Advances in the Sociology of Language: Volume 2 Selected Studies and Applications [Reprint 2018 ed.] 9783110880434, 9789027923028

Citation preview

Contributions to the Sociology of Language

Edited by

Joshua A. Fishman

Advances in the Sociology of Language VOLUME II

Selected Studies and Applications

Edited by

JOSHUA A. FISHMAN Yeshiva University

1972 - M O U T O N - T H E H A G U E • P A R I S

© Copyright 1972 in The Netherlands. Mouton & Co. N.V., Publishers, The Hague. No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publishers.

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 77-182463

Printed in The Netherlands by Mouton & Co., Printers, The Hague.

For Gele, Monele, Dovidl, and Avremele, with boundless affection

Preface

Joshua A. Fishman ADVANCES IN THE SOCIOLOGY OF LANGUAGE

In many fields of inquiry books of readings have recently proliferated to such an extent that it has become necessary to consider whether or not they contribute to scholarship (Onuf 1969). The sociology of language, I believe, is not yet overburdened with such collections and it is my hope that the present two-volume collection may be as well received and as much utilized by colleagues and students as was the one that it attempts to update (Fishman 1968). A FORMATIVE HALF-DECADE

Much has happened to the sociology of language in the six years that have elapsed since the summer of 1964 when the Committee on Sociolinguistics of the Social Science Research Council convened a group of linguists and social scientists for eight weeks of joint study (Ferguson 1965). Not the least of these has been the alacrity with which the coexistence of the sociology of language and sociolinguistics has been accepted by various segments of the scholarly community. Writing in 1964 (when the first Readings were actually completed, only to be delayed for four years on the publisher's assembly line) it seemed clear to me that sociology of language implied a broader field of interest, and one that was less linguacentric, than did sociolinguistic. In the last few years I have found linguists more willing to grant this point and sociologists more insistent in connection with it than I had ever dreamt would be the case. Therefore, after having succumbed briefly to the more exotic appellation (e.g. Fishman 1970), I have subsequently returned whole-heartedly to my original usage and to the one which is in closer

8

Joshua A. Fishman

agreement with my own interests and sympathies (Fishman 1971). I believe that many of those who still refer to "sociolinguistics" in the current collection will also come to prefer "the sociology of language" in the years to come, particularly as the differing implications of these two names and the approaches they signify become more widely recognized. SOCIOLINGUISTICS

After all is said and done, the differences between these two areas or emphases of specialization may well be far less significant than their similarities. Both are concerned with the interpénétration between societally patterned variation in language usage and variation in other societally patterned behavior, whether viewed in intra-communal or in inter-communal perspective. However, the adherents of sociolinguistics tend to stress the first part of this definition ("societally patterned variation in language usage"), finding in such emphasis a welcome expansion of the more traditional approaches to the underlying regularity of language. Sociolinguistics has been viewed, very largely, as a means of widening the contextual horizons of linguistics, beyond the phrase, beyond the sentence, beyond the utterance, to the speech act, the speech event and the speech occasion. Social units such as the latter (and other units by means of which they are ethnographically detailed) can be demonstrated to regulate, predict or generate systematic phonological, syntactic, morphological and semantic patterns where only free variation or weak structure would be evident without their aid. Essentially then, sociolinguistics has normally accepted the linguistic pursuit of system-in-language, although it has usually derived such system from the data of natural speaking (or natural writing) per se, rather than from more artificial corpuses elicited from informants. As a result, in the space of half a decade, erstwhile sociolinguists have come to claim or admit that what they were doing was "really linguistics" - perhaps of a somewhat broader, newer kind, a kind that recognized social-contextual units as well as the more traditional intra-code units - and that the term sociolinguistics might ultimately no longer be needed once most linguists came to recognize and accept the newer approaches and goals with which the broader contextualization of language structures was associated. Thus, the leading advocates and adherents of sociolinguistics are also commonly the ones that prophesy its earliest demise, not for lack of success but, on the contrary, as a result of hopefully carrying the day within the fold of linguistics proper.

Preface

9

SOCIOLOGY OF LANGUAGE

No such self-liquidating prophecy characterizes the sociology of language. It does not seek to capture or replace sociology as a whole or any of its specializations. Nor does it merely seek to relate communicative content or whole-code designations to social categories or social structures. Certainly it seeks whatever level of linguistic sophistication may be necessary in focusing upon micro-level or macro-level social processes and social problems. Rather than emphasize the ethnography of communication as an end in and of itself the sociology of language would hope to utilize the ethnography of communication, as it would utilize sociolinguistics and social science more generally, in order to more fully explain variation in societally patterned behaviors pertaining to language maintenance and language shift, language nationalism and language planning, etc. However, not only are supportive and adversary behaviors toward particular languages or language varieties close to the heart of the sociology of language but so are group self-identification behaviors, group formation and dissolution processes, network permeability differentials, referential membership behaviors, language attitudes and beliefs, etc. All in all then, the sociology of language is concerned with language varieties as targets, as obstacles and as facilitators, and with the users and uses of language varieties as aspects of more encompassing social patterns or processes. The relationship between the sociology of language and sociolinguistics is thus a part-whole relationship, with the whole not only being greater than any of the parts but also greater than the sum of all of the parts taken separately. While continuing to use the adjectival and adverbial modifier sociolinguistic it is now clearer to me than it was in the past that the sociology of language has a path of its own to follow. Those colleagues who joined and encouraged me in the organization of committees on the sociology of language in the International Sociological Association and in the American Sociological Association have in many instances seen this path more quickly and more clearly than I and have increasingly clarified the differences between the sociology of language and sociolinguistics which I initially recognized largely on an intuitive level in 1964. SEVERAL RECENT DEVELOPMENTS

The current two volume collection reflects the increased sophistication, subtlety and interdisciplinary grasp of specialists in the sociology of language in comparison to that which obtained half a decade earlier. In comparison to the collection that preceded the present one the

10

Joshua A. Fishman

sociology of language currently appears to be: (a) substantially more integrated around systematic questions rather than merely descriptive, fragmentary or accidental in focus; (b) substantially more data oriented, as distinct from merely programmatic or argumentative; (c) substantially more quantitative, as distinct from ethnographic, anecdotal or observational in so far as data analysis is concerned, and (d) substantially more interdisciplinary, combining both linguistic and social science skills at an advanced level, rather than referring to one parent discipline or the other in a purely passive or ceremonial vein. All four of the above mentioned characteristics or trends of more recent sociology of language strike the compiler of this collection not only as being essentially praiseworthy but also as further justifying and solidifying the designation of the field itself. The movement from bias to theory and from theory to data (and, by and large, to publicly verifiable data) should not only make possible better theory in the future but should also make possible a stronger movement from data to application. The tendency to quantify is not only a tendency toward greater precision and rigor but also makes possible more difficult questions as well as more complex models than would otherwise be feasible. The genuinely interdisciplinary nature of the work indicates that one who is merely a "linguistics appreciator" in the sociology of language will soon be as dated and as limited as one who is merely a "music appreciator" in the field of musicology. Hopefully, "sociology appreciators" will also become increasingly rare among linguists who claim to have serious interests in language and society. There is much evidence of such a trend too in many of the papers included in this collection. A final indication of the greater maturation and stabilization of this field relative to its position half a decade ago is the fact that most of those whose work is sampled in this collection are fully identified with the sociology of language or an allied field, are continuing to revise and advance their work in this connection, and may be expected to remain active in it for many years to come. SPECIALIZED AND GENERAL INTEREST in order to fully reflect tendencies (a) and (d), above, volume I of this 2 volume collection has been specifically devoted to Basic Concepts, Theories and Problems: Alternative Approaches. In order to fully reflect tendencies (b) and (c), above, volume I I has been particularly devoted to Selected Studies and Applications. Although there is undoubtedly a direct relationship between these four matters the separation into two

Preface

11

volumes should permit students and instructors to more intensively utilize one or the other, if that is in accord with their preference, and to do so with greater ease (and at lower cost) than would be the case with a doubly large one-volume collection. Instructors eager to stress theoretical issues and to illustrate them via their own favorite choice of up-to-date as well as "classical" readings may well prefer to require volume I, leaving volume II for less intensive library use in conjunction with journal articles of a varied nature. However, in those settings (academic and applied) in which a balanced variety of recent empirical studies and applied considerations is more difficult to come by than is a personally pleasing integrative approach, volume II may well represent required reading, while volume I may be consulted less intensively. Finally, the current and prospective specialist or devotee, for whom the sociology of language as a whole is a field of wide ranging and rather permanent interest and concern, may well find both volumes to be essential in his study and research. Such, at least, is my hope and expectation. INTERNATIONAL SOCIOLOGY OF LANGUAGE

The truly international nature of the sociology of language as a field of inquiry is somewhat masked in this collection by the monolingual nature of its contents. Nevertheless, the uniformity of language of publication is of far less significance with respect to the richness of this field than is the great and welcome diversity in the backgrounds of the scholars involved in it and the even greater diversity in the societies and social settings that these scholars have examined. THANKS

Obviously, a rather large group of students, colleagues and friends deserves to be thanked for helping with the selection of papers and with the preparation of this collection more generally. In this connection I would like to particularly thank not only those authors (and publishers) who gladly granted the permissions without which this volume could not have appeared, but also those authors who helped persuade various publishers to permit republication, and even those authors who sought to influence their publishers along similar lines but who were not successful in doing so. To all of them go my heartfelt thanks for their implied complement, both to me and to the sociology of language. Jerusalem, 1970

12

Joshua A. Fishman

REFERENCES Ferguson, Charles A. 1965 "Directions in sociolinguistics; report on an interdisciplinary seminar", SSRC Items, 19, no. 1, 1-4. Fishman, Joshua A. 1968 Readings in the Sociology of Language (The Hague: Mouton). 1970 Sociolinguistics; A Brief Introduction (Rowley, Mass.: Newbury House). 1971 "The sociology of language; an interdisciplinary social science approach to language in society", Volume I, this collection. (Also to appear in Current Trends in Linguistics, 12, in press, and as a monograph published by Newbury House, Rowley, Mass.) Onuf, Nicholas G. 1969 "Do books of readings contribute to scholarship?", International Organization, 23, 98-114.

Contents

Joshua A. Fishman Preface

7

Section I: Small Group Interaction Lawrence Greenfield Situational Measures of Normative Language Views in Relation to Person, Place and Topic among Puerto Rican Bilinguals

17

Muriel Hammer, Sylvia Polgar and Kurt Salzinger Speech Predictability and Social Contact Patterns in an Informal Group

36

Rolf Kjolseth Making Sense: Natural Language and Shared Knowledge in Understanding

50

Jacqueline Lindenfeld The Social Conditioning of Syntactic Variation in French

77

.

Emanuel A. Schegloff Sequencing in Conversational Openings

Section II: Large-Scale

91

Socio-CulturalProcesses

Basil Bernstein and Dorothy Henderson Social Class Differences in the Relevance of Language to Socialization

126

14

Contents

Joav Findling Bilingual Need Affiliation, Future Orientation and Achievement Motivation

150

G. Richard Tucker and Wallace E. Lambert White and Negro Listeners' Reactions to Various AmericanEnglish Dialects

175

Herbert C. Kelman Language as Aid and as Barrier to Involvement in the National System

185

Jonathan Pool National Development and Language Diversity

.

.

.

213

.

.

231

.

.

.

255

Roxana Ma and Eleanor Herasimchuk Speech Styles in Puerto Rican Bilingual Speakers .

.

.

268

Everett C. Hughes The Linguistic Division of Labor in Industrial and Urban Societies

296

Section III: Bilingualism and Diglossia

Stanley Lieberson Bilingualism in Montreal: A Demographic Analysis Robert L. Cooper and Susan Carpenter Linguistic Diversity in the Ethiopian Market

.

Section IV: Language Maintenance and Language Shift E. Glyn Lewis Migration and Language in the USSR John E. Hofman and Hccya Fisherman Language Shift and Language Maintenance in Israel

310 .

.

342

Andrée Tabouret-Keller A Contribution to the Sociological Study of Language Maintenance and Language Shift

365

Contents

15

Albert Verdoodt The Differential Impact of Immigrant French Speakers on Indigenous German Speakers: A Case Study in the Light of Two Theories

377

Val R. Lorwin Linguistic Pluralism and Political Tension in Modern Belgium

386

Section Applied Sociology of Language: Policy, Planning andV:Practice William F. Mackey A Typology of Bilingual Education Doris R. Entwisle Developmental Sociolinguistics: Inner-City Children

413 .

.

433

John DeFrancis Language and Script Reform (in China)

450

Joan Rubin Evaluation and Language Planning

476

Thomas Thorburn Cost-Benefit Analysis in Language Planning .

.

.

511

Subject index

520

Name index

529

Table of Contents of Volume I Joshua A. Fishman Preface

7

Susan M. Ervin-Tripp Sociolinguistics

15

Allen D. Grimshaw Sociolinguistics

92

16

Contents

William Labov The Study of Language in its Social Context

.

.

.

.

152

.

.

217

Joshua A. Fishman The Sociology of Language: An Interdisciplinary Social Science Approach to Language in Society

.

.

Subject index

405

Name index

413

Section I: Small Group Interaction

Lawrence Greenfield1 SITUATIONAL MEASURES OF NORMATIVE LANGUAGE VIEWS IN RELATION TO PERSON, PLACE AND TOPIC AMONG PUERTO RICAN BILINGUALS2 *

Two situationally based self-report instruments for measuring normative views with respect to bilingual usage were administered to groups of bilingual Puerto Rican youngsters in New York City. The results obtained indicate that the amount of Spanish and English claimed for conversational use with other bilinguals differs according to domain of interaction. Use of Spanish was claimed primarily in the domain of 'family', secondarily for the domains of 'friendship* and 'religion', and least of all in those of 'education' and 'employment', while the reverse was true for English. In more naturalistic situations normative language claims were related to differences in person, place and topic. These differences, when systematically studied in experimentally controlled situations, were found to be almost entirely attributable to interlocutor differences that were associated with the domains examined and minimally the result of differences in topics or locales.

In recent years, several studies have reported on the relationship between verbal behavior and a variety of psychological and social factors, such as the setting, the roles of the participants, the topics of conversation, the functions of interaction, and the views of interlocutors concerning each of the foregoing (Ervin-Tripp, 1964). Labov (1964), for 1

The author wishes to thank Dr. R. L. Cooper for his advice and encouragement during all stages of the work reported here. 2 The research reported herein was supported under DHEW Contract No. OEC-1-7-062817-0297, "The Measurement and Description of Language Dominance in Bilinguals", Joshua A. Fishman, Project Director. Data analysis was made possible by a grant from the College Entrance Examination Board. The preparation of the current presentation was facilitated by the Institute of Advanced Projects, East-West Center, University of Hawaii, where the Project Director spent the 1968-69 academic year as a Senior Specialist. * From Anthropos, 65 (1970), 602-618. Reprinted with permission.

IB

Lawrence Greenfield

example, found a series of phonological alternates in New York English speech which covaried with elicitation methods (that implied varying situational contexts of verbal interaction) and the socioeconomic status of the speaker. Fischer (1958), who studied the alternation between the use of the suffixes in and ing by New England children found that in was used to a greater degree than ing by boys than by girls, by children of lower than of higher socioeconomic backgrounds, in informal than in formal portions of the interview, and with informal verbs, such as chewin and hittin than with formal ones, such as correcting and reading. Brown and Gilman (1960) found that the use of the pronouns tit or vous (and their corresponding verb forms) in several Romance languages depended on relationships of power and solidarity existing between interlocutors. The social and psychological factors that are signaled linguistically by stylistic variation within a single language are frequently expressed by a complete switch in code in some bilingual settings (Ervin-Tripp, 1964; Hymes, 1966; Gumperz, 1964a). Rubin (1962) found that factors such as intimacy and informality were useful in describing the use of Spanish and Guarani in Paraguay. Thus, for example, young men used Spanish when first starting to court their sweethearts but as intimacy developed shifted to Guarani. Gumperz (1964b) and Blom and Gumperz (1966) reported that the use of the local dialect and national standard in a small Norwegian community was predictable from the social background of the interlocutors, the types of networks they formed and the topics discussed. Fishman has proposed the concept of domain in order to specify the larger institutional role-contexts within which habitual language use occurs in multilingual settings (Fishman, 1964, 1965a, 1965b, 1966). In gathering data appropriate to a given domain the investigator abstracts from or samples social situations at the level of face-to-face interaction involving domain appropriate places, role-relationships and topics. For example, in studying habitual language use in the family domain the investigator collects data regarding interactions between such domain appropriate interlocutors as husband-wife, parentchild, grandparent-grandchild, in such domain appropriate locales as 'home', concerning such domain appropriate topics as 'proper behavior of children'. Relevant domains for describing language use in many relatively complex multilingual societies would probably include family, friendship, religion, education, work sphere, and government (Fishman, 1966). Using this concept, Fishman has suggested that it is possible to differentiate between stable bilingual societies in which diglossia obtains (Ferguson, 1959), and unstable bilingual societies. In the former,

Situational Measures of Normative Language Views

19

languages tend to be reserved for different domains of life in the community, i.e., language varieties tend to have definite institutional associations which exist over and above their more fleeting metaphorical functions. In the latter, domain separation in language use vanishes and the 'other' tongue comes to be used alternatively with the 'mother' tongue, particularly in the family and friendship domains. The institutional separation of codes first becomes secondary to their metaphorical functions and then is re-established along quite different institutional lines. In general, unstable intragroup bilingualism has occurred in cases of immigrant languages in the context of rapid industrialization, urbanization, or other rapid social change, as for example, in the cases of Yiddish, Ukrainian, Hungarian and German in the United States (Fishman, 1965a). Examples of more stable intragroup bilingual speech communities have been described by Barker, 1948; Blom and Gumperz, 1966; Ferguson, 1959; Fishman, 1965b; Rubin, 1962; and Weinreich, 1951. Recently Fishman (1968) has developed a sociolinguistic model which suggests that in diglossia situations there generally exist two major clusters of complementary community values, called L and H, respectively, each of which is realized in a different speech variety or language. L-related values are usually those of intimacy, solidarity, spontaneity and informality, while H-related values usually involve an emphasis on status differences, ritual and formality. Furthermore, those members of the community who identify with or accept these two cultural value clusters tend to utilize the culturally approved speech variety or language in their domain appropriate behavior. Typically, the L-variety or language is used in domains such as family and friendship, while the H-variety is reserved for domains such as education, occupation and religion. Moreover, when two individuals interact in a locale or discuss a topic that is incongruent with their usual role-relationship, they struggle toward a redefinition of the situation and tend to use the speech variety or language which is congruent with such a redefinition. For example, a professor and student who are engaged in mountainclimbing may no longer view themselves as professor-student but as individuals interacting in some other role-relationship. Under such circumstances, the variety used would be appropriate to the perceived social relationship and to the re-defined total situation of which that relationship is a part. The present paper reports on two experiments which were designed to examine the possibility of incorporating into the measurement of normative language views via self-report instruments some of the interactional sensitivity derived from microsociolinguistic theory and ethnography. Studies of normative language views have thus far been inter-

