Acquire A Non-Native Phonology

Acquiring a Non-native Phonology
This page intentionally left blank

Acquiring a Non-native Phonology
Linguistic Constraints and Social Barriers
Jette G. Hansen
continuum
Continuum
The Tower Building 80 Maiden Lane
11 York Road Suite 704
London SE1 7NX New York, NY 10038
USA
First published 2006.
Jette G. Hansen 2006
All rights reserved. No part of this publication may be reproduced or transmitted in
any form or by any means, electronic or mechanical, including photocopying,
recording, or any information storage or retrieval system, without prior permission in
writing from the publishers.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library.
ISBN: 0-8264-6862-4 (hardback)
Typeset by Data Standards Ltd, Frome, Somerset, UK
Printed and bound in Great Britain by Antony Rowe Ltd., Chippenham, Wiltshire
Contents
1 Introduction
2 Investigating L2 Phonology
3 Researching the Nguyen Family
4 Acquiring a Non-native Syllable Structure
5 Linguistic and Task Constraints
6 Social Barriers
7 A Unified Approach
8 Conclusions
References
Appendices
Indexes
1
7
40
59
95
126
153
164
168
177
194
This page intentionally left blank
1
Introduction
In this introductory chapter to the book. Acquiring a Non-native
Phonology: Linguistic Constraints and Social Barriers, the purpose of the
book, as well as a brief introduction to the study and its background,
is presented. Finally, an overview of the structure of the book is given.
1.1 Purpose of the book
The following volume is a monograph study of the English second
language (ESL) phonological development of a Vietnamese family
across nearly one year. Specifically, the book examines the acquisition
of consonants (syllable-initial and syllable-final consonants and
consonant clusters) by a Vietnamese husband and wife who
immigrated to the United States one year prior to the commencement
of the study. In examining the acquisition of these consonants,
linguistic factors, such as transfer and markedness, and social factors,
such as opportunities for second language (L2) use, extent of first
language (LI) and L2 use, and gender, among others, were analysed
in order to examine their individual and combined effects on the
acquisition of consonants by the participants. Similarities and
differences between the two participants in their phonological
development were also analysed with respect to the linguistic and
social factors.
Although the study of the acquisition of an L2 English phonology,
and especially the acquisition of English consonants and consonant
clusters, has received attention from second language acquisition
(SLA) researchers, the studies in this area have been mostly one-off
rather than longitudinal, and thus have examined production, rather
than the process of acquisition. In addition, the focus has typically
been on either linguistic or social constraints, rarely both.
The acquisition of L2 English consonants and consonant clusters
has been a major area of study in L2 phonology as English has a
relatively complex phonology, with a wide range of consonants able to
occupy both syllable-initial (onset) and syllable-final (coda) position,
2 ACQUIRING A NON-NATIVE PHONOLOGY
as well as allowing a number of consonants to precede and follow
each other in the onset (up to three consonants in a row) and coda
(up to four consonants in a row) while other languages, such as
Vietnamese, have relatively simpler types and lengths of consonants
in onsets and codas. L2 learners of English whose LI is simpler, such
as Vietnamese, typically modify codas in production, via epenthesis,
devoicing, and/or absence in production, in favour of shorter, and
thus simpler, codas. While researchers agree that this is a common
process for learners of English, including LI learners of English, there
is a lack of agreement over what linguistic processes (e.g. LI transfer
or markedness) affect the learners' production and acquisitions, and
even what types of modification processes learners favour. One reason
for the lack of consensus is the fact that the majority of the research in
L2 phonology has been one-off, rather than longitudinal, with studies
being done with learners of different ages and language levels, making
comparisons among studies difficult. Therefore there is a lack of
knowledge about how processes interact and change over a longer
period of language acquisition.
Additionally, there has been little research on the influence that the
social context of language learning has on the acquisition on L2
phonology, although some one-off research has focused on the effect
of the social context on L2 production (cf. Adamson and Regan
1991; Beebe 1980; Beebe and Zuengler 1983; Dowd, Zuengler and
Berkowitz 1990; Schmidt 1974, 1983, 1987). Recent work (e.g.
Gumming and Gill 1992; Goldstein 1995; Norton 2000) in SLA has
also suggested that the socially structured and reinforced gender roles
of both the LI and the L2 culture may have an effect on opportunities
for L2 development. Additionally, work by Flege and colleagues (e.g.
Flege, Frieda and Nozawa 1997; Guion, Flege and Loftin 2000; Piske
and MacKay 1999; Piske, MacKay and Flege 2001) has suggested
that what is crucial in L2 accent is not the amount of L2 use but the
extent of LI use; that is, individuals may be more likely to retain an
accent in L2 if they have greater use of LI, whereas greater use of L2
may not minimize an accent in L2. A greater understanding of how
these social factors constrain L2 acquisition, and how social factors
and linguistic factors interact in the acquisition of an L2, is crucial to
theory testing and development in SLA as well as for L2 pedagogy.
While a number of long-term case studies and in-depth analyses of
individual learners acquiring an LI phonology (i.e. child LI phono-
logical acquisition) (e.g. Leopold 1939; Macken 1979) are considered
to be classics in the field, there have been few longitudinal or in-depth
analyses of individual learners acquiring an L2 phonology. Yet,
INTRODUCTION 3
studies of this kind may provide researchers and teachers with
information on developmental patterns for comparison with their
own studies and/or students, as well as a more comprehensive
overview of how both linguistic and social factors impact language
development.
Therefore, the purpose of this book is to illustrate - via a detailed
analysis of the acquisition of L2 consonants - developmental
sequences in acquiring an L2, and the social and linguistic factors
that influence this development at any given time in the acquisition
process. This book is unique not only due to its focus on both social
and linguistic factors, and their interaction within and across time,
but also due to its longitudinal nature, and finally, the comprehensive
and detailed analysis of the data due to a focus on one family. As
such, this book fills a unique void in the SLA literature - that is, it is a
comprehensive monograph study with detailed analyses across a
longitudinal time span of L2 acquisition.
The volume may be appropriate for a number of readerships,
including students in applied linguistics, linguistics, or TESOL
programmes; and researchers and teachers interested in the teaching
and learning of an L2 sound system. For the first audience, the
monograph study is designed to supplement SLA texts by examining
in detail the case of one family's acquisition of an L2, grounding the
examination of this acquisition within existing SLA theoretical
frameworks (e.g. psycholinguistic and socio-cultural theories), and
challenging these existing frameworks to account for the phenomena
observed in these data. This book should be of interest to teachers
since it will provide them with concrete examples of SLA processes
via a detailed discussion of actual language learners, and by making
theory and research accessible and meaningful to their needs via
specific examples and illustrations.
For language researchers, this book should be of interest because of
the very detailed analysis of SLA processes, and the extensive
presentation of actual data, which will allow researchers to gain a
deeper perspective of the phenomena observed in the study.
For the second audience, researchers and teachers interested in L2
phonology and pronunciation teaching in specific, this monograph
provides an in-depth analysis and discussion of the developmental
processes in acquiring a non-native sound system - presenting a rich
base of L2 phonological data across a longitudinal time frame - which
has not previously been presented in the literature. This comprehen-
sive database should be of interest to anyone concerned with the
acquisition of an L2 phonology. Additionally, it should also be of
interest to L2 language teachers as it provides them with concrete
examples of the pronunciation production and modification processes
of learners, as well as specific details about why these productions and
modifications take place.
1.2 Background to the study
The present study examines the acquisition of English L2 consonants
over nearly one year and focuses on learners of spoken English who
have entered the United States around one year before the onset of
the study. Consonants were the focus of the study because this is an
area of difficulty for many learners of English and therefore onset and
coda acquisition is an important pedagogical concern in L2 teaching
since modification of these can affect communicative efficacy and
debilitate communication and comprehension.
Speakers of Vietnamese were chosen in part because Vietnamese
has a relatively simpler phonological structure in comparison with
English, and English onsets and codas are difficult for these learners
to acquire. Participants who had been in the United States for around
one year before the commencement of the study were chosen as these
were still relatively new learners or in the early stages of L2
development while at the same time having some experience in L2
to enable them to communicate in English during the conversations
that were part of the data collection techniques.
The family under study was selected during my teaching at a
community college in Tucson, Arizona. One of my students, a
woman from Vietnam, was my student in an introductory ESL
speaking and listening class. After the course was completed, I
approached her about her interest in participating in the study, and
after agreeing to do so, she suggested meeting in her home and the
addition of her husband as a participant in the study. I agreed to
include her husband in the study as a family provides an ideal
opportunity to make comparisons between speakers as the members
of the family have similar LI backgrounds. Additionally, the family
provides a unique opportunity to research how the differential social
networks and identity formations of the members of the family
influence L2 acquisition, as well as what variety of English is targeted
for acquisition as based on social (e.g. peer) networks and in-group
identification. Communication within the family can also be exam-
ined. Insights into language use, language resistance and language
maintenance within a family are crucial in understanding how social
forces affect language acquisition and attrition.
INTRODUCTION 5
This study examines linguistic and social factors in order to
ascertain how linguistic and social factors individually and interac-
tively constrain, both positively and negatively, L2 acquisition. Data
on social factors were collected via interviews, participants' language
logs (language use journals), and observations, and data collection
took place approximately once a week for the duration of the study.
Interviews were always tape-recorded, and field notes were also taken
during the interviews and observations. The topic of the interviews
were the participants' LI and L2 language use patterns, social
interactions, and daily routines, which were analysed qualitatively in
order to understand the social constraints on the participants'
language use and acquisition. Speech samples for phonological data
analysis were selected from the one-on-one interviews at three month
intervals during the study and analysed via both descriptive and
inferential statistics.
1.3 Structure of the book
The structure of the book is as follows: this chapter has provided an
overview of the purpose of the book, the background to the study, and
the structure of the book. Chapter 2: Investigating L2 Phonology, is a
review of the literature in three major areas: linguistic constraints,
including LI transfer, developmental effects, markedness and lin-
guistic environment; social factors, including gender, social identity
and extent of LI and L2 use; and variation based on linguistic, social
and task effects. An overview of both English and Vietnamese
phonology is also provided in Chapter 2. Chapter 3: Researching the
Nguyen Family, details the selection of the participants, the
background of the family as well as that of Vietnamese immigrants
to the US, as well as how data - both phonological and social - were
collected and analysed. Chapter 4: Acquiring a Non-native Syllable
Structure, presents the findings from the linguistic analysis. Findings
from both onsets and then codas are first presented individually,
followed by a comparison of onset and codas findings to examine
developmental patterns. Chapter 5: Linguistic and Task Constraints,
presents an analysis of the linguistic findings in light of linguistic
constraints such as transfer, developmental effects and markedness,
as well as linguistic environment, grammatical conditioning and task
effects. Chapter 6: Social Barriers, first provides a detailed overview
of the social contexts of language use for the participants across the
duration of the study, and then an examination of the social factors,
including social identity, gender and opportunities for LI and L2 use,
that affect the participants' language development within and across
time. In Chapter 7: A Unified Approach, discussion of the stages of
onset and coda acquisition that the participants in this study show
across time is provided. This is followed by an analysis of the
interaction of social and linguistic constraints. Finally, the effect of
social barriers on the acquisition of L2 syllable margins by the
participants is discussed. In the final chapter. Chapter 8:
Conclusions, a summary of the findings from this study on the
acquisition of an L2 phonology is given. The limitations to this study
are then given, followed by suggestions for future research. Finally, an
epilogue to the story of the Nguyens, the participants in the study, is
given.
Investigating L2 Phonology
This chapter provides an overview of the major issues and findings in
research on L2 phonology relevant to the focus of this project. First,
theory and research on linguistic constraints, namely LI transfer,
developmental effects, markedness and linguistic environments, is
presented. Research and theory on social barriers, specifically social
identity, gender, extent of LI and L2 use, and attitudes, motivation
and investment, are explored. Additionally, research addressing
linguistic and social as well as task variation is also discussed.
Finally, an overview of first English phonology and then Vietnamese
phonology is provided.
2.1 Theoretical frameworks and research findings
2.LI Linguistic constraints
The discussion of linguistic constraints will first examine the effect of
LI transfer on the acquisition of an L2 phonology. This is followed by
a synthesis of developmental effects for LI child learners as well as
how these effects impact L2 phonological acquisition. Interaction
among LI transfer and developmental effects in L2 phonological
acquisition is then discussed. Markedness is then addressed in a
number of dimensions: preferences for an open syllable structure,
length of margins, type of margins, length and type of margins and
sonority. The effect of the linguistic environment is then examined.
Finally, research that focuses on the acquisition of an English L2
phonology by speakers of Vietnamese is presented.
2.1.1.1 LI transfer
As Leather and James (1991) explain, 'The role of LI in L2 speech
acquisition has formed a major, if not the major, focus of attention
almost as long as second language speech has been studied' (321,
authors' emphasis). Odlin (1989) defines LI transfer as follows:
'Transfer is the influence resulting from the similarities and differ-
2
ences between the target language and any other language that has
been previously (and perhaps imperfectly) acquired' (27). A major
finding in SLA research is that LI transfer is a prominent factor
affecting L2 phonological acquisition and production (Altenberg and
Vago 1987; Benson 1988; Broselow 1987; Flege and Davidian 1984;
Hancin-Bhatt and Bhatt 1997; Hansen 2001, 2004; Hodne 1985;
Major 1987a; Major and Faudree 1996; Odlin 1989; Sato 1984;
Skaer 1984; Tarone 1980, 1987; Vago and Altenberg 1977;
Weinberger 1987). As Skaer (1984) notes, 'the point still remains
that native language phonology does clearly affect target language
acquisition.. .transfer is usually detected through production errors in
speech...' (4). Many researchers state that similarity of sounds is a
criterion for transfer: the substituted sound is often the most
acoustically or articulatorily similar sound to the target language
sound.
In her work with Korean, Cantonese and Brazilian learners of
English, Tarone (1980, 1987) found that the majority of the errors in
word-final consonants produced by the learners could be attributed
to LI transfer effects. Tarone (1980) found evidence that the learners
modified one-fifth or 20 per cent of the final consonants they
produced. The majority (78 per cent) of these modifications could be
attributed to LI transfer. Tarone also found that the learners'
language backgrounds affected whether deletion or epenthesis was
the favoured modification process, with Cantonese and Korean
speakers favouring consonant deletion and Portuguese speakers
favouring epenthesis. As Sato (1984) explains, Tarone's research
provides clear evidence of'the predominance of LI transfer as a force
shaping [interlanguage] syllable structure' (45).
Sato's (1984) study of Vietnamese speakers' English syllable
structure production confirms Tarone's results. Sato found evidence
of LI transfer for syllable structures in Vietnamese interlanguage in
the learners' preference for closed over open syllables (open would
be a syllable ending in a vowel such as to and closed would be a
syllable ending in a consonant or consonant cluster as in hat) since
Vietnamese has closed syllable structures. LI transfer was also
demonstrated in the effect of syllable position on consonant cluster
production in Vietnamese interlanguage: word-final consonant
clusters were produced with less accuracy than word-initial con-
sonant clusters. In her study of task variation and final consonant
and consonant cluster production by a Vietnamese learner of
English, Sato (1985) also found LI transfer effects on production
accuracy.
INVESTIGATING L2 PHONOLOGY 9
In their work on the production of English final stops by speakers of
Spanish, Polish and Mandarin Chinese, Flege and Davidian (1984)
found that the native language had a significant effect on stop
deletion, and that speakers of the languages without final stops, for
example Mandarin Chinese and Spanish, had higher rates of deletion.
Major (1987a), in his study of the production of L2 English final
consonants and consonant clusters by native speakers of Japanese,
found that both positive and negative transfer affected production.
Positive transfer is defined as the facilitating effect on L2 acquisition
due to cross-linguistic similarities while negative transfer refers to
debilitating effects due to a cross-linguistic divergence (Odlin 1989).
Positive transfer was evident in the learners' higher accuracy rates of
voiceless obstruent cluster production while negative transfer was
evident in the greater difficulty the learners had with consonant
clusters containing a liquid over consonant clusters without liquids.
Hodne (1985) collected data from two speakers of Polish learning
English as an L2; 66 different production errors were found in the
666 word-final consonants she collected. Of these errors, 83 per cent
were found to be due to LI transfer. In research on the English
consonant cluster onset production by native speakers of Arabic,
Broselow (1987) also found that transfer played a significant role.
Additionally, she found that certain types of rules are more easily
transferred. In the case of Arabic learners of English, the general rule
of epenthesis was typically transferred from Arabic into the learners'
variety of English, since this rule 'functions to bring phonological
forms into conformity with restrictions on possible phonetic syllable
structures' (Broselow 1987: 303). Broselow added that rules that are
morphologically restricted tend not to transfer. Benson (1988)
examined the speech of two Vietnamese learners of English, and
collected data on 537 closed word-final consonants. Of those, 92
were modified into consonant-vowel (CV) syllables. Almost all of the
modifications (81 out of 92) were due to LI transfer. Osburne
(1996), in her research on the English syllable-final consonant
production of a native speaker of Vietnamese, also found that LI
transfer led to consonant cluster reduction.
In their research on four native speakers of Hungarian learning
English, Vago and Altenberg (1977) found that phonetic transfer
affected the production of the English sounds which did not exist in
Hungarian, with all the learners substituting the Hungarian sound /d/
for the English sound /5/ due to acoustic or articulatory similarity.
Altenberg and Vago (1987) examined the English consonant and
vowel production by two native speakers of Hungarian and found that
both phonetic and phonological transfer affected the participants'
production of English.
Major and Faudree (1996) examined both positive and negative
transfer in their research on the development of voicing contrasts in
L2 English by native speakers of Korean, and found that positive
transfer always applied in the production of voicing contrasts (e.g.
production of voiceless English obstruents in word-initial and final
positions due to their existence in Korean, and production of voiced
obstruents in medial position due to the Korean intervocalic voicing
rule). In contrast, negative transfer only occurred with voiced
obstruents in final position, as predicted, but not with voiceless
obstruents in medial position and voiced obstruents in initial position.
In summary, LI transfer appears to be a prominent factor affecting
L2 phonological acquisition and use (Altenberg and Vago 1987;
Benson 1988; Broselow 1987; Hodne 1985; Major 1987a; Major and
Faudree 1996; Sato 1984, 1985; Tarone 1980, 1987; Vago and
Altenberg 1977). As James (1988) states, 'the syllable structure of the
LI in question would seem to exert a strong influence in any case on
the syllable forms of an IL' (5). LI transfer also may influence
whether deletion or epenthesis is favoured as a modification process
(e.g. Tarone 1980). There also appears to be a constraint on which
LI rules tend to transfer, with those rules that simplify pronunciation
transferring more easily than morphologically restricted rules
(Altenberg and Vago 1987; Broselow 1987; Hammarberg 1997;
Vago and Altenberg 1977).
2.1.1.2 Developmental effects
Developmental effects can be defined as the sequence of and patterns
in the acquisition of a given linguistic phenomenon by children
acquiring a given language as their native language. As Jakobson
(1968) states:
Whether it is a question of French or Scandinavian children, of
English or Slavic, of Indian or German, or of Estonian, Dutch or
Japanese children, every description based on careful observation
repeatedly confirms the striking fact that the relative chronological
order of phonological acquisitions remains everywhere and at all
times the same. (46)
Both cross-linguistically and across children acquiring a particular
language as their LI, as Jakobson (1968) states, there are universal
tendencies in how children acquire their LI. In particular, voiceless
consonants are usually acquired before voiced consonants. Stops are
acquired before nasals, and nasals before fricatives. Fricatives are
usually not acquired before stops are acquired. When first acquiring
fricatives, a child changes the fricative to the corresponding stop, for
example /f/ to /p/ and /s/ to /t/, a process known as 'stopping'. The
acquisition of the back consonants such as the velars and palatovelars
presupposes the acquisition of the front consonants such as labials
and dentals. Additionally, front oral and nasal stops are acquired
before back oral and nasal stops. This is true for fricatives as well:
front fricatives are acquired prior to back fricatives and the acquisition
of back fricatives also presupposes the acquisition of front fricatives.
Therefore, one can say that the acquisition of back consonants
presupposes the acquisition of front consonants. One exception is the
acquisition of the dental fricatives /6/ and /5/; these two consonants
are usually acquired late and are often initially 'stopped' to /t/ and /d/,
respectively.
Initially, children substitute the back nasal consonants with /n/, and
also typically substitute back consonants with their corresponding
dentals (e.g. /t/ for /k/ and /c/), which is called 'fronting.' The
acquisition of/k/ is preceded by mistakes in the use of both /k/ and /t/,
and the often hypercorrection of the two phonemes in favour of /k/.
Therefore, when acquired, /k/ at first merges with /t/ and only later
develops into a separate phoneme. Affricates, also called half-stop
consonants, are acquired only after the acquisition of the corres-
ponding fricative, and before acquiring the affricate, a child may
substitute the sound with either its corresponding fricative or stop.
The latest phonemic acquisitions for the child are those oppositions
that rarely occur in the languages of the world, e.g. nasal vowels or the
second of one of the two liquids (either III or /r/) (Jakobson 1968, 47-
58). In development of English as an LI, children often substitute /w/
for/i/.
1
Other developmental processes include assimilation processes such
as consonant harmony, wherein consonants within the same word are
assimilated. Typically, velars and labials are most resistant to
assimilation while dentals are the most frequently assimilated
consonants (Macken and Ferguson 1981). Other assimilation pro-
cesses include consonant voicing in voiced environments and
devoicing in word-final position, as well as denasalization.
In terms of syllable structures, Ingram (1976) states that the
general tendency for children acquiring syllable structures is the
reduction of all syllables to a CV form via consonant deletion, vowel
epenthesis, unstressed vowel deletion, and reduplication of the first
CV syllable structure. The first strategy children appear to adopt is
the deletion of final consonants, especially in consonant-vowel-
consonant (CVC) syllable structures. Ingram notes that the first
consonants typically acquired in final position are velars. In terms of
the deletion of unstressed syllables, in the initial stages of language
acquisition the first words acquired are monosyllabic, and when
bisyllabic words are used, the unstressed syllable is deleted. When the
child is able to produce two syllables, several deletion processes
occur: initially, there is a reduction of the unstressed syllable (e.g. the
vowel is reduced to a schwa), and secondly, all unstressed syllables
existing in three syllable words are deleted. In the third stage, medial
stressed syllables may be produced but there is a continuation of the
deletion of unstressed initial syllables. In the fourth stage, unstressed
syllables may be produced in bisyllabic words, although not as
frequently in trisyllabic words. In stage five, the child approximates
correct adult pronunciation (Ingram 1976: 30-1).
The reduction of consonant clusters is another process that has
several stages and can continue for a longer period of time.
Consonant cluster reduction via consonant deletion is not random.
The first stage is marked by the deletion of the entire cluster, followed
by the reduction of the cluster to one consonant in the second stage.
Typically, the marked member of the cluster is deleted (in clusters
with /s/ plus a stop, the /s/ is deleted; in stop + liquid, the liquid is
deleted; in fricative + glide/liquid, the glide/liquid is deleted; and in
nasal + obstruent, the nasal is deleted) (Ingram 1976: 32-3). In stage
three, the cluster is produced, with the substitution of one of the
elements, a liquid to a glide, for example. Stage four marks the child's
correct pronunciation of clusters (Ingram 1976: 31-3). It is also
important to note that native speakers typically delete the medial
consonant in a final three-consonant cluster, usually a /t/ or a /d/
(Hieke 1987). In two-consonant final clusters, stops may be deleted,
especially in nasal-stop sequences.
Research on L2 phonological acquisition has examined the extent
to which L2 developmental processes (the sequences and modifica-
tions an L2 speaker makes in acquiring the L2) mirrors those for child
learners of the same LI. For example, Piper (1984) researched
whether LI and L2 developmental effects would be similar by
examining the consonant production of fifteen 5 year-old children of
various Lls (Portuguese, Punjabi, Mandarin Chinese, Vietnamese,
Serbo-Croatian and Italian) learning ESL. She found that the LI
developmental effects of substitution (stopping, fronting and gliding),
syllable structure mismatches (final consonant deletion, vowel and
consonant epenthesis and cluster reduction) and assimilation (voic-
ing, devoicing and consonant harmony) accounted for 86 per cent of
the participants' consonant production errors. However, some
discrepancies existed between LI and L2 developmental effects as
indicated by this study. Consonant harmony, the most common
assimilation effect in LI acquisition, was rare in the English L2
production in Piper's study. Additionally, LI learners commonly
reduce consonant clusters, but the L2 learners in Piper's study
deleted final consonants more often than they reduced consonant
clusters. Piper also did not examine the effect the learners' LI had on
the learners' production of consonants and consonant clusters, which
would have shed more insight into the different strategies learners
used in producing English sounds. Piper also found initial evidence of
a developmental stage in L2 phonological acquisition. Some of the
participants had a U-shaped curve of production, beginning with
imitation with few pronunciation errors, then deviation from the
norm as the learners began acquiring the feature, and finally a period
of resolution where production gradually became more target-like,
which correspond to findings on developmental sequences in
morphology (see Ellis 1994: 77).
Hieke (1987) researched consonant cluster reduction in both
native and non-native speech to compare rates of reduction, since
native speakers typically reduce final three-consonant clusters. One
limitation of this study is that it is not clear what the LI backgrounds
of the non-native speakers of English were, and what the syllable
position was of the consonant clusters Hieke researched. However,
Hieke found that native speakers reduced clusters approximately
twice as often as non-native speakers. This research not only
illustrates that consonant deletion is variable for both native and
non-native speakers, but also that cluster reduction is not just a
developmental process, but also a speech phenomenon that is shared
cross-linguistically.
In summary, one of the main findings in this line of research is that
the LI developmental process of word-final obstruent devoicing is
also common for L2 learners of English whose first languages do not
have word-final voiced (or voiceless) obstruents, such as speakers of
Mandarin Chinese (Flege and Davidian 1984), and Icelandic (Hecht
and Mulford 1982; Mulford and Hecht 1980).
2.1.1.3 Interaction between transfer and developmental effects
It has been found that transfer effects interact with developmental
effects in L2 phonological acquisition (Flege and Davidian 1984;
Hancin-Bhatt and Bhatt 1997; Hecht and Mulford 1982; Major
1987c; Mulford and Hecht 1980). In his work on Japanese native
speakers' production of English consonant clusters and final conson-
ants. Major (1987c) found that both transfer and developmental
effects applied to devoicing. Flege and Davidian (1984) also found
that both LI transfer and developmental effects influenced Spanish,
Chinese and Polish adult learners' production of English word-final
stops in CVC words. LI transfer effects influenced the deletion of
final stops, while both developmental and LI transfer effects were
found to affect learners' devoicing of final stops. Overall, there was a
higher frequency of devoicing than deletion for final stops.
Mulford and Hecht (1980) and Hecht and Mulford (1982)
researched a 6-year old Icelandic child's acquisition of English and
found that while LI transfer effects could predict which sounds may
be more difficult to acquire, developmental effects were better
predictors of what substitutions were adopted for L2 production. For
example, the participant devoiced final stops, an English LI devel-
opmental effect, and then strongly aspirated them, which is charac-
teristic of stops in Icelandic. Hancin-Bhatt and Bhatt (1997), in their
research on the production of English onsets and codas in monosyl-
labic words by native speakers of Japanese and Spanish, found that LI
transfer had a significant effect on L2 syllable structure production,
affecting both the error rates and types of errors made. They found
that positive transfer effects can override developmental effects, as
evidenced by Spanish speakers' production of specific L2 syllable
onsets similar to onsets in their LI. In the absence of positive transfer,
however, developmental effects were evidenced to have an effect on
production.
It has also been argued that LI transfer is more prominent in the
early stages of acquisition. Leather and James (1991) state that:
it would appear that, as acquisition proceeds, the influence of the
LI and the mechanism of transfer give way gradually to other
influences that shape developing L2 speech such as the mechan-
isms (i.e. "developmental processes") associated with the acqui-
sition of the mother tongue, (original parenthetical aside, 326)
Major (1987a) developed the Ontogeny Model2 to explain how
developmental and transfer effects interact in L2 phonological
acquisition. Transfer effects predominate in the early stages of L2
phonological acquisition, and then gradually decrease, while devel-
opmental effects increase in the middle stages of acquisition, and then
gradually decrease. In terms of phonological similarity of the LI and
the L2, Major postulates that for similar phenomena, transfer effects
will be more common but for phenomena that are dissimilar,
developmental effects may play a greater role in acquisition and
production. Major further states that production in different speaking
styles is affected differently by transfer from the LI and develop-
mental constraints: as the formality of the style increases, there is a
decrease in errors due to transfer, whereas errors due to develop-
mental effects increase and then decrease.
Major (1987a) tested his model by examining data from beginner
and advanced Brazilian Portuguese speakers' acquisition of final
consonants across three tasks designed to elicit different styles (in this
case, different levels of formality). Results indicate a trend, which was
not statistically significant, of advanced learners having more
instances of errors due to developmental effects and beginner learners
having more instances of errors due to transfer effects. Major's claims
regarding stylistic variation also received some support, but again this
support is best labelled as a trend as it was not statistically significant.
Major (1994) also tested the Ontogeny Model in his study of native
Portuguese speaking Brazilians' production of English L2 double
consonant onsets and codas in monosyllabic words over a four-week
period. His findings provided some support for the model: transfer
effects decreased over time, and a higher frequency of correct
production was achieved. However, developmental effects remained
stable instead of increasing when transfer effects decreased, and style
has no effect. As the duration of this study was only four weeks, it is
difficult to determine whether the data is truly able to show change
between the effects of transfer and developmental constraints.
In summary, research (e.g. Flege and Davidian 1984; Hancin-
Bhatt and Bhatt 1997; Hecht and Mulford 1980; Major 1987c;
Mulford and Hecht 1982) indicates that transfer and developmental
effects may interact in L2 acquisition, with LI transfer dominant in
the early stages of acquisition and developmental effects increasing as
LI transfer effects decrease. Both may also affect the production and
acquisition of a single segment; it has also been found that while LI
transfer may affect which sounds are difficult, developmental effects
may affect substitutions. Finally, it is possible that positive LI transfer
effects may override developmental effects.
2.1.1.4 Markedness
Markedness is a linguistic concept that can roughly be defined as the
tendency for certain linguistic elements to be more frequent in the
world's languages, based on issues such as naturalness, salience and
ease of articulation in terms of phonological features. SLA researchers
have employed several definitions of markedness in order to explain
the acquisition of phonological segments. Researchers have examined
the influence of markedness based on: a) preference for a universal
syllable structure; b) length of margin; c) type of margin; d) length
and type of margin; and e) the sonority hierarchy within a margin.
Each of these will be examined in turn.
a. A preference for a universal syllable structure
One area of markedness research is the universal preference for a CV
syllable structure, which is considered to be unmarked in relation to
other syllable structures such as CVC or CVCC, etc. Prior research
(Benson 1988; Osburne 1996; Sato 1984, 1985) on Vietnamese
learners of English has found that LI transfer effects exerted a greater
influence on the learners' production of English syllable structures
than a preference for an open syllable; that is, codas allowable in
Vietnamese were transferred into English, resulting in a closed
syllable structure.
b. Markedness defined by length
Another way researchers have employed markedness in SLA research
is to define markedness in onsets/codas by length. All languages have
a core consonant-vowel (CV) syllable structure (James 1988: 5;
Spencer 1996: 82). CV syllables are thus the least marked syllable
structure in languages, with increasing length increasing the marked-
ness of the syllable structure (Carlisle 1997). Although the CVC
syllable does not occur in all of the world's languages, it does occur
frequently and is therefore also considered universally unmarked
(Spencer 1996: 82). Additionally, the existence of codas or onsets
with a length of A implies the presence of codas/onsets with a length
of A-l. For example, if a language allowed a syllable with two
consonants in the onset, such as CCV (e.g. the word tree)., it would
also have to allow the onset as CC - 1 or as C (as in tee). One
exception to this rule is the CV syllable, which does not always imply
the presence of V (Greenberg 1978).
Research on L2 production has found that learners modify longer,
and thus marked, structures in favour of shorter, less marked,
structures by reducing the length of the margins (cf. Anderson, 1987;
Broselow and Finer 1991; Carlisle 1991, 1997, 1998; Eckman 1987,
1991; Hansen 2001; Sato 1984; Weinberger 1987). Weinberger
(1987), in his examination of four Chinese learners of English's
production of word-final codas, found that the number of modifica-
tions increased as the length of the codas increased. This indicates
that markedness had an effect on the number of modifications.
Anderson's (1987) research on word-final consonants and consonant
clusters also found that speakers of Egyptian Arabic, and Mandarin
and Amoy Chinese, made significantly more modifications on
margins based on the length - the more complex the consonant
clusters were, the more modifications were made. All the participants
performed better on unmarked than marked structures. Sato's (1984)
research on Vietnamese learners of English found that participants
reduced two-member onsets in favour of one-member onsets.
Eckman's (1987) research on native speakers of Korean, Japanese
and Cantonese's production of two- and three-member word-final
consonant clusters also corroborates both Weinberger's (1987) and
Anderson's (1987) findings that longer codas result in more frequent
modifications, and that longer codas (e.g. two- and three-member
codas), which are relatively marked, are reduced by one consonant in
favour of unmarked codas. Major (1987c) also found that his
Japanese participants produced the English less marked word-final
obstruents more accurately than the more marked consonant clusters,
and the less marked word-final voiceless obstruents more accurately
than the more marked voiced obstruents. Carlisle (1997) also found
that shorter onsets were not modified as frequently as more marked
onsets.
c. Markedness defined by type of margin
Another area of investigation in terms of markedness is whether
markedness relations between onsets/codas of the same length but
different constituents have an effect on acquisition. As Greenberg
(1978) states, two-member clusters comprised of two consonants
from a similar category, for example stop-stop or fricative-fricative,
are more marked in relation to consonants from different categories,
such as stop-fricative or fricative-stop, which are considered
unmarked in comparison. The following implicational hierarchy
exists: stop + stop (/pt/ as in rapt) implies fricative + stop (/st/fast) and
stop + fricative (/ts/ as in hats') and is therefore more marked; fricative
+ fricative (/fs/ as in hoofs') implies fricative + stop and stop + fricative
and is therefore more marked; stop-fricative is preferred in onsets and
fricative-stop in codas (both due to sonority, to be explained below);
word-final voiced obstruents imply word-medial obstruents, which in
turn imply word-initial voiced obstruents (see Greenberg, Ferguson
and Moravcsik 1978), resulting in the following order of syllable
position acquisition of consonants from most to least marked: final >
medial > initial. In research on native speakers of Vietnamese
acquiring English, Hansen (2004) found that learners had lower
accuracy ratings across time on more marked stop + stop and fricative
+ fricative clusters, than on stop + fricative and fricative + stop codas.
d. Markedness defined by length and type of margin
Research in this area focuses on markedness relationships between
clusters of different lengths based on the elements comprising the
codas: for example, a markedness relationship does not exist between
a three-member and a two-member cluster based on length alone -
the two elements of the two-member coda must also comprise part of
the three-member coda. Thus, /ill would be unmarked in relation to
/ild/ whereas An/ would not. Hansen (2004) examined acquisition of
codas by length and type, and found that the three member codas
learners had most difficulty with were liquid-fricative-fricative, stop-
fricative-fricative, nasal-fricative-fricative, nasal-stop-stop and stop-
stop-fricative, all comprising the more marked structures.
