u Our ma

..‘

‘<-- n v'

 

.
a
.
2
.1
'2

._1n 7 «v- -‘-. I

 

t

'11

J. .-.-r

 

 

.
u u
L

 

 

,4
{-5 $.12“ 1
'E: i: . t "'.
as?! .ng5; 3‘4};
.:;‘_ 'u,

.-
1
. .4
. S“

 

I: . H‘V-‘i .
33.3110
4’ ‘u-fi‘

.\
.4:-
M

v
I" An! A

 

‘V'lJ‘I ’
. .. ,4. ...,.
‘

J .
”.2: 11"» . .
'Eé:(£~é3‘s‘
v \‘ Inga, ,
‘53.?“ .12":

.
x

l‘ 51 4“. "A ‘
'9’“73£3"Mj.‘;.» ‘
:3?» Ni
4'35».
yr

 

 

s ..
1' ‘l a L
‘ 3A

95;“:gfi‘i;

“ '» ' “fin-F
awré'fik
*5

 

‘ Y. ~\ ‘
z “'n' ) at
‘9‘;ng A
Q'

.
u
1

51!“

. .555, ,3:
c8.

 

 

.4) u
x"

‘-.-i.r‘t

 

r
‘ J!
«9:

. 'f '
.;'.:,!.’1
.4',

 

1". o";

{mint-p};

. w.

. 0.-.... ;..

, . ..
m an",

”mm-..

”A . fl,

.
.0

9

5.1.} .

i.

«a.
«1-.‘7-
. .13 .2

 

G ‘5. ~ ',
”fw'g—‘hwr

32:1: .5114.

 

  ml

(

Milli mun

3 1293 00609

I)ate

0-7639

‘ TATE UNIVERSITY LIBPAFHES

WM Ni bl 5‘67 7

856

 

 

ll'lil

 

 

 

 

 

‘11
2

 

LIBRARY
Michigan State

dth |

 

 

This is to certify that the

thesis entitled
A Study of the Variable /t/ in the English
Dialect of Indian Immigrants in the United States

presented by

Lakshmi Jagadish Gogate

has been accepted towards fulfillment
of the requirements for

MaSLBLL— degree in W8

41%, 8 END

Major professor

616/89

MS U i: an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.

TO AVOID FINES return on or before date duo.

DATE DUE DATE DUE

  

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L_____
——__T

 

 

 

 

 

 

 

 

 

##

 

 

MSU Is An Affirmative ActionlEqual Opportunity Institution

 

A STUDY OF THE VARIABLE /t/
IN THE ENGLISH DIALECT
OF INDIAN IMMIGRANTS IN THE UNITED STATES

BY

Lakshmi Jagadish Gogate

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

MASTER OF ARTS

Department of Linguistics and Germanic, Slavic, Asian

and African Languages
1989

ABSTRACT
A Study of the Variable /t/ in the English Dialect

of Indian Immigrants in the United States

BY

Lakshmi Jagadish Gogate

This research concerns the variation and sound change from Indian
English to American English of the sociolinguistic variable /t/ in first
generation Indian immigrant speakers in the United States. Due to the
underlying features of LI (native language) of Indians, Indian English
/t/ consists of [t] with varying degrees of retroflexion; standard
American English uses [t] among other allophones.

One male American, one female American, one male Indian, and a
female Indian, interviewed and recorded speech samples from 24
subjects, 12 male and l2 female. As first used in Labov (I966),
informal Interviews elicited informal speech and a phrase list elicited
formal speech.

The 3840 occurrences of /t/ studied consisted of two sociolin-
guistic variants - a high-prestige norm [t], and a low-prestige norm [t].
Correlation of /t/ with subjects' period of residence in the US,
accommodation to interlocutors’ speech, subjects’ gender, interlocu-

tors’ gender, and speakers’ attitudes were significant (p_<_ .05).

To my teacher

Prof. Julia S. Falk

ACKNOWLEDGEMENTS

Many people have helped me during the preparation of this thesis
in particular, and during my life as a student in general. To them I wish

to express my deepest gratitude.

I would like to express my sincere thanks to the chairperson of my
thesis guidance committee, Professor Julia S. Falk, for her guidance

throughout this study and her advice during some very trying moments.
Without her encouragement I would not have been able to turn my

preliminary work into a thesis topic. Her suggestions on the art of
formal writing will always be useful in my future academic endeavors.

I would like to acknowledge the guidance given by the other
members of my thesis committee. My sincere thanks to Dr. Nancy
Ainsworth-Vaughn for her very useful suggestions on the methodology,
and for her encouragement. My sincere thanks to Dr. Carolyn Harford for
her useful comments on the thesis, and for her patience and promptness
during a rather busy period of her life.

I thank Professor Thomas W. Juntune, Chair, for allowing me the
use of the facilities in the department to conduct this study.

I am deeply indebted to Dr. David Magier for introducing me to the
study of Indian English. He taught me much about the linguistics of
Indian English and other Indian languages for which I will always be
grateful..

I am also deeply indebted to Professor James P. Wang, former
Chair, for giving me an opportunity to study and to teach in this
department. I thank him for believing in me when I had no experience
to support my intention to study Linguistics. I will always remember
him for the kindness he showed to me knowing how I felt in a new
environment.

I would like to thank Dr. Grover Hudson for his useful suggestions
on the distributional analysis of this study in particular, and for his

IV

support at other times.

I am much obliged to the Asian Studies Center for funding this
study. I am especially grateful to Ms. Lillian Kumata for her support
throughout my study at Michigan State University.

I am very grateful to all my kind Indian friends who took time off
from their busy schedules to be my subjects. Without their contribution
this study would not have been possible. I also thank the interviewers
for working with me during their vacation.

It is with some degree of inadequacy that I try to express my
gratitude to my husband for his support during my life as a student. With

regard to this study in particular I thank him for being one of the inter-
viewers and for doing the entire statistical analysis.

Lastly, I would like to express my thanks to my teacher at

Bangalore University, Dr. M. A. Yadugiri, for first introducing Linguis-
tics to me. For her inspiration I am eternally grateful.

TABLE OF CONTENTS

PAGE

LIST OF TABLES x
LIST OF FIGURES xi
CHAPTER
I INTRODUCTION I
2 BACKGROUND 3
2.1 Introduction ................................. 3
2.2.1 Indian English ................................ 3
2.2.1.1 Subvarieties of Indian English .................. 4
2.2.1.2 Significance of English in India ................. 5
2.2.1.3 Characteristics of Indian English Relevant

to this Study .................................. 6
2.2.2 /t/ in the English of Indian Immigrants

in the United States ............................ 7
2.3 Studies Relevant to the Current Study ............ 10
2.3.1 Some General Concepts ......................... 10
2.3.2 Selection of Sociological Parameters ............ 12
2.3.2.1 Speakers’ Attitudes to the New

Socio-cultural Environment ..................... 12

vi

2.3.2.2 Period of Residence in the United States ..........
2.3.2.3 Accommodation to Interlocutors ................
2.3.2.4 Gender of Subjects (speakers) ..................
2.3.2.5 Gender of Interlocutors (interviewers) ..........
2.3.2.6 Subject-Interviewer Gender Interaction . . . . . . . . .

2.3.2.7 Subject-Interviewer Gender and Ethnicity

Interaction ...................................
2.3.2.8 Contextual Style ..............................
2.4.1 The Pilot Study ...............................
2.4.2 The Goals of This Study ........................
2.4.3 Hypotheses ...................................

3 METHODOLOGY

3.1 Introduction ..................................
3.2.1 Subjects .....................................
3.2.2 Interviewers ..................................
3.3 Procedure ....................................
3.3.1 Data Collection ...............................
3.3.1.1 Taping of Data ................................
3.3.1.2 Number of Sessions ............................
3.3.2 Major Factors Controlled in the Study ............
3.3.2.1 Physical Setting ...............................
3.3.2.2 Gender .......................................
3.3.2.3 Order of Interviews ............................
3.3.3 Elicitation of Data .............................

3.3.3.1 Interviews ...................................

vii

14
16
17
18
18

19
19
2O
24
25

3o
30
31
33
3s
3s
36
36
36
36
37
38
39
4o

3.3.3.2 Selection of Samples of Casual Speech ...........
3.3.3.3 Selection of /t/ within Samples... . . . , ......... .
3.4 Data Analyses .................................
3.4.1 Distributional Analysis ........................ .
3.4.2 Statistical Analysis ...........................
4 RESULTS
4.1 Introduction ..................................
4.2 Correlation of Sociological

Parameters and the Dependent Variable /t/ ......
4.2.1 Period of Residence in the US and /t/ ..........
4.2.2 Contextual Style and /t/ .......................
4.2.3 Accommodation to Interlocutors and /t/ .........
4.2.4 Speakers’ Attitudes to

their new Socio-cultural Environment and /t/ . . . .
4.2.5 Subjects' Gender and /t/ .......................
4.2.6 Interviewers’ Gender and /t/ ....................
4.2.7 Subject-Interviewer Gender Interaction and /t/ . . .
4.2.8 Subject-Interviewer Gender

and Ethnicity Interaction and /t/ ................ .
4.3.1 Statistical Significance and

Interpretation of Analysis ......................
4.3.2 Sociolinguistic Explanation of Results .......... .

viii

41
45
46
46
5 1

S6
S6

S6
56
6O
62

65
67
70
72

75

79-
81

5

5.1
5.2.1
5.2.2
5.3

5.4

CONCLUSION
Introduction ...................................
Suggestions for Improvement on Current Study . . .
Suggestions for Further Research ...............
Implications for Theories of Sociolinguistics
and Second Language (Dialect) Acquisition ........

Summary and Discussion .......................

APPENDICES
APPENDIX A: Questionnaire
APPENDIX B : Subject List

BIBLIOGRAPHY
General References

List of References

87
87

9O

92
94

96
96
97

98
98
99

Table 1.1 :

Table 1.2:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

Table 9:

Table 10:

LIST OF TABLES

Percentage of occurrence of [t] ................. 22
Accommodation, Contextual Style and [t1 ......... 22
The frequency of [F] in subjects’ speech .......... SO
Coding of nominal variables ...................... 52

Distribution of [t] and [t] in subjects speech

in relation to the period of residence of subjects

in the United States ............................. 58
Distribution of [t] and [t] in formal and informal
contextual styles in categories A, B, C, and D ...... 61
Distribution of It] and [t] when correlated with
accommodation to interlocutors’ speech ........... 63
Distribution of [ti and [t] in subjects’ speech

in relation to their attitude towards their new
socIo-cultural environment ...................... 66
Distribution of [t] and [t] between male and female
subjects in categories A, B, C, and D .............. 68
Distribution of [t] and [t] in the speech of subjects

in relation to interviewers’(interlocutors’) gender. .71
Distribution of [t] and [t] in male and

female subjects in speech exchanges with

male and female interlocutors (I) ................ 73

Table II : Distribution of [t] and [t] in the speech of male and
female subjects when speaking with male and female
interlocutors (INT) both Indian (1) and American (A) 76

Table 12: Statistical significance of independent variables

and interactions .............................. 80

LIST OF FIGURES

Figure l ............................................. 57

XI

CHAPTER 1

INTRODUCTION

This research is concerned primarily with phonological variation
and change In the English dialect spoken by first generation Indian im—
migrants in the United States. This dialect is acquired as a second
language by that population. The research reported here is limited to
the study of the particular phoneme /t/ in the English dialect of those
immigrants, which has been noted to show Interesting variation due to
the posssible influence of certain sociological factors. These socio-
logical factors, discussed in 2.3.2, are commonly acknowledged in the
field of sociolinguistics to affect the state of a language or a dialect,
so as to bring about variation in the speech patterns of individuals that
ultimately leads to change in the language or dialect over time.

Let us consider the attention given to dialectal variation in socio-
linguistic research. Scotton ( I 985) has this to say about the beginning
of research on dialectal variation.

. . . dialectal variation for motivations other than region of
origin has been a subject for systematic study only since
the 1960’s [its main aim being] the correlation between
linguistic variation and social group membership and its
role in explaining the mechanisms and rationale of language
change. (106)

Scotton (1985) further explains the theoretical bias that encour-
aged the study of dialectal variation in the following -

Within any language, dialects are the linguistic reflection
I

2
of the more static features affecting linguistic choices. .

. . these are features identifying speakers as members of
specific speech communities. For example, dialects are
indexical of the speaker’s home ties and socioeconomic
class memberships. That the bulk of linguistic variation
can be accounted for by such static features has been the
dominant premise within sociolinguistic theory (eg. Labov
I972; Fishman I972). (107)

Labov (1972:160) mentions two important studies conducted in
Martha’s Vineyard (Labov: I 972b) and New York City (Labov: l 966) which
discovered regular relations between parameters such as stylistic
variation and class stratification in the study of linguistic variables
due to dialectal variation within a community. Labov (1972:160)
asserts that “the most general issues that have been raised so far
concern the explanation of linguistic change” or more specifically,
sound change. Further, under ’A Strategy for the Study of Linguistic
Changes In Progress' the discussion leads to ‘the transition problem’
for which a solution is proposed, which is to find “the route by which
one stage of a linguistic change has evolved from an earlier stage”
(161).

In studying certain phonetic variants of /t/ in the English spoken
by first generation Indian immigrants in the United States, the ultimate
aim of this study is to trace if possible the different stages of sound
change in process of that variable /t/ in the dialect of that population
of immigrants. Hence, there is the need for a crossectional study of the
population of Indians who have acquired English formally as a second

language in India, and continue to use it in the United States.

CHAPTER 2
BACKGROUND

2.1 INTRODUCTION

The main body of this chapter is composed of three major sections
namely, 2.2, 2.3, and 2.4. Section 2.2 is further divided into two sec-
tions. Subsection 2.2.1 is a general description of the English language
as it is used in India today. Subsection 2.2.2 deals particularly with
characteristics of the phoneme /t/ In the English dialect spoken by
Indian immigrants in the United States. Section 2.3 Is an outline of
previous research on similar and related topics. It redefines some fun—
damental concepts in sociolinguistic theory in 2.3.1. Subsection 2.3.2
explains the rationale for the selection of the sociological parameters
that concern this research. Section 2.4 is comprised of three subsec-
tions. 2.4.1 is a description of a pilot study conducted in 1987 as part
of a research report for a sociolinguistics course. Subsection 2.4.2
discusses the goals of this study and suggests ways in which to improve
on the pilot study. Lastly, subsection 2.4.3 consists of the statement

of the hypotheses for this research.

2.2.1 INDIAN ENGLISH

Kachru (I988) considers the current sociolinguistic profile of
English in the world in terms of three concentric circles, namely, the
inner circle, the outer circle, and the expanding circle. The inner circle
represents regions that are “the traditional, cultural and linguistic

bases of English” (I)- for instance, the United States and the United

3

4
Kingdom. The expanding circle represents those regions where active

varieties of English are restricted in their use to “foreign language
contexts” and do not have official status, - for example, China and
Israel. Further, Kachru ( I 988) describes the outer circle as represent-
ing,
. . . the institutionalized non-native varieties of English
spoken in regions that have passed through extended periods
of colonization. In these countries, English may be an
official language used in educational, commercial and trade
institutions, but is generally not the native language of the
citizens. (I)

The Indian English dialect belongs to the outer circle.

2.2.1.1 SUBVARIETIES OF INDIAN ENGLISH
According to Kachru (I983),
. . . there is a cline [or gradient] of Englishes in India ranging
from educated (or standard) Indian English to varieties such
as Babu‘ English, Butler English, Bearer English, and Kitchen
English. (70)
Kachru (1983:70) quotes from Quirk et al. which is an apt assessment
of the state of English in India among other countries of the outer circle.
. we find an even spectrum of kinds of English, which
extends from those most like Pidgin to those most like
Standard English, with imperceptible gradations the whole
way along. (1972249)
The following statement by Kachru regarding the varieties of English in

India 18 appropriate here I

5
At one end of the spectrum we have educated English and at

the other end we have Kitchen English. Other varieties such
as Babu English, appear at various points on the spectrum.
(198370)
For the purpose of this study, only speakers of Standard Indian
English have been considered. Henceforth in the current study, all

references to the Indian English dialect are to Standard Indian English.

2.2.1.2 THE SIGNIFICANCE OF ENGLISH IN INDIA

The following information is available in Kachru (1983) which
argues for the relative significance of Standard Indian English within
India in comparison with other major Indian languages.

Although only 3% (roughly 17 million out of 566.66 million)2of the
Indian population are users of the Standard Indian English variety, when
compared with the percentage of speakers of some other major official
Indian languages such as Kashmiri, Assamese, Sindhi, and Sanskrit3,
this percentage is relatively high. English along with Hindi is used as
the official language of the Union. Among educated Indians, English is
used as the lingua franca by Indians to communicate with those Indians
who are native speakers of other Indian languages and, therefore, have
no other common language. In addition, English is used as the official
language in two states of North Eastern India, Meghalaya and Nagaland.

The medium of higher learning in most universities in India is
English. English is taught at various levels of education in all thirty one
states and union territories“ of India.

