Preston

346

IMAGES, PERCEPTIONS, AND ATTITUDES

(1989: 3; 1993d: 193ff) is critical of the failure to include a request to respondents to “tell me where you think this voice is from.” Though this seems a simple technique to add to attitude surveys, it is rarely done, and language attitude results are made extremely difficult to interpret because the respondents’ areal taxonomy and identification of regional provenience of the voice samples are not known. (Preston 1989: 3)

Such studies have not all been altogether lacking in recognition procedures for the voice stimuli, as Preston himself mentions. In many studies, prior to the main data collection, voice samples have been “validated” by asking other groups of respondents (assumed to be comparable with the later respondents) to rate them in various ways, including for their geographical origins. The rationale for conducting this procedure prior to the main study is that it enables the researcher not to draw attention to this variable in the main study itself, thereby reducing, at least arguably, the likelihood that the respondents will judge the language variety rather than the speaker. However, as we shall illustrate below, designing recognition out of the main study has arguably tended to draw attention away from the theoretical significance and potential social psychological complexity of dialect recognition. Our chapter reports on one aspect of a language attitudes study investigating young adults’ evaluative reactions to regional dialects of Welsh English. We focus our attention on the responses to a recognition question. Before outlining the study, however, we first consider two preliminary issues: 1. 2.

What do we know about the development of attitudes toward languages and language varieties by early adulthood? What is the relationship between language awareness (and, whatever processes are subsumed in “recognition,” they must certainly include awareness) and language attitudes?

1.2 Developmental Aspects of Awareness and Attitudes Research into children’s ability to distinguish their own language from that spoken by others suggests that such awareness is developed around the ages of five or six years (Aboud 1976), if not by 3.6 to 4.0 years (Mercer 1975). There is some lack of accord in the findings regarding the development of language attitudes. Labov (1965), for example, claimed that children did not become aware of the social significance of their dialect until early adolescence. However, there is evidence in other studies that children are already making judgments about varieties before they begin primary education (see review in Day 1982). One pattern that emerges from such studies (Rosenthal 1974; Cremona and Bates 1977; Day 1980) is that minority pupils enter school with either neutral attitudes or a preference for their own speech code and then acquire the attitudes of the dominant culture as they grow older. But there is an exception to this pattern in the work of Schneiderman (1976) who found that not all minority pupils acquire favorable attitudes

DIALECT RECOGNITION

347

toward the majority language, possibly because they perceive their own speech as a valued symbol of in-group identity and pride (Giles et al. 1983). Labov’s claim that adolescence is the key period for attitudinal development should not be pushed aside, however. Language attitudes are seemingly not fully developed by puberty. Through adolescence, attitudes toward majority varieties may become far more positive. For example, during adolescence, there is a general tendency for Welsh bilingual children to have a more positive attitude toward English and a less positive attitude toward Welsh or to bilingualism itself (Lewis 1975; Baker 1988, 1985). Giles (1970) also found some significant differences between his 12– and 17–year-old judges in their evaluations of accented speakers. Or there may even be a more marked attitudinal reversal during adolescence, involving a swing away from the attitudes of the dominant culture toward those of the minority in-group. Hence, Lambert, Giles, and Picard (1975) found that, although 10–year-olds in the St. John’s Valley of Maine in the United States were rejecting French ethnicity and language in favor of English assimilation, high school and college students were gradually placing European and local French on a par with English. For the purposes of this chapter, we can reasonably draw some broad conclusions from this complex set of developmental studies. First, although there is evidence that children have become socialized into appreciating some aspects of the perceived socioeconomic correlates of speech varieties by the age of about 11 or 12, it is clear that young adulthood can also be an interesting developmental period from a language attitudes perspective. Young adults are sensitizing themselves to sociolinguistic norms and reappraising their own sociolinguistic identities. This is not surprising, because they are likely to be on the verge of moving out into more fluid life patterns, perhaps into employment and new possibilities in their social relationships. More generally, there appears to be an underlying assumption that awareness of the different varieties available in the community is a condition for the existence of attitudes toward them. And this does indeed seem to be a plausible assumption. 1.3 Awareness, Recognition, and Attitudes On the other hand, there is an argument that the ability to recognize what language or variety one is listening to makes no difference to attitudes. The argument is that the sounds of the language or variety, rather than where it is anchored geographically or socially, are what trigger the evaluative reactions in listeners: that is, that people respond to the inherent value of the language. The inherent value hypothesis (see Giles and Powesland 1975) is not totally lacking in supporting evidence. For example, Brown, Strong, and Rencher (1975) audio-recorded French Canadian speakers from a variety of social class backgrounds and presented the speech samples to a group of Anglo-American students. Although the students had no knowledge of French and could not recognize the varieties, they were able broadly to differentiate between the speakers in terms of social class. It is certainly plausible that there are sufficiently general correlations between phonetic/prosodic realizations and broad sociodemographic categories to allow interpretations of this sort to be made. In any event, we would expect listeners to transfer aspects

348

IMAGES, PERCEPTIONS, AND ATTITUDES

of their sociolinguistic appreciation of dialects they know to dialects they do not know, and this could generate reliable patterns of evaluation. Generally, however, the received position argues that judgments are not based on inherent value but on imposed social norms, or social connotations. For example, Giles, Bourhis, and Davies (1975) played an audio recording of various forms of Quebec French to people in Wales and to French Canadian listeners. In contrast to the French Canadians, the attitudes recorded by the Welsh respondents did not differentiate the varieties on measures of prestige and aesthetic qualities. (See also Giles, Bourhis, Trudgill, and Lewis [1974] for a similar study with Greek dialects.) If social connotations are filtering through into attitudes, then, arguably, attitudes and recognition should go hand in hand. But it is already clear that what we are calling “recognition” can be a relatively complex process. What has been called the “inherent” value of a feature or a style is necessarily an evaluation made in the light of particular sociolinguistic understandings and experiences. “Recognizing” a dialect is inseparable from that same cluster of affective and evaluative processes, and it is in these terms that our own study has addressed it. 1.4 The Data Context In this chapter, we consider data relating to young adults’ recognition of English dialect varieties in Wales as part of a series of integrated studies on the sociolinguistic stereotyping of English in Wales [see Chapter 20, this volume]. Although there has been a tendency in the literature to treat “Welsh English” as a single variety, English in Wales shows very considerable phonetic, and some morphological and lexical, variation from region to region within this small country of approximately three million people (Coupland 1990). Variation is conditioned by a variable history of anglicization (Williams 1990; Aitchison and Carter 1994) and by different degrees of contemporary influence from the Welsh language. There might be said to be two distinct rural “heartland” varieties of English, just as there are in the distribution of the Welsh language itself (one northwestern and one southwestern), that show contemporary influence from two principal dialectal varieties of the Welsh language. A large area of northeast Wales shows phonetic/phonological influences from the Liverpool conurbation (just over the national border). The inland valleys of the South and Southeast, which sustained the vibrant coal and steel economies of the Industrial Revolution in Wales, are culturally as well as sociolinguistically distinctive. “The Valleys,” with the Rhondda Valleys as their prototype, have carried the social stereotype of Welshness itself since this period, partly through media portrayals of harsh working-class coal-mining culture and male voice choirs. However, Cardiff (the capital city) and other closely linked centers on the southeast coastal plain are dialectally distinct from the Valleys, separated by a significant dialect boundary (not yet researched in any detail) just north of the city. For many people, the Cardiff accent is less successful than Valleys English as a signifier of Welshness. It shows features associated historically with the southwest of England, Liverpool, and even Irish English (Coupland 1988; Windsor Lewis 1990).

DIALECT RECOGNITION

349

Hence, the sociolinguistic map of Wales, not untypically of the United Kingdom as a whole, shows considerable diversity over relatively small geographical distances. Moreover, it is a diversity that we have found to be reflected in the perceptions of a sample of teachers from all over Wales (Coupland, Williams, and Garrett 1994; Garrett, Coupland, and Williams 1995; Garrett, Williams, and Coupland 1996; Williams, Garrett, and Coupland 1996). We found them making the broad divisions mentioned above: Northwest, Northeast, Southwest, Southeast Valleys, Southeast Urban (e.g., Cardiff), and a large Mid-Wales region. To emphasize this, the principal dialect communities that are the focus of this study are known to have linguistic as well as some psychological reality for community members [see Chapter 20, this volume]. Our objective in the present study is to begin assessing young adults’ perceptions of and attitudes toward these dialects. Specifically, in this chapter, we look at their responses to a recognition question of the type Preston recommends for such research (see above).

2.

Method

State secondary schools were contacted in towns in the six dialect regions of Wales: Northwest (Blaenau Ffestiniog), Cardiff, Southwest (Carmarthen), the Valleys (Merthyr Tydfil), Mid-Wales (Newtown), and Northeast (Mold) (see Figure 21.1). These, plus one private school in Cheltenham, England, were invited to participate in a project concerned with regional variation in English. Of the fifteen schools contacted, fourteen agreed to participate. The aim of the first phase of the investigation was to audio-record young people in Wales talking as naturally as possible. To achieve this, pupils (average age 14) were encouraged to tell the investigator a story that was newsworthy in some way. To avoid drawing attention to accent variation, they were told that the project was concerned with “stories that young people your age tell.” They were prompted with a version of the following instruction: All people are storytellers. You come to school every day and tell your mates about things that have happened to you. This is what I want you to do today. Think of something that has happened to you or someone you know and tell us about it. For example, a funny or embarrassing incident, a frightening story, accident or danger, or a time you got into trouble with your parents.

A total of 179 stories were collected from fourteen sites where recordings were made. Thereafter, two stories were selected from each of the six Welsh regions as representative (simply on the criterion of the phonetic/phonological characteristics of the speaker’s voice) of its dialect norms. A further two stories were selected from those told by pupils at the school in Cheltenham, where there was a far higher concentration of standard English accents (received pronunciation, or RP). In this case, the stories were selected on the criterion of speakers speaking RP. All the stories selected were by male speakers (in part because more males than females had volunteered stories).

350

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 21.1. Map of Wales showing places mentioned in this chapter

The next step was to prepare a listening tape for the attitude data-gathering task. Since the stories varied in length, and to guard against the fatigue factor in the data gathering, a thirty-second snippet was taken from each of the fourteen recorded stories selected. The second phase of the study involved a second round of visits to different schools in the same six regions of Wales. In this phase, the fourteen selected audio-recorded story snippets were played in randomized sequence to independent groups of listener judges of the same age, who were asked to complete a questionnaire evaluating each speaker and asking where they thought each speaker lived. For this recognition item, they were given a choice of seven answers (i.e., the six Welsh English dialect communities, plus RP), along with a further two: a “don’t know” and an open-ended option for other responses. We also collected data from a group of teachers (from South Wales only) for comparison.2

351

DIALECT RECOGNITION

3.

Results and Discussion

Overall, teachers were far more successful on the recognition item than the adolescents were (see Table 21.1). Table 21.1. Percentages of correct recognition of the community origins of the speakers Voice

Teacher’s answers

Young adults’ answers

Cardiff 1 Cardiff 2

Cardiff Cardiff

68% 53%

? Cardiff

21% 42%

N.East 1 N.East 2

England England

47% 30%

? England

28% 27%

S.West 1 S.West 2

S.West S.West

45% 47%

Valleys Valleys

39% 26%

Valleys 1 Valleys 2

Valleys Valleys

53% 68%

Vall/SW Valleys

22% 33%

MidWales 1 MidWales 2

MidWales ?

34% 28%

MidWales ?

28% 27%

N.West 1 N.West 2

N.West N.West

49% 47%

N.West N.West

29% 26%

RP 1 RP 2

England England

77% 85%

? England

34% 44%

Key: ? = don’t know.

For the adolescents, who are our main focus in this chapter, the recognition rates were generally low. Table 21.2 shows that, overall, the speakers were correctly identified by between 20% and 44% of all the young adult listeners. Two speakers were correctly identified by more than 40% of the adolescents: Cardiff 2 and RP 2. Given the considerable exposure to received pronunciation (RP) (e.g., most notably in the broadcast media), a high recognition performance for this variety was anticipated, and it is indeed striking that RP 1 was recognized by only 26.6% of all judges (see Table 21.2). Why Cardiff 2 was so well recognized compared with Cardiff 1 is initially also hard to explain.3 In general, given that the pairs of voices representing each dialect community were selected on phonetic/phonological criteria, it is difficult to establish why there should be divergent rates of recognition between members of these pairs. In Table 21.2, where we can see a breakdown of how each community of judges fared on this task, it is even more puzzling why adolescents’ recognition of voices from their own dialect communities (see bold figures) should in some cases differ so much between the pairs of speakers. McNemar’s tests4 showed that the differences in recognition between speaker pairs by judges in their own communities were significant in the cases of Cardiff (chi-square = 13.00; p < 0.01), the Valleys (chi-square = 5.33; p < 0.04), and the Northeast (chi-square = 13.20; p < 0.01).

352

IMAGES, PERCEPTIONS, AND ATTITUDES

Table 21.2. Percentage of young adults who achieved correct recognition of where speaker was from Young adults’ communities (judges) Cardiff n = 23

NE 30

SW 33

Valleys 29

Mid. 25

NW 29

All 169

Cardiff 1 Cardiff 2

043.5 100.0

13.3 30.0

21.2 36.4

17.2 34.5

20.0 36.0

10.3 27.6

20.1 42.0

NE 1 NE 2

08.7 21.7

73.3 23.3

03.0 18.2

06.9 06.9

04.0 40.0

20.7 13.8

20.1 20.1

SW 1 SW 2

08.7 21.7

10.3 14.3

72.7 54.5

03.4 13.8

24.0 28.0

31.0 13.8

26.6 24.9

Valleys1 Valleys2

34.8 26.1

23.3 03.4

39.4 45.5

13.8 41.4

04.0 64.0

13.8 17.2

21.9 32.5

Mid 1 Mid 2

17.4 21.7

30.0 03.3

36.4 09.1

10.3 00.0

44.0 28.0

27.6 31.0

27.8 14.8

NW 1 NW 2

17.4 08.7

23.3 16.7

54.5 60.6

10.3 03.4

24.0 36.0

37.9 24.1

29.0 26.0

RP 1 RP 2

43.5 82.6

20.0 40.0

39.4 54.5

17.2 27.6

32.0 44.0

10.3 20.7

26.6 43.8

Speakers

It is tempting to explain these results by attributing them to “simple misrecognition.” If recognition is simply a matter of individuals’ mapping the speech features they hear onto their perceptual records of speech community norms, then wrong answers to the recognition question could be caused by individuals’ not having reliable perceptual records of the out-group norms. In other words, perhaps these adolescents do not have sufficient experience and awareness of these varieties. The developmental studies reviewed earlier did not involve the elicitation of attitudes to a whole range of nonstandard varieties, as the present study does. It seems reasonable to assume that this does indeed account for some of the instances of wrong responses in our data. In the case of their own dialects, misrecognition might imply adequate experience but inadequate cognitive representation (or awareness) of in-group norms. The levels of in-group recognition are indeed striking, varying from 100% to 13.8%, with a mean for in-group recognition just below 45%. The picture becomes more intriguing when we take a closer look at some of the misrecognitions. One might expect misrecognitions to lead to fairly random responses, but in fact, there are some very consistent patterns in the data, not represented in Table 21.2 (which records only levels of correct recognition). Among the Cardiff listeners, we found that 44% thought Cardiff 1 was from southwest Wales, 39% thought Valleys 2 was from southwest Wales, and 20% thought

DIALECT RECOGNITION

353

Valleys 1 was from southwest Wales, and these are not untypical values for correctly recognized varieties. Such a regular pattern leads one to wonder what there might be about these speakers, or about the listeners, that leads Cardiff judges to attribute them to the southwest of Wales. In the opposite direction, 52% of the Cardiff judges thought Northwest 2 was from their own dialect community, when these are in fact two very distinct varieties of English.5 Among the Valleys listeners, it is fair to say that there was an overall failure to recognize Valleys 1. Conversely, 51% of the Valleys judges thought Southwest 1 was from the Valleys, and 10% of them thought Cardiff 2 was from the Valleys. Such regularities could be taken as evidence that we are not dealing with simple misrecognition here but that other processes are at work. These processes could be competing with and overwhelming what are in fact quite reliable perceptual records in the adolescents, or they could be taking compensatory effect where the adolescents’ perceptual records are not yet fully developed. To return, however, to the notion that recognition is a question of mapping the speech features heard onto a cognitive template, a further possibility that needs to be considered for misrecognition is that the speech features themselves misled the adolescents in our study (but the teachers far less so). Perhaps some of the individual dialect samples contained more, or more salient, phonological cues than others did, linking them to community norms. Or perhaps, since these were all authentic recordings, rather than controlled matched guises (Lambert 1967), they contained different content cues, narrative characteristics, and so on that might lead to faulty identification. There could be some degree of phonological explanation for the tendency seen in Table 21.1 for both teachers and pupils to identify northeast Wales speakers as English, due to the influences of the nearby Liverpool conurbation, which we pointed to earlier. Such an explanation might also apply to the above-mentioned confusion of Northwest 2 with Cardiff by Cardiff listeners, since both speech communities share the fronted and somewhat raised /a:/ feature (Coupland 1988), and the word car is repeated five times in the 30–second snippet from Northwest 2’s story, along with single instances of bar and go-cart. However, this has to be seen against relatively good rates of accurate recognition of Northwest 2 by southwest Wales (60.6%) and Mid-Wales (36%) listeners. 3.1 Welshness Although such phonological and content factors might account for some proportion of the lack of accurate identification, results relating to another item on the questionnaire given to respondents lead us to be skeptical of the ability of these factors to explain all of it. For the same reason, we have reservations about the idea that the results can be fully explained by the earlier suggestion that the perceptual records of these adolescents might not be adequate for this task. One item on the questionnaire asked our respondents to rate the voices according to how Welsh they sounded to them. The results for this item are set out in Table 21.3. Looking simply at the mean scores set out in Table 21.3, we see some signs of

354

IMAGES, PERCEPTIONS, AND ATTITUDES

Table 21.3. Mean scores of young adult and teacher judges on the item: How Welsh do you think this speaker sounds? Teachers South West South West Valleys North West North West Valleys Cardiff Cardiff Mid-Wales North East Mid-Wales North East RP RP

Adolescents 4.54 4.53 4.36 4.20 3.76 3.70 3.15 3.02 2.30 2.16 1.98 1.52 1.25 1.14

South West Valleys South West Valleys North West North West Cardiff Cardiff Mid-Wales Mid-Wales North East North East RP RP

4.47 4.01 3.91 3.76 3.44 3.41 2.88 2.83 2.29 2.14 1.88 1.60 1.56 1.51

pairing of the samples for each dialect. What is also evident here is that the pairing occurs for both the adolescents and the teachers, who achieved higher rates of recognition. It would seem from this that the adolescents do in fact recognize (at some level of awareness) that there are pairs of speakers from the same dialect community.6 The notion that the adolescents are operating with an awareness at some level of what these varieties are is not as problematic as it might first appear. Preston (1996b) sets out a taxonomy of folklinguistic awareness in which he identifies four independent continua along which awareness might be identified at different levels. These independent dimensions are availability, detail, accuracy, and control. Figure 21.2 shows these dimensions, with hypothetical settings marked on them to illustrate their independence. Preston also lists a number of factors that might contribute to these levels of awareness. Space precludes a summary of all of these, but they include formal training and/or knowledge, publicity (e.g., popular culture, media exposure), correctness (transmitted formally or informally), and folk culture artifacts. AVAILABLE -----X------------------ UNAVAILABLE DETAILED ---------------------X---- GLOBAL ACCURATE --------X---------------- INACCURATE FULL CONTROL ----------------X-- NO CONTROL Figure 21.2. Variable independent continua of the modes of non-linguists’ awareness, with hypothetical settings (after Preston 1996b)