Eckman (1991) examined the acquisition (defined as correct usage
80 per cent of the time) of two- and three-member onsets and codas
by native speakers of Japanese, Cantonese and Korean, as well as how
markedness relationships influenced acquisition. Eckman found that
for almost all the cases (98 per cent), if the more marked structure
(e.g. three-member codas) were acquired by the learner, the
subsequent less marked structure (i.e. a related two-member coda)
had already been acquired. Carlisle (1998), like Eckman (1991) and
Eckman and Iverson (1993), also used a criterion measure of 80 per
cent accuracy in production to examine the acquisition of English
syllable onsets in a markedness relationship (based on length of the
onsets) by native speakers of Spanish. His findings support Eckman's
in that in almost all of the cases (90 per cent) the more marked
structure had only been acquired after the corresponding unmarked
structure had been acquired.
e. Markedness defined by sonority
Another definition of markedness is based on the sonority hierarchy
within an onset or coda. Broselow and Finer (1991), for example,
examined markedness in terms of sonority by focusing on the distance
between the members of the consonant cluster. Sonority is denned as
the loudness and/or resonance of a given segment in relation to other
segments. The researchers posit that clusters closer in sonority are
more marked than clusters wherein consonants have a wider sonority
distance. The sonority hierarchy, from the most sonorous to the least,
is as follows: vowels > glides > laterals > nasals > fricatives > stops.
Their findings on Japanese and Korean learners of English's word-
initial consonant cluster production supported their hypothesis as the
participants in their study produced less marked onsets more
accurately than the more marked onsets.
Another approach to sonority is research that examines the effect of
the Universal Canonical Syllable Structure (UCSS) on the acquisition
of L2 onsets and codas. Margins meet UCSS 'if the sonority of
segments.. .increase[s] as the syllable nucleus is approached' (Herbert
1986: 62). Research by Tropf (1987) and Carlisle (1991) has found
that learners less frequently modify those margins that conform to
UCSS. Hansen (2004) found that the CCC codas learners had the
least difficulty with did not violate UCSS and that the CCC codas
that emerged first met UCSS; conversely, codas that did violate
UCSS were often modified to conform to UCSS via absence, feature
change (liquid-stop-fricative and nasal-stop-fricative were changed to
liquid-fricative and nasal-fricative, respectively). Hansen (2001)
found that for the native speakers of Mandarin Chinese learning
English in this study, 73 per cent of the codas modified violated
UCSS and that all the CCC codas that violated UCSS and were
modified were changed into CC codas that met the UCSS.
In summary, research has consistently found that L2 learners
modify marked syllable structures more frequently than they modify
unmarked syllable structures as based on length (Anderson 1987;
Carlisle 1997, 1998; Eckman 1987, 1991; Eckman and Iverson 1993;
Hancin-Bhatt and Bhatt 1997; Hansen 2001, 2004; Sato 1984;
Weinberger 1987) and sonority (Broselow and Finer 1991; Hansen,
2001, 2004). In addition, longer, and thus more complex syllable
structures are modified in favour of shorter, or less marked, syllable
structures and acquired after the acquisition of a related two-member
coda (Carlisle 1998; Eckman 1991; Eckman and Iverson 1993;
Hansen 2001; 2004).
2.1.1.5 Linguistic environment
Very few L2 phonology researchers have examined the effect of
linguistic environment on L2 consonant and consonant cluster
production. However, the research that has been done indicates
that linguistic environment has a significant effect on consonant
production, and may interact with other linguistic factors such as LI
transfer. Benson (1988), for example, in her research on Vietnamese
speakers' production of English syllable structures, found that the
preceding vocalic context was a factor in LI transfer processes, for
example, single consonants were always deleted after diphthongs, and
in Vietnamese, most diphthongs occur only in CV structures.
Weinberger (1987) found that Mandarin speakers' modifications of
English consonants and consonant clusters occurred most frequently
between consonants. Anderson (1987) found that the American
English III, one of the most commonly deleted consonants by
speakers of Mandarin, was most commonly deleted post-vocalically in
both final clusters and in word-medial sequences, while III and /d/
were most often deleted when in word-final position in consonant
clusters. Carlisle (1997) found that linguistic environment influenced
the type of modification of final consonants and consonant clusters
that took place, e.g. epenthesis was found to occur more frequently
after consonantal than vocalic environments. Edge (1991) compared
non-native speakers' production of final obstruents with the produc-
tion of native speakers since native speakers of English often devoice
final obstruents before pauses or before a voiceless sound. Portions of
her data revealed similar processes of vowel epenthesis by both native
speakers and Japanese speakers of English: after word-final voiced
stops. Carlisle (1994) re-examined Tarone's (1980) data and found
that over 50 per cent of the instances of epenthesis after word-final
consonants occurred before a pause, 40 per cent before a word-initial
consonant, and less than 5 per cent occurred before a word-initial
vowel. In his examination of -t/d production by Chinese learners of
English, Bayley (1996) found that -t/d was less likely to be deleted
following a liquid, over a nasal or obstruent. In terms of following
segments, the order of constraints favouring deletion, in higher to
lower rates, are: consonant > glide > vowel.
Additionally, grammatical conditioning may also exert a significant
effect on syllable coda production. In research on final /ps ts ks/
clusters, Saunders (1987) found that /s/ was more likely to be retained
in plural over third-person singular /s/, a finding corroborated by
Abrahamsson (2001). Osburne (1996) found that a grammatical
marker in CC and CCC codas was likely to be retained, even if the
coda violated the UCSS; in fact, codas that violated the UCSS were
more likely to be produced correctly by her participant due to
grammatical conditioning as the last member of this type of cluster
typically is a morphological marker.
These findings indicate that although the effect of linguistic
environment has not received a great deal of attention in L2
phonological research, findings from research in this area can
contribute significantly to our understanding of why some consonants
and consonant clusters are modified, and how they are modified.
Therefore, it is an area of L2 phonology that deserves more research.
2.1.1.6 The acquisition of English phonology by speakers of
Vietnamese
There have been several previous studies on Vietnamese learners'
production of syllable onsets and codas. Sato (1984), in a longitu-
dinal study of two Vietnamese adolescents, for example, examined
whether closed syllables would be preferred in the English inter-
language of Vietnamese learners, basing her hypothesis on the fact
that 77 per cent of Vietnamese syllables (3437 out of 4467) are
closed. Additionally, she hypothesized that syllable-initial clusters
would be produced more correctly than syllable-final clusters as
clusters are not allowed in final position in Vietnamese. Her results
confirmed both hypotheses: overall, syllable-initial consonant clusters
were produced more accurately than syllable-final consonant clusters,
and there also appeared to be a preference for a CVC syllable
structure over the CV structure due to LI transfer. The preferred
modification strategy for the production of clusters was cluster
reduction (one member was omitted), and this modification strategy
was favoured over deletion, epenthesis and feature change, and often
resulted in a closed syllable. In her study of task variation and final
consonant and consonant cluster production by a Vietnamese learner
of English, Sato (1985) also found LI transfer affected production
accuracy.
Benson (1988) also examined the universal preference for an open
syllable structure by Vietnamese speakers. Her data was elicited from
two native speakers of Vietnamese, and confined to monosyllabic
words ending in voiceless plosives and nasals, as these are the only
consonants allowed in syllable-final position in Vietnamese. She
found that the universal preference for an open-syllable structure had
a lesser effect on production, while native language influence exerted
a greater effect. For example, Benson found that native language
influence in terms of the effect of a previous vowel affected the
deletion of the final consonant: final consonants were deleted after
diphthongs and in Vietnamese, diphthongs only exist in open
syllables. Furthermore, Benson questioned Sato's (1984) conclusions
that learners' reductions of clusters by one member is due to LI
transfer exclusively, since it also is evidence of a universal preference
for an open syllable.
Osburne's (1996) research confirmed previous studies, as she also
found that for her Vietnamese speaker (there was only one participant
in this study), single syllable-final consonants were mostly produced
accurately, but syllable-final consonant clusters were often reduced,
with fricatives optionally deleted due to LI transfer effects (fricatives
are not allowed in syllable-final position in Vietnamese). Osburne's
research also confirms Benson's (1988) findings that native language
influences on syllable structure production in terms of the effect of the
preceding diphthong on consonant deletion played a major role in
English syllable production by speakers of Vietnamese. Osburne
provides insight into this process by explaining that in Vietnamese, a
consonant-diphthong syllable would be considered closed, since the
diphthong is realized as a vowel + glide, rather than a single segment
as it is in English. Therefore, this syllable is closed, and any following
consonants would violate Vietnamese syllable structure, and thus be
optimally deleted. Additionally, Osburne found that for her subject,
/i/ was never realized in English syllable codas. Either the segment
was absent or it was co-articulated with the vowel, and thus
considered part of the nucleus rather than the terminus. This
indicated that the learner was assigning 111 to the nucleus of the
syllable due to his perceptions of the realization of English 111 (as
Osburne states, there is some speculation that a postvocalic 111 should
be considered to be a glide). In fact, when 111 was the initial member
of a consonant cluster, it was not deleted even though the following
consonant was at times deleted since lit was considered part of the
nucleus by the Vietnamese speaker.
Osburne (1996) also provides other evidence of the effect of
linguistic environment on cluster reduction: clusters were reduced
before pauses 80 per cent of the time. In addition, sonority was
examined, and it was found that clusters that violated the sonority
hierarchy were significantly less likely to be reduced. This could be
due to the fact that in English, many final clusters violate the sonority
hierarchy due to inflectional endings, and it could be that these types
of clusters are more salient to the learners, and thus less likely to be
reduced. This confirms research by Young (1988) on English tense
marking by native speakers of English - redundancy in marking which
make the past tense ending more salient were found to increase the
likelihood of the past tense endings being produced. Finally, Osburne
also found that the final cluster /ks/, when representing the letter x in
orthography, was not reduced, possibly due to its orthographic
salience.
Overall, the research on Vietnamese learners of English indicates
that LI transfer plays a significant role in syllable structure produc-
tion, and that linguistic environment, especially preceding vocalic
environment, plays a great role in consonant deletion, providing
additional evidence for the primacy of LI transfer in the production
of syllable codas for native speakers of Vietnamese (cf. Benson 1988;
Osburne 1996; Sato 1984, 1985). Additionally., salience of past tense
marking and orthography may also be a factor in terms of which
consonant clusters are typically not reduced (Osburne 1996).
2.1.2 Social barriers
The discussion of social barriers in L2 acquisition, and specifically the
acquisition of an L2 phonology, starts with a discussion of social
identity. Gender, as one aspect of identity, is then addressed. The
social context of LI and L2 use, in the form of extent of usage of the
LI and the L2, is then explored. Finally, affective variables, namely
attitudes, motivation and investment, are explored.
2.1.2.1 Social identity
Social identity is comprised of different components such as
appearance, ethnicity, group memberships, language, culture, etc.
that play out in different ways in different contexts with different
interlocutors. Language is one medium through which we express our
social identity, but identity construction is not unidirectional - our
interlocutor(s) must recognize our constructed identity (our language
markers) as social identity must be constructed in relation to other
individuals. Lum (1982) states it concisely: 'identity is a social
process in which one balances what s/he thinks oneself to be and what
others believe that one has to be ...' (386).
Language and identity are extricably intertwined. Giles and Byrne
(1982) state that 'Accent serves as a powerful symbol of ethnicity and
'psychological distinctiveness'" (as quoted by Leather and James
1991: 310). In fact, linguistic behavior is 'a series of "acts of identity"
in which people reveal both their personal identity and their search for
social roles' (LePage and Tabouret-Keller 1985: 14). These acts can
be both conscious and unconscious, and resisting specific language
features is as important in marking one's social identity as is the use of
specific linguistic features, e.g. phonological markers, within one's
linguistic repertoire. As Pennington (1994) states, 'A pronunciation
of a certain sound that signals membership in a certain group is
termed a phonological marker of identity in that group' (author's
emphasis, 103). Research in both LI and L2 use have shown us that
'Speakers adopt various linguistic markers to identify themselves with
particular sociocultural groups' (Leather and James, 1991: 310). In a
summary of research on L2 social markings via phonology, Dowd,
Zuengler and Berkowitz (1990) state that social markings can occur
at any age group and level of L2 acquisition, despite a limited
repertoire. Additionally, 'several sounds marking the same social
factor may sift in different directions; a given sound, marking several
social factors, may shift in different directions' (Dowd, Zuengler and
Berkowitz 1990: 22).
One area of SLA in which we need a greater awareness of social
identity is learners' acquisition (or lack thereof) and use of specific
phonological features. Many SLA researchers and teachers compare
learners' language to the standard variety, assuming both that
learners are aiming at the standard as their target language and that
any deviations from the standard variety on the part of the learner
are 'errors.' Beebe (1985), however, suggests that learners' choices
of target language variety depend on the social context, and that
there are certain input preferences for language learning, for
example peers over parents and teachers, in-group over out-group,
friends over non-friends, high-contact over low-contact groups, and
higher-prestige over lower-prestige groups, and that learners decide
whether to give precedence to solidarity (for example for the peer
group,) or prestige (teachers'/parents' variety). However, as Hieke
(1987) states, 'the systematic variation of spoken English from ideal
forms (i.e., underlying representations) has not usually been
considered in IL phonology studies, nor have the wide range of
variations attributable to dialect within so-called Standard English'
(391).
Edge (1991), in her study of Cantonese speakers' production of
English obstruents states that the reason why some Cantonese
learners exhibit non-target variants in their obstruent production may
be because they are targeting a nativized variety of English, namely
Hong Kong English, which exhibits these variants, and therefore
analysing this data for interlanguage (IL) rules and processes is, as she
states, 'a questionably valid procedure' (391). She further states that,
'for many phonetic features, nativized varieties and other non-
standard Englishes offer different targets than those of what has been
called Standard English' (391).
Similarly, Hecht and Mulford (1982), in their research on an
Icelandic child's acquisition of English fricatives and affricates, found
that the learner's tendency to devoice final /z/ was partially affected by
the variety of English he was learning from his peers. Other L2
patterns in the learner's speech were also found to be influenced by
the typical Boston pronunciations the learner would have been
exposed to through school peers.
Retention of certain LI features and use of particular L2
variants may also be conscious choices of the L2 learners in order
to signal their identity. Gatbonton (1975) for example, found that
the production of interdental fricatives by French-Canadian learn-
ers of English varied based on the learners' self-identification as
either nationalistic and pro-French or non-nationalistic and pro-
English, with the latter employing greater English L2 dental
fricative usage.
The learners' level of integration and acculturation into the L2
culture may also affect the degree to which certain L2 variants are
produced, as Lybeck (2002) found in her research on the acquisition
of Norwegian by American women living in Norway. The women
who had the densest L2 social networks and were the most integrated
into the L2 community had the greatest use of the Norwegian /r/;
conversely, the women who had little contact with Norwegians and
had a strong American identity had a greater use of American III than
Norwegian /r/.
2.1.2.2 Gender
Social identity is also tightly connected with gender and gender
identity. Gender can be denned as 'social practice' (Ehrlich, 1997),
and be located within a social, historical and cultural space that is
dynamic across time and space. In this view, gender is 'a system of
culturally constructed relations of power, produced and reproduced
in interaction between and among men and women' (Gal 1991: 176).
One strand of research in this area has focused on how gender
constrains the level of access L2 learners have to linguistic resources,
with the majority of the studies examining how women are denied
access to linguistic resources, both at school and at work (cf.
Blackledge 2001; Gumming and Gill 1992; Ehrlich 2001; Goldstein
1995, 2001; Kouritzen 2000; Losey 1995; Teutsch-Dwyer 2001).
Findings indicate that both the first language (LI) and L2 community
may constrain the language development of immigrant women. In
many cases, women may be denied access to English as a second
language (ESL) classes or workplaces where the L2 could be
practiced due to LI cultural norms (cf. Gumming and Gill 1992;
Goldstein 1995, 2001; Kouritzen 2000). Additionally, when they do
attend the courses, they may not get as much attention in the
classroom as men (Losey 1995).
While research has begun focusing on how these actions, activities,
and behaviours are encouraged differently for women and men, and
the resulting access to L2 use and development opportunities, there
has been little research to date (cf. Teutsch-Dwyer 2001), that
directly examines the impact of the differential access to L2 use and
development opportunities on the acquisition of an L2 although recent
research in the acquisition of an L2 phonology has begun suggesting
that the degree of accent in the L2 may be influenced by extent of LI
use - the greater the LI use, the greater the degree of foreign accent
(cf. Flege, Frieda and Nozawa 1997; Piske, MacKay and Flege
2001).
Learners may also resist acquiring, and using, specific linguistic
forms because it conflicts with their LI identities, particularly gender
roles in the LI culture and the L2 culture. This is important for SLA
researchers and teachers to understand because students' lack of use
of certain forms may not always indicate a lack of acquisition of (or
ability to acquire) these forms, but rather acts of resistance to the
identity the learners perceive the forms portray. As Pennington
(1994) states:
.. .the identity signaled by a certain accent sometimes serves as a
psychosocial constraint pressuring the normative not to speak in a
nativelike way in order to avoid any identification with native
speakers - including symbolic identification with their symbolic
values. (103)
This phenomenon is well illustrated by research on the learning of
Japanese by Western women. One such study is Siegal's (1996) study
of white women, all native speakers of English, studying Japanese in
Japan. Although this is focusing on syntax, the study provides
interesting examples of the strategies several white women adopt in
resisting the L2 identity the Japanese language imposed upon them,
and how these women create a new L2 identity for themselves by
adopting aspects of both men's and women's speech. The three
women reported on in this study avoided the use of certain linguistic
features associated with women's language in Japanese, such as
honorifics (used by men as well, but to a lesser extent), and women's
sentential-final pragmatic particles, preferring instead to develop an
L2 linguistic identity by combining elements of both 'men's' and
'women's' speech rather than the invisible and passive identity they
associated with Japanese women's language. For example, one of the
women's language styles was marked with the overuse of the plain
form, which is associated with men's speech. She also avoided the use
of honorific language and used the polite statement 'I think' in place
of honorifics. Another woman also avoided the use of honorifics,
instead using the modal auxiliary 'desho' to soften her utterances in a
more feminine manner. A third woman, who inaccurately believed
that sentential-final particles functioned only in women's language.
avoided using pragmatic particles altogether. All three women were
conscious of their language choices, but avoided using 'correct'
Japanese forms in situations when they knew they should, because it
conflicted with their LI identity.
Ohara's (2001) work on the acquisition of intonation by American
women learning Japanese also illustrates how learners may resist
using certain L2 features if they perceive it conflicts with their LI
identity, in this case, also a gender-based identity. In Japanese,
femininity is expressed by women through the use of a high-pitched
voice and '.. .the use of a high pitched voice is an important way of
performing or "doing" gender' (234). Ohara examined to what extent
beginner learners of Japanese, Japanese (LI) - English (L2)
bilinguals, and English (LI) - Japanese (L2) bilinguals were aware
of these norms and willing to employ them to perform Japanese
gender. She found that while the beginner learners were unaware of
the norms, the Japanese (LI) - English (L2) bilinguals were aware of
the norms and employed them to perform gender. However, the
English (LI) -Japanese (L2) bilinguals were aware of the norms but
several (2 out of 5) of them were unwilling to employ a high pitch
because they perceived 'that the voice pitch levels they employed
correlated neatly with their attitude toward the kinds of images
typically associated with Japanese women' (Ohara 2001: 242) and an
identity they did not accept.
Finally, gender has also been linked to the variety targeted.
Adamson and Regan (1991) investigated the acquisition of commu-
nity speech norms by Vietnamese and Cambodian immigrants to the
US. The researchers examined the participants' use of the phono-
logical variable 'ing', specifically focusing on two variants, /irj/ and
/in/, hypothesizing that greater use of /in/, the non-prestige variant
and a variant not present in the participants' LI phonology (versus
/ i r j / , w h i c h i s p r e s e n t ) , w o u l d i n d i c a t e a g r e a t e r i n t e g r a t i n g o f
participants into their respective speech communities. The partici-
pants were all from either Philadelphia or Washington DC, where /in/
functioned as a non-prestige variant of '-ing'.
The researchers found men used /in/ more frequently than women,
a pattern that was similar to that found for native speakers in these
particular contexts. The researchers state that these findings suggest
that the participants are targeting different groups of speakers, i.e. the
non-native women are targeting the language of native speaking
women, and vice versa, and that men, native and non-native speakers
alike, use the nonstandard variant more frequently than either native
or non-native women.
2.1.2.3 Extent ofLl/L2use
Social interactions are also an important area of investigation within
SLA although it has as yet not received significant attention from SLA
researchers. Social interactions can be investigated through the
framework of a learner's amount of contact with speakers of both
the Ll and L2, as well as the type of contact that takes place and with
whom, in order to better understand how the context of language use
affects L2 acquisition as well as Ll maintenance. Within these
frameworks, social interactions are intricately connected with social
identity, since the person one chooses to associate with typically is a
member of the group with which one identifies, or feels solidarity
with. As illustrated in the above discussion of social identity, social
interactions can affect which language variety is targeted, and so have
a direct effect on both L2 acquisition and use. However, while most
SLA researchers acknowledge that L2 acquisition cannot take place
without learners' engagement in meaningful communication, there
are few studies that investigate the effect of learners' amount and type
of contact with speakers of the L2 on the learners' L2 acquisition. Of
these studies, even fewer focus on phonology, the majority of the
research centering on the acquisition of L2 syntax.
Oyama's (1976) research is one of the few studies that address
amount of language use and L2 phonological development. Oyama
investigated the effect of length of stay and age of arrival on the degree
of accentedness in English of 60 Italian-born immigrants to the US.
While age of arrival was a significant variable in degree of
accentedness, length of stay had no effect. This led Oyama to state
that, 'Considering the width of the number of years in the United
States intervals, the lack of effect of this variable in the analysis of
variance is striking. Amount of use of a language must obviously be an
important consideration,' (270, my emphasis). Tahta, Wood and
Loewenthal (1981) researched the transfer of accent from Ll to L2,
and found that after age of L2 acquisition, the top predictor of non-
accentedness was use of English at home and exposure to English
models outside of school.
In his research on the /t d/ production of Mandarin speakers,
Bayley (1996) included social interactions in the VARBRUL analysis
(see Chapter 3 for further explanation) of factors influencing
production and absence of /t d/. Participants were categorized into
two types of social networks: a) mixed Chinese and American, and b)
exclusively Chinese. Bayley found that social networks had a
significant effect on /t d/ absence, with speakers classified as having
an exclusively Chinese social interaction having a slight probability
for It d/ deletion, while for participants with mixed American and
Chinese social interaction the probability for It d/ deletion was
significantly higher.
There has also been a great deal of research on this area by Flege
and colleagues (e.g. Flege, Frieda and Nozawa 1997; Guion, Flege
and Loftin 2000; Piske and MacKay 1999; Piske, MacKay and Flege
2001), particularly on the effects of LI use, rather than L2 use, on
accent in the L2. Flege, Frieda and Nozawa (1997), in work on the
English of Italian immigrants to Canada, found that LI use affected
L2 accent: those who had a higher degree of LI use were rated as
having a stronger LI accent. These findings were confirmed by Piske
and MacKay (1999); Piske, MacKay and Flege (2001); and Guion,
Flege and Loftin (2000).
SLA researchers have also found that some aspects of social
interactions can limit or give learners opportunities to practice the
target language. Monshi-Tousi, Hosseine-Fatemi and Oiler (1980),
in their research on the relationship between English proficiency and
use of English outside the classroom, found that the number of
American friends the learners had was one of the best predictors of L2
proficiency. Peirce (1995), in her study of the English language
acquisition patterns and L2 social identity development of immigrant
women in Canada, found that the L2 (in this case Canadian)
community limited the interactions the L2 learners had with
members of that community, and that 'power relations play a crucial
role in social interactions between language learners and target
language speakers' (3). Goldstein (1995), in her study of Portuguese
women immigrant factory workers in Canada, found that it was the
LI community that limited the opportunities that L2 learners had to
interact with speakers of the L2. Use of the L2 was found to be
alienating and risk-inducing since it separated the women from the
social interactions and the 'sisterhood', which helped them to get and
maintain their jobs in the factory.
2.1.2.3 Attitudes, motivation, and investment
While the research on attitudes towards the target culture and L2
achievement as well as motivation in L2 language learning and L2
proficiency has been questioned on the basis of methodological
shortcomings (the majority of the research has been correlational
which can not establish cause-and-effect relationships) the overall
findings indicate both attitudes and motivation may have an indirect
effect on language learning.
In terms of attitudes, research has found that positive attitudes
towards the target language groups can enhance learning while
negative attitudes may impede learning (Gardner and Lambert 1972).
In some cases, however, negative attitudes can have a positive effect
on learning if the learners have strong reasons for learning (Lanoue
1991; Oiler, Baca and Vigil 1977). It is important to note that the
relationship between language learning and attitudes is inherently
cyclical in nature, in that positive attitudes may lead to learning which
in turn may lead to positive attitudes, and vice versa for negative
attitudes and a lack of learning. In addition, the social context of
language learning may influence the extent to which attitudes affect
language learning, in that attitudes may play a larger role when
language learning takes places in the second language context, and a
lesser role in foreign language contexts (Gardner 1980).
Motivation has typically been theorized as being comprised of two
separate components: integrative motivation and instrumental motiv-
ation. Integrative motivation is denned as the interest in learning an
L2 because of a personal interest in the target language culture and
people whereas instrumental motivation has been defined as an
interest in learning an L2 due to its practical value and advantages
(Lambert, 1974). Motivation is seen as the effort individuals put into
learning an L2. Correlational studies have consistently found a
positive relationship between levels of integrative motivation and L2
achievement (Gardner and Lambert 1972) although conflicting
results exist (Oiler, Baca and Vigil, 1977). Instrumental motivation
may also positively affect language learning, especially in contexts
where learners have little or no opportunity to interact with members
of the target culture (Gardner and Lambert 1972; Lukmani, 1972).
Research has also indicated that motivation may result from SLA,
rather than the other way around (Strong 1984).
Recently, however, the concept of motivation has come under
criticism for not capturing the complex relationship between the
language learner and the learning context (Peirce 1995). As Peirce
states, 'theorists have not adequately explored why it is that a learner
may sometimes be motivated, extroverted, and confident and
sometimes unmotivated, introverted, and anxious' (11). Peirce
instead proposes the term 'investment' to reconceptualize the
relationship between the individual and society, which is dynamic
and complex rather than a fixed property of the language learner.
Peirce defines investment by stating that:
.. .if learners invest in a second language, they do so with the
understanding that they will acquire a wider range of symbolic and
material resources.. .1 take the position that this return on
investment must be seen as commensurate with the effort
expended on learning the second language. (17)
Arguing that investment is not the same as instrumental motivation,
Peirce (1995) further states that learners' investment in the target
language is changing across time and space, and is intricately linked
with their social identity, which is also multiple and dynamic.
Peirce (1995) employed the concept of investment to explain why
the immigrant women in her study were unlikely to speak in certain
circumstances although all were highly motivated to learn English.
She found that the learners' motivation to speak was mediated by
investments in being a member of the target culture, for example
resisting to speak in order not to be identified as an immigrant in the
target culture. While interesting and insightful, this construct still
needs further investigation.
In summary, there is a scarcity of research on social constraints on
the development of an L2 phonology even though most researchers
acknowledge the importance of the social context in language
learning (cf. Ellis 1994). Research about social identity and second
language acquisition has focused mostly on language use, rather than
acquisition, and findings indicate that learners actively determine
their own target language variety, due to social networks and in-group
identification and solidarity (Adamson and Regan 1991; Beebe 1985;
Edge 1991; Hecht and Mulford 1982; Mulford and Hecht 1980).
Gender may affect access to L2 learning opportunities (cf. Gumming
and Gill 1992; Goldstein 1995, 2001; Kouritzen 2000). Learners may
also resist learning and using specific linguistic features due to
conflicts with their LI identity and gender (Lybeck 2002; Ohara
2001; Pennington 1994; Siegal 1996). Several aspects of social
networks have also been found to affect L2 acquisition. Amount and
type of contact with speakers of the target language has been
correlated with L2 proficiency (Bayley 1996; Heidelberger
Forschungsprojekt 'Pidgin Deutsch' 1978; Meisel, Clahsen and
Pienemann 1981; Monshi-Tousi, Hosseine-Fatemi and Oiler 1980;
Oyama 1976; St. Martin 1980; Spada 1986). Social interactions,
whether within the LI or L2 community, have also been found to
facilitate or limit the opportunities learners have to use the target
language (Goldstein 1995; Peirce 1995). Positive attitudes towards
the target culture may enhance learning (Gardner and Lambert,
1972), but this may differ across learners and learning situations
(Gardner 1980). Both instrumental and integrative motivation have
been found to influence language learning (Gardner and Lambert
1972; Lukmani 1972) although motivation may need to be
reconceptualized as a dynamic investment in the target culture that
is mediated by the learners' social identities and opportunities for
social interactions in the target culture (Peirce 1995).
2.1.3 Variation
Another area of research on L2 phonology has focused on determin-
ing which factors, such as linguistic environment, grammatical
conditioning, social class and task type, affect L2 production in
terms of the use of specific variants. Research that focuses on
linguistic and social constraints on production is discussed first,
followed by research on variation in production across tasks.
2.1.3.1 Linguistic and social factors
Research on variable production, also referred to as variationist
research, examines the effect of linguistic (e.g. preceding and
following linguistic environment) and social (e.g. gender, social
class) factors on variable production. Preston (1996) states that:
The central claim of this approach is that the alternative forms of
linguistic elements do not occur randomly. The frequency of their
occurrences is predicted by 1) the shape and identity of the element
itself and its linguistic context, 2) stylistic level (defined operation-
ally), 3) social identity, and 4) 'historical' position (i.e., an
assumption that, in much variation, one form is on the way in,
the other on the way out. (2)
Findings from this area of research indicate that a number of
linguistic and social factors may constrain production. In one of the
earliest variation studies, Dickerson (1975) found that task and
phonetic environment affected the production of Izl by Japanese
learners of English. Specifically, a following vowel promoted accurate
production of /z/, while a following pause or following consonant
promoted the deletion of/z/ or production of Izl as [s]
3
or [ds], for
example.
In research on the acquisition of English L2 syllable codas by native
speakers of Chinese, Hansen (2001) found that both grammatical
conditioning and linguistic environment affected the production of
codas. Grammatical conditioning was one linguistic factor: final /t d/
was deleted in lexical over inflectional clusters. Additionally, both
preceding and following linguistic environment were found to have an
effect on coda production. The study also found that homovoicing of
segments favoured absence while heterovoicing favoured retention.
Other L2 phonological studies, though not necessarily variationist,
have also found that linguistic factors such as phonological environ-
ment have an effect on L2 production: for example, both Gatbonton
(1978) and Major (1996) also found that a following vowel may
facilitate production (vs. deletion) of a given segment while other
researchers (Edge 1991; Major 1987c; Tarone 1980) have found that
a following pause may facilitate devoicing and/or epenthesis. Benson
(1988), Osburne (1996) and Yavas (1997) also found that the
preceding linguistic environment had an effect on production, with a
preceding diphthong promoting absence of the following coda for
Vietnamese speakers. Non-linguistic and social factors such as age of
L2 learning, home use, integrative motivation and work use (Flege,
Munro, and MacKay 1996), gender (Adamson and Regan 1991) and
social networks (Bayley 1996) have also been found to affect L2
variation.
2.1.3.2 Task variation
Finally, it is important to note that task type has a strong influence on
learners' production of consonant codas. Researchers (Beebe 1980;
Beebe and Zuengler 1983; Dickerson and Dickerson 1977;
Gatbonton 1978; Schmidt 1987) have long noted that L2 learners
systematically vary their production of certain variants based on level
of formality (e.g. speech style) and task (reading versus conversation).
Tarone's Capability Continuum Paradox, based on Labov's (1969)
Observer's Paradox, states that:
1. variation occurs for every speaker as the topic and situation
change;
2. there is a continuum of styles, which are defined according to
the amount of attention given to speech;
3. the most systematic patterns occur in the vernacular (colloquial
style), while more variability occurs in other styles. (Major 1994:
656-7)
There is some support for the model. Beebe (1980), for example,
found a greater accuracy in English for Thai speakers in word-initial
/i/ in casual speech. However, in terms of style and correct L2
production, many researchers (e.g. Dickerson and Dickerson 1977;
Gatbonton 1975; Major 1987c; Sato 1985) found that learners had a
greater accuracy in L2 production as style becomes more formal.
Dickerson and Dickerson (1977) for example, found that their
Japanese learners of English were twice as accurate in their produc-
tion of III when they were reading a word list than in conversation.
Sato (1985) also found limited support for the continuum paradigm.
In her study of a Vietnamese learner of English in terms of three tasks
(free conversation, oral reading of a short text, and elicited imitation
of words and short phrases), Sato found that for word-final
consonants, task variation was inconsistent. However, her findings
regarding task variation and consonant cluster production confirms
previous research that production is more accurate in formal styles.
Major (1994) found no significance for style variation; however, his
style was a difference between a word-list reading and text-reading,
which are perhaps too similar in nature to illustrate differences in
style. Therefore, his hypothesis that as style becomes more formal,
interference decreases and developmental processes stay unchanged,
and thereafter increase and then decrease, was not supported. But as
Schmidt (1974) demonstrates in his study of Egyptian Arabic learners
of English, greater accuracy in formal style may also be due to positive
interference of the formal LI variant which happens to be the target
in the L2. Schmidt found that the participants' production of
interdental fricatives, sibilant and stop variants was predictable
according to style: interdental fricatives predominated in more formal
styles such as reading word lists; while sibilants predominated in a
slightly less formal style of a reading passage; and stops predominated
in the casual style of informal conversation. Additionally, the
interdental fricatives were also a highly developed sociolinguistic
marker, as the more highly educated participants had a greater
frequency of use of those features than the working class participants,
more than half of whom did not produce the interdental fricative at
all. Schmidt's (1987) study of the stylistic and sociolinguistic use of
the variants of the English interdental fricatives by native speakers of
Arabic confirmed his earlier study that style determined variant -
again, the interdental fricatives had a higher frequency of use in more
formal styles, and that use of those features could be in part predicted
by social class and educational level. Beebe (1980) also found that
sociolinguistic patterns of /j/ usage was influenced by the LI - the
formal Thai variant of/r/ is transferred and used in English in formal
settings.