Finally, regarding the role of English in the Indian press, Kachru
(I983) affirms that the statistics of the 23rd Annual Report of the

6
Registrar of Newspapers of India (I979) show that, “the impact of

English is not only continuing but increasing" (71). Out of the total of
151,814 newspapers registered in India in 1978, 3085 (19.5%) were in
English. This percentage was second only to Hindi newspapers which
constituted 4196 (26.5%) of the publications. Also, while Hindi
newspapers had the highest percentage of circulation (23.8%), English
newspapers ranked next, their circulation being 22%. 53% of the total
number of periodicals published by the Union Government were in

English, while only 16% were in Hindi.

2.2.1.3 CHARACTERISTICS OF INDIAN ENGLISH RELEVANT TO THIS STUDY

According to the 1961 Census of India,5 Indo-Aryan and Dravidian
arethe two major language families; 87% of the Indian population Is
divided in the ratio of 3: I , respectively, as native speakers of languages
belonging to these two language groups. Languages of these two
families, which are spoken not only in India but in other countries of
South Asia, share a common phonological characteristic: the retroflex-
ion of stops. Thus, we find for instance, extensive usage of [t] and [d]
in languages which are either Indo-Aryan or Dravidian.

Kachru ( l 983) refers to certain generalizations that apply to South
Asian varieties of English (SAE), their basis being, solely, “the under-
lying features of the languages of the region" (27). In the phonetics of
SAE inclusive of Indian English, transfer or substitution of certain L2
(English) elements by LI (native language) elements occurs. One
exemplar of this substitution which is particularly relevant is the

substitution of the entire alveolar series of L2 by the retroflex series

of L,. This results in the substitution in Indian English of the phonetic

7
elements of /t/ by [t] in words such as, for example ‘particular’,

‘potential’, etc.5 While American or British English varieties use It]
among other allophones, Indian English uses It] with varying degrees of
retroflexion. The degree of retroflexion may vary from region to region
based on the native language of that region, and from person to person
based on the idiolect of that Individual. The degree of retroflexion may
also vary depending on the phonetic environment of /t/: for instance, [t]
in the Indian English word ‘little’ [litrl] will have a lesser degree of
retroflexion than the word ‘pot’ [pat] because, the preceding vowel [11in
‘little’ is a front vowel whereas [a] in ‘pot’ is pronounced further back
in the mouth. Here the preceding vowel progressively decreases or
increases the degree of retroflexion in each case of the following [t].

Retroflexion, among several other phonological features, disting-
uishes Indian English and other South Asian varieties from other English
dialects such as the Standard dialects of America, Britain, and
Australia. Hence, retroflex It] can be considered as a marker of ‘social
identity’ for the speakers of Indian English. Also, this particular
standard variety of Indian English is the prestige dialect for those
speakers of Indian English living in India. The existence of this
phenomenon in the phonetics of SAE and thereby in Indian English is the
critical criterion for the selection of /t/ as the dependent variable In

this study.

2.2.2 /t/ IN THE ENGLISH 0F INDIAN IMMIGRANTS IN THE UNITED STATES
Zwicky (1981) discusses the rationale for the transition within a
particular speaker from one dialect to another -

We grow up speaking the dialect of the region we live in and

8
the social group we live among. If we decide for some

reason to get rid of our native dialect - perhaps because its
. . . too unlike the way our friends speak - we can. .. replace
it by another dialect. (64)

On immigrating to the United States, at the very least there occurs
in the speakers of Indian English a transition in the perception of ‘the
standard set of norms' from Standard Indian English to Standard
American English.7 Assuming the above, various gradual sound changes
in the phonetics of English of those speakers may be predicted to fol low.
The sound change, if any, is an observable linguistic behavior facili—
tated by the unobservable personal/social motivation to assimilate to
the new dialectal environment. A minute part of this slow change may
be reflected in the speech of Indian English speaking immigrants in the
sound change in process from Indian English It] to the American English
It] and other allophones of the American English /t/.

Thus, at any given time, the speech of Indian English speakers in the
United States can be assumed to consist of It] as well as [t] in parallel
distribution. The degree of occurrence of either one of these free
variants within individual speakers may vary depending on time-
related, sociological, and/or socio-psychological factors such as pe-
riod of residence in the United States and thereby relative exposure to
the new dialect; gender of the speaker; gender of the interlocutor
(interviewer); ethnicity of the interlocutor; and speakers’ attitude
towards the new social setting and its dialect, reflected In their
attitude towards American culture, its people, and its lifestyle.

According to Hawkins (1984225), if the phonetic variation (as is

predicted in this case) occurs within individual speakers, the language

9
situation requires closer scrutiny. According to Zwicky (1981) -

Most of us make our style shifts largely within the bounds
of our native dialect, but some people become ‘bidialectal’.
They speak a standard dialect at work or with strangers, but
a regionally or socially marked dialect among friends and
“homefolks”. A bidialectal person can usually choose from
a range of styles within each dialect, but he can also change
his style by switching from one dialect to another. (64)

Hence, it is possible that the variation within individual speakers
is a result of true bidialectalism. It Is also possible as a result of
change in the ‘standard norm’ from Standard Indian English to Standard
American English that alveolar [t], along with other allophones of
American English /t/, becomes a “prestige norm’. The speaker uses It]
in ‘highly self-monitored’ (formal) speech, whereas the retroflex
Indian English It] Is used under conditions where ‘self-monitoring' does
not occur or occurs minimally, as in casual speech.

Labov ( 1972) refers to Sturtevant ( l 947: 74-84) which pertains to
the status of two phonemic variants before one overpowers the other
during the course of diachronic change.

Before a phoneme can spread from word to word . . . it is

necessary that one of the two rivals shall acquire some sort

of prestige. (3)
This statement is applicable to phonetic variants as well. However, an
objective data-based crossectional study of the population of Indian
immigrants is warranted in order to substantiate the above predictions
about /t/ In the English spoken by Indian immigrants in the United
States.

10

2.3 STUDIES RELEVANT TO THE CURRENT STUDY

2.3.1 SOME GENERAL CONCEPTS
Since the focus of this study Is the ‘variable’ /t/ and its phonetic

variants [t] and [t] which are “linguistic norms’ In the English spoken by
Indian immigrants, it would be appropriate at this time to consider
what Is meant by the terms ‘a linguistic norm’, ‘a linguistic variable’,
and ultimately, ‘a sociolinguistic variable’. Wald and Shopen (1981)
discusses the above terms In considerable detail. “ A linguistic norm is
any linguistic feature that occurs regularly In the speech of more than
one speaker in a community " (223). Accordingly, a standard norm is any
linguistic norm that is accepted and used as the standard by more than
one speaker in a community.

A linguistic variable is any linguistic unit realized by more

than one norm. The test for a variable is that one norm can

be substituted for another without any change in meaning.

. . . (1981:224)
The above test makes it more feasible to select phonological variables
for study, as they lack referential meaning unlike lexical and syntactic
variables. Lavendera (1978) discusses this methodological advantage
of choosing phonological variables for study in sociolinguistic research
in addition to the fact that the quantification of‘phonological variables
Is easier when the frequency of a variable is considered as an important
factor in linguistic analysis.

Both Lavendera (1978) and Wald and Shopen (1981) discuss the

importance of ‘same referential meaning’ in identifying a linguistic

l 1
variable and, therefore, the methodological advantage of selecting

phonological variables so that the results of the study are clear-cut and
not confounded. For instance, in the nasal phonetic variants [en~in] and
[in] In ‘laughing’ there is no change in the referential meaning of the
word. Lavendera (I978) considers Labov (1966) as being superior to
previous sociolinguistic research on Language Variation and Sound
Change, due to the successful isolation of cases of free variation on the
basis of similarity In referential meaning, or rather lack of it, of the
variants of the variable. Wald and Shopen (1981) defines a sociolin-
guistic variable as -
. . . a linguistic variable sensitive to social context. . . . we
have a sociolinguistic variable when some feature of a
social group or a social situation allows us to predict which
norms will be used from a variable. . . . characteristics of
speakers and listeners, and situations and topics, as well as
historical factors that might explain changes in linguistic
standards (norms), . . . [are used to explain language behav-
ior]. (227) ‘
Lavendera (1978: 176-177) refers to Labov (1966) which defines a
sociolinguistic variable based on the following criteria.
a. items which are high in frequency and can be easily quantified on a
linear scale;
b. items which are integral units of larger structures and are immune
to conscious suppression; and
c. in general a sociolinguistic variable must be socially stratifiable
within the community “should suggest an assymetric distribution over

a wide range of . . . ordered strata of society” (Labov 197228).

12
/t/, the Iinguisticvariable discussed In section 2.2.2, was selected

based on the above criteria and thereby may be defined as a sociolin-
guistic variable. Additional substantiation of the definition can only be
provided by an empirical study. If the phonetic variants are components
of a linguistic variable, a sociolinguistic variable must consist of at
least two phonetic variants which are themselves sensitive to social
context and, therefore, reliable indicators of sociolinguistic phenom-
ena. Thus, a phonetic variant of a sociolinguistic variable can be
considered as a sociolinguistic variant of that variable if it fulfills the
same criteria as a sociolinguistic variable in terms of frequency,

stratifiability, and quantifiability.
2.3.2 SELECTION OF SOCIOLOGICAL PARAMETERS

2.3.2.1 SPEAKERS’ ATTITUDES TO THE NEW SOCIO-CULTURAL
ENVIRONMENT

Let us consider firstly speakers’ attitudes toward the environ-
ment, its people, its culture, its lifestyle, and its role, either (a) in the
transition - the sound change in process from Indian English to
American English, or (b) In the effort by Indian immigrants to become
truly bidialectal (acquire the American English Dialect while retaining
the Indian English variety). Two studies in particular are pertinent.

Labov (1972:1-42) explains the pattern of occurrence of the
centralized diphthongs [aw] and [ay] in the speech of the population of
Martha’s Vineyard, an Island off the coast of Boston, In the following
manner. The centralized diphthongs are markers of social identity. The

frequency of their occurrence is higher in the speech of those natives

13
of Martha’s Vineyard who are hostile toward mainland Bostonians, “the

summer people” who come as tourists and buy property on the island
that has belonged to the island’s settlers for centuries. In contrast,
those who are not hostile to mainland Bostonians and those who intend
to leave the island for jobs on the mainland have a lower frequency of
centralized diphthongs in their English dialect.

Falk (1979) discusses the following concepts in Gardner and
Lambert (I972); ‘Integrative motivation’ as opposed to 'Instrumental
motivation' is an important prerequisite for adults to successfully
master a foreign language.

Integratively motivated learners study a language because
of an intense, personal desire to become familiar with, or
even part of, the society in which the language Is used; there
are strong attachment and positive feelings toward the
people and the culture represented by the language. Instru-
mental motivation on the other hand is more utilitarian. .
.. Integrative motivation almost invariably leads to greater
success In foreign language learning than does instrumen-
tal motivation. (445)

Extending the concepts of integrative motivation and Instrumental
motivation to foreign dialect acquisition, and applying the rationale
used in Labov (1972) and Falk (1979), one may presume that Indian
immigrants in the United States who possess Integrative motivation
and, thereby, a positive attitude to native Americans and their socio-
cultural environment will acquire American English to a greater extent.
Conversely, those who have a negative attitude or possess only instru-

mental motivation, will either not acquire American English or acquire

14
it to a minimal extent. This can be observed directly in the greater

frequency of one or the other of the corresponding phonetic variants
under consideration. Hence, the experimental design must include a
method by which to quantify speakers’ attitude as either positive or

negative.

2.3.2.2. PERIOD OF RESIDENCE IN THE UNITED STATES
Closely related to attitudes towards speakers of a particular
dialect is the matter of the speakers’ attitude to the dialect spoken.
Speakers are known to have ‘opinions’ and ‘feelings’ about their own or
other’s dialects. Fasold( I 984) mentions Bailey ( l 973) which states an
important factor that facilitates sound change. -
The course of sound change is apparently influenced by
whether the change is favored or disfavored by the speech
community. (148)
Therefore, language (dialect) attitudes and the period of residence of
informants In the United States must be considered as two related
factors contributing to sound change, if any. A sound change may occur
over a period of time in the speech of Indian immigrants from Indian
English to American English, utilized minimally, for the purpose of
crosscultural communication.
According to Giles (1979) mentioned In Bourhis (I984) -
More often than not, it is the high-prestige language of the
dominant group that is deemed most appropriate for cross-
cultural communication. (35)
Within the minority speech community of Indians in the United

States, the sound change is favored as the American English phonetic

15
variant [t] is the prestige-norm and therefore, the ‘favored norm’.

Studies on dialect attitude research have shown that the standard
variety used by the dominant group Is considered the high prestige
dialect even by those speakers who strongly identify with the nonstan-
dard regional variety. Fasold (1984:158) mentions (a) d ‘Anglejan and
Tucker (1973:22) -

. . . attitude research involving European French and two va-

rieties of Canadian French in Quebec, found that the Euro-

pean French speaker was rated not only more intelligent and

better educated, but more likeable than the Canadian French

speaker [based on the respective dialects they spoke].
(b) Carranza and Ryan (1975:99) show that both Anglo and Mexican
Americans rated English higher than Spanish on status scales and on
solidarity scales. In addition, the sound change from ‘Indian’ to
‘American’ is favored by the American society in which the immigrants
are a minority community. Thus, the sound change must take place in
that favored direction.

A crossectional study of the population can be used to test this
hypothesis by considering the period of residence of each subject in the
United States. In other words a relative increase In the length of time
in the United States of a subject can be shown to affect a sound change
in speech, reflected In the greater frequency of the American English
phonetic variant [t] and, correspondingly, a lower frequency of the
Indian English phonetic variant [t]. Allowing for length of time, during
which the sound change is likely to occur in the speech of those who are
the minority group, the extent of occurrence of either of the two

variants depends mainly on which one of the following psychological

I6
factors dominates within each subject: the need to use the prestige

norm or the need to identify with the regionally acquired non-prestige
norm. This may be studied only by correlating /t/ with the independent

variable period of residence in the United States.

2.3.2.3 ACCOMMODATION TO INTERLOCUTORS

Dialect choices, in addition to reflecting the language attitudes of
speakers, can, as mentioned earlier, “also reflect speakers’ like or
dislike of their interlocutors as individuals or as group members”
(Bourhis 1984235). A positive attitude towards interlocutors Is often
reflected when speakers adapt or ‘accommodate’ their speech toward
their interlocutors. Bourhis (1984) refers to several studies confirm-
ing the above phenomenon of ‘interpersonal accommodation’ -

Instances in which speakers tend to adopt the speech

patterns of their interlocutors have been shown to occur at

various linguistic levels including speech rate, vocal inten-

sity, speech silences, regional accents, and language

switches (Giles and Powesland. 1975). Such switches,

known as speech convergence, not only can allow for
efficient communication but may also reflect speakers'
conscious or unconscious need for social integration with

their interlocutor (Giles I979). (34)

Interpersonal accommodation may occur in speakers of Indian
English in the United States due to the “need for social integration"
within the American society. The occurrence of this phenomenon may
be indicated in the greater usage of the American English variant [t1

when the interlocutor is a native American, and a greater usage of the

17
Indian English variant [t] when the interlocutor is another Indian

immigrant. Ainsworth ( I 974:1 14) refers to Ervin-Tripp ( 1 964a) which
finds Japanese-American subjects showing greater interference from
Japanese when speaking English to a Japanese test administrator than
when speaking English to an American test administrator.

Thus, the experimental design requires the choice of American as
well as Indian interviewers as Interlocutors. The data gathered,
consisting of speech exchanges between Indian subjects and Indian
Interlocutor verses Indian subjects and American interlocutor, could be
considered as comparable parameters of the independent variable -

accommodation.

2.3.2.4 GENDER OF SUBJECTS

Trudglll (1974), on gender differences affecting language use,
affirms on the basis of previous studies carried out In Britain and
America, that

. . . allowing for other factors such as social class, ethnic
group, and age, women consistently use forms which more
closely approach those of the standard variety or the
prestige accent than those used by men. (85)

Thus, variables may be correlated with the sex of speakers to
illustrate sex differences in speech patterns. By balancing the data
equally for gender It would, therefore, be possible to logically verify
the above theory in the English dialect of Indians in the United States.
On the basis of the above claim, the tendency in the use of the phonetic
variants of /t/ by each sex is predictlble. Sex differences, if any, are

substantiated by the greater or lesser frequency of one or the other

norm by either sex.

2.3.2.5 GENDER OF INTERLOCUTORS

Sex differences in speech, In addition to sex of the subject
(speaker) may be studied with the parameter of sex of the interviewer
(interlocutor) and its effect on the speech of subjects of both sexes.
Wolfson ( I 976) has pointed out that sex of the interviewer may affect
the verbal behavior of the informant (speaker) (208). Thus, the

experimental design contains interlocutors of both sexes.

2.3.2.6 SUBJECT-INTERVIEWER GENDER INTERACTION

Ainsworth (I974) raises the issue in an experiment which uses

bilingual (Spanish and English) children:
...to control for sex of the administrator... would have been
desirable; [but] to Introduce four administrators with the
appropriate sex-ethnic identity combinations would have
greatly complicated the research design. (I 15)

Therefore, to study subject-interviewer gender interaction, the
research design for this study requires the following sex-ethnic
Identity combinations of interviewers; a male Indian, a female Indian,
a male American, and a female American. Studied in this manner,
subject-interviewer gender interaction reflects significant differ-
ences, if any, in the speech exchanges between subjects and interview-

ers of the same sex as well as the opposite sex.