DIALECT RECOGNITION

355

Facets of language differ in their degree of availability to nonlinguists. Some are common topics of discussion, whereas others are never discussed or are commented on only with some prompting (= availability). Second, their comments about language may differ in their degree of detail. Third, nonlinguists’ descriptions of language may differ in their degree of scientific accuracy. Fourth, people may or may not be able to perform a variety (= control). Most relevant to the concerns of this chapter are the first three continua, since our recognition task precludes performance. Recognition can be interpreted as one form of availability, in this case prompted by audio-recordings. Correct recognition of the dialect would reflect accuracy. The “detail” continuum is of interest here because the questionnaire predetermined the level of detail at which the adolescents were being asked to comment when they made their judgments of recognition. It is possible that some of the “wrong answers” can be explained by this. For example, we mentioned earlier a number of instances in which speakers from the Valleys were identified as coming from southwest Wales and vice versa. Our questionnaire gave multiple choice options for respondents to select from, which included three regions from South Wales: the urban Southeast (i.e., Cardiff), southwest Wales, and the Valleys. Conceivably, some respondents were able to operate only at a lower level of detail than we asked of them, and they might, therefore, have felt confident that the voice they heard was from a nonurban (or non-Cardiff) area of South Wales but not have been able to select confidently from the two alternatives of southwest Wales or the Valleys. Their recognition would be available and accurate but not sufficiently detailed. Some evidence that such a process might be at work is to be found in Table 21.3, where there appears to be more pairing of voices by dialect community for all the regions except southwest Wales and the Valleys. For the young adults in our study, these latter two intermesh. (However, we are looking only at descriptive statistics in this table.) On the other hand, it is still striking that percentages choosing southwest Wales rather than the Valleys (or vice versa) in response to the recognition item tend to point markedly more in one direction than the other in each instance. One would expect that if these responses were simply due to a lack of detail in their awareness, results would not tend to pull in one direction much more than in the other. Hence, again, it is likely that other factors are at work as well. The question that remains to be answered in our study, then, is what factors might lead to the degree of recognition we detect in the “How Welsh?” item that do not have the same influence on the responses to the recognition item itself. On the “How Welsh?” item, we might, for example, expect influence from several of the factors that Preston lists as affecting the settings on the continua: publicity, folk culture artifacts, correctness, and so on, all contributing to a cognitive mapping process. On the issue of recognition, however, we decided to explore the possibility that this did not simply require a mapping of speech features onto a cognitive template but that there are affective, group-level factors at work: recognition might be influenced by an active in-grouping process (Tajfel 1974). In Preston’s taxonomy, affective factors are not included in the list of factors influencing

356

IMAGES, PERCEPTIONS, AND ATTITUDES

settings on the dimensions, but if there is evidence of them in our data, they would make a useful addition to this theoretical model of dialect awareness. 3.2 Likability Accordingly, to test the hypothesis that the more recognized speaker would be rated higher on affiliation by their in-group community, further analysis was conducted on the speaker pairs in which one was significantly more recognized than the other. The hypothesis was tested by calculating t tests comparing speaker pair means for the three communities in question on two other items on the questionnaire that measured how much the listener liked the speaker and to what extent the listener thought the speaker was a “good laugh.” For Cardiff, this hypothesis was supported for both liking (M C1 = 2.39 vs. M C2 = 3.18; t = 2.94; p < 0.01) and “good laugh” (M C1 = 2.13 vs. M C2 = 4.13; t = 6.94; p < 0.01). The hypothesis was also supported in the case of northeast Wales for both liking (M NE Wales 1 = 3.00 vs. M NE Wales 2 = 3.70; t = 2.42; p < 0.01) and “good laugh” (M NE Wales 1 = 3.26 vs. M NE Wales 2 = 4.20; t = 3.56; p < 0.01). The hypothesis was not supported in the case of the Valleys speakers, however, for liking (M Valleys 1 = 1.59 vs. M Valleys 2 = 1.61) or “good laugh” (M Valleys 1 = 1.59 vs. M Valleys 2 = 1.56), as neither of these speakers was rated as particularly likable or a “good laugh” by the Valleys judges. There is further support for the hypothesis elsewhere in our data. For example, the results for the Valleys listeners show that, although 24.1% recorded a “don’t know” for the Valleys 1 speaker, who did not score high on likability, 51% thought Southwest 1 was from the Valleys, and 10% thought Cardiff 2 was from the Valleys. Both Southwest 1 and Cardiff 2 scored high on likability. Similarly, 52.2% of Cardiff listeners thought that Northwest 2 was from Cardiff, whereas there is a weaker misattribution for Northwest 1, a difference echoed in their likability ratings. There is evidence, then, of claiming (and indeed, in the case of misrecognition, denial) processes at work in the recognition data. Most striking of all in this respect are the results for the Cardiff 2 speaker, who scores highest of all on the likability ratings. Although he is correctly identified as a Cardiff speaker by the majority in all the judging groups, he is also the most claimed: 26.7% of the northeast listeners claim that he is from northeast Wales, 18.2% of the southwest Wales judges say he is from southwest Wales, 16% of Mid-Wales listeners think he is from Mid-Wales, and 10.3% of Valleys listeners claim he is from their dialect community. Also, in contrast to all our other results, he is correctly identified as a Cardiffian by 100% of the listeners from the Cardiff community. The processes of claiming and denial illustrated above find a theoretical foundation in theories of social identity and self-categorization (Hogg 1992) and are essentially concerned with social attraction. The focus of social (as opposed to personal) attraction is the set of prototypical properties of the group (although social and personal attraction may coexist). Intergroup relations define groups, and the groups to which one belongs are a resource for defining one’s personal identity. So there is a continual process of competition and innovation to establish and maintain a relatively positive evaluation of one’s own group. A consequence is that in-group prototypes are generally evaluated positively. In the

DIALECT RECOGNITION

357

task we gave to the adolescents, they were arguably able to decide for themselves whether the speakers they heard were in-group or out-group members. Hence, from the speech performances they heard, listeners were able to focus on the salient dimensions of group identity and to decide whether speakers approximated the prototype they valued most or not or even whether they more closely approximated some out-group prototype that the listeners knew well and valued negatively. The apparently simple task of giving a community label to a particular speaker may well have tapped into these group-level cognitions and influenced the frequencies with which particular speakers were recognized as members of in-group and out-group communities.

4.

Overview

Despite the long tradition of language attitudes research and the very welcome recent interest in perceptual dialectology, sociolinguistics still seems to lack a unified perspective on dialect in use. Awareness and recognition of dialects is certainly part of this broad agenda. In this chapter, we have considered various possible ways of defining recognition. We can construe recognition to be the cognitive mapping of audible speech features (or stylistic configurations of features in combination) onto individuals’ records of the usage norms of particular communities. By this account, recognizing a dialect involves identifying values of variable features and then succeeding or failing to make the appropriate mapping. In Preston’s terms (again see Figure 21.2), failing to recognize a dialect is one form of nonavailability at specifiable levels of (in)accuracy and detail. This language awareness model helps us explain several key aspects of our data. The important general finding that the young adults in our study produce lower levels of recognition than the teachers can be accounted for by their predictably lower levels of dialect experience. They probably have experienced lower geographical mobility and less access to dialect speakers, face-to-face or in the broadcast media. The young adults show lower levels of availability (in this particular, receptive sense of availability), which might be the result of less accurate or less detailed cognitive templates of English language sociolinguistic variation in Wales. Even if levels of exposure were similar between younger and older people, we might suggest that secondary school-age youth may have less inferential experience than their teachers — less experience in assessing the social significance of the sociolinguistic differences they encounter. There are some minor, but intriguing, trends in our data that a cognitive mapping model is not fully able to explain. Dialect recognition, to the extent that this label is an adequate one for the sociolinguistic processes activated in asking the question “Where is the speaker from?” is, of course, a cognitive process. But it is also part of social cognition — people’s cognitions about social group memberships, group identities, and group boundaries. Listeners did not merely recognize or fail to recognize speakers as belonging to specific communities. In some respects, their identifications responded to and manipulated the group designations that we offered them. Most notably, there was a tendency for a very likable speaker to be actively appropriated into the in-group.

358

IMAGES, PERCEPTIONS, AND ATTITUDES

We feel that the group-level, affective dimension of dialect recognition and of language awareness deserves more attention in future research. Affective factors could usefully be added to Preston’s summary list of factors that influence an individual’s set of locations along his or her independent modes of language awareness. It could be argued that such factors are likely to dominate in recognition tasks in which accurate cognitive mapping cannot be achieved: for example, when listeners are inexperienced. But social cognition about dialect can equally well be seen as a relatively sophisticated sociolinguistic activity, responding to preferences and ideologies that dominate in listeners’ own communities. We suspect that processes such as claiming and disavowing are an intrinsic part of dialect recognition processes, even when listeners’ records are well developed and speech cues are unambiguous. Sociolinguistic studies need to ask not only what speech community members know about varieties but also how they construct this knowledge and how they use it creatively to reflect and refine their group priorities and memberships. The study of dialect recognition, linked to studies of dialect variation and other aspects of dialect awareness, seems to be a profitable new direction.

Notes 1. This research was funded by a University of Wales Intercollegiate Research Award granted to Nikolas Coupland and Peter Garrett. Earlier versions of this chapter were presented at the Third International Conference on Language Awareness (held at Trinity College Dublin, Ireland, July 1996) and at AILA ‘96 (held at the University of Jyväskylä, Finland, August 1996). 2. We did not visit the same schools as in the recording phase for fear that speakers would be personally recognized, but instead we selected a second set of schools to offer the best social and regional matches to those where the original speakers were recorded, following advice from members of the school staff. In one or two instances, the selected best match declined the invitation to participate in the study, and we had to go further afield within the same region. For this reason, the northwest judges were in Bethesda, the Valleys judges were in Bedwellty, and the Mid-Wales judges were in Builth Wells. 3. We refer to the individual speakers as either 1 or 2 for a particular region, without any significance in the 1 versus 2. 4. McNemar’s test (1975) is a variant of chi-square for frequency and nominal data in which the same group of participants is measured twice (i.e., correlated samples). In this case, the frequency of correct recognition of one speaker compared with another within a pair is analyzed. 5. Northwest Wales English is very dependent on the Welsh language for its distinctive features, specifically modeled on the North Wales Welsh dialects, whereas Cardiff is not — or only marginally — influenced by Welsh. One feature, however, is a possible exception: long /a:/, as we mention later. 6. One might conceivably still argue that the pairings are due to chance or that they have more to do with story content than with the phonological qualities themselves. We would dismiss both these possibilities with the following arguments. First, the similarity between the adolescents’ and the teachers’ results, both in terms of the pairing and the rank ordering of the dialects, effectively rules out chance. Second, in an earlier stage of our work, we sent out questionnaires to teachers all over Wales asking them to record their attitudes toward the same Welsh English dialects. On that occasion, the varieties were not presented through speech samples but were instead presented conceptually. Hence, no story content and no audible phonological features were present. The results showed a very similar hierarchical structure for judgments of Welshness (Coupland, Williams, and Garrett 1994; Garrett, Coupland, and Williams 1995; Williams, Garrett, and Coupland 1996).

C 22 A Language Attitude Approach to the Perception of Regional Variety Dennis R. Preston

1.

Some Background

Although studies of the folk perception of regional varieties began in Japan and The Netherlands in connection with production dialect studies (see Parts I and II), their more recent use has been to supplement work on language attitudes. From this perspective, I complained (Preston 1989) that language attitude research did not determine where respondents thought regional voices were from and, worse, did not know if respondents even had a mental construct of a “place” where a voice could be from; that is, their mental maps of regional speech areas might not include one with which a sample voice could be identified. For example, if one submitted a voice from New England to California judges and the judges agreed that the speaker was “intelligent,” “cold,” “fast,” and so on, researchers could reasonably conclude that Californians judged that voice sample in that way. They should not conclude, however, that that is what Californians believe about New England voices, for a majority of the judges might not have agreed that the voice was from New England. (Perhaps they would have called it a “New York” voice.) More generally, Californians may not even have a concept of “New England” speech. Perhaps the most detailed mental map of regional U. S. speech available to them is one that simply identifies the “Northeast” (whatever their folk name for that region might be). It is odd that this long-standing problem has caused so little discussion among those who do language attitude studies. In perhaps the earliest “classic” study of attitudes toward regional and ethnic varieties in the United States, Tucker and Lambert (1969) note that neither northern nor southern European American judges identified the ethnicity of educated African American speakers better than chance (scores ranging from 47% to 54%), but in spite of their concluding plea in that article to investigate identification when ratings are done, their suggestion has been very seldom followed.

360

IMAGES, PERCEPTIONS, AND ATTITUDES

Of course, that failure does not vitiate all language attitude research work that has not asked for regional and/or ethnic identification of the sample voices presented for evaluation. Milroy and McClenaghan (1977) note an interesting consistency of ratings of Scottish, Southern Irish, Ulster, and received pronunciation (RP) varieties even when judges misidentified accents. They comment on this finding as follows: It has been widely assumed that an accent acts as a cue identifying a speaker’s group membership. Perhaps this identification takes place below the level of conscious awareness. … Presumably by hearing similar accents very frequently [one] has learnt to associate them with their reference groups. In other words, accents with which people are familiar may directly [italics in original] evoke stereotyped responses without the listener first consciously assigning the speaker to a particular reference group. (8–9)

Irvine (1996) has more recently commented on this transfer of linguistic features to social facts that apparently make the unconscious reactions that Milroy and McClenaghan (1977) note possible: Iconicity is a semiotic process that transforms the sign relationship between linguistic features and the social images to which they are linked. Linguistic differences appear to be iconic representations of the social contrasts they index — as if a linguistic feature somehow depicted or displayed a social group’s inherent nature or essence [italics in original]. (17)

In other words, the presumed social attributes of a group are transferred to the linguistic features associated with it (as Irvine notes), and an occurrence of those features may directly trigger recognition of those attributes without being filtered through (conscious) identification of the group (as Milroy and McClenaghan note). Perhaps responses to language and language variety may operate along a continuum (or several continua) of consciousness or “awareness” (just as language performances involve degrees of “monitoring” or “attention to form” [e.g., Labov 1972: 208]). In Preston (1996b) I review a number of these possibilities for “folk linguistics,” suggesting that folklinguistic facts (i.e., linguistic objects as viewed by nonlinguists) may be subdivided for awareness along the following clines. 1. 2. 3.

4.

Availability: Folk respondents range in their attention to linguistic features from complete disregard for to frequent discussion of and even preoccupation with them. Accuracy: Folk respondents may accurately, partially accurately, or completely inaccurately represent linguistic facts (and their distribution). Detail: Folk respondents’ characterizations may range from global (reflecting, for example, only a general awareness of a variety) to detailed (in which respondents cite specific details). Control: Folk respondents may have complete, partial, or no “imitative” control over linguistic features.

An important fact about these several clines is their relative independence. For example, a respondent who claims only a general awareness of a “foreign accent” may be capable

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

361

of a completely faithful imitation of some of its characteristics and a completely inaccurate imitation of others. On the other hand, a respondent who is preoccupied with a variety might have no overt information about its linguistic makeup but be capable of performing a native-like imitation of it.1 Perhaps the range of so-called language attitude effects ought to be treated in a similar way. That is, attitudinal responses based on the respondents’ association of a sample voice with a particular social group may be different from ones based on reactions to linguistic caricatures such as ain’t. Responses that may be based on some sort of cline (e.g., masculine-feminine, degree of “accent”) may be different from those based on the recognition of “categorical” features (e.g., correct-incorrect). Whatever the answer to these and other questions, there is no doubt that folklinguistic responses to regional varieties can profit from being investigated by using a variety of research methods. I have tried to piece out some of the attitudinal puzzle concerning responses to U. S. regional varieties by determining the mental maps of regional varieties (from a number of different regions) and by asking respondents to characterize regions of the U. S. (usually by state) on scales of language “correctness,” “pleasantness,” and “degree of difference” (from the home area of the judges [e.g., Preston 1996a]). In this research, I try to combine the results of these folk-linguistic investigations with the more classic language attitude research model. That is, I try to answer at least a part of my own criticism of language attitude research by employing the “cognitively real” mental maps of speakers in an assessment of their attitudes toward regional variation.

2.

Methods

The first requirement of this project was to make use of the respondents’ actual mental map of regional speech areas. In previous work (e.g., Preston 1989), following the lead of cultural geographers (e.g., Gould and White 1986), I have simply asked respondents to draw maps of where they believe varieties are different. Figure 22.1 is a typical example of such a hand-drawn map. Although one may profit from an investigation of these individual maps (e.g., by looking at the labels assigned various regions; see Hartley and Preston, 1999), their usefulness for general language attitude studies depends on the degree to which generalizations may be drawn from large numbers of such maps. This may be done by drawing an (approximate) boundary for each salient region from the first map and then “overlaying” each subsequent respondent’s map and drawing the “perceptual isoglosses” for each region.2 A more sophisticated version of this procedure makes use of a digitizing pad that feeds the outlined area of each salient region into a computer so that a more precise numeric determination can be made of the “boundary” of each hand-drawn region (Preston and Howe 1987). Figure 22.2 shows a computer-determined map for the mental map of U. S. regional speech areas derived from the hand-drawn maps of 147 southeastern Michigan respondents (from a variety of status and age groups, male and female).

362

IMAGES, PERCEPTIONS, AND ATTITUDES

Figure 22.1. A Michigan hand-drawn map

Figure 22.2. Computer-assisted generalizations of hand-drawn maps showing where southeastern Michigan respondents believe speech regions exist in the United States

Given this mental map determined in previous research (one, it was assumed, that would be typical of the respondents to be investigated in this research), it was next determined, in classic language attitude research style (e.g., Shuy and Fasold 1973), what labels would be relevant to an investigation of attitudes to those areas. For the respondents

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

363

studied here (young, European American, Michigan university students), the most frequently mentioned labels were the following: Slow – Fast Smart – Dumb Nasal – Not Nasal Drawl – No Drawl

Formal – Casual Polite – Rude Normal – Abnormal Twang – No Twang

Educated – Uneducated Snobbish – Down-to-Earth Friendly – Unfriendly Bad English – Good English

These descriptors were elicited by showing a large number of respondents (none of whom participated in the later evaluation task) a simplified version of Figure 22.2 and asking them to mention any characteristics of the speech of those regions that came to mind. The most frequently mentioned descriptors were selected and arranged into the pairs shown above. These pairs were then presented in a 6–point “semantic differential” task as shown below. The respondent judges (85 young, European American southern Michigan residents who were undergraduate students at Michigan State University) were shown a simplified version of Figure 22.2 and given the following instructions:

Instructions This map shows where many people from southern Michigan believe speech differences are in the U. S. We will give you a list of descriptive words which local people have told us could be used to describe the speech of these various regions. Please think about twelve3 of these regions, and check off how each pair of words applies to the speech there. For example, imagine that we gave you the pair “ugly” and “beautiful” ugly

____ a

____ b

____ c

____ d

____ e

____ f

beautiful

You would use the scale as follows: – – – – – –

If If If If If If

you you you you you you

very strongly agree that the speech of a region is “ugly,” select “a.” strongly agree that the speech of a region is “ugly,” select “b.” agree that the speech of a region is “ugly,” select “c.” agree that the speech of a region is “beautiful,” select “d.” strongly agree that the speech of a region is “beautiful,” select “e.” very strongly agree that the speech of a region is “beautiful,” select “f.”

Use the op-scan form (and the numbers on it) for all answers. 1. First, please tell us your sex a. female b. male Go on to Region 1 (which begins with Question 2 on the next page). Refer back to the map on this page whenever you like. Thank you very much for your cooperation.

364 3.

IMAGES, PERCEPTIONS, AND ATTITUDES

The Results

The first step in classic language attitude work is to determine whether or not the paired items used in evaluating the “samples” can be reduced. This is normally carried out by means of a factor analysis. The results of such analysis for all areas rated are shown in Table 22.1. Two robust factor groups emerge. The first (which I will call Standard) shows loadings from those categories that one associates with education and majority norms. Note, however, that the last three factors in this group (Formal, Fast, and Snobbish) are not necessarily positive traits. Factor Group 2 (which I will call Friendly) loads affective factors (including two that are negatively loaded — Down-to-Earth and Casual — in Factor Group 1). Table 22.1. The two factor groups from the ratings of all areas Factor Group 1 Smart Educated Normal Good English No drawl No twang Casual [Formal] Fast Down-to-earth [Snobbish]

Factor Group 2 −.76 −.75 −.65 −.63 −.62 −.57 −.49 −.43 −.32

Polite Friendly Down-to-earth (Normal) (Casual)

(.74 (.74 (.62 (.27) (.27)

Note: Parenthesized factors indicate items that are within the .25 to .29 range; ‘−’ prefixes indicate negative loadings and should be interpreted as loadings of the opposite value (given in brackets).4

These groups will not surprise old hands at language attitude research. As Ryan, Giles, and Sebastian (1982) note, “With regard to the structure of attitudes toward contrasting language varieties, the two major dimensions along which views can vary can be termed social status and group solidarity [italics mine]” (8). A full analysis of these data would, of course, go on to consider the realization of each of these factors (and groups) with regard to each of the areas rated. First, that would be too space-consuming, and second, I believe a sample of two particularly salient areas (for these respondents) will provide a good insight into the mechanisms at work here. I have chosen to look at the respondent ratings of areas 1 and 2 from Figure 22.2. The reasons are straightforward. Region 1 is the U. S. South, and Figure 22.2 shows that it was outlined by 94% (138) of the 147 respondents who drew hand-drawn maps. For these southeastern Michigan respondents, it is clearly the most salient regional speech area in the United States. Although one might note anecdotal or popular culture characterizations of why that might be so, a look at Figure 22.3 will provide an even more dramatic explanation. In an earlier task in which I asked southeastern Michigan respondents to rate

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

365

the states (and Washington, D. C., and New York City) for correctness, it is clear that the South fares worst. On a 1–to-10 scale (with 1 being least correct), Alabama is the only state that reaches a mean score in the 3.00 to 3.99 range, and, with the exception of New York City and New Jersey, the surrounding southern states (Texas, Arkansas, Louisiana, Mississippi, Tennessee, and Georgia) are the only other areas rated in the 4.00 to 4.99 range. In short, the salience of southern speech would appear to lie in its distinctiveness along one particular dimension — it is incorrect English. The second most frequently rated region (by 90 of 147 respondents or 61%) is the local one called North in Figure 22.2 but more accurately North Central or Great Lakes. At first, one might be tempted to assert that the local area is always salient, but a closer look at Figure 22.3 will show that these southeastern Michigan raters may have something else in mind when they single out their home area. Only Michigan scores in the heady 8.00 to 8.99 means score range for language “correctness.” In short, perception of language correctness (in the positive direction) determines the second most salient area for these respondents.