20

Lawrence

Greenfield

situational at best. The current paper reports two attempts to study the intra-situational normative views of a bilingual population. Several studies have suggested the possibility that unlike most previous immigrant groups in the United States, the Puerto Rican community in New York has many of the features that Fishman describes in his model of diglossic speech community. One factor that has been mentioned in favor of this possibility is that while adapting to life in the United States, the Puerto Ricans in New York continue to maintain close physical ties with their homeland and as a result, come to identify with the values prevalent in both countries (Padilla, 1958; Senior, 1965; Hoffman, 1968). According to these studies, Puerto Ricans learn from the U.S. the importance of social and economic advancement and from their Puerto Rican heritage the importance of maintaining close contact with family members and friends. Therefore, it is hypothesized that in the Puerto Rican community in New York, Spanish may have come to be associated with values such as intimacy and solidarity and to be used primarily in domains such as family and friendship, while English may have come to be associated with values such as status differentiation and to be used primarily in domains such as religion, education and employment. Within each of these domains we further hypothesize that Spanish and English will be normatively viewed as polarized in the directions of intimacy and status-stressing respectively. EXPERIMENT 1

Method Technique. - The technique used in the first experiment was derived from studies of the structure of conversations which were conducted by Hershkowitz and Krause (1965) and by Blass (1965). In these studies, lists of persons, places and topics were ranked by groups of American college students along the dimensions of intimate-distant, private-public and personal-impersonal, respectively. The students were asked to imagine themselves in a number of conversations of which two components were supplied by E and the third was to be filled in by them. When E supplied a pair of elements which were of the identical scale position (congruent), the Ss invariably selected the third one from the same end of the scale as the others. When the two provided elements were from opposite ends of the scale (thus being incongruent), there was a tendency for S to redefine one of them so as to be congruent with the other and then to select as the completing element one which was congruent with the perceived position of the first two. For example, when presented with the situation of talking to a friend (intimacy-

Situational Measures of Normative Language Views

21

distance rank # 1 or #2) in the park (private-public rank # 6 or #7), some of the Ss explained the situation by saying that "he wasn't really a good friend", while others explained it by saying that "no one was around". In selecting the third element, the former Ss tended to choose relatively impersonal topics, while the latter ones tended to select personal ones. Subjects. - The 5s included in this study were a group of boys and girls of Puerto Rican descent who belonged to a Puerto Rican youth organization, Aspira, which sponsors clubs (conducted primarily in English) in New York City high schools. This organization is a private educational agency designed on the one hand, to build career opportunities and leadership roles for Puerto Rican youth and on the other, to develop in them a positive self-image by strengthening their Puerto Rican identification. Accordingly, this group was used as a basis for securing subjects who were most likely to identify with the two major value clusters in the community. Domain

Interlocutor

Place

Topic

Family Friendship Religion Education Employment

Parent Friend Priest Teacher Employer

Home Beach Church School Workplace

How to be a good son or daughter How to play a game How to be a good Christian How to solve a math problem How to do your job in the most efficient way

Procedure. - Since domains are a higher order generalization derived from congruent situations (i.e., from situations in which individuals interacting in societally appropriate role-relationships with each other, and in the societally appropriate locales for these role-relationships, discuss topics that are considered societally appropriate to these role-relationships and locales), it was first necessary to test intuitive and rather clinical estimates of the congruencies that were felt to obtain in the Puerto Rican community of New York City. After more than a year of general participant observation as well as focused interviews and discussions with native informants it seemed to the authors that five domains could be generalized from the innumerable situations that they had encountered, namely, 'family', 'friendship', 'religion', 'education', and 'employment'.3 As a means of collecting self-report data on normative language use, a situation was selected which seemed to be typical s

For a complete protocol of a validating interview with a native informant see Fishman, Cooper, Ma et al., 1968, Chapter III-3-b (Instrument construction tryout: Tape A, Informant P2).

22

Lawrence Greenfield

of each domain. As indicated above each of these situations consisted of a seemingly congruent situational interlocutor, situational place and situational topic. An instrument was constructed which required S to complete a number of situations in which two of the three components were provided by E* Specifically, S was requested a) to select a third component (from among five alternatives) in order to complete the situation and b) to indicate the amount of Spanish and English they would be likely to use if they were involved in such a situation and if they and their Puerto Rican interlocutors knew Spanish and English equally well. For each situation, amount of each language used was to be rated on a fivepoint scale in which 1 = all in Spanish, 2 = more Spanish than English, 3 = roughly equal amounts of Spanish and English, 4 = more English than Spanish, and 5 = all in English. In some of the situations the components which were provided by E were seemingly congruent, as they appeared to belong to the same domains, and in others they were seemingly incongruent, as one of them appeared to belong to either the family or friendship domains (intimacy value cluster) and the other to the domains of religion, education, or employment (the status value cluster). In acoord with our hypothesis concerning the domains which existed in the community and the persons, places and topics that were congruent with these domains, it was expected that where the two components provided by E were congruent with each other the component selected by S would come from the same domain as both of those which were provided by E. Where the two components provided by E were incongruent with each other it was expected that the component selected by E would come from the same domain as one of the two provided components. Furthermore, in accord with the hypothesis that the community studied was a diglossic speech community, it was expected that a preference for Spanish would be reported when the third component chosen by S was appropriate to either the family or friendship domains; conversely, it was expected that a preference for English would be indicated when the component selected by E was appropriate to the domains of religion, education or employment. The data gathering instrument was entirely in English and consisted of three sections in each of which the situations described constantly lacked a given component, namely, either the person, place or topic. The sections were randomly distributed among the 5s who were tested in groups at the conclusion of their club meetings. 4

The layout of instrument 1 and the full scoring procedures utilized in connection with it are shown in Appendix VIII-2, Fishman, Cooper, Ma et al., 1968.

Situational Measures of Normative Language Views

23

Results Choice of the third component. Table 1 shows the percent of Ss who, for each of five seemingly congruent situations, selected the hypothesized domain-appropriate third component as the completing element. In the situation comprising 'friend' and the friendship topic, the hypothesized friendship locale, beach, was chosen by only 40% of the Ss. In each of the remaining seemingly congruent situations, however, the component which was hypothesized to be congruent with those provided by E was selected by at least 80% of the Ss. TABLE 1

Percent of Ss Selecting 3rd Components Congruent with Two Others Presented by E and Derived from Given Domains Congruent Component Selected Domain Family Friendship Religion Education Employment

Person (n - 16) 81 94 81 81 88

Place (n = 16) 100 40 100 93 100

Topic (n = 18) 89 100 83 100 100

Table 2 shows for each of the seemingly incongruent situations the number of Ss who chose a component which was hypothesized to be congruent with one or another of the two provided components and the number who chose one which was hypothesized to be congruent with neither of them. For each situation of the situation-types studied, at least 85% of the Ss chose an element which was congruent with one or another of those provided by E. Of the situations in which S was provided a seemingly incongruent Place and Topic, the Person selected was congruent with Place in 57%, with Topic in 28%, and with neither Place nor Topic in 15%. Of those situations in which a seemingly incongruent Person and Topic were provided, the Place selected was congruent with Person in 67%, with Topic in 18%, and with neither of the two provided components in 15%. In those situations in which an incongruent Person and Place were provided, the Topic selected was congruent with Person 41%, with Place in 50%, and with neither component in only 8%. All in all, the choice of third component was made congruent with Topic less often than with either Person or Place. Also noteworthy, is the fact that for most of the incongruent (as well as congruent) situations little variation was found in

24

Lawrence Greenfield ©

. 4

o

a o E u 0H

V)P, CJ

o

Ä

E R

> o

.ti

ä

* 8 2 2 60 c

M

Ü

!

E u

E

E S

i,

E

h

E E E 5 5 3 3 | | t 8 c 3 c « l , l H V H O U ' a T 3 C C

JS E •a Ä O

o

O sS o U cu

CS 00 W1 CS

a o H

5

o.

'

&

5

S

|

B

E

E

00 o

u - O C O - O b e U U i O l i - c S l - i

E

a

3

E

E

8

ä

>.

f

>> f

cd •w S

*

Situational Measures of Normative Language Views

25

TABLE 3

Spanish and English Usage Self-Ratings in Various Situations For Components Selected I. Congruent Situations: Two "congruent" components presented; S selects third congruent component and language appropriate to situation. 1 = all Spanish, 5 = all English. Congruent Persons Selected Mean S.D. N

Parent 2.77 1.48 13

Friend 3.60 1.20 15

Total 3.27 1.12 15

Priest 4.69 .61 13

Teacher Employer 4.92 4.79 .27 .41 13 14

Total 4.81 .34 15

Congruent Places Selected

Mean S.D. N

Home

Beach

2.33 1.07 15

3.50 1.26 6

Total 2.60 1.10 15

Church 3.80 1.51 15

School 4.79 .58 14

Work Place 4.27 1.34 15

Total 4.27 .94 15

Employment 4.44 1.12 18

Total 4.38 .73 18

Congruent Topics Selected

Mean S.D. N

Family Friendship 1.69 3.30 .92 1.20 16 18

Total 2.64 .95 18

Religious 3.80 1.47 15

Education 4.78 1.53 18

II. Incongruent Situations: Two "incongruent" components presented; 5 selects third component and language appropriate to situation. 1 = all Spanish, 5 = all English. Persons Selected Mean S.D. N

Parent 2.90 1.20 16

Friend 3.92 .64 16

Total 3.60 .70 16

Priest 4.68 .59 14

Teacher Employer 4.77 4.44 .48 .68 15 9

Total 4.70 .52 15

Places Selected

Mean S.D. N

Home 2.63 .77 15

Beach 3.86 .94 5

Total 2.77 .70 15

Church 3.71 1.32 15

School 4.39 1.20 15

Work Place 4.42 .96 15

Total 4.10 .82 15

Employment 3.81 .85 18

Total 3.49 .76 18

Topics Selected

Mean S.D. N

Family Friendsnip Total 3.81 3.26 2.83 1.02 1.04 1.13 18 18 16

Religious 3.07 1.00 18

Education 3.66 1.20 17

26

Lawrence Greenfield

the selection of the third component, regardless of whether it was a Person, Place or Topic. This is indicative of the existence of considerable normative consensus concerning the naturalness of the situations selected for study. Language choice. Table 3 shows the mean amount of Spanish and English that 5s claimed in connection with various hypothetically congruent and incongruent situations, following their selections of another congruent or any third component, respectively. In hypothetically congruent situations, Spanish was decreasingly claimed for family, friendship, religion, employment and education, regardless of whether the component selected was a person, place or topic. Similar results were found for hypothetically incongruent situations with only three exceptions (5s reported they would use a smaller amount of Spanish upon their selection of the friendship locale than upon the selection of the religious one, and upon the selection of the friendship topic than upon either the selection of religious or educational topics). In addition, all domains became somewhat less different from each other in language selection following hypothetically incongruent situations than following hypothetically congruent ones. However, this finding was less evident in those situations of which the selected third component was a Person, than in those in which it was either a topic or place. An analysis of variance of the mean language usage scores obtained for hypothetically congruent and incongruent situations in which the selected third component was related to Intimacy and Status, is shown in Table 4. The significant Value Cluster effect, F(l,135) = 161.28 (p 2 »0 u •s¡ -a s ° •s a ' S 2"S a ö ® f 9 «¿6.2 oc® '3 S ' a S o "o o< ¡ s.s a

3 o a o c3 •O e .0 a £ tu -s:

). Similar to the strengthening of the French language among the French ethnic population observed earlier is a sharp decline since 1941 among other ethnics in the percentage speaking English only and a doubling in the small percentage speaking French only. It is difficult to account for these trends in a quantifiable fashion since the necessary cross-tabulations are not available. We should recall, however, that there has been a growing movement from rural Quebec to urban Quebec - particularly to Montreal. The fact that rural French Canadians living in more homogeneous settlements are less likely to learn English than are their Montreal compatriots would help explain the drop in bilingualism and the rise in monolingual French-speakers among the French Canadians in Montreal. In 1961, for example, only 9.7 per cent of the French ethnic population in rural Quebec was bilingual. We must note that internal migration probably does not fully account for the decline in French bilingualism, since we have also

240

Stanley

Lieberson

observed a recent increase among immigrants in their propensity to learn French. It would appear that the French Canadians, although still subordinate to the British in the economic world, have risen somewhat, and this may make knowledge of French more useful than it has been for Montrealers. We are told, for example, of a department store that at one time made no attempt to hire clerks who knew French on the grounds that the French Canadians could not afford to shop in its high-priced store. The bilingual Montreal classified telephone directories are relevant here. Under the English-language category, "Department Stores", there are twenty-eight different listings in 1938. Under the French equivalent, "Magasins a Rayons", we find only four listings. In the 1964 Yellow Pages, we find forty-five companies listed under "Department Stores" and forty-nine under "Magasins a Rayons".15 To be sure, the number listed as true department stores must be taken with a grain of salt; however, it is clear from inspection that all of the major downtown stores are now in both the French and the English listings. The possibility of distortions in the census data due to French nationalism should be considered as accounting for the decline in bilingualism among French Canadians. We are inclined to minimize this explanation for several reasons. First, this decline was observed before the recent increase in nationalism. Second, the most recent census was conducted before French-Canadian separatism reached its apex. Finally, if a segment of the population knows English but refuses to speak it, then for many functional purposes such people are equivalent to those who know only French. Considering that in 1961 French was the mother tongue of nearly two-thirds (65 per cent) of the metropolitan population and English the mother tongue of little more than one-fifth (22 per cent) of the population, it is clear that the position of the French language is far from being as strong as we might expect simply on the basis of the composition of Montreal's population. The number speaking French only amounts to 60 per cent of the 1.4 million persons whose mother tongue is French (FMT). The monolingual English population by contrast, amounts to 93 per cent of the 495,000 persons whose mother tongue is English (EMT).16 For 1961, it is estimated that 28 per cent 15

Bell Telephone Company of Canada, Montreal Classified Telephone Directory, July, 1938, pp. 108, 223. I am indebted to G. L. Long, historian, Bell Telephone Company of Canada, for making this volume available. Yellow Pages, Montreal, September, 1964, pp. 291, 552. 16 This does not mean that 93 per cent of the EMT population speak English only. Many EMT persons are bilingual but are replaced in this figure by people with other mother tongues who, of the two official languages, speak only English

Bilingualism in Montreal

241 TABLE 2

Percentage Distribution of Official Language by Ethnic Origin Montreal Area, 1921-61

Population and official language A. All groups: English only French only English and French Neither B. British: English only French only English and French Neither C. French: English only French only English and French Neither D. Other Ethnics: English only French only English and French Neither

Montreal and Verdun, 10 years of age and older

Montreal, Verdun, and Outremont, all ages

1921

1931

1941

1961

1941

1951

1961

25.2 21.9 51.9 0.9

25.9 20.5 52.7 0.9

23.7 28.2 47.8 0.3

18.7 32.5 46.6 2.2

23.9 34.0 41.7 0.4

22.1 35.8 41.1 1.0

19.0 39.0 39.1 2.9

71.4 0.7 27.9 0.0

71.4 0.5 28.1 0.0

n.a.a n.a. n.a. n.a.

n.a. n.a. n.a. n.a.

70.2 1.3 28.5 0.0

70.6 2.0 27.4 0.0

67.1 2.8 29.9 0.2

0.2 34.3 65.6 0.0

0.5 32.7 66.8 0.0

n.a. n.a. n.a. n.a.

n.a. n.a. n.a. n.a.

0.7 51.7 47.6 0.0

1.1 52.8 46.1 0.0

1.0 55.4 43.4 0.2

53.3 5.0 34.5 7.3

53.8 3.7 35.9 6.7

n.a. n.a. n.a. n.a.

n.a. n.a. n.a. n.a.

55.9 5.3 36.1 2.7

50.7 5.6 37.2 6.6

44.4 10.8 31.4 13.4

» Not available.

of EMT British males were bilingual. On the other hand, about 48 per cent of the FMT French males were bilingual.17 Equally lopsided differences are found for females, although they are of a lower magnitude. In other Canadian cities, where the relative numbers with French and English mother tongues are reversed, French fares far worse as the lesser language than does English in Montreal. In Moncton, N.B., for example, where French is the mother tongue of about a third of the population and English the mother tongue of nearly all the remaining two-thirds, only about 4 per cent of EMT British males have learned French, whereas nearly 90 per cent of FMT French males were bilingual in 1961. 17

Lieberson, "Language Questions in Censuses", op. cit.

242

Stanley

Lieberson

INTERGENERATIONAL MAINTENANCE

If bilingualism is to lead to the demise of one of the languages, then the second language of some bilingual parents must be passed on as the mother tongue of their children. If bilingualism is to be the end product of linguistic contact, then the mother tongue of bilingual parents must be passed on as the mother tongue of their children. To simplify the problem, we will consider three elements in our model of linguistic maintenance. First, the proportion in each mother-tongue group that is bilingual clearly influences what we might call the "exposure to risk". We observed for the two leading ethnic groups that a far larger proportion of FMT residents learn English than EMT residents learn French. For all fathers of small children in Montreal-Verdun in 1961 we estimate that about 35 per cent of those with EMT have learned French, whereas about 49 per cent of FMT fathers are estimated to be bilingual.18 In the previous two decades these differences in exposure to risk were even greater. French is subject to a far greater 'exposure' than is English in the sense that proportionately more of the French are potential spreaders or carriers of the English language than the English are of the French language. The relative degree of bilingualism among the different speech groups is critical, since it is reasonable to assume that only bilingual parents face an option in the tongue they pass on to their offspring. Second, we should consider the degree to which each bilingual group passes on the acquired non-mother-tongue language to their offspring, what we might call the 'rate of intergenerational language-switching'. It is most important that the actual behavior of bilinguals with respect to language transfer be distinguished from the rate of bilingual exposure. The sheer fact that the FMT group is more bilingual than is the EMT group does not necessarily mean that a larger proportion of French bilinguals pass on a second language to their children. In fact, it is possible for the two sets of rates to be reversed so that there is a net decline in the language of the linguistic population with the smaller degree of actual bilingualism. This can occur if proportionately more of its bilingual members than of the bilingual component of the other language group actually pass on a second language. Third, we should examine the fertility rates separately for the monolingual and bilingual subpopulations of each mother-tongue group. 18

Estimates are based on the official-language distribution of the 'other* mother tongue population in 1931 (most recent year for which such data are available) and the assumption that all EMT or FMT persons in Montreal-Verdun retain speaking ability in their mother tongue.