19
2.3.2.7 SUBJECT-INTERVIEWER GENDER AND ETHNICITY INTERACTION

Balancing gender and ethnicity of speakers and Interlocutors
provides, additionally, comparative parameters subsumed within
subject-Interviewer gender and ethnicity interaction. The comparison
within these parameters could indicate in the occurrence of the
variants [tI/[t] the greater or lesser influence, due to the interaction of
one of three independent variables, namely, subjects’ gender, inter-
locutors’ gender and accommmodation in the speech of subjects result-
ing from variance in the ethnicity of interviewers. The ethnicity of all

speakers (subjects) is Indian and is, therefore, constant.

2.3.2.8 CONTEXTUAL STYLE

Trudgill (1974:100-122) argues, based on studies which discuss
linguistic variation and contextual style, that there are two types of
situational variation, the difference between them being more one of
degree than of kind.
(a) Variation within dialects occurs when situational shifts along the
scale of formality produce changes In pronunciation within the limits
of a dialect.
(b) Variation between dialects which occurs within the speech of
bilinguals. In cases of bidialectal switching according to situation,
according to Trudgill (1974:112), “one dialect will occur in formal
situations and another In informal situations.” The more formal
situation demanding the prestige variety is used among strangers while
the non-standard regional dialect Is used among friends and “home-
folks” (Zwicky 1981).

Thus, it may be assumed that speakers of Indian English in the

20
United States would use the American English It] in an attempt towards

bidialectalism, in formal contexts, while retaining the Indian English
norm [t] for use among “homefolks ”. This assumption can be illustrated
by isolating significantly distinct speech styles - formal and informal,
and studying the frequency of occurrence of the norms [t] and [t] In those
styles in the speech of subjects. This is possible to achieve using
Labov's technique for isolating distinct speech styles mentioned in
Trudgill ( I 974: I 08) and discussed in Labov (l 972270-109). By control-
ling the degree of self-monitoring, the amount of attention paid to one's
own speech at any given time, significantly distinct styles ranging at
different levels on a scale of formality/informality may be isolated
(refer to 3.3.3 of this thesis for isolation of contextual styles).

In conclusion, the following are the sociological parameters
chosen based on previous studies that either suggest or use such
variables to study phonological variation: speakers’ attitudes to the
new socIo-cultural environment, period of residence In the United
States, accommodation to interlocutors, gender of subjects, gender of
interlocutors, subject-Interviewer gender interaction, subject~
interviewer gender and ethnicity interaction, and contextual style. The
effect of these parameters on the verbal behavior of the speakers is

discussed In chapter 4 Of tI'IIS thesis.

2.4.1 THE PILOT STUDY

A preliminary study was conducted in May-June 1987 using fifty
two adult Indian English (IE) speaking informants residing in and around
areas of Lansing and Detroit, Michigan.

The purpose of the study was to examine the distribution of the

retroflex [t] in the speech of the Informants when correlated with

21
independent variables such as period of residence, contextual style, and

accommodation to American and Indian English speakers. While formal

speech was elicited from the subjects by their reading of a word list,

informal speech was elicited by asking the informants personal ques-
tions during the course of an interview. Both word list readings and
answers to personal questions were tape recorded.

The total number of informants was divided into four categories
based on the period of residence of each subject in the United States.
Despite imperfect balancing (uneven distribution) in the data for
subjects’ gender and number within these four categories, the study
elicited some significant results.

a. The study showed that [t] is sensitive to social parameters and
therefore is a sociolinguistic variant of the variable /t/. It is a
reliable indicator of a slow change In time in the speech of first
generation Indian English speaking immigrants, from Indian Eng-
lish to American English.

b. The study indicated, more specifically, a changing standard norm
from [t] to [t] provided by the inversely proportional relationship
between the occurrence of the variant [t] and the length of
residence in the United States of each informant.

c. Additional substantiation of this change in the standard norm was
given by the presence of a greater frequency of [t] in informal
speech and a relatively lower frequency of [t] in formal speech. The
American English variant It], being the ‘prestige norm’, occured
more In the formal speech style.

d. The study also suggested the speaker’s need to integrate within the

new society, reflected in the English of a majority of IE speakers

22

Table 1.1 : Percentage of occurrence of [t]

 

 

 

 

CATE- TIME OF RESIDENCE INDIAN INTERVIEWER AMERICAN INTERVIEWER

GORY IN THE US (in yrs.) (I FEMALE) (2 FEMALES)
INFORM AL FORMAL INFORM AL FORMAL

I 0 - 2 69.30 64.30 53.50 50.00

2 2 - 5 55.00 52.80 52.80 46.40

3 5 - 10 46.25 41.25 50.80 44.10

4 10 & ABOVE 26.25 2 I .90 10.00 9.00

 

 

 

 

 

 

Table 1.2 : Accommodation, Contextual Style and [t]

 

 

 

 

SPEECH INTERVIEWER B INTERVIEWER A
STYLE (INDIAN) (AMERICAN)
INFORMAL

(questions) 49.20% 41.78%
FORMAL

(word list) 45.06% 37.38%

 

 

 

 

 

e.

23
who accommodated to American English speakers by lowering the

occurrence of [t] in their speech. When Indians interacted with
another Indian, a higher frequency of It] was noted. This was
attributed to speakers’ attitude towards American culture, its
people, etc.; and the relative linguistic security of the informants,
resulting from their relative socioeconomic stability in the ‘alien’
society. Table 1.1 illustrates the results of this study.

Finally, the results suggested that accommodation to American
English interlocutors was a more influential factor in lowering the
occurrence of [t] in IE speakers than formal speech context. This
is illustrated in the figures In Table 1.2. Formal speech with the
Indian English speaking Interlocutor (interviewer) produced a
higher percentage of [t] (45.06%) than Informal speech with Ameri-

can English speaking interviewers (42.78%).

Disadvantages of the Pilot Study:

1.
2.

The physical setting for data collection was uncontrolled.
The data were not balanced for gender within the four categories

Into which the informants were grouped.

. The number of informants in each group varied depending on avai l-

ability. The results were quantified In terms of total percentages.
Three female interviewers were used to study accommodation to
Indian English speakers verses accommodation to American Eng-
lish speakers (see Table 1.1). Ideally, four Interviewers could have

been used with the appropriate sex-ethnic combinations.

5. The data for studying accommodation were gathered from different

informants. In order to control for personality of subjects, the

24
same informants would have been desirable.

6. Finally, the word list used to obtain samples of formal speech was
too revealing. The specific variable being studied needed to be
concealed so as to avoid conscious self-monitoring by the subjects.
Hence, a study controlling the above mentioned factors under more

ideal conditions was warranted for examining the variable /t/.

2.4.2 THE GOALS OF THIS STUDY

The principal aim of this study is to examine in detail the socio-
linguistic distribution of the variable /t/ in the English dialect of
Indian immigrants in the United States. The general goal of the study
is to explain the variation and thereby change in /t/ from [t] to [t] in the
dialect spoken by first generation Indians in the US. This variation
occurs due to the effect of certain social factors. This in turn results
in the sound change which is currently in process in the speech of these
speakers, from [t] to [t]. The study of variation, therefore, involves
determining those factors that motivate speakers to use the opposing
phonetic variants of the variable /t/ which are [t] and [t].

This study is intended to Improve and expand on the preliminary
study (2.4. l ). To be able to obtain results that are unbiased, the current
study contains in its experimental design an equal number of male and
female informants. Within a sample of 24 subjects, it- is possible to
balance the data gathered for number and gender, with 6 subjects in
each category - 3 male and 3 female informants. Also, a sample from
the population of Indian English speaking informants can be selected in
a manner so as to achieve even distribution of subjects within the

aforementioned four categories in which the subjects are grouped.

25
Four interviewers are used in this study with the appropriate sex-

ethnic combinations to study accommodation to interlocutors of Indian
and American ethnic backgrounds, subject-interviewer gender and
subject-interviewer gender and ethnicity Interactions. The four inter-
viewers Interview all twenty-four subjects. In this manner, any change
in the subjects’ speech patterns due to accommodation could be noted.

Finally, the questionnaire includes a phrase list instead of the word
list used in the pilot study, so that the variable /t/ under consideration
could be disguised.

In addition to examining the variable /t/ in the speech of Indian
immigrants in the United States in relation to factors such as period of
residence of the speaker, accommodation to interlocutor, and contex-
tual style, the current study also correlates /t/ with factors such as
gender of subjects (informants), gender of interviewers (interlocu-
tors), speakers’ attitudes, subject-interviewer gender interaction, and
subject-interviewer gender and ethnicity Interaction. The effect of
these factors, If any, can be observed in the change In frequency of
occurrence, within the speech of informants, of the two opposing

phonetic variants [t] and it] of the variable /t/.

2.4.3 HYPOTHESES

The following are the hypotheses which need to be tested concern-
ing the distribution of the variable /t/ and its two phonetic variants [t]
and [E]. -

If /t/ is a sociolinguistic variable consisting of two opposing
sociolinguistic variants [t1 and [t] in the English spoken by first

generation Indian immigrants in the United States -

26
a. the frequency of occurrence of [t] is inversely proportional to the

period of residence in the United States of the informants. Con-
versely, the frequency of occurrence of It] Is directly proportional
to the period of residence of the Informants in the United States.

b. [t] is the ‘high-prestige norm’ in the society in which the informants
currently reside. Therefore, due to self-monitoring within one’s
speech, the frequency of occurrence of [t] is higher in formal speech
styles and lower in informal speech styles. As It] is the ‘low-
prestige norm’, the opposite is true for [t].

c. Due to accommodation of speakers to their interlocutors’ speech,
the American English variant It] occurs more frequently in infor-
mants’ speech when speaking with an American English speaking
interlocutor and less frequently when speaking with an Indian
English speaking interlocutor. Inversely, the Indian English variant
[t] occurs more frequently when speaking with an Indian English
speaking Interlocutor than when speaking with an American Eng-
lish speaking Interlocutor (interviewer).

d. Informants’ (subjects’) positive attitude towards American soci-
ety, its people, its culture, and its lifestyle, is reflected in the
higher frequency of [t] and a lower frequency of [t]. Informants’
negative attitude towards American society is reflected in the
relatively higher occurrence of the IE variant It] and a lower
occurrence of the American English [t].

e. Allowing for other sociological factors, in general, the frequency
of [t] is higher in the speech of male Indian informants than female
Indian informants. Inversely, the frequency of [t] is higher in

women’s speech than in men’s speech.

27
f. The total percentage of [t] is higher in the speech of informants

when the interlocutor (interviewer in the experiment) is male than
when the Interlocutor is female. The inverse Is true of [t].
g. Due to interaction of subject-interviewer gender, the percentage
of it] decreases in descending order as follows. The subject (5)-
interviewer (I) combinations are -
Male S/Male I > Male S/Female I > Female S/Male I > Female S/
Female 1.
The percentage of It] increases in the following order of subject
(S)- interviewer (I) combinations:
Male S/Male I < Male S/Female I < Female S/Male I < Female S/
Female 1.
h. Due to subject-Interviewer gender and ethnicity interactions, the
percentage of It] decreases in the following descending order of
subject(S)-interviewer(l) gender and ethnicity cominations:
Male S/Male Indian I > Male S/Male American I > Male S/Female
Indian I > Male S/Female American I > Female S/Male Indian 1 >
Female S/Male American I > Female S/Female Indian I > Female S/
Female American I.
The opposite is true for It], therefore it] increases in the following
orden

Male S/Male Indian I < Male S/Male American I < Male S/Female
Indian I < Male S/Female American I < Female S/Male Indian I <
Female S/Male American I < Female S/Female Indian I < Female S/

Female Indian I.

28
NOTES

.Babu - a pejorative term used by the British in India before 1947 to
refer to a westernized Indian.

. Today, according to the International Institute for Population
Sciences, India’s population Is 800 million.

. Sanskrit does not have a spoken component used for the purpose of
everyday speech yet the Eighth Schedule of the Indian Constitution
includes Sanskrit as one of the official languages and as the native
language of 2544 speakers.

. Territories directly governed by the center at New Delhi headed by
the President, unlike the states which are headed and governed by
the chief minister as the head of state.

. Due to non-availability of recent census information I use the 1961
census. Kachru (1977) refers to the 1961 census of India.

. Sahgal and Agnihotri (1988) conclude from their study of fortyfive
subjects from a "relatively affluent and smart area of South Delhi“
that "retroflexion of alveolar stops is not a pan-Indian feature of
Ind E,- In fact, educated urban middle class speakers generally use
a non-retroflexed variety."

Non-retroflexion Is perhaps a particular feature of the middle class
In South Delhi. A comparative study of other urban communities is
required to apply this trend to the speech of all urban middle class
Indians. If this finding is true in the speech of all urban middle class
Indians (the social strata most likely to come to the United States
for higher education) it is possible that there is an increase in the

use of [t] during the initial period of residence in the United States.

- 29
As mentioned in Labov (1972 : 29), phonetic differences become

stronger as the group whose speech is experiencing a sound change
fights to maintain its identity.

7. This assumption that speakers minimally perceive the difference
between It] and [t] was substantiated when subjects were told about
the specific intent of the study after the interviews and they agreed
that Indians pronounced /t/ differently from the [t] of American

English.

CHAPTER 3

METHODOLOGY

3.1 INTRODUCTION

This chapter presents the experimental method and its underlying
principles. The discussion includes three major topics In sections, 3.2,
3.3, and 3.4.

Section 3.2 is a detailed description of the subjects (3.2.1 ) and the
interviewers (3.2.2) who participated In the experiment.

Section 3.3 Is a discussion of the procedure followed in conducting
the experiment. This section has three major subdivisions. Subsection
3.3.1 discusses the data collection which subsumes taping of the data
as well as the number of sessions taken to collect the data. In
subsection 3.3.2, I describe the major factors that were controlled
during data gathering, and the rationale for doing so. Subsection 3.3.3
is a discussion of the principles and the manner in which the relevant
data were elicited from the subjects. This subsection discusses the
format of the interviews, and the selection of the appropriate speech
samples for analysis.

Finally, section 3.4 consists of the analyses of the data, made up
of two subsections. Subsection 3.4.1 discusses the distribution of the
various phonetic components of /t/ in the data. Amore detailed account
of It] and [t], the two sociolinguistic variants when correlated with
other social parameters and their results, is given in Chapter 4.
Subsection 3.4.2 Involves a discussion of the statistical tests used to
provide additional substantiation for the results of this experiment.

30

31

3.2.1 SUBJECTS

Out of the population of first generation Indian immigrants in the
United States, 24 subjects, 12 male and 12 female, were selected for
this study.

The subjects were chosen based on the following criteria -

a. The subjects possessed a speaking knowledge of English at the time
the study was conducted.

b. The subjects were first generation Indian Immigrants currently
residing in the United States.

c. All subjects had resided in India from the time of their birth until
at least eighteen years of age, after which they had Immigrated to
the United States. Therefore, their first exposure to the English
language was the Indian English variety.

(1. Consequently, all subjects had acquired English formally, as a
second language in India and not In any other country. All subjects
were native speakers of Indian languages which belonged to the
Indo-Aryan or Dravidian language families only. The subjects were
native speakers of one of the following languages - Hindi, Punjabi,
Marathi, Oriya, Bengali, Assamese, Gujarati, Tamil, Telugu, or
Malayalam.

All subjects currently resided in or around Lansing, Michigan. 18
out of 24 subjects were either students or faculty members at Michigan
State University. All students were graduate students. The remaining
6 were permanent residents of the United States.

The pilot study (2.4.1) showed clear differences, In the results

obtained, between each of the four groups which were defined based on

32
the period of residence of each subject In the US. Therefore, the same

guidelines as In the pilot study were used, to group the 24 subjects
according to their respective time of residence in the US. The subjects
were divided Into four categories,

Group A: 0 - 2 years

Group B : 2 - 5 years

Group C : 5 - l 0 years, and

Group D :10 years & above.
6 subjects, 3 male and 3 female, were selected in each of the above
groups. An attempt was made to achieve even distribution of subjects
within each category and within the entire list for gender and period of
residence in the United States. It was possible to attain some degree
of even distribution when selecting subjects with respect to period of
residence within the above groups A, B, and C. This, however, became
increasingly difficult within Group D. The period of residence of
subjects in Group D ranged between I I years 6 months and 35 years with
uneven spacing between subjects’ period of residence (refer to list of
subjects In the appendices).

Considering the limitations In experiments involving human sub-
jects, gender was equally distributed within each category and In the
entire list of subjects; 12 male and 12 female informants were equally
distributed within the four groups with 3 male and 3 female subjects
in each group. However, gender was not evenly distributed within the
four categories. By even distribution for gender, I refer to the regular
alternation of male and female subjects within each category and
throughout the subject list (refer to subject list In the appendices).

Whenever possible, however, even distribution of subjects was

33
achieved.

3.2.2 INTERVIEWERS (INTERLOCUTORS)

Wolfson (1976) has pointed to the fact that sex, age, occupation,
attitude and/or dialect of the interviewer may affect verbal behavior
of the Informant<208). Brouwer, Gerritsen and De Haan( 1979), as a con-
clusion to the study, strongly advocates the significance of the inter-
viewer In sociolinguistic research In the following -

Sociolinguistics must start from the principle that the
selection of the interviewer needs as much attention as the
selection of the informants, and. . . both need a similarly
detailed description In the research report . . . . (49).