Figure 22.3. Means of ratings for language ‘correctness’ by MI respondents for US English (on a scale of 1 to 10, where 1 = least and 10 = most correct)

Although investigation of the ratings of other areas will doubtless prove interesting, a careful look at those of the high-prestige local area (North) and of the most highly

366

IMAGES, PERCEPTIONS, AND ATTITUDES

stigmatized area (South) will prove most revealing. Table 22.2 shows the means scores for the individual attributes for the North and South. Perhaps the most notable fact is that the ranked orders are nearly opposites. “Casual” is lowest-rated for the North but highest for the South. “Drawl” is lowest-rated (meaning “speaks with a drawl”) for the South but highest rated (meaning “speaks without a drawl”) for the North. In factor group terms, the scores for Factor Group 2 (and −1 loadings) are the lowest-ranked ones for the North; these same factors (e.g., Casual, Friendly, Down-to-Earth, Polite) are the highest ranked for the South. Similarly, Factor Group 1 scores are all low ranked for the South; the same attributes are all highest ranked for the North. Table 22.2. Means scores of attributes Means scores (ordered) North

Means scores (ordered) South

Rank Factor

Mean

Attribute

Rank Factor

Mean

Attribute

12 11 9.5 9.5 8 7 6 5 4 3 2 1

3.53 3.94@ 4.00 4.00@ 4.09 4.12 4.19 4.41 4.53 4.94 5.07 5.11

Casual Not nasal Friendly Polite Educated Fast Down-to-earth Good English Smart Normal No twang No drawl

1 2 3 4 5

4.66 4.58 4.54 4.20@ 4.09@

Casual Friendly Down-to-earth Polite Not nasal

‡3.22 ‡3.04 ‡2.96# ‡2.86 ‡2.72 ‡2.42# ‡2.22

Normal [Abnormal] Smart [Dumb] No twang [Twang] Good English [Bad English] Educated [Uneducated] Fast [Slow] No drawl [Drawl]

−1 & 2 Ø 2 2 1 1 2 & −1 1 1 1&2 1 1

6 7 8 9 10 11 12

−1 & 2 2 2 & −1 2 Ø * 1&2 1 1 1 1 1 1

Note: * marks the only significant (p < 0.05) break between two adjacent scores (determined by an analysis of variance with a Tukey comparison of means); ‘‡’ marks values below 3.5 (which indicate the opposite polarity, shown in brackets here); ‘#’ indicates the only scores significantly different for gender (p < 0.05, determined by a series of t tests); ‘@’ marks the only two attributes (Nasal and Polite) for which there was no significant difference (p < 0.05 on a series of paired t tests) between the ratings for North and South.

These scores are not just ordered differently. As indicated by the “@” in Table 22.2, a series of paired t tests shows that there is a significant difference (p < 0.05) between the attribute ratings for the North and the South, except for Nasal and Polite. For those attributes that load on Factor Group 1 (No Drawl, No Twang, Fast, Educated, Good English, Smart, and Normal), the means scores are all higher for the North. In other words, these Michigan raters consider themselves superior to the South for every attribute of the “Standard” factor group. This is not very surprising, considering the results from earlier research on correct English shown in Figure 22.3. Before considering the scores for the attributes in Factor Group 2 (Friendly), let’s see what Michigan raters have done previously in a direct assessment of the notion “pleasant”

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

367

(as was shown above in Figure 22.3 for correctness). As Figure 22.4 shows, the South fares very badly again. Alabama (actually tied here by New York City) is the worst-rated area in the United States, and the surrounding southern states are also at the bottom of this 10–point rating scale. One may note, however, that the ratings for the “pleasantness” of the English of southern states are 1 degree less harsh than those for “correctness.” Similarly, there is no “outstanding” (8.00–8.99) rating as there was for correctness, making Michigan no longer the uniquely best-thought-of area (since it is joined here by Minnesota, Illinois, Colorado, and Washington). In previous work (e.g., Preston 1996a), I have taken this to indicate that northern speakers have made symbolic use of their variety as a vehicle for standardness, education, and widely accepted or mainstream values. On the other hand, southern speakers (who are well aware of northern prejudices against their variety) use their regional speech as a marker of solidarity, identity, and local values.5

Figure 22.4. Means scores for ‘pleasant’ English by Michigan respondents. Note: Ratings as in Figure 22.3.

For those attributes that load on Factor Group 2 (or −1), the means score is higher for the South for Casual, Friendly, and Down-to-Earth. There is no significant difference for Polite (as noted above), and the North leads the South in Factor Group 2 attributes only for Normal, but it is important to note that Normal loaded (positively) on both groups.

368

IMAGES, PERCEPTIONS, AND ATTITUDES

This is indeed a new finding for rankings of the prejudiced-against South by linguistically secure northerners. These data suggest that, for these eighty-five young Michiganders, the Friendly attributes (excepting only Polite) are more highly associated with southern speech than with speech from the local area. A few other statistical facts confirm and add to the results reported so far. Sex (the only demographic variable testable in this experiment) played little or no role. No rating of any attribute for the North differed by sex, and only Twang and Fast differed for the South. (Female raters found the South “twangier” and slower.) More important, note (in Table 22.2) that no attribute rating for the North falls below 3.5 (the median value of the 6–point scale), whereas all of the Factor Group 1 (Standard) attributes are rated below that score for the South. Perhaps even more dramatically, analysis-of-variance tests of the means scores for North and South independently show that there is no significant break (p < 0.05) between any two adjacent means scores for ratings of the attributes for the North. On the other hand, there is such a significant difference for the South between the Factor Group 2 (and −1) attributes and the Factor Group 1 attributes, as shown by the “*” in Table 22.2. In other words, there is a continuum of relatively positive scores for the North and a sharp break between the two factor groups for the South. This break can be even more directly shown in Table 22.3, which displays the combined means scores for the two factor groups and the two areas focused on here. Only Factor Group 1 (Standard) for the South is very different from any other. Unfortunately, this representation of the results hides the important fact that ratings of individual attributes for North and South are nearly all significantly different (as shown in Table 22.2 above). What it does reveal, however, is that this model of research found considerably better ratings of the South by northerners along the affective dimension than did previous research on pleasantness ratings (as shown, for example, in Figure 22.3). Table 22.3. Factor group means (all attributes combined) North

South

Factor Group 1

4.44

*3.18

Factor Group 2

4.13

*4.24

4.

Conclusions

What does this approach contribute to language attitude study? Most important, I believe we can be relatively assured that the judges have rated regions that are cognitively real for them; that is, they have rated areas for which the notion regional speech has been shown to have folk-linguistic status. Unlike classic matched-guise attitude studies, this research provides respondents with the category name and mapped outline of regions rather than actual voice samples. The

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

369

obvious benefit of this is that I do not have to use what could only be gross, stereotypical imitations of varieties (if one speaker tried to imitate all the varieties studied here or even the two more carefully looked at). Since some recent language attitude research has shown that there is little or no difference in evaluations when the stimulus is a category name or an actual speech sample (e.g., Coupland, Williams, and Garrett, Chapter 20 [this volume]; Williams, Garrett, and Coupland, Chapter 21 [this volume]), I have not considered this manner of presenting the stimulus to be a deficiency. Of course, the question of whether or not respondents can identify varieties is still an open one and requires independent study. Here, I chose to investigate the “stereotypes” respondents have of regional voices (without submitting a sample), and I consider this to be one approach to the larger question of identification of and attitudes toward regional varieties. It is the case that the respondents in this study were all young, college-enrolled undergraduates, but the mental map (Figure 22.2) that was shown to them was derived from a study of a variety of age groups and social classes in southeastern Michigan. In fact, however, there were found to be very few important differences in age, gender, and status representations of dialect regions or even in the evaluations of them (e.g., Preston 1988c). I believe, therefore, that there are no important differences between the cognitive map of regional speech for the respondents studied here and the generalization shown them (derived from Figure 22.2). The major finding of this study, however, is that there is a considerable difference in the rankings here of the affective dimension of attitudes of Michiganders to the South and those given by similar respondents in my earlier research. As Figure 22.4 shows, the ratings for the local area for pleasantness were among the highest, and the ratings for the South along the same dimension were among the lowest. In the present study, however, the South actually did significantly better than the local area in three key characteristics of the affective factor group (Factor Group 2 — Friendly, Casual, and Down-to-Earth) and was not significantly different on a fourth (Polite). As Table 22.3 shows, there is hardly any difference between the overall ratings of Factor Group 2 for the North and South. That is a very different picture from the one seen in Figure 22.4. What accounts for this amelioration of attitude toward the South among these raters? I think there are several possibilities. First, one might assume that the global label “pleasantness” (used in the earlier research) does not as subtly (or perhaps as covertly) elicit the attitudes along this dimension. That is a real possibility, but I cannot resolve it here. Second, one might attribute this amelioration to the age of the respondents. Although they are the same age as the youngest group studied in the earlier research, they are certainly not their contemporaries. Those earlier data were collected in 1986–87; the data for the current study were collected in 1996. Since I do not have ratings from older respondents in the current research, however, it is difficult to make this comparison straightforwardly. There is some evidence (although it is confounded by region) that the respondents of the late 1990s may behave differently from those of the late 1980s, but the difference would appear to be in the area of evaluations of correctness (more closely parallel to Factor Group 1, Standard) than along affective lines. In her work in Oregon,

370

IMAGES, PERCEPTIONS, AND ATTITUDES

Hartley (1996) notes that a number of respondents (citing what sounds very much like a somewhat sophisticated linguistic relativism) refused to evaluate regions of the United States for correctness (or rated them all the same).6 She notes, however, no such accompanying reluctance for rating pleasantness, and the student field-workers for the present study reported no such reluctance among these young southeastern Michigan judges in rating correctness or pleasantness. Third, one might suspect that some sort of “covert prestige” attaches itself to southern speech (since it is clearly seen as incorrect). If that were the case, however, one might expect to see a strong gender differentiation (with a male preference for the stigmatized variety), but as Table 22.2 shows, there is little gender significance in the ratings. In addition, high ratings for attributes such as Friendly hardly point to “tough” characteristics. I believe, however, that this last possibility moves in the right direction, but I also believe that previous definitions of covert prestige are too tough and male oriented to cover the entire territory. Let us consider another possible interpretation. Although many hand-drawn maps of U. S. dialect areas by Michigan respondents label the local area standard, normal, correct, and good English, some also treat it as seen in Figure 22.1 (where it is called boring). Since there is obviously no dissatisfaction with the local variety as a representative of correct English, what is the source of the preference for other varieties along affective dimensions? Recall that I have suggested (e.g., Preston 1996a) that a group has a tendency to use up what might be called the “symbolic linguistic capital” of its variety in one way or another (but not both). Speakers of majority varieties have a tendency to spend the symbolic capital of their variety on a Standard dimension. Speakers of minority varieties usually spend their symbolic capital on the Friendly dimension. I suggest that northerners (here, southeastern Michiganders) have spent all their symbolic linguistic capital on the standardness of local English. As such, it has come to represent the norms of schools, media, and public interaction and has, therefore, become less suitable for interpersonal use. In short, these young Michiganders don’t identify other varieties for their covert prestige on the basis of anti-establishment or tough characteristics alone; they also assign covert prestige to a variety that they imagine would have more value than theirs for interpersonal and casual interaction, precisely the sorts of dimensions associated with Factor Group 2. Of course, I do not doubt the existence of covert prestige along the traditional masculine or tough lines that Trudgill (1972) points out; I simply suspect that there are other kinds of covert prestige, or at least one in which friendship, solidarity, trust, informality, strong emotion, and such factors are highlighted. Southern U. S. English would appear to be such a variety for these judges. I will not develop here the popular culture, folkloristic, and qualitative evidence for this interpretation, although I am sure such caricatures (many encoded in the notion “southern hospitality”) are well-known, and northerners indeed comment that southern speech “sounds nice.”7 From a language variation point of view, of course, we are ultimately more interested in the general social and linguistic mechanisms at work here. Ryan, Giles, and Sebastian (1982: 9) outline the following evaluative possibilities for majority (LV1) and minority (LV2) speakers:

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

371

Type of preference

A. Majority group B. Majority group for status/ in-group for solidarity C. In-group D. Majority group for status/ minority group for solidarity

Judges LV1 Speakers Status Solidarity LV1 LV1

LV2 Speakers Status Solidarity LV1 LV1

LV1 LV1

LV1 LV1

LV1 LV2

LV2 LV2

LV1

LV2

LV1

LV2

In these terms, I wonder if speakers of inland northern U. S. English (i.e., the Michiganders studied here) have changed from Type B to Type D. Perhaps speakers of some of these varieties have moved in the direction of RP speakers in Britain (the group that Ryan, Giles, and Sebastian use to illustrate the LV1 pattern of their Type D). In other words, the inappropriateness of their own (Standard) variety to interpersonal modes of communication has caused them to evaluate other (nonstandard) varieties higher for the characteristics identified as belonging to the Friendly factor group.8 Space will not allow a thorough discussion of other interpretive dimensions of this finding, but I will briefly mention some. In Preston (1992) I note that, although young European American imitations of African American speech might be regarded as racist, many appear to have other motivations — sounding not only tough and “cool” but also casual and down-to-earth. This motivation among younger speakers is complex. Although adolescents are often presented with a dichotomous choice between mainstream (“approved”) and nonmainstream (“rebellious”) behaviors, a middle ground exists in which there is a desire to succeed along traditional lines but another to display egalitarian principles, ones that require, on the linguistic front, the (at least partial) use of varieties seen as stigmatized. As a result of other associations with both the standard and the perceived nonstandards, these latter varieties also seem to be more appropriate for casual, interpersonal use. I believe the “in-betweeners” in Eckert’s (1989) suburban Detroit study (i.e., those who want to be neither the mainstream “Jocks” — perceived as “snooty” — nor the anti-establishment “Burnouts” — associated with drug culture) display just such an attitude. One of them characterizes this dilemma of such “Neutrals” as follows: They [i.e., the Neutrals, neither Jocks nor Burnouts] just don’t want to seem to turn to drugs to cope with their problems, and, uh, they want to, they want to have good grades, you know, but not be stuck up where you’ll look at someone and say “Well, you are lower than me,” and stuff like that. (174)

In Britain as well, Rampton’s work (e.g., 1995) evaluates the occurrence and meaning among adolescents of cross-ethnic language use (i.e., “code-switching”) and concludes that a principal function is its reflection of a desire to do away with ethnic boundaries.

372

IMAGES, PERCEPTIONS, AND ATTITUDES

In other words, in resolving the adolescent tension between mainstream and nonmainstream behavior, a linguistic option might be the use of Standard English in settings that require that variety and a mixing of the speaker’s native variety with perceived nonstandards in settings that require casual use.9 In short, I do not believe that the use of or preference for nonstandard (or stigmatized) varieties by adolescents is uniquely associated with the “anti–language” interpretation offered by Halliday (e.g., 1976) and apparently embedded in most interpretations of covert prestige. I will not press this favoring of stigmatized varieties into service for general sociolinguistics too much further, but I want to mention that it is one (alternative) option in the search for accounts for the introduction and spread of novel elements (particularly into the mainstream speech community); that is, it is another option in the search for answers to the problems of “actuation” and “embedding” (Weinreich, Labov, and Herzog 1968). Rampton’s work referred to above also notes that the adolescent use of other varieties provides an opportunity for “practice” (while making the social statement that such selection implies). If that is true, then all the cases of accommodation, imitation, and acquisition of socially motivated nonmainstream varieties discussed here (and in many other places in the sociolinguistic literature) are opportunities for the introduction and spread of alternative forms. This may be particularly true when those elements are introduced into the wider speech community by adolescents whose own native variety is closer to the mainstream (although their initial motivation in acquiring them was accommodation to nonmainstream varieties as a part of the solution to the “adolescent dilemma” outlined above). This interpretation is potentially an addendum to the “weak ties” argument for linguistic change outlined in J. Milroy (1992). In suggesting that persons with low-density networks are likeliest to be the “early adapters” (e.g., 183) in linguistic change, Milroy appears to overlook the age factor (in which younger, even adolescent, speakers seem to lead). If we use Eckert’s (1989) terms to characterize the members of adolescent social networks, both Jocks and Burnouts have high-density group relations (corresponding to the high solidarity values of the highest- and lowest-status speakers shown in Figure 7.1 of Milroy 1992: 213). The weak ties group (like the “lower-middle” and “upper-working” status sectors in Milroy’s representation) would be the Neutrals of Eckert’s suburban Detroit adolescents. They are not among themselves a tightly knit cohesive group, and they “borrow” linguistically from the lower-status (or “nonmainstream”) Burnouts. As such, they are in a privileged position to introduce such elements into the wider speech community.10 The young southeastern Michiganders reported on here have similar privilege, and although we cannot know their various social status backgrounds, we can assume that their university status will have some effect on their later social position. That they prefer a stigmatized variety to their own for affective characteristics suggests that they are not only changing their attitudinal perception to the Type D outlined above but that they are also (potential) borrowers of norms from stigmatized speech communities in their own attempts to achieve a more casual, interpersonal style. Of course, a great deal more quantitative and qualitative work will need to be done to establish this direct link between attitudes and language change, but I believe such work

ATTITUDE APPROACH TO THE PERCEPTION OF REGIONAL VARIETY

373

will be a productive enterprise, and I urge attitudinal work that focuses on the occurrence (and frequency) of specific linguistic features, as done in, for example, Labov (1966) and Graff, Labov, and Harris (1986) but so seldom replicated in more recent studies. Even if this direct link cannot be shown conclusively to exist, the patterns of language regard outlined here form an important part of the study of variation in its social context, one that has implications for both the more broadly based ethnographic approach to sociolinguistics and linguistic intervention in schools, law, medicine, communications, and other areas of public concern.

Notes 1. In Preston (1996a) I provide a detailed analysis of the phonetic and phonological failures of a northerner (Michigander) talking “southern” (saying the phrase “Y’all know what I’m talking about now, don’t you”) (347–49). 2. I did this for respondents from Hawaii, southern Indiana, southern Michigan, western New York, and New York City in Preston (1986). 3. Texas and California (areas 8 and 12 in Figure 2) were excluded from the rating so as to limit the task to one large op-scan form (“electronically scorable answer sheet”). Since “Texas” and “Southwest” and “California” and “West Coast” overlapped considerably in the generalization of the hand-drawn task, this was not seen as especially detrimental. 4. Although the paired opposites were presented to the respondents with “negative” and “positive” sides randomly distributed, the positive poles were all moved to the high (i.e., “6”) end of the scale for all the quantitative analyses reported below. I realized after I did this that there might be cultural misunderstandings of what I consider to be the positive end. They are Fast, Polite, Down-to-Earth, Educated, Normal, Smart, Casual, Good English, Not nasal, Friendly, Speaks without a drawl, and Speaks without a twang. I apologize to readers who disagree with my assignments. That should not detract from the contents of the chapter. 5. In fact, maps of southern respondents’ ratings of “correctness” and “pleasantness” show just the opposite pattern of Michigan raters. For example, Alabama raters regard their own state’s variety as uniquely pleasant but rank themselves in the middle on the correctness scale. 6. Hartley suggests that Oregonians are aware of their diverse US origins (including southern backgrounds) and are reluctant on those grounds to evaluate other regions, but this interpretation would appear to be confounded by their not being reluctant to rate the same regions for “pleasantness.” 7. In fact, there are many such comments in Niedzielski and Preston (forthcoming). 8. In fact, these northern raters may have been Type D for some time, and the caricaturistically blunt “pleasant” assessment I asked for in earlier work was simply not sensitive enough to elicit that aspect of their evaluation of southern speech. 9. Of course I know that there is such a “linguistic” object as “Standard U. S. southern English,” but it does not exist for the nonlinguist northern respondents under discussion here. It is simply one of a variety of “incorrect” Englishes (e.g., Preston 1996a). 10. Labov’s “lames” might also be seen as such “early adapters.”