243

Bilingualism in Montreal

There can be a higher rate of language-switching among the bilingual segment of a mother-tongue group without a decline in the language if this segment's fertility rates are relatively low compared with those of the monolingual members. In this sense, we must always distinguish between aggregate and individual assimilation. It is possible for individuals to assimilate and yet for their group to maintain itself or to expand in size.19 There are additional variables that might be added to this simplified analytical model, such as differentials in mortality and internal and international migration. However, the three elements described above - exposure to risk rate of intergenerational language-switching and fertility - appear to be the major factors to consider in the Montreal situation. TABLE 3

Percentage Distribution of Mother Tongue of Children under 5 (Actual and Expected) and of Women in the Child-Bearing Ages Montreal and Verdun, 1941-61

Year

1941 1951 1961 1961 (MA)a

Children , under 5 years (actual)

J3 o au.

,„ „„ Women,

15 4 4

00 e W (1)

OH

o

(2)

19.8 24.8 16.5 22.6

73.0 70.1 70.7 68.3

Children under 5 years (expected)

s

.2

M

JS u g

h

ja O

(3)

eo a W (4)

h (5)

(6)

•3 W (7)

tt. (8)

(9)

7.2 5.0 12.9 9.0

23.9 23.0 17.3 22.0

66.1 67.9 68.4 65.9

10.1 9.0 14.3 12.1

20.4 21.2 14.5 20.2

70.9 70.0 70.8 68.5

8.7 8.8 14.7 11.3

•C

a

Entire Census Metropolitan Area. Note: Cols. (7)-(9) based on ethnic origin of children under five and the relationship between ethnic origin and mother tongue for all males in the area.

If we compare the mother-tongue distribution of small children (Table 3, cols. [l]-[3]) with that of women in the child-bearing ages (Table 3 cols. [4]-[6]), we find that the higher degree of bilingualism among the FMT population has not led to a net switch to English among the children. In all periods, a larger proportion of small children than of women in the child-bearing ages have French as their mother tongue. In 19 Stanley Lieberson, Ethnic Patterns in American Cities (New York: Free Press of Glencoe, 1963), pp. 8-10.

244

Stanley

Lieberson

1951, for example, 67.9 per cent of the women had French as a mother tongue, whereas 70.1 per cent of the children were reported as having French. The position of the English language is somewhat inconsistent, in some decades being slightly better than its rank among women of child-bearing ages would suggest and in other instances poorer (cf. Table 3, cols. [1] and [4]). The erosion of other mother tongues is quite apparent; in each decade their over-all position among small children is weaker than the position among adult women. The net results of our intergenerational analysis indicate that French is not merely holding its own but is actually gaining between generations. Before interpreting this finding, it is necessary to consider the differentials in fertility between language groups - particularly since the French have higher fertility rates than the non-French ethnic component of Montreal.20 As a crude approximation we have determined the mother tongues we would 'expect' among children under five years of age knowing their ethnic origins and the cross-tabulation between mother tongue and origin for males in the metropolitan area. Using Westergaard's method of standardization, we derive the 'expected' mother-tongue distribution of children under five given in columns (7) through (9) of Table 3. Several comments are called for before discussing the results. First, we standardized on the basis of males rather than females because ethnic origin in the case of mixed offspring is traced through the male lineage in the Canadian census enumerations. Second, the cross-tabulation between origin and mother tongue is not available for specific age categories. Consequently, the standardization rates incorporate to some extent the very data examined for small children. Since children under five do not comprise a large segment of the population, we are not too greatly hampered by this departure from the ideal.21 Finally, the available data do not allow for the control of differences in fertility between monolinguals and bilinguals within each ethnic population. This has important implications, which we will discuss shortly. Controlling for fertility differences between ethnic groups, we find the French language has about held its own (Table 3, cols. [2] and [8]). In both 1951 and 1961, the position of French among children is within 0.2 per cent of the figure we would expect on the basis of the ethnic origins of the youngsters. In 1941, the number of FMT children is even greater than what we would expect on the basis of ethnic fertility differentials. 20 The French have higher fertility even with education and income held constant. See Enid Charles, The Changing Size of the Family in Canada (Ottawa: Dominion Bureau of Statistics, 1948), p. 105. 21 Moreover, where there are any trends through time, the child-bearing ages would probably fall close to the mean for each ethnic group.

Bilingualism in Montreal

245

The gain for French in the two most recent decades among children compared to its position among adult women, however, is not due to a net transfer to French among the bilingual parents of Montreal. Rather, it appears largely to reflect the higher fertility rates of ethnic groups with sizable numbers of native French-speakers. By contrast, ethnic fertility differentials are clearly unfavorable to the maintenance of English. In each period the percentage of children we would expect to have English as a mother tongue falls below the percentage of women in child-bearing ages with the mother tongue (cf. Table 3, cols. [4] and [7]). In the last two decades, however, the English language has more than held its own when we consider the lower fertility of ethnic groups with large numbers of EMT adults (cf. Table 3, cols. [1] and [7]). In 1951, for example, taking fertility into account, we would expect 21.2 per cent of children to have English as a mother tongue whereas actually 24.8 per cent were reported as learning this language first. The excess was more than enough to compensate for lower fertility in 1951 and results in the percentage of small children with English exceeding the percentage of women with this language (respectively, 24.8 and 23.0 per cent). The results for 1961 are inconsistent, depending on whether one includes the entire metropolitan area in the analysis. What is clear, however, is that the net intergenerational switching has been favorable to English but that much or all of the gain is wiped out by the lower fertility rates. Although the data relevant to the question of whether fertility differentials aid or handicap other mother tongues are inconsistent (Table 3, cols. [6] and [9]), there is little doubt that other languages in general fail to hold their own. Whether the mother tongues of children (col. [3]) are compared with those of women in the child-bearing ages (col. [6]) or with those expected among children on the basis of ethnic differentials in fertility (col. [9]), the results clearly show a drop in other mother tongues among small children. If anything, the extent of this drop is probably understated, since many small children were either born in a country where English was not spoken or were born in Canada of relatively recent immigrants who had not yet sufficiently mastered one of the official languages to use it for speaking to their children. It is very difficult to estimate fertility differences between the linguistic subpopulations within each ethnic group. Overall, on the basis of indirect evidence, we would be inclined to guess that the bilingual segment has lower fertility rates. This guess is based on the following facts: First, Enid Charles finds an inverse association between fertility and income within each major ethnic population.22 Second, there is 12

Charles, loc. ctt.

246

Stanley Lieberson

some evidence that bilinguals tend to have higher incomes than monolingual within both the French- and the English-speaking populations.23 This would suggest, although by no means conclusively, that the bilingual segments of Montreal's FMT and EMT populations have lower fertility than their monolingual compatriots. Under such circumstances, the exposure-to-risk rates overstate the danger to the mother-tongue populations of bilingualism. The intergenerational language-switching rate is more elusive than the two elements of the model considered thus far. Indeed, we do not have available the basic data needed for determining the switching rates for the bilinguals of each mother-tongue group. We do wish to note, however, that if the languages are maintained between generations in a simple two-language-contact setting with some switching between generations, then if fertility is equal it can be shown that actually the bilinguals of the group with lower exposure to risk are experiencing a higher rate of intergenerational language-switching than is the group with the larger percentage of bilingual parents. It is reasonable to assume that, in a bilingual setting such as Montreal, some parents do in fact pass the second language on as their children's mother tongue. Under such circumstances, if the FMT population is more bilingual than the remainder of the population and if switching is equal in each direction, then it follows that the rate of switching is higher for the bilingual non-EMT population than for the bilingual FMT population. One important consideration in interpreting the maintenance of both English and French through the generations is the fact that men in Montreal are more bilingual than women. This sex difference means that many of the bilingual members of each ethnic group are married to monolingual mates who share only the same mother tongue. Insofar as this occurs, the high incidence of bilingualism is reduced as a contributing force to language transfer, because only one of the mates can use the second language with their children. Sex differences in bilingualism within a largely endogamous population help to maintain the common mother tongue of the mates in the next generation, since this would be the one language that both parents could use with the children. Of course, marriages across mother-tongue lines are considerably more complex in their possibilities. Marriages between EMT and FMT mates will often involve a cross between Catholic and non-Catholic. If the ties of the Roman Catholic church are strong enough so that frequently mates will agree to a Catholic education for prospective is

In a current study of want ads in Montreal newspapers, jobs that require bilinguals tend to be concentrated in categories that offer higher income.

Bilingualism in Montreal

247

offspring, intermarriage will not harm French-language maintenance in Montreal, since the Catholic school system is overwhelmingly Frenchspeaking.24

COHORTS

Thus far our demographic examination has focused on intergenerational dimensions of bilingualism, namely, the language transferred by bilingual parents to their offspring. At this point, we turn to viewing second-language-learning in connection with the life span of residents of Montreal. The terms 'cohort', 'longitudinal', and 'generational' will be used interchangeably to describe the approach employed. As opposed to the more commonly encountered cross-sectional analysis, a longitudinal study allows us to determine true social change since the linguistic behavior of a population is actually traced through time.25 By considering where in the life span second-language-learning tends to occur, we may infer the social conditions leading to bilingualism. There are several serious difficulties involved in tracing the bilingualism of age groups through time. First, we have no way of directly controlling for the foreign-born population in Montreal and cannot readily prevent immigration between decades from affecting the analysis. Similarly, lack of available data prevents the application of controls for emigration and internal migration. The results presented are, therefore, crude cohorts which describe the process of linguistic change by age groups but which do not easily allow for the decomposition of these changes into the contributions that selective migration and mortality may have contributed to the results. With respect to mortality differentials, however, it is unlikely that these would significantly alter the results. We shall have more to say about the influence of immigration later. Shown in Table 4 are the bilingual proportions by age and sex of 24

Only two of the fourteen elementary-school districts of the Montreal Catholic School District are English. See Bureau of Statistics, School Directory, 1963-1964 (Montreal: Montreal Catholic School Commission, n.d.). 25 Among recent examples of the application of cohort analysis to empirical data are Pascal K. Whelpton, "Trends and Differentials in the Spacing of Births", Demography, I (1964), 83-93; Hope T. Eldridge, "A Cohort Approach to the Analysis of Migration Differentials", Demography, I (1964), 212-19; Beverly Duncan, George Sabagh, and Maurice D. Van Arsdol, Jr., "Patterns of City Growth", American Journal of Sociology, LXVII (January, 1962), 418-29; N . B. Ryder, "The Influence of Declining Mortality on Swedish Reproductivity", in Cur rem nesearcn in human Fertility (New York: Milbank Memorial Fund, 1955), pp. 65-81.

248

Stanley Lieberson

the four decades, 1931-61. For 1941-61, the analysis is based upon the combined cities of Montreal-Outremont-Verdun. Since data are not available for Outremont in 1931, the 1941 figures were recomputed for the Montreal-Verdun area alone. Male Cohorts If we assume that the proportion of boys who are bilingual at the age they usually begin school falls somewhere between the proportions for the population under five and five through nine, then we would estimate that less than 10 per cent speak both English and French by the time elementary school is begun. This means that the linguistic heterogeneity of Montreal has a limited effect in introducing bilingualism among boys at an early age. What appears to operate is a steady push in the learning of a second official language during the ages of mass education, some additional gain in bilingualism in the early adult ages, and then a steady decline in bilingualism through the remaining years. Tracing male age groups between 1931 and 1941, for example, we find 4.1 per cent of children under five are bilingual (col. [1]), but ten years later boys ten through fourteen years of age are 22.2 per cent bilingual (col. [2]). Similarly, 18.2 per cent of boys five through nine are bilingual in 1931, but more than half (51.4 per cent) are bilingual ten years later. This result holds up as well for the 1941-51 and 1951-61 comparisons (cols. [3]-[4], [4]-[5]). In addition to the rapid buildup of bilingualism through the teens, there is some increment in the early adult ages. This surge in bilingualism tapers off for each cohort somewhere in the twenties and early thirties and is then followed by an actual net decline among men in the middle and later years of their life. Men thirty-five and older in 1931 all show declines in bilingualism during the following decade (col. [1] and [2]). This interdecade decline starts even earlier in the 1941-51 span, with men twenty-five and over in 1941 showing declines in bilingualism ten years later. Indeed, the age at which the decline in bilingualism begins is even earlier by 1961. As we shall see, the decline in the age at which male bilingualism begins to erode is explainable in terms of immigration. These results suggest that the educational system of Montreal, which requires courses in the official language not used as the language of instruction in the school, meets with success. This success is far from complete, since in most decades about half of the boys fifteen through nineteen are unable to speak both official languages. The decline that follows the school years indicates that economic and occupational forces in combination with internal and international migration fail to fully support or maintain bilingualism among a fair number of males. Once

249

Bilingualism in Montreal

§

vo o -) «

s

Í

H on

o

00 •Q « S 60

i 1

e

CQ S C H N T f f n o i N n ^ f « H^toiov^inviiov)

I

® s a si

i44 4 3

250

Stanley

Lieberson

the supports of education and the early years in the labor force are past, ability in both official languages begins to decline. It is as if we were to examine knowledge of algebra by age cohorts over time. We would expect to find a rapid rise through the school ages, followed by a decline after the educational period, since many find little or no need for algebra in either their occupational or their social worlds. In similar fashion, through a fair part of the adult male span, more bilingual speakers lose their knowledge of the second language than monolinguals acquire a knowledge of the second official language. Female Cohorts Further support for this interpretation comes from examination and comparison of bilingualism among the female age cohorts (Table 4, cols. [6]-[10]). First, we observe that the degree of bilingualism among girls under 15 is very similar to that found for boys of the same ages. With the exception of ten- through fourteen-year-old girls in 1931, the sex differences are well under 1 per cent. This suggests that the experiences of boys and girls are very similar at these ages so that the net influences of home language, neighbors, mass media, playmates, and early years of formal second-language instruction in the schools are nearly identical. Beginning with the late teens, at the age when formal education ends for many and participation in the labor force begins, we find increasing differences between the sexes in their bilingualism. By the early twenties, the sex differences in bilingualism are considerable. In 1931, for example, 43.4 and 41.4 per cent of the males and females ten through fourteen are bilingual; in 1941, 67.1 and 51.5 per cent, respectively, of males and females are bilingual. Percentage-point differences of about 15 are also found for 1951 and 1961 between males and females in the twenty through twenty-four age category. The influence of male participation in the labor force and female withdrawal into the home and child-rearing can also be seen in the points at which a net decline in bilingualism occurs as well as in the lower increases in bilingualism among women. In all three interdecade comparisons we find that declines in the bilingual percentage occur at earlier ages for women than for men. This is most striking for the 1931-41 comparisons. Not until we reach men thirty-five through fortyfour in 1931 do we find a cohort that declines in the succeeding ten years (62.2 in 1931 and 60.3 in 1941). By contrast, females fifteen through nineteen and older decline in bilingualism by 1941. 26 These differentials are compatible with the contention that the main supports " Here and elsewhere, graphs have been employed to compare cohorts over time that are in different-size age categories in the two periods.

Bilingualism in Montreal

251

of bilingualism are school and occupational systems, although the reader should be reminded that the evidence is hardly incontrovertible. Immigration Unfortunately, limitations in the available data preclude an analysis or weighting of the specific contributions made by such diverse institutional and demographic forces as education, occupational demands, internal migration, and immigration in influencing the bilingualism of cohorts. It is possible, however, to gauge through indirect means the influence of immigration in interdecade changes in bilingualism. There is little need to consider the impact of recent immigration on bilingualism among the younger age groups, since the rapidly rising bilingualism rates run counter to any expectation we might have on the basis of the low bilingualism of recent immigrants. The decade between 1931 and 1941 affords the easiest test, since there was very little net immigration to Montreal-Verdun. For some age categories in 1941, less than 1 per cent of the population were foreign-born who had arrived in the preceding decade. Recent immigrants (since the 1931 census) were strongest among the thirty-five through forty-four-year-old population, comprising 2.2 and 2.8 per cent, respectively, of males and females in 1941. Even if no immigrants in the preceding decade had become bilingual - which is far from the case - their number would have been too small to account for the interdecade declines in bilingualism that we observe in 1941 among men who were thirty-five and over in 1931 and among women who were fifteen and older. We can conclude that the inferences based on cohort changes between 1931 and 1941 are substantially unaltered by immigration during the decade. Heavier immigration since World War II definitely influences the cohort patterns described earlier. Among both men and women in the age category thirty-five through forty-four in 1961, about 20 per cent were immigrants who had arrived in Canada during the preceding ten years. For both 1951 and 1961 we have estimated the age-specific bilingual percentages, excluding for each year immigrants first arriving in the preceding decade ("recent immigrants").27 We shall examine these rates in order to determine whether the inferences made earlier are valid after the effect of immigration is eliminated. The bilingual percentage for the population, excluding recent immigrants, was obtained through indirect standardization. Using the component of recent immigrants in each group as weights, the ratio of actual and standardized bilingual percentage was applied to each specific age category to estimate the bilingual percentage of the population who had lived in Canada in the preceding decade.