In this study, four interviewers were chosen at first. Later, due to
non-availability of the female native American Interviewer after the
first session, a second female native American was used.

Accommodation to interlocutors (2.3.2.3), subject-Interviewer
gender interaction (2.3.2.6), and subject Interviewer gender and ethni-
city Interaction (2.3.2.7) were three sociological parameters chosen
for this study, due to the possibility that the ethnic group Identity,
gender of the interviewers, and gender of the subjects may affect the
verbal behavior of the subjects. To be able to study the above in detail,
interviewers of the following four sex-ethnic Identity combinations
were selected : one male Indian English speaker, one female Indian
English speaker, one male native American English speaker, and one
female native American English speaker. This selection of interview-
ers, therefore, controlled for two major factors, namely, sex and

dialect of the interviewers, which could affect the verbal behavior of

34
the 24 subjects to be Interviewed.

Other variables of the Interviewer were accounted for, in the
following manner.
a. SocIo-economic class: This factor was controlled for by selecting
all interviewers from the same socIo-economic strata of society. All
Interviewers were either senior undergraduate or graduate students.
b. Affect: The personality factor was not possible to control within the
interviewers due to the uniqueness of each individual. However, an
attempt to control the attitudes of all the interviewers towards the
subjects was made by selecting only those Interviewers who had prior
interviewing experience and/or relative experience In communicating
with ethnic minorities In the United States. Interviewers were either
personally acquainted with people from such minority groups or lived
among members of these groups. The Interviewers lived in student
housing at the university, which Is largely an international community.
Thus, they were not unfamiliar with ‘foreigners’. The native American
Interviewers were noted as having a highly positive attitude towards
the subjects, Indicated by the enthusiasm they expressed for having the
opportunity to interact with people from a different ethnic background.
The Indian Interviewers also had a highly positive attitude towards the
subjects expressed In their desire to meet people of their own ethnic
' background.
c. Age: The age range of the interviewers was between 23 years and 32
years. No deliberate effort was made to select interviewers of the
same age. But an effort was made to select an interviewer whose age
was close to that of the others. It was difficult to find interviewers

who fulfilled all the above criteria and in addition were available for

35
a month or more, during which time the data for this study were to be

gathered. Hence, among the available interviewers an arbitrary age
range was selected which was also representative of the ages of
persons who were most willing to be hired on an hourly basis.

Thus, some major variables of the interviewers that could be
Identified and could possibly affect the subjects’ speech patterns were
accounted for. The idea was to attain uniformity; that Is, if a particular
variable of the Interviewer could affect the speech of the subjects, by
maintaining a certain degree of uniformity within the interviewers, one
could predict that the data would be affected In a similar manner. Also,
the effect of an unforeseen variable if any, could be nullified within the
experiment. However, because factors such as personality cannot be
controlled, perfection can never be achieved In an experiment involving

human subjects or interviewers.

3.3 PROCEDURE

3.3.1 DATA COLLECTION

The data gathered for the current study consists of a total of 96
interviews with 24 subjects, I.e., four interviews with each subject.
All subjects were interviewed by interviewers of all four sex ethnicity
combinations, that Is, once each by a male Indian English speaker, a
female Indian English speaker, a male native American English speaker,
and a female native American English speaker, but not necessarily In
that order.

The total time taken to Interview 24 subjects was approximately

14 hours. The average length of time per Interview was 35 minutes.

36

3.3.1.1 TAPING OF DATA

All data were recorded on tape by the Interviewers themselves,
with the prior consent of the subjects. The four tape recorders used for
this purpose were of different name brands. Therefore, they were
utilized in such a manner that no tape recorder was used for the same
subject twice. In this way any possible discrepency In the quality of
recording of one particular set of interviews was avoided. The tapes
used were high frequency 90 minute casettes manufactured by Sony

Magnetic Products Inc. (Japan).

3.3.1.2 NUMBER OF SESSIONS

The data were collected during three sessions between 17th June
and 28th July 1988 on the Michigan State University campus. Each of
these sessions was arranged largely on the basis of availability of the
interviewers and subjects at that particular time. The interviews were
set up so that every subject was Interviewed by the different inter-
viewers In a single session. The first session was for a duration of
8 1/2 hours, the second session for 3 1/2 hours and the final session for

2 hours.
3.3.2 MAJOR FACTORS CONTROLLED IN THE STUDY

3.3.2.1 PHYSICAL SETTING

Four classrooms were assigned for data collection in the Depart-
ment of Linguistics and Germanic, Slavic, Asian and African Languages,
Wells Hall.

37
Each of the interviewers was assigned a particular classroom. The

subjects were directed to the different rooms until each subject had
completed all four Interviews, so that a subject was Interviewed in a
room only once. Both male and female subjects were interviewed in the
same four rooms. In this way the physical setting was controlled for
both male and female subjects.

The furniture in the classrooms was placed in a manner so as to
create an Informal atmosphere.

Although the rooms were not sound proof, the noise level was
minimized by keeping all doors closed during the course of each
Interview. Also, during the summer no classes were being conducted on
the floor in which the classrooms were located. Therefore, the noise

level was minimal.

3.3.2.2 GENDER

The possible effect of gender as a factor Influencing the verbal
behavior of subjects was controlled by considering two aspects,
namely, (a) gender of subjects, and (b) gender of Interviewers.

(a) The possible effect/s of the gender of subjects, was controlled by
selecting an equal number of male and female subjects for the
study; 12 female and 12 male subjects. In addition, within the four
groups A, B, C, and 0,3 male and 3 female subjects were selected
for all four groups (refer to appendices for list of subjects/
gender)

(b) The possible effect/s of the gender of interviewers was accounted
for, by choosing 2 males and 2 females as interviewers, each

consisting of one American and one Indian interviewer.

38

3.3.2.3 ORDER OF INTERVIEWS

For the purpose of analysis, the data were divided Into four sets of
Interviews, grouped under each of the four sex-ethnicity combinations
of Interviewers. For all subjects the four Interviews were identical,
that is, the format of the questionnaire for each interview was
Identical so as to maintain a certain uniformity of context for all
subjects.

The manner in which questions were answered and the specific
choice of questions varied from Interview to Interview with some
degree of similarity between subjects, and a greater degree of similar-
Ity within each subject. But human nature being as It Is, answering
identical questions and reading the same phrase list four times was
bound to be disinterestlng to the subjects In the long run, resulting in
subject fatigue. This in turn would result in an imbalance In the data
from each subject between successive Interviews with increasing
disinterest and practice due to learning In progressive order. “To
counterbalance the effects of subject fatigue and learning due to
practice, the order of the four successive Interviews for every subject
was arranged so that every fourth consecutive subject starting from
the top of the subject list (refer to appendices for subject list) was
interviewed In the same order. In this manner, the number of shortened
uninteresting interviews due to repetition of the substance of speech
was equally shared between the four sets of interviews, thereby
neutralizing the adverse effects of ‘subject fatigue' and “the practice

effectfi

39

3.3.3 ELICITATION OF DATA

Brouwer, Gerritsen and De Haan( l 979) discuss the various methods
used to gather data for research on sex-based differences in language
use. The discussion of the method of elicitation in particular is relevant
as this method was utilized to obtain the data for the current study. The
specific Intent of data collection was concealed from the subjects. The
only information given was that the data collected would be used to
study certain aspects of Indian English.

Contextual style (2.3.2.8) was selected as one of the Independent
variables In this study. Therefore, in order to be able to study
correlation of the dependent variable /t/ and contextual style, an
interview situation necessitated two contrastive speech contexts.
These are casual and careful speech contexts, which elicit informal and
formal speech styles respectively. Therefore, the Interview was made
up of two parts, one to elicit casual speech and the other to elicit highly
self-monitored speech (2.3.2.8).

A phrase list and answers to personal questions provide samples
of formal and casual speech respectively; this concept is borrowed
from sampling methods first used in Labov’s 1966 study of English in
New York City. Labov ( 1972) discusses the concept of specific samples
as representing a continuum with informal on one end and formal on the
other end of the scale, on the basis of his 1966 and other studies. Thus,
dependent variables could be tested along the fol lowing scale borrowed

from Labov (1972: 99) and modified In order to study situational

4O
variation according to context.

CASUAL / CAREFUL / READING / WORD LISTS/ MINIMAL PAIRS

(Passages)
< >

INFORMAL FORMAL

 

Labov (I972) concludes this discussion with the following.
We... therefore put forward the hypothesis that the various
styles of speech we are considering are all ranged along a
single dimension of attention paid to speech, with casual
speech at one end of the continuum and minimal pairs at the

other. (99)

3.3.3.1 INTERVIEWS

The two part Interviews, firstly, consisted of questions of pos-
sible interest to the subjects, in an attempt to produce near spontane-
ous speech. While the questions were asked verbally, they were also
given to the subject in written form as part of a questionnaire at the
beginning of each interview. The questions asked were identical for all
Interviews so as to maintain a certain uniformity. Only questions that
required the subjects’ personal Involvement while answering were
chosen.

The subjects were asked to speak about or discuss one or more of
the following three broad-based questions for a total time of approxi-
mately 10 minutes. However, the subjects' responses actually ranged
from a lower bound of I 0 minutes to an upper bound of 30 minutes. The

questions asked were as follows (refer to the appendices):

41
1. How did you feel about the United States when you first came here?

2. Do you like/dislike the United States, Its people, the culture, the
lifestyle, etc. ? Please elaborate.
3. Have you had an experience In your life which has caused or seemed
to cause danger to your life? If yes, please narrate.
All subjects answered questions I and 2 once or more than once In the
four Interviews. Some subjects chose not to answer question 3.
Question 2 was asked with the specific intent of evaluating speakers'
attitude toward their new socio-cultural environment.
The data gathered In this manner, constituted the first part of the
interviews and contained samples of casual speech.
In addition to the above questions the questionnaire consisted of
a second part, a phrase list (refer to appendices). Formal speech
samples were elicited by making each subject read aloud the phrase list
four times, once In each Interview. Thus the total number of phrase list
readings was 4 times 24, I.e. , 96. The phrase list was chosen instead
of the word list so as to disguise the actual variable being studied. The
variable /t/ was embedded within phrases, whereas concealing the
specific Intent of the study, /t/, may not have been as successful within

a word list.

3.3.3.2 SELECTION OF SAMPLES 0F CASUAL SPEECH

A major disadvantage of using any method of elicitation, as
Brouwer, Gerritsen and De Haan ( I 979) points out, Is that although this
method Is aimed at eliciting everyday speech, often the speech Is not
adequately spontaneous, as the Interview situation makes the subjects

uncomfortable.

42
In this study, an attempt to counterbalance the above disadvantage

was made by analyzing only those stretches of conversation which
showed signs of uninhibited behavior such as,
(a) laughter,- or display of other emotions such as
(b) personal involvement while recol Iecting a past Incident, Indicated
by long stretches of intricate narration of some kind; or
(c) display of strong resentment towards a particular Issue.
Such samples of speech are characterized by certain channel cues which
are markers of casual speech. According to Labov (1972) -
A change In tempo, a change In the pitch range, a change in
volume or rate of breathing, form socially significant signs
of shift towards a more spontaneous or more casual style
of speech. (95)
In addition to the above mentioned nonphonological‘ channel cues, Labov
(1972) also characterizes casual speech within the Interview in terms
of five contextual situations. This characterization Is used In Labov’s
1966 study of English In New York City to Identify samples of casual
speech. Two of the contextual situations pertain to this study.
a. Speech not In response to questions: In the current study, this
contextual situation typically consists of some subjects who showed
signs of wanting to talk more and digress from the main topic during the
course of the Interview. Many of the long stretches of speech in
response to questions I. and 2 reflected such a situation. The longer the
speaker digressed, the better were the chances of capturing the
speaker‘s natural speech pattern. For Instance, In the following
conversation between a female subject and the female American

Interviewer the speaker directed the entire conversation while the

43
interviewer encouraged the subject to speak.

Speaker - I’ve lived for one and a half years in down South and two
years - three years in Michigan.

Interviewer - Do you find a big difference between the South and the
North?

Speaker - [slowly] Yah, South people are, you know, more relaxed,
and they have more social life. . . They have more time
so they have more human tendencies. Here everybody Is
busy you know, [faster] taking two shifts, three shifts,
and going to college and nobody has any time. [faster] And
whatever spare time, it goes sprinkling the lawn, clean-
ing the house. . . .

Interviewer - [laughs] more busy work, right?

Speaker - [digresses] Yah, and a family life is completely different
than our [Indian] way, you know, [digresses] man Is a...
you know, women are . . . . I appreciate American women
more than any other women in the whole world. . . they
are the most hardworking . . .

As the speaker digresses her speech becomes more spontaneous.

b. The danger of death: In the current study question 3 of the
questionnaire was chosen from Labov (1966) and rephrased slightly.
The purpose was to create a contextual situation in which, a subject,
who had experienced a situation which had caused danger to her/his life
became very “ Involved in the narration”. He/she seemed to be “rel iving
the critical moment” and showed signs of highly uninhibited behavior
such as, nervous laughter, heavy and irregular breathing due to tension,

or a change In pitch. For example, consider the following conversation

44

between a male subject and the female American interviewer In re-

sponse to question 3 (see appendices).

Speaker

. . . he suddenly was moving around and decided to point
It at me, and, [clears his throat]. . . I was standing about,
altogether five feet away from him, a little more than
that. And this other friend told him “its loaded, don’t
point It at him.” [heightened pace] Apparently he didn’t
realize that he had ahh . ., it was a very fine trigger, and
almost like feather touch [nervous laughter], and perhaps
he didn’t realize but the trigger went off, and the next

thing I knew was, there was a ringing sound in my ear and

Interviewer - He was only five feet from you?

Speaker

— Yah, It was a very short trip, and fortunately It was a gun

which didn’t have a ahh, it was a old gun. . the one I had
was a very new one but ahh . . from the face I could see
that something really bad had happened [shaky voice] and
from that ringing sound I knew that something had gone
wrong but I didn’t feel a thing. When I touched my
forehead I found this hard thing stuck to my forehead.

[very shaky, intense, low pitched voice] That was a really

- frightening experience . . .

Further, Labov (1972) says that,

The formal definition of casual speech within the Interview

requires that at least one of the five contextual situations

prevail, and also one of the five nonphonological cues. (87)

Therefore, samples were selected for analysis of casual speech if

45
and when they fulfilled the above requirement. Out of 96 Interviews

there were only three In which such definite characteristics were
absent. In these cases, only the last few minutes of the conversation
In each Interview were taken into consideration. This was done to avoid
the portion of the subject’s speech when he/she was still in the process

of familiarizing himself/herself with the interview environment.

3.3.3.3 SELECTION OF /t/ WITHIN SAMPLES

The selection of /t/ within samples of highly self-monitored
speech, which comprised 96 readings of a phrase list, was as follows.
There were 30 occurrences of /t/ within each phrase list. In each of the
96 cases only the first 20 successive instances of /t/ were considered
for the analysis of formal speech. In similar cases In previous studies,
when subjects were made to read lists of words It has been observed
that, as subjects‘ reading progresses down the list, self-monitoring of
speech decreases and as a consequence, the reading becomes more
spontaneous. Therefore, for all samples of formal speech, only the first
20 occurrences of /t/ were considered so as to obtain the most formal
speech available within each Interview, of which the phrase list was a
part. The number 20, however, Is arbitrary.2

The selection of /t/ from samples of casual speech was as fol lows.
The selection of subsamples of casual speech was discussed in section
3.3.3.2. From these subsamples that best represented casual speech, in
each of the 96 cases, 20 successive Instances of /t/ were selected for
all 96 interviews.

Thus, for each of the 96 interviews, a total of 40 occurrences of

/t/ 20 from Informal speech and 20 from formal speech were

46
considered for analysis. The phonetic nature of these selected occur-

rences was determined as to whether [t], [t] or any other variant of

/t/ appeared.

3.4 DATA ANALYSIS

3.4.1 DISTRIBUTIONAL ANALYSIS

A total of 3840 occurrences of /t/ were analysed in this study. 18
(47%) of the 3840 occurrences were determined as unclear for analy-
sis.

Of the remaining 3822 instances of /t/ 1416 (37.10%) were rec-
ognized as the variant [t]. The criteria discussed in Labov (1966)
referred to in Lavendera ( I 978) and mentioned in 2.3.1 of this thesis are
used here. [t] could be defined as a sociolinguistic variant of the
variable /t/ because:

1.[t] is easily stratifiable within the sample of the population being
studied;
2. [t] is high in frequency;
3. [t] Is immune to conscious suppression;
4. [t] Is an integral unit of a larger structure, namely, the word in
which It occurs; and
5. [t] being a phonetic variant carries purely nonreferential informa-
tion, occurs in all possible phonetic environments of /t/, and Is
easily quantifiable within the data.
8 occurrences of /t/ in the data were realized as [ti]. The lateral
affricate [ti] occurred in the restricted environment of word final
position of a single word “petal’ pronounced as [‘petl]. All [ti] occurred

In samples of formal speech. This was the only possible environment

 

47
for [t4] in samples of formal speech. In addition, these 8 occurrences

were found In the formal speech of female subjects only, restricted to
categories A (0-2 years) and B (2-5 years). Other words with similar
phonological environments such as ‘little' elicited [-t1l] or [-tr.l] In
informal speech.