In summary, style variation (as dependent on the type of task
learners were required to do, each of which supposedly supported a
different style, from formal to vernacular, based on amount of
attention to the task) has been seen to depend partly on the linguistic
variable examined (in this case consonants or consonant clusters);
systematic task variation was demonstrated for word-final consonant
clusters but not for word-final consonants due to LI transfer.
2.2 English and Vietnamese Phonology
2.2.1 English phonology
English has 24 consonant phonemes, which are outlined in Table 1.1.
English syllable structure is relatively complex since all English
consonants, with the exception of the velar nasal /rj/, and the palatal
fricative 1^1 (except in loan words of French origin such as genre), are
allowed word-initially, though / rj / is allowed syllable-initially word-
medially in an unstressed syllable (Jensen 1993). All consonants,
except for IbJ, and the glides /w y/, which are realized as part of the
diphthong vowel, are allowed syllable-finally (Hammond 1999).
Syllable onsets in English may consist of one to three consonants or a
vowel (Jensen 1993: 65). As Spencer (1996) states, the nearly most
complicated syllable structure in English is CCCVCCC (e.g. the
Table 1 . 1 An overview of English consonants
Place of
Articulation
Manner of
Articulation
Stops
voiceless
voiced
Fricatives
voiceless
voiced
Affricates
voiceless
voiced
Lateral
Nasals
Approxi-
mants
Bilabial
P
b
m
w
Labio-
dental
f
V
Dental
6
5
Alveolar
t
d
s
z
1
n
i
Palato-
alveolar
I
3
tj
d3
Palatal
y
Velar
k
g
q
Glottal
h
From Jensen (1993)
word 'strengths'). In terms of two-consonant onset clusters, there are
a number of linguistic constraints. Initial clusters must consist of the
combinations of plosive plus approximant (either a liquid or glide).
The combination of sonorant plus obstruent is not allowed (Spencer
1996: 83). These constraints result in the following 49 syllable onset
clusters: /sp/, /st/, sk/, /sf/, /sm/, /sn/, /si/, /sw/, /sy/, /Jp/*, /Jm/*, /Jl/*,
I I I / *, /Jw/*, /pi/, /pj/, /pw/*, /py/, /bl/, /bi/, /bw/*, /by/, /ti/, /tw/, /ty/,
/dj/, /dw/, /dy/, /kl/, /ki/, /kw/, /ky/, /gl/, /gj/, /gw/, /gy/, /fl/, /&/, /fw/,
/vl/*, Au/*, /Qif, /9w/, %/, /hy/, /my/, /ny/, and /ly/ (Jensen, 1993: 66).
The notation * denotes clusters that are foreign in origin but that have
been incorporated into the phonotactics of English (Jensen 1993: 66).
Three-consonant onsets also exist in English, and are formed by
adding /s/ (and only /s/ is allowed in the beginning of a three-
consonant onset) to already permissible two-consonant onsets,
resulting in the following nine three-member onsets: /spl/, /spj/,
/sti/, /ski/, /ski/,, /skw/, /spy/, /sty/, and /sky/ (Jensen 1993: 67). The
three-consonant onsets consist of clusters in which the first two and
the last two consonants respectively can form two-consonant onsets.
Additionally, /!/, /j/, /w/, and /y/, with the exception of /ly/, do not
appear as the first consonant in a two- or three-consonant onset.
Codas in English are even more complex since inflectional suffixes
at the end of words create complicated codas (Jensen 1993: 68).
Codas can consist of sets of two stops or two sonorants, which is not
possible in onsets. The set of two-consonant codas in English are as
follows: /pt/, /p6/, I ps/ , / bdl, /bz/, /t6/, /d6/, /dz/, /kt/, /ks/, /gd/, /gz/,
/tft/, /ds/, /ft/, /ft/, /fs/, /vd/, /vz/, /6t/, /6s/, /Sd/, /5z/, /sp/, /st/, /sk/, /zd/
s
/ft/, /3d/, /mp/, /mt/, /mf/, /mz/, /nt/, /nd/, /ntJY, /nds/, /n6/, /ns/, /nz/,
/qd/, /qk/, /q6/, /qz/, /lp/, /lb/, /It/, /Ik/, /Id/, /ItJ/, /Id
3
/, /If/, /lv/, /IB/, /Is/,
/lz/, /1J/, /1m/, /In/, /jp/, /ib/, /at/, /id/, /jk/, /Jg/, /atj/, /ids/, /if/, /iv/, /i9/,
/is/, /iz/, /jj/, /im/, / in/ , l\\l Qensen 1993: 69). There are fewer three-
consonant codas, namely: /dst/, /kst/, /ks6/, /mps/, /Ipt/, /Ikt/, /Iks/,
/IfB/, /ikt/, /nst/, /qst/, /qkt/, /qks/, /imG/, /ipt/, /Jps/ (Jensen 1993: 70).
Some rules can be noted: all the three-consonant clusters end in I I I ,
/s/, or /6/, which are all voiceless coronal obstruents (i.e. stops, nasals
and fricatives). The majority of two-consonant codas also end in
coronals; furthermore, vowels occurring before both two- and three-
consonant codas are mostly short and lax. In fact, long vowels do not
occur before three-consonant codas, although they may occur before
two-consonant codas if the two-consonant cluster contains only
coronals Qensen 1993: 70). Four consonant codas also exist; in these
cases, the final consonant is a bound morpheme, either /s/ for plural
or third person singular, or /t/ or /d/ for past tense.
Table 1.2 An overview of Vietnamese consonants
Place of
Articulation
Manner of
Articulation
Plosive
plain
aspirated
voiced
Fricatives
voiceless
voiced
Lateral
Nasals
Trill
Semi-
vowels
Bilabial
b
m
w
Labio-
dental
f
V
Dental
t
i
h
d
n
Retroflex
t
;
r
Alveolar
1
Jl
From Santry (1997) and Nguyen (1970)
Alveo-
Palatal
c
j
Velar
k
X
Y
q
Glottal
h
2.2.2 Vietnamese phonology
There are a total of 22 consonant sounds in Vietnamese, not
including allophonic variations, which are illustrated in Table 1.2.
Each of these sounds will be described in turn, beginning with the
plosives: /b/ is a voiced bilabial plosive in initial position. It has an
allophone in final position, /p/, which is a voiceless, unreleased,
bilabial plosive. In terms of the dentals, /t/ is a voiceless, unaspirated
dental plosive which occurs both initially and finally, while /d/ is a
voiced apical-alveolar plosive which only occurs in initial position.
However, /t
h
/ is a voiceless, aspirated apical-dental plosive which is
released with a puff of air, which is very audible. Another plosive, Id,
is a voiceless lamino-palatal plosive, which is unreleased in final
position, /k/ is a voiceless, unaspirated, dorsovelar plosive with five
allophonic variations, depending on the preceding vowel. In initial
position, /k/ is produced as [k]; however, in final position after the
vowels /u ra/ it is produced as [k
p
], which means it has a simultaneous
bilabial closure; in final position after lot it is produced as [k
w
], which
means it is produced strongly labialized, i.e. with rounding of the lips.
In final position after hi, it is weakly labialized, and produced as [kw].
After all other vowels, it is unreleased and produced as [k
1
].
Vietnamese also has three retroflex sounds, /r/, /{/, and /JY, the
former of which is a voiced alveolar retroflexive in initial position,
with three variants: [r], which is typically produced as a series of rapid
taps or trills; [r], which is produced as a single tap; and [z], where
friction is caused by the stream of air. The second retroflex, l\l is a
voiceless, apical alveolar retroflex plosive, which occurs in initial
position only. It is slightly affricated. The third retroflex, /JY, a
voiceless retroflex fricative, also only occurs in initial position.
In terms of fricatives, /f7 is a voiceless labio-dental fricative which
occurs in initial position, while /v/ is a voiced labio-dental fricative,
which also only occurs initially, /s/ is a voiceless fricative, also
occurring only in initial position, which is the same for the voiceless
dorsovelar fricative /x/. The phoneme /y/, which is a voiced dorsovelar
fricative, may be produced as the stop [g], especially when the
preceding syllable ends in either /k/ or /q/. Another fricative, also
occurring only in initial position, is the voiceless glottal fricative /h/.
As for the nasals, /m/ is a voiced bilabial nasal with two variants,
[m] in initial position, which is released, and [m
n
] in final position,
weakly released. The voiced alveolar nasal /n/ has similar allophonic
variants, as well, [n], released, in initial position, and [n"
1
]. /jl/ is a
voiced palatal nasal while /q/ is a velar nasal in final position. Both of
these phonemes can occur both initially and finally. The voiced
alveolar lateral /!/ is unreleased (it is released in English) and only
occurs initially. There are also two semi-vowels, /w/ and /j/, both of
which can occur in initial or final position. The former is a bilabial
semi-vowel while /j/ is a palatal semi-vowel (Santry 1997).
Overall, all the phonemes except for /p/, which is an allophone of
/b/, can occur initially, while the phonemes that can occur in final
position are restricted to: /p t c k m r^n w j/, i.e. only plosives, nasals
and semi-vowels. Additionally, all the plosives are voiceless. There is
an overall agreement that the final consonants are unreleased or
weakly released; in addition, there are no consonant clusters in
Vietnamese. For the few consonants with allophones, allophonic
variation is caused by vocalic environment (Santry 1997).
In addition, Vietnamese is very restricted in the consonants that
can occur finally (see the description above) and there are no
consonant clusters. Syllables in Vietnamese can be reduced to the
following formula, which illustrates canonical syllable structure as per
Santry (1997): O[ + C W] + N [ V] + T <C, W>
3
with O
representing the Onset, C as Consonant, W as semi-vowel, N as
nucleus, V as vowel or diphthong, and T as terminus. This formula
states that a syllable in Vietnamese can optionally consist of an onset,
which may consist of either a consonant or a consonant plus a semi-
vowel, but must consist of a nucleus, which is composed of a vowel or
diphthong, and optionally a terminus, which may be composed of
either a consonant or a semi-vowel. This formula produces the
following possible syllable structures: V, VW, VC, VWC, VS, VW,
cv, cwv, cvc, cwvc, cvw.
Vietnamese is a monosyllabic tone language and the nucleus of the
syllable, which can be composed of vowels, diphthongs and
triphthongs, carry one of the five Vietnamese tones (Nguyen 1970).
The five tones are as follows: high-level, high-rising, low-level, low-
rising and mid-rising. The tones are employed to semantically
distinguish words from one another.
Notes
1 /if is usually used to denote the American English /r/.
2 See Major (2001) for a more recent discussion of this model.
3 It is standard IPA practice to employ slant / / brackets to denote the
underlying phoneme or phonemic transcription while square [ ]
brackets are used to denote the actual phonetic (physical) articu-
lation of a sound/word. This practice will be employed throughout
this book.
Researching the Nguyen
Family
This chapter introduces the participants of the study, the Nguyen
family, as well as the methods of data collection and analysis for both
the linguistic and the social data analysed in the study. First, the
setting of the study and the selection of the participants is described.
This is followed by a detailed explanation of how linguistic data and
social data, respectively, were collected and analysed. Thereafter, the
immigration patterns of the Vietnamese in America is presented, and
finally, the participants of the study, the Nguyen family, are
described.
3.1 Setting and selection of participants
The selection of participants was based on judgement sampling in
which 'the researcher identifies in advance the types of speakers to be
studied and then seeks out a quota of speakers who fit the specified
categories' (Milroy 1987: 26). The goals of the research commonly
dictate the methods employed to select participants for the study,
with these goals translated into criteria that individuals must meet in
order to be included in the study. In selecting the participants for the
study, I had the following judgement sampling criteria: first of all,
participants had to be native speakers of a language such as
Vietnamese with restricted syllable margins (onsets and codas).
Secondly, the participants should have arrived in the United States
within one year of the onset of the study so that the early stages of
onset and coda acquisition could be observed while at the same time
they would have enough English to be able to communicate
sufficiently during the interviews. A third criterion for participant
selection is that the potential participants were planning to reside in
the Tucson area for the following 12 months, which would help to
decrease the possibility of participant attrition. A final criterion for
participant selection was that participants must be willing to engage
3
RESEARCHING THE NGUYEN FAMILY 41
in weekly to bi-weekly data collection for the duration of the study,
which was scheduled for approximately ten months.
This study was conducted in Tucson, Arizona, as this was my
residence for the duration of the study. One of the participants for this
study was selected from one of the ESL courses (a beginner level
speaking and listening course I taught at Pima Community College in
Tucson, Arizona) after the course had ended. After this participant
agreed to become involved in the study, she further suggested that her
husband - also a native speaker of Vietnamese - also participate in the
study, and that the data would be collected at their home. I agreed to
collect data at their home as this would potentially be a more relaxed
data collection environment for the participants and since it allowed
entry into at least one aspect of the participants' lives, their home
life.
The data collection for this study took place on a weekly basis,
except for one or two missed sessions due to vacation or illness of
either the participants or myself, from March 21, 1999 until
December 19, 1999, for the duration of 10 months. Data collection
took place in the participants' home as per their suggestion. There
were a total of 28 interviews. Each data collection session lasted a
minimum of one to one and a half hours, producing a total of 42
hours of data. Each interview was tape-recorded with the partici-
pants' prior consent, using an AIWA tape recorder and SONY
microphone. Each tape was transcribed word-for-word after each
session. Two types of data were collected, phonological and social. A
discussion of each type of data collection, and analysis, follows. On a
final note, the participants were not paid for their time; rather, from
the very beginning both participants were very motivated to
participate. I believe their motivation and enthusiasm was due to
the limited opportunities both of them felt they had to learn and use
English: my weekly one to one and a half hour conversational
interviews were one such a rare opportunity, and they valued this
time greatly.
3.2 Data collection and analysis
3.2.1 Phonological data
Two types of phonological data were gathered, interview data and
reading data (comprising a reading passage and a word list), at three
intervals during the duration of the study. These intervals were
spaced evenly three months apart. The interview data were selected
from one of the interviews held closely to each of the word list and
reading passage data collection times, and each complete data set
(word list, reading passage and interview data) necessarily took place
across three successive interview times. As such, each data set was
approximately three to four hours in length, for a total often hours of
data for phonological analysis.
The interview data was the primary data source for the study as I
was interested in determining what onsets and codas the participants
could freely produce. Word list and reading passage data (see
Appendix A for the word list and Appendix B for the reading passage)
were also collected for a number of reasons: a) to investigate task
variation; b) to elicit certain consonants and consonant clusters the
learners may not freely produce during the interview sessions; and c)
to augment the number of complex clusters in the study. However,
this data is employed only as a comparison with the interview data
and is therefore not part of the primary data analysis. Additionally,
when the word list and reading passage were selected and developed,
the focus of the study was coda structures only and therefore the word
list and reading passage data is available only for coda structures. The
same word list and reading passage were given at each of the three
data collection intervals; however, as there was a period of three
months between each data collection time, there was only a slight
likelihood that the prior reading of the passage and word list had an
effect on subsequent readings. Word list and reading passage data
were collected individually from each participant, with one partici-
pant instructed to leave the room while the other read the word list
and reading passage, in order to prevent the participants from
influencing each other's pronunciation.
The word list was designed to maximize the type of consonants
and consonant clusters the speakers produced. The word list (see
Appendix A) was compiled by employing the English Pronouncing
Dictionary, 15th edition (Jones 1997) and utilizing the American
English pronunciation of words. For the most part, one or two-
syllable words were used in order to increase the likelihood that
the participants were familiar with all the words and that
orthographic representation would not impede pronunciation.
Each word was listed on a separate notecard, and each participant
was instructed to read each word three times. Prior to their
reading the word, I pronounced the word for them in order to
minimize reading errors.
The reading passage (see Appendix B) was selected from the
textbook Accurate English: A Complete Course in Pronunciation (Dauer
1992). The reading passage had been devised as an 'accent inventory'
for students studying American English pronunciation. The reading
passage was selected on the basis that it has a wide range of
consonants and consonant clusters. However, the reading passage did
not elicit all onset and coda structures and therefore was used to
supplement the word list. After having a few minutes to read the
passage silently to themselves, the participants read the passage one
time on a separate basis.
In terms of the interview data employed for phonological analysis, I
selected a longer interview (one to two hours) that closely corres-
ponded in time to the collection of the word list and reading passage
data. The discussion during these data collection times focused on
the participants' social interactions, LI and L2 language use,
attitudes and motivation, etc. In other words, the interviews probed
the participants' social constraints on English language use.
The speech samples were transcribed using the International
Phonetic Alphabet (IPA). Following Sato (1985) and Hansen
(2004), data were coded for the following: a) native-like production;
b) absence
1
of one or more member; c) epenthesis; and d) feature
change (voicing, place or manner changes). For example, if the
consonant under study was III,2 native-like production would be [t],
absence would be as the lack of III in [las] for lots. Epenthesis would
mean that the speaker has inserted a vowel, typically a schwa, after
the consonant, as in [latas] for lots. Feature change would mean that
the speaker produced the consonant in an articulatorily different
manner, place or voicing. For example, many learners of English
produce the voiceless interdental fricative /6/ as a voiceless alveolar
fricative [s], thereby changing the place of articulation from
interdental to alveolar. A fifth category emerged during the data
analysis - two types of modification of a single onset or coda. For
example, one consonant in a two-member cluster may be absent
while a vowel was inserted after the remaining consonant as in the
production of dogs as [do.zs], which actually occurred in the data. In
this case, the modifications were classified as two types, rather than
as separate examples of both absence and epenthesis. The data were
transcribed twice with a few days interval between each transcription,
in order to ensure that the tokens were transcribed accurately, with
an average intra-rater reliability of 93 per cent. Additionally, another
IPA trained researcher randomly transcribed 20 per cent of the data,
with an inter-rater reliability of 94 per cent. When discrepancies
arose, a third rater was asked to transcribe the word in question. In
the rare cases that there was disagreement among the three raters
(less than 1 per cent of the time), the word in question was
eliminated from analysis.
A total of 15,749 tokens of both syllable-initial and syllable-final
consonants and consonant clusters resulted from the data sets. In the
interview data, there were 4,400 syllable codas and 6,673 syllable
onsets; there were 4,676 syllable codas in the word list data.
Descriptive statistics were utilized to examine correct production
and production modification based on length and type of onset or
coda, for each participant individually and combined, and within and
across time. Acquisition was measured by an 80 per cent criterion
method, as employed by Cancino, Rosansky and Schumann (1978),
which has been used in L2 phonological research by Eckman (1991),
Eckman and Iverson (1993) and Carlisle (1997, 1998). In this
method, if the participants use a certain structure correctly 80 per
cent of the time, it is considered to be acquired. Correct usage was
calculated as number of correct attempts over total number of
attempts. The 80 per cent criterion acquisition method was
employed rather than measuring acquisition as onset of usage since
language acquisition is not necessarily a linear process, and one-off
usage of a form may not indicate that the form has actually been
acquired.
A number of VARBRUL analyses were also run on the data.
VARBRUL is a loglinear regression inferential statistics program that
has been developed by linguists in order to examine probabilities for
the use of certain linguistic features, depending on both internal
factors (e.g. linguistic factors such as linguistic environment) and
external factors (e.g. social factors such as gender and/or social class).
It is increasingly used in SLA by researchers such as Adamson and
Regan (1991), Bayley (1996), Major and Faudree (1996) and Young
(1988). In VARBRUL analysis, the researcher's question is 'not
which single factor is associated with variation, but what the relative
weight of the different factors associated with variation is' (Young and
Bayley 1996: 254). VARBRUL calculates the input probability that a
specific variant, for example, deletion rather than /t/, will be produced
based on the proposed factors such as following phonological
environment and social class. An input probability of 1.00 means
that in this context, the variant will be categorically produced. In
other words, there is no variation in this particular context. If this
occurs, the program will not run the analysis as VARBRUL requires
that the data must exhibit variation. When the input probability is
.50, there is an even likelihood of a variant being produced or not in
this environment. An input probability between .01 and .49 indicates
that a factor inhibits the production of the variant, while the reverse is
true of input probabilities ranging from .51 to .99. VARBRUL also
subjects the variables and the factors to three tests of statistical
significance: Total chi-square, Chi-square/cell and Log likelihood.
The VARBRUL version used was Goldvarb Version 2: A Variable
Rule Application for the Macintosh (Rand and Sankoff 1990).
The dependent variable was classified into five categories: a)
accurate production; b) absence (of one or more members); c)
epenthesis; d) feature change; e) two types of modification. Two
types of independent factor groups were investigated: internal
(linguistic) factors (based on Bayley 1996 and Guy 1980) and
external (extralinguistic) factors. These factors are outlined in detail
below. One important note is that since Vietnamese does not allow a
coda following a diphthong, and as this has been found to promote
absence in prior research on Vietnamese learners of English (Osburne
1996), diphthong and vowel were selected as factors for preceding
linguistic environment for codas and preceding linguistic environ-
ment for onsets. Additionally, during the descriptive analysis of the
data, it was noted that the mid back vowel hi also promoted absence
in the following segment and therefore this preceding vowel was
included as a separate category in the analysis. Finally, not all factor
groups or factors within each group were appropriate for each
VARBRUL analysis (e.g. onsets were not coded for grammatical
conditioning) and therefore, this list of factor groups and factors given
below indicate only the range of variables explored. The actual factor
groups and factors employed will be discussed in each VARBRUL
analysis in the following chapter.
Internal (Linguistic) Factor Groups:
1. Preceding phonological environment (for onsets: Pause as in a
noticeable break in connected speech, Vowel as in you tell where
initial /t/ is preceded by a vowel; and Consonant as in hand you
where initial /y/ is preceded by a consonant; for codas:
Diphthong vowel as in hide, Monopthong vowel as in hid, hi
as in dog).
2. Following phonological environment (for onsets: Diphthong
vowel as in hide, Monopthong vowel as in hid, hi as in dog, for
codas: Pause as in a noticeable break in connected speech,
Vowel as in tell us where final /!/ is followed by a vowel, Glide as
in tell you where final HI is followed by the glide /y/, and
Consonant tell me where final HI is followed by the nasal /m/).
3. Grammatical environment (Monomorphemic words as in tell,
Bimorphemic words as in the past tense ending of talked).
4. Stress (Unstressed syllable as in problem, Stressed syllable as in
Vietnam}.
5. Voicing agreement with the following segment (Homovoicing as
in tell us where both final III and the initial vowel are voiced;
Heterovoicing as in tell Paul where III is voiced but /p/ is
voiceless).
External (Extralinguistic) Factor Groups:
1. Speech sample (Time 1, Time 2, Time 3).
2. Participant (Anh or Nhi).
3. Length of cluster (one-member, two-member, three-member).
5.2.2 Social Data
A grounded theory approach (Strauss and Corbin 1990) was used in
social data collection and analysis. A grounded theory is one that is:
.. .inductively derived from the study of the phenomenon it
represents. That is, it is discovered, developed, and provisionally
verified through systematic data collection and analysis of data
pertaining to that phenomenon. Therefore, data collection, analy-
sis, and theory stand in reciprocal relationship with each other.
One does not begin with a theory, then prove it. Rather, one begins
with an area of study and what is relevant to that area is allowed to
emerge. (Strauss and Corbin 1990: 23)
In grounded theory, research questions start out broad and then
progressively narrow in focus as concepts and their relationships
emerge as relevant. In collecting data, the researcher typically
employs interview or observations. As data collection and analysis
is cyclical in nature in this approach, researchers analyse the data
after each data collection in order to discover salient themes or
patterns which can guide further data collection (Strauss and Corbin
1990).
A number of data gathering techniques were employed to gather
information on social factors and to triangulate data collection in
order to better understand how social factors affect language
acquisition. My initial research question was broad, concerning
social constraints on language learning, which narrowed throughout
the study as salient themes, such as identity, gender, LI and L2 use
opportunities, acculturation, motivation and attitudes towards Ian-
guage use and interactants, began to emerge. A questionnaire (see
Appendix C) was employed as an initial data gathering instrument.
The questionnaire was based in part on Peirce (1993) and Day
(1985). The administration of the questionnaire was in a one-on-one
oral formatj and therefore, the administration of the questionnaire
was actually in the form of an interview, and as such, tape-recorded.
Information gathered from the questionnaire served as a basis for
further probing during subsequent interviews with the participants.
Special focus was on the participants' contact with native speakers of
English and members of the LI community, as well as affective
reactions to English and Vietnamese language use and the
Vietnamese community in Tucson. Additionally, I created a list of
follow-up questions based on the participants' answers to the
questions about their perceived levels and desired levels of
acculturation, access to and opportunities for LI and L2 use,
identity, motivation in learning English and becoming American and
Americanized, and attitudes towards Americans and other
Vietnamese in Tucson, etc. (see Appendix D for examples of
follow-up questions). While these questions were discussed through-
out the duration of the study, they were a focus for more intensive
probing during each of the three phonological data collection times in
order to gather a more complete picture of the participants' language
use contexts during each of the three data collection times.
Additionally, I had the participants keep track of their English and
Vietnamese language use by employing a 'Language log' question-
naire that they filled out on a weekly basis (see Appendix E) that I
then utilized as a basis for probing their social interactions each week.
As the interviews took place in the participants' home, I also had an
opportunity to observe family interactions and home language use,
and therefore field notes were taken during all the interviews. In
addition, the participants came to my house for social gatherings on a
number of occasions, at which point I was able to observe their
interactions with other individuals and their language use patterns in
naturalistic communication with other individuals. While these
interactions were not tape-recorded, I took field notes immediately
after each of these occurrences, and also discussed the participants'
interactions with other individuals who interacted with the partici-
pants. I also took notes on all the phone conversations I had with the
participants, and collected email and other written communication in
order to compare their spoken and written language.
Data on the social constraints on language learning were analysed
within a grounded theoretical framework (Strauss and Corbin 1990)
as described above. Specifically, the social interaction and language
use patterns of each participant were probed in more detail through
interviews (and observations) over the duration of the study.
Interviews and observations were transcribed and coded for emerging
patterns and trends immediately after each interview and observation.
These patterns and trends then formed the basis for questions for
subsequent interviews. Additionally, I frequently checked the infor-
mation I had received from the participants and my interpretations of
the information with the participants in order to maximize the
accuracy of the findings.
3.3 The Vietnamese in America
The settlement of Vietnamese refugees in America can be divided
roughly into three waves of immigration, stemming from political
conflicts. Relatively few Vietnamese were living in the United States
before the fall of the South Vietnamese Government. The first wave,
in 1975, occurred after the fall of Saigon on 30 April 1975, and the
collapse of the Thieu regime. Many of these immigrants either fought
alongside the American forces against the North Vietnamese or
served in the South Vietnamese Government and therefore feared
retaliation from the communist Government of North Vietnam,
which took over South Vietnam and reunified Vietnam.
Approximately 150,000 Vietnamese fled to the United States, aided
by the American Armed Forces, who first brought them to refugee
centres in Guam, and then to the US mainland. Other countries such
as Germany, China and Japan also took Vietnamese refugees. Of
these refugees, most were young, well-educated and urban dwellers.
Many of these refugees spoke English as they had worked with
American personnel in the Armed Forces or in American companies
in Vietnam. Many had also received some education in the United
States. Interestingly, about 55 per cent of these immigrants were
Catholic. This first wave of refugees assimilated fairly easily into
American society due to their English language skills and in part, their
Christianity, as this served as a point of friendship with some
American citizens. Many of this wave of immigrants settled in
California, Texas and Florida (Rutledge 1992).
The second wave of immigration took place between 1979 and
1983 because of the Vietnamese invasion of Cambodia, military
actions against the Laotian hill people and the anti-Catholic policy of
the Vietnamese Government. In this wave of immigration, approxi-
mately 455,000 south-eastern Asian refugees settled in the United
States. This group of refugees was diverse in terms of language,
ethnicity, religion and nationality (Rutledge 1992). The third wave of
immigration took place between 1985 until present as family
reunification programmes began bringing Vietnamese people to the
US.
The overall statistics of the Vietnamese population in the United
States is approximately 480,000 immigrants and 107,000 US-born
Vietnamese (76 per cent of these are aged between 0-9).
Approximately 54 per cent of the Vietnamese in the US have settled
in the west (i.e. California, Texas, Arizona and Nevada). In terms of
the economic situation of the Vietnamese population in the US, in
1979 35 per cent of Vietnamese families were living below the poverty
level. This had increased to 50 per cent by 1985.
The state of Arizona has approximately 7,200 south-east Asian
refugees as of 1989 (Rutledge 1992). The city of Tucson, Arizona,
had a total population of 799,375 as of 1997, the latest statistics
before the onset of the study. Although there are no official records
available, members of the Vietnamese community in Tucson estimate
that there are approximately 3,000 Vietnamese in Tucson.
3.4 The Nguyen family
The Nguyen family, husband Nhi, wife Ann, and daughter Minh (all
pseudonyms), immigrated to the United States in March of 1998. As
such, they were part of the third wave of immigrants, and came in
order to be reunited with family members already residing in the US.
The story of how the Nguyen family ended up in the US begins
during the Vietnam war. Both Anh's and Nhi's fathers served in the
US military; Anh's father died serving the US in the war against
North Vietnam. Since both Anh's and Nhi's families had been
involved in the war and were allies of the Americans, they felt insecure
and decided to leave Vietnam after the war ended.
Anh comes from a family of five brothers and sisters, of whom she
is the oldest. The family began immigrating in 1975. First, one sister
and one brother left Vietnam, and fled to Thailand, where they had to
wait for sponsorship. Nhi comes from a family of 12, of which he is
also the oldest. In Nhi's case, his entire family, with the exception of
him, left for Thailand as well. At this time, Nhi's father was in prison,
where he stayed for seven years due to his role in the war as he was a
high-ranking official in the South Vietnamese army. As the oldest son,
Nhi could not leave Vietnam at this time because he had to take care
of his father in prison. Because Anh had married Nhi (in 1979), she
could not leave either. However, both of them knew that they would
eventually leave Vietnam - they just had to wait.
In Thailand, many nations were sponsoring the Vietnamese
refugees, in effect randomly picking them out of the crowd. As
there was a great uncle in Nhi's family who lived in Germany, Nhi's
whole family elected to go to Germany. They have now been there for
20 years and are all German citizens. While facing some difficulties in
Germany due to cultural adjustment and language barriers, Nhi's
family has been successful in adapting to this new culture. They have
all become fluent in German, and many speak English as well. One of
Nhi's brothers is a Catholic priest and one of his sisters is a Catholic
nun. He has a brother who is a psychiatrist and another who owns a
clothing store.
Ann's brother and sister were sponsored by an American woman
who lived in Tucson. Slowly, the brother and sister were able to
sponsor the other relatives, and eventually all of Ann's family came to
the US, including her mother, who has been here for 13 years already.
Anh's mother lived in Tucson but eventually moved to San Diego in
order to live in a larger Vietnamese community.
Like Nhi's family, Anh's family has also integrated into their new
life very successfully, attaining college degrees and middle-class
incomes. All Anh's brothers and sisters attended the University of
Arizona. The first brother and sister who came to the US became
engineers. In fact, both of them work for the government, one as a
satellite engineer and the other as an electrical engineer, and live in a
small town, though not together, outside Tucson. The sister has a
family, having married another Vietnamese, while the brother is still
single. Anh's other brother also went to the University of Arizona,
though without finishing his degree in computer science, and works
as a technician for a weapons factory in Tucson. One of Anh's sisters
went into the Navy after getting a degree in computer science from
the University of Arizona. She worked as a computer trainer in the
Navy, having the rank of Captain, and lived in Virginia. A third sister
works for IBM and owns a nice house with a swimming pool in an
exclusive area of Tucson.
Back in Vietnam, Nhi's father was finally released from jail after
being imprisoned for eight years. Due to poor nutrition and health
care in jail, he died one month after his release, which was apparently
quite common for many released prisoners due to illnesses suffered
and left untreated in jail. While their families were adapting to and
becoming successful in their new countries, life was very hard for the
Nguyens. Nhi was a high school chemistry teacher and Anh was an
elementary school teacher. In Vietnam, the education system was
controlled by the communist Government who stipulated that it was
the teachers' responsibility that all their students should pass their
courses. If all the students did not pass, then the teacher would be
criticized and lose benefits or get a lower pay. This resulted in
immense pressure for teachers and also gave rise to some teachers
cheating the system or offering outside tutoring in order to survive.
Teachers had to work very hard and most had students come to their
house for extra tutoring in order to ensure that the students would
pass the exams. After school, Anh normally had 10-12 students at her
house that she took care of and tutored all for no extra pay. Both Nhi
and Anh were under extreme pressure and never felt they could relax.
Anh said she had to take sleeping pills every night in order to finally
be able to sleep. After being a teacher for 22 years in Vietnam, Nhi
still only made $500 a month, and got a yearly benefit/bonus of only
$5. As college for their daughter could cost about $500 per semester,
life was very hard.
Finally, after waiting for over 14 years, their families were finally
able to sponsor them and the Nguyens came to the US. There had
never been a choice of whether to go to Germany or to the United
States since: a) German immigration policy had changed and only
young single Vietnamese under the age of 18 could obtain permis-
sion to immigrate to Germany; and b) both Nguyens wanted to live
in the United States since they believed it would be easier to adapt to
and become successful in this society based on their families'
experiences.
The transition to live in the United States has not been easy for the
Nguyens, especially for Anh. While their families initially supported
Anh and Nhi financially and helped them to find a place to stay (they
currently rent a house in a fairly nice area of Tucson), Anh and Nhi
were surprised when they realized that while their family would help
them, they were expected to live on their own, rather than with a
member of the family, as was traditional in Vietnam. Anh's brothers
and sisters had successfully adapted to the American lifestyle, and
they expected Anh and Nhi to do the same. However, as the Nguyens
had come at a later age (Anh was 41 when they came and 42 at the
time of this study while Nhi was 46 when they came and 47 at the
time of this study), they had a more difficult time adjusting than
Anh's siblings who had also had the opportunity to attend college in
the US.
Initially, Anh found work as an 'inserter' for the local newspaper:
she inserted the ad pages into the daily newspaper before it got
delivered. She worked at this for several months before saving enough
money to attend a nail technician school. She had decided to study to
be a nail technician because it was a relatively easy job to find, and she
could get a job after studying for only three months. Additionally, in
Tucson, the majority of the nail shops are Vietnamese-owned, with
almost all Vietnamese employees. Therefore, language difficulties
would not hinder her in seeking employment. While Ann had
difficulty in understanding the classes at the nail technician school,
she survived by watching videos and watching what the teacher was
doing. At night, she would translate the reading word-by-word into
Vietnamese until she understood the meaning. After three months,
she finished attending the school and easily found a job in a
Vietnamese nail salon. Nhi found a job in a factory where he began
working as an order filler - as an invoice for a product came in, he
would go and find it and put it on a cart. This was tough work, but as
his English became better, he was moved to order taker, where he
now interacts with customers who come to the factory. He also held a
second job at weekends as a dishwasher in a Chinese restaurant;
however, after six months he quit this job to enable him to have more
time to study English in order to improve his life.