252

Stanley Lieberson

The results, shown in Table 5, indicate that no substantial changes need be made in the cohort analysis presented earlier. In 1951 we find that the decline in bilingualism begins among men who were forty-five and older in 1941 (cols. [1] and [2]). Among women, the decline now begins among those twenty and older in 1941 (cols. [5] and [6]). In 1961, the decline for men starts with those who were thirty-five and over ten years earlier (cols. [3] and [4]). The results for women are somewhat inconsistent; there is a decline for women twenty through twenty-four in 1951 but a rise between 1951 and 1961 among women twenty-five through thirty-four at the beginning of the period. However, a steady decline is observed among women thirty-five and over in 1951. TABLE 5

Percentage Bilingual for Selected Age Groups, by Sex, Excluding Recent Immigrants, Montreal-Outremont-Verdun, 1941-61 Females

Males Age

10-14 15-19 20-24 25-34 35-44 45-54 55-64 65-69 70+ a b

1941

1951a

1951

1961b

1941

1951a

1951

1961b

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

22.6. 51.7,0 67.20 ^66.8 6 8 . 8 ^ 72.2 63.7^. ^ 7 1 . 2 6 0 . 3 ^ ""64.3 53.8. ""58.0 49.6^ > 5 0 . 1

S2.5

20.5. 50.6/" 64.90 ^64.7 6 8 . 8 ^ 67.6 68.1-^ ""71.6 6 2 . 7 \ ""67.5 5 7 . 3 ^ """59.0 49.7-C > 5 3 . 0 44.5

22.3. 43.5,0 51.7-0 N9.5 4 8 . 1 ^ 50.0 41.2^ ^ 4 6 . 4 36.0 ^ ""38.1 31.6. ""31.1 28.50 >26.7 "^23.7

20.1. 44.50 48.20: N9.1 4 7 . 8 ^ 46.0 4 5 . 2 ^ ""49.0 3 7 . 4 ^ ""44.7 30.8 ""35.7 26.5 0 > 2 9 . 3 24.9

Excluding immigrants since 1941 Excluding immigrants since 1951

In brief, the existence of unlearning is still found among the population who were not immigrants to Canada in the intervening decades. The process of immigration tends to lower the age at which bilingualism starts to decline. In other words, the effect of large-scale immigration on cohorts is the same as if bilingualism were to start its decline at earlier ages. The decline among men is more of a middle-age phenomenon, while women tend to begin their decline in bilingualism at earlier ages. This result is, of course, compatible with the earlier interpretation of the institutional forces that maintain bilingualism in Montreal.

Bilingualism in Montreal

253

COMMENT

Racial and ethnic contact is frequently accompanied by the confrontation of peoples who speak different tongues. In the United States, with a few exceptions, we have been able to take more or less for granted the linguistic outcome of contacts with non-English-speaking peoples. Although the experience of the United States is not unique, there are far more complex linguistic situations in many parts of the world. Indeed, Canada is relatively simple compared to such nations as Nigeria, India, or South Africa. The examination of linguistic trends in Montreal indicates a process quite contrary to the view of the city as a great mixer and melting pot of diverse cultures. Unlike most American cities, where populations with diverse linguistic origins have moved toward a monolingual status in a matter of a few generations, Montreal has maintained an equilibrium. This equilibrium is similar to that of Switzerland in the sense that it is based on dynamic demographic forces that tend to counterbalance each other.28 It is a precarious equilibrium, since the exposure to risk is greater among the French ethnic population than among the British. If we make the assumption that most people will not really master a second tongue unless it is learned at a relatively early age, then secondlanguage-learning is less of a threat to the mother tongue than might otherwise be the case, since English and French languages are able to hold their own as the first language of children, and since much of the bilingualism in Canada does not occur in the very early ages or as a result of informal social contacts. Keyfitz has pointed to the handicaps that bilingual French Canadians face in the upper echelons of management or in white-collar office settings because of their lack of fluency or ease in English, which is often the language used.29 The constant replacement of a population through the introduction of new generations means that social change may occur without any individual changes.30 However, we should not lose sight of the fact that generations may be instruments of conservation. Insofar as new members are socialized by older members, then it is equally possible for the 28

For an excellent analysis of the demographic factors, see Kurt Mayer, "Cultural Pluralism and Linguistic Equilibrium in Switzerland", reprinted in Joseph J. Spengler and Otis Dudley Duncan (eds.), Demographic Analysis (Glencoe. 111.: Free Press, 1956), pp. 478-83. 29 Nathan Keyfitz, "Canadians and Canadiens", Queen's Quarterly, LXX (Summer, 1963), 163-82. 30 Karl Mannheim, Essays on the Sociology of Knowledge (New York: Oxford University Press, 1953), pp. 276-320.

254

Stanley Lieberson

system to maintain itself over time indefinitely as the newcomers who gradually replace the older members are socialized.31 In the case of Montreal, we witness a social process of second-language-learning in each generation that does not lead to change through the years in the linguistic structure of Montreal. From the perspective outlined earlier, transfer from parents to children and the stage in the life span at which bilingualism occurs are both critical for understanding the outcome of language contact.

** See George Simmel, "Social Interaction as the Definition of the Group in Time and Space", in Robert E. Park and Ernest W. Burgess, Introduction to the Science of Sociology (Chicago: University of Chicago Press, 1921), pp. 348-56.

Robert L. Cooper and Susan Carpenter LINGUISTIC DIVERSITY IN THE ETHIOPIAN MARKET*

Marketing requires bilingualism in Ethiopia in view of the fact that different products are disproportionately produced, sold and/or bought by quite different linguistic groups. A method is described for determining the frequencies with which different languages are used in various markets as well as in various parts of markets. Languages of wider communication - whether indigenous or foreign - remain highly useful in Ethiopian markets, particularly in connection with products that have general appeal or that are associated with particular educational or economic statuses.

Like most of Africa, Ethiopia is linguistically diverse: its population of approximately 24,000,000 speak about 95 languages.1 One arena for the observation and description of Ethiopia's linguistic diversity is the marketplace, where buyers and sellers who speak different first languages come together to trade. This paper summarizes the results obtained from a survey of the languages used in 23 Ethiopian markets, conducted as a part of a larger enterprise, the Language Survey of Ethiopia.2 * Prepared for the Joint Annual Meeting of the African Studies Association of the United States and the Committee on African Studies in Canada, October 15-18, 1969, Montreal. From: Journal of African Languages, 1969, 8, 160-168. Reprinted with permission. 1 The Central Statistical Office of the Ethiopian Government estimated Ethiopia's 1967 population as 23,667,400 (Ethiopia: Statistical Abstract, 1967 and 1968. Addis Ababa: Central Statistical Office). Marvin L. Bender has gathered word lists for 95 Ethiopian languages as part of the Language Survey of Ethiopia. 2 The Language Survey of Ethiopia is one of five country studies conducted by the Survey of Language Use and Language Teaching in Eastern Africa, a project supported by the Ford Foundation. The first author was a member of the Ethiopian survey from August 1968 through August 1969.

256

Robert L. Cooper and Susan Carpenter

PROCEDURE

All of the 23 markets which were observed were in towns. These towns were Addis Ababa, the capital, located in the center of the country; Asmara and Keren, in the north; Alemaya, Dire Dawa, and Harar, in the east; and Jimma and Soddo in the south. None of these towns is in the north central plateau, the ancestral homeland of the Amharas, whose language, Amharic, is the country's official language. Asmara and Keren are in Eritrea, where the most important language is Tigrinya. Soddo is the capital of a Wollamo-speaking district. The other towns are located in areas where the most commonly spoken language is Galla. All of these towns contain administrative units of the central government. Thus each town contains a core of Amharic speakers due to the schools and other governmental agencies centered there. The procedure followed in each town can be described as follows. First, enumerators were enlisted. These were local secondary school students who were familiar with the languages spoken locally. That is to say, they could recognize if not speak all of the local languages. All were fluent in Amharic, which is the language of instruction in all Ethiopian public schools through the sixth grade (English is the medium thereafter). The number of enumerators employed depended on the size of the markets to be surveyed. The larger markets were covered by 12 students organized into two teams. The smaller markets were surveyed by a single team of six enumerators. Each team was supervised by a local secondary school teacher who knew both the enumerators and the town. Second, for each of the markets to be studied, a list was drawn up of the most important commodities and services sold, e.g., red peppers, chickens, barley, tailoring. For the first few markets, this list was determined by means of a preliminary survey. The items sold were identified and the vendors of each item were counted. The items with the largest number of vendors were considered to be the most important. Inasmuch as the lists compiled by these preliminary surveys appeared to be similar and inasmuch as these preliminary surveys consumed an entire market day, the lists for the subsequent markets were compiled by submitting a standard list of commodities and services, representing an amalgam of the previous lists, to local informants who were asked to add items which were important locally but which were not on the standard list and conversely to delete items which were of small importance locally. The third step consisted in training the enumerators and their supervisors. The training session required about three hours and was held on the morning of the day on which the first market was to be sur-

Linguistic Diversity in the Ethiopian Market

257

veyed. During this session, the procedures were explained and the enumerators practiced filling out the tallies which they were to use that afternoon. With only a few exceptions, the day of the week which was chosen for the survey was the 'market day', that is the day of peak market activity. For a given market, the survey would begin at about one p.m. and continue until five or six p.m.3 The language survey itself was carried out as follows. Each enumerator was given a tally sheet, small enough to be held in the palm of his hand. On each sheet was listed the names of a set of languages, including Amharic and the most commonly spoken local languages. He was then assigned, from the list which had been prepared, a commodity or service, e.g., barley, and was asked to walk slowly through the area in which that item was sold. His task was to note the number of transactions carried on in that item in each of the language on his tally. (There were spaces on his tally for him to indicate the names of languages other than those on the list.) A transaction was defined as an interaction between buyer and seller, i.e., a buyer and seller talking to each other for the purpose of making a purchase or sale. The enumerator was also asked to note the number of bilingual transactions observed (transactions in which the buyer and seller did not speak the same language to each other). The enumerator was instructed to wait until he left the assigned area before writing the numbers on his tally. If the area was a large one or if there were too many transactions for him to keep them all in mind, he would walk through part of the area, leave it to tally his observations, then return to the area continuing at the place at which he had stopped, and he would leave and reenter as many times as was necessary to make a complete circuit. After the enumerator had completed his tally, he returned to his supervisor, who then gave him another tally and assigned him to another commodity. Although the number of commodities varied from market to market, typically about 30 items were covered by each team. Every commodity was assigned to every enumerator in the team but the assignments were made in such a way that no item was visited by more than one enumerator at a time. In this fashion hundreds of transactions were observed in the smaller markets and thousands in the larger ones. The numbers of markets and transactions observed in each town are presented in table 1. 3

The Addis Ababa Markato was an exception to the rule of one-day surveys. Due to the Markato's great size and complexity, three days were required to carry out the survey, one day for the open market and two days for the shops of the Markato. Before the language survey of the Markato was carried out, rough maps were made indicating the locations of the items to be covered.

258

Robert L. Cooper and Susan

Carpenter

TABLE 1 Number of Markets and Transactions in Eight Ethiopian Towns Town Addis Ababa Alemaya Asmara Dire Dawa Harar Jimma Keren Soddo

Observed

Number of markets

Number of transactions

4 1 7 4 4 1 1 1

11,088 1,337 11,564 1,792 5,912 4,151 1,010 2,233

TABLE 2 Language of Market Transactions in Eight Ethiopian

Towns

80

Soddo

8

Keren

5

Jimma

Total

13 14

11 3 2

Harar

64

Dire Dawa

Amharic Adari Arabic Bilen English Galla Gurage Italian Kaffa Kullo Saho Somali Tigre Tigrinya Wollamo Other Bilingual

Alemaya

Addis Ababa

Language

Asmara

Percentage of Transactions Town

36 4 6

25 27 1

33

2

17

37

23 1

53 6

5

1

1

1 5 14

10

12 17 1 2

6

1

6

2 67

24 22

1 1

3

1 11

1 2

1 12

1 1

2 17

75 1 1

99

100

100

100

100

100

100

100

Linguistic Diversity in the Ethiopian Market

259

RESULTS

For each market and each town the proportion of all transactions observed in each language was computed. The proportions for each town are summarized in table 2. Some notion of Ethiopia's linguistic diversity can be obtained from this table, which requires the names of 15 languages to describe language usage in the markets of eight towns. As has been pointed out, however, the country which is linguistically diverse at the national level may be homogeneous at a smaller unit of analysis (Criper and Ladefoged 1971). This is clearly not the case for the towns which were studied. In each town, transactions in several languages were observed. Even in the most linguistically homogeneous market, Alemaya, where 80% of the transactions were observed in Galla, transactions were also observed in four other languages. Between-Town Variation Although each market town proved to be linguistically diverse, the pattern of diversity varied, the proportion of transactions in a given language changing from town to town. The proportion in Galla, for example, ranged from 80% in Alemaya to zero in Asmara and Keren,4 and the proportion in Amharic ranged from 64% in Addis Ababa to 2% in Keren. In general, the direction of these differences could be expected on the basis of differences in the geographical distribution of the languages involved. Within-Town Variation To a lesser extent, variation was also observed between the markets of a single town. Tables 3-6 summarize the observations obtained for each of the markets surveyed in Addis Ababa, Asmara, Dire Dawa, and Harar. In Asmara's 'Italian' market (table 4), for example, 46% of the observed transactions were in Italian, whereas the proportion of transactions observed in Italian in the other Asmara markets did not exceed 5%. Similarly, in the one Dire Dawa market (table 5) which has a relatively high proportion of Somali vendors and customers, the Afatesa 4

It might be argued that the enumerators would not have been able to recognize certain languages not widely spoken in or near their towns, e.g., Galla in Asmara. There was, however, a space on the tally for enumerators to indicate the number of transactions observed in languages which were 'unknown'. In each market, the number of transactions in languages which the enumerators could not identify accounted for less than one per cent of the observed transactions.

260

Robert L. Cooper and Susan Carpenter TABLE 3

Language of Market Transactions: Addis Ababa Percentage of Transactions Market Markato shops (N=5532)

Markato open mkt. (N=4138)

Kirkos Kebele (local) (N.—1000)

Emmanuel (local) (N=418)

Total ( N = 11088)

Amharic Galla Gurage Tigrinya Other Bilingual

73 8 11 6 2 1

55 19 18 6 1 2

52 21 14 11 1 2

61 9 25 0 0 4

64 13 14 6 1 1

Total

101

101

101

99

99

TABLE 4

Language of Market Transactions: Asmara Percentage of Transactions Market

Language

Spice (N=1096)

Italian (N = 895)

Hadish Adi (local) (N=1945)

Adega Arbi (local)

Amharic Arabic Italian Saho Tigrinya Tigre Other Bilingual

6 15 5 3 58 1 1 11

6 6 0 0 72 1 1 15

3 2 1 3 77 6 1 6

5 9 2 1 72 2 1 9

4 3 46 0 25 6 2 15

6 6 0 1 74 1 0 14

2 3 0 0 86 1 0 9

5 8 5 1 67 2 1 11

Total

100

101

99

101

101

102

101

100

Grain + vegetable (N —3052)

Sheep + goi (N=1339)

a

h "> W oo >

274

Roxana Ma and Eleanor Herasimchuk

The co-occurring variables in F2 are almost identical to the group which co-occur in F l , the difference being in their phonetic realizations and in the fact that the F2 items come from only the conversational speech contexts B and A. In F2, the variants show that interference patterns predominate and that speakers are likely to use UH-3 [o] and EH-3 [a] in such words as lunch and language in their conversational English. They also pronounce final r and final t with their full standard values. It is interesting to note that Item 138 or word-final r correlates highly with this factor. As we suggested earlier in our separate discussion of PRE (R) variable, R-l pronunciation could possibly be interpreted as an interference equivalent for Spanish-dominated speakers, so that it is thus quite natural to find it correlating here so highly with the vowel interference variants. Factor analysis inter-correlates common variation over the total corpus of our speakers. This means that the data in F l and F2 could presumably be uttered by some of the same people. From this, we can tentatively conclude that there are two conversational styles in PRE. In one style (Fl), the variables (EH), (UH), (R) and (T) are realized as the set of sounds [A a; 0 ?], respectively, and in the other style (F2), as the set [ o a r t], respectively. A sentence such as "he bought another fancy car" can have these two ranging pronunciations: [hi bo? SIIASS feensi ka] or [hi bot anoSar fansi ka r ]. The most characteristic predictors of this second style are the variables (EH) and (UH), Items 151 and 153, respectively, which are the very same variables which are the best predictors of the first style; see F l . Factor 3 can be interpreted as yet another English conversational style, which we are calling 'Nonstandard English'. Out of 18 items, only Factor 3. Nonstandard English Item #

Speech Context

81 91 127 152 154 165 166 179

PRE-C

5 This item AY-1 as a separate item, on this factor. The notwithstanding, is not have as high a

PRE-B PRE-A

»

Variant AY-1 CC-reduced OH-1 EH-1 OH-1 R#-0 VN-0 CC-reduced

Loading .38 5 .71 .65 .87 .77 .80 .80 .59

is included here on Factor 3 despite its low loading because, it correlated highly ( > .60) with the other individual items very fact that it showed up on the factor at all, low loading evidence that it correctly "belongs" to this factor F3. It did loading on any other factor.