The occurrence of [ttl can be accounted for by arguing that some
Indian English speakers with relatively less exposure to American
English used the Standard British English [ti] which Is perhaps fossil-
Ized In IE and Is used In certain hyper-formal styles of Indian English
as the ‘super-standard norm’ in those contexts in that environment.
[ti] In the speech of Indians, and therefore in the speech of Indian im-
migrants In the United States, can be traced historically to be the result
of “‘extended periods of colonization " (see 2.2.1 ) of India by the British.
Jones (1943: 323) describes “petal’ as a single syllabic word [‘petll
which suggests that [till is a single phone in that environment In British
English. Whether [ti] belongs to /t/ or to /I/ can only be answered by a
detailed phonological analysis of the two phonemes in British English.
Similar words such as ‘metal’ and “little’ are described in Jones ( l 943)
as disyllabic words, [‘met/l] and [‘lit/l] respectively; the slant line
indicating syllable boundaries within morphemes. At this point, this
apparently arbitrary syllabIc division remains inexplicable. The above
explanation however, accounts for

(a) the occurrence of [ti] In the data, in samples of formal speech and
Its absence in similar environments within Informal speech; and

(b) the presence of [ti] in the speech of female subjects, only In
categories A (0-2 years) and B (2-5 years) and not in C (5- I 0 years)
and D (10 years & above).

48
The absence of [ti] in categories C and D could be explained as follows.

Beyond 5 years the prestige norm changes from the hyper-formal
(British English) norm [ti] in that environment to American English [t]
in the speech of these subjects at least In formal speech. It] Is used in
‘petal’ as [‘petzl] Instead of [‘pett] In the speech of subjects whose period
of residence In the United States Is 5 years or more.

Two occurrences of /t/ in the data are realized by the unaspirated
dental stop It]. It occurs In the speech of one particular subject (no. 20
in the subject list in the appendices). it] is also restricted to a single
word ‘pistol’. This English word has been borrowed by Indian languages
and indianized by substituting the unaspirated dental stop [t] for the
alveolar stop [t]. Thus, the Indianized version [‘pistol] 'is used in the
standard dialects of certain Indian languages such as Hindi, Marathi,
Gujarati, etc., to mean ‘pistol’. The subject In whose speech [1;] occurred
(two out of four times in formal speech) was a native speaker of
Gujarati and used the Gujarati [‘pistol] in her Idiolect bf Indian English
Instead of the Standard Indian English [‘pistal] or the Standard American
English [‘pistal].

Bhatia (1967:75) mentions the loanword [‘pistol] as occurring In
Standard Hindi. The following discrepency is noted in Bhatia (1967)
regarding a similar word ‘hospital’ in Gujarati. Bhatia (1967:162)
gives a comparative table of contrastive Hindi phonemes in the Deva-
nagari Script,3 and the phonetic script equivalents of these phonemes.
The unaspirated dental stop ("T In the Devanagari Script Is given the
equivalent symbol [t]. The same symbol 6T Is used In Bhatia ( I 967:
18096) in the Gujarati word for ‘hospital’ which Is a loanword from

English. The equivalent phonetic transcription however, does not have

49
the symbol tbut Is given as [Ispita:l] with an alveolar stop [t]. According

to Kachru ( l 983227-28), the alveolar series in SAE is substituted by the
retroflex series due to the underlying features of LI of the region. This
Implies that In Indian languages alveolar It] does not exist. Bhatia's
transcription of [t] In the English loanword ‘hospital’ In Gujarati Is
probably Incorrect.

Besides, that the loanword [‘pistol] exists In Gujarati was substan-
tiated when native Gujarati speakers were asked for the word ‘pistol’
In Gujarati. Two speakers were asked and both speakers confirmed the
presence of the word [pistol] in Gujarati.

The three variants [t], [tit], and [t], as discussed above, are not
phonetic components of the American English /t/; therefore, I treat
them here as belonging to Indian English. The Indian English variant It]
occurred freely In the data with the American English variant [t] In all
possible environments of /t/. Thus, [1;] and [t] are sociolinguistic
variants occuring In parallel distribution in the dialect of the twenty
four Indian immigrants whose speech was studied.

2315 (62.4%) Instances of /t/ In the data were realized as the
alveolar stop [t]. It] may also be considered as a reliable sociolinguistic
variant of /t/ as it occurred In the speech of all subjects and was
frequent In both contexts and In the speech of both sexes. Also, It] Is
an allophone of the American English /t/ and a reliable indicator of the
acquisition of the American English dialect; the percentage of [t] In-
creased proportionally as the period of residence of the subjects
increased (see Table 4). [t] showed “assymetric distribution over a
wide range of . . . ordered strata of society” (Labov 1972: 8).

81 Instances of [F] were found In formal and informal speech styles

50
of both male and female subjects in all four categories A, B, C, and D.

Table 2 below shows the occurrence of [F] In the four categories. [1“] is
an allophone of the American English phoneme /t/; hence It may be
considered here as belonging to American English and not to Indian

English.

Table 2: the frequency of [r] in subjects’ speech.

 

 

CATEGORY NUMBER or [F]
A 4
B 8
C 24
D 45

 

 

 

 

[F] like [ti] and [3], however, did not occur in the speech of all subjects
in the data. [1‘] occurred in the restricted intervocal ic position in words
such as, [WD is r] ‘water’ and [hwni‘a] ‘what a’. For reasons discussed
earlier in thissection, [F], [ti], and [t] cannot be treated as sociol inguis-
tic variants of /t/.

Thus, In the data analyzed, the sociolinguistic variable /t/ con-
sisted of five variants - [t], [ti] and [t] belonging to IE and [t] and [F]
belonging to American English.4 However, only two of these variants,
It] and It] could be characterized as sociolinguistic variants of /t/.
Chapter 4 of this thesis discusses the distribution of the sociol Inguis-

tic variants [t]~ [t] when correlated with sociological parameters.

 

51

3.4.2 STATISTICAL ANALYSIS

The following are the Independent variables in this study - the
period of residence of the subjects In the United States-X,, subjects’
gender-X2, interviewers’ gender-X3, (accommodation due to) ethnicity
of Interviewers (interlocutors)-X4, Speakers” attitude to their new
socIo-cultural environment-X5, and contextual style-X6. The two
Interactions, subject-interviewer gender interaction, and subject-
interviewer gender and ethnicity interaction are Z,and 22, respectively.
These variables were coded in the following manner. X,, the period of
residence of subjects is a continuous variable and It takes the values
as was reported when Indian English speakers were requested to be
Informants In this study. The other variables X2, X3, X4, X.5 and X5 are
nominal variables which were coded consistently, for the purpose of
analysis as given in Table 3. Since the variables X2, X3, and X4are dicho-
tomous, that is, they take the values -I or I, the Interaction between
them, as is calculated In similar analyses, Is the product of those
variables. Thus, subject-Interviewer gender Interaction -21 = X2 times

X and subject-interviewer gender and ethnicity Interaction -22 =

3.
x2 times x3 times x4.

To study the effect of the Independent variables X,, X2, X3, X4, X5, and
X5» and the interactions 2, and 22 on the covariants [t] and [t] of the
dependent variable /t/, the following linear models were proposed for
the data:

(1)

Y1 = BO+B 1X1 +82x2 +8 3X3 +B4x4 +BSXS +BGX6 +5 721 +BBZZ+ 6‘1

52

Table 3: Coding of nominal variables.

 

 

 

LABEL VARIABLE VALUES TAKEN CODING
Sex of Subjects X2 -1, I —1= Female
l= Male
Sex of Interviewers X3 -1, l -I= Female
(Interlocutors) l= Male
Ethnicity of Interviewer X4 -1, 1 -I= Indian
(accommodation) 1= American
Subjects’ Attitude X5 1,0, -I l= Positive
to their new O= Undetermined
environment - l = Negative

 

 

 

 

 

53
(2)

Y2= 80 +8 1X1 +8 2X2 +B3X3 +B4X4+BSX5 +BGX6 +8721 +BBZZ+ €2

In equation (1) above,

Y, is the frequency of occurrence of the retroflex variant [t];
B o is the Intercept;

B . . 3

independent variables, and interactions); and

, . 8 are the sociological parameters (coefficients of the
Q, Is the random unobservable error.

In equation (2),

Y2 is the frequency of occurrence of the alveolar variant [t];

B o is the intercept;

B, ...B8 are the sociological parameters (coefficients of the independ-
ent variables, and interactions); and

E 2 is the random unobservable error.

To study the significance of the parameters, multiple regression
analyses were performed on both models. F-tests were conducted to
determine whether the models suited the data. From Table 12,431, it
is evident that both models (I) and (2) are significant; the models
proposed fitted the data Ideally; the significance level of both models
being .0001. The R-squares for both models were not high indicating
that there Is not much variation in the data (refer to Table I2). R-square
values close to I are generally preferred. However, for the purpose of
this study, R-squares of.44l 1 for[t]and.4395 for[t] were acceptable.

The significance of the statistical analysis and its interpretation

is presented In chapter 4.

54

NOTES

1. Labov (I972: 87) uses this term. See quote on page 44. Rate and
pitch are phonological, i.e., suprasegmental elements, therefore,
the term nonsegmental Is perhaps more appropriate than nonphon-
ological.

2. The greater the number of /t/, the better are the chances of
capturing regularities. When sampling /t/ in casual speech, for
instance, /t/ occurs very often and a smaller number, if used, will
Include a very small stretch of conversation which is less likely to
capture regularities within speech. On the other hand, one might
argue that a smaller number could have been more confined to self-
monitored speech in the phrase list.

3. The Script of Sanskrit and also of languages derived from it such as
Hindi and Gujarati.

4. For the purpose of the statistical analysis and correlation of /t/
with the sociological parameters, the comparatively small number
of occurrences of [ti] and [g] were Included within the socIol inguis-
tic variant [t] and the occurrences of [F] were Included within [t],
so as not to exclude these occurrences from the study.

The small number of the variants [t], [t1] and [F] when compared with
that of [t] and [t] Indicate the possibility of considering the socio-
linguistic variants [t] and [t] at a level Intermediate to the variable
/t/ and Its five variants (which then, may be treated as subvari-
ants of the phonetic variants [tland[t]). However, the basis for such

an analysis In this study Is insufficient; it would be based solely

55
on quantitative differences between the occurrences of these

variants. Among the factors that might justify such an analysis is
speakers' perception of one variant verses the other, for instance,

[t] verses [ti], or [t] verses [F].

CHAPTER 4

RESULTS

4.1 INTRODUCTION

This chapter reports the results of this study in 4.2. Section 4.2
has eight subsections which report the correlation of the dependent
variable /t/ with various parameters presented In 2.3.2.

Section 4.3 is divided into two subsections. Subsection 4.3.1 dis-
cusses the statistical significance of the results and the interpreta-
tion of the statistical analysis conducted. Subsection 4.3.2 gives a
sociolinguistic explanation for the various patterns In the occurrence
of the variants [t] and [t] of /t/.

4.2 CORRELATION OF SOCIOLOGICAL PARAMETERS AND THE DEPENDENT
VARIABLE /t/

2396 occurrences of the alveolar [t] and 1426 occurrences of the
retroflex [t] were separately correlated with the Independent variables
and Interactions, discussed in 2.3.2 and mentioned in 3.4.2. This
permits the study of the effect of these parameters on the two opposing
covariants [t] and [t] which are phonetic components of the sociolin-
guistic variable /t/. The results obtained are discussed In the follow-

Ing subsections.

4.2.1 PERIOD OF RESIDENCE IN THE UNITED STATES AND /t/
[t] and [t] were correlated with the continuous Independent variable

period of residence in the United States. Figure I shows a considerable

56

35:2. .... .ms 2.. ... 8.32.2 .6 .85..

owv oov can own cvm own com com com ovm cmméow oar cop of. our cop om om ov cm 0
n

D nIPhb P h- F - Db b - up D P. h I h. n I. Pb h n b b bePhn h b - I o

 

57

 

t

row

row
[on

1

ice

..8
:8

We

too

 

 

 

    

 

 

 

.u .u n. .8
m ” u 4
.. . .. .n . u r8,
... u .. . u m .. .
.. a; m ....m n." - a:
.. .. a: u ......u mu .
.. I m... .. m." ..s.
Vb. ... .... ... m n ...... m u. I 00w
mn—r loll-lul on. one no. in Q m“. f
N... 3.1:... ............ .... ...... m We:
Ph- onic-8 coo so“ temp

 

 

8..

P 952“.

[3] put [1] to oauounooo

58

 

 

 

 

 

 

 

 

 

 

 

 

oo.oo_ comm 00.09 cm: 359
8. .8 3E omv 8
oo om omv oom mm
NE NW .8: wow mm
. . of . . . Bum .N
co .m. t. On 0? cm comm mmm. omm vs 0 Evow No. om
B. m mm. 9
mm. mm co. o.
no .3 380m. oo. .2
on— om -.8 mo @—
o _ .62 3.3 mmo no omdv mmdm 3m 60 u 3.8 mm 9
Go 3 on 3
mm. vm .K 3
K0. mm mm N—
mm. mm 358 om _ _
om 09 .m8 mv o.
2.00. mth moo B mwwm mndm mmm 00 m omdv on o
91 _ _ .m m
oo_ vm om N.
MK no ..N 0
3 no. BE 2 m
66 66 XS 9 v
66.3. 66.6. 69. mm 8.3 Ban omm mm < no.2 o. m
E. N. _ m6 N
mm mm o _
E E E to E to a. a. a. no 2 to E8 2: a: 35.8: .oz
2...“: do a .02 358 823586 25: do a .62 358 BzEasuuo LED 25: to 8.3.. .9

 

 

 

 

6.33m 63:5 65 E 36632... co 66:628.. ..o 60:66 65 S
8:28 E 566% .3833 S E 6cm 3 B 8:2:me ”v 655

 

59
degree of individual variation in the occurrence of [t] in T2 and [t] In T4.

However, when the mean of the individual scores for [t] and [t1 within
the categories A, B, C, and D are considered (refer to Table 4), a more
regular pattern emerges in T1 and T3 for [t] and [t], respectively, as can
be seen In Figure l.

The results of this correlation confirm the hypothesis a. postulated
in 2.4.3: the frequency of occurrence of [t] Is Inversely proportional to
the period of residence In the United States of the Informants. Con-
versely, the frequency of occurrence of [t] is directly proportional to
the period of residence of the informants In the United States.

As the period of residence increases the mean scores for the
alveolar phonetic variant [t] in the speech of speakers belonging to the
four categories Increases progressively which Is observable In T 1. The
mean score values for [t], the retroflex variant, decreases with an
Increase In the subjects’ length of residence as is indicated by T3
(Figure I)

This result Is additionally substantiated by the results of the t-
test for statistical significance discussed In 4.3.1. The correlation of
/t/ with X, - the period of residence is Indicated In Table 12 which
shows the parameter estimates as -.3867 for [t] and .3781 for [t], and
a probability of .0001 for both [t] and [t]. A p-value of .05 (5%) or less
generally shows that the result is significant.

Table 4 shows a rapid decrease of 13.82% In'the occurrence of [t]
in speakers, from category A (38.57%) to category B (24.75%). There-
after, the decrease Is gradual. Similarly, a rapid increase In the
occurrence of the American English variant [t], can be observed In the

speakers studied, between category A (l 6.94%) and category B (25.33%)

60
which is an increase of 8.39%. The increase from group B to C and from

C to Dis gradual.

4.2.2 CONTEXTUAL STYLE AND /t/

The results, illustrated in Table 5, confirm the hypothesis b. stated
in 2.4.3: [t] is the ‘high-prestige norm’ In the society in which the
informants currently reside. Therefore, due to self-monitoring within
one’s speech, the frequency of occurrence of [t] is higher in formal
speech styles and lower in Informal speech styles. As [1;] is the “low—
prestige norm’, the opposite is true for [t].

The Indian English variant [t] occurs more frequently In informal
speech (51.47%) and less frequently in the subjects’ formal speech '
(48.53%). The American English variant It] being the prestige norm
occurs more in formal speech (50.75%) and less In Informal speech
(49.25%), of the Indian informants studied. The difference between
percentages, however, is not significant. This is also confirmed by the
t-test for statistical significance. The parameter estimate for [t]
when correlated with the independent variable X,5 - Contextual style Is
-.O484 and the probability .3815. The parameter estimate for [t] is -
.0065 and the probability of the hypothesis being rejected is .4213
showing that the results are not significant (refer to Table 12, 4.3.1).

Table 5, however, shows some deviation In the pattern of occur-
rence of [t] and [t] within the categories A, B, C, and D.

In categories A, B, and C the occurrence of [t], the low prestige
norm, Is lower In formal speech than in informal speech, for instance
in C the percentage of occurrence of [t] in formal speech is 10.24%

whereas In Informal speech it is 10.59%. Category D however, Indicates

Table 5 : distribution of [t] and [t] In formal and informal

61

contextual styles In categories A, B, C, and D.