While they admit that life in the US is hard, they like life here
because they feel success can be achieved by working hard. At the nail
salon where she works, Anh determines how much she earns by how
hard she works; she is on a commission which means she gets 60 per
cent of what her client pays. Anh knows she can work more hours
than she currently does, but it is not that important to her in the long
run because she wants to take time off to learn more English as her
long term plans are to go to the University of Arizona and get a degree
in computer science. Her job as a nail technician enables her to
survive financially at the moment while she is learning English, but
her end goal is to work in an office as a computer programmer. At the
time of the study, Anh made about $500 a week in the off-season
(May through December) and about $600 a week during the winter
since it is peak season due to the increase in tourism. She is the main
breadwinner in the family.
As for Nhi, he was paid $1000 a month at the factory and his goal is
also to take classes at the local community college for two or more
years until his English is 'so-so' as he said, and then go to the
University of Arizona to get a BSc in chemistry. He is not sure what
he wants to do after that, but his dream is to go to university full-time.
Their daughter Minh is currently attending the University of
Arizona and double majoring in engineering and computer science. A
lover of literature and languages in Vietnam (she studied both
Chinese and English in Vietnam), Minh Nguyen adjusted extremely
quickly to her new life in the US. She attended 12th grade in an
American high school and quickly realized that in order to make
friends and get good grades, she had to practise her English even if
she made mistakes. Outgoing, confident and assertive, Minh quickly
made friends and by the time I met her parents and began this study,
Minh was almost fluent in English, which precluded her from being a
participant in the study. Minh is very loquacious, and she speaks very
well, overall, though she makes some grammatical mistakes, but her
pronunciation is usually clear. When together with her mum and dad,
she tends to dominate the conversation because: a) she is very
outgoing (this probably contributed to her learning English so quickly
and making friends in the US); and b) because they sometimes defer
to her when they can not think of a way to express something, which
then tends to reinforce her role as the 'talker' of the family. She may
also correct their English, which also reinforces her role as the
superior English speaker and changes the balance in the family
relationship. She works as a maths grader at Pima Community
College, and at times, such as during university breaks, works as a
waitress at a Chinese restaurant. In her first year at the University of
Arizona, she enrolled in 22 credits per semester.
Anh is extremely soft-spoken and initially appears very shy. She is
serious most of the time and appears worried about life, her English
and her job. Anh tends to speak quietly, which can make her difficult
to understand. She has an extensive vocabulary but she speaks less
frequently than Nhi. However, when she speaks she is quite fluent,
though she does tend to make more grammatical mistakes than Nhi.
She speaks more carefully, and in words rather than full sentences.
This is probably also a result of interacting primarily with Mexican
customers at work, as she probably communicates word-by-word in
order to get her meaning across. In terms of listening comprehension,
Ann's is actually better than Nhi's at times, though it is unclear from
her facial expression whether she understands. However, once in a
while she will respond with a very articulate sentence or word, or she
sometimes translates my meaning for Nhi.
Nhi is very eager to talk, he is easy-going and always smiling and
happy. He is very eager to learn English and all about American
culture, and is more of a risk-taker in using English. Overall, Nhi
appears a bit disfluent as he has many hesitations and false starts,
repeating the initial word(s) over and over again several times before
finishing his idea. However, his ideas are typically expressed in perfect
grammatical form. Therefore, while he has a harder time expressing
his ideas, he tends to do so with perfect grammar while Anh expresses
herself quite easily, but not always in complete and grammatical
sentences. In terms of listening comprehension, Nhi tends to jump in
more quickly to answer, and he does not always interpret the question
correctly, so he does not always respond correctly. So overall, his
listening comprehension may not be as good as Anh's.
When Anh and Nhi talk with me together, there is usually intensive
scaffolding between the two of them. For example, they tend to finish
each other's sentences, add words in either Vietnamese or English,
and correct each other's pronunciation, grammar and interpretations
of my questions. As such, they are very successful in communicating
their meaning, and while I had some difficulty understanding them in
the beginning, once I adjusted to their speech patterns and pronun-
ciation, and as they began acquiring more English, I understood them
most of the time.
The following data samples are brief excerpts from interviews that
took place in March, August and December, respectively. The
purpose of these excerpts is to provide an overall illustration of the
learners' grammar and vocabulary in order to provide a fuller
depiction of the learners' language abilities at each data set time. The
selections from each data set were also chosen to provide further
insight into the different personalities of the learners, as well as their
different viewpoints and approaches to life in the United States. This
transcription is not phonetic, as the goal is to indicate their
communication styles and linguistic competence in terms of grammar
and vocabulary. In this transcription the following symbols are used:
... means a shorter pauses of a few seconds while // indicates a break
in communication.
1. Data Set One: 21 March 1999
Jette: What did you do yesterday?
Anh: I. . . working... yes. . . and umm. . . I was no. . . eh. . . little
customer...
Nhi: Few. . . f ew. . .
Anh: Few customer... they are. . . um. . . spring break. . . and
because I. . . I go home early and then and visit visit ehh... a friend
... together in class... yeah. . .
Jette: A classmate?
Anh: Yeah... yes... yes... this is the Kurnia.
Jette: And what did you do Nhi?
Nhi: I got clean up around house // grass.
Jette: You cut the grass?
Nhi: Cut... [laughter].
Jette: It looks very nice. What else?
Nhi: Ehh... visit my... I ehh... spend... party... I eh... I stay no
I spend... no I go to I went to party... my eh. . . my friend...
birthday party... birthday... eh... birthday party... eh... he party
... eh... he party... party yeah... party yeah... he son... birthday
his son...
Comment:
While this sample is relatively short in word count, the actual time
length was several minutes. As the data indicates, both Anh and Nhi
struggled with vocabulary, and there were many pauses and
hesitancies while they searched for the appropriate words (more so
for Nhi than Anh) and grammatical structures. At this point, it
appears that Anh had an easier time communicating than Nhi while
Nhi was a perfectionist with grammar, correcting both Anh and
himself and repeating his message over and over until he was satisfied
with the grammatical accuracy. In fact, Ann's grammar was actually
better than Nhi's at this point, though Nhi seemed to be monitoring
his (and Ann's) speech more carefully than Anh, judging by his
consistently self- and other-corrections.
2. Data Set Two: 22 August 1999
Jette: So tell me about your week Anh. What did you do?
Anh: With Kurnia... in the phone with Kurnia // Last Monday I off I
off. . . I... went to dental... Tuesday... 111 work I work after seven
PMI met with my friend Kurnia and Yan Ma... we are we are we are
... going drink coffee in downtown... yes... after... we are we are
visit... a Yan Ma house... listen listening her practice //
Jette: Music?
Anh: Yeah... this last week... I . . . I . . . I am watch I watch the
movie film TV in English In and Out... story a man teacher is gay...
funny... yeah funny yeah...
Jette: And Nhi... how was your week?
Nhi: Friday and Saturday I talked with my niece... niece... and //
Jette: What did you talk about?
Nhi: Ehhh... watching TV... and... I watch TV and I asked they I
asked they... I asked you... I asked them... tell me. . . tell us. . .
tell me. . . any idea fr om. . . from film yeah. . . yeah... and. . . I
watch... and Saturday... Saturday yesterday yesterday I watched...
videotape titled Virus II
Jette: And what will you do today?
Anh: Mmm. . . today 11 off yeah...
Jette: So what will you do?
Anh: I ehh... I going to church to church and... and visit my friend
... yeah... because this summer little customer...
Nhi: A few a few customer no little
Anh: A few
Jette: What about you Nhi? What will you do?
Nhi: I and my wife visit will visit... my. . . our friend... yeah... and
maybe maybe... maybe tonight... maybe tonight... we are we are
... we will we will eat Vietnamese restaurant together//
Comment:
At this time, both the Nguyens had a much easier time commu-
nicating their ideas, and were able to speak without as many
hesitations and pauses. Overall, Nhi had greater grammatical accur-
acy than Anh, and he was also much more aware of his (and Ann's)
grammatical errors, as indicated by his intensive self-monitoring and
correction. Anh had more difficulties with verb tense, and commu-
nicated more in single words (nouns, main verbs) than in complete
sentences, unlike Nhi, who was able to communicate in longer
phrases and clauses.
3. Data Set Three: 19 December 1999
Jette: Is American life stressful compared to Vietnamese life?
Anh: Before we are move here... to my brother... visit my family in
... in Vietnam... he told me. . . American life is very stress very
stress... and very hard very hard... and and... we are... prepare
the new new life in here... when when when... we are move here...
I am worry everything... because I don't hear don't speak and... I
don't we are no j ob. . . but... he... my husband has a j ob... my
daughter have a job and my daughter... begin confident in the...
her school she... I study nail technician... when I... graduate nail
technician I confident... 111 feel better I have a job then I... don't
stress yeah.
Jette: What do you think Nhi?
Nhi: Yeah I like I l ike... I like the live here
Jette: Why?
Nhi: Because... in Vietnam... I al ways... I always conf use...
confuse about communism politics politics yeah always about politics
yeah... I I teach... I I taught... always controlled always was
controlled... and... don't have... free time t o. . . rel ax... and in
here... when I have job... I take money... small amount of money I
f eel ... I f eel ... my life in here... enough... don't worry... and...
everybody around in Tucson... always friendly yeah... and always
help me... I study... when I want to study what... I can... I can do
that... for future yeah... so I ... I am happy.
Comment:
While both Anh and Nhi have improved since August, it is still
clear that Anh's grammatical accuracy was much lower than that of
Nhi's. She still had a fairly restricted language base in terms of
grammatical structures, and while she was beginning to use more
function words, she still tended to communicate in words or short
phrases, rather than in complete sentences. She still had difficulty
with verbs, both tense and aspect. Nhi, on the other hand, spoke in
fairly grammatical sentences, though his speech was still slow and full
of hesitations. His grammar was quite good and much more complex
than Anh's.
Notes
1 The term 'absence' rather than 'deletion' is used in this study as
the term 'deletion' implies that a segment may in fact exist in the
learner's underlying grammar and has been deleted at the surface
level. However., whether this is actually the case is not always clear.
Therefore, the more ambiguous term 'absence' is used as it does
not imply that the segment exists in the learner's grammar.
2 Following standard IPA transcription practice, slant / / brackets
will be used to denote the underlying phoneme or phonemic
transcription while square [ ] brackets will be used to denote the
actual phonetic (physical) articulation of a sound/word.
Acquiring a Non-native
Syllable Structure
In this chapter, the linguistic analysis - both descriptive and
inferential - of the data will be discussed. The chapter will be
organized as follows: first, the acquisition of onsets will be discussed,
followed by a discussion of the acquisition of codas. Within each of
these sections, the production of all the onsets (or codas) by
participant, length and time will first be discussed, followed by a
discussion of production accuracies of each onset/coda structure.
Finally, a discussion of production type and modifications of each
onset/coda will be discussed. The chapter will conclude with a
synthesis of the findings for both types of syllable margins in order to
determine possible developmental sequences and patterns.
4. 1 Onsets
4. 1. 1 Production of all onsets by length and participant
There were a total of 6, 673 onsets in the interview data. There were a
total of 5, 877 singleton onsets, 785 CC onsets, and 33 CCC onsets.
The onsets were analysed by production, length, time and partici-
pant, as is illustrated in Table 4. 1 below. As indicated in Table 4. 1,
production accuracy was consistent across time, length and partici-
pant: C onsets were produced more target-like than CC onsets, which
were in turn produced more target-like than CCC onsets, with the
resulting accuracy hierarchy of C> CC> CCC. At time 1, onsets
were produced at 91 per cent accuracy by Anh and 90 per cent by
Nhi; at time 2, this was 93 per cent for Anh and 94 per cent for Nhi
and at time 3, 89 per cent for both Anh and Nhi. Overall, therefore,
singleton onsets could be said to be acquired as they reached the 80
per cent criterion.
For CC onsets, target-like percentiles were lower: At time 1, Anh
produced CC onsets at 69 per cent accuracy and Nhi at 60 per cent;
4
Table 4. 1 Onset production by length, participant and time
Time:
Participant:
Time 1
Ann Nhi Both
Time 2
Anh Nhi Both
Time 3
Anh Nhi Both
C = 5855
Target-
like
Absent
Epenthesis
Feature
Change
Total
Target-
like
Absent
Epenthesis
Feature
Change
Total
790
91%
11
1%
0
0%
68
8%
869
70
69%
28
27%
0
0%
4
4%
102
720
90%
10
1%
0
0%
73
9%
803
58
60%
30
31%
8
8%
1
1%
97
1510
90%
21
1%
0
0%
141
8%
1672
CC
128
64%
58
29%
8
4%
5
3%
199
837
93%
0
0%
0
0%
64
7%
901
= 785
58
59%
16
16%
3
3%
21
21%
98
580
94%
3
0%
0
0%
35
6%
618
44
56%
20
26%
0
0%
14
18%
78
1417
93%
3
0%
0
0%
99
7%
1519
102
58%
36
20%
3
2%
35
20%
176
1407
89%
6
0%
0
0%
165
11%
1578
162
63%
74
29%
7
3%
15
6%
258
968
89%
0
0%
0
0%
118
11%
1086
115
76%
19
13%
4
3%
14
9%
152
2375
89%
0
0%
0
0%
283
11%
2664
277
68%
93
23%
11
3%
29
7%
410
CCC = 33
Target-
like
Absent
Epenthesis
Feature
Change
Total
0
0%
11
100%
0
0%
0
0%
11
0
0%
5
100%
0
0%
0
0%
5
0
0%
16
100%
0
0%
0
0%
16
3
60%
2
40%
0
0%
0
0%
5
0
0%
0
0%
0
0%
0
0%
0
3
60%
2
40%
0
0%
0
0%
5
8
89%
1
11%
0
0%
0
0%
9
2
67%
1
33%
0
0%
0
0%
3
10
83%
2
17%
0
0%
0
0%
12
time 2 these numbers were 59 per cent for Anh and 56 per cent for
Nhi and at time 3, 63 per cent for Anh and 76 per cent for Nhi. For
both participants, CCC onsets were produced at 0 per cent accuracy
at time 1, with 60 per cent accuracy at time 2 for Anh and 0 per cent
for Nhi, and 89 per cent accuracy at time 3 for Anh and 67 per cent
for Nhi. Therefore, overall, both CC and CCC onsets could be said
not to be acquired (CCC at time 3 for Anh is based on only 11
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE 61
continued
Time:
Participant: Anh
Time 1
Nhi Both
Time 2
Anh Nhi Both
Time 3
Anh Nhi Both
ALL = 6673
Target-
like
Absent
Epenthesis
Feature
Change
Total
860
88%
50
5%
0
0%
72
7%
982
778
86%
45
5%
8
0%
74
8%
905
1638
87%
95
5%
8
0%
146
8%
1887
898
89%
18
2%
3
0%
85
8%
1004
624
90%
23
3%
0
0%
49
7%
696
1522
90%
41
2%
3
0%
134
8%
1700
1577
86%
81
4%
7
0%
180
10%
1845
1085
87%
20
2%
4
0%
132
11%
1241
2662
86%
101
3%
11
0%
312
10%
3086
tokens, which is not large enough to make any claims about CCC
onsets in general).
In terms of other types of production as well as absence, for C
onsets, for both Anh and Nhi, feature change was the most common
production type after target-like production - there was very little
absence (0-1 per cent) and no epenthesis. For CC onsets, absence
was common, between 13 per cent and 31 per cent, followed by
feature change, from 1-21 per cent. Finally, there were a few cases of
epenthesis for CC codas, from 0-8 per cent. For CCC onsets,
absence was the most common, ranging from 11 per cent to 100 per
cent. There were no cases of feature change, epenthesis, or two types
of modifications. Therefore, it appears that type of production or
lack therefore is conditioned by length of coda, though not by
participant.
A VARBRUL analysis, outlined in Table 4. 2, was conducted to test
whether the differences in production by time, participant and length
were significant. In this analysis, type of production or absence was
the dependent variable. A VARBRUL analysis was run independently
for each type of production - target-like, epenthesis and feature
change - as well as absence. A number of independent variables,
linguistic and extralinguistic factors, were tested to determine
whether they affected production. These factors were: preceding
linguistic environment, following linguistic environment, stress, time
and participant (see Chapter 3 for a more detailed discussion of these
factors). Several VARBRUL runs were conducted for each produc-
tion type to find the best goodness-of-fit; while the factor groups
found not to be significant were eliminated from further statistical
Table 4. 2 VARBRUL analysis of syllable onsets
Factor Target-like
Group Production
n = 5822
87%
Absence
n = 237
4%
Epenthesis
n = 22
0%
Feature
Change
n = 592
9%
Total
N = 6673
1. Preceding linguistic environment
Pause
P**
#
%
Vowel
P
{
#
%
Consonant
P*
#
%
. 48tns**
1608
28%
. 52ns
2932
50%
. 48|ns
1282
22%
. 47tns
66
28%
. 47tns
103
50%
. 58ns
68
22%
ko
0
0%
. 49
14
50%
. 53
8
28%
. 53fns
203
28%
. 46ns
261
50%
. 53|ns
128
22%
1877
28%
3310
50%
1486
22%
2. Following linguistic environment
Vowel
$
#
%
Diphthong
P*
#
%
/D/
$
#
%
3. Participant
Anh
P*
#
%
Nhi
P*
#
%
4. Length
C
P
f
#
%
. 46
4896
85%
. 68
514
8%
. 76
412
6%
. 50ns
3335
57%
. 51ns
2487
43%
. 48fns
5302
88%
. 53tns
214
85%
. 53tns
19
8%
. 36ns
4
6%
. 53ns
149
57%
. 46ns
88
43%
. 37
30
12%
ko
22
100%
ko
0
0%
ko
0
0%
. 56
12
55%
. 45
10
45%
ko
0
0%
. 55
558
85%
. 24
18
8%
. 27
16
6%
. 50ns
337
57%
. 50ns
255
43%
. 51ns
523
88%
5690
85%
551
8%
432
6%
3833
57%
2840
43%
5855
88%
continued
Factor
Group
CC
/
#
%
ccc
y
#
%
5. Stress
Stressed
f
#
%
Unstressed
P'
#
%
6. Time
Tz Vw e /
/> ''
#
%
Tz Vw e 2
1*
#
%
Time 3
P'
#
%
Input Weight
Total
X
2
X
2
Cell
df***
Significance
Target-like
Production
n = 5822
87%
. 55ns
507
12%
. 48-fns
13
0%
. 47
4117
72%
. 59
1705
28%
. 48tns
1638
28%
. 55ns
1522
25%
. 48tns
2662
46%
. 006
8. 3921
1. 32986
3
p <. 05
Absence
n = 237
4%
. 98|
187
88%
. 98f
20
0%
. 44ns
168
72%
. 64ns
69
28%
. 64
95
28%
. 43
41
25%
. 45
101
46%
. 008
7. 4049
1. 2241
3
p <. 06
Epenthesis
n = 22
0%
ko
22
100%
ko
0
0%
. 46
13
59%
. 62
9
41%
. 56f
8
36%
. 34
3
14%
. 56f
11
50%
. 004
28. 6882
1. 1953
5
p <. 001
Feature
Change
n = 592
9%
. 47ns
69
12%
ko
0
0%
. 55
482
72%
. 37
110
28%
. 46|ns
146
28%
. 46fns
134
25%
. 54ns
312
46%
. 082
5. 8965
. 9828
3
p <. 25
Total
N = 6673
785
12%
33
0%
4780
72%
1893
28%
1887
28%
1700
25%
3086
46%
*input probability **ns = not significant ***degrees of freedom w ere calculated
by subtracting the number of factor groups in the best statistical model from the
number of factors w ithin these groups, f These factors w ere collapsed and run as
one factor due to similar input probabilities in earlier runs.
analysis, their numbers, percentages and input probabilities are
included in Table 4. 2 for comparison.
1. Target-like production: A total of 5, 822 of the onsets, or 87 per cent,
were produced target-like. Two factor groups, following linguistic
environment and stress, were found to have a significant effect on
target-like production (%2 = 8. 3921, 3 df, p <. 05). A following h/ (p{
. 76) and a following diphthong (pl. 68) had a more favourable effect
on production than a following vowel (pl. 46). An unstressed syllable
promoted target-like production slightly at (pl. 69) in comparison to a
stressed syllable (pl. 47).
2. Absence: A total of 237 onsets had absence. Two factor groups,
length and time, had a significant effect on absence (%
2
= 7. 40498,
df, p <. 06). A singleton onset, C, disfavoured absence ( p
l
. 37) in
contrast to CC and CCC onsets, which both strongly favoured
absence (p1. 98). Time 1 favoured absence (p1. 64) more than time 2
(p*. 43) or time 3 (p*. 45).
3. Epenthesis: There were only 22 cases of epenthesis in the data.
VARBRUL statistics were run on the data to indicate general trends
although they should be interpreted with caution due to the small
number of tokens in this data set. Four factor groups were found to
affect epenthesis: preceding linguistic environment, participant, stress
and time (%2 = 28. 6882, 5 df, p <. 001). While there were no cases of
a preceding pause before epenthesis, a preceding consonant favoured
epenthesis (p1. 53) over a preceding vowel (p1. 49). Anh favoured
epenthesis (p1. 56) slightly over Nhi (p1. 45) and unstressed syllables
(p1. 62) favoured epenthesis over stressed syllables (p1. 46). Finally,
time 1 and 3 favoured epenthesis (p1. 56) over time 2 (p1. 34).
4. Feature change: There were 592 cases of feature change in the
data. Two factor groups, following linguistic environment and stress,
had an effect on feature change (%
2
= 5. 8965, 3 df, p <. 25). A
following vowel had a favourable effect on feature change (p1. 55) in
contrast to a following hi (p1. 27) or diphthong (p1. 24). A stressed
syllable favoured feature change (p1. 55) over an unstressed syllable
(P * -37).
In summary, every factor group was significant for one or more
types of production or absence. The factor group that had the most
effect on production was stress, which affected target-like production,
epenthesis and feature change. Time affected both absence and
epenthesis and a following linguistic environment affected target-like
production and feature change. Length only affected absence, and a
preceding linguistic environment and participant affected only
epenthesis.
4. 1. 2 Production accuracies of individual onsets
Onsets were also analysed individually in order to examine which
onsets were more likely to be produced accurately as a means of
determining any acquisition orders or patterns. These patterns are
outlined in Table 4. 3 below.
As Table 4. 3 indicates, the most accurate onsets, with accuracy
percentages ranging between 75-100 per cent across time for both
participants, were the nasals /m n/, the voiced stops /b d g/, the
voiceless stops /t k/, the voiceless fricatives /s f h/, the voiced fricative
/v/, the glides /w y/ and liquids /i I/. A number of CC and CCC onsets
were also produced with an overall high level of accuracy: /st si sk sp
vj tw py dy kw fl fy ny sm skw/ - all of them comprising single onsets
that were produced with a higher level of accuracy (with the exception
of/p/ for Nhi), e. g. /t k d s f v n m y w 11/. Onsets produced with a mid
level of accuracy were the voiceless stop /p/, the voiceless fricative /J/,
voiceless affricate /tj/ and voiced fricative /z/. A number of CC onsets
were also produced with a mid level or fluctuating level of accuracy:
/di bi ti gj pj ki bl kl ft pi by vy hy sw sti/. The least accurately
produced onsets were the voiced affricate /ds/ and the interdental
fricatives /5 0/ and the CC and CCC onsets /6j dsy spl spj/. Not
surprisingly, the CCC onsets that were produced with some degree of
accuracy (/skw/ and /sti/) comprised CC onsets that had higher
accuracy levels, /sk kw st ti/. The CC and CCC onsets with the lowest
accuracy percentiles comprised the C and CC onsets that also had the
lowest accuracy percentiles, especially with /0 3/.
Overall, while most of the singleton onsets, especially the nasals,
stops, glides, liquids, and /v/ as well as /s f h/ were produced
accurately, the participants did have difficulty with the affricates,
interdental fricatives and alveo-palatal fricatives. In terms of the types
of CC and CCC onsets, a number of patterns can also be found (the
following notation will be used: L = liquid, S = stop, N = nasal,
F = fricative, and # = syllable margin so that #FL stands for a
fricative-liquid onset. Note: affricates are classified as fricatives in this
analysis): the most accurately produced CC and CCC onsets were
#FS and #FL, particularly with III as the liquid, #SG, and the CCC
Table 4. 3 Accuracy Percentiles by Onset, Time, and Participant
Time:
Participant:
100%
75 - 99%
50 - 74%
Time 1
Anh Nhi
d, k, g,
m, n, h,
f, tj, w,
j, y, tw,
fy, si, sk,
sm, vi
p, b, t, s,
z, v, f, 1,
st, tr, sp
gJ
b, k, m,
y, 1, hi, J 7 7 "5
kw, pj,
gi, sm,
py, st
t, d, g, n,
f, v, s, J,
tj, h, w,
j, vy, sp
bl
Time 2
Anh Nhi
t, d, b, k, t, b, k, g,
g, m, n, m, f, v,
h, v, f, z, d3, w, i,
i, sm, sk, sp, kw,
ti, ki, fl, di, vi
hy, by,
fy, sw,
kw, tw,
VJ
p, s, tf, d, n, s, h,
w, y, 1, 1, y, st
st, sp
di, pi, sti p, z, kl,
sm
Time 3
Anh Nhi
b, g, m,
f, v, z,
w, y, ny,
kw, fy,
dy, si,
di, sk,
skw
t, d, p, k,
n, s, tf,
h, i, 1, st,
sp, tw,
ti, sti
f, sw, gj,
bi, pi, vi
t, d, g, n,
m, h, v,
1, y, w,
gj, st, ti,
tw, si,
sk, skw
b, k, s, f,
tf, J, fy,
sp, pi, vi
p, fc
25 - 49% di, fi z, dj, di, f, vy, fi 5, 61, pi f, z, ds,
tw, fi, kl kl
1 - 24% ds, 9, kl p, ti, pi ds, pi f 6, dj, pi, 8, di
fi
0% a, bi, 0i, e, a, sw, a, e, ei tf, pi, hy, a, ki, 0i, a, ea, pi,
pi, by, 61, spl, ki, spi sm, sti
dsy, ki, sti
pi, bi,
spl, sti
onset #FSG, which comprises #FS and #SG, both of which are
highly accurately produced. The least accurately produced CC and
CCC onsets were #SL, #FL with III as the second member, #FG,
and the CCC onsets #FSL, which comprises #SL, which is also
difficult to produce.
A VARBRUL analysis was also conducted to determine whether
accuracy percentiles were statistically significant for each onset. As
there were very few tokens of individual CC and CCC onsets
produced accurately, only singleton onsets were included in this
analysis. There were no tokens of/3/. The onset /m/ was not included
in the VARBRUL run as it was produced with 100 per cent accuracy,
and therefore categorically. The factor groups employed for the
previous VARBRUL analysis as well as type of onset was included in
the analysis. These results are outlined in Table 4. 4, below. Only type
of onsets as well as following linguistic environment were significant,
at (x
2
= 60. 000, 22 df, p <. 001).
Based on the VARBRUL results, it is possible to establish a
continuum of accurate production (the types of onsets are compared
against each other) from most to least: /m/ (100 per cent accurate)
> /b/ (p*. 87) > /k/ (p*. 79) > /n/ (p'. 78) > /y/ (p*. 77) > III (p*. 74)
> /d/, /w/ (p*. 69) > Kl (p1. 67) > /g/ (p*. 61) > hrl (p*. 58) > /t/, /h/
(p*. 57) > III (p1. 51) > /s/ (p*. 41) > /t|7 (p*. 13) >/z/ (p*. 09) > /p/
(p*. 08) > /// (p*. 04) > /d3/ (p*. 01) > /9/ (p*. 002) > /3/ (p'. 001).
The onsets that were more likely to be produced accurately were the
nasals, glides, liquids, voiced stops, the fricatives /s f v/, and /k/ while
the onsets least likely to be produced accurately were the affricates,
dental fricatives and alveo-palatal fricative.
4. 1. 3 Production type of individual onsets
In order to determine whether the production types of individual
onsets were patterned as well, individual analyses of each onset by
type and length were conducted.
C:
Voiceless stops: Both /t k/ were produced target-like most of the time.
/t/ underwent a few word-specific changes: feature change (/t/ to [s] in
sometimes} or absence in customer (for both Anh and Nhi). /k/ was
almost always correct; there was one instance of absence (in customer)
and feature change (to [1] in unlucky and to [t] for the second /k/ in
communicate - both for Nhi). /p/ had a lower accuracy percentage,
particularly for Nhi - in each case, this was a feature change,
particularly lack of aspiration when it was a stressed syllable initial
singleton onset.
Voiced stops: Voiced stops had a high accuracy percentile. For /b/,
there was one case of devoicing and one of absence; for /g/ two cases of
absence (in register), and for /d/, a few cases of absence in understand.
Nasals: Iml was produced target-like 100 per cent of the time, and /n/
also had a high accuracy percentage. There were a few cases of feature
Table 4. 4 VARBRUL analysis of target-like production by singleton onset
Accurate Number Total Number Percentage Probability
/ P /
M
N
W
N
/ g /
I f f
M
1 * 1
/ z/
I I I
/ e /
/ s /
A J 7
w
N
M
M
N
/ V
/ m/
/ n/
119
382
429
351
417
100
216
213
323
24
7 7
7
4
98
17
386
386
313
167
260
495
518
158
384
440
356
420
102
219
218
338
32
141
104
200
120
65
396
391
316
169
268
495
523
7 5
99
98
99
99
98
99
98
96
7 5
55
7
2
82
26
97
99
99
99
97
100
99
. 09
. 87
. 57
. 69
. 7 9
. 61
. 67
. 58
. 41
. 09
. 04
. 002
. 001
. 13
. 01
. 57
. 69
. 7 7
. 7 4
. 51
1. 00
. 7 8
Following linguistic environment
Vowel 4074 4606 8 8 . 4 9
Diphthong 3 3 5 3 4 8 9 6 . 3 8
/ o / 3 9 8 4 0 6 9 8 . 7 1
Input Weight . 97
Total ft2 60. 000
X
2
Cell 1. 1321
df 22
Significance p <. 001
change (to [1] in no, [m] in the second syllable in Vietnamese and [rj] i
knowledge.
Voiceless fricatives: HI was 100 per cent target-like for Anh while Nhi
had a few cases of feature change (to [p] in feel and physics and [h] in
family), /s/ had a number of instances of feature change for both Anh
and Nhi: to [f] in Mexican, sometime, since and Saturday and to [t] in
nieces. /JV had frequent cases of feature change, all to [s], in words
such as she, Russian, shop and sugar, /h/ had a high percentage of
target-like production, but a few cases of absence, such as in he and
have (after a pause), possibly an overgeneralization of /h/ deletion in
auxiliary verbs and pronouns in casual speech. The interdental
fricative /0/ was rarely produced accurately; it usually underwent
feature change to [s] or [t], most usually to [s] in the word think and
to [t] in other cases. The voiceless affricate /tJY also had frequent
feature changes, in this case to [f] in children, [t] in structure, capture
and nature and to [ds] in future.
Voiced fricatives: /z/ was produced as [f] in disease and Joseph and had
absence in lazy, /v/ was usually produced accurately, with a few
changes to [w] (very) and [f] (Voltaire). The voiced dental fricative /5/
was produced target-like in only a very few cases; it was most
commonly feature changed to [d]. The four cases of target-like
production were in the word father (there were no cases of inaccurate
production of this word in the data). The voiced affricate /dj/ was
rarely accurately produced; it most commonly underwent feature
change, in all cases devoicing to [tf].
Liquids: III was usually produced accurately, with a few cases of
feature change (to [w] in writing}. Ill had a few cases of absence (in
only) and feature change to [g], especially after /n/.
Glides: /w/ was almost always produced target-like, with a few cases of
feature change to [v] (Washington, will), lyl was also target-like most
of the time, with a few cases of feature change to [ds] in year (after /n/
only).
CC:
Fricative-Stop: (/sp st sk/). These onsets were typically produced
target-like (/sk/ was 100 per cent target-like) with a few cases of
absence (/st/ to [s], /sp/ to [s] or [p]). /sp/ had a few instances of
epenthesis, only in the word speak.
Fricative-Nasal: (/sm/). This was usually produced target-like, except
for the word smoke, which Nhi produced with feature change to [fm].
Fricative-Glide: (/fy vy dsy hy sw/). /fy/ was usually produced
accurately, with one case of absence to [f]. /vy/ had high target-like
production but was epenthesized in interview and view 3 times. There
was only one instance of /d3y/ and it was produced as [y]. /hy/ was
produced as [h] once and target-like once, /sw/ was usually target-like
with a few reductions to [s] in sweatshirt and sweets.
Fricative-Liquid: (/fi 61 si fl/). /fl/ was alternatively produced target-like
and produced as [fj. /6i/ was only accurately produced once; in all
other cases, it was feature changed to [ti]. /si fl/ were both 100 per
cent target-like.
Stop-Liquid: (/pi ti ki hi di gj pi kl bl/). /ki bi di gi/ had a few instances
of target-like production but were usually reduced to /i/. /di/ also had
a few cases of feature change to [ti]. /pi/ also had a few reductions to
[j] but was also often featured changed to [hi] word-initially (priest,
pregnant, present) after a vowel. /ti/ had a higher level of target-like
production than the other stop-liquid onsets with /i/, but also had a
number of reductions to [i]. Overall, stop-liquid clusters with III were
produced more target-like than those with /i/, with /pi/ sometimes
undergoing feature change to [bl] (after a nasal), /bl/ often reduced to
[1], and while /kl/ was often produced target-like, there were also a
number of reductions to [1] or feature changes to [gl].
Stop-Glide: (/py by dy tw kw/). There were only two cases of/by/ in the
data: one was produced target-like and the other produced as [by].
The few cases of/py/ (2) and /dy/ (1) in the data were target-like, /tw/
had a high degree of target-like production but was produced as [t] in
a few cases; /kw/ was 100 per cent target-like.
Nasal-Glide: (/ny/). There was only one case of/ny/ in the data and it
was produced target-like.
CCC:
Fricative-Stop-Glide: (/skw/). /skw/ was produced 100 per cent target-
like.
Fricative-Stop-Liquid: (/sti spi spl/). Both /spi/ and /spl/ were produced
with absence, with /spi/ produced as [sp] and /spl/ as [pi], /sti/ had
some target-like production, but was also often reduced to [s] (Nhi)
or [ti st] (Anh).
In summary, the analysis of the production of C, CC and CCC
onsets indicates that production type differs by onset type and length.