References

Aboud, F. E. 1976. Social and developmental aspects of language. Papers in Linguistics 9:15–37. Aitchison, J., and H. Carter. 1994. A geography of Welsh. Cardiff: University of Wales Press. Allen, Harold. 1964. Aspects of the linguistic geography of the Upper Midwest. In Studies in language and linguistics in honor of Charles C. Fries, edited by Albert Marckwardt, 303–13. Ann Arbor: University of Michigan. Bach, Adolf. [1934] 1950. Deutsche Mundartforschung: Ihre Wege, Ergebnisse und Aufgaben [German dialect research: Its paths, results, and problems]. 2nd ed. Heidelberg: C. Winter (3rd ed., published 1969). Bahner, Werner, Joachim Schildt, and Dieter Viehweger eds. 1991. Proceedings of the Fourteenth International Congress of Linguists, Berlin/GDR, August 10–15, 1987. Berlin: Akademie-Verlag. Baker, C. 1985. Aspects of bilingualism in Wales. Clevedon: Multilingual Matters. ———. 1988. Key issues in bilingualism and bilingual education. Clevedon: Multilingual Matters. Ball, P. 1983. Stereotypes of Anglo-Saxon and non-Anglo-Saxon accents: Some exploratory Australian studies with the matched-guise technique. Language Sciences 5:163–84. Balsom, D. 1985. The three Wales model. In The national question again: Welsh political identity in the 1980s, edited by J. Osmund, 1–17. Llandysul: Gomer. Baumgartner, Heinrich. 1922. Die Mundarten des Berner Seelandes [The dialects of Bern Seeland]. Beiträge zur schweizerdeutschen Grammatik XIV. Frauenfeld: Huber & Co. Becker, Ulrich, Horst Becker, and Walter Ruhland. 1992. Zwischen Angst und Aufbruch: Das Lebensgefühl der Deutschen in Ost und West nach der Wiedervereinigung [Between fear and awakening: Germans’ perceptions of life in East and West after reunification]. Düsseldorf, Vienna, New York, and Moscow: ECON. Belic, J. 1965. Zur Problematik der Grenzen zwischen verwandten Sprachen (dargestellt am Material slawischer Sprachen) [On the problem of the boundaries between related languages (on the basis of the Slavic languages)]. Zeitschrift der Universität Leipzig, Gesellschafts- und sprachwissenschaftliche Reihe 14(1):105–10.

376

REFERENCES

Benhardi, K. [1844] 1849. Sprachkarte von Deutschland [Linguistic map of Germany]. 2nd ed. Kassel: J. J. Bohn. Bezoen, H. L. 1948. Taal en volk van Twente [Language and people of Twente]. Assen: Van Gorcum. Boal, F. W. 1976. Ethnic residential segregation. In Social areas in cities, edited by D. Herbert and R. Johnston, Vol. 1, 41–79. London: Wiley. Bonnemason, Jean. 1993. Les langues de France. Paris: Publications de l’ecole moderne francaise. Bourhis, R., and Howard Giles. 1976. The language of co-operation in Wales: A field study. Language Sciences 42:13–16. Bourhis, R., Howard Giles, and H. Tajfel. 1973. Language as a determinant of Welsh identity. European Journal of Social Psychology 3:447–60. Brook, G. L. 1963. English dialects. London: A. Deutsch. Brown, B., W. Strong, and A. Rencher. 1975. Acoustic determinants of the perceptions of personality from speech. International Journal of the Sociology of Language 4. Büld, H. 1939. Sprache und Volkstum im nördlichen Westfalen: Sprachgrenzen und Sprachbewegungen in der Volksmeinung [Language and nationality in northern Westphalia: Perceptions of linguistic boundaries and language movements]. Emsdetten: H. & G. Lechte. Butters, Ronald R. 1995. Some current issues in variation theory and practice. In Viereck (ed.), 3–36. Carton, Fernand et al. 1983. Les accents des français. Paris: Hachette. Carver, Craig M. 1987. American regional dialects: A word geography. Ann Arbor: University of Michigan Press. Ceuleneer, A. de. 1909. Pieter Willems, zijn leveln en werken [Pieter Willems, his life and work]. Jaarboek Koninklijke Vlaamse Academie 23:213–337. Chambers, J. K. 1995. Sociolinguistic theory. Oxford: Blackwell. Chambers, J. K., and Peter Trudgill. 1980. Dialectology. Cambridge: Cambridge University Press. Chloupek, J. 1980. Die Stellung des Dialekts im Rahmen der Nationalsprache und beim kommunikativen Akt [The position of dialect in the framework of national language and in the communicative act]. In Dialekt und Dialektologie, Wiesbaden, edited by J. Göschel et al., 165–75. Zeitschrift für Dialektologie und Linguistik, Beihefte 26. Coupland, Nikolas. 1988. Dialect in use. Cardiff: University of Wales Press. ———, ed. 1990. English in Wales. Clevedon: Multilingual Matters. Coupland, Nikolas, and Alan Thomas. 1990. Introduction: Social and linguistic perspectives on English in Wales. In Coupland, 1–16. Coupland, Nikolas, Angie Williams, and Peter Garrett. 1994. The social meaning of Welsh English: Teachers’ stereotyped judgements. Journal of Multilingual and Multicultural Development 15(6):471–89. Cremona, C., and E. Bates. 1977. The development of attitudes toward dialect in Italian children. Journal of Psycholinguistic Research 6:223–32.

REFERENCES

377

Crochetière, André, Jean-Claude Boulanger, and Conrad Ouellion, eds. 1993. Les langues menacées: actes du XVe Congrès international des linguistes, Québec, Univeristé Laval août 1992 [Endangered languages: Proceedings of the XVth International Congress of Linguists, Quebec, Laval Univeristy, August 9–14, 1992]. Vol. 3. Sainte-Foy, Québec: Les Presses de L’Université Laval. Crystal, David. 1988. The English language. London: Penguin. Daan, Jo. 1955. De Amsterdamse olievlek [The Amsterdam oil stain]. Taal en Tongval 7:120–29. ———. 1957. Betrekkingen tussen het Fries en het Noordhollands [Relationships between Frisian and the North Holland dialect]. It Beaken 19:197–205. ———. 1963. Défaites et victoires: La lutte des parlers régionaux contre le néerlandais cultivé aux Pays Bas. Orbis 12:398–407. ———. 1969. Dialektatlas van Noord-Holland [Dialect atlas of North Holland]. Reeks Nederlandse dialektatlassen 13. ———. 1970. Dialekten [Dialects]. In Van randstad tot landrand [From “city-side” to “country-border”], edited by Jo Daan and D. P. Blok, 7–43. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 37. Daan, Jo, and K. Heeroma. 1965. Zuidhollands [South Holland dialect]. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 30. Dailey-O’Cain, Jennifer. 1997. Geographic and socio-political influences on language ideology and attitudes toward language variation in post–unification Germany. Ph.D. diss., University of Michigan. Dakin, Robert. 1966. The dialect vocabulary of the Ohio River Valley. Ph.D. diss, University of Michigan. Davis, Lawrence. 1990. Statistics in dialectology. Tuscaloosa: University of Alabama Press. Day, Richard. 1980. The development of linguistic attitudes and preferences. TESOL Quarterly 14:27–37. ———. 1982. Children’s attitudes towards language. In Ryan and Giles, 116–31. Diercks, Willy. 1988. Mental maps: Linguistisch-geographische Konzepte. Zeitschrift für Dialektologie und Linguistik 55:280–305. Eckert, Penelope. 1989. Jocks and burnouts: Social categories and identity in the High School. New York: Teachers College Press. Edwards, John. 1979. Judgements and confidence reactions to disadvantaged speech. In Giles and St. Clair, 22–44. Edwards, V. 1991. The Welsh speech community. In Multilingualism in the British Isles: The older mother tongues and Europe, edited by S. Alladina and V. Edwards, 107–25. London: Longman. Emnid-Institut. 1995. Stolz aufs eigene Leben. Spiegel 27:40–52. Entjes, H. 1974. Dialecten in nederland [Dialects in the Netherlands]. Haren-GN: Knoop and Niemeijer.

378

REFERENCES

Faries, Rachael, and Donald M. Lance. 1993. Regional variation in Missouri. In “Heartland” English: Variation and transition in the American Midwest, edited by T. Frazer, 245–56. Tuscaloosa: University of Alabama Press. Fox, J. 1982. Selective aspects of measuring resemblance for taxonomy. In Classifying social data: New applications of analytic methods for social science research, edited by H. C. Hudson et al. San Francisco. Frings, Theodor. 1956. Sprache und Geschichte, parts I-III [Language and History]. Reprints of papers published 1922ff. Halle: Max Niemeyer. Funk, W. P. 1991. Applying “Multidimensional Scaling” to linguistic variation: Accents of Standard English. In Bahner et al., 1496–99. Garrett, Peter, Nikolas Coupland, and Angie Williams. 1995. “City Harsh” and “The Welsh Version of RP”: Some ways in which teachers view dialects of Welsh English. Language Awareness 4:99–107. Garrett, Peter, Angie Williams, and Nikolas Coupland. 1996. Adolescents’ judgements of Welsh English dialects. Paper presented at Sociolinguistics Symposium 11, University of Wales Cardiff, September. Gazier, Augustin. 1969. Lettres à Grégoire sur les patois de France, 1790–1794. Geneva: Slatkine Reprints. Giles, Howard. 1970. Evaluative reactions to accents. Educational Review 22:211–27. ———. 1971. Patterns of evaluation to RP, South Welsh and Somerset accented speech. British Journal of Social and Clinical Psychology 10:280–81. ———. 1979. Accommodation theory: Optimal levels of convergence. In Giles and St. Clair, 45–65. Giles, Howard, R. Bourhis, and A. Davies. 1975. Prestige speech styles: The imposed norm and inherent value hypotheses. In Language in anthropology IV: Language in many ways, edited by W. McCormack and S. Wurm. The Hague: Mouton. Giles, Howard, R. Bourhis, Peter Trudgill, and A. Lewis. 1974. The imposed norm hypothesis: A validation. Quarterly Journal of Speech 60:405–10. Giles, Howard, and Nikolas Coupland. 1991a. Language: Contexts and consequences. Buckingham: Open University Press. ———. 1991b. Language attitudes: Discursive, contextual and gerontological considerations. In Bilingualism, multiculturalism and second language learning, edited by A. Reynolds, 21–42. Hillsdale, NJ: Lawrence Erlbaum. Giles, Howard, C. Harrison, C. Creber, P. Smith, and N. Freeman. 1983. Developmental and contextual aspects of British children’s language attitudes. Language and Communication 3:1–6. Giles, Howard, and Peter Powesland. 1975. Speech style and social evaluation. London: Academic Press. Giles, Howard, and Robert St. Clair, eds. 1979. Language and social psychology. Oxford: Blackwell. Ginneken, J. van. [1913] 1928. Handboek der Nederlandsche taal [Handbook of the Dutch language]. ’s Hertogenbosch: L. C. G. Malmberg.

REFERENCES

379

Goebl, Hans. 1982. Dialectometrie: Prinzipien und Methoden des Einsatzes der Numerischen Taxonomie im Bereich der Dialectgeographie [Dialectometry: Principles and methods of the use of numeric taxonomy in the area of dialect geography]. Vienna: Verlag der Österreichischen Akademie der Wissenschaften. Goeman, A. C. M. 1984. Klank- en Vormverschijnselen van het Dialect van Zoetermeer [Phonology and morphology of the dialect of Zoetermeer]. Publications of the P. J. Meertens Institute 4. Amsterdam: P. J. Meertens Institute. ———. 1987. Perceptie van de afstand dialect-standaardtaal in relatie tot nieuwe gegevens betreffende de Nederlandse dialecten [Perception of the distance between dialect and standard language in relation to new data on Dutch dialects]. In Eigen en Vreemd: Identiteit en ontlening in taal, literatuur en beeldende kunst [Identity and loans in language, literature and arts], Handelingen v. h. 39ste Nederlands Filologencongres 1986, Amsterdam, 411–22. ———. 1989. Dialectes et jugements subjectifs des locuteurs. Quelques remarques de méthode a propos d’une controverse [Dialects and the subjective judgments of speakers: Remarks on controversial methods]. Espaces Romans (Études de dialectologie et de géolinguistique offertes à Gaston Tuaillon. Vol. II. Université Stendhal — Grenoble 3: Ellug, 532–44. Goossens, J. 1965. Die niederländische Strukturgeographie und die “Reeks Nederlandse Dialectatlassen” [Dutch structural geography and the “Royal Dutch Dialect Atlas”]. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam, 29. ———. 1970. Dialectindeling en structuurgeographie [Dialect divisions and structural geography]. Driemaandelijkse Bladen Nieuwe Serie 22:49–73. Gould, Peter. 1977. Changing mental maps: Childhood to adulthood. Ekistics 255:100–21. Gould, Peter, and Rodney White. 1986. Mental maps. 2nd ed. Boston: Allen & Unwin. Graff, David, William Labov, and Wendell A. Harris. 1986. Testing listeners’ reactions to phonological markers of ethnic identity: A new method for sociolinguistic research. In Diversity and diachrony, edited by David Sankoff, 45–58. Current Issues in Linguistic Theory, Vol. 53. Amsterdam and Philadelphia: Benjamins. Grootaers, Willem. 1957. Les débuts de la géographie linguistique au Japon [The beginnings of linguistic geography in Japan]. Orbis 6(2):342–52. ———. 1958a. Saishô shûrakukan no hôgen tairitsu [The distribution of dialects in small villages]. Paper presented at the dialect sub-group of Nihon Kotoba no Kai [Japan Language Society], February 15. ———. 1958b. Nihon hôgen kenkyû ni okeru futatsu no atarashii hôhô [Two new approaches in Japanese dialectology]. International Congress of Orientalists in Japan 3:71–74. ———. 1958c. Itoigawa chôsa no hôkoku A [Report “A” on Itoigawa research]. Paper presented at the dialect sub-group of Nihon Kotoba no Kai [Japan Language Society], October 11. ———. 1958d. Methods to study Japanese Dialects. Paper presented at the Asiatic Society Meeting, Sophia University, November 10.

380

REFERENCES

———. 1959a. Etymology through maps. Folklore Studies 17:187–198. ———. 1959b. Hôgen chizu no kakikata to yomikata [How to draw and read dialect maps]. Paper presented at the First International Christian University Summer Language Conference, July 7 (published as Grootaers 1962). ———. 1959c. Origin and nature of the subjective boundaries of dialects. Orbis 8(2):355–84. ———. 1959d. Hôgen Kyôkaisen wo megutte [In search of dialect bounderies]. Lecture, Tokyo Metropolitan University. ———. 1962. Hôgen chizu no kakikata to yomikata [How to draw and read dialect maps, translated into Japanese by Takesi Sibata]. In Hôgengaku gaisetsu [General dialectology], edited by Kokugo Gakkai [Japanese Linguistic Association], 344–73. Tokyo: Musashino Shoin. ———. 1963. Les premiers pas à la recherche des unités dialectales. Orbis 12:361–80. ———. 1964. La discussion autor des frontières dialectales subjectives. Orbis 13:380–98. Grosse, Rudolf. 1955. Die Meissnische Sprachlandschaft. Halle: Saale. Gueunier, N., E. Genouvrier, and A. Khomsi. 1978. Les Français devant la norme. Paris: Champion. Haag, Karl. 1898. Die Mundarten des oberen Neckar-und Donaulandes [The dialects of the upper Neckar and Danube regions]. Reutlingen. Haeringen, C. B. van. 1955. Is tat juist, op tie manier [Is that correct, in this way?]. Taal en Tongval 7:49–57. Halliday, M. A. K. 1976. Anti-languages. American Anthropologist 78(3):570–83. Hammarström, Göran. 1961. Inquéritos Linguisticos [Linguistic Inquiries]. Revista de Portugal Series A, 26:9–32. (In Swedish: 1964. Dialektalogiska undersökningar. Nordisk tidsskrift for tale och stemme 24). ———. 1963. Réflexions sur la linguistique structurale et la phonétique expérimentale [Reflections on structural linguistics and experimental phonetics]. Phonetica 9(1):11–16. Hartley, Laura C. 1996. Oregonian perceptions of American regional speech. Master’s thesis. Department of Linguistics and Languages, Michigan State University. Hartley, Laura C., and Dennis R. Preston. 1999. The names of U. S. English: Valley girl, cowboy, Yankee, normal, nasal and ignorant. In Standard English, edited by Richard Watts and Anthony Bex, 207–238. London: Routledge. Haugen, Einar. 1966. Dialect, language, nation. American Anthropologist 68:922–35. Hayashi, Chikio. 1952. On the prediction of phenomena from qualitative data and the quantification of qualitative data from the mathematico-statistical point of view. Annals of the Institute of Statistical Mathematics 3:69–98. ———. 1954. Multidimensional quantification with the applications to analysis of social phenomena. Annals of the Institute of Statistical Mathematics 5:121–43. Heeroma, K. 1942. Beoordeling van A. Weijnen, De Nederlandse dialecten. Review of A. Weijnen, “Dialects of the Netherlands”. Nieuwe Taalgids 36:38–41. ———. 1961. Structuurgeografie [Structural geography]. In Structuurgeografie [Structural geography], edited by K. Heeroma and K. Fokkema, 1–15. Bijdragen en Mede-

REFERENCES

381

delingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 23. ———. 1963. De geografische indeling der Oostnederlandse volkstaal [The geographical division of the East Netherlands vernacular]. Taal en Tongval 15:175–81. Heeroma, K., and K. Fokkema, eds. 1961. Structuurgeografie [Structural geography]. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 23. Henzen, W. 1954. Schriftsprache und Mundarten: ein überblick über ihr Verhältnis und ihre Zwischenstufen im Deutschen [Standard language and dialects: An overview of their relationship and their intermediate stages in German]. 2d ed. Bern: Francke. Hertog, C. H. den. 1897. De Nederlandsche taal. Practische spraakkunst van het hedendaagsche Nederlandsch [The Dutch language: A practical grammar of modern Dutch]. Amsterdam: W. Versluys. Hoenigswald, Henry. 1966. A proposal for the study of folk-linguistics. In Sociolinguistics, edited by William Bright, 16–26. The Hague: Mouton and Co. Hof, J. J. 1933. Friesche dialectgeografie [Frisian dialect geography]. ’s-Gravenhage: Martinus Nijhoff. Hogg, M. 1992. The social psychology of group cohesiveness. London: Harvester Wheatsheaf. Hol, A. R. 1965. De g in hij heeft het en in ik, gij (enk.), hig en gij (meerv.), doet het en dergelijke ww. Vormen [/g/ in verbal forms such as he has it, I, you (sing.), he and you (plur) do it]. Taal en Tongval 17:32–40. Honey, John. 1989. Does accent matter? The Pygmalion factor. London: Faber and Faber. Houck, Charles L. 1967. A computerized statistical methodology for linguistic geography: A pilot study. Folia Linguistica 1:80–95. Inoue, Fumio. 1972. Chiiki meisho to hôgensha ishiki [Names of districts and consciousness of dialect differences]. Chimeigaku Kenkyu 1:18–30. ———. 1973. Hôgen’on Ninchi no Mekanizumu [Mechanism in the recognition of dialectal sounds]. Kokugogaku 95:94–106. ———. 1977/8. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of dialect image] (part 1). Gengo Seikatsu 311:82–91. ———. 1978/9. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of dialect image] (part 2). Gengo Seikatsu 312:82–88. ———. 1986. Sociolinguistic aspects of new dialect forms: language change in progress in Tokyo. International Journal of the Sociology of Language 58:73–89. ———. 1988. Dialect image and new dialect forms. Area and Culture Studies 38:13–23. ———. 1989. Kotoba-zukai Shin-fukei [New landscape of spoken Japanese]. Tokyo: Akiyama Shoten. ———. 1991. Dialect image and the diffusion of new dialect forms. In Bahner et al., 1508–11. ———. 1992. Tôkaidô ensen no hôgen shiyô to hôgen ishiki [Dialect usage and dialect image along the Tokaido Line]. Area and Culture Studies 45:11–50.

382

REFERENCES

———. 1993. Word-accent change in progress in Japanese: Multi-variate analysis of sociolinguistic groups. In Crochetière et al., 355–58. ———. 1995. Classification of dialects by image — English and Japanese. In Viereck, 355–68. ———. 1996. Subjective dialect division in Great Britain. American Speech 71(2):142–61. Inoue, Fumio, and Chitsuko Fukushima. 1997. A quantitative approach to English dialect distribution: Analyses of CLAE morphological data. In Viereck 1992–, 57–65. Inoue, Fumio, and H. Kasai. 1989. Dialect classification by standard Japanese forms. In Japanese quantitative linguistics, edited by S. Mizutani. Quantitative Linguistics 39:220–35. Irvine, Judith T. 1996. “Style” as distinctiveness: The culture and ideology of linguistic differentiation. Paper presented to the NSF Workshop on “Style,” Stanford University, February. Iwai, Ryûsei. 1961. Hôgen no jittai to kyôtsûgo-ka no mondai-ten “Toyama, Ishikawa” [The current dialect situation and the problem of standardization in Toyama and Ishikawa]. In Hôgengaku kôza 3 [Lectures on dialectology 3], edited by Misao Tôjô, 84–110. Tokyo: Tokyodô. Jaberg, K., and J. Jud. 1928–40. Sprach- und Sachatlas des Italiens und der Süsschweiz [Speech and object-atlas of Italy and Southern Switzerland]. Ringier: Zofinger. Jernudd, Björn H. 1968. There are no subjective dialects. Kivung 1:38–42. Kandori, Takehiko, and Shigeru Takanaga. 1988. Hôgen ni taisuru kôaku no ishiki: “Hiroshima hôgen” ni taisuru baai [Good-bad consciousness towards dialects: the case of Hiroshima dialect]. Kokubun gakkô 120:1–18. Karsten, G. 1931. Het dialect van Drechterland [The dialect of Drechterland]. Purmerend: J. Muusses. Kate, Lambert ten. 1723. Aenleiding tot de kennisse van het verhevene deel der Nederduitsche sprake 1 [An introduction to the study of the elevated elements of Low Dutch]. Amsterdam: Rudolf & Gerard Wetstein. Kiliaan [Cornelius Kilianus]. 1574. Etymologicum teutonicae linguae 1e dr. 1574; meest verspreide uitgave [most widely circulated edition] 1777. Antwerp: Christophorus Plantinus (printed by Gerardus Smits). Kirk, John M. et al. 1985. Studies in linguistic geography: The dialects of English in Britain and Ireland. London: Croom Helm. Kitamura, Hajime. 1952. Kodomo no kotoba wa ijû ni yotte dô kawaruka [How children’s language changes due to migration]. Gengo Seikatsu 8:20–27. Kloeke, G. G. 1919. De apokopeeringslijn in Groningen en Drente: Proeve van dialectgeografisch onderzoek [The apocope line in Groningen and Drente: A specimen of geographical dialect investigation]. Driemaandelijke Bladen 19:1–40. ———. 1927. De Hollandsche expansie in de zestiende en zeventiende eeuw en haar weerspiegeling in de hedendaagsche Nederlandsche dialecten [Expansion from Holland in the sixteenth and seventeenth centuries and its reflection in modern Dutch dialects]. ’s-Gravenhage: Nijhoff.