 

 

 

 

 

 

CATEGORY OCCURRENCE OF It] OCCURRENCE OF [t]
(In years) INFORMAL FORMAL INFORMAL FORMAL
No. of It] % of 1;] No. of It] % of [t] No. of [t] % of [t] No. of [t] % of [t]
A
(0-2) 285 19.99 265 18.58 193 8.05 213 8.89
B
(2-5) 187 13.11 166 1 1.64 293 12.23 312 13.02
C
(5-10) 151 10.59 146 10.24 328 13.69 327 13.65
D
(10 & 1 1 l 7.78 1 15 8.07 366 15.28 364 15.19
more)
TOTAL 734 51.47 692 48.53 1 180 49.25 1216 50.75

 

 

 

 

 

 

 

 

 

 

 

62
otherwise. The percentage of occurrence of [t] is more in formal speech

(8.07%), and the percentage of [t] is less in informal speech (7.78%). The
difference in percentage, however, Is very marginal (29%).

The high-prestige norm [t] occurs more frequently in formal speech
and less frequently in informal speech In categories A and B - for
Instance, the percentage of occurrence is 8.89% and 13.02% In groups A
and B in formal contextual style; and 8.05% and 12.23% in groups A and
B in informal contextual style (refer to Table 5). However, in categories
C and D the percentage of occurrence of [t] is higher In Informal speech
(13.69% In C and 15.28% in D) and lower In formal speech (13.65% in C
and 15.19% in 0). Again, the difference In the percentage of occurrence
of [t] between formal and Informal styles of speech is marginal.

Table 5 also shows that the difference In percentage of occurrence
of [t] and [t] between the two contextual styles is more in groups A and
B and decreases in groups C and D. However, the total difference Is

negligible and Is, therefore, not statistically significant.

4.2.3 ACCOMMODATION TO INTERLOCUTORS AND /t/

The figures in Table 6 confirm hypothesis c. in section 2.4.3: due to
accommodation of speakers to their Interlocutors’ speech, the Ameri-
can English variant [t] occurs more frequently in informants’ speech
when speaking with an American English speaking Interlocutor and less
frequently when speaking with an Indian English speaking interlocutor.
Inversely, the Indian English variant It] occurs more frequently when
speaking with an Indian English speaking interlocutor than when
speaking with an American English speaking Interlocutor.

The Indian immigrants studied accommodated to the speech of

Table 6: Distribution of [t] and [t] when correlated with
accommodation to interlocutors’ speech.

63

 

 

 

 

 

 

CATEGORY OCCURRENCE OF [1;] OCCURRENCE OF [t]
(In years) INDIAN 1 AMERICAN 1 INDIAN I AMERICAN I
No. of It] % of [t] No. of [t] % of [t] No. of [t] % of [t] No.01 It] % of [t]

A

(0-2) 321 22.51 229 16.06 145 6.05 251 10.48
B

(2-5) 195 13.68 158 11.08 283 11.81 322 13.44
C

(5-10) 169 11.85 128 8.97 306 12.77 349 14.56
D

(10 8. 132 9.26 94 6.59 355 14.82 385 16.07
more)

TOTAL 817 57.30 609 42.70 1089 45.45 1307 54.55

 

 

 

 

 

 

 

 

 

I - INTERLOCUTOR

 

64
American Interlocutors, both male and female, by lowering the fre-

quency of the Indian English variant [1;] In their speech to 42.70% when
speaking with American Interlocutors. When speaking with “home-
folks” (Indian interlocutors), however, the frequency of It] Is higher
(57.30%). Along with lowering the frequency of [t], subjects increased
the frequency of the American English variant [t], when speaking with
American Interlocutors (54.55%), and lowered the frequency of [t] in
their speech when speaking with Indian Interlocutors (45.45%).

The significance of the above result Is additionally confirmed by
the t-test for statistical significance. Table 12, 4.3.1 shows the
parameter estimates for [t] and [t] when correlated with X,, - Accommo-
dation in the speech of speakers to interlocutors of different ethnic
identities (Indian-and American). The parameter estimate for It] Is -
.2356, and for [t] It Is .2431. The probability of the hypothesis being
rejected Is .0001 for [t] and [t] showing that accommodation as an
Independent variable has a significant effect on the dependent variable
/t/.

The maximum difference in the percentage of occurrence of [t] and
[t] due to accommodation in the speech of subjects Is seen In category
A (0-2 years). In the speech of subjects In category A, the percentage
of occurrence of [t] when speaking with Indian interlocutors is 22.51%,
and when speaking with American Interlocutors is 16.07%. The differ-
ence between these two percentages is 6.46%. Also, In category A, the
percentage of occurrence of [t] is 6.05% in speakers when communicat-
ing with Indian Interlocutors and 10.48% when speaking with American
interlocutors. The difference between these values Is 4.43%. Thus, the

speech of subjects in category A shows maximum fluctuation.

65
From Table 6 it can be seen that, In category D speakers raise the

number of [t] in their speech minimally by 1.25% (16.07% - 14.82%),
when speaking to American Interlocutors, Indicating most linguistic
security in the speakers of that category. However, when [t] is
considered It is not speakers of category D but those in category B who
lower the occurrence of [t] in their speech minimally, when speaking
with American interlocutors (2.60%), which is unexpected. However,
the difference In the percentage of [t] between categories 8, C, and Dis
marginal: In category B it Is 2.60%; in C it is 2.88%; and In D It is 2.67%.

Thus, the difference is not significant.

4.2.4 SPEAKERS’ ATTITUDES TO THEIR NEW SOCIO-CULTURAL
ENVIRONMENT AND /t/

The positive and negative values for this independentvariable were
determined based on subjects’ responses to question 2 (refer to the
appendices). Subjects who responded positively to question 2 are
included under ‘positive attitude’, and subjects who responded nega-
tively to question 2 are Included under ‘negative attitude’ in Table 7.
The attitude of four subjects toward their new environment could not
be determined. Out of 24 subjects studied, 12 were determined as
having a positive attitude and 8 were determined as having a negative
attitude.

The results given In Table 7 show that the hypothesis d. postulated
In 2.4.3 was supported: Informants’ positive attitude towards American
society, Its people, its culture, and Its lifestyle, is reflected in the
higher frequency of [t] and a lowerfrequency of[t]. Informants’ negative

attitude towards American society is reflected in the relatively higher

66

Table 7: Distribution of [t] and [t] In subjects’ speech in relation
to their attitude towards their new socio-cultural
environment.

 

 

 

 

 

 

 

ID. SPEAKERS WITH A ID. SPEAKERS WITH A ID. SPEAKERS WHOSE
NO. POSITIVE ATTITUDE NO. NEGATIVE ATTITUDE NO. ATTITUDE WAS NOT
DETERMINABLE
No. of [t] No. of [t] No. of It] No. of It] No. of It] No. of It]

3 75 85 1 75 83 4 94 64
6 87 73 2 1 12 48 9 99 61
7 54 106 5 107 53 14 63 95
10 100 59 8 I l 149 24 53 107
1 1 37 123 15 96 63

12 52 107 17 63 92

13 24 135 20 74 86

16 29 130 23 56 99

18 22 138

19 3 157

21 22 139

22 18 142

12 523 1385 8 594 673 4 309 327

av. 43.58 1 15.42 av. 74.25 84.13 av. 77.25 81.75

 

 

 

 

 

 

 

 

 

av. = average number of [t] or [t] per speaker.

 

67
occurrence of the IE variant [t] and a lower occurrence of the American

English It].

The effect of speakers’ attitude can be seen in the difference in the
average number of [t] and [t] per person between subjects with a
positive attitude toward their environment and subjects with a nega-
tive attitude.

The average number of it] per person for subjects with a positive
attitude is 43.58, and the average number of [t] Per person is I 15.42. For
speakers with a negative attitude, the average number of [t] per person
is 74.25 and the average number of [t] Is 84.13.

The significance of this result is substantiated by the statistical
analysis. The parameter estimates when speakers’ attitude - X,5 was
correlated with the variable /t/ are as fol lows (refer to Table 12,431)
-.4454 for [t] and .4517 for [t]. The probability of the hypothesis being
rejected for both [t] and [t] is .000 1 , showing that the results are signi-
ficant.

The results, therefore, show that subjects with a positive attitude
to their new socIo-cultural environment have a higher average occur-
rence of [t] and a lower average occurrence of [t] than subjects who have

a negative attitude toward their new environment.

4.2.5 SUBJECTS' GENDER AND /t/

Table 8 gives the distribution of [t] and [t] In the speech of male and
female subjects within categories A (0-2 years), 8 (2-5 years), C (5-
10 years), and D (10 years & above). The total percentage of occurrence
of the Indian English variant [t] In the speech of male subjects is 55.54%,

which Is higher than the percentage of occurrence In female subjects

Table 8: Distribution of [t] and [t] between male and
female subjects In categories A, B, C, and D.

68

 

CATEGORY

(In yrs.)

OCCURRENCE or it]

OCCURRENCE OF It]

 

MALE SUBJECTS

FEMALE SUBJECTS

MALE SUBJECTS

FEMALE SUBJECTS

 

 

 

 

No. of [t] % of It] No. of It] % of It] No. of It] % of [t] No. of [t] % of It]
A
(0-2) 306 21.46 244 17.11 174 7.26 232 9.68
B
(2-5) 251 17.60 102 7.15 227 9.47 378 15.78
C
(5-10) 142 9.96 155 10.87 336 14.02 319 13.32
D
(10 & 93 6.52 133 9.33 388 16.20 342 14.27
more)
TOTAL 792 55.54 634 44.46 1 125 46.95 1271 53.05

 

 

 

 

 

 

 

 

 

 

69
(44.46%).

Inversely, the total percentage of occurrence of the American
English variant [t] is higher in the speech of female subjects (53.05%)
and lower in the speech of male subjects (46.95%). These results
support hypothesis e. stated in 2.4.3: allowing for other sociological
factors, In general, the frequency of [t] is higher In the speech of male
Indian informants than in female Indian informants. Inversely, the
frequency of [t] is higher in womens’ speech than in mens’ speech.

The significance of this result is confirmed by the t-test that was
conducted to test the hypothesis. The parameter estimates for [t] and
[t], as given in Table 12 In 4.3.1, when correlated with X2 - Subjects’
gender, are .31324for[t], and -.2975 for [t]. The p-value or probability
of the hypothesis being rejected is .0001 for [t] and [t]. The above values
Indicate that the independent variable, subjects’ gender, has a signifi-
cant effect on the variance of the dependent variable /t/.

Within the four categories, there Is considerable variation from
the general trend in the occurrence of [t] and it], particularly within the
speech of female subjects In category B. The frequency of [t] decreases
from A(17.11%)to group D (9.33%) with one exception - female subjects
in group B have the lowest frequency of [t] among female subjects
(7.15%). This is because, in group 8 subject No. 8 (see Table 4) has only
11 [t] In her speech. The Individual characteristic of this subject
strongly affects the percentage of [t] in group B.

In male subjects, the percentage of [t] is maximum in group A
(21.46%) decreasing In that order to a minimum of 6.52% In group 0
without any deviations. The decrease, however, is not gradual because

the decrease in the percentage of [t] between group B and C is 7.74%.

70
The same exception in linguistic behavior can be seen in the

frequency of [t], in the alveolar phonetic variant, in the speech of female
subjects In group B. The frequency of [t] increases from 9.68% In group
A to 14.27% In group D. However, the Increase Is not steady, as group
B has the highest frequency of [t] In the speech of females (15.78%).
Then, between group B and group C, there Is a decrease of 2.46%, from
15.78% to 13.32%.

In the speech of male subjects, however, there is an Increase from
group A to group D, from 7.26% to 16.20%, in that order. Again, the
increase in the percentage of It] is more rapid (4.55%) between group B

and group C than between the other groups (see Table 8).

4.2.6 INTERVIEWERS’ GENDER AND /t/

The results illustrated in Table 9 confirm hypothesis f. stated In
2.4.3: the total percentage of [t] is higher in the speech of informants
when the interlocutor is male than when the interlocutor Is female. The
Inverse Is true of [t].

The total percentage of occurrence of the Indian English variant [t]
Is more (54.28%) in the speech of subjects when the interviewers
(interlocutors) are male Irrespective of the interlocutors ethnicity.
When, the interviewers are female the total percentage of [t] in the
speech of Indian subjects Is less (45.72%). Conversely, the American
English variant [t], the high-prestige norm occurs less (47.50%) in
subjects when they communicate with male interlocutors. And [t]
occurs more In subjects’ speech when they communicate with female
Interlocutors (52.50%).

This result is statistically significant. The parameter estimates

71

Table 9: distribution of [t] and [t] in the speech of subjects in

relation to interviewers’(interlocutors’) gender.

 

 

 

 

 

 

CATEGORY OCCURRENCE OF It] OCCURRENCE OF [t]
(in yrs.)
MALE I FEMALE 1 MALE I FEMALE I
No. of It] % of [$1 No. of It] % of [E] No. of [t] % of It] No. of [t] % of It]
A
(0-2) 292 20.48 258 18.09 186 7.76 220 9.18
B
(2-5) 199 13.96 154 10.80 281 11.73 324 13.52
C
(5-10) 164 11.50 133 9.33 314 13.11 341 14.23
D
(10 & 1 19 8.34 107 7.50 357 14.90 373 15.57
more)
TOTAL 774 54.28 652 45.72 1 138 47.50 1258 52.50

 

 

 

 

 

 

 

 

 

 

I - INTERLOCUTORS

72
given in Table 12, 4.3.1 are as follows. When [t] and [t] are correlated

with x3 - Interviewers’ gender, the estimate for [t] 13.13868 and the
estimate for [t] is -.1344. The probability of the hypothesis being
rejected 13.0130 for[t]and.0161 for[t]. As mentioned in 4.2.1 ap-value
of .05 or less is accepted as significant.

There Is a decrease from categories A to D in the occurrence of [t]
when Interviewers are male as well as when the interviewers are
female. However, there is a rapid decrease In [t] of 6.52% from 20.48%
in category A to 13.96% In category B in subjects when Interlocutors are
male. Also, there is a rapid decrease of 7.29% from A (18.09%) to B
(10.80) In the occurrence of [t] in speakers when the interlocutors are
female.

The opposite trend is observed in the difference In frequency of [t]
between groups A and 8. With male Interlocutors the frequency of [t]
increases from category A to B by 3.97%. Thereafter, the percentage of
occurrence increases very gradually to 13.1 1% in C and to 14.90% In
group D. When the interviewers are female, there Is an Increase of 4.34%
between group A (9.18%) and group B (13.52%) in the percentage of [t].
In groups C and 0, once more, the increase In [t] Is gradual (from 14.23%
to 15.57%). A possible explanation for this pattern of occurrence Is

given In 4.3.2.

4.2.7 SUBJECT-INTERVIEWER GENDER INTERACTION AND /t/

The figures In Table 10 confirm the hypothesis 9. postulated in
2.4.3: due to interaction of subject (S)-interviewer (1) gender, the
percentage of It] decreases and that of [t] Increases.

The decreasing order of the total percentages of [t] when correlated

73

 

 

 

 

 

 

 

 

 

 

 

 

 

..(
:NN 00.0N ONVN 0 _ .NN 0N.ON o _ .VN 0v.0N 00.00 APO...
00.5 00.0 NN.0 no.5 0V? vm.v No.0 00.0 o
0 _ N M _ .0 00.5 50.0 NV? 0V0 3.? 00.0 0
00.0 NVN NN.0 0N.v 00d 0N6. NON 00.0 m
00.? 00.? 00.? 0o.N 99.0 00.0 00.0 00.. _ (
_ 5(me _ 04(2 _ 04(me _ m._(_>_ _ 0.320“. _ w._(Z _ 5(me _ u._(_)_
>a00
0.505030 3(me 0.505030 m..(T_ 0HUmBDO 3(me 0.505030 04(2 .m._.(u
E ...O mUZmamDuuo ...o 00(h200mma n; ...o mUZm¢¢DUUO “.0 00(thumma

 

 

A S 0L0uvuotmuc—

29:8 6cm BEE 52> mwocmcuxm 58% S 36638

22:8 6% was 5 E 26 E .6 8:32:20 No_ 638

 

74
with subject-interviewer gender interaction confirms the order of

subject interviewer gender combinations predicted. The order for [t],
therefore, is - Male S/Male I (30.08%) > Male S/Female I (25.45%) >
Female S/Male I (24.I9%) > Female S/Female I (20.26%).

Inversely, the order of subject (S)/interviewer(|)gender combina-
tions increases in the following order as given in g., 2.4.3. The order
for [t], therefore, is - Male S/Male I (22. I 6%) < Male S/Female I (24.79%)
< Female S/Male I (25.33%) < Female S/Female I (27.7l%).

The regression analysis shows, however, that this interaction is
not significant. The parameter estimate for [t] (see Table 12, 4.3. I)
when correlated with Z1 = X, (subjects’ gender) and X3 (interviewers’
gender) is .007. The probability of the hypothesis being rejected is very
high (.8866). The parameter estimate for [t] is -.0065 and the
probability is .9058. These values indicate that the interaction does not
have an effect on the percentage of occurrence of [t] and [t]. This can
be seen in the marginal difference between Male S/Male I (30.08%) and
Female S/Male I (20.26%) for [t], and Male S/Male I (22.16%) and Female
S/Female I (27.71%) for [t].