While C onsets were typically produced target-like, some of the more
difficult onsets were produced with feature change. Epenthesis did
not occur and absence was rare. For CC and CCC onsets, patterns
existed by type of onsets. #FS onsets were usually target-like, with
some production to #F, with the stop absent. #FN were usually
target-like as were #FG, with some reduction to the #F and absence
of the glide. #FL had different patterns depending on whether the
liquid was III or 1 1 1 . If it was /!/, the cluster was usually target-like, but
with 1 1 1 , was often produced as #F, with the liquid absent. #SL had
the same pattern - with /!/, they were usually target-like with a few
productions as III and loss of the stop; with lit, the stop was usually
absent and it was produced as #L. #NG and #SG were usually
target-like, with #SG produced as #S in a few cases. For CCC codas,
#FSG was target-like, and #FSL was produced differently based on
the onset: /sti/ was produced as #F, /spl/ as #SL and /spi/ as #FS.
4. 2 Codas
4. 2. 1 Production of all codas by length and participant
There were a total of 4, 400 codas in the interview data, with a total of
3, 703 singleton codas, 607 CC codas, and 130 CCC codas. There
were no CCCC codas in the interview data. Table 4. 5, below,
illustrates coda production individually and in aggregate form across
the three coda lengths, time and production type.
There was a consistent pattern of accuracy of production by length
for both participants and across all three data sets, in that singleton
codas were produced more target-like than CC codas and CC more
target-like than CCC codas, with the resulting hierarchy of C > CC
> CCC. At time 1, C codas were produced by Anh with 45 per cent
accuracy, and by Nhi with 53 per cent accuracy. At time 2, the
accuracy percentages were 40 per cent for Anh and 48 per cent for
Nhi. Finally, at time 3, the accuracy percentages were 49 per cent for
Anh and 47 per cent for Nhi. For the CC codas, accuracy was very
low across time for both participants: For Anh, it was 5 per cent at
time 1, 4 per cent at time 2 and 2 per cent at time 3; while for Nhi it
was 9 per cent at time 1, 7 per cent at time 2 and 11 per cent at time
3. Finally, both participants had a 0 per cent accuracy percentage
across time for CCC codas.
In terms of other types of production for the codas, for C codas,
absence was the most common after target-like production, ranging
Table 4. 5 Production of codas by length, time and participant
Time:
Participant:
Time 1
Anh Nhi Both
Time 2
Anh Nhi Both
Time 3
Anh Nhi Both
C = 3703
Target-
like
Absent
Epenthesis
Feature
Change
2+ Types
Total
281
45%
173
28%
57
9%
103
16%
12
2%
626
276
53%
117
22%
34
9%
79
15%
19
4%
525
557
48%
290
25%
91
8%
182
16%
31
3%
1151
232
40%
194
34%
49
8%
92
16%
12
2%
579
219
48%
121
26%
45
10%
61
13%
14
3%
460
451
43%
315
30%
94
9%
153
15%
16
2%
1039
369
49%
160
21%
65
9%
161
21%
5
0%
760
356
47%
125
17%
98
13%
131
17%
43
6%
753
725
48%
285
19%
163
11%
292
19%
48
3%
1513
CC = 607
Target-
like
Absent
Epenthesis
Feature
Change
2+ Types
Total
6
5%
89
76%
3
3%
1
1%
18
15%
117
11
9%
89
71%
0
0%
3
2%
23
18%
126
17
7%
178
73%
3
1%
4
2%
41
17%
243
3
4%
54
79%
3
4%
2
3%
18
25%
73
5
7%
47
64%
0
0%
0
0%
9
13%
68
8
6%
101
72%
3
2%
2
1%
27
19%
141
3
2%
104
87%
2
2%
2
2%
10
8%
121
11
11%
59
58%
2
2%
6
6%
24
24%
102
14
6%
163
73%
4
2%
8
4%
34
15%
223
CCC = 130
Target-
like
Absent
Epenthesis
Feature
Change
2+ Types
Total
0
0%
22
100%
0
0%
0
0%
0
0%
22
0
0%
19
95%
0
0%
0
0%
1
5%
20
0
0%
41
98%
0
0%
0
0%
1
2%
42
0
0%
16
76%
0
0%
0
0%
5
24%
21
0
0%
19
61%
0
0%
1
3%
11
35%
31
0
0%
35
67%
0
0%
1
2%
16
31%
52
0
0%
5
83%
0
0%
0
0%
1
17%
6
0
0%
29
97%
1
3%
0
0%
0
0%
30
0
0%
34
94%
1
3%
0%
1
3%
36
continued
Time:
Participant:
Time 1
Anh Nhi Both
Time 2
Anh Nhi Both Anh
Time 3
Nhi Both
ALL = 4440
Target-
like
Absent
Epenthesis
Feature
Change
2+ Types
Total
287
38%
284
37%
60
8%
104
14%
30
4%
765
287
43%
225
34%
34
5%
82
12%
43
6%
671
574
40%
509
35%
94
7%
186
13%
73
5%
1436
235
35%
257
38%
52
8%
94
14%
35
5%
673
235
40%
194
35%
45
8%
62
11%
35
5%
559
459
37%
451
37%
97
8%
156
13%
69
6%
1232
372
42%
269
30%
67
8%
163
18%
16
2%
877
367
41%
213
24%
101
11%
137
15%
67
8%
885
739
42%
482
27%
168
9%
300
17%
83
5%
1772
from 17 per cent to 34 per cent. This was followed by feature change,
which ranged from 13-21 per cent, and epenthesis, from 8-13 per
cent. There were also a few cases of two or more types of
modifications, from 0-6 per cent. For CC codas, absence was the
most common, ranging from 58-87 per cent, followed by two or more
types of modifications, ranging from 8-24 per cent. Both feature
change and epenthesis occurred as well, though not as often, with
feature change ranging from 0-6 per cent and epenthesis from 0-4
per cent. For CCC codas, absence was the most common, but there
were a few cases of two or more types of modifications, ranging from
0-35 per cent. There were very few cases of epenthesis and feature
change for CCC codas.
The differences in production by time, participant and syllable
coda length were tested for statistical significance using a VARBRUL
analysis. The VARBRUL analysis was also set up to determine
whether other linguistic factors affected production (e. g. whether
linguistic environment affected rate of epenthesis or feature change)
by length, participant and time (see Chapter 3 for a more detailed
discussion of these factors). The analysis also tested whether stress
(e. g. an unstressed vs. stressed syllable), grammatical conditioning
(monomorphemic vs. bimorphemic), and following linguistic envir-
onment (pause, consonant, vowel) had an effect on production.
Preceding linguistic environment was also tested for the entire coda
structure. Several VARBRUL runs were conducted for each produc-
tion type to find the best goodness-of-fit; while the factor groups
74 ACQUIRING A NON-NATIVE PHONOLOG
Table 4. 6 Variable rule analysis of syllable codas
Factor Target-like
Group Production
n = 1772
41%
Absence Epenthesis
n = 1442 n = 359
32% 8%
Feature 2 Types of Total
Change Production
n = 642 n = 225 N =
14% 5% 4440
1. Preceding linguistic environment
Diphthong
P2*
#
%
Vowel
P
l
#
%
hi
P*
#
%
. 12
23
1%
. 57
1676
95%
. 24
73
4%
. 94
247
17%
. 42
969
67%
. 75
226
16%
. 20
8
2%
. 52
296
82%
. 58
55
15%
0
0%
. 50ns
593
92%
. 47ns
49
8%
. 49ns
16
7%
. 51ns
192
85%
. 38ns
17
8%
294
7%
3726
84%
420
9%
2. Following linguistic environment
Pause
P*
#
%
Consonant
i
#
%
Glide
P*
#
%
Vowel
i
#
%
3. Participant
Nhi
P
l
#
%
Anh
i
#
%
4. Length
C
i
#
%
cc
p*
#
%
. 41ns**
547
31%
. 58ns
912
51%
. 52ns
86
5%
. 50ns
227
13%
. 53
878
50%
. 47
894
50%
. 59
1733
98%
. 09
39
2%
. 44ns
560
39%
. 55ns
601
42%
. 45ns
71
5%
. 57ns
210
15%
. 45
632
44%
. 55
810
56%
. 40
890
62%
. 87
442
31%
. 66
218
61%
. 36
88
25%
. 58
20
6%
. 43
33
9%
. 50ns
180
50%
. 50ns
179
50%
. 59
348
97%
. 16
10
3%
. 54
278
43%
. 43
248
39%
. 55
33
5%
. 49
83
13%
. 48ns
281
44%
. 52ns
361
56%
. 60
627
98%
. 14
14
2%
. 68ns
158
70%
. 38ns
405
20%
. 46ns
8
4%
. 36ns
14
6%
. 59
144
64%
. 42
81
36%
. 42
105
47%
. 84
102
45%
1761
40%
1894
43%
218
5%
567
13%
2115
48%
2325
52%
3703
83%
607
14%
continued
Factor
Group
ccc
p *
#
%
Target-like
Production
n = 1772
41%
0
0%
Absence
n = 1442
32%
. 95
110
7%
5. Grammatical conditioning
Mono-morp hemic
P *
# . 5 0 n s . 4 4 n s n
%
Bi-morp hemic
P *
#
%
6. Stress
Stressed
y
#
%
[7wsres. sec?
P *
#
%
7. Time
Time 1
t?
#
%
Time 2
P f
#
%
Time 3
P
f
#
%
Input Weight
Total ft
2
X
2
Cell
dP**
Significance
1727
97%
. 55ns
45
3%
. 51ns
1311
74%
. 48ns
461
26%
. 52ns
574
32%
. 47ns
459
26%
. 51ns
739
42%
. 37
11. 2648
. 9387
4
p <. 01
1298
90%
. 56ns
144
10%
. 50ns
1149
80%
. 50ns
293
20%
. 54ns
509
35%
. 55ns
451
31%
. 43ns
482
33%
. 30
12. 2906
. 7230
5
p <. 025
Epenthesis
n = 359
8%
. 07
1
0%
. 51ns
353
98%
. 37ns
6
2%
. 62
341
95%
. 18
18
5%
. 47f
94
26%
. 47f
97
27%
. 55
168
47%
. 04
37. 7122
0. 5312
9
p <. 0001
Feature
Change
n = 642
14%
. 06
1
0%
. 51ns
631
98%
. 49ns
11
2%
. 44
362
56%
. 69
280
44%
. 47tns
186
29%
. 47fns
156
24%
. 55ns
300
47%
. 11
24. 5871
1. 0690
6
p <. 0005
2 Ty pes of
Production
n = 225
5%
. 79
18
8%
. 49ns
185
82%
. 63ns
40
18%
. 59ns
196
87%
. 42ns
29
13%
. 49ns
73
32%
. 50ns
69
31%
. 51ns
83
37%
. 03
5. 8358
. 9726
3
p <. 025
Total
N =
4440
130
3%
4194
94%
246
6%
3359
76%
1081
24%
1436
32%
1232
28%
1772
40%
*input probability **ns = not significant ***degrees of freedom were calculated
by subtracting the number of factor groups in the best statistical model from the
number of factors within these groups. jThese factors were collapsed and run as
one factor due to similar input probabilities in earlier runs.
found not to be significant were eliminated from further statistical
analysis, their numbers, percentages and input probabilities are
included in Table 4. 6 for comparison.
/. Target-like production: A total of 1, 772 codas were produced target-
like. Three factor groups, preceding linguistic environment, partici-
pant and coda length, were found to have a significant effect on
target-like production (%
2
= 11. 2648, 4 df, p <. 01). A preceding
vowel (pi. 57) was found to slightly favour target-like production over
a preceding diphthong (pi. 12) and a preceding /o/ (pi. 24). Nhi's
production (pi. 53) slightly favoured target-like production over Anh's
(pi. 47). Finally, a singleton coda (pi. 59) favoured target-like
production over CC codas (pi. 09).
2. Absence: 1, 442 codas had absence. Preceding linguistic environ-
ment, participant and coda length had a significant effect on absence
(%
2
= 12. 2906, 5 df, p <. 025). A preceding diphthong (p*. 94) and hi
(pi. 75) both favoured absence over a preceding vowel (pi. 42). Nhi's
production (pi. 45) slightly disfavoured absence over Anh's (pi. 55).
The longer coda structures both strongly favoured absence with CC
having a. 87 input probability and CCC with an almost categorical
. 95 input probability over the singleton coda (pi. 40).
3. Epenthesis: 359 codas were epenthesized. Five out of the seven
factor groups in the analysis were found to have a significant effect on
epenthesis (%
2
= 37. 7122, 9 df, p <. 0001): preceding linguistic
environment, following linguistic environment, length, stress and
time. A preceding diphthong (pi. 20) was found to have a strong
disfavouring over a preceding vowel (pi. 52) and a preceding /o/ (pi
. 58). A following pause (pi. 66) and a following glide (pi. 58) favoured
epenthesis over a following consonant (pi. 36) and vowel (pi. 43).
Shorter codas also favoured epenthesis over longer codas: Singleton
codas had an input probability of. 59 while CC codas had one of. 16
and CCC codas one of. 07. A stressed syllable (p1. 62) promoted
epenthesis over an unstressed syllable (p
1
. 18). Finally, both time 1
and time 2 slightly disfavoured epenthesis (pi. 47) over time 3 (pi. 55).
4. Feature change: There were 642 cases of feature change in the data.
Three factor groups had a significant effect (X2 = 24. 5871, 6 df, p <
. 0005) on feature change: following linguistic environment, length
and stress. While a following vowel (pi. 49) seemed to have little effect
on feature change in contrast to a following glide (pi. 55) and pause
(pi. 54), which both slightly promoted feature change while a
following consonant (pi. 43) disfavoured feature change. Feature
change was favoured for singleton codas (pi. 60) in contrast to both
CC (pi. 14) and CCC (pi. 06) codas. It was also disfavoured for
stressed syllables (pi. 44) over unstressed syllables (pi. 69).
5. Two production types: 225 codas had two or more types of
production. Two factor groups significantly (X2 = 5. 8358, 3 df, p <
. 05) affected two types of production: participant and length. Nhi's
production (pi. 59) slightly favoured two types of production over
Ann's (pi. 42). The singleton coda (pi. 42) slightly disfavoured it over
both the CC (pi. 84) and CCC (pi. 79) codas.
In summary, only one factor group, grammatical conditioning, was
found not to be significant for any type of production. The factor
group that affected the most production types was coda length, which
affected target-like production, absence, epenthesis, feature change
and two types of production. Preceding linguistic environment was
found to have a significant effect on target-like production, absence
and epenthesis, and following linguistic environment had a significant
effect on epenthesis and feature change. Participant significantly
affected target-like production, absence and two types of production
while stress affected epenthesis and feature change. Time was a
significant factor only for epenthesis.
4. 2. 2 Production accuracies of individual codas
As with the onsets, the production accuracy of codas was also
analysed individually across time and participant. The coda data from
the interviews were also compared against the coda data from the
word list task in order to determine to what extent the participants
could produce the codas under a more controlled task. These are
shown in Table 4. 7. The comparisons were based on the 4, 440
syllable codas from the interview data and 4, 676 syllable codas from
the word list data.
In comparing the codas from the interview data with the codas
from the word list data, it becomes clear that strong similarities exist
between data from the two tasks as well as between participants in
terms of which codas are produced most and least target-like, and
which consonants are members of the CC and CCC codas that are
beginning to emerge. For both tasks, codas that are produced more
target-like contain nasals, voiceless stops and the voiceless fricatives
Time
Participant
Time 1
Anh Nhi
Time 2
Anh Nhi
Time 3
Anh Nhi
Interview Coda Data
100%
75 - 99%
50 - 74%
25 - 49%
1 - 24%
0%
f, jz, jd
P
t, m, n, q, s, nz
k, d, j, 1, ts
v, J, z, 0, Jk
b, nds, ntj, nd, nt, st,
qk, mz, Jt, It, zd, Id, ks,
dz, jst, ndz, nts, its,
nte
p, JZ
m, n, f, nz
k, n, s
t, d, v, j, 1
b, d, z, J, ks, nt, jk
0, ds, ntj, nd, st, qk,
mz, it, zd, dz, jn, bz,
kt, sk, pt, Jd, ts, 5z, jst,
kst, ndz, nts
f, jl
n, m, q, s, Jz
t, k, j, 1
d, J, v, z, ks
b, g, tj, 0, ntj, nd, nt,
st, mz, Jt, ts, nz, jk, JtJ,
vd, sk, J0, Iz, qk, jld,
jst, ndz, nts, kst, jkt,
jlz, nt6, jnd
p, f, nz, jz
m, q
t, m, n s, Iz
k, v, j, 1
J, ks
d, z, tj, 0, ntj, nd, st,
mz, ts, sk, jn, kt, nt, Iv,
zd, Jt, nts, jkt, jnd,
nt0, ndz
P, J
t, n, q
g, m, f, s, ts
k, v
d, z, 1, j, jk, nt
0, ds, tj, ntj, nd, st,
mz, nz, sk, j0, JtJ, jz,
ft, jn, jd, Id, Iv, Jt, ks,
jst, ndz, nts, nt0
p, Id
n, m
t, k, q, f, s, tj, ts
d, v, ks, nz, jn, jz
z, j, jd
b, g, J, ds, 0, If, nd, nt,
st, qk, jk, jt, Ip, pt, J0,
Iv, Iz, nds, jst, ndz,
kst, Idz, jdz, jnd, nts,
nt0
Word List Coda Data
100%
75 - 99%
50 - 74%
25 - 49%
1 - 24%
Iz, n0, ns, mt, ntj, jm6
t, k, m, n, rj
p, d, f, s, 9, Jt, ps, j6, 16
b, z, 1, st, If, sk, 1J, Ip,
t0, jf, bz, nz, zd, an, jz,
pt, qk, nst, jst
b, v, 0, tj, id, mz, nt
m, Ip, Is, Iz, mt, Ik, ItJ,
gz, ps, fs, ns, nz, qz,
ntj, zd, jmz, jkt, nst,
nts, jbz
t, n, q, f, jst
k, s, z, v, tj, 3, j, jl, It,
nt, bz, mz, dz, st, ts,
ks, kts
p, d, 1, jf, mp, vz, jm,
1m, Jt, JZ, kst, Jm0, jpt
b, If, tjt, jnz
mf, ns, Ip, ntj, nst
p, t, k, m, n, qk
q, f, st, 10, 1m, J0, JtJ,
sk
s, 3, j, 1, If, ps, fs, bd,
1J, mt, ItJ, mz, jz, jn,
mps, qst, qks
d, tj, z, v, nz
3, Ip, JtJ, Ik, ps, bz, mt,
st, ns, qz, sk, jmz, nst,
jst, qst
p, t, k, n, m qk
k, q, z, st, jz, Iz, ks, nz,
1J, md, sp, Ip, ft, mf, Is,
It, vz, mp, bz, mz, jkt,
Iks
d, s, v, j, 1, In, J, Jt, st,
jnz, jdz, dst
b, f, If, sks, qks
p, Ip, ps, ItJ, mf, ns,
mt, mp, qd, ntj, n0,
nst, qst
t, n, m, f, qk
k, q, 1, 0, js, J0, 10, JtJ,
nd3, 1J, jb, bz, Is, If, jn,
Ipt
b, d, s, tj, 3, vz, In, mz,
nz, sk, st, ts, jnz, jmz
Z, V, J, tjt
p, m, ft, gz, Ik, ntj, 10,
ItJ, vz, ps, mt, jm, mp,
md, qz, Iv, JtJ, jnz,
jmz, nst, qst
t, n, jf, mz, Iz, qk, jst
k, q, f, s, tj, 1, mf, nz,
sp, bd, sk, bz, Ip, jl,
dz, nd3, st, ks, ts, jn,
sks, mps, Ipt, jkt
b, d, z, v, tjt, zd, ns, Is,
1m, Ib, 1J, kt, jz, Iks,
kst, ntjt
j, If, jd, nt, qks
Table 4. 7 Accuracy Percentiles by Coda, Task, Time, and Participant
Time
Participant
0%
Time 1
Anh
g, 3, 8, ds, J, jt, Ik, qz,
ks, Iv, 8z, Ib, bd, dz,
kt, is, 5d, ip, ItJ Is, vz,
tjt, jj, Id, gz, gd, md,
nd3, ds, im, Jb, In, nd,
dsd, sp, Jg, Jt, 3d, 1m,
fs, 0s, JtJ, jl, vd, f6,
jd3, qd, Id3, d6, ft, ts,
mp, mf, dst, kst, jld
qkt, jnz, qst, jmz, n0s
nipt, mps, ndz, qks,
nts, jpt, J0s, If9, Jkt
Ikt, Ipt, its, jdz, Ik
jbz, jps, Idz, kts, sts,
sks, ntjt, t6s kste, IfG s,
jldz
Nhi
g, 0, 5, d3, J, sk, in, pt,
Ib, qk, id, iQ, t0, 10, kt,
1J, Jt, jp, jj, Id, gd, js,
jg, Iv, JtJ, jb, Jds, Ids
bd, d0, sp, ft, vd, 3d,
6d, f0, 0s, 5z, n0, mf,
md, nds, nd, qd, dsd,
In, jk, dst, jld, qkt,
nipt, qst, mps, n0s,
ndz, qks, J0s, If0, Ikt,
Ipt, Idz, jdz, jts, jp
Iks, sks, sts, t0s, ntjt
kst0, If0s, Jldz
Time 2
Anh
b, g, J, 0, 5, ds, bz, nt,
pt, it, Iv, It, ft, js, Ik,
dz, Is, Iz, ts, tjt, jf, Ib,
ks, jj, jp, Jm, kt, In, Jt,
Id, il, jb, mp, jg, jd,
gz, md, jds, d0, sp,
Ids, t0, zd, 3d, 5d, dsd,
vd, f0, 0s, jk, 5z, vz,
n0, qz, gd, nd3 nd, qd,
dst, kst, jld, qkt, jnz
n0s, jmz, ndz, jm0,
nts, nipt, jlz, jnd, J0s
If0, jst, jpt, Jkt, Ikt
Ipt, jdz, Jts, jps, Ik
Idz, kts, ntjt, sts, sks
t0s, kst0, If0s, jldz
Nhi
g, 0, 5, ds, J, Jt, Iv, 5z,
Ib, bd, dz, kt, js, 5d,
jp, tjt, Id, gz, gd, nd3,
jm, Jb, nd, dsd, 3d, jg,
1m, fs, 0s, jl, vd, f0,
jd3, qd, Ids, d0, jn, pt,
jd, ts, J0, jk, 10, t0, jf,
zd, n0, Itjt, nt, kst, jld,
qkt, jm0, n0s, mpt
mps, ndz, nts, J0s, If0,
jpt, Ikt, Ipt, Jts, jps,
Idz, ntjt, kts, sts, t0
kst0, If0s, Jldz
Time3
Anh
g, J, 8, ds, kt, bd, 5z,
dsd, nd, cfs, Jz, Iz, fs,
pt, sd, jf, ft, 6d, dz, It,
vd, Ik, jj, zd, qz, jp,
1m, ks, Jt, Jt, f0, gd,
md, jl, Ib, gz, jg, jd,
d0, Jds, Ids, sp, t0, Id,
st, nt, Iv, jm, jk, dst,
kst, jld, qkt, Jm0, kts,
mpt, n0s, mps, qks,
ndz, nts, J0s, If0, jst,
jpt, jkt, Ikt, jdz, Jts,
sks, jbz, jps, Iks, Idz,
ntjt, t0s, sts, kst0, If0s,
jldz
Nhi
g, J, 3, 0, 5, ds, It, Jt,
pt, jk, t0, Jt, jj, Id, In,
jp, jg, jb, gd, J0, is,
d0, Jds, Id3, 3d, 5d, fs,
f0, 0s, 5z, n0, nd, qd,
dsd, dst, jld, qkt, jm0,
mpt, nts, n0s, jnd, ndz,
J0s, If0, jpt, Ikt, Jdz,
jts, Jbz, jps, Idz, kts,
t0s, sts, kst0, If0s, jldz
continued
/s ft. These phonemes range in the 50-100 per cent production
accuracy percentile in the interview data, and between 75-99 per cent
for the word list data and are the most target-like produced singleton
codas for both data sets. This is followed by the liquids, the voiced
fricatives /z v/ and the voiced stop /d/ - these are produced in the 1-49
percentile in both the interview and word list data. While not present
in the interview data, the voiced interdental fricative is at 0 per cent in
the word list task and the voiced affricate is at 0 per cent across both
tasks. For both tasks, voiced consonants - whether in singleton, CC
or CCC codas - have much lower accuracy percentages than the
corresponding voiceless consonant.
The CC codas that have begun emerging in the interview data are
also the codas that are produced at higher accuracy percentages in the
word list task, and they contain the consonants that have begun
emerging as singleton codas in the interview data - the nasals,
voiceless stops, voiceless fricatives, and the voiced fricative /z/; for
example, Nhi was able to produce /nz jz ks/ with 25-49 per cent
accuracy. While no CCC codas were produced target-like in the
interview task, it is interesting to note that those that have begun to be
produced target-like in the word list data consist of nasals, voiceless
stops, voiceless fricatives, liquids and /z/. For example, for Anh
/nst qst inz/ and for Nhi /nst qst jmz ist jkt mz rjks Iks/ had begun to be
produced target-like in the word list task.
In terms of codas by type, across both tasks and both partici-
pants, there are a number of patterns (note: the following notation
will used for the codas: L = liquid, S = stop, N = nasal, F =
fricative and # = margin so that LF# stands for a liquid-fricative
coda. Note: affricates are classified as fricatives in this analysis). In
the interview data, Nhi had begun producing LF#, NF# and SF#
codas with some accuracy. These are also the CC codas with the
highest accuracy percentages in the word list data. The CCC codas
that emerge first are as follows: LNF#, NFS# and LFS#. The most
difficult CC and CCC codas are those consisting of the least target-
like singleton consonants such as the voiced stops /b g/, voiceless
fricatives /J 0/ and voiced fricatives /5 3 ds/ either in combination
with each other or in combination with more target-like consonants.
Even the more easily produced CC and CCC clusters, such as NF#
and LF#, have very low accuracy percentages if they consist of the
more difficult consonants. Regardless of which consonants com-
posed them, the most difficult CC and CCC codas to produce were
as follows: LL#, SS#, FF#, LSF#, FSF#, NSF#, SFS#, NSS#,
NFF#, LSF#, LFF#, SSF# and SFF#. A VARBRUL analysis was
conducted in order to determine whether target-like production
numbers were statistically significant for each coda. Only interview
codas and of those, singleton codas, were included in this analysis
as only 39 CC and 0 CCC were produced target-like. This is
shown in Table 4. 8, below. There were a total of 3, 703 singleton
codas in the interview data; however, there were no tokens of /3 5/
in the data set. The analysis included the same factor groups from
the previous analysis; however, only type of coda (e. g. /p/ versus /b/,
etc. ) and participant were significant (%
2
= 28. 7153, 18 df, p <
. 025). A VARBRUL analysis of each participant's target-like
production was not conducted as this would result in a single
independent variable. However, numbers and percentages for each
individual are given for each coda type. In fact, as Table 4. 8
indicates, the participants had similar production accuracy percent-
ages for each coda and when there were differences that could
explain the significance of participant in the VARBRUL analysis, it
was typically due to a lack of tokens of a given singleton for one of
the two participants.
Based on the VARBRUL input probabilities, which were statistic-
ally significant at p <. 025, it is possible to establish a pattern of
target-like production: /p/ (pi. 90) > /n/ (pi. 82), /m/ (pi. 77), /?/
(pi. 72) > /f, s/ (pi. 64) > kl (pi. 58) > /k/ (pi. 51) > /v/ (pi. 33), /I/
(pi. 33), /r/ (pi. 31) > /g/ (p'. 32), /d/ (pi. 21), /b/ (pi. 12), /t?/ (pi. 18),
/z/ (pi. 12), /?/ (pi. 10), /?/ (pi. 03), /d3/ (pi. 00). In sum, the consonants
that both participants produced most target-like in syllable coda
position were the nasals, voiceless fricatives and voiceless stops. The
consonants that both participants had some difficulty producing in
syllable coda position were the liquids, the voiced fricative /v/ and the
voiceless affricate. The most difficult consonants in syllable coda
position for both participants were all the voiced stops, the voiced
fricative /z/, the voiceless fricatives /[/ and /0/ and the voiced affricate.
4. 2. 3 Production type of individual codas
As examined above, there were very consistent patterns in the
ordering of which codas emerged in terms of their accuracy
percentiles; similarly, there were consistencies in the production
type of each coda type (e. g. epenthesis or feature change). Unlike
previous research that has suggested that learners from specific
language backgrounds produce codas in one specific manner, this
study found that the production types were patterned by specific
Table 4. 8 VARBRUL analysis of target-like production by singleton coda
Coda
Both
P
1
'
#
n
%
Anh
#
n
%
Nhi
#
n
%
/p/
. 90
30
34
88%
17
21
81%
13
13
100%
Ibl
. 12
2
19
11%
0
7
0%
2
12
17%
Itl
. 58
204
377
54%
104
199
52%
100
172
58%
Idl
. 21
24
126
19%
8
57
14%
16
69
23%
/k/
. 51
129
272
47%
50
131
38%
79
141
56%
/g/
. 32
4
15
27%
4
12
33%
0
3
0%
Isl
. 64
93
159
58%
65
115
57%
28
44
64%
Izl
. 12
28
273
10%
14
135
10%
14
138
10%
/fif
. 64
37
61
61%
18
27
67%
19
34
56%
M
. 33
62
210
30%
18
90
20%
44
120
37%
/J V
. 10
12
75
16%
8
29
28%
4
46
9%
/tf/
. 10
12
75
16%
8
29
28%
4
46
9%
Id
0
19
0%
0
3
0%
0
16
0%
/e/
. 03
1
41
2%
1
19
5%
0
22
0%
Iml
. 77
209
282
74%
91
134
68%
118
148
80%
Inl
. 82
552
700
79%
311
397
78%
241
303
80%
W
. 72
103
150
69%
45
70
64%
58
80
73%
Idl
. 31
180
666
27%
97
397
24%
83
269
31%
III
. 33
62
212
29%
31
113
27%
31
99
31%
consonant type. A discussion of the most important and consistent
production types is presented below.
C:
Voiceless stops: All three voiceless stops were produced target-like the
majority of the time, with accuracy increasing across time. When not
produced target-like, they were epenthesized, which occurred pri-
marily before a pause. At times, /k/ was produced as [tf] (in headache),
[s] (in speak and pack), and [t] (in take), and thus appears to be word-
specific. In a pattern that would occur across other codas as well
(discussed more below), the two types of modifications for /t k/ were
absence and epenthesis after tense vowels and diphthongs.
Grammatical conditioning, which will also be discussed in more
detail below, also had an effect on production of /t/: while there were
only 7 tokens of /t/ as a past tense marker in the data, 57 per cent of
these were absent and only 1 (14 per cent) was produced target-like,
compared with only 27 per cent (100/270) absent monomorphemic
/t/ and 55 per cent (203/370) target-like monomorphemic /t/.
Nasals: Nasals were usually produced target-like. While there were
cases of absence for both /m/ and /n/, which decreased across time,
these cases only occurred following the diphthong /ay/. All three
nasals were also epenthesized before a pause and /m q/ were
assimilated when they preceded alveolar consonants. Finally, words
ending in /m/ had absence and epenthesis when /m/ followed a
diphthong as in the word time, which was produced as [ta. ys]. There
were only a few examples of syllabic nasals in the data (4 tokens of/m/
and /n/ each) and therefore patterns are difficult to discern. For the
syllabic /m/, 75 per cent (3/4) were produced target-like and 25 per
cent (1/4) were epenthesized. A similar pattern occurred for syllabic
/n/: 2 of each were target-like and epenthesized, respectively.
Voiced stops: /b d/ were commonly produced as [p t] across time, as
well as being epenthesized to fall syllable-initially before a pause.
Words ending in /d/ typically had absence of the /d/ and were
epenthesized when following a tense vowel; /g/ was typically absent
from production but also began emerging as [k] and target-like at
time 3. Grammatical conditioning (discussed below) could also be
factor in the absence of/d/: 57 per cent (17/30) past tense markers
were absent compared with only 17 per cent (17/96) of the
monomorphemic /d/.
Voiceless fricatives: The voiceless fricatives /s f/ were usually target-like
or epenthesized while /JY was commonly produced as [k] for the word
English and as [s] in the words wish, finish and fish. The voiceless
interdental fricative /6/ was typically produced as [. te] or as [t] or [s].
The voiceless affricate /tf/ was produced variously based on specific
words: as [ts] in beach, in watch as [t], in each to [k] or [f].
Voiced fricatives: For both Izl and /v/, target-like production was fairly
stable while rate of epenthesis increased and absence decreased.
Grammatical conditioning (discussed below) may also have been a
factor in the absence of Izl: 74 per cent (20/27) of the bimorphemic
tokens of Izl were absent compared with 56 per cent (138/246) of the
monomorphemic tokens. The voiced affricate /ds/ typically was
produced as [. ts] or [. ks].
Liquids: III was typically absent after /o/ (60/177 at time 1 and time 2,
and 94/312 at time 3). In every instance of this, the III was co-
articulated with the vowel, resulting in [a], which increased across
time. /!/ tended to be produced as [n] after back and central vowels
and as [ow] or [uw] after front vowels and diphthongs. The analysis
of the 34 tokens of syllabic /!/ in the data show that 2 (6 per cent) were
absent in production, 7 (22 per cent) were target-like, 2 (6 per cent)
were epenthesized, 11 (34 per cent) had two production types, and 10
(31 per cent) were produced as either [n] or [ow], following a similar
pattern as non-syllabic /!/.
As there were a number of similarities in production type by
singeton coda, a VARBRUL analysis was conducted in order to
determine whether production type was statistically significant. For
the VARBRUL runs, the singletons were grouped by type when
production was similar - e. g., the voiceless stops /t/ and /k/ were run
together while /p/ was not added to this group due to its higher
production accuracy. Similarly, the velar nasal /rj/ was run separately
from /m/ and /n/ due to its higher incidence of feature change. Some
consonants such as /0/ and /ds/ were run together as they were
produced fairly similarly across time. The results are outlined in
Table 4. 9, below (note: this type of analysis was not run for onsets
due to the high target-like percentages for onsets). The input
probabilities for each group of consonants are given, as well as the
input probabilities for the other significant factor groups for that
group of consonants. Finally, chi-square statistics (total chi-square,
chi-square per cell, degrees of freedom and statistical significance
level) are given for each group of consonants.