Speech Styles in Puerto Rican Bilingual Speakers

275

three were Spanish and had relatively low factor loadings. The most characteristic English items are given above. By looking at all the variants in this factor, we can see that they are quite representative of the nonstandard lower-class Negro speech of New York City investigated by Labov, Cohen, and Robins (1965). In vowels, speakers who use a very short lower front vowel for (AY), Item 81, will also use higher closed vowels for (EH) and (OH), Items 127, 152, and 154, respectively. The vowel qualities in a sentence such as "I had a black dog" will be realized most likely as [a hed a bleak dosg]. In consonants, these speakers will also drop word-final r, reduce their final consonant clusters, as well as nasalize their vowels. This speech style of English is one of the most accessible to our PR population since they live in the same neighborhoods as do Negro speakers and have social and hence linguistic contacts with them. The results of these interaction and acculturation processes are becoming apparent in the speech habits of some of our PRE speakers, as demonstrated by F3. The best predictor for this English speech style is Item 152, an [e] or [es] pronunciation of the variable (EH). Factor 4 might best be called 'Spanish and English Ellipsis' and presents strong evidence for demonstrating that the same socio-linguistic process can underlie the speech production of both languages among bilingual speakers. In the first place, the 24 items were almost evenly divided between Spanish and English, covering three styles of each. Secondly, the seven most characteristic items from each language involve realizations in which phonetic distinctions are reduced by dropping them entirely, most generally at syllable and word-final positions, Factor 4. Spanish and English Ellipsis Item # 23 26 38 46 53 54 61 67 98 107 109 113 119 162 169

Speech Context PRS-WN PRS-B >>

PRS-A r> »»

„ »

PRE-C PRE-WN

„

>>

PRE-A »

Variant S#-0 S#-0 VN-0 SCO RC-3 R#-0 VN-0 S#-0 CC-non-reduced AY-2 AY-3 Ng-1 TS-reduced RC-0 T-0

Loading .73 .74 .50 .72 .61 .45 .47 .73 -.35 .51 -.59 .66 .72 .75 .78

276

Roxana Ma and Eleanor Herasimchuk

a linguistic process known as ellipsis or apocopation. The most characteristic items are given above. Looking at the Spanish items first, it is immediately noticeable that all variants here have a zero code for the (S) and (R) variables. This means that these sounds are phonetically realized as [0]. In Item 53, the subvariable (RC) is coded as R-3, the assimilation variant, which means that r takes on the same phonetic quality as the consonant following it. In both zero and assimilation variants, the phonetic process is the same, namely one in which the phonological distinctiveness of the variable is lost or in some way attenuated. There is also evidence in PRS of an assimilating (S) variant, although this was not included as an independent variant in our study of variable (S) due to its relatively low frequency of occurrence in our speech samples (where it was counted as S-0). In variable (VN), Items 38 and 61, we have yet another type of phonetic weakening, the loss of syllable-final nasal consonants accompanied by nasalization of the preceding vowel. The convergence of all these Spanish variants into one co-occurrence cluster enables us to define rather precisely a prevalent colloquial speech style of PRS which Puerto Ricans themselves have characterized as 'eating the (ends of) words'. It is the most casual conversational style (note that most of the items are from PRE-A) and it is certain to be the one most frequently used in the everyday social interaction of Puerto Ricans with each other. Ellipsis as the most common marker of casual style has been discussed by Joos (1959). We now turn to a discussion of the English variants in this factor.® The vowel values in Items 107 and 109 are complements of each other, since they carry opposite values. The negative Item 109 is the long upglide in [ai] and its negative value means that it is characteristically not used for speakers who use all the positively loaded items in the factor. Instead, they use Item 107, which represents the short upglide [ai]. For consonant clusters, the same process of shortening or reduction applies. Item 98 is negatively loaded (although low) with respect to Item 119, meaning that speakers who reduce (TS) are not as likely to have co-occurring non-reduced (CC), but presumably reduce these as well. Other single final consonants are also dropped, final t and syllable-final r, Items 169 and 162, respectively. Finally, Item 113 is the pronunciation of final unstressed -ing in words like "something, anything" as "somethin"' and "anythin"', pronunciations which, in • We must acknowledge beforehand, however, that while the English data is interesting and certainly relatable to the evident apocope in the co-occurring Spanish style, we are less certain of its statistical validity, since some of these items had much smaller frequency scores than did the Spanish items.

Speech Styles in Puerto Rican Bilingual Speakers

277

some psychological if not linguistic sense, constitute a shortening of these words. The two co-occurring styles of Spanish and English are comparable in that both contain speech variants which either reduce or delete underlying phonological distinctions. The total results of this process of apocopation are speech styles which are abbreviated, almost codelike styles well suited for conversational discourse. Factor 5 is called 'Standard Conversational Spanish'. Of the 19 items which showed up in this factor, more than half were from all 5 style contexts of Spanish and carried the highest loadings in the factor. The remaining English items were both too heterogeneous and carried lower loadings to contribute to an overall characterization of English, and have therefore not been used in interpreting F5. Factor 5. Standard Conversational Spanish Item # 11 12 20 43 47 50 52 55 63

Speech Context PRS-D PRS-C PRS-WN PRS-B PRS-A »

»

Variant

Loading

R#-l RR-3 RR-1 S#-2 SC-1 S#-l RC-1 R#-l RR-3

.51 -.44 .47 -.61 .77 .55 .70 .56 -.31

It is immediately evident that all the variants coded with the number 1 are positively loaded, in contrast to variants with other numerical codes which are negatively loaded. This indicates that a high usage of one group of sounds automatically excludes any high usage of the other group of sounds. Since code 1 is the code for all the standard variants of our Spanish variables, and since it occurs throughout all of the speech contexts, we conclude that its speakers are the most conscious in adhering to correct or standard pronunciation norms in their overall speech patterns. Variable (R) is most consistently pronounced with the flap [ r ] in both syllable and word-final position. For the (S) variable, S-2 or [h] pronunciation in Item 43 is negatively related to S-l, indicating its relative absence in the speech of i-pronouncers. Item 47, or SC-1, appears to be most characteristic of F5. Finally, Items 12 and 63, or RR-3, the velar variant, are negatively loaded in relation to RR-1, Item 20. The relatively low loadings in all three (RR) items indicates that they are

278

Roxana Ma and Eleanor Herasimchuk

not as typical of this 'Standard Spanish' factor in the same sense as are the R-l and S-l variants. Thus we might conclude that velarization of (RR) is never completely absent from nor wholly present in a given speech type, regardless of context, although individual speakers may fail to produce it altogether. We can assume that not only some speakers who may have a standard pronunciation for most of the Spanish sounds, but also speakers who do not, can be alike in that velar (RR) cannot be reliably predicted in their total speech styles. F5 confirms the observation of many investigators of PRS (see particularly Navarro-Tomás, 1948 and Alonso, 1953) that the velar [X] of (RR) is most peculiar and characteristic to the PRS dialect and randomly affects its speakers regardless of their social status, education, or geographic origin. Factor 7 appears to deal with a formal speaking style in both Spanish and English and hence is called 'Formal Spanish and English'. Half of the 14 items in the factor came from the speech context WN in both languages, where they were among the most highly loaded in the factor. The most characteristic items on this factor are: Factor 7. Formal Spanish and English Item # 17 18 100 104 112 120

Speech Context PRS-WN >>

PRE-WN

»

Variant RC-1 Vdo-1 UH-2 OH-2 VN-1 TS-non-reduced

Loading .60 .54 .52 .65 .85 .52

In Spanish, the standard pronunciation of preconsonantal r co-occurs with the standard pronunciation of intervocalic d. These two, in turn, co-occur with standard English vowels (Items 100 and 104), nonnasalized vowels, and no reduction of certain consonant clusters. Taken as a whole, F7 represents a careful pronunciation norm. F7 can best be interpreted in the following way. In a very formal speaking contexts such as the recitation of a list of isolated words, speakers have an opportunity to concentrate on their speech output via the feedback mechanism of self-monitoring. Under these circumstances, standard variants are automatically favored to occur, all other things being equal. F7 shows the co-occurrence of several standard variants in both Spanish and English word-naming tasks, indicating that, regardless of language, bilingual speakers interpret the 'same' social situation by like behavior. As with F4, we again see that such similar socio-

Speech Styles in Puerto Rican Bilingual Speakers

279

linguistic behavior is a manifestation of one sociolinguistic process which cuts across two languages. We assert that this is only possible in speakers for whom bilingual interaction is a stable, functionally-distributed intra-group phenomenon. We can conclude our discussion of the various styles or language varieties in PR bilingualism by a summary of the kinds of information and insights gained from our factor analysis. It has enabled us to define 'style' rather precisely as a co-occurrence cluster of variants. The definitions given in the six factors or styles were both statistically valid and linguistically unified. It has also given us a method for future measurement of bilingual dominance or usage of one language relative to another, as well as for defining the range of stylistic variation in each. Finally, it has shown that a cross-language examination of co-variation can reveal underlying sociolinguistic processes common to both languages.

DEMOGRAPHIC CHARACTERISTICS OF LINGUISTIC GROUPS

Having demonstrated the usefulness of factor analysis for arriving at a precise and reliable definition of speech styles in Puerto Rican bilinguals, we now turn to the problem of finding out which of our speakers use which of these styles and whether those speakers who behave linguistically alike also share demographic and global linguistic characteristics. In other words, we are interested in showing what are the demographic correlates of linguistic groups (those speakers who have the most similar linguistic behavior) and what are some of the most striking phonological factors and variables on which these groups differ. It is hoped that answers to questions such as these will add a dimension of social reality to our analysis and enrich our understanding of some of the phonological features of PRS and PRE. The statistical method known as Q group analysis was utilized in order to establish our linguistic groups. This type of analysis yields groups of speakers which, on the one hand, are maximally alike in their linguistic variation and, on the other hand, are maximally different in their linguistic behavior from other groups of speakers. Actually, Q group analysis is also a type of factor analysis. The ordinary factor analysis (also known as R analysis) is based on the intercorrelation of behavior and was used in defining our bilingual speech styles by disclosing what clusters of phonological variants (or factors, as we called them) were maximally independent of other clusters of variants at the same time that each cluster itself was composed of maximally interdependent variants. Thus our R analysis is based upon the intercorrela-

280

Roxana Ma and Eleanor Herasimchuk

tion of phonological variants whereas our Q analysis is based upon the intercorrelation of our speakers using these variants. Our Q group analysis yielded four clusters of speakers (or linguistic groups) which are known as Ql, Q2, Q3, and Q4. For each of these Q groups, cross-tabulations were obtained with six demographic variables (age, sex, birthplace, educational level, occupation, and years in the continental United States) and four global linguistic variables (scale of Spanishness, reading ability, Spanish repertoire range and English repertoire range). The demographic data was obtained from the sociolinguistic census reported in Chapter III of the Final Report (Fishman, Cooper, Ma, et el. 1968). The global sociolinguistic data was derived from ratings made after the scoring and recording of individual phonological variables had been completed for all of our subjects.7 Table 1 indicates the percent of distribution within each Q group across each of the demographic and sociolinguistic parameters mentioned above. Ql was found to be composed of seven people, mostly young adult females, all of whom were educated in Puerto Rico, with only two having received fewer than seven years of schooling. All but one person in this group originated in the coastal area of Puerto Rico or San Juan; the exception was an adolescent who left her upland town recently enough to have repeated some of her schooling in a New Jersey junior high school. Virtually the entire group reads both Spanish and English fluently, and the group can be characterized as bilingual, with moderate to heavy phonological interference from Spanish in their English, and a flexibility of styles in Spanish. The twelve people falling into Q2 were evenly divided between males and females, showing greater than proportional representation for males in this 'most-English-fluent' group. A substantial proportion of the group was born in the United States (41%), while another one-third 7 Global evaluations were made of the performance of each respondent in regard to command of English phonology (Scale of Spanishness) and demonstrated range of variations in both languages. The Spanishness scale evaluation was made by inspecting the English transcripts for high incidence of interference vowels or the signs of Spanish phonology, such as flapped [r]. Evaluation of repertoire ranges was made by an impressionistic comparison in both languages of speaking versus reading contexts. Those using heavy proportions of nonstandard items in reading were considered one-style speakers, unless spoken style showed even greater informality, such as widespread apocope. Those who used secondary or zero values in conversation but standard values in reading were considered two- or three-style speakers according to the proportional variation between elicitation contexts or the presence of an intermediate spoken style for interview questions. Reading ability was rated from performance on word reading and paragraph reading tasks. See Appendix 10.4 of the final Report for a complete list of all these rating scales.

Speech Styles in Puerto Rican Bilingual Speakers

281

TABLE 1

Composition of Q Groups by Demographic and Global Linguistic Characteristics

Sex 1. M. 2. F. Birthplace 1. U.S. 2. San Juan 3. Coastal and lowland 4. Highland Age 1. 2. 3. 4. 5. 6.

13-18 19-24 25-34 35-44 45-54 55-64

Education 1. 6 th gr. or less PR 2. 7th-9th gr. PR 3. 10th-12th gr. PR 4. 6th gr. or less U.S. 5. 7th-9th gr. U.S. 6. 10th-12th gr. U.S. 7. College (U.S.) U.S. 1. 2. 3. 4. 5.

Residence 5 yrs. or less 6-10 years 11-20 years 20+ U.S. born

Occupation 1. Operative, service, laborer, welfare 2. Craft, foreman, blue collar 3. Self-empl., white collar, clerk

* •

Ql(n=7)

Q2(n=12)

Q3 ( n = 9 )

Q4(n=17)

28.57 71.43

50.00 50.00

44.44 55.56

29.41 70.59

00.00 14.29 71.45 * 14.29

41.67 8.33 33.33 *16.67

11.11 22.22 22.22 44.44

00.00 5.88 41.18 52.92

14.29 28.57 28.57 14.29 14.29 00.00

75.00 16.67 00.00 8.33 00.00 00.00

22.22 11.11 33.33 22.22 11.11 00.00

5.88 5.88 41.18 17.65 23.53 5.88

28.57 28.57 28.57 00.00 •14.29 00.00 00.00

8.33 00.00 8.33 25.00 41.67 00.00 16.67

22.22 33.33 11.11 11.11 11.11 11.11 00.00

47.06 35.29 17.65 00.00 00.00 00.00 00.00

14.29 28.57 42.86 14.29 00.00

8.33 16.67 41.67 00.00 33.33

22.22 33.33 33.33 00.00 11.11

17.64 23.53 58.82 00.00 00.00

14.29

16.67

55.56

58.82

14.29

8.33

11.11

11.76

14.29

00.00

00.00

00.00

Highland-born young people who came to U.S. at ages 10-14. 9 years ed. PR with two years repeated in U.S.

282

Roxana Ma and Eleanor Herasimchuk TABLE 1 (continued)

4. Professional, mngr., official, college stu. 5. Housewife 6. Unemployed minor (16+ non-stu.) 7. Students Scale of Spanishness 0. Span, monolingual 1. Heavy phon-synt. interf. S-E 2. Moderate phon-synt. interf. S-E 3. Slight phon-synt. interf. S-E 4. Maximal differentiation 5. Slight two-way interference 6. Slight interf. E-S Reading Ability 0. Neither S nor E 1. S n o t E 2. Both, E w. difficulty 3. Both fluently 4. Both, S w. difficulty 5. E n o t S 6. Unknown Repertoire Range - English 0. Virtual Spanish monolingual 1. One-style, limited fluency 2. One-style, informal fluent 3. Two-style shifters 4. Maximally fluent in English Repertoire Range - Spanish 1. One-style, no shifting 2. Two styles 3. Three styles

Q1 ( n = 7 )

Q2(n=12)

Q3 ( n = 9 )

Q4(n=l

00.00 42.86

16.67 8.33

11.11 11.11

00.00 29.41

00.00 00.00

16.67 33.33

00.00 11.11

00.00 00.00

14.29

00.00

22.22

23.53

28.57

00.00

44.44

64.71

57.14

8.33

11.11

11.76

00.00 00.00

33.33 16.67

00.00 11.11

00.00 00.00

00.00 00.00

25.00 16.67

00.00 11.11

00.00 00.00

00.00 14.29 00.00 85.71 00.00 00.00 00.00

00.00 00.00 00.00 50.00 16.67 33.33 00.00

22.22 22.22 11.11 22.22 11.11 00.00 11.11

17.65 29.41 23.53 17.65 5.88 00.00 5.88

14.29

00.00

22.22

41.17

14.29

00.00

22.22

35.29

42.86 28.57

41.67 50.00

33.33 22.22

17.65 5.88

00.00

8.33

00.00

00.00

00.00 57.14 42.86

50.00 16.67 33.33

44.44 55.56 00.00

11.76 47.06 41.18

Speech Styles in Puerto Rican Bilingual Speakers

283

are natives of coastal Puerto Rico. All but two of these respondents have received some U.S. schooling, four being ongoing students. This is in keeping with the fact that Q2 is the youngest of the groups; 75% of its members are under 18 years of age. Half of them read Spanish and English fluently, with the others showing difficulty or inability in Spanish reading. Only one respondent has as much as moderate interference from Spanish into English, with the rest distributed toward the English end of the Spanishness scale. While 50% command only one style in Spanish, 58% command two or more styles in English. Q3, composed of nine people, eludes definition by demographic criteria alone, with its most characteristic feature being a limited range of styles in Spanish. It is the only group which has no three-style speakers in Spanish. This obtains in spite of the fact that the group is primarily Spanish-speaking, with 44% of its members of highland origin. Likewise, 44% show heavy phonological interference from Spanish in their English speech. In education, as in most other respects, the members of Q3 are distributed over the range of categories. The group takes definition mainly as a function of its R factor profile (see below). Q4, with seventeen people, is generally the most highland, the oldest, and the least educated group, with a rather high degree of stylistic flexibility in Spanish in spite of its lack of education. 53% of Q4 are highland born; and while in absolute terms this group contains just 57% of all highland-born people, the percentage rises to 69% when those young people under 23 in other groups who left Puerto Rico at an early age are discounted. All but two members of the group are over 25 years of age, with 47% well over 35. No one in this group has any U.S. education, and a plurality, 47%, received fewer than six years of schooling in Puerto Rico. Thus it is not surprising that on the Spanishness scale 64% are considered to have heavy phonological interference from Spanish in what English speech they produce. Virtually, the entire group displayed two or more speech styles in Spanish, with 41% showing three, while 50% of all three-style Spanish speakers in the sample belong to this group. In summary, the four Q's divide into two basic groups according to language ability, the bilinguals (Q1 and Q2) versus the Spanish dominant speakers (Q3 and Q4). Within the bilinguals, there are those for whom English is identifiably a second language, essentially people for whom Spanish was the sole language of instruction and who finished or came close to finishing high school in that language, in contrast to those rather younger people for whom English was or is the language of instruction, and in most cases is the language of primary usage. Within the essentially Spanish monolingual group, for whom English is to

284

Roxana Ma and Eleanor Herasimchuk

varying degrees marginal, there are those who command a very narrow range of styles in their native language as opposed to those skilled at switching Spanish styles. This appears to be regardless of education. In fact, the individuals who display a greater Spanish linguistic repertoire, perhaps by virtue of being both older and more rural-highland in origin, have an even lower overall level of education than those whose Spanish is essentially one-style. Two other demographic variables which appear on Table 1 but which do not enter into the differentiation of the Q groups are U.S. residence and occupation. Two interesting points do come up in regard to these variables. In all groups, more people fall into the 11-20 years residence category than into any other, with the exception of Q3, in which only one-third falls into that category and another one-third falls into the 6-10 years' residence category. Thus U.S. residence does not function as an independent variable in determining the acquisition of English. The other point is that the two least bilingual groups show a majority of their members to be in the operative category of occupations. This is most interesting in relation to sex. Q1 and Q4 have the same proportional representation of females, yet Q1 is occupationally 43% housewives, while Q4 is 59% operatives, with only 29% housewives. Obviously, more women in the non-English and low education Q4 group work than do women in the Spanish-educated bilingual Q1 group. For the purpose of contrasting the specific linguistic performance of the four Q groups, we cross-tabulated all the items in the factors described in the first section against each group. However, only those linguistic items were considered for which (a) data were available for at least 28 individuals and (b) for which the difference between the highest and the lowest ranking groups was at least equal to the standard deviation of the entire sample. A loading of over ±.50 was also required, although exceptions were made in the case of items 84, 54, and 61, with loadings of .40, .45, and .47, respectively. These were included because of their relevance to the factors in question and also because they fulfilled the other two criteria. Almost all of the items that were selected on the basis of significant differentiation and N had loadings over .50. Out of the six factors described in the first section, five contained sufficient items that met the above criteria, although only four of these, F l , F2, F4, and F7, seemed to clearly indicate Q group differentiation. A fifth, F5, is included in this discussion because its items are relevant to the contrast between Q groups based upon the other four factors and factor items. Table 2 presents the cross-tabulation of factor items by Q groups. The degree to which a Q group relates to a factor or factor item is represented in terms of mean frequency scores.