Once again, among the four groups A (0-2 years), 8 (2-5 years), C
(S- I 0 years), and 0 (IO years & above), female subjects in group B have
the minimum frequency of [t],- 4.28% when the interlocutors are male
and 2.88% when the interlocutors are female. Also, the female subjects
in group B have the highest percentage of [t] in their speech,- 7.47% when
speaking with male interlocutors and 8.30% when speaking with female
interlocutors. As the number of male and female subjects in each group
is only three, the possibility of the individual characteristics of one

subject affecting the occurrence of either variant [t] or [t] is very high.

75
This is a possible reason for the deviation seen in the speech of female

subjects in group B, which may be due to the particular characteristics
of subject No. 8 (see Table 4).

Also, the percentage of occurrence of [t] in the speech of female
subjects actually increases in group D, when speaking with female
interlocutors which is contrary to the usual trend of decrease in the
frequency of [E] from category C to D. The increase, however, is very

marginal and does not affect the total percentage.

4.2.8 SUBJECT-INTERVIEWER GENDER AND ETHNICITY
INTERACTION AND /t/

The following order of subject (S)-interviewer (I) gender and
ethnicity combinations was proposed in hypothesis h., 2.4.3. The
descending order of occurrence of [t] in the S-I gender and ethnicity
combinations is:

Male S/Male Indian I > Male S/Male American I > Male S/Female
Indian I > Male S/Female American I > Female S/Female Indian I >
Female S/Male American I > Female S/Female Indian I > Female 8/
Female American I.

The ascending order of occurrence of [t] in the fol lowing S-I gender
and ethnicity combinations is:

Male S/Male Indian I < Male S/Male American I < Male S/Female
Indian I < Male S/Female American I < Female S/Male Indian I <
Female S/Male American I < Female S/Female Indian I < Female 8/
Female American I.

The above orders were based on the assumption that subjects’ and

interviewers‘ sex would have more influence on the decrease in [t] and

76

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

..(P

8.: ...a. on: mod. 2%. on... 8... Rd. av... on... 93 8.: m... on... 8.2 :3. -2
o.v Foam «an on m...» oo.v o... 85 6.. 3.. mm. 8.». 21 3.. mm... mm. a
:3 Fan mom 3N mum van BM can we. 8N SN own ta KN .3 EN 6
mm... Now 63 mum mow EN mam ow 8.. 3.. now EN own no.» new mo.m m
Ba .3 on...” on. com on. ca 8. 86. on? and m3 .2» 9%. mom 8.... <

< _ < _ < _ < _ < . < _ < . < _

5. fish. EMS. Hz. 5...sz IE. 5...: #2. BEN: IE. 5...: E. “.25“. Hz. 3.2.. ,

983% 356... 382.8 .35. $838 SEW. $6038 SS. :5

-8

.26

 

 

 

 

E “.0 MUZmamDuuo ...o 00(thumwa

 

a. “.6 Bzéaauo “.0 35285:.

 

 

 

.3: c8265.. 6cm :0 £265 58 :2: 8332.35 0388 cam 29: 5.2, 9.28%

..ch 8832.... 29:8 ocm ENE ..o 53%. on”. E E van E ..o 8:2:me .2 2an

77
the increase in [t] in subjects’ speech than the“ ethnicity of the inter-

viewers. For instance, the percentage of [t] in the combination Male 8/
Male American I was predicted to be higher than in the combination Male
S/Female Indian I. This assumption underestimated the influence of
ethnicity of interlocutors on the the speech of Indian informants.

The results show in Table I I that interlocutors’ ethnicity was the
dominating factor; both male and female subjects accommodated to the
speech of American interviewers (interlocutors) by lowering the occ-
urrence of [t] in their speech, and simultaneously, increasing the
frequency of the American English variant [t], in their speech.

The new descending order, of subject-interviewer gender and
ethnicity combinations, in which [t] occurs is:

Male S/Male Indian I (16.41 %) > Female S/Male Indian I (I4.80%) >
Male S/Female Indian I (I 4.30%)> Male S/Male American I ( I 3.68%)
> Female S/Female Indian I (I I.78%) > Male S/Female American I
(I I.l5%) > Female S/Male American I (9.40%) > Female S/Female
American I (8.49%).

Female subjects accommodate more than male subjects to the
speech of American interlocutors,- for instance, the frequency of [t] in
the speech of female subjects decreases by 5.90% ( I 4.80% - 9.40%) and
by 3.29% (l I 78% - 8.49%), whereas the decrease in the speech of male
subjects is only 3.03% ( I 6.4I % - 13.38%) when speaking with the male
American Interviewer and 3. I 5% ( I 4.30% - I I . I 5%) when speaking with
the female American interviewer. Also, female subjects accommodate
more to the male American interviewer (5.90%) and male subjects
accommodate more when speaking to the female American interviewer
(3. I 5%).

78
When speaking with Indian interviewers the percentage of [t] is

higher in the speech of subjects than when speaking with American
interviewers (see Table I I).

Correspondingly, Indian subjects increase the frequency of the
American English variant [t] in their speech when speaking with
American interlocutors in the following ascending order of subject (5)-
interviewer (I) gender and ethnicity combinations:

Male S/Male Indian I (10.27%) < Female S/Male Indian I (10.98%) <
Male S/Female Indian I (I 1.39%) < Male S/Male American I (I 1.89%)
< Female S/Female Indian I (12.81%) < Male S/Female American I
(13.40%) < Female S/Male American I (14.36%) < Female S/Female
American I (14.90%).

Once again, female subjects accommodate more in their speech to
that of American interlocutors than male subjects. The percentage of
It] increases by 3.38% (14.36% - 10.98%) in the speech of female
subjects when speaking with the male American interviewer, and by
2.09% (14.90% - 12.81%) when the interlocutor is a female American.
Once more, female subjects accommodate more to the male American
interlocutor (3.38%) than to the female American interlocutor (2.09%),
and male subjects accommodate more to the female American inter-
locutor (2.01%) and less to the male American interlocutor (1.62%).

When speaking with Indian interlocutors the percentage of occur-
rence of [t] is lower in the speech of subjects than when speaking with
American interlocutors.

These differences in percentages, however, are very marginal. The
results of this interaction are, therefore, not statistically significant.

This is indicated in Table 12, 4.3.1, which gives the parameter

79
estimates. The estimate for [t] when correlated with 22 the interaction

of X2 (subjects’ gender), X3 (interviewers’ gender), and X4 (accommoda-
tion in the speech of subjects due to ethnicity of the interviewer) is .04,
and that of It] Is -.0433. The probability of the hypothesis being
rejected is .4513 for [t] and .4350 for [t], which is very high.

Thus, while the difference in percentages are interesting because

they refute the hypothesis postulated, they are not significant.

4.3.1 STATISTICAL SIGNIFICANCE AND INTERPRETATION OF ANALYSIS
Table 12 was obtained from the multiple linear regression analy—
ses conducted. It is evident from the table in both models (for [t] and
[t]) that variables X, to X5 are significant and X6 and the interactions 2,
and 22 are not statistically significant. However, some interesting
patterns, as described in 4.2 were observable even within the insignifi-
cant parameters. Since the number of observations was only 24, and the
total number of occurrences of [t] and [t] are subdivided under the
interactions as indicated in Table 10 and II, the statistical analysis
did not find the interactions significant as it did with variables X, to
X.
The effect of each of the significant independent variables on the
occurrence of It] and It] may be interpreted as follows:

a. X : as the period of residence of subjects in the United States

1
increases, It] in subjects’ speech decreases and [t] increases.

b. X2: In the speech of female subjects [t] Is lower, and [t] is higher.
In the speech of male subjects [t] is higher and [t] is lower.

c. X3: In the speech of subjects when speaking with female inter-
viewers Itl Is lower and It] is higher. When speaking with male

interviewers [t] is higher and [t] is lower.

80

Table 12: Statistical significance of independent variables

and interactions.

 

RETROFLEX It]

ALVEOLAR It]

 

 

VARIABLE LABEL pARAMETER p- PARAMETER p-
ESTIMATE VALUE ESTIMATE VALUE
x, PERIOD OF RESIDENCE or
SUBJECTS IN THE US. -.3867 .0001 .3781 .0001
, SUBJECTS' GENDER .31324 .0001 ~.2975 .0001
3 INTERLOCUTORS’
GENDER .13868 .0130 -.1344 .0161
x, ACCOMMODATION IN
SUBJECTS’ SPEECH DUE
TOINTERLOCUTORS’
ETHNICITY -.2356 .0001 .2431 .0001
x5 SUBJECTS' ATTITUDE TO
THEIR NEw SOCIO-
CULTURAL ENVIRONMENT -.4454 .0001 .4517 .0001
x,5 CONTEXTUAL STYLE -.O484 .3815 .0446 .4213
z, INTERACTION BETWEEN ,
x2 AND X3 .007 .8866 -.0065 .9058
z, INTERACTION BETWEEN
x,, x,, AND x, .04 .4513 -.O433 .4350

 

 

 

 

 

 

 

R-SOUARE FOR THE MODEL

SIGNIFICANCE OF THE MODEL - [E] - .0001, [t] - .OOOI.

- It] - .4411, It] - .4395.

 

 

81
d. X4: When speaking with Indian Interviewers [t] is higher in subjects’

speech and [t] is lower. When speaking with American interview-
ers [t] is lower and [t] is higher.

e. X - If the speakers’ attitude toward their new socio-cultural

5 .
environment is negative, [t] is higher and [t] is lower. If the

speakers’ attitude is positive, then [t] is lower and [t] is higher.

4.3.2 SOCIOLINGUISTIC EXPLANATION OF RESULTS

In section 2.4.3, in the hypotheses, and in previous sections, certain
factors were assumed to affect the frequency of occurrence of the
variants [t] and [t]. The numerical pattern of occurrence of these
variants reported in 4.2, therefore, suggests the definite presence of
sociolinguistic phenomena which are the result of the underlying
sociological factors.

The decrease in the percentage of the Indian English variant [t]
when the period of residence of subjects in the United States increases,
as seen in Table 4 in categories A to D from 38.57% to 15.85%, and
simultaneously, the increase in the percentage of [t], the American
English variant, from 16.94% to 30.47% suggest the following. This
pattern of occurrence is due to the transition in process in the speech
of the Indian immigrants studied, from Indian English acquired in India,
to American English which is currently being acquired by those speak-
ers. The state of flux in the dialect of these first generation immi-
grants Is evident in the fluctuation between [t] and [t], facilitated by
other factors such as gender, and accommodation to speakers of
different ethnic groups.

The figures in Table 4 also suggest that the above transition is

82
maximum between categories A (0-2 years) and 8 (2-5 years); the

percentage of decrease in [t] is 13.82 (38.57% - 24.75%) from A to B. The
percentage of increase of [t] is 8.39 (25.33% -16.94%) from A to B. The
decrease in [t] and the increase in [t] in categories C and D is relatively
gradual (see Table 4).

It is possible to conclude from the above figures that the transition
in the dialect, or an attempt towards bidialectalism, occurs most in
group B and not in groups C or D. This conclusion seems reasonable when
the social status of these speakers in the current society is considered.
Speakers studied in groups C (5-10 years) and D ( 10 years & more) were
professionals, mostly professors who are well established within
American society. The subjects considered in those groups were either
permanent residents or citizens of the United States and, therefore, are
socially secure. Their socioeconomic stability 15 reflected in the
relative stability in their dialect usage. This in turn, indicates their
linguistic security, which is evident when compared with subjects In
group 8.

Speakers in group B were graduate students (on a foreign student
visa). This sample of the population is in the process of establishing
itself within the society. Within this group there is thus a crucial
motivation to be accepted by the majority community of native Ameri-
cans, depending on whether or not the speaker has a genuine liking for
the social environment (see 2.3.2.1 and 4.2.4). Hence, the transition in
the speech of group B is more drastic. Also, their relative socio-
economic instability and social insecurity as members of an unfamiliar
society are reflected in the linguistic insecurity in their speech seen

in the maximal fluctuation in speech due to sociological factors.

83
Relative linguistic security/insecurity is, therefore, directly

related to the rate at which a particular social stratum within the
society undergoes a change in its speech patterns, or the extent to
which it responds to social factors. The more linguistic security, and
thereby, social stability, there is, the less it is susceptible to linguis-
tic fluctuations; the less linguistic security and therefore social
stability there is, the more it will respond to linguistic change. For
instance, the relative linguistic insecurity of subjects in group A
accounts for the high percentage of the Indian English variant It]
(22.51%) in group A (0-2 years) In Table 6, when speakers communicate
with Indian interlocutors, and the relatively lower percentage of [E]
(16.06%) when the same speakers accommodate to the speech of
American interlocutors. The maximum speech convergence is found in
category A - 6.45% for [t] and 4.43% for [t], and the minimum conver-
gence in speech 15 found in group D - 2.67% for [t] and 1.25% for [t].

In the study of English in New York City, Labov (1972:122-142)
discusses the phenomenon of hypercorrection by the lower middle
class. The phenomenon is explained in, for instance, the usage of (r),
the prestige norm of that society, by the lower middle class in
phonological environments where it Is not normally used, as a result of
an “extreme sensitivity to the norms of an exterior reference group"
(52). The lower middle Class Is determined as'that strata within the
society which has the least linguistic security.

In the current study, in 3.4.1, it was noted that only subjects in
groups A and 8 used the lateral affricate [ti], originally a British English
phoneticvariant of /t/, in their formal speech. Using the same rationale

as in Labov (1972) it can be argued that the occurrence of [tic] in the

84
groups which have the least linguistic security is an instance of 'hyper-

accommodation". The phenomenon of 'hyperaccommodation' differs
from that of hypercorrection as [ti] is not an existing norm of the
exterior reference group of native American English speakers. How-
ever, [t4] is used by the informants in categories A and 8, probably, as
a better alternative to the Indian English [t] and as the next best
alternative to American English [t], in an attempt to accommodate to
the dialect of the exterior reference group. Thus, it is the very
existence of [ti] and not the occurrence of [tat] in an unusual phonological
environment that is in question. The explanation for the existence of
[tic] in the informants’ speech is given in 3.4.1.

The influence of subjects’ gender and interlocutors' gender on /t/
can be interpreted as follows. The study of [t] and [t] in relation to
subjects’ gender (Table 8) and interviewers” gender (Table 9) shows
maximum differences in percentages in groups A and 8, indicating once
more that there is maximum linguistic insecurity in these groups. The
change in frequency from group C to D is more gradual, indicating
maximum linguistic security.

From Table 8 and Table 9 one can conclude that subjects’ gender as
an independent variable is more effective than interlocutors’ gender in
influencing the occurrence of [t]/[t] in speakers. Out of the total number
of /t/ (3840), I 426 were [t] and 2396 were [t] in the Speech of subjects
studied. Out of 1426 [t], correlation with speakers' gender elicited
55.54% in the speech of male subjects and 44.46% in the speech of
female subjects. Of the 2396 [t] 46.95% occurred in the Speech of male
subjects and 53.05% in that of female subjects.

The result of the correlation of /t/ with interlocutors' gender was

85
as follows. Out of 1426 [t], 54.28% occurred when subjects spoke with

male interlocutors and 45.72% occurred when subjects spoke with fe-
male interlocutors. Of 2396 [t], 47.50% occurred when the interlocu-
tors were male and 52.50% occurred when the interlocutors were
female.

Lastly, the study confirms the presence of integrative motivation
discussed in 2.3.2.1. A positive attitude in Indian immigrants toward
their current environment - its culture, its people and lifestyle-
provides an incentive to acquire the American English dialect. On the
contrary, a negative attitude inhibits the process of acquisition of the
new dialect. In this study, [t] and [t] are reliable indicators of the extent
of retention of Indian English and the extent of acquisition of American
English. Thus, a positive attitude is represented by a higher average
occurrence of the American English [t] -1 15.42 per person, and a low
average occurrence of Indian English [t]- 43.58 per person (see Table 7).
A negative attitude is represented by a relatively lower average
occurrence per person of [t] - 84. 13, and a higher average occurrence per
person of [t] - 74.25.

Chapter 5 considers the implications that the above phenomena

have for certain linguistic theories.

86

NOTES

1. To date there is no documentation of the different norms of the IE
/t/. Perhaps, [t2] is a hyper-prestige norm of Indian English as it is
originally a British English phonetic variant of /t/. [t%] may be
fossilized in IE as the postvocalic (r-zi) of /r/ (Sahgal and Agnihotri
1988) and is used in the speech of only those IE speakers whose
proficiency is relatively high.

The term ’hyperaccommodation’ is used here to refer to the phenome-
non when (immigrant) speakers of a social stratum with the least lin-
guistic security use the norm of a previous exterior reference group,
in this case standard British English speakers or high-prestige IE
speakers, to accommodate to the speech of the current exterior
reference group of American English speakers, in formal speech

styles.

CHAPTER 5
CONCLUSION

5.1 INTRODUCTION

This chapter has three major subdivisions in sections 5.2, 5.3, and
5.4.

Section 5.2.1 deals with ways in which this study could have been
improved, so as to achieve better results. 5.2.2 discusses topics for
further research.

Section 5.3 discusses the implications this study has for certain
theories proposed in the areas of Sociolinguistics and Second Language
(dialect) Acquisition.