REFERENCES

383

———. 1950. Herkomst en Groei van het Afrikaans [Origin and Growth of Afrikaans]. Leiden: Universitaire Pers. ———. 1956. Op tie manier, is tat algemeen Hollands? [In this way, is that common Dutch?] Tijdschrift voor Nederlandse Taal en Letterkunde 74:296–300. Kloeke-van Lessen, M. J. 1968. G. G. Kloeke. Bibliografie en dagboekfragment [Bibliography and journal fragment]. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 35:29–58. Kremer, Ludger. 1979. Grenzmundarten und Mundartgrenzen: Untersuchungen zur wortgeographischen Funktion der Staatsgrenze im ostniederändisch-westfälischen Grenzgebiet [Border dialects and dialect boundaries: Studies of the word-geographic function of the national border in the east and Westphalian border region]. Two vols. Cologne, Vienna: Bählau. ———. 1983. Standardsprachliche Transferenz und die Definition niederländischer und/oder deutscher Dialekte [Transference of the standard language and the definition of Dutch and/or German dialects]. In Een spyeghel voor G. Jo Steenbergen, edited by F. Daems and L. Goossens. Leuven. ———. 1984. Die niederländisch-deutsche Staatgrenze als subjektive Dialektgrenze [The Netherlands-German national border as a subjective dialect boundary]. In Grenzen en grensproblemen, 76–83. Een bundel studies uitgegeven door het Nedersaksisch Instituut van de R. U. Groningen ter gelegenheid van zijn 30–jarig bestaan (essentially the same as in Nedersaksische Studies 7; zugleich: Driemaandelikse Bladen 36). Kuitert, R. 1965. Romantiek en realiteit [Romanticism and reality]. Taal en Tongval 17:1–14. Kurath, Hans. 1949. A word geography of the Eastern United States. Ann Arbor: University of Michigan Press. Kurath, Hans, and Raven I. McDavid Jr. 1961. The pronunciation of English in the Atlantic States. Ann Arbor: University of Michigan Press. Labov, William. 1965. Stages in the acquisition of standard English. In Social dialects and language learning. Champaign, IL: National Council of Teachers of English. ———. 1966. The social stratification of English in New York City. Arlington, VA: Center for Applied Linguistics. ———. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. Lafeber, A. P. M. 1967. Het dialect van Gouda [The dialect of the city of Gouda], met medewerking van [with the assistance of] L. B. Korstanje. Gouda: Oudheidkundige Kring “Die Goude.” Lambert, Wallace E. 1967. A social psychology of bilingualism. Journal of Social Issues 23:91–109. Lambert, Wallace E., Howard Giles, and O. Picard. 1975. Language attitudes in a FrenchAmerican community. International Journal of the Sociology of Language 4:127–52. Lambert, Wallace E., R. C. Hodgson, R. C. Gardner, and S. Fillenbaum. 1960. Evaluational reactions to spoken language. Journal of Abnormal Social Psychology 60:44–51.

384

REFERENCES

Lance, Donald M. 1994. Variation in American English. In American pronunciation, edited by Donald M. Lance and S. Kingsbury, 345–74. 12th ed. expanded, by John Kenyon. Ann Arbor, MI: George Wahr. Landheer, H. C. 1955. Het dialect van Overflakkee [The dialect of Overflakkee]. Assen: Van Gorcum & Comp. Leenen, J. 1954. Een Limburgse en Nederlandse uitspraakregel [A Limburg and Dutch pronunciation rule]. Taal en Tongval 6:1–24. Lewis, G. 1975. Attitudes to language among bilingual children and adults in Wales. International Journal of the Sociology of Language 4:103–21. Lippi-Green, Rosina. 1994. Accent, standard language ideology, and discriminatory pretext in the courts. Language in Society 23:163–98. Lodge, Anthony. 1993. French: From dialect to standard. New York: Routledge. Long, Daniel. 1990a. Hôgen ninchi chizu no kakikata to yomikata [The drawing and reading of perceptual dialect maps]. In Proceedings of the 50th meeting of the Dialectological Circle of Japan, 7–16. Tokyo: Dialectological Circle of Japan. ———. 1990b. Ôsaka to Kyôto de seikatsu suru tachihô shusshinsha no hôgen juyô no chigai [Differences in the dialect adaptation of students from other regions to the Kyoto and Osaka dialects]. Kokugogaku [Studies in the Japanese Language] 162:12–25. ———. 1993. The role of linguistic features in perceptual dialect regions. In Crochetière et al., 371–74. ———. 1997. The perception of “standard” as the speech variety of a specific region: Computer-produced composite maps of perceptual dialect regions. In Issues and methods in dialectology, edited by Alan Thomas, 256–70. Bangor: University of Wales, Department of Linguistics. ———. 1998. Nihon no Hôgen Ninchi Chizu Senkô Chizushû [A selected atlas of Japanese perceptual dialect maps]. Nihongo Kenkyû Sentâ Hôkoku [Japanese Language Research Center Reports] 5:45–84. Marckwardt, Albert. 1957. Principal and subsidiary dialect areas in the North-Central States. Publication of the American Dialect Society 27:3–15. Mase, Yoshio. 1964a. Hôgen ishiki to hôgen kukaku [Dialect Consciousness and Dialect Divisions: Examples in the Nagano-Gifu Boundary Region] In Nihon hôgen kenkykai, 270–302. ———. 1964b. Hôgen ishiki ni tsuite: Washa no genkyûshita hôgenteki tokuchô [On dialect consciousness: Dialect characteristics given by speakers]. Nagano-ken Tanki Daigaku Kiyô [Collected Papers of the Nagano Junior College] 18:1–12. ———. 1964c. Une nouvelle tentative pour tracer frontières subjectives des dialectes [A new proposal for the determination of subjective dialect boundaries]. Orbis 13:357–79 (translation from Japanese by Willem Grootaers of selected parts of Mase 1964a and b). ———. 1992. Gengo chirigaku kenkyû [Studies in linguistic geography]. Tokyo: Ôfusha. Mather, J. Y., and H. H. Speitel, eds. 1975. The linguistic atlas of Scotland: Scots section (Vol. 1). London: Croom Helm.

REFERENCES

385

———. 1977. The linguistic atlas of Scotland: Scots section (Vol. 2). London: Croom Helm. Mattheier, Klaus J. 1980. Pragmatik und Soziologie der Dialekte: Einführung in die kommunikative Dialektologie des Deutschen [Pragmatics and sociology of dialects: Introduction to the communicative dialectology of German]. Heidelberg: Unitaschenbücher. ———. 1985. Dialectologie van de dialectsprekers [The dialectology of dialect speakers]. In Dialect, standaardtaal en maatschappij, edited by J. Taeldeman and H. Dewulf, 23–45. Louvain-Amersfoort: Uitgeverij Acco. Maurer, F. 1930. Sprachschranken, Sprachräume und Sprachbewegungen im Hessischen [Linguistic barriers, language areas, and language movements in Hessian]. Hessische Blätter für Volkskunde 28:43–109. McNemar, Q. 1975. Psychological statistics (8th ed.). New York: Wiley. Meertens, P. J. 1952. Rumoer in Friesland [Uproar in Friesland]. Nieuw Vlaams Tijdschrift, 877–86. Opgenomen in [reprinted in] B. W. Schippers. (1968). Taal en spraak in stad en streek [Language and speech in city and region] 4:3–12. Mercer, G. 1975. The development of children’s ability to discriminate between languages and varieties of the same language. Unpublished master’s thesis, McGill University, Montreal. Milroy, James. 1992. Linguistic variation and change. Oxford: Blackwell. Milroy, Lesley. 1980. Language and social networks. Oxford: Blackwell. Milroy, Lesley, and Paul McClenaghan. 1977. Stereotyped reactions to four educated accents in Ulster. Belfast Working Papers in Language and Linguistics 2:1–11. Moulton, William G. 1960. The short vowel systems of Northern Switzerland. Word 16(2):155–82. ———. 1961a. Lautwandel durch innere Kausalität: die ostschweizerische Vokalspaltung [Sound change through inner causality: The eastern Swiss vowel split]. Zeitschrift für Mundartforschung 28(3):227–51. ———. 1961b. The dialect geography of “hast,” “hat” in Swiss German. Language 37(4):497–508. ———. 1962. Dialect geography and the concept of phonological space. Word 18(1–2):23–32. ———. 1963a. Phonologie und Dialekteinteilung [Phonology and dialect divisions]. In Sprachleben der Schweiz, 75–86. Festschrift Hotzenkôcherle. Bern: Francke Verllag. ———. 1963b. Review of Sprachatlas der Deutschen Schweiz. Journal of English and Germanic Philology 17(4):828–37. Naarding, J. 1947. Terreinverkenningen inzake de dialectgeografie van Drente, in 1948 uitgegeven onder de titel: De Drenten en hun taal [Reconnoitering of the terrain concerning the dialect geography of Drente, published in 1948 under the title “The inhabitants of Drente and their language”]. Assen: Van Gorcum. NHK. 1980a. Gendaijin no hanashikotoba: “Kotoba ni kansuru ishiki” Chôsa kara [Modern people’s spoken language: from the survey on language consciousness]. Bunken Geppô [NHK report on broadcast research] 30(2):1–15.

386

REFERENCES

NHK. 1980b. Nihonjin no gengo ishiki [Japanese people’s language consciousness]. NHK hôsô bunka kenkyû nenpô [Annual bulletin of NHK radio and television culture research] 25:173–250. Niedzielski, Nancy, and Dennis R. Preston. Forthcoming. Folk linguistics. Berlin: Mouton de Gruyter. Nihon hôgen kenkyûkai, ed. 1964. Nihon no Hôgen Kukaku [Dialect division in Japan]. Studies presented to Professor Misao Tôjô on his 80th birthday. Tokyo: Tokyodô. Nomoto, Kikuo. 1953. Mie ken Ueno-shi ni okeru Keigo Chôsa no Shûkei, Bunseki [Calculation and analysis of the Honorifics Survey of Ueno-city, Mie Prefecture]. Kokuritsu Kokugo Kenkyûjo Nenpô 6:88–110. ———. 1954. Aichi ken Okazaki-shi ni okeru keigo chôsa no Shûkei, Bukseki [Calculation and analysis of the Honorifics Survey of Okazaki-city, Aichi Prefecture]. Kokuritsu Kokugo Kenkyûjo Nenpô 6:66–77. ———. 1957. Rigen chôsa no hi-chosha sû [The number of informants in dialect lexical item surveys]. Keiryô Kokugogaku 3:36–41. ———. 1963. Kotoba no ishiki no kyôkai to jissai no kyôkai [Consciousness of linguistic boundaries and actual linguistic boundaries]. Jinruikagaku 15:271–81. Oki, Hiroko. 1986. Hôgen imêji no keisei [The construction of dialect images]. Kokubungaku 63:158–82. Okumura, Mitsuo. 1958. Hôgen no kukaku [Dialect regions]. Kokugo kokubun 27(3):20–35. Osgood, Charles E., G. J. Suci, and P. H. Tannenbaum. 1957. The measurement of meaning. Urbana: University of Illinois Press. Ôta, Eitarô. 1929. Toyama-shi kinzai hôgenshû [The collection of dialects in the Toyama area]. Tokyo: Kyôdo Kenky-sha. Pauwels, J. L. 1956. De Nederlandse Taalstudie in 1955 [Dutch language studies in 1955]. Handelingen van de Koninklijke Commissie voor Toponymie en Dialectologie 30:63–143. Pederson, Lee et al. 1992. Linguistic atlas of the Gulf States, Volume 7: Social pattern for the Linguistic Atlas of the Gulf States. Athens: University of Georgia Press. Peeters, F. J. 1952. Het taalkarakter in de spiegel [The character of language mirrored]. Driemaandelijkse Bladen Nieuwe Serie 4:33–38. Polle, Friedrich. 1889. Wie Denkt das Volk über die Sprache? Leipzig: B. G. Teubner. Pool, Jonathan. 1979. Language planning and identity planning. International Journal of the Sociology of Language 20:5–21. Pop, Sever. 1950. La dialectologie. Louvain: Universite catholique de Louvain. Preston, Dennis R. 1981. Perceptual dialectology: Mental maps of United States dialects from a Hawaiian perspective (summary). In Methods IV/Méthodes IV (Papers from the Fourth International Conference on Methods in Dialectology), edited by Henry Warkentyne, 192–98. British Columbia: University of Victoria. ———. 1985. Southern Indiana perceptions of “correct” and “pleasant” speech. In Methods/Méthodes V (Papers from the Fifth International Conference on Methods in

REFERENCES

387

Dialectology), edited by Henry Warkentyne, 387–411. British Columbia: University of Victoria. ———. 1986. Five visions of America. Language in Society 15(2):221–40. ———. 1988a. Methods in the study of dialect perception. In Methods in dialectology, edited by Alan Thomas, 373–95. Clevedon, Avon, and Philadelphia: Multilingual Matters. ———. 1988b. Sociolinguistic commonplaces in variety perception. In Linguistic change and contact: NWAV-XVI, edited by Kathleen Ferrara et al., 270–92. Austin: University of Texas, Department of Linguistics. ———. 1988c. Change in the perception of language varieties. In Historical dialectology: Regional and social, edited by Jacek Fisiak, 475–504. Berlin: Mouton de Gruyter. ———. 1988d. The nicest English is in Indiana. Studia Germanica Posnaniensia 14:169–93. ———. 1989. Perceptual dialectology. Dordrecht: Foris. ———. 1992. Talking black and talking white: A study in variety imitation. In Old English and new: Studies in language and linguistics in honor of Frederic G. Cassidy, edited by Joan Hall, N. Doane, and D. Ringler, 326–55. New York: Garland. ———. 1993a. Folk dialectology. In Preston, 1993d, 333–77. ———. 1993b. Folk dialect maps. In Language variation in North American English: Research and teaching, edited by Wayne Glowka and Donald M. Lance, 105–18. New York: Modern Language Association of America. ———. 1993c. Two heartland perceptions of language variety. (Revision of “Standard English spoken here” [1989]). In “Heartland” English, edited by T. Frazer, 23–47. Tuscaloosa: University of Alabama Press. ———, ed. 1993d. American dialect research. Amsterdam and Philadelphia: Benjamins. ———. 1993e. The uses of folk linguistics. International Journal of Applied Linguistics 3:181–59. ———. 1996a. Where the worst English is spoken. In Focus on the USA, edited by Edgar Schneider, 297–360. Amsterdam: Benjamins. ———. 1996b. Whaddyaknow?: The modes of folk linguistic awareness. Language Awareness 5(1):40–74. ———. 1997. The South: The touchstone. In Language variety in the South revisited, edited by Cynthia Bernstein, Tom Nunnally, and Robin Sabino, 311–52. Proceedings of the LAVIS II Conference, Auburn University, April 1–3, 1993. University: University of Alabama Press. ———. Forthcoming. A renewed proposal for the study of folk linguistics. In Language in action: New studies of language in society, edited by Peg Griffen, Joy Kreeft Peyton, Walt Wolfram, and Ralph Fasold (papers presented to Roger W. Shuy). Creeskill, NY: Hampton Press. Preston, Dennis R., and George M. Howe. 1987. Computerized studies of mental dialect maps. In Variation in language: NWAV-XV at Stanford, Stanford University, edited by Keith Denning et al., 361–78. Palo Alto, CA: Department of Linguistics, Stanford University.

388

REFERENCES

Price, S., M. Fluck, and Howard Giles. 1983. The effects of language of testing on bilingual pre-adolescents’ attitudes towards Welsh and varieties of English. Journal of Multilingual and Multicultural Development 4:149–61. Protze, Helmut. 1957. Das Westlausitzische und Ostmeissnische. Halle: M. Niemeyer. Rampton, Ben. 1995. Crossing: Language and ethnicity among adolescents. London: Longman. Rensink, W. G. 1955. Dialektindeling naar opgaven van medewerkers [Informant classification of dialects]. Amsterdam Dialectbureau Bulletin 7:20–23. (Mededelingen der Centrale commissie voor Onderzoek van het Nederlandse Volkseigen no. 7). Rey, Alberto. 1977. Accent and employability. Language Sciences 47:7–12. Rosenthal, M. 1974. The magic boxes: Pre-school children’s attitudes toward Black and Standard English. Florida FL Reporter 12:55–62. Ryan, Ellen Bouchard, and Howard Giles, eds. 1982. Attitudes towards language variation. London: Arnold. Ryan, Ellen Bouchard, Howard Giles, and R. J. Sebastian. 1982. An integrative perspective for the study of language attitudes toward language variation. In Ryan and Giles, 1–19. Schirmunski, V. M. 1962. Deutsche Mundartkunde: Vergleichende Laut- und Formenlehre der Deutsche Mundarten [German dialect studies: Comparative sound and structure studies of German dialects]. Berlin: Akademie Verlag (translated from Russian). Schneiderman, E. 1976. An examination of the ethnic and linguistic attitudes of bilingual children. ITL Review of Applied Linguistics 33:59–72. Schrijnen, J. 1902. Benrather-, Uerdinger- en PanningerLinie [Benrather, Uerdinger, and Panninger line]. Tijdschrift voor Nederlandse Taal- en Letterkunde 21:249–52. ———. 1907. Taalgrenzen in Zuidnederland — Het Mich-kwartier [Speech boundaries in South Holland — the mich quarter]. Tijdschrift voor Nederlandse Taal- en Letterkunde 26:81–85. Shibatani, Masayoshi. 1990. The languages of Japan. Cambridge: Cambridge University Press. Shuy, Roger W., and Ralph Fasold. 1973. Language attitudes: Current trends and prospects. Washington, DC: Georgetown University Press. Sibata, Takesi. 1957. Hôgen chôsa kiso-zu no kôsô [The organization of basic dialect research maps]. Paper presented at Kokugakuin Kokugo Kenkyûkai [Japanese Language Study Group], October 19. ———. 1958a. Hôgen no furui sô to atarashii sô. [The old and new layer of dialect]. Gengo seikatsu 8:28–37. ———. 1958b. Gengo henka no yôin to katei [Factors in language change and its processes]. In Kôza Gendai Kokugogaku [Lectures on modern Japanese linguistics], Vol. 3, 208–32. Tokyo: Chikuma shobô. ———. 1958c. Nihon no Hôgen [Dialect in Japan]. Tokyo: Iwanami. ———. 1959. Hôgen kyôkai no ishiki [Consciousness of dialect boundaries]. Gengo Kenkyû 36:1–30.