Speech Styles in Puerto Rican Bilingual Speakers

285

TABLE 2 Differentiation Factor # 1

2

4

5

7

Item #

Speech Context

25 34 36 39 48 56 76 79 101 125 137 145

PRS-B

11 124 138 151 153

PRS-C PRE-B

26 38 46 53 54 61

PRS-B

50 52 55 80 84

PRS-A

96 100 104 112 120

of Q Groups by Linguistic Factor

99 99 99

PRS-A 99

PRE-C »

PRE-WN PRE-B » 99

99

PRE-A 99

99

PRS-A » 99 99

99 99

PRE-C 99

PRE-C PRE-WN 99 99

»

Variant ; 1» sc-2 Vdo-l N-2 VN-1 sc-2 R#-2 UH-3 EH-3 EH-2 EH-i R#-0 T-2 #RR-1 UH-3 R#-l UH-3 EH-3 S#-0 VN-0 SC-0 RC-3 R#-O VN-0 S#-l RC-1 R#-I OH-2 R#-L

; ; : 1 ; J : ! i : ! 1\ 1 j 1 ! j ! 1 ' ! ; j ; 11 i ! ;

11 CC-non Red UH-2 > OH-2 ! VN-1 ; TS-nonRed1

] = highest ranking;

Items

Ql

Q2

Q3

Q4

6.8 1.8 6.5 8.2 15.3 5.3 7.0 11.6 1.8 1.8 16.6 10.4

6.5 ; 1.3 i 5.8 1 5.4! 6.2 2.4 1.9 .9 7.5 5.3 26.6 13.3

14.9 2.7 11.8 13.1 9.7 5.3 3.4 5.6 1.0 2.3 3.3 6.8

14.8 4.3 17.2 15.1 19.8 10.7 5.9 8.5

! -°l i_5.6J

1.9 2.2 3.8 1.8 2.2

1.6 2.8 2.3

2.8 3.4 2.6

¡"8"; ¡3.61 ¡1.0; ! .3 ¡2.8

1.0 5.4 .6 .3 .8 6.2

6.0 20.8 3.7 1.8 5.2 17.7

2.9 9.1 2.5 .5 1.6 12.1

1.8 .5.

2.1

1.0 7.5 4.2 2.6 .7

.2 2.7 2.2 1.8 .3

1.5 5.0 I M 1.5 1.6

rijs\ ! l.i ! ¡1.9! ¡5.0) !_3.8J

3.4 5.0 4.3 10.7 10.9

1.0 1.5 1.8 5.1 4.3

1.8 .8 2.1 6.2 2.8

3.9 14.0 14.4 14.6 20.4

H I! 3.0

1 ; . ! !

!

.81 4

j = lowest ranking

Scores are given as mean frequencies, obtained by dividing the number of occurrences bf the variant by the number of people responding to the variant.

286

Roxana Ma and Eleanor Herasimchuk

Factor One (Q2 and Q4) The first factor in a factor analysis based on frequency of occurrence data is generally viewed as related to size or to the productivity of a set of items, and this obtains in the present case. F1 happens also to be the factor of English dominance which we discussed earlier, since it was noted that there is a preponderance of Spanish items in negative relation to the factor including two English interference variants. (Negative loadings indicate that the people who speak the most English use these items least, and conversely people who use these items do not speak a great deal of English.) It is no surprise that Q2, the largely U.S. born and educated group, ranks highest on this factor (as it does on the Standard English factor, F7 below). But the reverse ranking is of greater importance since it helps define which among the Spanish dominant groups is least English-speaking and enumerates the features of Spanish articulation which are associated with least Englishness. Therefore, we shall discuss Factor 1 primarily in relation to the lowestranking or least English group, Q4, whose low ranking on Factor 1 validates the findings of the global linguistic measurements applied to this group in our demographic Q group analysis. The only two items on which this group is not absolutely lowest are the two English items diat show Spanish interference, UH-3 and EH-3 (talk for tuck, Kant for can't), on which Q4 ranks next to lowest. Looking at the Spanish items on F l , we find that they are all from continuous speech, indicating that Q4 used a larger volume of informal Spanish than any other group. The kind of Spanish used is of some interest. The items comprise the most common variety of informal Spanish typical of the Puerto Rican speaker. It is neither a 'corrected' Spanish nor a nonstandard Spanish in terms of norms for the Island. That is, it contains those 'deviant' features most typical of the Island dialect: I for final r, and h for s before a consonant (e.g., pahtol for pastor). It is a variant which retains underlying morphophonemic shapes of words, so that even in the most informal speech there is some articulatory signal, in contrast to a possible zero realization such as pa'to'. Further, also in relatively informal speech, intervocalic (D) is maintained, so that hablado does not become hablao. Likewise, (VN) receives primary articulation so that gente does not become [hete], while final (N) receives the very common velar articulation, as in Bayamong (but not Bayamo) for Bayamdn. It is important to note that regardless of the informality of the speech contexts represented here, no item in this configuration shows a zero value, and it is for this reason that we describe it as correct for the dialect in these speech contexts. This grouping of features relates rather neatly to our demographic and

Speech Styles in Puerto Rican Bilingual Speakers

287

global linguistic categorization of Q4 as strong on traditional Spanish regardless of formal schooling. If we may look ahead to an item on Factor 5, where no single Q group was found to predominate on all items, we will find some further support for the linguistic identification of Q4 outlined above. We see that Item 55 is Spanish R # - l , or the standard variant, [mar] for mar. Primary value articulation of a variable such as ( R # ) might be expected of a group with much formal schooling, because such a group would presumably be exposed to the standard variant, in contrast to the Island dialect. This might especially apply to a feature for which standard orthography clearly indicates a correction of the dialect pronunciation. But we see here, rather surprisingly perhaps, that the group with the most formal education in Spanish ranks lowest in the realization of R # - l in Context A, the most informal one, while Q4, the least educated group in any language ranks highest. We can infer from this that, at the overall social level in question, schooling is not seen to be necessarily related to the ability to render certain standard speech forms. From the information we have so far we might say, however, that schooling is positively related to the acquisition of English as a second language. But there is no a priori reason to expect lowest schooling to correlate positively with a somewhat more careful Spanish. We therefore must refer to an additional, perhaps more elusive variable, that of traditional culture. We ascribe this to Q4 by virtue of its being the oldest and most highland group, two aspects which, not surprisingly, are found to be closely associated with lowest level of education. We might say, that traditional culture tends to take the place of formal education in the maintenance of certain standard speech features. This implied linguistic conservation prevails against the acquisition of English, regardless of the subject's length of residence in the United States. Turning to the English items found on Factor 1, we find that they comprise a set of the most commonly used English sounds, EH-2 ([se] as in standard American bad) occurs in context WN; the more local variant, EH-1 (as in New York City [be^d]); R # - 0 or r-lessness as in [brASs]; and T-2 or glottal t (in a monomorpheme such as meet) all occur in the interview style context. This set of variants is the essence of the style of local English speech, including style shifting on one variable, (EH). Moreover, with the exception of R#-0, these sounds are not available to the person whose phonology remains to any substantial degree Spanish, as they do not resemble any sounds of the Spanish phonological repertoire. Thus the largest volume of English speech also contains the most representatively English sounds (by local standards), without any presence of interference features. Q2 which ranks highest on these items, also ranks lowest in the use of interference sounds (see

288

Roxana Ma and Eleanor Herasimchuk

UH-3 and EH-3 with means of 1.9 and .9, respectively). In other words, the most English is spoken by those who have mastered English phonology. Furthermore, the most representative sounds in English are all sounds that might be termed nonstandard or lower-status sounds on the sociolinguistic scale for English alone; it is a very colloquial style of English. We shall discuss this again in the section on F7, the moststandard-English factor, in which Q2 ranks highest as it does here. Factor Two (Ql) Factor 2 contains five items, only three of which are encountered in all four Q groups. The factor can be described as one of 'accented English'. In Table 2, the presence of Item 11 or #RR-1 shows that there is also an element of Standard Spanish as well and, reference to the original set of factor items from which these five were chosen shows that the other Spanish items were also the standard Spanish variants. We might thus describe our single Spanish item as a 'schooled Spanish' feature, since it is produced in the context of consecutive reading and since it is produced most by Ql, the group with highest educational attainment in schools in Puerto Rico. Since this item occurred in a closed corpus, we have an absolute standard of scoring. For each respondent, there were five opportunities to use initial / r r / (Spanish [rriko]), as opposed to one of the velar variants [xiko] in this context. Members of Q l each used the standard variant close to four out of the possible five times. This is quite high, considering the commonness of the velar variant, which accounted for 76% of all instances of the variable in context A. We can also note that the articulatory distance between the two variants is so great that some respondents are unable to produce the standard apico-alveolar trill. For these reasons, therefore, we can say that this single item serves adequately as an indicator of ability to produce standard Spanish in a formal context, fulfilling our expectation of Ql's educational attainment in this regard. Not to overemphasize the importance of schooling as a sole influence in the realization of standard phonological variants, however, we note that Q4, the lowest in educational achievements, ranks second on this item. The other four items (representing three different sounds) show Spanish interference in English speech. UH-3 and EH-3 are both substitutions of Spanish vowels for English ones which the respondent does not control. R # - l represents the standard retroflex articulation of final / r / of words like mother, brother. Flap [ r ], as we pointed out earlier, was also counted as instances of R # - l . It is most interesting to compare Q l with Q2 on these English items. First of all, Q2 ranks substantially lower than Q l on all of the items presented here. This is to be expected

Speech Styles in Puerto Rican Bilingual Speakers

289

for the items of Spanish interference, as we have already found Q2 to rank lowest on the two interference items in F1. But the case for R # - l is somewhat more definitive. Not only did Q1 rank higher than Q2 on this standard English item, but it ranked very substantially higher (14.4 as compared to 3.8). As we saw on F l , Q2 ranked highest on R # - 0 , indicating that R # - 0 is most English and most in use by the most English-speaking; Q2 actually had very little variation on this variable. If it were not for Ql's behavior, the standard English R # - l realization would be very poorly represented in our sample. We can conclude from this that, for the present population, articulation of final (R) in English words joins EH-3 and UH-3 in the delineation of interference or in the separation of native and near-native speakers from accented speakers of a certain kind. It could also be said to separate those with secondary education in Puerto Rico from those educated in the U.S. On the one hand, our analysis indicates that use of final (R) in English conversational style seems to be prescribed by native fluency and pressure from the local dialect. On the other hand, we can explain the high incidence of final (R) in Q1 by pointing out the influence of spelling pronunciation. Q l , with its relatively high educational achievement, is 85% biliterate, so that its members have easy access to the written word as a source of information and reinforcement of the underlying shapes of words. The influence of orthography obtains only in the absence of peer group dialect pressure in English at the time education takes place, as is in fact the case with Ql. For this reason let us hasten to add that the same orthographic influence is not likely to prevail in the spoken informal Spanish of Ql; see Item 55 in F5 below, where we find Ql lowest in the articulation of R # - l in Spanish in the context given. This clearly implies then, that peer group is most likely to determine articulation in the primary language, while orthography forms a substantial influence, for those who have access to it, in the articulation of the secondary language. Our conclusion as to the role of orthography in providing pronunciation norms is also supported by the behavior of the other two Spanish-speaking groups, who rank low on English R # - l and on education. Factor Four (Q3) Factor 4, comprising five variants of unmistakable linguistic homogeneity, has already been described as the ellipsis factor. It will be noted from Table 2 that every item describes phonetic zero realizations of the variable in question, including Item 53, the Spanish assimilation variant (RC), which we regard as a form of zero realization. Not only do the items in question describe a distinct pattern of behavior, but they

290

Roxana Ma and Eleanor Herasimchuk

also represent virtually all the important Spanish variables, a possible exception being (RR), which was a variable with a fairly low functional load. (We do note that the highest ranking group in question here, Q3, ranked lowest on the realization of RR-1 on the preceding factor.) Variable (RR) does not appear to be highly correlated with the items for which zero is a possible variant. All the items occur in the contexts of continuous speech B and A. They are: S#-0, or no articulation for final / s / (i.e., má' for más, as opposed to the very common intermediate variant, máh); VN-0 (hete for gente)-, SC-0 (puetto for puesto, instead of the possible intermediate articulation pushto); RC-3 (patte for parte, as distinguished from the most common informal articulation paite); and R # - 0 (hablá for hablar or dialect hablat). Q3 takes on definition as a linguistic group because of its consistent high ranking on this well-defined set of items, as well as being the most limited of all the groups in its Spanish repertoire range. We might generally describe Q3 as the most stylistically limited group of all, for not only is it the only group for whom none of the members commands three Spanish styles by our global measures, but it also has a substantially higher percentage of its members in the one-style Spanish repertoire range category: 44% as compared with 00% for Q1 and 12% for Q4 (see Table 1). It is exceeded only by Q2 in its percentage of one-style Spanish speakers; and Q2 members, we recall, are all fluent in English. Although Q3 appears to command more English than Q4, it is not considered a bilingual group on the whole, as two-thirds of the group show heavy phonological interference from Spanish in their English or are virtual Spanish monolinguals. But Q4 has far greater stylistic flexibility in its native language, with 41% of its members commanding three styles as opposed to 00% for Q3. Thus we can characterize Q3 both as most limited in terms of linguistic repertoire and as exhibiting the most colloquial variants within this stylistic limitation. Our analysis of F4 as being the definitive style of Q3 speakers has an additional dimension beyond the present discussion, in that it describes what many informal Spanish-speaking observers, whether continental South Americans or educated Puerto Ricans, find most 'at fault' with the Puerto Rican popular dialect. The dialect is criticized not by "They use I for r", as it might be, but by "They drop everything; they eat their words (Se comen las palabras)". Our analysis has shown that this description applies to a demographically diffused segment of our sample population, Q3. It is not known how much Q groups 1 and 2 would exhibit the same behavior if as much continuous speech had been elicited from them as from Q groups 3 and 4, but we do have an indication of Ql's stylistic range and we do know that Q4 most

Speech Styles in Puerto Rican Bilingual Speakers

291

typically produces value 1 or value 2 variants rather than value 0. We can probably assert with some confidence that widespread zero realization is a phenomenon restricted to a given population segment, even though we have not been able to correlate it with shared demographic attributes. The fact that Q3 cannot be defined by birthplace nor by age nor by educational level makes it difficult to label zero realization as a function of a specific dialect. Further investigation might reveal it to be a sub-dialect of some form, but at present we can only say that it is found in a variety of native speakers of Spanish as well as in some non-native speakers. There is some indication that whatever its dialect origin or social definition in Puerto Rico, this tendency toward zero realization offers certain conveniences to younger Puerto Ricans learning Spanish in the New York area. The fact that apocope exists as a version or stratum of an actual Puerto Rican speech group provides the justification for young speakers to simplify generally in their own adaptation of apocope Spanish. We are led to believe, therefore, that by virtue of its very demographic heterogeneity Q3 is the group representing New York Spanish and the dialect version that gives rise to it. We would alternatively describe Q3 as the group of linguistic instability or flux, the group that refers least to any standard of formal pronunciation norms. Since this lack of standard reference represents the state of the Spanish language in New York, we might by extension say that Q3 stands at the point of change for Spanish as a New York City language, other influences (such as educational mobility or cultural revival) remaining equal. Factor Seven (Q2) Factor 7 is the factor of standard or formal English, and differs from Factor 1 in that the latter represents volume of productivity for English of whatever kind, while the items of F7 are clearly those representing certain features of standard articulation. While not all variables considered sociolinguistically representative of standard English are encountered on F7, it should be noted that no item on F7 shows any zero or interference variants. We shall also discuss below the implications of a major item of standard English which does not appear on this selected factor, that of R # - l . The absence of R # - l is of particular diagnostic interest. Returning for a moment to the first factor, we recall that R # - 0 was a highly productive item in English, and that group Q2 ranked highest on English items throughout. Then in view of the very high frequency of R # - 0 for Q2 in F l , we should not be surprised that F7, the most

292

Roxana Ma and Eleanor Herasimchuk

standard English factor by the present analysis, does not include this feature. Evidently, variable (R) does not count very strongly as a salient characteristic of Q2's formal English. If we go back to Figures 22 and 23 (Section 5), we find that the standard variant of (R) is more likely to occur before a consonant than word finally, i.e., R # - l is even less likely than RC-1 to be correlated to a factor which is describable as 'Formal English'. And in fact, this is the case. On the original list of items to appear on F7, RC-1 had an extremely low loading of .33 and R # - l did not appear at all. As we mentioned in our discussion of Factor 2, R # - l is a marker of formal English only for Q l , not for Q2. Q2 is largely r-less, in any style of English, and we must look to other items on F7 to find out what the defining linguistic variants are for Q2's formal English. Items 96, 112, and 120 appear to be the most diagnostic, so that our fluent young bilinguals fully articulate final consonant clusters (in words like kept, hats, and milk) and do not nasalize their vowels (in words like pen and pencil) when they speak in a formal style. By contrast, the next most bilingual group, the somewhat older Ql, ranks uniformly low on all these variants. The information on F7 provides us with some clues as to the distribution of sociolinguistic norms in relation to the different generations of PRE speakers. Looking at Ql first, we see that these Puerto Rican educated speakers seem to have norms of correctness and formality for English which are unrelated to those of Q2. Although F7 shows only the items for which Ql scored lowest, not highest, we suspect that in their formal English, they articulate final single consonants such as (T) and ( R # ) but not final clusters of consonants, and that their vowels will most resemble the standard Spanish ones. In other words the norms for their second language will most likely be set up exclusively in terms of or by reference to their primary language. Turning to the younger, U.S.-educated group, Q2, we see a somewhat more complex pattern of sociolinguistic norms. On the one hand, the extent to which Q2 bilinguals are r-less in their formal English speech style reinforces our notion that highly localized peer group pressures play an all-important part in shaping and enforcing linguistic norms. The greater New York City prestige variant, R-l, is conspicuously absent from their speech. On the other hand, the fact that they do not simplify their consonant clusters is some evidence that they are aware of certain features which differentiate them as a linguistic subgroup from their Negro neighbors (cf. Labov's study on cluster simplification in Negro speech), features which they share with other white speakers.8 8

This obviously does not apply for all respondents. Lucy R., a Puerto Rican

Speech Styles in Puerto Rican Bilingual Speakers

293

TABLE 3

Summary of Q and R Analyses Demographic and Global Linguistic Characteristics

Phonological Behavior

Q group Q1 Young adult females of coastal origin with more than 7 years schooling in Puerto Rico, speaking fluent but accented English and using two or more Spanish styles.