Section 5.4 concludes this thesis witha brief summary and discus-

sion.

5.2.1 SUGGESTIONS FOR IMPROVEMENT ON THE CURRENT STUDY
This study has certain shortcomings, especially in the experimen-
tal design, which may be avoided in future research of a similar nature.
1. Physical Setting : A more informal environment than that of a
Classroom could have allowed for more free flowing conversation,
and thereby, everyday speech within the interview situation. It is .
not clear, however, that this was the cause of the difficulty in
obtaining much difference between formal and informal contextual

styles in this study.

2. Sample Size: A larger sample size would have been desirable. For

the study of difference in language use based on gender within the

87

88
four categories A, B, C, and D, only 3 male and 3 female subjects

were used in each cell. Inspite of the small number of observations
the result was significant (see 4.2.5). The small numbers within
each category resulted in the insignificance of subject-interviewer
gender interaction and subject-interviewer gender and ethnicity
interaction. Studying Speech within a larger number of male and
female speakers, however, would strengthen the possibility of
capturing regularities in the speech patterns of male and female
subjects within each category: the Chances of the interactions being
significant are more when the number of observations is large.

3. Reliability : A second phonetically trained ear, besides myself,
would have provided a check on the accuracy of my judgement as to
the phonetic nature of /t/. An acoustic analysis would have further
ascertained the phonetic nature of /t/.

4. Contextual Style : The scale of formality/informality, in section
3.3.3, shows that the contextual styles used in this study - conver-
sation during an interview (informal speech), and the reading of a
phrase list (formal speech) - are too close to each other on the scale.
While informal speech was obtained, most formal speech was not
obtained due to ‘the practice effect’ discussed in 2.3.3. This perhaps
was the reason that speech in these two styles did not show much
difference in the occurrence of [t]/[t]. Two contextual styles further
apart on the scale, such as casual speech and minimal pairs, could
have shown greater variance between formal and informal styles.

5. Duration of Sessions : Section 3.3.1.2 discusses the time taken for
each session. The first session was 8 U2 hours long while the

second was only 3 1/2 hours and the third was only 2 hours long. It

89
is possible that long sessions such as the first caused interviewer

fatigue. As a result there probably were some interviews in which
interviewers did not actively participate. This in turn could have
caused the subjects not to speak as in a regular conversation. This
is not evident In the study as it can be noted that many of the last
interviews in the first session were long interviews. Because
interviewers encouraged subjects to speak, subjects dominated the
conversation, and were not affected by interviewer fatigue, if any.
However, this factor could affect interviews and it would, therefore
be desirable to have shorter sessions of equal length.

6. Tape Recordings : A11 tape recorders were checked before each
session and were in good working condition. The tape recorders were
interchanged between sessions but were not interchanged between
interviews. As the three sessions were not of equal duration it is
possible that the quality of the recording was biased towards a
particular set of interviews depending on which tape recorder was
used.

7. The Phrase List : This was identical for all interviews and for the
four interviews of each subject. Although the interview question-
naire was Identical, speakers were asked to choose between three
questions. Thus, there was more flexibility in answering questions
than in reading the phrase list. The same target words used in
different phrases could have reduced the effects of ‘subject fatigue’
and ‘practice due to learning’.

8. Determination of speakers' attitudes: This factor was determined
based on whether or not subjects said that they liked or disliked the

United States, its people, its culture and its lifestyle. The deciding

90

factor was whether speakers intended to stay in the United States
or wanted to return to India. However, this method to determine
speakers' positive or negative attitudes was not sufficiently valid.
More recent studies on attitude research need to be consulted and

better methods of evaluation could be used.

5.2.2 SUGGESTIONS FOR FURTHER RESEARCH

I .

During the course of this study other possible variables were
observed. These variables occur due to the specific nature of the
phones of Indian languages which in turn results in the particular
nature of English spoken by Indians. Thus, a number of variables
may be studied in the speech of Indian immigrants in the United
States. The existence of the following variables needs further
substantiation in the form of a pilot study prior to a large scale

study.

a. The voiced counterpart of /t/ which is /d/ in English is pronounced

in IE as retroflex [d] - for instance, in ‘dinner’ as [d1 na]. Id] like [t]
is likely to undergo a transition in the Speech of immigrants in the

United States which may be studied.

. Indo-Aryan and some Dravidian languages such as Kannada, (in

Tamil, a dravidian language, aspirated stops do not occur) differen-
tiate phonemical ly between aspirated dental stops such as /tp/, and
unaspirated dental stops as /t/. Native speakers of Indian lang-
uages may use either [tan] or [t] in Indian English in the place of the
voiceless interdental fricative [O] of American English, in words
such as ‘thin’ and ‘theater’. The use of [th] or [1;] along with [O] in the

speech of Indians in the United States can form the basis for

91
monitoring variation and sound change in their speech.

c. In Indian languages, there is no phonemic distinction between [w]
and [v]. In environments where American English Clearly requires
distinguishing between [w] and [V] as in ‘wine’ and ‘vine’, IE seems
to rely entirely on context to distinguish between such minimal
pairs. Firstly, the factors that condition the pronunciation of either
[w] or [V] in Indian languages and as a consequence in IE, if any, needs
examination. Secondly, if these phones are not phonetically condi-
tioned allophones in IE, then this seemingly arbitrary use in the
English of the immigrant population, and a transition, if any, may be
studied using [w] and [v] as possible indicators, and by measuring a
change in their use. At present the entire topic is available for an
empirical study.

d. In Indian English, the long, central, low vowel [a2] is used in the
environment of C __ sC,,‘ and C _ s, as in, for instance, ‘fast’ -
[fast], and ‘pass’ - [pazs]. Standard American English speakers use
[881111 these environments. A study of [a:]/[ae] in the speech of first
generation Indian immigrants is, therefore, possible.

e. Finally, /r/ in Indian languages is either a trill or a flap. As a result,
Indian English speakers use [1‘] in all environments where standard
American English speakers use [1‘] which is a retroflex consonant.
Sahgal and Agnihotri ( l 988) mentions the fossil ization of the post-
vocalic (r-0‘) in IE. Thus, the transition from (r)/(r—21) to [r], if any,
could be studied in the speech of the immigrant population.

2. The current study raises the following methodological question
which could be a topic for further study. An analysis could be done

of where the the critical sample of casual speech is most likely to

92
occur within the four interviews of each subject. One could

speculate that the sample of the most informal speech would occur
either in the beginning of the first interview when the subject is
first given an opportunity to speak on one or more broad based
questions, or beyond a point in the conversation when the subject is
overcome by fatigue and does not care to self-monitor anymore.
This is most likely to happen in the third or fourth interview.
Studying the data for signs of most uninhibited verbal behavior

COUIG substantiate the above speculation.

5.3 IMPLICATIONS FOR THEORIES OF SOCIOLINGUISTICS AND SECOND
LANGUAGE (DIALECT) ACQUISITION

Based on the results obtained in this study it would be desirable to
make general inferences about the entire population of first generation
Indian immigrants. However, due to the small sample size used in this
study it is not possible to make such generalizations. If, in the future,
similar studies of other variables in the English spoken by Indian
immigrants support the same theories as this study, then the possibil-
ity of generalizations would be strengthened. At this point it must
suffice to say that the study of /t/ within the 24 samples supports
certain theories in the field of sociolinguitics and supports a theory of
second language acquisition.

The study supports the theory proposed in Bailey (1973) and Giles
(1979) that in a language (dialect) contact situation, given time, a
sound change will take place in the direction of the language or dialect
which is spoken by the majority community in a native capacity, and,

therefore is the favored language or dialect of that majority community

93
(see 2.3.2.2 for discussion and 4.2.1 for results).

‘ The study strongly supports the theory of Speech convergence
proposed in Giles and Powesland (1975) and Giles (1979). Refer to
2.3.2.3 for discussion on interpersonal accommodation. The degree of
accommodation of immigrants to speakers of the majority community
once again, among other factors, depends on speakers' attitudes to their
new environment.

The current study also favors the proposition in Trudgill (1974)
that allowing for other factors, women tend to use the standard variety
more than men (refer to sections 2.3.2.4 and 4.2.5).

The extent to which a new language or dialect is acquired is
directly related to the speaker’s attitude to his/her new environment
and definitely depends on whether or not the speaker is integratively
motivated to acquire the language or dialect (see section 2.3.2.1 for
discussion and 4.2.4 for results). In the current study, all subjects had
acquired IE formally as a second language and had immigrated to the
United States only after the age of 18 (see section 3.2. 1 C). Although,
a comparative study between adult and child immigrants would give a
better perspective, the study shows that at least some adults are able
to successfully acquire the American English pronunciation(s) of /t/.
The crucial factor determining the rate of acquisition as observed in
this study and as proposed in Falk (1979), is whether or not the speaker
possesses integrative motivation, instrumental motivation or no
motivation at all. For instance, subject No. 19 (see Table 4, 4.2.1)
possessed integrative motivation, therefore, her rate of acquisition of
the American English pronunciation was higher than, for instance,

subject No. 20 (see Table 4) who did not possess integrative motivation

94
i.e., an inherant liking for the new environment.

5.4 SUMMARY AND DISCUSSION

The current study shows a sound change in process from [t] to [t]
and, thereby, a transition in progress from Indian English to American
English in the speech of first generation Indian immigrants. Given time,
it is possible that the transition will be complete. The data in this
study, however, do not indicate a completion as can be seen in the study
of other immigrant populations. The following is perhaps the reason
for the incomplete sound change. Immigration of Indians to the United
States, when compared with other ethnic groups, is a relatively recent
phenomenon which began in the 1950’s and became popular in the 1960’s
and 1970's. Due to the new laws of immigration and naturalization in
effect in the United States since the 1980’s, the proportion of perma-
nent immigrants (permanent residents and naturalized citizens) to that
of temporary immigrants (students who return to India on completing
their education) In the United States since that time, has decreased
though not considerably.

Hence, studying the transition in the Speech pattern of permanent
immigrants would give a more realistic picture of the extent of sound
change in the speech of the population of first generation immigrants.
This would, however, exclude from the study, the student population and
perhaps others who have resided in the United States less than five
years.

In conclusion, an extensive study of a number of variable phonemes
in transition would monitor more clearly the acquisition of American

English by first generation Indian immigrants.

95

NOTES

1. C denotes - consonant, s refers to /s/, and , indicates word

boundary.

APPENDICES

APPENDIX A
QUESTIONNAIRE
Please answer ONE OR MORE questions speaking for approximately 10
minutes.
I. How did you feel about the United States when you first came here?

2. Do you like /dislike the United States, its people, the culture, the
lifestyle, etc? Please elaborate.

3. Have you had an experience in your life which has caused or seemed
to cause danger to your life? If yes please narrate.
Please read ALOUD the following phrases.
as wet as the weather
as light as a petal
the wide western desert
a water pistol
a winter storm
what a content person
a widely practiced custom
one particular waiter
for contempt of court
with potential skills
the United States Postal Service
the Post and Telegraphs Department

a nice teacher

96

APPENDIX B

SUBJECT LIST

ID. NO SUBJECTS’ GENDER PERIOD OF RESIDENCE (in months)

1 female 6

2 male 9.5
3 female 10
4 female 15
5 male 17
6 male 21
7 female 26
8 female 31
9 male 34
10 male 45
11 female 50
12 male 57
13 male 71
14 female 78
15 male 83
16 female 95
17 female 100
18 male 106
19 female 138
20 male . I92
21 female ' 247
22 male 288
23 female 300
24 male 420

97

BIBLIOGRAPHY

GENERAL REFERENCES

Anshen, Frank. 1978. Statistics erLinguists,Rowleszassachu-

setts, Newbury House.
Chengappa, Raj. 1988. “Family Planning : The Great Hoax” India

May, (issue: October 31) Vol.XIII:20., New Delhi : Living
Media India Ltd.

98

LIST OF REFERENCES

Ainsworth, Johnson, N. 1974. “Zombies and other problems: Theory
and Method in Research on Bilingualism,” Language Learn-
tm, 24.1:105-133.

Bailey, N. 1973. Yaniatien anti Linguist]; lheery, Arlington: VA,

Center for Applied Linguistics.

Bhatia, K. 1967. Hian mm 6932': keagats'and'é Isa bbasa tatuik
annuauan, Allahabad : Hindustani Academy.

Bourhis, Richard, Y. 1984. “Cross-Cultural Communication in
Montreal: Two Field Studies Since Bill 101,” Intecnatienal

JQUFDQI at the Sneielngy QI Language, 46:33-47.

Brouwer, D., Gerritsen, M. and De Haan, D. 1979. “Speech
Differences between Men and Women : On the Wrong

Track?,” Language in Seeietu, 46:33-50.

Daniel, Jones. 1943. An English Enennuneing Qietinnanu, New York
: E. P. Dutton & Co., Inc.

Ervin-Tripp, Susan. I964a. “ An Analysis of the Interaction of
language, topic, and listener,” John, J., Gumperz and Dell,
Hymes (eds), The Ethnography of Communication,

American Anthrnnnlegist, 66286-102.

Falk,J.,S. I979. “LanguageAcquisitionandtheTeachingandLearning
of Writing,” Cellege English, 41:4.436-447.

Fasold, R. I984. “Language Attitudes,” h 1' lin I 1' f
SQQlQEX, I46-I79, Oxford: Blackwell.

99

lOO
Gardner»R- andtambertiw. 1972. WWW

Secdnd Language Leacning, Rowley, Massachusetts

Newbury House.

Giles, H. and Powesland, P., F. 1975. “Speech style and social
evaluation,” r i i o ,
New York : Academic Press.

Giles, H. 1979. “Ethnicity Markers in Speech,” i r r
Sneecn, K., R., Scherer and H., Giles (eds), 251-290,
Cambridge : Cambridge University Press.

Hawkins, Peter. 1984. Lntnddugngendnnldgyfioverfiew Hampshire
: Hutchinson &Co. Ltd.

Kachru, Braj, B. 1977. "Linguistic Schizophrenia and Language
Census: a Note of the Indian Situation", Linguistics, I86: 17-
32.

Kachru, Braj, B. 1983.1netnd1anlzatldncLEngjjsn,New Delhi :Oxford

University Press.

Kachru, Braj, B. 1988. “Teaching World Englishes,” W
Bulletin, 12:1.1-8, Washington, DC. : ERIC Clearinghouse on
Languages and Linguistics.

Labov, William. l966a. ijne Sdcia15tratiticatidn Qt Englisn in New
Ydrk City, Washington, DC. : Center for Applied Linguistics.

Labov, William. 1970. “The logic of non standard English,” Frederick,
Williams (ed), Language and Edvecty, 159- I 89, Chicago:

Markham Publishing House.

Labov, William. 19720 “The recent history of some dialect markers
on the island of Martha’s Vineyard, ” L., Davis (ed.), Studies
presented td Raven McDauid, University, Ala.: University of

101
Alabama Press.

Labov, William. 1972. “The Social Motivation of Sound Change,”
WW 1-42, Philadelphia:University of

Pennsylvania Press.

Labov, William. 1972. “The Isolation of Contextual Styles,”
Sncjnungtnstjcflattenns, 70-1IO,Philadelphia:University

of Pennsylvania Press .

Labov, William. 1972. “Hypercorrection as a Factor in Linguistic

Change,” Wm, 122-142, Philadelphia:

University of Pennsylvania Press.

Labov, William. 1972. “The Social Stratification of (r) in New York
City Department Stores, ” W 43-69.

Philadelphia: University of Pennsylvania Press.

Lavendera, B. 1978. “ Where does a Sociolinguistic Variable Stop?, ”

W 72171-182.

Quirk, R., Sidney, G., Geoffrey, L., and Jan, Svartvik. 1972. A
W. London : Longman.

Sahgal, A. and Agnihotri, R., K. 1988. Indian English phonology: a
sociolinguistic perspective, Englisn Wdrld-Wide, 9:51 -64,

Amsterdam : John Benjamins B. V.

Scotton, C., Myers. 1985. “What the heck, sir: style shifting and
lexical colouring as features of powerful language, " Street,
R. and J. N. Cappel (eds), Sequence and Battem__in
WW, 103-1 19, London: Arnold.

102
Sturtevant, E. 1947. An |ntrdductidn td Linguistic Science, New

Haven : Yale University Press.

Trudgill, Peter. 1974. “Language and Sex," 1 i i ’ : A
intcdductidn td | anguage and Sdciety, 78-99, Reading: Cox
and Wyman.

Trudgill, Peter. 1974. "Language and Context," Sdcidljnguistics ;An
Intcdductidn tn Language and Sdciety, 100-122, Reading:

Cox and Wyman.

Wald, B. and Shopen, T. 1981. “ A Researcher’s Guide to the
Sociolinguistic Variable (ING),” T. Shopen and J. Williams
(eds), Style and Varianles inEnglisn, 219-249, Cambridge
: Winthrop.

Wolfson, N. 1976. “Speech events and natural speech : some
implications for Sociolinguistic Methodology," Language in
Sccietu, 5:189-209.

Zwicky, A. 1981. “Styles,” T. Shopen and J. Williams (eds), Styles
and Vanianles in Englisn, 6383, Cambridge: Winthrop.

HIGRN STRTE UNIV.

IIIIIIIIIIIIIIIIII92 :ISIII