As Table 4. 9, below, indicates, production type was statistically
significant for each group of singleton codas. For each group, the
favoured production type is given along with the input probability:
For /p/, target-like production was favoured (pi. 53); for /t/ and /k/,
target-like production (p1. 58) was slightly favoured while absence (pi
. 61) and epenthesis ( p
l
. 69) were more strongly favoured; for the
nasals /m/ and /n/, target-like production was strongly favoured (p
l
. 75); for the velar nasal /?/, target-like production (pi. 77) and feature
change ( p i . 82) were both strongly favoured; for the voiced stops /b d
g/, feature change ( p i . 86) and two types of production (pi. 73) were
strongly favoured; for the voiceless fricatives /s f/, target-like produc-
tion (pi. 60) and epenthesis ( p i . 61) were favoured; for the voiced
fricatives /z v/, epenthesis ( p i . 85), two types of modifications (pi. 78)
and absence (p
l
. 78) were strongly favoured; for the voiceless fricative
/J Y and the voiceless affricate /tj/, feature change (pi. 90) and two types
of modifications (pi. 94) were both strongly favoured; for /6/ and /ds/,
feature change and two types of modifications (pi. 97) were strongly
favoured; and for the liquids, feature change (pi. 82) was strongly
favoured and absence ( p i . 60) was favoured. In summary, the most
favoured production type for each consonant is as follows: target-like
production /p m n/; epenthesis /t k s f z v/; feature change /q b d g tj ds
j I/, and two types /J 0 tj d3/ (for /0 d3/, feature change and two types of
production had the same input probability). In terms of other
significant factor groups for the different consonant types, a few are
important to note: grammatical conditioning was significant for
voiced stops and voiced fricatives (discussed more below) and
following linguistic environment was significant for /q/.
CC:
Fricative-Stop: (/st ft sk J t vd zd/). 63 per cent (66/104) had one
member absent and 33 per cent (34) had both absence and were
epenthesized. Regardless of grammatical conditioning (e. g., /vd zd J t/
clusters were bimorphemic), the first member of the cluster, the
fricative, was produced while the stop was absent.
Stop-Stop: (/kt pt/). 75 per cent (16/20) had one member absent while
25 per cent (4) had absence and were epenthesized. All tokens of/pt/
were monomorphemic and the first stop was produced while the final
stop was absent. At time 1, all /kt/ codas were monomorphemic and
Table 4. 9 VARBRUL analysis of production by singleton coda
Factors
/Pi
(n: 34)
y
#
%
/ t k/
(n: 642)
y
#
%
/ mn/
(n: 982)
y
#
%
to/
(n: 150)
y
#
%
/ bdg/
(n: 160)
P*
#
%
/ s f /
(n: 220)
y
#
%
/ 2V/
(n: 483)
^#
%
/j tj y
(n: 87)
P'
#
%
T-L
. 53
30
88%
. 58
332
52%
. 75
761
78%
. 77
103
69%
. 35
30
19%
. 60
130
59%
. 35
90
19%
. 39
13
15%
Abs.
0
. 61
186
29%
. 30
114
12%
. 06
1
1%
. 53
39
24%
. 42
51
23%
. 78
226
47%
. 34
4
5%
Ep.
. 36
4
12%
. 69
96
15%
. 52
69
7%
. 16
1
1%
. 32
5
3%
. 61
26
12%
. 85
122
25%
. 21
1
1%
Fc.
0
. 19
12
2%
. 20
36
4%
. 82
45
30%
. 86
79
49%
. 29
13
6%
. 27
17
4%
. 90
57
66%
2T
0
. 51
16
2%
. 05
2
0%
0
. 73
7
4%
0
. 78
28
6%
. 94
12
14%
Other significant factor
groups
Following linguistic
environment: Pause, Glide*:
. 68 Consonant:. 36 Vowel:
. 38; Time: Time 1:. 65 Time
2:. 50 Time 3:. 30;
Participant: Anh:. 57 Nhi:
. 43
Time:
Time 1:. 55, Time 2:. 48,
Time 3:. 47
Preceding linguistic
environment:
Vowel:. 56, Diphthong:. 83,
/o/:. 04
environment:
Pause:. 40, Vowel:. 36,
Glide:. 52, Consonant:. 62
Grammatical conditioning:
Monomorphemic:. 48
Bimorphemic:. 97
environment:
Vowel:. 50, Diphthong:. 85,
/o/:. 25; Time: Time 1:. 42,
Time 2:. 44, Time 3:. 61
Grammatical conditioning:
Monomorphemic:. 49
Bimorphemic:. 80
Participant: Anh:. 46 Nhi:
. 55
environment: Vowel:. 54, /o/:
. 20
Time: Time 1:. 58, Time 2:
. 63, Time 3:. 35
Statistical
significance
Input weight:. 012
Total x
2
: 20. 1147
X
2
/cell:. 8381
df: 5; p <. 0001
Input weight:. 15
Total x
2
: 11. 6052
X
2
/cell:. 7737
df: 6; p. <. 05
Input weight:. 176
Total x
2
: 13. 3545
X
2
/cell:. 9538
df: 6; p <. 025
Input weight:. 176
Total x
2
: 13. 3545
X
2
/cell:. 9538
df: 6; p <. 025
Input weight:. 025
Total x
2
: 9. 9535
X
2
/cell:. 9953
df: 7; p <. 05
Input weight:. 047
Total x
2
: 30. 8839
X
2
/cell:. 9359
df: 7; p <. 0001
Input weight:. 076
Total x
2
: 11. 6202
X
2
/cell:. 6835
df: 6, p <. 05
Input weight:. 009
Total x
2
: 21. 6941
X
2
/cell:. 7231
df: 7; p <. 001
i
continued
Factors
/ e < b /
(n: 60)
P'
#
%
/ j I/
(n: 878)
y
#
%
T-L
. 17
1
2%
. 38
242
28%
Ab s.
. 61
4
7%
. 65
261
30%
Ep.
. 50
1
2%
. 19*
23
3%
Fc.
. 97*
19
32%
. 82
347
40%
2T
. 97*
35
58%
. 19*
5
1%
Othe r significant factor
groups
environment:
Pause :. 33, Glide :. 51,
Consonant:. 59, Vowe l:
environment:
Diphthong:. 08, Vowe l:
hi:. 67
Statistical
significance
Input we ight:. 003
Total x
2
: 13. 3000
5C
2
/ ce ll:. 8313
. 67 df: 6; p < . 025
Input we ight:. 19
Total i2: 9. 7433
. 53, x
2
/ ce ll:. 8857
df: 5; p < . 05
T-L: Targe t-like , Ab s: Ab se nce , Ep.: Epe nthe sis, Fc.: Fe ature change , 2T:
Two type s of production. *Run as one factor to achie ve a b e tte r mode l of
variation due to similar input prob ab ilitie s in e arlie r runs.
production was word-spe cific: as [t]: expect, correct', as [k]: connect;
ab se nce of b oth: correct. At time 2, all the cluste rs we re b imorphe mic
and [k] was produce d.
Stop-Fricative: (/ ks ts b z/ ). 68 pe r ce nt (49/ 72) had one me mb e r
ab se nt and 15 pe r ce nt (11) we re targe t-like ; / b z/ was produce d as
[ps] in b oth case s. If one me mb e r was ab se nt, it was the fricative .
Bimorphe mic cluste rs had a highe r accuracy pe rce ntage (50 pe r ce nt)
than monomorphe mic cluste rs (10 pe r ce nt).
Fricative-Fricative: (/ 5z/ ). The re we re only 5 toke ns of this cluste r, all
b imorphe mic and with [z] produce d for the e ntire cluste r.
Liquid-Liquid: (Iill}. In the only e xample of this coda, only [1] was
produce d.
Liquid-Fricative: (/If 16 Iv iz Iz JtJ/). 24 per cent (9/38) were target-like,
39 pe r ce nt (15) had one me mb e r ab se nt (the fricative in e ve ry case ),
16 pe r ce nt (6) we re produce d as diffe re nt consonants (e . g. [jt] for /iQ f)
and 21 pe r ce nt (8) had ab se nce and e pe nthe sis following a te nse
vowe l or ab se nce and fe ature change . The highe st accuracy pe rce nt-
age s we re for /iz/ - 50 pe r ce nt (9/ 18) we re targe t-like .
Liquid-Nasal: (/ in/ ). Eight of the 10 toke ns of this coda we re produce d
as [n]. At time 3, 1 ml was produced target-like one time (1/6-13 per
cent).
Liquid-Stop: (/it jk id Id lp/). Of the 95 tokens of this coda, only 8 were
with III + stop. For both the III and III + stop clusters, the most target-
like were those with /d/; /ik/ was also target-like at times. For III
clusters, the III was produced when one member was absent.
Nasal-Stop: (/nd nt qd/). All examples of and as well as all cases of
contractions (e. g., can't, don't) were eliminated from analysis as they
typically have absence of the final stop. For nasal-stop clusters, the
patterns were similar: 86 per cent (160/187) were produced as the
nasal; 11 per cent (21/187) were produced as an epenthesized nasal;
and 2 per cent (4/187) were target-like (only /nt/ clusters). All /nd/
clusters were monomorphemic.
Nasal-Fricative: (/ntj nds n0 mz nz/). There were 70 tokens of this
coda: 90 per cent (63) were produced as [n] or absent (for /mz/ after a
diphthong). Only /nz/ was target-like.
CCC:
Liquid-Fricative-Stop: (/jst/). There was only one type of coda for this
cluster and it was monomorphemic and produced as [i] or [iz].
Liquid-Stop-Fricative: (/idz Idz its/). There were only 4 tokens of this
coda. For 3 of these, a liquid-fricative coda ([Iz iz]) was produced. In
one case, /idz/ (in words) was produced as [k]. All were bimorphemic.
Stop-Fricative-Stop: (/kst/). In the 7 tokens of this cluster (all
monomorphemic), 56 per cent (5) were produced as /k/, while 2
were produced as [. fcs].
Liquid-Nasal-Stop: (/md/). Of the 3 tokens, 2 were produced [n] and
1 was epenthesized.
Nasal-Stop-Fricative: (/nts ndz/). 92 per cent (72/78) were produced
as either [n] or [nz] (only for /ndz/). The remaining 8 per cent (6)
were produced as [. no]. All were bimorphemic.
Liquid-Liquid-Fricative: (/ilz/). The one example of this was produced
as [z] (bimorphemic).
Liquid-Stop-Stop: (/ikt/). Of the 5 tokens of this, all bimorphemic, 3
were produced as [. ko] and 2 were produced only as [k].
Nasal-Fricative-Fricative: (/n0s/). The 22 tokens of this were all
bimorphemic; 91 per cent (20) were produced as [n], [nz], or [n9]
(Ann only). Two were produced as [n. zs].
In summary, the analysis of the production C, CC and CCC codas
indicates that there are differences in the type of production for
different codas, and that these differences are statistically significant
in the case of singleton codas (CC and CCC codas were not analysed
statistically due to limited tokens of each type). In summary, the
favoured production type for each consonant is as follows: target-like
production /p m n/; epenthesis /t k s f z v/; feature change /q b d g tf ds
i I/, and two types /J 0 tj ds/. For CC codas, one segment was typically
absent. For the majority of the codas (e. g., FS#, SS#, LF#, NS#,
NF#), the first segment was produced; for a few codas (e. g., LL# and
LN#) the second segment produced. Finally, for the LS# coda, the
liquid was typically produced if it was /!/ while for /j/, the stop was
typically produced and the liquid was co-articulated with the vowel or
absent. Finally, CCC codas most typically were produced as a C or
CC coda; if produced as a CC coda, it was a coda that had begun
emerging, such as /nz Iz jz/. Production of CCC codas differed by
coda type: LFS# and LSF# were both produced as LF#; SFS# was
produced as the first stop; LNS# was produced as a nasal; NSF# was
produced as a nasal or NF#; LLF# was produced as a fricative; LSS#
was produced as the first stop; and NFF# was produced as the nasal.
Overall, the nasal was favoured for production across all coda types.
Two more things bear examination: the role of grammatical
conditioning and the production of tense vowels and diphthongs in
open syllables given their effect on absence and epenthesis in closed
syllables. First, in terms of the role of grammatical conditioning, as
the statistics on singleton consonants (described above) indicate,
grammatical conditioning was a significant factor for voiced stops and
voiced fricatives. Additional statistics on the codas that could be
bimorphemic were run. For the singleton codas, grammatical
conditioning was significant at p <. 05 (total %
2
8. 6483, %
2
/cell
. 9609, 4 df). Specifically, a monomorphemic coda slightly disfa-
voured absence at. 47 while a bimorphemic coda favoured absence at
. 74. For CC codas, grammatical conditioning was also significant,
this time at p <. 001 (total %2 20. 5332, %2 /cell. 5867, 7 df). However,
for CC codas, monomorphemic codas had a slight favouring effect on
absence at. 56 while bimorphemic codas strongly disfavoured absence
at. 28. For CCC codas, grammatical conditioning was also statistic-
ally significant at p <. 001 (total %2 12. 4123, %2 /cell. 8865, 5 df).
An examination of absence and accuracy patterns by inflection type
(plural, possessive, third person singular, regular past tense, irregular
past tense) yielded few patterns. First, there were too few tokens (e. g.
less than ten tokens of each) of the irregular past tense, the possessive
form and the third person singular form for analysis. Secondly, for the
CC and CCC codas, the last member(s) were typically absent,
meaning that both the tense and the plural markers (or any other)
were absent, thus making it difficult to discern any differences among
the inflectional morphemes. This left only the singleton coda Izl for
the plural marker and /d/ for the past tense marker for analysis (all
seven of the bimorphemic marker /t/ were irregular past tense; of
these 1/7 (14 per cent) were target-like and the rest were absent). For
these codas, 11 per cent (3/27) of the Izl plural marker were target-
like while 43 per cent (13/30) of the past tense marker /d/ were target-
like, which is interesting since only 11 per cent (11/96) of the
monomorphemic /d/ were target-like (10 per cent or 25/246 of the
monomorphemic Izl were target-like).
Finally, for both C and CC codas, it was noted that tense vowels
and the diphthongs /ay/ and /au/ were epenthesized after absence of
the C or CC cluster, resulting in [o. ba. ws] for about, for example. An
analysis of CV codas ending in the diphthongs /ay au/ and the tense
vowels /iy ey/, the vowels that seemed to promote epenthesis in closed
syllables, was conducted in order to determine whether these vowels
were also epenthesized in open syllables. Of the 1, 555 tokens of these
vowels in the data, 180, or 12 per cent, were epenthesized; 77 per
cent of these (139/180) preceded a pause. However, not every
instance of these vowels before pauses resulted in epenthesis: there
were a total of 495 of these vowels before a pause and only 28 per cent
(139) were epenthesized. The other 356 (72 per cent) were produced
without epenthesis.
4. 3 A comparison of onset and coda production
In comparing the onset and the coda data, it is clear that onsets
overall have a much higher percentage of target-like production than
codas, with an 87 per cent target-like production for onsets when data
are collapsed across participant and time; codas were at 41 per cent
(interview data only). Therefore, onsets were produced twice as
accurately as codas. Within each margin type, there are also clear
patterns: for both onsets and codas, singletons were produced more
accurately than CC margins, and these were in turn produced more
accurately than CCC margins. If the data are collapsed by time and
participant (interview data only), the following accuracy percentages
are found for onsets and codas: singleton onsets 91 per cent, CC
onsets 65 per cent, CCC onsets 39 per cent, singleton codas 47 per
cent, CC codas 6 per cent, CCC codas 0 per cent. It is therefore
possible to posit a hierarchy of production and acquisition based on
length of margin: C > CC > CCC. Additionally, it is interesting to
note that both C and CC onsets are produced more accurately than C
codas, and that CCC onsets are produced more accurately than CC
and CCC codas. Therefore, both length and type of margin has an
effect on accuracy.
Because onsets have a higher accuracy percentage overall, there are
fewer modifications in their production and less absence than for
codas. In fact, onsets exhibit fewer modifications in number as well as
fewer types of modifications: for C onsets, accuracy or feature change
was common, while for CC and CCC codas accuracy or absence was
dominant. There were only a few cases of epenthesis for onsets and
no examples of two types of modifications. For codas, the number
and type of modifications were greater: for C codas, target-like
production, absence, feature change and epenthesis were found while
for CC codas, absence, two types of modifications, as well as a few
cases of feature change and target-like production were found.
Finally, for CCC codas, both absence and two types of modifications
were found.
In terms of the linguistic and extralinguistic constraints that
affected the different types of production for onsets and codas, a
number of interesting patterns were found. For target-like produc-
tion, the same linguistic factor affected both onsets and codas: for
codas, this was the preceding linguistic environment, specifically that
vowels favoured target-like production over diphthongs and /o/; for
onsets, this was the following linguistic environment, and in this case,
diphthongs and hi promoted target-like production over vowels. For
absence, length was a factor for both onsets and codas, with CC and
CCC margins favouring absence over C margins for both onsets and
codas. For epenthesis, stress was a factor for both onsets and codas,
with an unstressed syllable favouring epenthesis for onsets while a
stressed syllable favoured it for codas. Preceding linguistic environ-
ment for onsets and following linguistic environment for codas also
affected epenthesis; for onsets, a preceding consonant promoted
epenthesis over a preceding vowel while for codas, a following pause
or glide promoted epenthesis over a consonant or vowel. Finally, time
affected epenthesis for both as well: for onsets, time 1 and 3 favoured
epenthesis over time 2 while for codas, time 3 favoured epenthesis
over time 1 and 2. For feature change, the only common factor was
stress, with a stressed onset promoting feature change over an
unstressed one while an unstressed coda promoted feature change
over a stressed coda.
It is also possible to determine whether there are similarities
between onsets and codas in terms of which consonants and
consonant clusters emerge before others as evidenced by their
production accuracies. For onsets, the most accurately produced
onsets were the singleton onsets / mnt kbdgs f hwyj l / , that is, the
nasals, voiceless stops, voiced stops, voiceless fricatives, glides and
liquids. While fewer of these sounds were produced with a high
degree of accuracy in coda position, there is an overlap in terms of
which type of sound emerged first: nasals, voiceless stops and
voiceless fricatives. The voiced fricative Izl was more difficult for the
participants in both onset and coda form, and the affricates /J 3/ and
the interdental fricatives were the most difficult singletons regardless
of margin position. While a number of CC and CCC margins were
emerging for both onset and coda clusters, a longer margin was
produced only after its constituents had been produced in singleton
form with a higher degree of accuracy. For the onsets, these clusters
were #SG, #FS and #FL (with III as the liquid). More difficult,
though still emerging, CC onsets were #SL, #FL (with I i f ) and #FG
while for codas it was LF#, NF# and SG#. Both #FSG (comprised
of #SG and #FS, both of which were produced with a high degree of
accuracy) and #FSL (comprised of #FS and #SL, which were also
emerging) onsets had begun to emerge. While there were no cases of
CCC codas emerging in the interview data, in the word list data,
LSF#, LNF# and NFS# codas are beginning to emerge. Most
difficult margins were the #FSL for onsets and NFF#, LSF#, SFS#,
NSS#, SFF#, FSF#, LSF#, LFF# and FSF# for the codas.
It is also possible to compare the interview coda data with the word
list coda data. From this analysis, it becomes clear that the accuracy
orders and emergence sequence of the codas is nearly identical
regardless of length, time and participant, although the accuracy
percentages in the word list data are higher. The CC codas that
emerged in the word list task are also similar to those of the interview
data, and also consist of nasals, voiceless stops and voiceless
fricatives. The CCC codas that began emerging in the word list
data also consisted of the nasals, voiceless stops, voiceless fricatives,
liquids, /z/ and /d/, also with close similarity between the two
participants. This finding appears to suggest that there are not any
significant differences between the more spontaneous and more
controlled tasks in indicating phonological development within and
across time. However, while the actual emergence order is strikingly
similar between the two tasks, the accurate percentages differ greatly,
especially for the more complex codas (e. g. some CC and CCC codas
are produced with 100 per cent accuracy in the word list but are at 0
per cent in the interview data). Therefore, there is still a question as to
what extent accurate production on a more controlled task can be
generalized to the acquisition of the segment in question, and data
from a more controlled task may be a better indicator of develop-
mental patterns than actual acquisition. This accuracy order
sequence is outlined in Table 4. 10 above.
Finally, it is also possible to compare the production of the
different onsets and codas if they underwent feature change or
absence of one or more members in the consonant clusters in order to
determine similarities of how these sounds may be represented in the
participants' linguistic repertoires. While voiced stops typically
underwent devoicing (e. g. /d/ to [t], /b/ to [p], and /g/ to [k]) as
codas, they were typically produced accurately as onsets; however, /p/
was sometimes produced as [b], without aspiration. In both onset and
codas, /JV was often produced as [s], /6/ to [s] or [t], and /t|7 to [t], [f]
Table 4. 10 Hierarchies in target-like production for singleton codas
Most target-like
Fairly target-like
Somewhat target-like
Poorly produced
Very poorly produced
Unclear
Voiceless stop
Nasals
Voiceless fricatives
Voiceless stops
Liquids
Voiced fricative
Voiceless affricate
Voiced fricative
Voiceless fricative
Voiced stops
Affricates
Voiced fricatives
P
n > m > q
s, f
t> k
u
V
tj
z
J > e
b, d, g
tf> d3
3, 9
or [k] depending on the word. For both onsets and codas, CC and
CCC margins were often reduced to one member, as outlined below
in Table 4. 11. As Table 4. 11, below, illustrates, a number of patterns
are apparent: in FS or FG margins, the fricative is usually retained.
Nasals are retained in clusters, regardless of type and length. Stops
are often absent in clusters, regardless of position (1st or 2nd or 3rd
member), length or margin.
In summary, while onsets and codas were produced with very
different levels of accuracy, there nonetheless were a number of
similarities in terms of emergence orders as well as how the margins
were being produced as they emerged. These developmental patterns,
and the linguistic factors that affected these patterns, will be discussed
in Chapter 5, below.
Table 4. 11 Onset and coda cluster production patterns
Onsets
#FS - #F
#FG - #F
#FL (/I/) -* #L
#FL (/i/ -> #
#SL -> #L
#SG -> #S
#FSL - #F, #SL,
#FS
Codas
FS# -> F#
SS# - S# (1st)
SF# -* S#
FF# -> F# (2nd)
LF# -> L#
LN# _>N#
NS# -> N#
NF# - N#
LFS# -> L#
3
LF#
LSF# -> LF#
LLF# -> F#
LSS# - S# (1st)
LNS# - N#
SFS# -> S# (1st)
NSF# ? N#, NF#
NFF# ? N#, NF#
5
Linguistic and Task
Constraints
This chapter will examine the effect of linguistic and task constraints
on the developmental patterns evidenced for the acquisition of
syllable onsets and codas by the two participants within and across
time. First, LI transfer will be discussed, followed by developmental
effects and then markedness. These are followed by a discussion of
the effect of linguistic environment, grammatical conditioning and
task type. Finally, a synthesis of the effects of these constraints is
provided.
5. 1 LI transfer
This study confirms prior research (e. g. Benson 1988; Osburne 1996;
Sato 1984, 1985) that LI transfer is a primary factor in the
development of English L2 onsets and codas for native speakers of
Vietnamese. In fact, LI transfer appears to constrain production
accuracies and the order of development of syllable margins,
confirming prior research (Benson 1988; Osburne 1996; Sato 1984,
1985). In Vietnamese, only voiceless stops and nasals are allowed
syllable-finally, while voiced stops, voiceless stops (not /p/, which is an
allophone of /b/ that occurs syllable-finally), voiced and voiceless
fricatives, nasals and the rhotic /r/ are allowed syllable-initially
(Nguyen 1970). It appears that the consonants are initially repre-
sented in the learners' L2 phonologies in a similar manner as they are
represented in Vietnamese. This study found that initially, the
learners' emerging L2 repertoire consisted of the sounds they
transferred from their LI and that those sounds that are allowed
syllable-initially and syllable-finally in the LI emerge in their
respective syllable positions with a higher accuracy than other
consonants.
In terms of the singleton onsets, this is evidenced by the higher
production accuracies of the nasals /m n/, the voiced stops /b d/, the
voiceless stops /t k/, the voiceless fricatives /s f h/, the voiced fricative
/v/, the glides /w y/ and the liquids /j I/, all of which exist in
Vietnamese in syllable-initial position. Interestingly, while the par-
ticipants had a higher accuracy percentage on /g/ as well, this
consonant does not exist in Vietnamese; the Vietnamese /k/ is
produced as the plain, unaspirated [k] in initial position, which is
largely the same as /g/, which explains the higher accuracy of /g/ for
the participants. Additionally, transfer explains the lower production
accuracy of /p/ and its production as [b]. In Vietnamese, /p/ is a
syllable-final allophone of/b/; therefore, the learners are producing /p/
as the nearest equivalent in their LI repertoire, which is /b/. The
lower accuracy percentage (though it is emerging) of /[/, which does
exist in Vietnamese but with a different articulation, can be explained
by difficulties the participants may have in shifting its articulation
from a retroflex (in Vietnamese) to an alveo-palatal (in English). The
consonants that do not exist in Vietnamese are the most difficult
consonants for the participants to produce in syllable-onset position:
/tj ds 6 5/.
As for singleton codas, the consonants which exist syllable-finally in
Vietnamese, the voiceless stops and nasals, are the consonants with
which the participants appear to have less difficulty, as evidenced by
the higher accuracy percentages for /p t k m n rj/ across time for both
Anh and Nhi and the fact that only /p m n/ reach the 80 per cent
criterion point of acquisition across time. The fricatives and voiced
stops which do exist in Vietnamese but not in final position, are
typically (though not absolutely) produced with greater accuracy than
those consonants that do not exist in Vietnamese, such as /tj ds 0 5 3/.
This is true for the consonants /f s b v j z J/. Interestingly, all of these
consonants have begun emerging in syllable-onset position in the
learners' L2 repertoire, and therefore the participants could be said to
be building on their emerging L2 phonologies. The devoicing of the
voiced stops /b d g/ so they are produced as [p t k] respectively, which
are allowed syllable-finally in Vietnamese, may be affected by LI
transfer. By producing these consonants in a voiceless manner, the
participants are conforming to Vietnamese phonotactics.
Furthermore, the common modification process for /s z f v/, all of
which exist in Vietnamese but only syllable-initially, was initially
absence, followed by epenthesis, indicating that the participants were
resyllabifying the syllable structures by inserting a vowel after the
consonant and creating a new syllable boundary in order for these
codas to conform to Vietnamese phonotactics. For example, a word
such as if /If/ would be pronounced as [I.fs]. This also helps to
LINGUISTIC AND TASK CONSTRAINTS 97
account for the fact that as absence decreased across time, epenthesis
increased: the phonemes were emerging in the learners' L2 but
initially only in syllable-initial position, as in Vietnamese. This pattern
diverged slightly for /s/, which had a greater percentage of absence
across time, rather than epenthesis. This consonant also had a higher
level of accuracy than the other fricatives, and thus having begun to
be acquired in syllable-final position, the level of epenthesis for /s/
may be decreasing across time, as in fact the data indicates.
In comparing the production of several consonants in onset and
coda position, the effects of LI transfer become more apparent. The
consonant /v/, which exists syllable-initially but not syllable-finally in
Vietnamese, is produced with a high degree of accuracy in onset
position, but with a much lower degree of accuracy in coda position.
The consonant /p/, conversely, which exists as an allophone of /b/ in
syllable-final position, is produced with nearly 100 per cent accuracy
in syllable-final position, but with a much lower accuracy percentage
in syllable-initial position, where it is produced as [b], most likely
because it is /b/ in syllable-initial position in Vietnamese. Therefore,
while both /p v/ exist in the participants' L2 repertoires, their different
levels of production accuracies based on the syllable margin is
affected by their consonants' syllable positions in the learners' LI.
Overall, it appears that the participants make a number of
equivalence classifications between Vietnamese and English, com-
paring consonants in their LI repertoire against their perceptions of
English consonants, thus effectively matching L2 consonants to
consonants they already can produce. This can be evidenced by the
type of production modification common for each margin. While
statistics were not run for singleton onsets as there were few samples
of different types of modifications, for the singleton codas the type of
production was statistically significant for each group of consonants.
The patterns that emerge as a result of the production modifications
are indicative of acquisition processes. It appears that as consonants
begin to emerge in the learners' L2 phonology, they are initially
produced based on the initial state of the L2 phonology, which is
largely based on the LI; the nasals /m n/, for example, favour target-
like production, for both onsets and codas, as does the voiceless coda
stop /p/ - all of which are allowed syllable-finally in Vietnamese (the
velar nasal also strongly favours target-like production; the high rate
of feature change for this consonant is due to nasal assimilation to the
following consonant, a common linguistic process). The voiceless
stops /t k/ also favour target-like production but also have a high
incidence of absence and epenthesis, indicating that while they are
entering the L2 phonology in syllable-final position, this emergence is
not as rapid as that of the nasals and /p/. In onset position, /t k/ are
produced with a high degree of accuracy, while /p/, as discussed
above, emerges initially as [b], based on the LI phonology. One
possible explanation for the absence of /t k/ in coda position is that in
Vietnamese, stops are weakly released or unreleased (Nguyen 1970)
in final position and the participants may be transferring the
unreleased articulation of the final stops to their production of the
English stops, and the stop may therefore have been inaudible and
categorized as absent. It is not clear why this is not the case for /p/,
which has a high accuracy. It is possible that as a syllable-final
allophone of /b/ in Vietnamese, it has more salience; it may also have
more visual salience as a bilabial stop (cf. Hardison 1996).
Singleton coda fricatives initially emerge with epenthesis, thereby
conforming to the acceptable syllable structure in Vietnamese since
these consonants are allowed syllable-initially in Vietnamese; in onset
position, they were generally accurate, due to LI transfer. The voiced
stops are typically produced as a voiceless stop in coda position and
accurately in onset position, thus conforming to an acceptable
Vietnamese syllable structure. The consonants that do not exist in
Vietnamese - /5 0 tJ ds/- are typically produced as a consonant that
has begun emerging (e. g. /6/ as [s t], /t|Y as [t k]) as they are 'classified'
based on the emerging L2 phonology. This same pattern exists
regardless of syllable margin position. Once the participants' phon-
ologies begin expanding, they are able to again reclassify these sounds
based on the sounds that at a given stage comprises their L2
phonology, as evidenced in the production of /ds/ as [tj] at time 3 in
coda position. The coda /6/ also begins emerging syllable finally once
it has begun emerging in onset position. Initially, both /6 5/ are
produced as [t] or [d], and once /6/ begins to be accurately produced,
/5/ begins to be devoiced and produced as [0]. The same pattern exists
for the affricates /tj ds/; initially, both are typically produced as [t] or
[k] when modified. However, as /tj/ begins to be accurately produced,
the voiced affricate begins to be produced in a voiceless manner. It is
interesting that the phoneme /J/, which does exist in Vietnamese but
only syllable-initially, would undergo reclassification rather than
epenthesis like other consonants that exist in syllable-final position in
Vietnamese. Instead, /J/ is modified and produced as either [k] or [s],
which is word-specific, as described above. However, the participants
may have reclassified /J/ based on a perceptual auditory similarity with
/s/ and /k/ as it is a retroflex in Vietnamese and an alveo-palatal in
English. Its voiced counterpart /3/ was reclassified to /g/ for Ann and
to /z/ for Nhi, also possibly on a perceptual auditory similarity with
these sounds for each participant.
In onset position, the liquids /I il are typically produced with a high
degree of accuracy as both of these consonants exist syllable-initially
in Vietnamese. However, in coda position, there is a high percentage
of feature change for both /l/ and /i/, which may also be due to LI
transfer. Prior research on the acquisition of English syllable codas by
a Vietnamese speaker (Osburne 1996) found that III was never
realized in syllable codas. Either it was absent or it was co-articulated
with the vowel. As Osburne suggests, the Vietnamese speaker may
have perceived the articulation of III after a vowel as a glide, and
produced it as such. In fact, as Osburne states, there is some
speculation among linguists about whether /I/ in fact should be
classified as a glide in postvocalic position in English. Kahn (1976),
for example, classifies the English III as a glide rather than as a
sonorant consonant because of the lack of obstruction in the
articulation of /J/, the reduction of III in post-vocalic position to a
glide in non-rhotic dialects, and similar patterning with other glides in
terms of being able to syllabify in the nucleus.
This study confirms Osburne's findings that III was sometimes not
realized in syllable codas and further finds that the type of preceding
vowel affects whether absence or co-articulation occurs. The fact that
feature change appears to be increasing over time as absence
decreases for III may indicate that the participants are not receiving
counter-evidence and therefore the co-articulation perceptual classi-
fication is strengthening.
This analysis can also be tentatively extended to help explain the
increase in feature change of III across time as accurate production
decreases. As described earlier, the preceding environment appeared
to affect how III was modified: after front vowels such as /iy/ and after
diphthongs, III was produced as a rounded back vowel. However,
after a central or unrounded back vowel, III was produced as [n]. I
want to suggest that III is co-articulated with the glide that occurs in
diphthongs and American front vowels such as /iy/. However, after lax
central and back vowels, which lack the glide characteristic, III is
articulated as [n], changing in manner of articulation from an oral to a
nasal consonant.
LI transfer effects also explain why codas are often absent after
diphthongs or the low back vowel. Vietnamese consonants can only
occur after monopthong vowels; Vietnamese does have diphthong
vowels, but they only exist in open syllables. In other words, a
consonant is not allowed after a diphthong in syllable-final position.
Osburne (1996) explains that in Vietnamese, the diphthong is
realized as a vowel + glide rather than as a single segment, and the
glide would thus be viewed as a final consonant, closing the segment.
Both Osburne (1996) and Benson (1988) found that Vietnamese
speakers frequently did not produce a final consonant after a
diphthong; this research study confirms these prior results.
Additionally, this research project found that these effects do not
appear to decrease across time.
This brings up the question of the effect of the low back vowel /h/,
3
which promotes absence. Vietnamese does have this vowel, but it is
classified as a short vowel before consonants while often produced as
a diphthong in an open syllable. As Santry (1997) states, before
consonants, this vowel is actually shorter than its English counterpart.
It is therefore possible that relying on perceptual cues, the partici-
pants are classifying the English /o/ as a diphthong as it appears to be
longer in duration that the Vietnamese /o/ and thus closer in duration
to the Vietnamese diphthong. This would then explain the high rate
of absence of consonants following /o/; if the learners are classifying it
as a diphthong rather than as a monopthong vowel, then the high rate
of absence of consonants after this consonant is indicative of LI
transfer effects.
After these effects were found for codas (regardless of length), the
effect of these vowels were investigated for onsets. They were found
to have a significant effect on target-like production on the preceding
onset in a reverse pattern to that for a following coda: diphthongs and
/o/ promoted target-like production over vowels. It appears that since
diphthongs and /o/ favour absence in a following coda, they
necessarily favour target-like production - or retention - in a
preceding onset since absence of the onsets would reduce the syllable
to just the vowel. In other words, it is necessary to retain the onset in
order to be able to communicate the syllable, especially if the coda is
absent.
Overall, it appears that LI transfer exerts a significant effect on the
development of singleton consonants in terms of accuracy of
production, developmental sequences and the effect of a preceding
diphthong. Prior research studies (e. g. Major 1986, 1987a, 1994)
have suggested that LI transfer effects decrease across time as
acquisition proceeds. This study in part confirms this finding.