High incidence of English vowels showing Spanish interference. Highest ranking group on English final (R) indicating possible orthographic influence. Standard Spanish [r] in reading.

Q2

Mostly under 18, American-born or U.S. educated, fluent in local English, but 50% one-style Spanish speakers.

High incidence of standard English vowels and local English variations, r-lessness in all styles of English but do not simplify consonant clusters; reveals affinities to Negro dialect but not to any white standard beyond the neighborhood dialect.

Q3

Range of age, education and birthplace and, as a result, no real demographic definition. Range of English ability with i showing heavy Spanish interference or being Spanish monolinguals. The only group with no 3-style Spanish speakers.

Uniform ellipsis for all key Spanish variables on which zero is a phonetic variant.

Q4

More than half highland-born, of low education, with very little English mastery, but 88% with two or three styles in Spanish.

Use of standard Spanish variants except for Island dialect features [ r -V1], [s - > h], but no zero realization. Secondary use of standard [ r ]# in word final position.

college girl, corrects for (R). In the other direction, luan H., a gifted styleswitcher and something of a hipster, seems to prefer the Negro dialect as measured by reduction of consonant clusters. He uses a great many of these reductions as the prestige manner for discussing certain hip topics, though he corrects these features for reading.

294

Roxana Ma and Eleanor

Herasimchuk

SUMMARY

We have defined the notion 'speech style' among bilingual speakers by studying the cross-language co-occurrence or co-variation patterns of different PRS and PRE phonological variables, using the technique of factor or R analysis. Six statistically and linguistically well-defined clusters of variants (factors) emerged, which we may think of as speech styles existing in our Puerto Rican community. Then by inter-correlating our speakers based on their linguistic behavior, we were able to divide our speech community into four behaviorally different groups called Q groups. Each of these groups correlated with distinctive demographic and global linguistic characteristics. The four groups were then compared with each other in terms of their behavior or particular items from five out of our six factors. By thus contrasting the discrete phonological behavior of demographically well-defined Q groups, we have been able to provide a meaningful picture of sociolinguistic variation in the Puerto Rican speech community. Table 3 summarizes the main points of the Q and R analyses. While there remain many 'bugs' in the system, our hope has been to show that applications of factor analysis to sociolinguistic research are quite promising and merit further use and refinement as analytic techniques.

REFERENCES Alonso, Amado 1953 Estudios Lingüísticos - temas hispano-americanos (Madrid). Anastasi, A. 1961 Psychological Testing (New York). Ervin-Tripp, S. 1964 "Interaction of Language, Topic, and Listener", in J. Gumperz and D. Hymes, (eds.), Ethnography of Communication. American Anthropologist, 66, no. 2: 86-101. Ferguson, S. 1959 "Diglossia", Word, 15: 34-41. Fishman, J., R. Cooper and R. Ma et al. 1968 Bilingualism in the Barrio. Final Report on Contract No. OEC-1-7062817-0297 to DHEW (New York: Yeshiva University). Published edition: Bloomington, Laguage Sciences Series, (Publication 7), 1971. Gumperz, J. 1967 "The Linguistic Markers of Bilingualism", Journal of Social Issues, 23, no. 2: 48-57. Labov, W. 1964 "Hypercorrection by the lower middle classes as a factor in linguistic change", in W. Bright (ed.), Sociolinguistics (The Hague: Mouton), 84-113.

Speech Styles in Puerto Rican Bilingual Speakers 1965

295

"On the mechanism of linguistic change", Georgetown Monograph Series, no. 18: 91-113. 1966 The Social Stratification of English in New York City (Washington: Center for Applied Linguistics). Labov, W., P. Cohen and C. Robins 1965 A Preliminary Study of the Structure of English Used by Negro and Puerto Rican Speakers in New York City, Cooperative Research Project No. 3091 (New York: Columbia University). Navarro-Tomás, T. 1948 El español en Puerto Rico (Rio Piedras).

Everett C. Hughes THE LINGUISTIC DIVISION OF LABOR IN INDUSTRIAL AND URBAN SOCIETIES *

Language encounters, hence bilingualism, are a function of social organization. Bilingualism, as a social phenomenon, must be described in terms of the organization of communication in a particular society. Montreal yields many clear examples of between group bilingualism as well as within-group bilingualism, i.e. of bilingualism bridging the gap between two separate monolingual populations interacting in the same organization and of bilingualism between individuals interacting in organizations most of whose members are bilingual. The linguistic division of labor in complex social settings can be clarified only if both types of bilingualism are recognized and recorded, sometimes for the very same participants.

In urban and industrial societies in which two or more languages are widely current who speaks what language to whom, about what and in what situations? I refer not merely to verbal speech in personal encounters but to all sorts uses of language and not only between individuals but between institutions and the public as well. By linguistic division of labor I mean that in a society where two languages are used, they are not used for precisely the same purposes, one being used more in certain contexts and for certain purposes than the other. The two make a more or less complete system of communication in that community. Of course, where there are two languages of wide use there are probably also other languages used by fewer people in perhaps fewer contexts. We are ignoring the languages of lesser use in this discussion. My illustrations will be taken from the one very large city of two main languages which I know fairly well, Montreal, Canada.1 It is a city * From: Georgetown Monograph Series on Languages and Linguistics, 23 (1970), pp. 103-119. Reprinted with permission. 1 The reader is warned that Montreal is in a period of conflict over the ap-

The Linguistic Division of Labor

297

in which Mackey's statement about the world is apt: "More and more people are tending to be bilingual through the necessity of becoming poly-social; that is, belonging to one group for one thing and to another for another."2 Two official languages are widely used in Montreal. One is the language of that empire on which, for a number of decades, the sun never set, of most of the continent of North America, thus of the world's greatest industrial complex, and which is at present the dominant language of world business, science, and diplomacy. It is the language of about one-quarter of the population of Montreal but of a much larger proportion of its financial and industrial transactions. The other language, that of about three-quarters of the population is also a world language; it had its day as the dominant language of the world, and can now hold its own as a medium for communicating all sorts of human concerns and thought. M. de Gaulle hoped to restore it to world dominance. It is that language which, more than any other, people pretend they can understand, or could understand if only the French were to speak their own language with a true Parisian accent. While it is the language of a social élite, theatre, theology, local and provincial politics, of universities and learned professions in Montreal, it is better designated as the language of the masses of the people and of small rather than of larger transactions. W. F. Mackey calls such languages as English and French "great languages" and "languages of wide distribution". "They are often the same languages in which most of the world's knowledge is available propriate use of the two main languages and over the place of the ethnic groups in business, industry, government and education. A great deal of research on use of language is being done by sociologists at the Université de Montréal, especially by Prof. Jacques Brazeau, who has also done similar study in Belgium with respect to the Flemings and the French-speaking Walloons. A commission of the Government of Quebec is also considering the matter. A Royal Commission on Bilingualism and Biculturalism, appointed by the Government of Canada some years ago, has issued a number of volumes dealing with the place of French and English in various regions and institutions. Of the several volumes it has thus far issued the one most pertinent to our discussion is Report of the Royal Commission on Bilingualism and Biculturalism, Book III, The Work World, Part 1, Socio-economic Status; Part 3, The Private Sector (Ottawa, The Queen's Printer, 1969). I regret that I received these volumes so recently that I have not been able to make use of them in this discussion. I should also mention a pertinent thesis: Lussier, Yvon, La Division du Travail selon l'origine au Québec, 1931-1961, M A . thesis. Université de Montréal, 1967. This is an exhaustive analysis of the numbers and proportions of people of the various ethnic origins in the occupations and industries of the Province of Quebec. 2 W. F. Mackey, Bilingualism as a World Problem (Montreal, Harvest House, 1967), pp. 19-20.

298

Everett C. Hughes

languages which can easily express and communicate not only what is known but also the new findings of modern science and technology. They are also the native languages of countries with the economic wealth and population to be able to diffuse this learning in the form of books, periodicals, films, broadcasts, and to spread this knowledge throughout extensive areas over which the country has some influence. Because of this, these languages develop into very flexible communications media, carrying a varied culture which becomes more and more universal." Contact between the dominant languages and those which are tribal, regional, or limited to one continent or subcontinent occurs nowadays wherever these languages of limited currency are found; likewise contacts of people who use languages of limited repertoire with those of wider functions are to be found wherever such limited languages are in use. Where a more limited language is in contact with a dominant language, the dominant language will likely take over some functions which could perfectly well be carried out in the limited language. In the case of the meeting of world languages, as English and French in Montreal, the dominance of one in certain areas of activity may be such that one can say, as does Jacques Brazeau, that "in several respects French is an unused language".3 The starting point of sociological study of languages is the fact that human societies are systems of interaction; interaction occurs only where there is communication even though it be only the communication of motion from one billiard ball to another.4 Spoken and written languages are among the chief media of communication, hence of interaction. In the rhetoric of science presumably each symbol used has but one clearly defined and strictly limited reference to each of the givers and receivers of a message; one hears that even in that case gestures of aggression creep in. It is probably possible to say "you idiot" or "I told you so" in algebraic symbols. But, in any case, such limited communication of knowledge is, if not rare, a rarefied form of communication. Situations are generally not so pure, and the symbols and gestures are richer in meaning. One of the purest of situations is that of trade; and trade has been one of the recurrent occasions of meeting of peoples of different languages; hence of inventiveness in communication. 3 J. Brazeau, "Language Difference and Occupational Experience", Canadian Journal of Economics and Political Science, XXIV (1958), p. 536. See also his "La Question Linguistigue à Montréal", Revue de L'Institut de Sociologie, 1968, pp. 31-52. 4 R. E. Park, and E. W. Burgess, Introduction to the Science of Sociology (Chicago, 1921). Chapter on "Social Interaction", p. 341 et passim. "Society stated in mechanistic terms reduces to interaction . . . the limits of society are coterminus with the limits of interaction. Communication (is) the medium of social interaction."

The Linguistic Division of Labor

299

Among the most fascinating phenomena of the contacts of peoples are the 'pidgin' languages, and their offspring, the Creoles, which arise when there results from migrations new settled classes of people in such contact with each other and in such isolation from others that they develop a code of their own. The sociolinguists and anthropologists are working hard on the pidgins and the Creoles. Others are working on the loss or maintenance of languages in countries that receive immigrants. Still others are working on the bilingualism which results when peoples are in contact. But we sociologists, pioneers as we have often been in study of the contacts of peoples as well as in study of social interaction - which requires communication - have done little with language except to count the numbers of people who claim to speak some one language or combinations of more than one.5 R. E. Park and R. D. McKenzie8 were pioneers in study of ethnic division of labor. McKenzie noted the functions of various Indian, Chinese and other ethnic groups in the rubber plantation economy of the Malay Peninsula. Park and his students described the symbiotic relations of the various ethnic and racial groups in the plantation and other large-scale commercial agricultural economies of the U.S.A., Brazil, Hawaii and other regions. The races worked with and for each other with a minimum of social contact. Their communication was limited to that necessary to getting the simple labor of the plantation done. It was the work of these two men that led me to look at French Canada as an economy and society in which French and English each performed certain functions, and to show in tabular form what these functions were in various industries and professions. My students William Roy and Stuart Jamieson had a hand in our early analysis of these relations. But we did not pay much attention to the use of language in the industries and occupations we studied, although of course we used both French and English in gathering our data on the occupational distribution of the ethnic groups in industry.7 5

The readers of this doubtless know that a team has been working on choice of language in market transactions in Ethiopia. It is an ideal experimental setting for sociological study of linguistic division of labor. See this Volume, pp. 255-267. * R. D. McKenzie, "Migration in the Pacific Area", in American Foreign Relations, C. P. Howland (ed.) (New Haven, Yale University Press, 1930). McKenzie, "Cultural and Racial Differences as Bases of Human Symbiosis", in Social Attitudes, Kimball Young (ed.) (New York, Henry Holt, 1931). McKenzie, "Industrial Expansion and the Inter-relations of Peoples", in E. B. Reuter (ed.), Race and Culture Contacts (New York, McGraw-Hill, 1934). McKenzie's work has recently been reissued in A. Hawley (ed.), R. D. McKenzie on Human Ecology (Chicago, University of Chicago Press, 1969). 7 Everett C. Hughes, French Canada in Transition (Chicago, University of

300

Everett C. Hughes

As students of Park carried out parts of his program of studying ethnic relations throughout the world, it became more evident that contacts between peoples often brought about an industrial revolution for one of them; sometimes for both.8 Seen from the other end, seldom has any country become urban and industrial without becoming ethnically diverse. "Industry is always and everywhere a grand mixer of peoples."9 But it does not beat them at once into a homogeneous batter. Rather it comes about that in the industries management and labor are not of the same ethnic background; and the various staff services are manned often by still other ethnic groups. This has long been known and has been much studied; ethnic succession in industries was one of the topics of interest to the early Chicago human ecologists. Ethnic differences ordinarily include difference of language. In newer industrial regions, or in older ones which receive immigrant labor, the language of management is more likely a world language while the new labor more likely uses a language of less dominance, one less widely diffused and less versatile in its uses. One could devote much work to observation and analysis of communication in the early phases of linguistic adjustment between management and labor. This may be conceived in many ways. It may be a sort of Robinson Crusoe-Good Man Friday case where there is no previous history of communication. More often there is some supply of people who are bilingual enough to pass on commands and to teach the greenhorns. In a recent novel about life on a Paris automobile assembly line,10 the French foreman gets the word to a bilingual Algerian worker who initiates newcomers who speak only Arabic. But not much language is required. Given a wrench and a supply of nuts and bolts, one can point to the hole, insert the bolt, tighten the nut and hand the wrench to the greenhorn without any talk. He can make signs to show the nut must be drawn tight. In fact, the Algerians even initiate a Hungarian worker who has neither French nor Arabic. They even get it over to him to take it a bit easy when the time-study man is approaching. Our purpose, however, is not to go into the natural history of communications where ethnic groups meet in industry and cities, but to talk of the particular case of cities and industrial systems already in a very advanced stage of development but Chicago Press, 1943). Phoenix Edition, 1963. Stuart M. Jamieson, "French and English in the Institutional Structure of Montreal", M.A. thesis, McGill University, 1938. William H. Roy, "The French-English Division of Labor in Quebec", M.A. thesis, McGill University, 1935. 8 See Hughes and Hughes, Where Peoples Meet, 1952. Chapter 5, "Industrial Revolutions and Ethnic Frontiers". • Ibid., p. 63. 10 Claire Etcherelli, Elise ou la vraie vie (Paris, Denoel, 1967).

The Linguistic Division of Labor

301

in which there are two main languages in use both at work and in the city and region. These background conditions will and do vary greatly throughout the world, and even between cities within a given country. The assembly line is but one of many functioning parts of modern industrial and other organizations. It has its own problems of communication on the small scale among fellow workers, and of communication with the rest of the factory, with unions, families and public. One way of looking at a community, especially a large one, is as a constellation of institutions, that is, a constellation of going concerns all the way from families to huge corporations, public agencies and government. Communication goes on inside these institutions. Each of them has its own situations of confrontation of people of various categories (offices, roles). For many of these situations there is an appropriate social rhetoric. It may be instruction, command, persuasion, threat. The people in the various roles learn the equations relating words to action, the amount of salt to be taken or to recognize understatement and to be prepared for the worst. I will come back to internal communication and language. The institutions also communicate with each other at various levels: research institutes with client industries; business with government; unions with management. Some of these inter-institutional communications are highly technical, some are privileged and supposedly locked in top drawers; some are public relations pronouncements meant to be consumed, if not swallowed whole, to be overheard as well as heard. The institutions also communicate with the public to whom they furnish goods or services, from which they seek a mandate to carry on and to control certain features of society, goods and services. Each institution, each going concern may be considered a system of communications, with its confrontations between communicators and audiences, and with its own degrees, kinds and speeds of feed-back; with its degrees of intimacy and exclusiveness. Some communications are meant only for a few, and it is assumed that those few instinctively keep the secret. Some communications are made in terms which can be understood by only those few who have a particular kind of technical training. My small contribution to the study of bilingualism is merely the proposition that bilingualism, as a social phenomenon, must be described in the terms of the organization of communication in the particular society. It is the same proposition I insist upon in study of race and ethnic relations in general; that the significant statements concerning them can be made only in terms that have social significance in their own right. Until now most of what we have on bilingualism is statistics on the

302

Everett C. Hughes

statements of people concerning the language they speak, when they learned them and perhaps how well they know them. Some of the applied linguists have already gone a good deal beyond that; many studies are being made of the kind I suggest, but usually in small settings. We are all customers or clients of the great institutions. Few are the makers and distributors of goods and services who do not appeal to us to believe in their good will towards us and to accept their notions of what is best for us. In a multi-lingual society, they have to decide in what language to address us, and in what rhetoric. This requires a judgment about the relative number, spending power, social influence and political power of the publics of the different languages.11 But these same service and commercial institutions tend to take on the quality of utilities, which deliver universally desired or required goods and services in highly standardized forms to everyone. The customer-client must be taught how to shop for the goods and how to use the services. A combination of simple instruction in the commoner languages with signs and symbols makes it possible to dispense goods and services with a minimum of talk and writing.12 As services are mechanized and made more automatic, language encounters of persons are eliminated. The arrangement of goods in a super-mart, the packages and the pictures make it possible for the initiated to shop efficiently without talk. A bilingual taped message tells us what to do next about a phone call. The picture on the slot-machine, the male or female shoe on the toilet door, the international traffic signs in European countries make it possible for people initiated into modern life to travel, buy groceries, make telephone calls. A Montreal acquaintance tells of seeing a new immigrant woman from North Africa taking something from a shelf in a supermarket to one of the people who put things on the shelves and trying to haggle with him over the price. Elihu Katz reports that Yemenites newly arrived in Jerusalem try to haggle with the bus driver over the fare. Haggling requires a richer language. This is why I emphasize that this standardized communication without talk depends upon the initiation of the public into this scheme of things. The special11 Frederick Elkin of York University, Toronto, has done work not merely on the language of advertising in Montreal, but also on the symbolic rhetoric of advertising directed to the two ethnic groups. See Elkin, "Advertising Themes and Quiet Revolutions: Dilemmas in French Canada", American Journal of Sociology, Vol. 75 (July 1969), pp. 112-122. 12 At the meeting of the Canadian Sociological-Anthropological Association of June, 1969, a paper was presented on the image of the woman shown in various kinds of advertisements in Montreal. If clothing is being advertised, the woman is young and slim. In advertisements for bread she is allowed to be more motherly in appearance. This suggests that the language of advertising goes far beyond use of words.