REFERENCES

389

———. 1960. Eastern Dialect and Western Dialect in Japan. Zeitschrift für Mundartforschung 27(2):97–103. ———. 1963. Kyôdô Kadai: Nihon no Chiikisei. Tango no Zenkoku bunpu [Common Issue: Localism in Japan, the distribution of words throughout Japan]. Jinruikagaku 15:149–76. Sibata, Takesi, and Willem Grootaers. 1988. Itoigawa Gengo Chizu [The Linguistic Atlas of Itoigawa], Vol. 1. Tokyo: Akiyama Shoten. ———. 1990. Itoigawa Gengo Chizu [The Linguistic Atlas of Itoigawa], Vol. 2. Tokyo: Akiyama Shoten. ———. 1995. Itoigawa Gengo Chizu [The Linguistic Atlas of Itoigawa], Vol. 3. Tokyo: Akiyama Shoten. Sibata, Takesi, and Kikuo Nomoto. 1955. Hôgen no shosô: 4. On’in taikei no bunpu [Aspects of dialects: 4. Distribution of phonological systems]. In Noto: Shizen, Bunka, Shakai [Noto: Nature, Culture and Society], edited by Kyûgakkai Rengô [Federation of Nine Academic Associations], 175–81. Tokyo: Heibonsha. Taalatlas van Noord- en Zuid-Nederland [The linguistic atlas of the northern and southern Netherlands]. 1939–1987. Leiden: E. J. Brill (from 1981 on, Taalatlas van het Nederlands en het Fries [The dialect atlas of Netherlandic and Frisian]). Tajfel, H. 1974. Social identity and intergroup behaviour. Social Science Information 13:65–93. Terwey, A. T. 1885. Nederlandsche spraakkunst [Dutch grammar] (6th printing). Groningen: J. B. Wolters. Tôjô, Misao. 1927. Dainippon hôgen chizukokugo no hôgen kukaku [The great Japanese map of dialect divisions]. Tokyo: Ikvei Shoin. ———. 1933. Hôgengaku gaisetsu [Outline of Japanese dialectology]. (Kokugo Kagaku Kôza, vol.1). Tokyo: Meiji Shoin. ———. 1938. Hôgen to Hôgengaku [Dialects and dialectology]. Tokyo: Shunyôdô. ———. 1950. Hôgen shûkenron to hôgen kukaku ron [Wave theory and dialect regions]. Kokugogaku 4:1–12. ———. 1951. Zenkoku hôgen jiten [A dictionary of general Japanese dialects]. Tokyo: Tokyodô. ———, ed. 1954. Nihon hôgengaku [Japanese dialectology]. Tokyo: Yoshikawa Kôbundô. Tokugawa, Munemasa. 1958a. Itoigawa chôsa no hôkoku B [Report “B” on Itoigawa research]. A paper presented at the dialect sub-group of Nihon Kotoba no Kai, November 1. ———. 1958b. Nihon Gengo Chizu Sakusei no tame no Chôsa [A survey for the Linguistic Atlas of Japan]. Kokuritsu Kokugo Kenky-jo Nenpô 9:38–61. ———. 1959a. “Kamakiri” no hôgen bunpu wo kaishaku suru [Dialect distribution and interpretation of “praying mantis”]. Kokuritsu kokugo kenkyûjo ronshû 1, Kotoba no kenkyû 287–302. ———. 1959b. Hôgen chôsa no jissai [The current state of dialect research]. Paper presented at Gakushûin University kokugo kokubun gakkai, June 7.

390

REFERENCES

———. 1959c. Hiki-gaeru hôgen no rekishi(jô) [On the history of dialect forms for “toad”]. Gakushûin Daigaku Kokugo Kokubun Gakkai-shi 4:11–21. ———. 1977. Hôgen kenkyû no rekishi [A history of dialect research]. In Iwanami kôza Nihongo 11, Hôgen, edited by Susumu Ôno and Takesi Sibata, 327–78. Tokyo: Iwanami. ———. 1981. Nihongo no sekai 8, Kotoba Nishi to Higashi [The world of Japanese language, vol. 8, Words east and west]. Tokyo: Chûô kôronsha. Tomosada, Kenji. 1995. Kansai-ben ha hôgen wo hen’yô saseru ka? [Does the Kansai dialect change other dialects?] Gengo 24(11):122–135. Tourtoulon, Ch. de. 1890. Le classification des dialectes. Revue des langues romanes 34:130–75. Trudgill, Peter. 1972. Sex, covert prestige and language change in the urban British English of Norwich. Language in Society 1:179–95. ———. 1974. The social differentiation of English in Norwich. Cambridge: Cambridge University Press. ———. 1975. Accent, dialect and the school. London: Arnold. ———. 1983. On dialect: Social and geographical perspectives. Oxford: Blackwell. ———. 1990. The dialects of England. Cambridge, MA: Blackwell. Tucker, G. Richard, and Wallace E. Lambert. 1969. White and Negro listeners’ reactions to various American-English dialects. Social Forces 47:463–68. 25 jaar Dialectbureau [Department of Dialectology 25 years]. 1955. Mededelingen der Centrale commissie voor onderzoek van het Nederlandse volkseigen VII. Umegaki, Minoru. 1964. Hôgen kukakuron shôshi [A short history of dialect division theory]. In Nihon Hôgen Kenkyûkai, 23–45. Veen, T. van. 1964. Utrecht tussen oost en west [Utrecht between east and west]. Assen: Van Gorcum & Comp. Viereck, Wolfgang. 1985. Linguistic atlases and dialectometry: The Survey of English Dialects. In Kirk et al., 94–112. ———. 1986. Dialectal speech areas in England: Orton’s phonetic and grammatical evidence. Journal of English Linguistics 19:240–57. ———. 1992. The Computer developed linguistic atlas of England (CLAE): A preview of Volume 2. Nihon Hôgen Kenkyû-kai Genkô-shû 55:55–71. ———, ed. 1992–. Computer developed linguistic atlas of English. Tübingen: Max Niemeyer Verlag. ———, ed. 1995. Verhandlungen des Internationalen Dialektologenkongresses, Bamberg, 29.7.-4.8.1990 [Proceedings of the International Congress of Dialectologists, Bamberg, July 29–August 4, 1990]. Zeitschrift für Dialektologie und Linguistik 75–77. 4 vols. Stuttgart: Franz Steiner Verlag. Voegelin, Carl F., and Zellig S. Harris. 1951. Methods for determining intelligibility among dialects of natural languages. Proceedings of the American Philosophical Society 95:322–29.

REFERENCES

391

Vragenlijsten met register [Questionnaires with index] (1931–1958). 1960. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 22. Weber, Eugen. 1976. Peasants into Frenchmen: The modernization of rural France. Stanford, CA: Stanford University Press. Wegener, P. 1880. über deutsche Dialectforschung Dialectforschung [On German dialect research]. Zeitschrift für Deutsche Philologie 11:450–80. Weijnen, Antonius A. 1946. De grenzen tussen de Oost-Noordbrabantse dialecten onderling [The borders between the dialects of eastern North Brabant]. In OostNoordbrabantse dialectproblemen [Eastern North Brabant dialect problems], edited by Antonius A. Weijnen, J. M. Renders, and Jac. van Ginneken, 1–15. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 8. ———. 1947. De onderscheiding van dialectgroepen in Noord-Brabant en Limburg [Differentiating dialect clusters in North Brabant and Limburg]. Akadamietdagen I, Voordrachten gehouden te ’s-Hertogenbosch op 11 en 12 April 1947, Amsterdam, 69–99. ———. 1961. Het bewustzijn van dialectverschil [The awareness of dialect difference] In Voordrachten Gehouden voor de Gelderse Leergangen te Arnhem. Vol. 5. Groningen: J. B. Wolters. ———. 1966. Nederlandse dialectkunde [Dutch dialectology]. Assen: Van Gorcum. ———. 1968. Zum Wert subjektiver Dialektgrenzen. Lingua 21:594–96. ———. 1989. De invloed van P. Willems op de Nederlandse dialectologie [the influence of P. willems on Dutch dialectology]. Taal en Tongval (Themanr. 2: Honderd Jaar Enquête Willems), 16–28. Weinreich, Uriel, William Labov, and Marvin Herzog. 1968. Empirical foundations for a theory of language change. In Directions for historical linguistics, edited by Winfred P. Lehmann and Yakov Malkiel, 95–195. Austin: University of Texas Press. Wells, John C. 1982. Accents of English. Cambridge: Cambridge University Press. Wijk, N. van. 1911. Een Oudwestnederfrankies æ-dialekt [An old western Low Franconian æ-dialect]. Tijdschrift voor Nederlandse Taal- en Letterkunde 30:161–89. Willems, P. 1886. De enquête werd gehouden in 1886, de antwoorden zijn het eigendom van de Koninklijke Vlaamsche Academie voor Taal- en Letterkunde, en worden daar bewaard [The inquiry was done in 1886; the responses are the property of the Royal Flemish Academy of Languages and Literature in Ghent where they are preserved]. Microcopies are at the Institute of Dialectology and Phonetics in Leuven, the Catholic University Nijmegen, and the P. J. Meertens Institute Amsterdam. Williams, Angie, Peter Garrett, and Nikolas Coupland. 1996. Perceptual dialectology, folklinguistics and regional stereotypes: Teachers’ perceptions of variation in Welsh English. Multilingua 15:171–99. Williams, C. 1985. On culture space: Perceptual culture regions in Wales. Études Celtiques 22:273–95. ———. 1990. The anglicisation of Wales. In Coupland (ed.).

392

REFERENCES

Williams, Frederick, Nancy Hewett, Robert Hopper, Leslie M. Miller, Rita C. Naremore, and Jack L. Whitehead. 1976. Explorations of the linguistic attitudes of teachers. Rowley, MA: Newbury House. Windsor Lewis, J. 1990. The roots of Cardiff English. In Coupland (ed.), 105–08. Winkel, J. te. 1895. De taalkaart voor Noord-Nederland [The dialect map for the northern Netherlands] Tijdschrift Koninklijk Nederlandsch Aardrijkskundig Genootschap 2e serie, dl. XII. ———. 1898. Geschichte der niederländischen Sprache [History of the Dutch language]. Grundriss der germanischen Philologie V 2; ook afzonderlijk verschenen [also a separate edition]. Strassburg: Trübner. ———. 1899–1901. De Noordnederlandsche tongvallen [The northern Dutch dialects]. Atlas van taalkaarten met tekst, uitg. d.h. Kon. Ned. Aardr. Genootschap, 2 volumes, met twee kaarten. Leiden: E. J. Brill. ———. 1901. Geschiedenis der Nederlandsche taal. Naar de tweede hoogduitsche uitgave, vertaald door F. C. Wieder [History of the Dutch language, based on the 2nd German edition, translated by F. C. Wieder, a Dutch translation of Te Winkel 1898]. Culemborg: Blom & Olivierse. Winkler, Johan. 1874. Algemeen Nederduitsch en Friesch dialecticon [General Low Dutch and Frisian dialect anthology], 2 dln [2 parts]. Wood, G. 1971. Vocabulary change: A study of variation in regional words in eight of the Southern states. Carbondale: Southern Illinois University Press. Zahn, C., and R. Hopper. 1985. Measuring language attitudes: The speech evaluation instrument. Journal of Language and Social Psychology 4:113–23. Zinsli, Paul. 1957. Berndeutsche Mundart [Bern German dialect]. Bern: Berner Staatsbuch.

Additional Readings

The following is a comprehensive list (unpublished lectures generally excluded) of readings in the area of perceptual dialectology (and may, therefore, include items not referred to in any of the chapters). Readings that are principally concerned with perceptual dialectology are listed first, followed by ones at least partially concerned with the topic. A number of “perceptual” works have been excluded on the basis that their principal focus is on the study of “language attitudes” (or the “social psychology of language”), although, admittedly, as suggested in the introduction, this boundary is often difficult to draw. The editor welcomes any additions and/or corrections to this list.

Principally Concerned Büld, H. 1939. Sprache und Volkstum im nördlichen Westfalen: Sprachgrenzen und Sprachbewegungen in der Volksmeinung [Language and nationality in northern Westphalia: Perceptions of linguistic boundaries and language movements]. Emsdetten: H. & G. Lechte. Butters, Ronald R. 1991. Review of Preston 1989. Language in Society 20(2):294–99. Carver, Craig M. 1991. Review of Preston 1989. American Speech 66(4):432–37. Coupland, Nikolas, Angie Williams, and Peter Garrett. 1994. The social meaning of Welsh English: Teachers’ stereotyped judgements. Journal of Multilingual and Multicultural Development 15(6):471–89. Coupland, Nikolas, Angie Williams, and Peter Garrett. 1995. “Welshness” and “Englishness” as attitudinal dimensions of English language variation in Wales. Paper presented to the International Conference on Language Contact, Valencia, Spain (see Chapter 20, this volume). Daan, Jo. 1970. Dialekten [Dialects]. In Van randstad tot landrand [From “city-side” to “country-border”], edited by Jo Daan and D. P. Blok, 7–43. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 37. Dailey-O’Cain, Jennifer. 1994. The wall in the mind: Language attitudes in Eastern and Western Germany. Stanford, CA: NWAV 23, Stanford University, October.

394

ADDITIONAL READINGS

———. 1996. “That terrible Saxon dialect”: Standard language ideology in post-unification Germany. In Texas Linguistic Forum, edited by R. Ide, R. Parker, and Y. Sunaoshi, 36. Austin: Department of Linguistics, University of Texas (essentially the same as Dailey-O’Cain 1994). Diercks, Willy. 1988. Mental maps: Linguistisch-geographische Konzepte. Zeitschrift für Dialektologie und Linguistik 55:280–305. Frazer, Timothy. 1990. Review of Preston 1989. Language 66(3):650–51. Garrett, Peter, Nikolas Coupland, and Angie Williams. 1995. “City Harsh” and “The Welsh Version of RP”: Some ways in which teachers view dialects of Welsh English. Language Awareness 4:99–107. Garrett, Peter, Angie Williams, and Nicolas Coupland. 1996. Adolescents’ judgements of Welsh English dialects. Paper presented at Sociolinguistics Symposium 11, University of Wales Cardiff, September. Goeman, A. C. M. 1987. Perceptie van de afstand dialect-standaardtaal in relatie tot nieuwe gegevens betreffende de Nederlandse dialecten [Perception of the distance between dialect and standard language in relation to new data on Dutch dialects]. In Eigen en Vreemd: Identiteit en ontlening in taal, literatuur en beeldendekunst [Identity and loans in language, literature and arts], Handelingen v. h. 39ste Nederlands Filologencongres 1986, Amsterdam, 411–22. ———. 1989. Dialectes et jugements subjectifs des locuteurs. Quelques remarques de méthode a propos d’une controverse [Dialects and the subjective judgments of speakers: Remarks on controversial methods]. Espaces Romans (études de dialectologie et de géolinguistique offertes à Gaston Tuaillon.) Vol. II. Université Stendhal — Grenoble 3: Ellug, 532–44. Grootaers, Willem A. 1959. Hôgen kyôkaisen wo megutte [In search of dialect boundaries]. Lecture, Tokyo Metropolitan University. ———. 1959. Origin and nature of the subjective boundaries of dialects. Orbis 8(2):355–84. ———. 1964. La discussion autor des frontières dialectales subjectives. Orbis 13:380–98. Hartley, Laura C. 1996. Oregonian perceptions of American regional speech. Master’s thesis. Department of Linguistics and Languages, Michigan State University. Hartley, Laura C., and Dennis R. Preston. 1999. The names of U. S. English: Valley girl, cowboy, Yankee, normal, nasal and ignorant. In Standard English, edited by Richard Watts and Anthony Bex, 207–238. London: Routledge. Hout, R. van, and H. Münstermann. 1981. Linguistische afstand, dialekt en attitude [Linguistic distance, dialect and attitude]. Gramma, Nijmeegs tijdschrift voor taalkunde 5:101–123. Inoue, Fumio. 1972. Chiiki meisho to hôgensha ishiki [Names of districts and consciousness of dialect differences]. Chimeigaku Kenkyu 1:18–30. ———. 1977/8. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of dialect image] (part 1). Gengo Seikatsu 311:82–91. ———. 1978/9. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of dialect image] (part 2). Gengo Seikatsu 312:82–88.

ADDITIONAL READINGS

395

———. 1988. Dialect image and new dialect forms. Area and Culture Studies 38:13–23. ———. 1995. Classification of dialects by image — English and Japanese. In Verhandlungen des Internationalen Dialektologenkongresses, Bamberg, 29.7.-4.8.1990 [Proceedings of the International Congress of Dialectologists, Bamberg, July 29–August 4, 1990], edited by Wolfgang Viereck, 355–68. Zeitschrift für Dialektologie und Linguistik 75–77. 4 vols. Stuttgart: Franz Steiner Verlag (see Chapter 11, this volume). ———. 1996. Subjective dialect division in Great Britain. American Speech 71(2):142–61 (see Chapter 12, this volume). Jernudd, Björn H. 1967. A listener evaluation experiment of variants of Australian English (Unpublished report). Jernudd, Björn H. 1968. There are no subjective dialects. Kivung 1:38–42. Kontra, Miklos. 1997. Hol beszélnek legszebben és legcsúnyábban magyarul? [Where are the most beautiful and ugliest Hungarian spoken?]. Magyar Nyelv 93:224–32. Kremer, Ludger. 1984. Die niederländisch-deutsche Staatgrenze als subjektive Dialektgrenze. In Grenzen en grensproblemen, 76–83. Een bundel studies uitgegeven door het Nedersaksisch Instituut van de R. U. Groningen ter gelegenheid van zijn 30–jarig bestaan. Nedersaksische Studies 7; zugleich: Driemaandelikse Bladen 36 (see Chapter 3, this volume). Lance, Donald M. 1992. This is what we think: Folk perceptions of dialect regions. A paper presented to Section 299 (Session I, American Dialect Society) of the Modern Language Association, New York, December 28. Long, Daniel. 1990. Hôgen ninchi chizu no kakikata to yomikata [The drawing and reading of perceptual dialect maps]. Proceedings of the 50th meeting of the Dialectological Circle of Japan, 7–16. Tokyo: Dialectological Circle of Japan (included and revised in Chapter 13, this volume). ———. 1993. The role of linguistic features in perceptual dialect regions. In Les langues menacées: actes du XVe Congrès international des linguistes, Québec, Université Laval août 1992 [Endangered languages: Proceedings of the XVth International Congress of Linguists, Quebec, Laval University, August 9–14, 1992], edited by André Crochetière, Jean-Claude Boulanger, and Conrad Ouellion, 371–74. Vol. 3. Sainte-Foy, Québec: Les Presses de L’Université Laval (included and revised in Chapter 13, this volume). ———. 1995. Hôgen ninchi chizu [Perceptual dialect maps]. In Pasokon kokugo kokubungaku [Japanese linguistics and language on the personal computer], edited by D. B. West, 157–71. Tokyo: Keibunsha. ———. 1997. Nihon no Hôgen Ninchi Chizu Senkô Chizushû [A selected atlas of Japanese perceptual dialect maps]. Nihongo Kenkyû Sentâ Hôkoku 5:45–84. ———. 1997. The perception of “standard” as the speech variety of a specific region: Computer-produced composite maps of perceptual dialect regions. In Current methods in dialectology, edited by Alan Thomas, 256–70. Bangor: University of Wales, Department of Linguistics.

396

ADDITIONAL READINGS

Mase, Yoshio. 1964. Hôgen ishiki to hôgen kukaku [Dialect consciousness and dialect divisions]. In Nihon no Hôgen Kukaku [Dialect division in Japan], edited by Nihon hôgen kenkyûkai [Dialectological Circle of Japan], 270–302. Studies presented to Professor Misao Tôjô on his 80th birthday. Tokyo: Tokyodô (see Chapter 6, this volume). ———. 1964. Hôgen ishiki ni tsuite: Washa no genky-shita hôgenteki tokuchô [On dialect consciousness: Dialect characteristics given by speakers]. Nagano-ken Tanki Daigaku Kiyô 18:1–12 (see Chapter 7, this volume). ———. 1964. Une nouvelle tentative pour tracer frontières subjectives des dialectes. Orbis 13:357–79 (translation from Japanese by Willem Grootaers of selected parts of Mase 1964a and b). Mattheier, Klaus J. 1980. Pragmatik und Soziologie der Dialekte: Einführung in die kommunikative Dialektologie des Deutschen [Pragmatics and sociology of dialects: Introduction to the communicative dialectology of German]. Heidelberg: Unitaschenbücher. ———. 1985. Dialectologie van de dialectsprekers [The dialectology of dialect speakers]. In Dialect, standaardtaal en maatschappij, edited by J. Taeldeman and H. Dewulf, 23–45. Louvain-Amersfoort: Uitgeverij Acco. ———. 1987. Regional dialects and folk linguistics. Abstracts: AILA (8th World Congress of Applied Linguistics). Sydney, August 16–21, 12. Nomoto, Kikuo. 1963. Kotoba no ishiki no kyôkai to jissai no kyôkai [Consciousness of linguistic boundaries and actual linguistic boundaries]. Jinruikagaku 15:271–81 (see Chapter 5, this volume). Oki, Hiroko. 1986. Hôgen imêji no keisei [The construction of dialect images]. Kokubungaku (Suita: Kansai University) 63:158–82. Preston, Dennis R. 1981. Perceptual dialectology: Mental maps of United States dialects from a Hawaiian perspective (summary). In Methods IV/Méthodes IV (Papers from the Fourth International Conference on Methods in Dialectology), edited by Henry Warkentyne, 192–98. British Columbia: University of Victoria (summary of Preston 1982). ———. 1982. Perceptual dialectology: Mental maps of United States dialects from a Hawaiian perspective. Working Papers in Linguistics 14(2):5–49. ———. 1985. Mental maps of language distribution in Rio Grande do Sul (Brazil). Geographical Bulletin 27:46–64. ———. 1985. Southern Indiana perceptions of “correct” and “pleasant” speech. In Methods/Méthodes V (Papers from the Fifth International Conference on Methods in Dialectology), edited by Henry Warkentyne, 387–411. British Columbia: University of Victoria. ———. 1986. Five visions of America. Language in Society 15(2):221–40. ———. 1988. Methods in the study of dialect perception. In Methods in dialectology, edited by Alan Thomas, 373–95. Clevedon, Avon, and Philadelphia: Multilingual Matters.