Acquisition does not appear to be proceeding if we examine
production percentages only; however, through the examination of
types of production modifications and absence, it is clear that the
consonants are slowly emerging in the learners' L2 phonology, and
that even though overall accuracy percentages remain stable, and low,
across time, the learners are in fact beginning to acquire all the L2
consonants, in similar stages but in different rates across time. As
these consonants do begin to emerge (i. e. the emergence of the
interdental fricatives and affricates, for example), the learners do
appear to rely more on their existing L2 repertoire, rather than their
LI, in modifying and producing these consonants. In other respects,
however, the effects of transfer remain stable across time, in this case
in terms of the effect of the preceding diphthong on production. Even
if an L2 consonant has begun emerging in the learners' L2
phonology, it will be absent if preceded by a diphthong, as in the
case of the consonant /m/, which is typically produced accurately with
the exception of when it is preceded by a diphthong.
The analysis for singleton onsets and codas can be extended to CC
and CCC onsets and codas although the effects of transfer are most
clearly evidenced for the codas as many of the onsets are produced
with a high degree of accuracy. Although there are no onset clusters
allowed in Vietnamese, a number of CC (and CCC) onsets were
produced with some degree of accuracy: /st sp sk si vi tw py dy kw fl fy
ny sm/. All of these CC onsets comprised singleton onsets allowable
in Vietnamese, with the exception of the /p/ in /sp py/ although the /p/
in /sp/ is unaspirated in English, and therefore similar to /b/, which is
produced highly accurately. The least accurately produced CC onsets
were /6i dsy/, both of which had consonants which do not exist in
Vietnamese. In terms of CCC onsets, there were only a few examples
but /skw/ was produced highly accurately, while /sti/ had a mid-level
of accuracy and /spl spi/ had a low degree of accuracy. The onsets
that were not produced accurately were produced as emerging C or
CC onsets, particularly as [s sp pi st ti].
For CC codas, as in the singleton coda data, the participants
appear to be making equivalence classifications and production
modifications based on the LI, especially at time 1 and 2 when they
are in the early stages of acquiring the L2 consonants. This is
especially true for Anh across time while Nhi appears to be making
equivalence classifications based both on the LI and on the
consonants emerging in his L2 repertoire to a greater extent than
Anh. Additionally, similar to the single coda data, epenthesis was
typically employed to resyllabify those consonants to be syllable-
initial which were allowed in this position in Vietnamese. This was
especially common for the fricatives such as /s z f v/ which were the
second member of the two-member codas, for example after liquids
or nasals. For example, the cluster /lv/ was often produced as [l. vs].
Unlike the final voiced stops in the single coda data, which were often
devoiced to conform to Vietnamese phonotactics, the final voiced
stops were often epenthesized to fall syllable-initially, similar to the
fricatives. This different strategy for the production of voiced stops in
CC codas was probably due to the fact that devoicing the consonant
would still not make the coda conform to Vietnamese phonotactics as
the coda would still have two members, whereas epenthesis would
resyllabify the coda into an allowable LI structure. This strategy,
which appears to be based on LI transfer in terms of not allowing
voiced stops syllable-finally, was especially common for Anh. Nhi
appears to have less difficulty in producing a two-member coda.
Similarly to the singleton coda, the production of CC (and CCC
codas) builds upon the emerging L2 phonology. In the few cases (397
607, 6 per cent) of target-like production for the CC codas, the CC
codas comprised consonants already produced target-like in singleton
form. In the few cases of feature change, the codas were produced as
other CC that had begun emerging (e. g. /bz/ to [ps]). The few codas
that were epenthesized also fit emerging patterns for the singleton
codas, such as /st/ to [s. te] and /ks/ to [k. ss]. This also occurred for
cases of absence and feature change (e. g. /ft/ /zd/ to [s], and /Id/ to
[n]) and absence and epenthesis (e. g. /st vd/ to [. so. vs]).
When only one member was produced in a CC coda, it was
typically the first member that was produced, which resulted in a coda
composed of a consonant that was already produced with some
degree of accuracy in singleton form (e. g. /pt/ to [p]; /ts/ to [t]; /nd nt
qk/ to [n q]). There were a few cases where only the second member
was produced; in each of these cases, the consonant that was
produced was the consonant that had a higher accuracy in production
in singleton form (e. g. /kt/ to [t]; /jn/ to [n]; /it/ to [t]; /ik/ to [k]).
As with the single codas, there is also an effect of the previous
phonological environment. As discussed previously for the singleton
coda data, consonants are not allowed syllable-finally after diph-
thongs in Vietnamese. It appeared that the participants were
transferring this phonotactic rule into their production of English
syllable-final consonants as there was a strong likelihood of coda
absence after diphthongs. In fact, both members of the coda were
typically absent after diphthongs in this data set. Additionally, as in
the single coda data, the participants appeared to be perceiving the
English low back vowel /o/ as a diphthong as in English it is produced
with longer duration than the Vietnamese equivalent. Similar to the
single coda data, this preceding low back vowel also promoted
absence of following consonants.
There were also similar results on the production of the liquids as
in the single coda data. As mentioned in the discussion of the single
coda data, previous research on the production of English codas by
speakers of Vietnamese (Osburne 1996) has suggested that
Vietnamese speakers may perceive the articulation of III after a
vowel as a glide and therefore co-articulate the liquid with the vowel.
In the single coda data, it was found that the participants tended to
co-articulate the liquid with the vowel after all vowels with the
exception of the low back vowel. The CC data confirms this finding:
lit is typically co-articulated with the preceding vowel, especially if it
is a mid or low central vowel, and typically absent after a low back
vowel. In the cases when III begins to emerge in the CC cluster, it
emerges only after a mid central vowel. Interestingly, when the til is
co-articulated with the vowel, the second member of the coda is
typically produced correctly, possibly because the coda structure is
interpreted as a single coda.
Similar to the single coda data, III was also modified based on the
preceding vowel in the CC coda data: after front vowels and
diphthongs, it was produced as a rounded back vowel while after a
central or unrounded back vowel, it was produced as a nasal. In the
discussion of the single coda data, I suggested that these modifica-
tions were possibly due to the liquid being co-articulated with the
glide in diphthongs and the tense vowels. After lax vowels, the III is
articulated as [n], which is allowed syllable-finally in Vietnamese.
Interestingly, in the CC coda data this change is more common for
Anh than Nhi. While Nhi is able to produce codas with this liquid
correctly, for the most part, across time, Anh consistently modifies
the liquid.
Although there were no cases of accurate production of CCC codas
in the interview data, findings from the word list data suggests that LI
transfer effects also appear to constrain the development of CCC
codas, though it appears that Anh is more affected by LI transfer.
Nhi, on the other hand, is more likely to build on his emerging L2
repertoire, and appears to be less affected by LI transfer. One way in
which LI transfer effects appear to constrain the development of
CCC codas is in the higher production accuracy of codas which
contain consonants that also exist in Vietnamese, especially those
allowed syllable-finally in Vietnamese. This is similar to the effects of
LI transfer for the onsets and shorter codas. As with the C and CC
codas, the CCC codas that were produced the most accurately
comprised voiceless stops, nasals and voiceless fricatives and
affricates. For example, at time 3, both Nhi and Anh produced the
clusters /nst qst/ with 100 per cent accuracy in the word list data.
Additionally, the codas /Iks kst qks qtjt jst/, which have begun
emerging in Nhi's speech, but with lower accuracy percentages
(ranging from 20-33 per cent), all comprise voiceless stops, alveolar
fricatives, nasals, liquids and the voiceless affricate. While these
transfer effects are not straightforward, since CCC codas do not exist
in Vietnamese, it is significant that across all coda types and lengths,
the codas (and onsets) with which the participants have the least
difficulty are those containing consonants that also exist in
Vietnamese, and for the most part, are allowed in syllable-final
position in Vietnamese. Of course, it is also likely that the participants
are building on their emerging L2 repertoires after first making
equivalence classifications between the LI and L2, and in effect,
transferring those LI consonants into their emerging L2 phonologies
which are similar to L2 consonants, such as the nasals, and voiceless
stops, fricatives and affricates.
Another effect of LI transfer is that some CCC codas are produced
in a manner that conforms to an acceptable LI syllable structure.
This is more prevalent for Anh than Nhi. For example, Ann is more
likely to produce the CCC clusters as a single consonant and she is
also more likely to employ two types of modifications, for example
absence and epenthesis, in order to resyllabify the CCC syllable
structure into one which is acceptable in Vietnamese. This indicates
that she is relying on the LI, to some extent, in producing CCC
clusters. For example, for the cluster /t0s/, Nhi always produced two
consonants across time, though he does employ epenthesis at time 1
and 2. He initially produced the cluster as [. tes], which does indicate
some LI transfer effect because of the resyllabification of the coda to
a CVC structure. At time 2 and 3, however, Nhi has begun to
produce the cluster as [ts], indicating he is building upon his
emerging L2 repertoire, as he is beginning to produce this CC coda
accurately as well. Anh, on the other hand, has more difficulty with
the cluster. She typically produces a single stop, fricative, or affricate
in place of the CCC. This seems to indicate that overall, she has more
difficulty with longer, and more complex clusters, and favours a
Vietnamese syllable structure, even though she is in effect building on
her emerging L2 repertoire by producing the voiceless fricative
syllable-finally. These patterns exist across other CCC structures,
such as /sks sts/, with which both Nhi and Anh have difficulty.
However, while Nhi is able to produce two and even three consonants
in a row, Anh still tends to produce only one consonant, typically the
voiceless fricative. Again, it appears that while she is beginning to
acquire the voiceless fricative in syllable-final position, she still has
trouble with syllable structures beyond a single coda, an indication of
LI retention. Therefore, while LI transfer appears to affect L2
development, it does seem that Anh has more LI retention, overall,
than Nhi. In summary, LI transfer effects do appear to constrain the
development of those codas that contain consonants that exist in
Vietnamese, and can be modified, via absence and epenthesis,
towards an acceptable LI syllable structure.
The effect of LI transfer can also be examined for CCCC codas in
the word list data. Overall, the effect of LI transfer appeared to be less
for the longer codas than for the shorter coda structures. The
modification of the /kstG/ coda to a single voiceless stop, which is
allowed syllable-finally in Vietnamese, does indicate that at least for
this coda, LI transfer has an effect (however, this may interact with
developmental effects, as explained below). The typical modification
of this SFSF# to a single consonant resembles the participants'
production strategy for the FSF# and SFS# codas, and indicates that
the learners have difficulty building upon their emerging L2 reper-
toires for this coda as it has not been acquired as a CCC coda. The
SF# and FS# CC codas have begun emerging, but while they can be
produced accurately at times, they are often modified to a single
consonant. Therefore, it may be difficult for the learners to build on
their existing L2 repertoire for this coda as its component CC and
CCC codas are still in the process of emerging in the L2.
5. 2 Developmental effects
This study also found that a number of developmental effects
constrained acquisition, confirming prior findings that L2 learners of
English are subject to the same constraints as those for LI child
learners of English (Flege and Davidian 1984; Hancin-Bhatt and
Bhatt 1997; Hansen 2001, 2004; Hecht and Mulford 1982; Major
1986, 1987c; Mulford and Hecht 1980; Piper 1984). First of all,
onsets are acquired before codas: child LI learners (regardless of the
LI) acquire a CV syllable structure first, then VC, followed by CVC.
As evidenced by this data set, the same pattern may hold true for L2
learners: the participants produced onsets with a higher degree of
accuracy across time than codas. C onsets were produced more
accurately than C codas; additionally, CC onsets were produced
more accurately than C codas. Finally, CCC onsets were produced
more accurately than even C codas at time 3.
Secondly, regardless of syllable position or length of margin, in
general, front voiceless stops were produced more accurately than
back voiceless stops, i. e. l\i and /p/ were produced with a higher
degree of accuracy than /k/. The same was true for front voiced stops
and back voiced stops, /d/ and /b/ were produced with higher degrees
of accuracy than /g/. The same pattern existed for fricatives, in that
voiceless front fricatives were produced with greater accuracy than
back (relatively speaking) voiced fricatives, i. e. /z/ and /v/ were
produced more accurately than /3/. Additionally, voiceless front
fricatives such as /s/ and /f/ were produced more accurately than the
voiceless back fricative /J7. This pattern also extended to nasals: both
/m/ and /n/, front nasals, were produced with greater accuracy than
the velar nasal /q/. This follows the general developmental sequence of
acquisition of consonants described by Jakobson (1968) in that front
consonants are acquired before back consonants. Thirdly, also
regardless of syllable margin or margin length, voiceless consonants
appear to be acquired before their voiced counterpart, which is also
an effect of developmental processes as this follows the same pattern
as for child LI learners of English (Jakobson 1968). This is the case
across all stops, fricatives and affricates with the exception of /J7 and
/3/, where the voiced consonant in coda position had a slightly higher
accuracy rating at time 1 and time 2 due to Nhi's fairly accurate
production of this phoneme. However, at time 3, the voiceless
consonant has a higher overall accuracy rating than the voiced
consonant.
Fourthly, child LI learners of English also first acquire stops before
nasals, and nasals before fricatives, a pattern which the learners in this
study follow as well. Developmental effects may also explain the
relatively lateness of acquisition of the interdental fricatives /0/ and /5/
as well as the affricates /tj ds/, and the fricative /s/, which are also
acquired late in comparison with other consonants by native speakers
of English (Owens 1996).
Developmental effects may also explain the devoicing of some
voiced consonants that are not in the LI, which is also a common
developmental effect. For example, the learners in this study
commonly devoiced the voiced interdental fricative and affricate as
they began emerging in their L2. Production of the CVC syllable as a
CV structure is also a common developmental effect that occurred for
both participants as evidenced by the high rate of absence for
singleton codas within and across time (this will be addressed more
below in section 5. 3). Finally, CC and CCC (and CCCC coda)
cluster 'reduction' (or absence of one or more members of the cluster)
is a common process in child LI acquisition; this was commonly
evidenced in the data for both onsets and coda cluster. For FS#, SS#
and SF# clusters where both members have begun emerging in
singleton form, the pattern is to produce the first member - even
though in the SF# cluster this choice leaves a less sonorous coda.
Transfer does not appear to be a factor in these patterns since a
cluster like lip/ is produced as [1] even though /p/, and not /!/, is
allowed syllable-finally in Vietnamese. Instead, these patterns may be
indicators of developmental effects since these reductions mirror
those of child LI learners of English, possibly due to sonority for the
retention of the liquid in LS# and LF# and the fricative in FS#
clusters. In a discussion of child LI acquisition, Olmsted (1971)
notes that it is typically the second member of the CC cluster that is
affected either by deletion or modification (Olmsted's terminology),
and that 'frequently-retained members of consonant clusters are
consonants that are usually gotten correct when they stand alone'
(241). In her research on child LI learners of English, Ohala (1994)
also found that in FS# clusters, the F# is typically retained. This is
also confirmed for L2 learners by Osburne (1996). Some patterns of
CC onset deletion may also be due to LI developmental effects; as
Ingram (1976) notes, in #FG and #FL clusters, the #F is usually
retained and the L or G is deleted. This pattern was found for the
participants for the clusters /fy hy sw fl/, which were (at times)
produced only as the fricative.
In short, these findings suggests that while LI transfer effects
influence the development of the consonants that already exist in the
learners' LI repertoire, developmental effects may influence the rate
of acquisition and devoicing of the English consonants that do not
exist in Vietnamese.
5. 3 Markedness
The effects of markedness on the acquisition of syllable onsets and
codas can been found at a number of levels: a preference for a
universal syllable structure, a preference for shorter over longer
margins, the preference for some types of margins over others
regardless of length, the acquisition of longer margins only if its
constituents have been acquired in shorter margins, and the effect of
sonority. Each of these will be discussed in turn.
One area of markedness research is the universal preference for a
CV syllable structure, which is considered to be unmarked in relation
to other syllable structures such as CVC or CVCC, etc. Prior research
(Benson 1988; Osburne 1996; Sato 1984, 1985) on Vietnamese
learners of English has found that LI transfer effects exerted a greater
influence on the learners' production of English syllable structures
than a preference for a universal syllable structure. This study
confirms these findings in terms of the relative accuracy of final
consonant production for voiceless stops and nasals, which conforms
to the allowable CVC syllable structures in Vietnamese. This suggests
that Vietnamese syllable structure, and thus LI transfer, exerts a
greater influence on the English syllable structure production than
universal preference for a CV structure.
As noted earlier, there also appears to be a preference for
epenthesis as a modification strategy after the fricatives /f v z/, all of
which exist in Vietnamese but only in initial position. The effect of
the epenthesis is to create a new syllable in which the fricative is
placed syllable-initially. While this initially appears to indicate a
preference for a universal syllable structure as this creates a CV
structure, we must note that this preference only exists for the
fricatives that exist in Vietnamese in syllable-initial position.
Therefore, I suggest that this is an effect of LI transfer, rather than
a preference for a universal syllable structure.
The effect of length of onset and coda structures on production is
another area of markedness that has been investigated by L2
researchers. Longer onset and coda structures are considered to be
less marked than shorter onsets and codas. Thus, a coda ending in
only one consonant would be considered relatively unmarked in
comparison with a CC coda. Prior research (Anderson 1987;
Broselow and Finer 1991; Carlisle 1991, 1997, 1998; Eckman
1987, 1991; Hansen 2001, 2004; Sato 1984; Weinberger 1987) has
found that learners appear to have less difficulty with shorter
structures. This was found for both onset and coda structures in
this data set, regardless of participant and time, as outlined in Table
5. 1.
As Table 5. 1 indicates, the shorter syllable margins are produced
more accurately than the longer margins, and CC margins are
produced more accurately than CCC margins, with the following
hierarchy for both onsets and codas: C> CC> CCC. At time 3,
however, there is a slightly higher accuracy percentage for CCC
onsets than CC onsets; however, as the number of tokens the CCC
onset percentage is based on is very small (8 for Anh and 2 for Nhi),
the conclusion that CCC onsets are acquired before CC onsets is not
substantiated.
Additionally, the study found that learners modify longer onsets
and codas in favour of shorter, and less marked, onsets and codas.
The high percentage of absence across time for both participants
provides clear evidence that more marked (i. e. longer) onsets and
codas were modified, resulting in a shorter, and less complex, onset or
coda structure. Therefore, this study does confirm prior findings that
markedness, defined by length, is a factor in the production of syllable
codas.
Another area of investigation in terms of markedness is whether
markedness relations between onsets/codas of the same length but
different constituents have an effect on acquisition. As Greenberg
(1978) states, two-member clusters comprised of two consonants
from a similar category, for example SS or FF, are more marked in
relation to consonants from different categories, such as SF or FS,
which are considered unmarked in comparison. This was investigated
for codas (since #SS and #FF onsets are not allowable in English).
Overall, the data indicate that the participants were able to produce
the unmarked codas with more accuracy than the more marked
codas. This study found that LF#, NF#, SF# and LS# codas
emerged earlier than LL#, SS#, NS#, LN# and FF# codas, but only
if these clusters consisted of the more accurate consonants. If they
consisted of more difficult consonants, they had very low accuracy
percentages or were at 0 per cent. If the codas consist of the
consonants that emerged earlier and were produced more accurately,
they had a higher accuracy. For example, in all three data sets, the
voiceless FF# cluster had a lower accuracy rate than any of the
following clusters: voiceless FS#, voiceless (fricative) LF#, voiceless
SF#, voiceless (fricative) NF#. In most cases, the voiced FF# also
had a lower accuracy ratio than the voiced SF#, though it was
typically produced with greater accuracy than the FS#. Sequences
based on accuracy of production percentages could also be estab-
Table 5. 1 Onset and coda accuracy percentages by time, length and
participant
Time
Part.
C onset
CC onset
CCC onset
C coda
CC coda
CCC coda
Time 1
Anh
91%
69%
0%
45%
5%
0%
Nhi
90%
60%
0%
53%
9%
0%
Botii
90%
64%
0%
48%
7%
0%
Anh
93%
59%
60%
40%
4%
0%
Time 2
Nhi
94%
56%
0%
48%
7%
0%
Both
93%
58%
60%
43%
6%
0%
Anh
89%
63%
89%
49%
2%
0%
Time 3
Nhi
89%
76%
67%
47%
11%
0%
Both
89%
68%
83%
48%
6%
0%
lished for codas with stop consonants. The voiceless SS# coda was
typically produced less accurately than the voiceless SF#, voiceless
FS#, voiceless (stop) LS# and voiceless (stop) NS#. The voiced SS#
was produced with less accuracy than the voiced SF#, voiced (stop)
NS# though with less accuracy, overall, than the voiced FS#
sequence. In terms of the LN#, for all three data collection times,
they were produced with less accuracy than voiceless (fricative) LF#
and voiceless (affricate) LF#. In general, they were produced with
less accuracy than LL#. At time 1 and 2, they were also produced
with less accuracy than voiced (fricative) LF# and voiceless (stop)
LS#. For all three data sets, however, they were produced with more
accuracy than either the voiced (affricate) LF# or the voiced (stop)
LS# clusters.
In terms of how the constituents of the CC codas affected
production by type, for both participants, for example, the NF#,
SF#, LS# and LF# clusters that emerged first were /nz ts ks jd iz/,
with nasals, voiceless stops, voiceless fricatives and /d/ and /z/.
However, NF# clusters with the voiceless affricate such as /ntf/ and
LF# clusters with /v/ or the voiceless interdental fricative such as
/lv J0/ were at 0 per cent accuracy for both participants. Even the
more easily produced CC clusters had very low accuracy percentages
if they consisted of the more difficult consonants. Overall, the most
difficult CC codas for the participants were those consisting of the
least accurate consonants such as the voiced stops /b g/, voiceless
fricatives /J 0/ and voiced fricatives /5 3 d$/ either in combination with
each other or in combination with the more accurate consonants,
even in clusters such that have begun emerging with more accurately
produced consonants.
If we examine the word list data, the same patterns could be found
for CCC codas. Overall, the participants do have more difficulty with
the more marked codas. For example, both the LFS# and LSF#
codas had higher accuracy percentages than LFF#. Additionally, the
NSF# and the NFS# had higher accuracy percentages than NFF#.
In fact, the clusters with which the participants had the most difficulty
were LFF#, SFF#, NFF#, LLS#, NSS# and SSF#. All of these
clusters, with the exception of SSF#, which has an accuracy
percentile of 25 per cent at time 1, have 0 per cent accuracy of
production across time. These clusters have a common feature: they
all comprise two consonants of the same category in a row. Clusters
comprised of a FF# sequence seem particularly difficult for the
participants. In fact, with the exception of the LSS# cluster, which
has a 29 per cent accuracy percentage at time 3, all clusters with two
members of the same category are more difficult than clusters
comprised of a FS# sequence or SF# sequence.
There also appears to be an effect of length and coda type on the
acquisition of onsets and codas. Prior research (e. g. Carlisle 1998;
Eckman 1991) has indicated that learners may be constrained in
acquiring a longer syllable structure until they have acquired shorter
structures with the same constituents. For onsets, the most accurately
produced singletons were / mn t k b d g s f h wy j l / while /p J tj z/
were more difficult and /ds 5 0/ were the most difficult. A number of
CC and CCC onsets were also produced with a high degree of
accuracy: /st si sk sp vi tw py dy kw fl fy ny sm skw/ - not surprisingly,
all of these CC and CCC onsets consisted of the singleton and CC
(for CCC onsets) that were already produced with a high degree of
accuracy: / t k p d s f v l i wy n ms k kw/ (notice here that the CCC
onset /skw/ comprises /sk/ and /kw/ onsets, both of which are
produced highly accurately) (the exception here is /p/; however, /p/ as
the second member of a cluster is produced without aspiration, and
therefore as /b/, which is produced accurately in singleton form).
More difficult CC and CCC onsets were /da bj ti gi pi ki bl kl fa pi by
vy hy sw sti/. It is interesting to note that the CCC onset /sti/
comprises /st/ and /ti/, and as the latter of these two is not produced
highly accurately, neither is /sti/ although /st/ is highly accurate.
Finally, the most difficult CC and CCC onsets were /0i dsy spl spj/.
The CC onsets comprise the voiceless interdental fricative and the
voiced affricate, both of which are difficult to produce in singleton
form, which also suggests that these longer onsets cannot be acquired
before their constituents can be produced in shorter form. As for the
CCC onsets, they comprise /sp pi pi/ CC clusters, and while /sp/ is
produced accurately, the second CC cluster within each of the CCC
onsets, /pi/ for /spl/ and /pi/ for /spj/ are only beginning to emerge
(they are produced with a mid level range of accuracy), again
indicating a ordering of acquisition based on length and type of
constituent. Therefore, in the case of onsets, there are no examples of
longer clusters (CC and CCC) that are produced with any degree of
accuracy before their constituent C and CC members have emerged.
For codas, there also appears to be an effect of markedness based
on coda length and type: the CC codas produced with the greatest
accuracy comprise those consonants that are produced with a higher
level of accuracy in single coda form. Additionally, the CC codas
comprised of consonants produced with 0 per cent accuracy as
singleton codas are also produced with 0 per cent accuracy, perhaps
indicating that consonants comprising CC codas must first be
acquired in single coda form before they can be produced as
constituents in a CC coda.
Additionally, in most cases, the CCC codas that were produced
correctly in the word list data were those whose constituent
consonants had begun emerging in coda position in both singleton
and CC codas. For example, the CCC coda /qtjt/ has begun emerging
in Nhi's speech. This coda was produced with 25 per cent accuracy
for Nhi at time 3, (0 per cent for Anh). In CC form, Nhi produced
/qtj/ with 100 per cent accuracy and /tjt/ at 40 per cent at time 3,
while in single coda form, he produced /n/ with 89 per cent accuracy,
I I I with 76 per cent accuracy, and /t|7 with 57 per cent accuracy at
time 3.
Overall, Nhi is also able to produce more CCC with higher degrees
of accuracy than Anh, possibly because he is able to produce the
constituent CC clusters with greater accuracy. For example, at time
3, Nhi also produced /jmz/ with 100 per cent accuracy. Additionally,
the clusters /mps ikt/ are produced with 67 per cent accuracy and /sks/
with 60 per cent accuracy. These codas are comprised of two-
member codas which Nhi also produces with a higher degree of
accuracy at time 3: /im/ - 100 per cent; /mz/ - 83 per cent; /mp/ - 100
per cent; /ps/ - 100 per cent; /kt/ - 25 per cent; /sk/ - 67 per cent; /ks/
- 50 per cent.
Therefore, it appears that there is an order of acquisition in that CC
onsets and codas are not produced accurately until their constituent
members have begun to be acquired as singleton onsets or codas.
Additionally, CCC onsets or codas are not produced or acquired until
their constituent singleton and CC onsets or codas have begun to be
acquired. This confirms research by Eckman (1991), who also found
that in 98 per cent of the cases of his data, if the CCC coda had been
acquired, then the CC codas that comprised the CCC coda had
already been acquired. This is also supported by research on onsets
by Carlisle (1998).
We can also extend the comparison between the production of C
and CC clusters with the production of CCC clusters to examine the
resulting onset or coda when a CCC onset or coda is modified. In
most cases, when modification occurs, the result is a CC onset or
coda that has begun being produced with some accuracy. For
example, the coda /ndz/ is commonly produced as [nz] rather than
[nd] even though the final consonant is a plural marker, and thus a
separate morpheme. The NF# cluster is produced with some degree
of accuracy while the nasal-stop cluster has 0 per cent accuracy across
time.
Finally, there also appears to be an effect of markedness defined by
sonority in terms of acquisition of onsets and codas and absence of
members within the onset or coda structure. As Carlisle (1997) states,
absence may be more dominant in margins that violate the Universal
Canonical Syllable Structure (UCSS). Margins meet UCSS 'if the
sonority of segments... increase [s] as the syllable nucleus is
approached' (Herbert 1986: 62). The order of sonority for conson-
ants is as follows, from the most sonorous to the least sonorous:
liquids > nasals > fricatives > stops. For example, a #FL onset or a
NFS# coda would not violate the UCSS since the syllable onset
increased in sonority towards the nucleus while the coda decreases in
sonority in a rightward direction, i. e. moving away from the syllable
nucleus. On the other hand, the cluster NSF# would violate the
UCSS since there is an increase in sonority from the stop to the
fricative consonant. As previous research has indicated, learners tend
to have less difficulty with clusters that do not violate the UCSS
(Carlisle 1991; Tropf, 1987). Since many CC and all CCC onsets in
English begin with a fricative, and usually /s/ - creating such onsets as
#FS and #FSG - onsets typically violate UCSS and therefore this
discussion will focus on codas.
First of all, CC coda acquisition patterns appear to be partly
explainable based on markedness and sonority: the more sonorous
Consonant + Fricative# codas appear to emerge before the less
sonorous Consonant + Stop# coda (e. g. NF# before NS#, SF#
before SS#). The less marked LF#, NF#, SF# and LS# codas appear
to emerge before the more marked FF#, SS# and LL# codas, and the
less marked voiceless codas before voiced codas, which corroborates
previous research (e. g. Major 1987b). Interestingly, this study also
found that SF# codas, which violate sonority constraints, were
produced fairly accurately - a finding that corroborates Major (1996),
who found that SF# codas emerged before both FS# and FF# codas,
which he explains by stating that the /s/ is extrasyllabic and not
attached to the syllable node and therefore this coda does not violate
the UCSS. However, this high production accuracy may also be due
to grammatical conditioning, which will be discussed below.
Similar patterns emerged for CCC codas, which were typically
produced as a C or CC coda that had begun emerging accurately or
with epenthesis: LFS# codas were produced as LF# as was the LSF#
coda. While the production of both CCC codas as LF# resulted in a
more sonorous coda, the production of the LSF# coda as a LF# also
resulted in a change from a coda that violated the UCSS to a form
that did not, which confirms prior research that has found that
margins that violate the UCSS are typically produced as a form that
does not violate the UCSS by L2 learners (cf. Abrahamsson 2001;
Hansen 2001). Similarly to the CC codas, all the codas with nasals -
LNS#, NSF#, NFF# - were produced as the nasal, regardless of
whether the nasal was the second or first segment, possibly also a
sonority effect.
This study also found that with only one exception, /Iks/, which
Nhi produced accurately, all the clusters with which the participants
had the least difficulty in the CCC codas did not violate the UCSS.
In other words, the codas that emerged first did in fact meet the
UCSS. Additionally, in cases where CCC violated the UCSS, the
codas were often modified in favour of a coda that did meet UCSS.
For example, both LSF# and NSF# codas tended to have absence of
the stop consonant, even though the fricative is a bimorphemic plural
marker and the stop was part of the monomorpheme. This resulted
in a LF# and NF# coda, respectively, both of which do meet the
UCSS.
These findings can also be corroborated by examining the
production of CCCC codas from the word list data. There does
also appear to be an effect of sonority on member absence for the
CCCC coda, similar to the CCC codas. Both the SFSF# and the
LLSF# codas violate the UCSS. When produced, the LLSF# is
typically modified to a LF# or LLF#, both of which do meet the
UCSS. Similarly, the SFSF# coda was produced as a single stop,
which also meets UCSS. As for the LFFF# cluster, the most marked
member, the interdental fricative, is the member which is either
modified or absent, which may also be a markedness effect.
5. 4 Linguistic environment
For both onsets and codas, linguistic environment had an effect on
production. These will be discussed in turn for each type of
production and margin type. Across all codas, regardless of length,
type and participant, preceding linguistic environment had a signifi-
cant effect on both target-like production and absence due to the
influence of a preceding diphthong and mid back vowel /o/, which
both strongly promoted absence and strongly disfavoured accurate
production. This is due to LI transfer as in Vietnamese, a consonant
is not allowed after a diphthong in syllable-final position. Osburne
(1996) explains that in Vietnamese, the diphthong is realized as a
vowel plus glide rather than as a single segment, and the glide is
viewed as a final consonant, closing the syllable. Both Osburne
(1996) and Benson (1988) found that Vietnamese speakers fre-
quently did not produce a final consonant after a diphthong; this
study confirms these results. This brings up the question of the effect
of /o/, which also promoted absence. Vietnamese does have this
vowel, but as Santry (1997) states, before consonants, this vowel is
actually shorter than its English counterpart. It is therefore possible
that relying on perceptual cues, the participants are classifying the
English /o/ as a diphthong as it appears to be longer in duration than
English /o/ as a diphthong as it appears to be longer in duration than
diphthong. This explains the high rate of absence of consonants
following /o/.
There was also an interesting phenomenon of absence and
epenthesis after the tense vowels /iy/ and /ey/ and the diphthongs
/ay/ and /au/, and their epenthesis in open syllables, primarily before a
pause. As Santry (1997) notes, the tense vowels /iy ey/ are actually
categorized as diphthongs in some descriptions of Vietnamese and
are typically produced as lax vowels in closed syllables and as
diphthongs in open syllables. Similarly to the longer duration of/o/ in
English, the longer duration of English /iy ey/ in closed syllables
versus their Vietnamese counterparts may lead them to be perceived
as diphthongs by the learners, which would then promote absence as
well.
For onsets, these sounds - in this case as a following linguistic
environment - also significantly affected target-like production as well
as feature change, although not absence. However, conversely to the
effect on codas, the vowel /o/ as well as diphthongs promoted target-
like production while a following vowel promoted absence. I suggest
that these patterns exist due to the effect of these vowels on the coda:
since diphthongs and /o/ promote absence in codas, they promote
'retention' or target-like production of onsets because otherwise the
syllable structure would have no margins and communicative efficacy
would be lost. In terms of the effects of following linguistic
environment on feature change, a following vowel promoted this
type of production in comparison to diphthongs and /o/. This pattern
is understandable in light of the findings for the effect of this linguistic
environment on target-like production: as diphthongs and /o/
promote target-like production, they consequently inhibit feature
change.
Following linguistic environment also had a significant effect on
production of codas, particularly on feature change and epenthesis.
For example, a following pause or glide promoted epenthesis while a
following vowel and consonant disfavoured epenthesis. The higher
rate of epenthesis before a following pause has been found for other
L2 learners (Tarone 1980). The question then emerges as to what
extent the high rate of epenthesis that was found for the fricatives
/s f z v/ and which were postulated to be due to LI transfer were
actually caused by the linguistic environment of a following pause. As
shown in the VARBRUL analysis of singleton coda production,
however, linguistic environment is not a significant factor for /s f z v/;
therefore, whether these fricatives are followed by a pause, vowel,
consonant or glide does not significantly affect the rate of epenthesis.
The following linguistic environment also had a significant effect
on feature change, with a following pause and glide promoting feature
change, typically devoicing before a following pause, which has been
noted in prior L2 research (Edge 1991; Major 1987b) and is a
common phenomenon for native speakers as well. The effect of a
following glide on both epenthesis and feature change is interesting;
when an analysis was done to determine why these patterns existed,
however, it was found that the number of tokens for these environ-
ments were very small - 9 per cent (20/359) of the following
environments for epenthesis and 5 per cent (33/642) of the following
environments for feature change were a glide and therefore there may
not be a significant pattern. Finally, following linguistic environment
was also statistically significant for the velar nasal, which is not
surprising due to its high rate of nasal assimilation to the following
consonant.