The Linguistic Division of Labor

303

izing and standardizing of organization and of positions in organizations - known as bureaucracy - depends upon a complementary standardizing of goods, services and of the wants and behavior of the public. The apparatus for delivering many kinds of goods and services and 'savvy' about that apparatus is so widely diffused that the people of the post-industrial world can get themselves to and from work, can buy food, drink, cigarettes, can post letters, see the movies and heaven knows what else (sailors never learn the languages of all those ports) with scarcely any verbal exchange. It is as if the day of the silent trade had returned. But if in certain transactions language has been automated out, in other transactions it returns with all of its power of nuance, nicety, elaboration and sophisticated imagination and precision. The choice of language becomes itself a gesture of power, of acquiescence, of intimacy, of self-defense, within the limits of the bilingualism of the parties to the transactions. Further, the number and proportion of bilinguals in a population is itself a function of power and of other social relations. It is characteristic of large business and industrial organizations in Montreal that the top management and technical personnel are unilingual Anglophones. They exchange messages with their counterparts in similar organizations - often with parent companies — in the United States. They belong to clubs of their peers. If they are technical and scientific staff, they read the technical journals and attend scientific meetings and consort with other such staff people and with research people. Their language is English. (At a meeting of medical educators in New Delhi, I overheard the Swede, the Norwegian and the Dane talking English to each other. When I inquired why, they said if we were talking medicine at home, we would probably talk English among colleagues, but not to patients.) There is thus horizontal communication which, in the upper strata of large organizations, is apt to be done in an international language; and in the dominant language of the economic system of which the local going concerns are part. Even at this upper level there is horizontal communication with Francophones in Montreal. In our study of white collar careers in Montreal we found the bilingual private secretary. Her function is liaison communication. Usually of upper middle-class French family, she has taken the bilingual secretarial course offered by a certain convent. She has developed to a fine point the art of answering a phone call in the right language, French or English, and of performing all those confidential and diplomatic functions required of a private secretary. These include translation of messages from either language to the other, but also the translation of social gestures. This art of liaison communication is a standard requirement of most large organizations;

304

Everett C. Hughes

bilingualism is an extra skill required in performing it in Montreal. One also finds in certain management offices a bilingual executive assistant, a personnel man and other staff people usually playing second fiddle to the number-one man of the department. This bilingual man has some of the liaison functions of the bilingual secretary, but he also deals with people lower in the ranks inside the organization. As one goes down the ranks, it is likely that horizontal communication will circulate in a smaller orbit, and be confined to the one concern or enterprise, except insofar as the members (workers) belong to a union which transcends that enterprise. With the decline of trade or craft unions, and with the emphasis on seniority, guaranteed wages, and on pensions - i.e., on security - workers are more and more bound by interest to a given organization and may thus have less occasion for communication with people of their own level in other organizations. We may state this as the proposition that, in the measure that itineracy occurs less frequently, there is less close communication among the people of a given level but in different organizations. As one goes downward in large organizations one might expect that there would be less inter-organizational communication. On the other hand, the need for internal horizontal communication probably does not decrease as one goes downward through the strata. In so far as it concerns a small group of people who are very much in the same boat, they can use a language peculiar to themselves, or at least not shared with people at other levels or in other sections of the organization in which they work. In Montreal the men in a shop can talk to each other in that special variety of Canadian French called joual, from the way of pronouncing what is in standard French, cheval, a horse. I am told that the boys gathering around the water-fountain talk joual to each other, saying things the girls must delicately pretend they do not understand. In many Montreal establishments, however, use of standard Canadian French is sufficient guarantee against eavesdropping by management of the higher ranks. The question then is at what point in the system there is need of bilingualism for internal communication. There is translation at some level in the line of command of many organizations; that is, a point of bilingualism. Sometimes it is a clear point, with an individual foreman passing the English word downwards in French; in some cases he uses his bilingualism to keep a monopoly on both downward and upward communication. Of course, many supervisors manage such a monopoly of upward and downward communication in organizations where but one language is used. Bilingualism is an extra device for the common practice of keeping control of communications. The minor boss who uses bilingualism for controlling downward and upward communication

The Linguistic Division of Labor

305

can, by refusing to speak English with aspiring workers under his supervision, make it difficult for them to become functionally bilingual enough to rise in the organization. This possibility exists wherever bilingualism is required for upward mobility. In the retail establishment, unless it be a small one in a linguistically homogeneous neighborhood, bilingualism occurs at the bottom and at the periphery where customers confront salespeople. While many goods are dispensed with little or no verbal exchange, the large department store must still talk to its customers. In earlier times in Montreal the large department stores catered largely to English customers. That day is long past, and with it, the unilingual salesperson. The customer is always right; one must try to answer him in his own language. But some customers are righter than others. There are few retail establishments in Montreal which can ignore or turn away French customers, although some may have some departments which cater to a clientele that might be largely English or at least bilingual enough to be served in English. A department store which did not advertise at all in the French newspapers when I first knew Montreal, this year even carried a full-page goodwill advertisement in the French papers congratulating the French-Canadian people on the occasion of their national holiday, St. Jean Baptiste. The store is thus not merely seeking French as individual customers, but seeks their collective goodwill. This is accompanied by an almost complete elimination of unilingual English from retail selling, from the point of contact with the public. Previously the book section of a department store was usually completely or almost completely English, and the sales people usually did not speak French. The more general question here is what kinds of goods are retailed in a modern city to one ethnic group only, to two or more ethnic-linguistic groups separately, and finally, what kinds are distributed to all ethnic groups by the same distributing agencies. The trend in the modern consumer society is toward common distributing agencies, since the various groups want essentially the same goods and are so distributed spatially and by social class that there is less and less place in the system for enterprises that cater to only one group. The enterprises which cater to one language group tend to be small and residual; although some can survive on being exotic. A French restaurant may thrive in Montreal by catering to tourists, both local and foreign, but it must be ready to translate its menu verbally into English for customers who can't quite read French. Schools of all levels, hospitals and clinics, institutions which dispense professional services, social agencies - public and private - form an increasing part of the constellation of institutions. They communicate constantly with individual members of the public. Although there is a

306

Everett C. Hughes

demand for protection against the evil eye among Italian businessmen in Boston, and although many Wall Street operators retain the services of fortune tellers and astrologers, and although the chiropractors of Montreal - nearly all French - have recently declared their willingness to give their services free to indigents, the medical services demanded by various ethnic-linguistic groups tend to be more and more alike. The Province of Quebec has three French universities with medical schools which, of course, teach and treat patients in French. Hospitals still are related to religions. Ethnic and religious groups differ somewhat in their demands for medical services; the religious establishments also differ in their conception of disease and treatment, and even as to what services they are willing to perform. Tying off the tubes was formerly not permitted in Catholic hospitals, although there appears to be a change. (In Boston, I was told by a gynecologist that it is no longer Catholic women, but black women who refuse this operation. They want to make sure of the future of the black race.) There is a dialectic in medicine between the peculiar symptoms and endemic troubles of various cultures, expressed in some of the most particular phrases and concepts of the appropriate language, and the more universal aspects of human physiology, disease, the medical sciences and modes of treatment. A medical team is in a sense multi-lingual even if its several members and the patient all speak one standard historic language. The translation from patients' complaints to medical diagnosis and effective management may involve a practical nurse, a registered nurse, a clergyman, a social worker, other patients on the ward, and physicians and surgeons of widely differing viewpoints as to what ails people. Add difference of basic culture and language to the differences of professional and lay culture and language and one has a difficult situation. Sick people are often afraid of death; they wish to die in proper religious and cultural surroundings, in cultural and linguistic comfort. In bilingual Montreal there have been two systems of hospitals, English-Protestant and French-Catholic, each staffed from top to bottom by people of the appropriate language and religion. Hospitals are also training places for nurses and doctors. There is accordingly a full complement of higher educational institutions and training schools in each language and attached to the religious establishment to some extent." 15 There is a Jewish hospital in Montreal. Such a hospital serves several purposes. Hospitals are hotels for sick people; Jewish sick people require special food and kitchens. There is often a problem for Jewish doctors to find hospitals they can practice in. Jewish medical graduates do not always find it easy to get internships and residencies of the calibre they are prepared to accept. The language of the Jewish hospital in Montreal is, in general, English; Yiddish is probably used for the comfort of some of the older patients.

The Linguistic Division of Labor

307

Some French patients get into the English system of hospitals and clinics, through referral or for other reasons; the reverse occurs, but probably less frequently. Doctors work rather exclusively in one system or another. Nurses, as their occupation becomes more and more professionalized and secular, become more interchangeable. French laboratory technicians and physiotherapists, practicing new trades, are more likely to have had training in the same schools as their English colleagues. There is, however, a strong movement to fill out the French system of higher education to include all the specialty training programs of the North American system. That movement requires bilingual teachers, trained in English-speaking schools but now working with French students. The hoped-for result is that a French Canadian may have all of his medical needs provided for in institutions of his own language. Each ethnic medical system might get along with one language internally, if each accepts patients of only one language. Even so, there remains the problem of horizontal communication across the ethnic and language line. There is, in Quebec, a single official College of Physicians which licenses all members of the profession. The staff of that organization is, need one say, bilingual in a very complete way - in technical language and in diplomatic language. Medical sciences are largely international. The recent rash of heart transplants is certainly multilingual; the surgeons of South Africa and Montreal who have performed these operations are all bilingual members of groups which have been known as ethnic minorities, although they are not numerically in the minority in their immediate cities or regions. This brings up the question of communication among colleagues of two language groups working in the same city or region. One's closest colleagues are generally those with whom he is in daily contact, and with whom he must talk about particular cases and problems. But there is a larger colleagueship of a profession within a country or language community. English-speaking physicians and nurses in Montreal are members of the English-speaking national fellowships, and even of the inclusive North American fellowship. French-speaking physicians are in some measure members of the same fellowship, but they have a more intense provincial fellowship; and they have the alternative of fellowship in the worldwide French community centering in France. To which of these fellowships do they refer their wish for recognition, mutual confidences, for all that is implied in closer colleagueship? If they were to refer only to the French medical community, they would be practically free of verbal encounters which would require use of English. They could lapse into verbal unilingualism but would have to be bilingual in the reading of medical literature since so much of scien-

308

Everett C. Hughes

tific and medical work is published in English only. The nurses, now that even the Catholic nursing orders have joined actively in modern nursing education and professional organization, seem to find their closer colleagues not among the nursing sisters of Europe but among the more aggressively professional nurses of North America. All of the professions, while they may carry on much of their practice with clients of their own language, and may have some closer colleagueship which can be nourished in that language, may have some point or points of contact with a larger community in which other languages are used. Thus bilingualism of various degrees of power of colloquial understanding and expression, of varying degrees of oral and written communications is to be found in modern professional systems. The pressure for bilingualism is the greater on those whose primary language is less dominant in the systems in question. One could carry this on to the law, where there are confrontations with clients with opposing counsel in the courts, with colleagues, with juries and with judges of lower and higher courts. As a rule where two historic literary languages are current, there are in some respects two systems of law. One who is tried before a court and judged by a jury of another language than his own is likely not to be content with the verdict. One might also go on to analysis of language and religion. Liturgical bilingualism is very widespread. The priest performs his mysteries in obsolescent language; the laymen live, sin and are guilty in a vernacular. This institutionalized diglossia becomes holy; people are really torn over whether they want the Mass said in the vernacular. Yet religion is tied with ethnic and linguistic identity. The clergy of various denominations and of various religions are almost by definition not colleagues; at least, not if they and their flocks take their doctrinal differences as heaven or hell choices. But we have seen a growing colleagueship of the clergy of many denominations and nations. How far they can speak to each other in the same linguistic and theological codes is to be seen. In our city of Montreal, there is a huge colleagueship of French-speaking Catholic clergy; let no one believe that they agree with each other on all things! Yet I suppose a Catholic priest, or a brother or nun, can pass more days than almost anyone without use of the English language. There are in Montreal complete, separate systems of public communication - newspapers, periodicals, radio, television. The public for each language system is so large that there is no resort to the device of programmes or publications which use both languages. The bilingual connection is that of getting the news which comes over the wire in one language turned into the other. English dominance is clear; it is the

The Linguistic Division of Labor

309

French who must translate news dispatches. How much bilingual readership there is I do not know; most of it is, I suspect, done by French readers. Another connection is that of advertising, an especially interesting case because the words of advertising are meant as stimuli to arouse feeling and move people. Simple translation won't do, as Frederick Elkin has shown. In this paper I have meant to demonstrate the obvious: that language encounters, hence bilingualism, are a function of social organization. It is social organization which makes the occasions for communication between people. In many parts of the world drastic changes in social organization are bringing people of different languages into contact. In some cases, through various media, one language encounters people of another without personal confrontation. The case from which I have taken examples is of long standing, but the pattern of confrontation is being drastically changed as society itself changes. These changes have in turn aroused old sentiments of nationalism to new liveliness; new demands are made by the French - that Quebec have but one language, that English people and all immigrants learn French, and that the transactions of higher management be carried out in French. I have presented no organized empirical findings. My own knowledge consists of my experience of linguistic encounters with French speakers in many situations, in various roles and with opposites of various roles and statuses in relation to my own. I have had students report their encounters. But I have no systematically organized findings. Since I first began considering this problem, there has been a tremendous surge of studies similar to what I have proposed and would like to see done. My proposal, once made but not carried out, was to train an army of observers to pick up language cues quickly and send them to observe a large sample of places where people meet in various roles to carry out transactions, and to record choice and use of language. They were also to have initiated transactions in various situations and to have noted the reactions. Such a study would have included detailed observation of the characteristic linguistic encounters inside various institutions. It is certain that I am far behind the times with respect to the linguistic changes and politics of the État de Québec as well as concerning the research being carried on there. It is gratifying to learn that research is being carried on in experimentally created situations and that many natural situations are being studied with what amounts to experimental control of pertinent variables. The intent of my paper is, in part, to develop a frame of reference for study of the linguistic division of labor in complicated settings; in part to alert my fellow sociologists to the importance and the fascination of a field they have neglected.

Section IV: Language Maintenance and Language Shift E. Glyn Lewis MIGRATION AND LANGUAGE IN THE U.S.S.R.*

Planned migrations from the European to Eastern and especially Central Asian Republic of the U.S.S.R., as well as the more localised migration into towns, have added a new dimension to the existing linguistic heterogeneity of the Soviet Union. The effects on language maintenance in different Republics have been accentuated by differential growth rates and reinforced by educational policy as well as by the ideological movement towards 'merging' the nations and the cultural traditions of the nationalities. The level of language maintenance varies between urban and rural areas, migrant and non-migrant populations in the Republic, according to the size of the ethnic group and the extent of its dispersal among other factors. Where another language is substituted for the native tongue Russian is usually chosen, but this too varies according to some of the factors already suggested. In spite of considerable migration, the salient feature is the high degree of language maintenance. (1)

ETHNIC COMPOSITION OF THE U.S.S.R.

"God alone can count the multilingual peoples who live in the mountains of the Caucasus" it has been said, and the Caucasus region is only a small reflection of the vast complex of nationalities, ethnic groups and different linguistic communities which make up the U.S.S.R. today. Since the criteria employed in interpreting the ethnic affiliations of the population have been changed,1 it is difficult to relate estimates of the number of nationalities and of their sizes which have been produced at different times. In 1926 it was claimed that there were upward of 169 * From: International Migration Review 1971, 5, 147-179. Reprinted with permission. 1 In the 1926 Census the individual was asked to state his ethnic group, objectively in terms of origin - namely his narodnost. In 1939 a subjective criterion was substituted, namely his ethnic preference or affiliation - natsionalnost.

Migration

and Language

in the

311

U.S.S.R. TABLE 1

Major Nationalities and Selection of Minor Nationalities Numerical Composition - 1926, 1939 and 19592

O oi yj

.05 —

o r - © VO o VO t i—i Tf"

vi

o

"t vo"

OV.

o o o

< VI o en VO m es

o vi

" 1 en e s 00 r-

o o ""

00 B g g

X a? '

t-; Tf Ov vo' rH rH rH t - rH

z

vo e s e s ov en 00

o o o *-H

o v> O rH i-H e s Vi o . es vo"

Vi TT t— vo"

o o o

o < e s r-; en Ov vo es

O o" o

o o o **

w-i o

v> VO

t

00

-

to

cj o en f - ; VO VO VO