ADDITIONAL READINGS

397

———. 1988. Sociolinguistic commonplaces in variety perception. In Linguistic change and contact: NWAV-XVI, edited by Kathleen Ferrara et al., 270–92. Austin: University of Texas, Department of Linguistics. ———. 1988. Change in the perception of language varieties. In Historical dialectology: Regional and social, edited by Jacek Fisiak, 475–504. Berlin: Mouton de Gruyter. ———. 1988. The nicest English is in Indiana. Studia Germanica Posnaniensia 14:169–93. ———. 1988. New trends in perceptual dialectology. Hong Kong Conference on Language and Society, Hong Kong, April (Abstracts, 39). ———. 1989. Perceptual dialectology. Dordrecht: Foris. (Revs: T. Frazer, Language 66(3):650–51; Craig Carver, American Speech 66(4):432–37; Ronald Butters, Language in Society 20:294–99 (minimally revised versions of Preston 1981, 1982, 1985a, 1985b, 1986, 1988a, 1988d; see Reference list). ———. 1989. Standard English spoken here. In Status and function of languages and language varieties, edited by Ulrich Ammon, 324–54. Berlin: Walter de Gruyter. ———. 1993. Folk dialectology. In American dialect research, edited by Dennis R. Preston, 333–77. Amsterdam and Philadelphia: Benjamins. ———. 1993. Folk dialect maps. In Language variation in North American English: Research and teaching, edited by Wayne Glowka and Donald M. Lance, 105–18. New York: Modern Language Association of America. ———. 1993. Two heartland perceptions of language variety (Revision of “Standard English spoken here” [1989]). In “Heartland” English, edited by T. Frazer, 23–47. Tuscaloosa: University of Alabama Press. ———. 1996. Where the worst English is spoken. In Focus on the USA, edited by Edgar Schneider, 297–360. Amsterdam: Benjamins. ———. 1997. The South: The touchstone. In Language variety in the South revisited, edited by Cynthia Bernstein, Tom Nunnally, and Robin Sabino, 311–52. Proceedings of the LAVIS II Conference, Auburn University, April 1–3, 1993. University: University of Alabama Press. Preston, Dennis R., and George M. Howe. 1987. Computerized studies of mental dialect maps. In Variation in language: NWAV-XV at Stanford, Stanford University, edited by Keith Denning et al., 361–78. Palo Alto, CA: Department of Linguistics, Stanford University. Rensink, W. G. 1955. Dialektindeling naar opgaven van medewerkers [Dialect boundaries according to respondents]. Amsterdam Dialectbureau Bulletin 7:20–23. (Mededelingen der Centrale commissie voor Onderzoek van het Nederlandse Volkseigen no. 7) (see Chapter 1, this volume). Sibata, Takesi. 1959. Hôgen kyôkai no ishiki [Subjective consciousness of dialect boundaries]. Gengo Kenkyû 36:1–30 (see Chapter 4, this volume). Tôjô, Misao. 1959. Watashi no tsukatta “Hôgen ishiki” to iu kotoba ni tsuite [On the expression “dialect consciousness” as I used it]. Gengo Kenkyû 36:31–32 (translated in Chapter 8, this volume).

398

ADDITIONAL READINGS

Weijnen, Antonius A. 1946. De grenzen tussen de Oost-Noordbrabantse dialecten onderling [The borders between the dialects of eastern North Brabant]. In OostNoordbrabantse dialectproblemen [Eastern North Brabant dialect problems], edited by Antonius A. Weijnen, J. M. Renders, and Jac. van Ginneken, 1–15. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 8. ———. 1947. De onderscheiding van dialectgroepen in Noord-Brabant en Limburg [Differentiating dialect clusters in North Brabant and Limburg]. Akadamietdagen I, Voordrachten gehouden te ’s-Hertogenbosch op 11 en 12 April 1947, Amsterdam, 69–99. ———. 1961. Het bewustzijn van dialectverschil [The awareness of dialect difference] In Voordrachten Gehouden voor de Gelderse Leergangen te Arnhem. Vol. 5. Groningen: J. B. Wolters. ———. 1968. Zum Wert subjektiver Dialektgrenzen. Lingua 21:594–96. Williams, Angie, Peter Garrett, and Nikolas Coupland. 1996. Perceptual dialectology, folklinguistics and regional stereotypes: Teachers’ perceptions of variation in Welsh English. Multilingua 15:171–99.

Partially Concerned Butters, Ronald R. 1995. Some current issues in variation theory and practice. In Verhandlungen des Internationalen Dialektologenkongresses, Bamberg, 29.7.-4.8.1990 [Proceedings of the International Congress of Dialectologists, Bamberg, July 29–August 4, 1990], edited by Wolfgang Viereck, 3–36. Zeitschrift für Dialektologie und Linguistik 75–77. 4 vols. Stuttgart: Franz Steiner Verlag. Entjes, H. 1974. Dialecten in nederland [Dialects in the Netherlands]. Haren-GN: Knoop and Niemeijer. Gooskens, Charlotte. 1997. On the role of prosodic and verbal information in the perception of Dutch and English language varieties. Ph.D. diss., Catholic University of Nijmegen (privately printed). Graff, David, William Labov, and Wendell A. Harris. 1986. Testing listeners’ reactions to phonological markers of ethnic identity: A new method for sociolinguistic research. In Diversity and diachrony, edited by David Sankoff, 45–58. Current Issues in Linguistic Theory, Vol. 53. Amsterdam and Philadelphia: Benjamins. Grootaers, Willem A. 1958. Nihon hôgen kenkyû ni okeru futatsu no atarashii hôhô [Two new approaches in Japanese dialectology]. International Congress of Orientalists in Japan 3:71–74. ———. 1958. Methods to study Japanese Dialects. Paper presented at the Asiatic Society Meeting, Sophia University, November 10. ———. 1963. Les premiers pas à la recherche des unités dialectales. Orbis 12:361–80. ———. 1976. Nihon no hôgen chirigaku no tameni [Contributions to Japanese dialect geography]. Tokyo: Heibonsha.

ADDITIONAL READINGS

399

Hammarström, Göran. 1961. Inquéritos Linguisticos [Linguistic Inquiries]. Revista de Portugal Series A, 26:9–32. (In Swedish: 1964. Dialektalogiska undersökningar. Nordisk tidsskrift for tale och stemme 24). ———. 1966. Linguistische Einheiten im Rahmen der modernen Sprachwissenschaft [Linguistic units on the basis of modern linguistics] (Kommunikation und Kybernetik in Einzeldarstellungen Band 5). Berlin, Heidelberg, New York: Springer-Verlag. ———. 1967. Zur sozialen und dialektalen Funktion der Sprache Sprache [On the social and dialectal functions of language]. Zeitschrift für Mundartforschung 34(3/4):205–15. Inoue, Fumio. 1986. Sociopsychological characteristics of users of “new dialect forms.” Journal of Pragmatics 10:327–45. ———. 1986. Sociolinguistic aspects of new dialect forms: language change in progress in Tokyo. International Journal of the Sociology of Language 58:73–89. ———. 1989. Kotoba-zukai Shin-fukei [New landscape of spoken Japanese]. Tokyo: Akiyama Shoten. ———. 1991. Dialect image and the diffusion of new dialect forms. In Proceedings of the Fourteenth International Congress of Linguists, Berlin/GDR, August 10–15, 1987, edited by Werner Bahner, Joachim Schildt, and Dieter Viehweger, 1508–11. Berlin: Akademie-Verlag. ———. 1992. Tôkaidô ensen no hôgen shiyô to hôgen ishiki [Dialect usage and dialect image along the Tokaido Line]. Area and Culture Studies 45:11–50. Labov, William. 1966. The social stratification of English in New York City. Arlington, VA: Center for Applied Linguistics. ———. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. Léonard, Jean Léonard. 1987. Demarcation linguistique, conscience de la variation dialectale et dialectologie du locuteur: Approche de la conscience linguistique dans l’isle de Noirmoutier (Vendée) [Linguistic demarcation, the awareness of dialect variation and dialectology by the speaker: An approach to linguistic awareness on the island of Noirmoutier (Vendée)]. Mémoire de D. E. A. Faculté des Lettres d’Aix-enProvence. ———. 1990. Variation dialectale et microcosme anthropologique: l’ile de Noirmoutier (Vendée, France) [Dialect variation and an anthropological microcosm: The island of Noirmoutier (Vendée, France)]. Doctoral dissertation, Sciences du Langage (ILGEOS), Université de Provence — Aix-Marseille 1. ———. 1991. Distances et dialecte: Approche des représentations ethnolinguistiques … Noirmoutier (Vendée) [Distances and dialect: An approach to the ethnolinguistic representation of Noirmoutiers (Vendée)]. In Les Français et leurs langues (Colloque tenu a Montpellier les 5, 6 & 7 Septembre 1988, sous la direction de Jean-Claude Bouvier, Actes rassamblés par Claude Martel). Publications de l’Université de Provence, Aix-Marseille 1:59–83. Mase, Yoshio. 1992. Gengo chirigaku kenkyû [Studies in linguistic geography]. Tokyo: Ôfusha. Nakamata, Hitoshi. 1997. Hôgen kukakuron to gengo chiiki kubun [Dialect regions and actual linguistic regions]. Jinbun Chiri 49(1):20–31.

400

ADDITIONAL READINGS

Niedzielski, Nancy and Dennis R. Preston. Forthcoming. Folk linguistics. Berlin: Mouton de Gruyter. Nihon hôgen kenkyûkai [Dialectological Circle of Japan], ed. 1964. Nihon no Hôgen Kukaku [Dialect division in Japan]. Studies presented to Professor Misao Tôjô on his 80th birthday. Tokyo: Tokyodô. Peeters, F. J. 1952. Het taalkarakter in de spiegel [The character of language mirrored]. Driemaandelijkse Bladen Nieuwe Serie 4:33–38. Preston, Dennis, R. 1992. Talking black and talking white: A study in variety imitation. In Old English and new: Studies in language and linguistics in honor of Frederic G. Cassidy, edited by Joan Hall, N. Doane, and D. Ringler, 326–55. New York: Garland. ———, ed. 1993. American dialect research. Amsterdam and Philadelphia: Benjamins. ———. 1993. The uses of folk linguistics. International Journal of Applied Linguistics 3:181–59. ———. 1996. Whaddyaknow?: The modes of folk linguistic awareness. Language Awareness 5(1):40–74. ———. Forthcoming. A renewed proposal for the study of folk linguistics. In Language in action: New studies of language in society, edited by Peg Griffen, Joy Kreeft Peyton, Walt Wolfram, and Ralph Fasold (papers presented to Roger W. Shuy). Creeskill, NY: Hampton Press. Romanello, Maria Teresa. 1996. Sulla rappresentazione dei confini linguistici [On the representation of linguistic areas]. Rivista Italiana Dialettologia 20:7–33. Schirmunski, V. M. 1962. Deutsche Mundartkunde: Vergleichende Laut- und Formenlehre der Deutsche Mundarten [German dialect studies: Comparative sound and structure studies of German dialects]. Berlin: Akademie Verlag (translated from Russian). Sibata, Takesi. 1958. Nihon no Hôgen [Dialect in Japan]. Tokyo: Iwanami. ———, ed. 1984. Amami Ooshima no kotoba: bunpu kara rekishi e [The language of Amami Ooshima: From distribution to history]. Tokyo: Akiyama Shoten. Tachibana, Shôichi. 1936. Hôgengaku gairon [General dialectology]. Tokyo: Ikuei Shoin. Tôjô, Misao, ed. 1954. Nihon hôgengaku [Japanese dialectology]. Tokyo: Yoshikawa Kôbundô. ———. 1957. Hôgengaku no hanashi [Topics in dialectology]. Tokyo: Meiji Shoin. Tourtoulon, Ch. de. 1890. Le classification des dialectes. Revue des langues romanes 34:130–175. 25 jaar Dialectbureau [Department of Dialectology 25 years]. 1955. Mededelingen der Centrale commissie voor onderzoek van het Nederlandse volkseigen VII. Central Commission. Umegaki, Minoru 1964. Hôgen kukakuron shôshi [A short history of dialect division theory]. In Nihon no Hôgen Kukaku [Dialect division in Japan], edited by Nihon hôgen kenkyûkai [Dialectological Circle of Japan], 23–45. Studies presented to Professor Misao Tôjô on his 80th birthday. Tokyo: Tokyodô. Vragenlijsten met register [Questionnaires with index] (1931–1958). 1960. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam 22.

ADDITIONAL READINGS

401

Weijnen, Antonius A. 1952. Dialect-atlas van Noord-Brabant [Dialect atlas of North Brabant]. Antwerp: De Sikkel. ———. 1966. Nederlandse dialectkunde [Dutch dialectology]. Assen: Van Gorcum. Willems, P. 1886. De enquête werd gehouden in 1886, de antwoorden zijn het eigendom van de Koninklijke Vlaamsche Academie voor Taal- en Letterkunde, en worden daar bewaard [The inquiry was done in 1886; the responses are the property of the Royal Flemish Academy of Languages and Literature in Ghent where they are preserved]. Microcopies are at the Institute of Dialectology and Phonetics in Leuven, the Catholic University Nijmegen, and the P. J. Meertens Institute Amsterdam. Zinsli, Paul. 1957. Berndeutsche Mundart. Bern: Berner Staatsbuch.

About the Contributors and Translators

K B, an Afrikaner, is Assistant Professor at the University of San Francisco. Her research focuses on the works of contemporary French women poets (most recently Anne-Marie Albiach and Marie Etienne). Feminocentric in her choice of authors, her approach is predominantly feminist. Although her papers concentrate on individuals, she continues to question the frequent omission of women poets in critical discourse. She has also made forays into the Francophone world, reading the Belgian poet-philosopher Claire Lejeune and the Franco-Ontarian poet Andrée Christensen. In addition to working on Marie-Claire Bancquart, she is currently exploring the highly poetic novels of Cameroonian Werewere Liking. N C is Professor and Founding Director of the Centre for Language and Communication Research at the University of Wales, Cardiff. With Allan Bell, he is founding editor of the Journal of Sociolinguistics. He is author of Dialect in Use: Sociolinguistic Variation in Cardiff English (1988), coauthor (with Howard Giles) of Language: Contexts and Consequences (1991), editor of English in Wales: Diversity, Conflict and Change (1990), and author/editor of seven other books in sociolinguistics, discourse studies, and social gerontology. J (J) C. D is the former head of the Dialectology Department of the P. J. Meertens Institute in Amsterdam. She has written a monograph on the dialect of the island (then) of Wieringen, with special emphasis on the ethnological aspects of island life, and she published the volume on North Holland in a series of dialect atlases (RND), for which she did all fieldwork. She was the principal author of a dialect atlas (ANKO) tracing the historical phonology of Dutch at the dialectal level (2 volumes). Her main interests, however, are (and have always been, even long before 1962) in sociolinguistic questions. She coauthored a study on dialect and standard language among primary school children, and initiated the study of emigrant Dutch in the United States (in the mid 1950s, carrying out fieldwork in Michigan). Her special interests, after retirement, are the standardization process of the Dutch standard language since the 18th century and the social stereotyping of dialect speakers. She has studied the sometimes difficult linguistic and dialectological terminology of the famous 18th-century Dutch linguist L. ten Kate, who discovered the

404

ABOUT THE CONTRIBUTORS AND TRANSLATORS

ablaut system in the Germanic derivational and verbal system and who was the first to suggest the dialectal basis of the emergent spoken standard language. J D-OC is Assistant Professor of Germanic Linguistics at the University of Alberta. The study in this volume is part of a larger study of language attitudes and language ideology in postunification Germany, which used perceptual dialectology, qualitative, and matched-guise methodologies. Although her primary emphasis is on German, she has also worked on a sound change in progress in Michigan, as well as the discourse marker “like” in American English. Presently, she is working on two studies: one of sociolinguistic variation in the city of Hanover, Germany, and one of attitudes toward, usage of, and acquisition of the discourse marker “like” in Western Canada. M D is Assistant Professor of English at Illinois Normal University. She is interested in sociolinguistics, cross-cultural communication, and second-language acquisition (the latter from the point of view of universal grammar). She has presented findings from her research at a number of national and international conferences and is coauthor of a more extensive study of the perception of Turkish dialects, to appear in Language Awareness. B E. E is a doctoral candidate in linguistics at Michigan State University. She has a BA and an MA in French from Kent State University and is interested in translation, language teaching and acquisition, and French phonetic variation. More recently, her interests have included work on minority ethnic group acquisition of features of the vowel change in the United States known as the “Northern Cities Shift” and an acoustic study of variety imitations. P G was a lecturer in the Linguistics Department at the University of Wales, Bangor, from 1986 until 1994, when he moved to the Centre for Language and Communication Research at the University of Wales, Cardiff, where he is now Senior Lecturer and Director of Research. His research field is in sociolinguistics, language attitudes, and language awareness, and he has held a number of research awards to work in these areas from the Economic and Social Research Council, University of Wales, European Community. He has published numerous research articles in journals such as Language in Society; Language and Communication; Journal of Multilingual and Multicultural Development; Language Awareness; Language and Education; Language, Culture and Curriculum, and Multilingua. He has also published chapters in various edited volumes. He is Editor of the journal Language Awareness (Multilingual Matters) and Secretary of the Association for Language Awareness. He coedited (with Carl James) the book Language Awareness in the Classroom (1991) and is currently coediting (with Allan Bell) a volume of papers titled Approaches to Media Discourse. He teaches courses at Cardiff on sociolinguistics, language attitudes, and communication research methods. He has also taught in Argentina, Austria, Gambia, Germany, Hungary, Sudan, and Sweden. A. C. M. G is a member of the P. J. Meertens Institute of the Royal Netherlands Academy of Arts and Sciences in Amsterdam. His main focus of interest is on the

ABOUT THE CONTRIBUTORS AND TRANSLATORS

405

morphology and phonology of Dutch dialects in a variationist setting. He is currently working with others on a morphological dialect atlas of the Netherlands and Flemish Belgium (3 volumes planned, 320 maps), the first volume of which will contain pluralization and diminutivization. Other interests include subjective dialectology, presumed great vowel shifts in Dutch dialects, and verbal inflections on subordinating conjunctions — so called comp-agreement. W A. G currently divides his time between his duties as a Catholic priest and his linguistics research. He received his doctorate in 1939 from Université catholique de Louvain, conducting extensive dialectology fieldwork both in his native Europe and China before coming to Japan in 1950. In Japan, he has been involved in the planning and execution of several large-scale dialect surveys, including the Itoigawa survey conducted with Takesi Sibata. He has published numerous articles on the dialects of China and Japan and has worked on extensive translations into Japanese of the works of European dialectologists. He has taught dialectology and related courses in Europe, China, and the United States, as well as at several universities in Japan. His most recent book is The Sanctuaries in a North-China city: A Complete Survey of the Cultic Buildings in the City of Hsüan-hua (Chahar) (1995). L C. H received her PhD from the Department of Linguistics and Languages at Michigan State University and is Director of Research at Lesley College in Boston MA. In addition to her language attitude studies, her research and teaching interests include a focus on cross-cultural pragmatics and intercultural communication. She has presented her research findings at AILA and NWAV conferences, among others. She is coauthor of a study of labels of perceived dialect areas from several regions of the United States. F I is Professor of Japanese Linguistics at Tokyo University of Foreign Studies. His research interests include language variation, in particular, language attitudes and “new dialect” phenomena. In addition to contributing English articles to numerous linguistics books and journals, he is theauthor of Kotobazukai no Shinfûkei [New Landscapes of Spoken Japanese] (1989) and Hôgengaku no Shichihei [New Horizons in Dialectology] (1994). He is currently coediting Sociolinguistics in Japanese Contexts, an English collection of Professor Takesi Sibata’s most important papers. He is actively involved in several international linguistics organizations and currently serves as a Vice President of the International Society for Dialectology and Geolinguistics. B K received his PhD in May of 1997 from the Department of Linguistics at Michigan State University. He specializes in sociolinguistics and discourse analysis and has published articles in the Journal of Language and Social Psychology, Journal of Pragmatics, Folia Linguistica, and Discourse and Society, among others. His current research interests involve the ways in which racist ideology is reproduced in everyday discourse. L K is professor of German at the University of Antwerp, where he has been since 1969. He has studied at the University of Aberdeen and received his MA from the University of Freiburg and his PhD from the University of Münster. His research deals

406

ABOUT THE CONTRIBUTORS AND TRANSLATORS

mainly with dialectology, sociolinguistics, and onomastics of the regions on both sides of the German-Dutch border and with Dutch-German language contact. He has published several books and a number of articles in these fields, among them Sprache und Geschichte im westfälisch-niederländischen Grenzraum (1978), Grenzmundarten und Mundartgrenzen: Untersuchungen zur wortgeographischen Funktion der Staatsgrenze im ostniederländisch-westfälischen Grenzgebiet (1979), and Diglossiestudien, Dialekt und Standardsprache im niederländisch-deutschen Grenzland (1993). One of his major projects is a large-scale microtoponomastic survey of the West-Munsterland region, in cooperation with the West-Munsterland Institute Research Committee of Regional Studies at Vreden, Germany. He chairs the West-Munsterland Institute Research Committee and is a board member of the Westphalian Commission for Dialectology and Onomastics in Münster. L K is a PhD candidate at Michigan State University, in the Department of Romance and Classical Languages. His dissertation, interdisciplinary in nature, deals with perceptual dialectology and French and Francophone cultural studies. He is currently at the University of Wisconsin at Milwaukee, where he is a course coordinator of French instruction. D M. L, Professor Emeritus of English linguistics at the University of Missouri — Columbia, received his PhD in English Language and Linguistics from the University of Texas in 1968. He has published articles in the areas of Spanish-English bilingualism, pedagogical linguistics, and dialectology and is editor of the Missouri Folklore Society Journal. He also recently coedited American Pronunciation, 12th edition, expanded (1994), by John S. Kenyon and a pedagogically oriented anthology on language variation for the Modern Language Association of America (1993). D L is Associate Professor of Japanese Linguistics at Tokyo Metropolitan University. In 1995, he received his PhD from Osaka University for a dissertation on attitudes toward variation in Japanese. He is the coeditor of a bilingual text titled Japanese Sociolinguistics Illustrated (1997) and is a contributing author to several Japanese dialectology and linguistics books. He currently serves on the editorial boards of American Speech and the Journal of English Linguistics. He has begun compiling a bibliography of English books and articles on Japanese linguistics that may be viewed on the Internet at www.age.or.jp/x/oswcjlrc/jlrc/sl-lib-e.htm. Y M held teaching posts at Nagano University and Hiroshima Jogakuin University before assuming his current post as Professor of Japanese Linguistics at Ferris University in Yokohama. He has done extensive research on the geographical and social factors involved in language usage and change, particularly on the dialects of Nagano. His references works include A Dictionary of Tone-Accent on Words in the Tokyo Dialect (coeditor, 1985) and A Dictionary of Tone-Accent on Words in the Hiroshima Dialect (editor, 1994). His major articles dealing with dialect distribution and dialect divisions have been collected and published as Gengo Chirigaku Kenkyû (1992).

ABOUT THE CONTRIBUTORS AND TRANSLATORS

407

K N is Professor of Japanese Linguistics at Kobe Shoin Women’s University and Director Emeritus of the Japanese Language Research Institute. His work in sociolinguistics has focused on language behavior, particularly honorifics. His publications include Nihonjin to Nihongo [The Japanese and the Japanese Language] (1978) and Keigo wo Tsukaikonasu [Using Honorifics Well] (1987). He currently serves as president of the Society for Teaching Japanese as a Foreign Language. D R. P is a sociolinguist and Professor of Linguistics at Michigan State University. His primary interests lie in the areas of U. S. dialectology, folk linguistics, discourse, and second-language acquisition. Within the area of folk linguistics, he has developed a number of approaches in perceptual dialectology — techniques for acquiring and generalizing (1) mental maps of respondents’ notions of dialect distribution, (2) respondents’ evaluations (along affective lines) of regional varieties, (3) respondents’ understandings of the degree of difference between the speech of their home areas and others, and (4) respondents’ recognition and differentiation strategies in identifying language varieties. More recently, he has devised means of dealing with the qualitative (discoursal) contributions of respondents in this general area. His published contributions to this subfield are listed in the References section of this volume. W. G. R was, from 1960 to 1966, in the dialectology department of the P. J. Meertens Institute in Amsterdam and studied mainly lexicographic matters. He did considerable fieldwork (taped archives) in the northeastern part of the Netherlands as part of a joint project with the ethnology department on “Wörter und Sachen” in agrarian settings, and he had an interest in the function of “things.” He drafted the manuscript version of the original arrow map that formed the basis of his article (Chapter 1, this volume) and of Daan’s later color map (Chapter 2). T S is Professor Emeritus of Tokyo University and Saitama University. While a researcher at the National Language Research Institute, he directed numerous large-scale linguistics research projects, including the monumental Linguistic Atlas of Japan. He has authored and edited dozens of books, both in the technical and the popular vein, on subjects ranging from dialectology and sociolinguistics to semantics and lexicology. His most recent publications include the twelve-volume Linguistic Atlas of Itoigawa (1988–1995), which includes three volumes of English analyses authored by W. A. Grootaers. He has served on groups such as the Government Advisory Committee on the Japanese Language and the NHK Public Broadcasting Network’s Language Usage Committee. He is a member of the distinguished Japanese Council of Sciences. A A. W, formerly chair of Dutch linguistics and dialectology at the Catholic University of Nijmegen is a prolific author. His doctoral work was a study on the historical phonology and the division of the dialects of North Brabant. He published the volume on southern South Holland and western North Brabant in a series of Dutch dialect atlases (RND), for which he did all fieldwork. For educational purposes, he edited two anthologies of 16th-century and 17th-century Dutch with extensive introductions to the phonology, morphology, and syntax that appealed more to his linguistic colleagues. He

408

ABOUT THE CONTRIBUTORS AND TRANSLATORS

published two reference books — a manual of Dutch dialectology (the foundation of which was laid in his doctoral work and which has two forerunners, the first one also in book format) and, rather recently, after retirement, a general manual of the historical and synchronic phonology of Dutch dialects. He is the founder of two successful dialect dictionaries (ordered according to semantic criteria) — one for the dialects of Brabant (WBD) and one for Limburg (WLD). He is the principal cofounder and first president of the European Dialect Atlas (ALE: 4 volumes already published). His other research centered on dialect contacts between Dutch and Old English and Old French, subjective dialectology, phonology, and dialect syntax. His most recent two books concern etymological and lexicographic matters — onomasiology of disease names and an etymological dialect lexicon. A W completed her MA and PhD in Communication at the University of California, Santa Barbara. Prior to her appointment as Associate Professor in Communication at the University of Oklahoma, she spent a postdoctoral year as a research fellow at the Centre for Language and Communication Research, University of Wales, Cardiff, researching attitudes toward Welsh English on a University of Wales-funded project directed by Peter Garrett and Nikolas Coupland. She is currently a lecturer at the Centre for Language and Communication Research, where she is Director of the MA in Language and Communication Research. At the University of Wales, Cardiff, she teaches courses in research methodology, statistics, and relational communication. Her main area of research interest is at the interface of language and social psychology, particularly as it relates to aging and intergenerational communication, but she also maintains an active research interest in language attitudes. She has published numerous articles concerned with these issues in journals such as Human Communication Research, Language and Communication, Multilingua, Journal of Language and Social Psychology, and Journal of Multilingual and Multicultural Development. She is currently preparing a book with John Nussbaum (University of Oklahoma) on intergenerational communication. M Y has a BA in English and international relations from Kyoto Sangyo University and an MA in linguistics from the University of Michigan and is currently a doctoral student in linguistics at Michigan State University. She has taught Japanese as a second and/or foreign language in the Republic of China, Tokyo, Kanagawa, Shiga, Osaka, and at Michigan State University for about fourteen years. She has been involved in training teachers of Japanese for four years and has published Japanese language teaching materials. In addition to the teaching of Japanese, she is interested in Japanese linguistics in general and Japanese sociolinguistics in particular.

Index

Boldface type identifies entire chapters; names (of authors) are generally not cited in the index if the occurrence in the text is a reference only (with no discussion or quotation) A AAVE (African American Vernacular English) xxxviii, 369 abstractness (of language) 12 age xxxvi, 18, 31, 36, 64, 244, 248, 264, 276–278, 280–281, 351–353, 371–373 age-grading 282–283 Alps (Japanese), see Gifu, Nagano applied linguistics xxiv-xxv, xxxvii, 147 attitudes (language) xxiv, xxxiii, xxxviii, 12, 85, 138, 142, 147–149, 152–153, 163, 192, 199, 213–21, 227–228, 240–241, 244, 250–261, 264–268, 319–331, 333–343, 345–346, 359–373 Australia 152–155 authenticity (of variety) 334 B Bach 61 Ball 337 Balsom 335–336 Bates 346 Baumgartner 53–54 Beekman 22 Belgium 10–12, 23, 138–139, 250, 251, 254–255 Benhardi 14 Bezoen 23 Boal 333

Bourhis 333, 348 Britain (perception of dialects in) 151–156, 162–176, see also Wales Brook 155–156 Brown 347 Büld xl, 32, 131, 136 C Cajun 297, 308 Carver 312, 325 Chambers xxxiii, 345 change (in language) 18–19, 31–32 cluster analysis xxxvi-xxxvii, 150, 169, 251–256, 260, 265–269, 318–329 community 53, 60, 101, 123 comprehension (of varieties), see degree of difference contact (with other languages) 258–260, 265, 267, 270, 272–273, 307–309, 324, 358 contact (with other varieties) 82–83 conversational data xxxiv-xxxv, 35, 73–74, 76–77, 103–104, 228, 231, 235, 240, 248, 255–260, 264, 323–324, 327, 329–331 “correctness” xxiv, xxxvi, 10–12, 41, 228–232, 244, 247, 250–253, 264–267, 319–331, 338–343, 361, 364–373 Coupland xxxvii, 333–343, 334, 345–358

410 covert prestige 278, 370 Cremona 346 Creole 297, 308 cursing, see taboo D Daan xxix, 9–30, 20, 22, 24, 31, 33, 137–139 Dailey-O’Cain xxxvi, 227–242 DARE (Dictionary of American Regional English) 310–312 Davies 348 Day 346 De Tourtoulon 135 degree-of-difference xxxiv-xxxvi , 227, 244, 247, 253–255, 264, 269, 316–331, 361, see also Japanese methods (Grootaers-Sibata), “little arrow” method Demirci xxxvi, 263–281 Den Hertog 11–12 dialect image, see image dialect status (e.g., versus “regional speech”) 13 dialectology (dialect division) xxvi-xxxi, xxxvi, xl, 3–5, 9–30, 31–32, 35–36, 39, 47, 49–50, 58, 60, 66, 88–96, 116–121, 127–128, 135–136, 243–244, 246, 254, 310–312, 345, 348–349 Diercks xl Dutch dialectology 3–7, 9–30, 135–144 history 10, 14–16, 137 E Eckert 371–373 education 67–68, 271 Edwards 225, 334–335 “empty” areas 189, 260–261 England, see Britain Entjes 30 Eskimo (language) 308 ethnicity xxxviii evaluation, see attitudes, “correctness” and “pleasantness” F factor analysis 364–366 Fasold 362

INDEX Fluck 333 Fokkema 14, 16 folk linguistics xxiii-xxv, xxxix, 334, 360 folk speech 13 Fox 142 France xxxvi, 243–262 French (in the US) 308 G Garrett xxxvii, 333–343, 345–358 gender xxxvi, 244, 264, 274–280 general linguistics xxiii Germany xxxvi, 31-36, 227-242 Gifu (Japan) 71–99, 101–113 Giles xxxiii-xxxiv, 225–226, 323, 333–334, 337, 347–348, 366, 369 Gilliéron 128 Goeman xxviii-xxix, xxxii, xxxix, 7, 135–144 Goosens xxviii-xxix, 16, 141 Gould xxxiv, 200, 333, 361 Graff xxxviii, 373 grammar 22 (Dutch), 58, 74, 76, 88, 91–92, 94, 103–104, 106–109, 213 (Japanese), 265 (Turkish) Great Britain, see Britain Grosse 120, 127 Grootaers xxx-xxxii, xxxv, 33, 60, 115–129, 131–132, 141–142, 199, 221–222 H Haag 61 Halliday 372 Hammarström xxxi, 118–121, 128 hand-drawn maps xxxiii-xxxvi, 164–174, 178, 200, 228, 236–238, 244, 248–250, 264, 283–314, 315–319, 335, 361–363 Harris xxxviii, 373 Hartley xxxvi, 315–332, 370, 373 Haugen 339 Hayashi 3, see multidimensional scaling Heeroma 14, 16–17, 24, 128 Herzog 372 historical-political (and other “nonlinguistic” factors) xxix-xxx, xxxvi, 10, 21–22, 31–32, 46–49, 50–53, 58, 60, 65, 96, 101, 116–123,

411

INDEX 125–127, 141–142, 162, 174–176, 229, 239–241, 257, 264, 270–273, 283 Hoenigswald xxiv Hof 23 Hogg 356 Hol 22 home area 181, 186, 196–197, 204–214, 218–220, 224, 253, 255–254, 265, 267–269, 271, 323 Hopper 339 Howe xxxiv, 361 I identification (of varieties) xxiv-xxxv, 19–20, 162–163, 345–359, 361 image (dialect) xxxiii, 149–159 imitation xxxv, xxxviii, 4, 32, 110, 240, 248, 255, 258, 329, 371 inferiority, see linguistic security (and insecurity) Inoue xxxiii, 149–159, 161–176, 222–223 insecurity, see linguistic security (and insecurity) intonation 4, 20, 25, 213, 216, 258 Irish 166, 168 Irvine 360 isoglosses xxvi-xxvii, 9, 17–18, 24–25, 47, 50, 58, 66–69, 88–96, 118, 361 isolated regions xxvii, xxxvi, 46–49, 270 Itoigawa (Japan) xxx-xxxi, 39–62, 81, 96, 102, 115, 117, 122, 127, 162, 178 J Japan xxxiii, xxv-xxxvi, 39–62, 63–69, 71–99, 101–113, 115–129, 148–150, 199–226 Japanese dialects 39–62, 63–69, 71–99, 101–113, 115–129, 150, 177–198 Japanese methodology Grootaers-Sibata xxx-xxxii, 39–45, 63–64 Inoue xxxiii, 148–151, 162–164 Long xxxv-xxxvi, 178–179, 181–186, 200 Mase xxxi, 72–81, 86, 102 “Rule 1” 76–78 “Rule 2” 78–80 Nomoto xxxi, 64–69 Weijnen (criticism) xxxii, 131–133

Jernudd xl K Kern 15 Kiliaan 17 Kitamura 61 Kleiner xxxvi, 263–281 Kloeke 19, 21, 23, 29, 128 K-means cluster analysis, see cluster analysis Kremer xxix, 31–36, 31, 35–36, 138 Kuiper xxxvi, 243–262 Kuitert 18–19 L labels xxxv, 166, 180–185, 190–191, 200, 213–221, 228, 246–247, 255–260, 286–288, 340–341 Labov xxxviii, 176, 273, 281, 345, 347, 360, 372–373 Lambert 347, 359 Lance xxxvi, 283–314 language attitudes, see attitudes (language) levels (of dialect consciousness), see degree of difference Lewis 348 lexicon 18–19, 23, 33 (Dutch), 49–50, 64, 66, 73, 88–89, 91–96, 104, 106–109, 213 (Japanese), 120 (French) Limburg 7, 23, 136 linguistic features (dialect characteristics) xxxviii, 14, 16–17, 104–105, 162, 176, 341, 373 Dutch 3–4, 9, 14–25 English 273 French 257–259 Japanese 47, 49–50, 58, 60, 66–68, 73–74, 76, 87–96, 103–109, 113, 162, 193 Swiss German 53–54 linguistic security (and insecurity) xxxvi, 148, 225, 267, 271, 323 “little arrow” method (Pfeilchenmethode) xxv-xxx, 3–7, 10, 17–18, 21–22, 32–33, 39–40, 131, 136, 139 Long xxxv-xxxvi, 177–198, 199–226

412 M Mase xxx-xxxii, xl, 71–99, 101–113, 115, 121–123, 132, 141–142, 162 matched-guise, 336–337, see also attitudes (language) Mather 158 Mattheier 32, 135 McClenaghan 360 Meertens 12 methodology, see hand-drawn (perceptual) maps, Japanese methods, “little arrow” Milroy, L. 279, 360 Milroy, J. 373 morphology, see grammar Moulton 16, 128 multidimensional scaling (“Hayashi 3”) xxxiii, xxxvi-xxxvii, 148, 163, 168–170, 251–254, 318–329 N Naarding 20 Nagano (Japan) 71–99, 101–113 Native-American (Amerindian) 308 Netherlands xxvi-xxix, 3–7, 9–30, 31–36, 136–144 Netherlands-German border 31–36 network 279 Niedzielski xxxv, xxxix Nomoto xxx-xxxi, xl, 63–69, 126–127 North Brabant (Netherlands) xxvi-xxviii, 18, 21, 136–137, 141 Northern Ireland 161 O Okumura 62 Oregon 315–332 “outsiders” 252 P Peeters xxviii, 136–137 phoneme, see structural dialectology phonology, see pronunciation Picard 347 pitch accent (Japanese) 47, 85, 88–89, 91–96, 106, 112–113, 198, 213 “pleasantness” xxxiii-xxxvi, 12, 149, 199–221, 232–236, 244, 247, 255–255,

INDEX 264, 267–268, 319–331, 338–343, 356–357, 361, 364–373 popular culture 148, 174–175, 243 Powesland 347 prescriptivism xxxvii, 4, 110, 243–244, 254–255 Preston xxxiii-xxxv, xxxviii-xxxix, 143, 163, 193, 199, 223, 244, 263–266, 271, 315, 322–323, 327, 330, 335, 345–346, 354–355, 361–363, 370 Price 333 pronunciation 3–4, 9, 15, 17–22, 24–25 (Dutch), 49–50, 58, 66, 73, 76, 91–92, 95, 99, 103, 106–109, 113, 162, 213 (Japanese), 53 (Swiss German), 258–260 (French), 265, 267 (Turkish), 273 (US English), 351, 353 (Welsh English) Protze 120, 127 R Rampton 376 recognition (of varieties), see identification regional identity 283–284, 330 regional variety 14 Rencher 347 Rensink xxix-xxx, 3–7, 33, 39–40, 131, 137 Rosenthal 346 Ryan xxxiii-xxxiv, 366, 370 S Sado (island, Japan) 63–69, 126–127 salience (of linguistic features) xxix, 91, 106, 111 Schirmunski 140 Schneiderman 346–347 school districts 65, 125, 127 Schrinen 23 Scotland 161, 166, 168 Sebastian xxxiii-xxxiv, 364, 370 security (linguistic), see linguistic security (and insecurity) semantic-differential (scale), see attitudes (language), “correctness,” degree of difference, and “pleasantness” sensitivity (to dialect differences) 35, 48 sex, see gender

413

INDEX Shimo-kita (peninsula, Japan) xxxii, 124–126 shopping (zones) 50–51, 82 Shuy 362 Sibata xxx-xxxii, xxxv, 39–62, 60, 63–65, 71–72, 96, 101–102, 112, 115, 117, 121, 123–126, 131–132, 148, 162, 199, 221–222 social network, see network solidarity, see “pleasantness” Spanish (in the US) 307–309 Speitel 158 standard language 10–11, 13, 31–32, 36, 40, 85, 137–138, 143, 155, 157, 186, 191, 210–212, 222–223, 247, 253, 259–260, 271, 281, 370 status (social) xxxviii, 11–14, 53, 85, 106, 244, 264, 275, 278–280, 364, 371–373 stress (accent) 20, 257 Strong 347 structural dialectology 16–17, 128 swearing, see taboo Swiss German 16, 53–54 Switzerland (French-speaking) 250, 252, 254–255 T taboo 258 Tajfel 333, 355 talent (for variety perception) 68 Te Winkel 14–17, 19 Terwey 11 Thomas 334 Tôjô xxx, 39, 50, 62, 71, 101, 116–121, 177 Tokugawa 60, 127, 177 Trudgill xxxiii, 171, 173, 273–274, 278–279, 345, 348, 370 Tucker 359 Turkey xxxvi, 263–282 U United States xxxvi-xxxvii, 157, 283–314 Alabama 373 Alaska 324 Appalachia 299 Cajun 297

California 306–307, 323–324 Creole 297 Florida, 297, 328 Hawaii 324 Michigan 363–366 Midwest 219, 291–296, 318, 324–325 New England 284–287 New York City (and surrounding areas) xxxviii, 286–287, 326–327 Northeast 284–287, 317, 325–326 North 287–291, 364–373 Northwest 306 South 219, 296–302, 317, 327–329, 368–377 Southwest 303 Texas 301, 303–304, 328 West Coast 306–307 West 302, 304–306, 319, 321–324 urban dialects 12–13, 20, 24–25, 35, 53, 137, 151, 155, 157, 230, 247, 255–258, 262, 266, 268 V Van Ginneken 15–17, 139 Van Veen 25 vernacular 13 Viereck 172 W Wales xxxvii, xl, 161, 166, 168, 333–343, 345–359 Wegener 32 Weijnen xxvi, xxviii-xxxi, xl, 5, 7, 16–18, 21–22, 24, 32–33, 131–133, 136–137, 140–142 Weinreich 372 Wenker 60 White xxxiv, 200, 361 Willems xxxix, 138–139 Williams, A. xxxvii, 333–343, 345–358 Williams, C. 333, 335 Winkler 10, 14–15 written language 11, 255 Z Zahn 339 Zinsli 53–54, 60