This environment - in this case the preceding linguistic environ-
ment and therefore limited to a pause, vowel, and consonant - was
only significant for epenthesis for onsets. A preceding consonant
promoted epenthesis while a vowel did not (pauses were eliminated
from analysis as there were no cases of a preceding pause before
epenthesis). It appears that if the preceding word/syllable has a coda,
the participants employ epenthesis to break up the sequence of
consonants resulting from a coda followed by an onset.
Stress was also a significant factor for codas for epenthesis as well as
feature change, with a stressed syllable favouring epenthesis and an
unstressed syllable disfavouring epenthesis and an unstressed syllable
favouring feature change. The employment of epenthesis for a
stressed syllable creates an unstressed syllable, which results in a
stressed-unstressed rhythm, which is the stress pattern of Vietnamese
(Nguyen 1970). The employment of feature change in unstressed
over stressed syllable may be due to the lack of salience of the final
consonants and therefore greater likelihood of assimilation to the
following consonant.
For onsets, stress was significant for target-like production,
epenthesis and feature change, with opposite patterns from those
found for codas: an unstressed syllable promoted epenthesis while a
stressed syllable promoted feature change. These patterns are best
explained in light of the coda findings for stress, as explained above. It
appears that since a stressed syllable promotes epenthesis for codas,
epenthesis would not occur also for the onset in a stressed syllable; for
feature change, since an unstressed syllable favours this type of
modification in the coda, it is inhibited in the onset. Finally, an
unstressed syllable promoted target-like production; this may be due
to the same reasons: while not significant for codas, there is a
tendency for stressed syllables to promote target-like production.
Therefore, it appears that the same type of production modification
does not appear to occur in both the onset and the coda in the same
type of syllable; therefore, if a production type is promoted for the
coda, it is inhibited for the onset and vice versa.
5. 5 Grammatical conditioning
There were some very interesting patterns with the effects of
grammatical conditioning since for singleton codas and CCC
codas, absence was promoted while for CC codas, absence was
disfavoured for the bimorphemic coda. Additionally, it was found
that for the singleton codas, the only pattern that could be discerned
for the differences in the production of inflectional morpheme types
was that bimorphemic /d/ had a much higher accuracy percentage
(e.g. 43 per cent) over monomorphemic /d/ (II per cent). The plural
/z/ was about evenly target-like produced regardless of whether it was
morphemic (10 per cent) or bimorphemic (11 per cent). One
interpretation is that final /d/ is salient as a morphological marker -
especially in a singleton coda - and that the participants pay more
attention to it in production, whereas the plural form - perhaps due to
redundancy in the syntax - is not as salient. As for the differences
between grammatical conditioning and length, it did appear that for
the CC codas that emerged with some target-like production, e. g.
/nz iz ts/, the final segment was a morphological marker - in this case,
a plural marker. This does appear to indicate that the plural form is
emerging in the participants' grammatical system and that there is a
morphophonemic interface in their production of codas. This may
also help to explain the common production of the CCC codas to a
shorter form that typically, though not absolutely, had the final
consonant - especially when it was a plural marker (e. g. /idz/ to [iz];
/Idz/ to [lz]). The question then remains as to why this marker does
not have a higher accuracy percentage in singleton form; however, the
fact that /z/ is only beginning to be produced in singleton form, and
that it has an almost equal percentage in both monomorphemic and
bimorphemic form, does indicate that the plural morpheme is
emerging and produced across all codas, regardless of length.
5. 6 Task variation
Another important finding was that data type had an effect on
production. Overall, there was a greater accuracy on the reading data
(word-list and reading passage) compared with the interview data.
This confirms prior findings (Dickerson and Dickerson 1977;
Gatbonton 1975; Major 1987c; Sato 1985) that learners have a
greater accuracy in more formal speech styles such as reading a
passage, than for more casual styles, which may be the more
predominant speech style in a relaxed interview. There were several
other interesting findings that emerged from this analysis: first of all,
the reading data tended to promote both accurate production and
feature change while the interview data promoted absence and
epenthesis. As for the former, reading a list or passage tended to make
the participants pay closer attention to their speech, which therefore
promoted a greater accuracy in production. However, reading a list or
passage can also create 'reading errors' in the form of orthographic
miscues. As English does not have a direct one-on-one correspond-
ence between orthographic and phonetic representation, reading
errors are common for second language learners and this explains the
greater number of feature change errors for this type of data.
The interview data promoted greater absence and epenthesis than
the reading data in part because the participants were producing
conversational speech, with concern for meaning rather than accur-
acy, for the most part. The many hesitations and pauses in the
interview data also promoted epenthesis.
Furthermore, there were differences between Anh and Nhi in
production for the two data sets across time. Anh had a U-shaped
curve for accuracy for both types of data while Nhi only had a U-
shaped curve for the reading data; his interview data accuracy
decreases across time. In fact, by time 3, Anh had a higher level of
accuracy in the interview data, though only by 1 percentage point,
than Nhi. I want to tentatively suggest that this indicates two things:
first of all, Nhi and Anh are two very different types of language
learners, and while Nhi has greater linguistic accuracy overall, he is
also more careful in monitoring his speech, to the point that he
appears disfluent because of his hesitations and pauses. Ann, on the
other hand, speaks in words rather than full sentences. However, she
usually has the right vocabulary word and appears to comprehend
what others say; therefore, while she in fact speaks with less
grammaticality overall than Nhi, she appears much more fluent
because she does not hesitate or repeat herself as she speaks and she
has a large vocabulary base. When the reading passage was taped, Nhi
took a while to read through the passage and carefully articulated his
sounds; Anh, on the other hand, rushed through it, and I had to ask
her to slow down. These two different reading and speaking patterns
give some indication that they approach learning in different ways:
while Nhi is more concerned with accuracy, Anh is more concerned
with overall understanding.
There also appear to be different production strategies based on the
data type, with higher accuracy percentages on the reading data
across all coda types. This confirms prior research findings
(Dickerson and Dickerson 1977; Gatbonton 1975; Major 1987c)
that careful, monitored speech, which may be elicited on a reading
task, promotes accurate production in contrast to a more casual and
informal speech style, which may be a more indicative speech style of
the interview data. For example, while the participants are able to
produce the nasal-voiceless affricate cluster /ntJV 100 per cent
accurately across all three reading data sets, both tended to only
produce the nasal in the interview data. In other words, while they
could produce the nasal-voiceless affricate in careful, monitored
production, they typically produced only the nasal in casual,
unmonitored speech. In another example, both Anh and Nhi were
able to produce the /itJV coda accurately at time 3 in the reading data.
In the interview data, however, they produced only the [j] at time 3.
Interestingly, the coda was not modified - rather absence, again of the
second consonant, was common. This pattern occurred for many of
the codas that were elicited in both the reading and the interview
data. While the reading data indicate the stages of development of the
codas, including absence of either the first or second member of the
coda as well as epenthesis and feature change, the interview data,
especially at time 3 where there were more CC tokens in the interview
data, had a consistent pattern of absence of the second member. This
absence of the second member appears to be a production effect,
which like the effect of linguistic environment on coda production,
appears to be a secondary constraint on production. That is, while the
coda structure may be acquired and exist in the learners' underlying
grammar and they are able to produce it in careful, monitored speech,
casual speech promotes absence of the second member of the coda,
for a number of reasons. First of all, the learners are not monitoring
their speech in the interviews and thus not as careful in producing the
codas. Secondly, the absence of a consonant in a cluster is a natural
speech phenomenon for native speakers of English as well. As Celce-
Murcia, Brinton and Goodwin (1996) state, in connected speech
such as the interview data, deletion is a common phenomenon. One
motivation for these adjustments in connected speech is ease of
articulation. It appears that, at least to some extent, the differences
among the production accuracy in the reading and interview data sets
can be explained by natural phonological processes in connected
speech, rather than being an indicator that the final member of the
coda has not been acquired.
5. 7 Interaction of linguistic constraints
To summarize, the emergence of syllable margins was sequenced in a
consistent manner across time, participants and tasks. The sequence
is constrained not only by LI transfer, developmental and marked-
ness effects, but also by grammatical conditioning. The learners
initially transferred the syllable structure of their LI, Vietnamese, into
the L2, and this becomes the basis of their emerging L2 phonology.
Segments that are allowed in syllable-initial position emerge in that
position first; this is followed by the emergence of syllable-final
margins, for those consonants that are allowed syllable-finally in
Vietnamese. Consonants that are in the LI but only in syllable-initial
position are initially produced in light of the LI and the current state
of the L2. As the L2 phonology continues to develop, these
consonants do begin to emerge in syllable-final position, but
developmental and markedness effects constrain the ordering of
emergence (voiceless over voiced, front over back consonants).
Consonants that are not part of the LI phonology are initially
represented as a consonant that is allowed in Vietnamese. As the
phonology develops, these consonants then begin to be produced as
the closest consonant that has already begun emerging in the L2
based on manner or place of articulation. The CC and CCC onsets
and codas are initially produced as one of the emerging singleton
consonants, usually with the plural marker. Along the way, the
participants are constrained in both their acquisition (e. g. the effects
of a preceding diphthong and mid back vowel) and production of the
L2 by linguistic features such as inflectional morphemes, preceding
and following linguistic environment and syllable stress.
In terms of specific onsets and codas, the following patterns were
found. Stops, nasals, glides, liquids, some voiceless fricatives and /v/
were initially allowed in singleton onset position in the learners' L2
phonologies as this was transferred from Vietnamese. Nasals and
voiceless stops transferred from the learners' LI phonology to form
one of the bases of the L2 phonology's syllable-finals as they are
allowed syllable-finally in Vietnamese. While these codas did not have
100 per cent accuracy percentages across time, the variation in their
production can be explained by a number of factors. All these
consonants had statistically significantly higher absence rates after
diphthongs, tense vowels and the mid back vowel /o/ as well as
statistically significantly higher rates of epenthesis before a following
pause. The rates of absence for /t/ are also due to grammatical
conditioning as the past tense was in the process of being acquired
and therefore there were higher rates of absence when /t/ represented
a past tense marker. The high rate of feature change of/q/ was also not
due to lack of acquisition of this consonant but rather to the natural
phonological process of nasal assimilation to the following consonant,
a process that also explains the feature change of /m/ before alveolar
consonants.
The voiced stops /b d g/ are initially represented in the underlying
grammar as /p t k/ syllable-initially, respectively, as these latter
consonants are part of the L2 phonology, or are absent, especially in
the case of /g/ since /k/, its voiceless counterpart, is just beginning to
emerge across time. The higher rate of absence of/d/ can be explained
by grammatical conditioning. The voiceless fricatives /s fl are initially
represented in the underlying grammar and produced with a high rate
of accuracy in syllable-initial position as they are allowed in that
position in Vietnamese but with epenthesis syllable-finally which is a
transfer of syllable structure from Vietnamese (following linguistic
environment was not a significant factor in epenthesis for these
consonants). Absence rates can be explained by the effect of a
preceding diphthong. The phoneme /J7 does not initially exist in the
underlying grammar due to its different place of articulation in
Vietnamese and its representation in the grammar is word-specific
(usually as /s/ except for in the word English when it is represented as
/k/). The same is true for /t|Y. The voiceless interdental fricative /6/ is
beginning to emerge in the underlying grammar, particularly in onset
position, although it is also at times produced as /t/ or /s/. The voiced
fricative /z/ began emerging in the L2 phonology although it initially
had a high degree of feature change in initial position and absence in
final position, the latter due in part to grammatical conditioning
(absence rates were higher when it was a morphological marker). As
absence rates decreased, epenthesis increased as Izl began emerging
syllable-initially, thus conforming to an emerging onset syllable
structure. This same pattern was found for /v/, which also initially had
a higher absence rate, and then an increased epenthesis rate as it
began emerging syllable-initially. The voiced affricate Afe/ was initially
represented in the learners' grammar as /t k/ (there were too few
tokens to determine any patterns) and later as /tf/ as the voiceless
affricate began emerging. Finally, the liquids were present in the
learners' underlying grammar and produced accurately in onset
position, but they were only accurately produced in coda position in
some linguistic environments, as they were constrained both in
development and in production by LI phonotactics.
Linguistic environment also had an effect on onsets and codas. The
strongest effect of linguistic environment was the effect of a preceding
diphthong on the coda as this environment consistently promoted
absence while inhibiting accurate production; it had the opposite
effect on onsets. As discussed above, this can be explained as an effect
of LI transfer as Vietnamese does not allow final consonants after a
diphthong. As the data indicates, there was an especially strong effect
of a preceding phoneme on the production of a liquid, which can also
be explained by LI transfer, as discussed above.
In terms of explanations for these patterns, which are consistent
across time, participant and syllable margin, as well as length of
margin, there are a number of linguistic factors that affect the
acquisition of the learners' L2 phonology. First of all, confirming
prior research (Altenberg and Vago 1987; Benson 1988; Broselow
1987; Flege and Davidian 1984; Hancin-Bhatt and Bhatt 1997;
Hodne 1985; Major 1987a; Major and Faudree 1996; Sato 1984;
Tarone 1980, 1987; Vago and Altenberg 1977; Weinberger 1987),
LI transfer effects appear to be a primary factor in the development of
onsets and codas, especially at the early stages as may be evidenced at
time 1. LI transfer effects influence the development of the
consonants that already exist in the learners' LI repertoire, in that
the consonants that are allowed syllable-initially and syllable-finally in
Vietnamese have a high accuracy percentage in their respective
syllable position. In fact, for all onset and coda structures, onsets and
codas with these elements are produced more accurately than any
other margins. Also, LI transfer affects the emerging syllable
structures of the L2, as the participants employ different production
modifications such as feature change, epenthesis and absence, in
order to modify L2 onsets and codas to an acceptable LI syllable
structure. Those consonants that exist in Vietnamese in syllable-
initial position are typically epenthesized in syllable-final position,
resulting in an acceptable Vietnamese syllable structure. Finally, LI
transfer also affects the production of final consonants as a preceding
diphthong promotes absence, as in Vietnamese, and liquids are
modified, or co-articulated, based on the preceding vowel. However,
LI transfer, while still a dominant factor in the production of both C
and CC syllable structures, does not appear to have as strong an effect
on the development of CCC and CCCC structures. The exception to
this is Ann, who appears to still rely on the LI to some extent in
modifying the longer clusters, indicating that LI retention is more
common for her than for Nhi.
Developmental effects and markedness also appear to constrain the
development of English syllable margins for speakers of Vietnamese.
Developmental effects may influence the acquisition and production
of the English consonants that do not exist in Vietnamese. Similar to
previous research, (e. g. Flege and Davidian 1984; Hecht and Mulford
1982; Major 1987a, 1994; Mulford and Hecht 1980; Piper 1984),
this research project found that developmental effects promoted
obstruent devoicing and fricative 'stopping', and that developmental
effects were more dominant as transfer effects decreased (cf. Major
1987a, 1994), as illustrated by Nhi's data set. Additionally, this
research also found that developmental effects constrained the
sequence of development of English consonants as the participants
appeared to acquire English consonants in the same sequence as
children learning English as a native language, in that front
consonants were typically acquired before back consonants and
voiceless consonants before voiced consonants (cf. Jakobson 1968).
Furthermore, those consonants which are acquired later for children
learning English as a native language, such as the interdental fricatives
and voiced affricates, were also acquired later for the participants,
which is also indicative of developmental effects.
Markedness may affect the development of onsets and codas based
on a preference for a universal syllable structure (CV), length, as well
as length and type, and sonority. The CV syllable structure is
acquired before a CVC or VC structure. Additionally, shorter, and
thus less complex and less marked, onsets and codas are produced
more accurately than the longer, and thus more complex and more
marked, onsets and codas. A clear developmental sequence can be
established: C > CC > CCC > CCCC (the latter for codas only, as
based on the word list data). Additionally, longer onsets and codas
are typically modified in favour of shorter, and less marked, onsets or
codas, confirming prior research (Anderson 1987; Broselow and
Finer 1991; Carlisle 1991, 1997, 1998; Eckman 1987, 1991; Sato
1984; Weinberger 1987). This can be analysed even further as it is
also clear that the type of consonant within each margin combined
with length has an effect on development: a longer margin does not
appear to be produced with some accuracy until shorter onset or coda
structures containing the same consonants have begun being
acquired. This also confirms prior research (Carlisle 1998; Eckman
1991). Additionally, if modified, a longer onset or coda is usually
modified in favour of a shorter onset or coda that has begun to be
acquired.
There also appears to be some effect for sonority in terms of which
codas are more easily produced and how a coda is modified: codas
which violate the Universal Canonical Syllable Structure (UCSS) are
produced with less accuracy than those which do not. Additionally,
these codas are also often modified in favour of a coda that does meet
the UCSS, confirming prior research by Carlisle (1991) and Tropf
(1987). Ease of articulation also appears to be a major factor in the
production of longer codas, which are often reduced to a less
articulatorily complex coda (one which has been more easily acquired
as a shorter coda), by the elimination of the medial consonant or
more marked consonant. The consonant that violates the UCSS is
typically the one that is absent. In codas which do meet the UCSS,
the medial consonant, often the least perceptually salient, is typically
absent, since the consonants at the edge of the coda are typically more
auditorily salient.
In terms of the role of grammar in the development of the codas, it
appears that if the bimorphemic grammatical marker is analysed
grammatically, it is typically absent. This is especially the case in
single codas. However, once the grammatical marker has begun being
acquired, it appears to be analysed phonologically, and as the last
member of the coda, it is typically produced while the medial member
may be absent. Overall, it does appear that the learners have acquired
the plural marker, and are in the process of acquiring the past tense
marker.
Finally, it appears that there are two levels of constraints operating
on the development and production of syllable margins. There are
primary linguistic constraints such as LI transfer, developmental
effects and markedness, which appear to operate on the development,
or acquisition, of the syllable margins across time. These constraints
are considered to be primary since they affect acquisition and
development, i. e. which consonants are acquired, in what order, and
how consonants are modified across the stages of development. There
are also secondary effects, which are production effects. They include
the effect of preceding linguistic environment (which is also an effect
of LI transfer) and data type. These effects are considered to be
secondary since they appear to operate on the onset or coda after the
primary effects have already affected the onset or coda structure. For
example, even though a coda structure may be produced accurately in
the reading data, and thus appear to be acquired to some extent,
secondary effects such as a preceding diphthong, or data type (i. e.
connected, unmonitored speech) may promote absence, or modifi-
cation of the onset or coda which otherwise could be produced
accurately.
The question of which social constraints affect the development of
syllable onsets and codas and to what extent remains to be addressed.
This is examined below in the analysis of the social data.
6
Social Barriers
A description of the overall social context and factors that affect each
participant at all three linguistic data collection times will be
discussed, followed by an analysis of specific social constraints such
as social identity, social interactions, acculturation, affective variables
and investment. Throughout this section, verbatim quotes from the
interviews will be used in order to both present the participants' own
voices and also to provide more examples of their language use at
each of the linguistic data collection times.
6. 1 Social contexts of language use across time
6. 1. 1 Spring
At the beginning of the study, Ann's interactions in English are fairly
limited as she is surrounded by Vietnamese work mates at her job in
the nail salon. Even though her customers are not Vietnamese, the
majority are Mexican, and like Anh herself, have limited English
communication skills. While she talks with her customers (in fact, her
boss requires all the employees to converse with their customers while
doing their nails), her communication is restricted as she has difficulty
expressing her ideas, and also understanding the English her Mexican
customers speak. She also has an additional difficulty - as she is new
at being a nail technician, she also experiences problems in perform-
ing her job well. Coupled with the need to speak English, the work
environment is pressure-filled for Anh, and she is often stressed about
her lack of English skills. She is also under increasing pressure from
her boss to speak more English with her customers, and this pressure
fuels her anxiety, culminating in her switching nail shops in late April.
As she says:
I have just nail technician because my owner talk talk me and why
why you don't talk customer... are you nervous... are you afraid
customer... sometime I... yeah... because I am talk a little bit
... speaking English with customer... sometime I... I made in...
SOCIAL BARRIERS 127
I make I make no good... no good... I... control machine... yes
because I worry... I nervous in talk customer.
She often feels frustrated at work, partly because of the pressure from
her boss to have 'small talk' with the customers. Ann must talk to her
customers about nail diseases and fungi, but does not feel that she has
the vocabulary or the grammatical structures to communicate her
ideas. She forces herself to try to communicate with them, but her
experiences are not usually successful because:
I don't understand American with a customer because I talk...
vocabulary no sentence no structure... I can't explain I can't
explain their understand nail and disease... I speak vocabulary...
no sentence no structure sentence... customer don't understand
and I don't understand them... I think it is bad experience.
These experiences are very depressing and discouraging for Anh, and
she breaks down crying when she relates them. Her strategy is to
memorize sentences to use with her customers, and to write down her
sentences so her customers can read them instead. This latter
practice, however, is strongly discouraged by her boss, and she is only
able to use it when her boss is not around. When she talks with her
customers, she says she understands about 50 per cent, especially if
they speak slowly.
Anh craves a supportive and stable work environment so she can
relax and spend her energy on improving her English and learning
more about life in the US. Contrary to Nhi, however, this is not easy
for her to find. Most of her co-workers are young, and have come to
Tucson from California in order to seek work in the nail salons. They
change jobs quickly, moving to the nail salon where the money is best.
This leaves Anh feeling a little lost, since her co-workers, and her
potential support network, change constantly. In fact, at this point in
time, Anh does appear to be struggling not only in developing her nail
technician skills and English communication ability, but also in
finding a place for herself in this new society. While she wants to learn
English and to interact more with Americans, her opportunities are
extremely limited and appear to be reachable only in the distant
future when her English has improved. Therefore, interacting with
Americans is not a means to improving her language skills, but rather
a goal once her language skills have strengthened. Instead of putting
her energy into creating more opportunities to use English with native
speakers, which seems hopeless, Anh instead focuses on developing
social ties inside her work environment. In effect, she attempts to
acculturate not to the American style of life, but rather to the life of
the Vietnamese immigrants in the US who, for the most part, are
immersed within a Vietnamese community and way of life. As such,
she changes her clothing style to fit in with her co-workers, favouring
the short skirts and cropped tops that the young Vietnamese women
who are 20 years younger than her are wearing, and joining their
frequent social gatherings such as trips to Las Vegas and the many
casinos in the Tucson area. She is adapting to the Vietnamese lifestyle
in the US rather than the American lifestyle.
At the beginning of the research project, Anh was enrolled in an
intermediate listening and speaking ESL class at Pima Community
College, which provided her with opportunities to interact with non-
Vietnamese speakers, though to a limited extent as the class was fairly
large (33 students) and only held for 40 minutes a day, Monday to
Thursday. In fact, when asked to list her friends in the US, Anh
names three of her classmates, one from Indonesia, one from China,
and one from Mexico, as her closest friends in the US. She typically
speaks with her friends for 5-10 minutes a day, before and after class.
She desperately wants more chances to speak English and more
American friends, but does not know how to meet more Americans
since her social interactions are highly restricted. To her, the key to
success in the US is speaking English well and while she does not feel
that her English is good, she is happy she came to the US.
At home, she tries to improve her English by watching TV for an
hour a day with captions, stating that without captions she would not
be able to understand anything. With captions, she understands
about 20-30 per cent. At work, there is always a film on the TV to
entertain customers while they wait for a technician. She therefore
also watches TV when she is waiting for customers, but does not
understand very much. She feels that she speaks English about the
same as other Vietnamese, but much worse than native speakers,
especially in terms of fluency, pronunciation and knowledge of
American culture.
Nhi's social situation differs markedly from Anh. He is very happy
in the US, and unlike Anh, he does not feel the same level of stress
that she does, partly due to the different natures of their work
environments, and also different personalities. While Nhi also wants
more opportunities to speak English, he does have a fairly supportive
work environment in terms of English language practice, unlike Anh.
He has four good friends at work, two of whom are American while
the other two are Mexican. He talks with them everyday during their
break times and lunches, and as he says, 'they teach English... if I if
SOCIAL BARRIERS 129
I speak wrong they correct for me. ' They teach him job terminology
and they often joke during breaks, as he says, ' when... break
time... we have we talk we talk together funny. ' While one of his
Mexican co-workers speaks English better than him, and therefore is
easy to understand ('Jose speak I hear easy'), the other Mexican co-
worker speaks English worse than him ('Eddie speak I hear very very
hard... you speak English... no... no speak English no structure no
sentence'), which in a way gives Nhi more confidence since he is not
the only person struggling to communicate in English. He does
understand everything when he speaks with his friends at work as
they tend to speak very slowly so he can follow and join their
conversation.
At this time, he also has a weekend job as a dishwasher in a Chinese
restaurant. His co-worker, a Mexican man, also helps him practice
English. While Nhi does have more chances to speak English,
personality is also a factor. Nhi is much more relaxed and overall
friendlier in personality than Anh; in fact, Nhi might be called 'happy-
go-lucky' due to his constant smile, even in the face of adversity. This
attitude and personality makes him easy to interact with. Anh, on the
other hand, tends to be more serious, and takes adversity very heavily.
Nhi is very busy with both jobs and does not have friends outside of
work. In fact, he rarely even sees his family 'because I work... from
... seven o'clock until five o'clock... I don't meet... my... family'
since by the time he gets home from work, Anh and his daughter are
both at work.
Nhi's goal is to study to be an electrician at Pima Community
College, so he can change jobs from the factory to be an electrician.
However his English skills are also not sufficient for him to be
successful in taking these courses, for as he says, 'I speak very bad
English. ' Though he also feels that he speaks English much worse
than native speakers, unlike Anh, he feels that his English is better
than other Vietnamese. His biggest problem is his listening compre-
hension; in April, his company had a meeting about safety at which
his boss spoke and all the employees watched a video. Nhi stated that
he only understood about 10 per cent. If he watches TV with
captions, he can understand about 30 per cent. He usually under-
stands his co-workers if they speak slowly, and says he understands
everything I say; however, I also speak very slowly with them.
Nhi tends to equate speaking English with Americans with
happiness, which is a more simplistic view of American life than
Anh holds. As he says, 'I happy when when I speak... English... and
I hear English my friend and everybody understand. ' In fact, he says:
'In American English is my life. I I want to I want to I want to
understand Ameri American culture. . . and I want American
American understand me. . . yeah. . . so my life is happy. ' Nhi
equates adjusting to American life with happiness, and being
understood in speaking English is a symbol of this adjustment: 'I
need to adjust to American way because. . . because I want I want
every American understand me. '
Like Anh, Nhi also changes his appearance in order to fit into
American life. To him, wearing American clothing is symbolic of his
Americanization, and is a way for him to be viewed as American by
other Americans: 'I must I must wear. . . I must wear. . . American
style clothes... to. . . together... I must wear American style clothes
... the same with. . . everybody everybody i n. . . American. ' When I
asked why, he responded with 'friendly friendly, ' meaning that other
Americans would be friendly towards him if he appeared 'American'
via his clothing. More importantly, he also begins smoking, a habit he
had in Vietnam but had stopped before moving to the US. His co-
workers smoke during breaks, and Nhi begins to smoke again, much
to Anh's chagrin. When I asked why he smoked again, he said, 'my
friends... work friends, ' indicating that smoking with them gave him
an opportunity to interact with them more during breaks.
By mid-April, Nhi has quit his weekend job in the restaurant in
order to concentrate more on learning English, though he is not
enrolled in any ESL classes during that time.
6. 1. 2 Summer
By summer, both Anh and Nhi's social contexts of language use
change, in opposite directions. Anh's ESL classes end in May, and
she does not enrol in any summer classes as she is busy at work,
working almost every day. As such, she loses her tenuous connection
to her friends from Pima, whose friendship provided her with her only
opportunity for English language use in a supportive and relaxed
atmosphere. She has very little contact with them during the summer.
In fact, her only friends at this time are her workmates, who are all
Vietnamese. When she has free time, she usually spends it with her
own family, or her sisters and brothers.
She still has difficulties at work, not as much due to the job itself,
but rather still due to her problems in communicating in English: 'I
been difficult my work yes. . . because... many nail customer... I
must explain... their nail. . . difficult... I must my owner help help
me. '
SOCIAL BARRIERS 131
Another major change in Anh's life is the move of her sister's family
from Virginia to Tucson, and their rental of the house next door to
the Nguyens. This creates an even tighter Vietnamese environment
around Anh for a number of reasons: her social interactions outside of
work are now focused on her sister's and her own family, and when
she goes shopping, she often goes with her sister, who speaks better
English than Anh and usually takes over the communication in shops,
etc.
True to his happy-go-lucky nature, Nhi views the move of his
sister-in-law and her family next door as a new English language use
opportunity because of his two nieces, aged 8 and 11, who
understand Vietnamese but do not speak it. Instead, the family
language is English. Nhi views his nieces as 'English teachers' and
makes the most of any opportunity when he is around them to speak
with them in English and ask them to teach him vocabulary. Thus,
while the move of Anh's family to next door is tightening the
Vietnamese social structures around Anh, they actually open up new
language use opportunities for Nhi since he interacts more with the
children than with Anh's sister or brother-in-law. Nhi also expands
his social interactions by enrolling in an ESL class in the summer, the
same listening and speaking class Anh took in the spring, but at a
different place. His main goal is still to surround himself with many
American friends once his English is better: 'as soon as I speak
English... so-so... I will try to find... many many. . . American
friends. ' He finds some new friends through his classmates, 'because I
go to school I have many new friends' with whom he speaks before
and after class for a few minutes. At work, his social interactions
expand as well, as he has a new co-worker, an American woman who
is extremely talkative: 'my my f actory... my group have. . . a new
friend a new f riend... woman... she she talk all the way. ' He credits
her talkativeness to his English improvement: ' III speak English with
my my workmates better. . . better yes. . . because my my my
friends... she is talks very talks a lot. '
Aside from at school and work, however, Nhi does not speak
English very much, except with his nieces:
Friday and Saturday I meet my niece... I. . . I try in hear hear hear
them speak and I. . . 111... we are we are we are speak... together
everything... everything... for example... I. . . you are you you
you pass me glass.
On a material level, the Nguyens are now beginning to settle into
American life as well. Prior to this time, their house had been mostly
unfurnished, and they had used an outdoor patio table and chairs as
their dining room furniture. Around the beginning of August, their
house begins to become populated with furniture. First, they buy
second hand sofas for the living room, and then a formal dining room
set complete with a matching china cabinet. The transformation of
the Nguyens' home is symbolic, in a way, of their own transformation
from newly arrived immigrant, initially nervous and unsure, to
permanent residents who are planning their future in this country.
Their reception of their green cards, marks of permanence, also
symbolizes this period. Additionally, they have invested in two new
cars, one of which is a van, so they are now a three-car household. In
some ways, they materially represent the American dream. This may
be directly related to their family in the US, who are American,
complete with the house with swimming pool and two-car garage.
Not only do the Nguyens wish to emulate them, but it is as if their
family here expect them to settle in a similar fashion once they have
adapted to their new life. In effect, since the earlier family who came
to the US successfully adapted to and assimilated into American life,
the newly arrived Nguyens are expected to do so as well, after a
certain initial time of adjustment. While this did come as a shock to
Anh, especially, at first, she did begin to appreciate her independence
as well as the future independence of her daughter, and the furnishing
of their home in a way represents their attempts at permanency in the
US.
In another symbol of their continuing adjustment to American life,
in August, the Nguyens decide to go to the cinema to watch the movie
The Sixth Sense, which one of Ann's co-workers had recommended.
This was their first trip to the cinema since they had moved to the US,
though they often went to movies and the theatre in Vietnam. Going
to the movies is not only a way to relax from their busy lives, but also
to spend time together, which they rarely do. As Nhi states,
Our plan our plan... our plan... every weekend every weekend
... we are we are we will go to cinema together... because every
day every day. . . we work and don' t... have free time... I don't
see her... you are you come back and... you come back then I
sleep.
The change to this plan also represents the fact that the Nguyens are
no longer as worried about their lives in the US, and feel they can
relax. As Anh states, ' we... don't like go to cinema or go to visit my
friend, ' to which Nhi adds, 'because we are... because we have worry
... worry... now now my life our lives... better... yeah. '
SOCIAL BARRIERS 133
6. 1. 3 Autumn
By autumn, Anh re-enrols in English classes at Pima, taking
elementary and intermediate listening and speaking classes simultan-
eously, in order to work on improving her ability to communicate in
English, which she is now desperate to do as she has had few
opportunities to communicate in English in the summer. However,
her classes do not provide her with many opportunities to practise
English: 'I I watch the TV video 'Contact with English' yeah. . . I
answer answer when... video and practise... we are don't talk topic
... yeah yeah... speak... so I not conversation practise. '
The pressure to communicate with her customers is still present,
even though she is at a third nail salon. She still struggles to
communicate with her customers at work: 'Friday and Saturday...
many customers my owner told me. . . you you you must... talk with
customer... but I don't hear I don't answer. ' She tries to
communicate with them:
I I always... their name name and. . . what do you work work?
You you live around live around... yeah I ask. . . their... mmm
... family family... and children... sometimes... because in the
salon... only one me talk... talk a little customer and my owner
attention me. . . she and he. . . and they are always... complain...
complain me. . . why you don't speak English...
When she does talk with her customers, she understands only about
half: 'my my customer always ask me. . . 11 understand about fifty or
seventy... yeah. . . but I answer... a little. ' She especially has
difficulty with pronunciation and listening, but overall, she does feel
she is improving and understanding more of what her customers have
to say. Additionally, she is becoming more comfortable with her work
as a nail technician, which makes communicating with her customers
also a bit easier as she does not have to concentrate so much on what
she is doing: 'my my job is easily easily my work more and more better
... 11 feel... my life is. . . my life is. . . more and more little difficult. '
She has finally gained confidence in her job, experiencing difficulty
only with pronunciation. This difficulty, she believes, is in part
because of her restricted opportunities to interact with native speakers
in English as she only communicates with speakers of Spanish in
English:
I feel confident in my my j ob. . . yes... mmmm... the the difficult
i n. . . with me are now pronunciation my my . . . t he. . . my
pronunciation... is harder... because I always... communication
with with customer Mexican... although my classmates in study
English... almost Mexican....
She feels it is difficult to improve her pronunciation since she almost
speaks English exclusively with other non-native English speakers.
She tries to speak English everyday:
But . . . customer Mexican... sometime they are don't speak
English... or speak a few English... I very hard to speak...
customer... 111 feel I speak... slow slow slowly yeah... cause 11
I want contact with Americans... I can speak a lot of . . .
She struggles to improve her English, but finds that her only
resources are other Vietnamese such as her sister and her nieces,
along with the radio and TV:
Sometimes III ask my sister and my niece... yes. . . because I
work all day long long time and 11 no free time... talk with sister
and my niece... but I try yeah I try everyday... I . . . when I free
time... 11 try watching TV and hear radio when I drive car t o. . .
to go. . . my my shop.
At this time, she understands about 40-50 per cent of TV without
captions and about 50-60 per cent with captions, indicating that her
listening skills are improving.
She has lost contact with her friends from the spring semester at
Pima, and does not see them as she is now at a different campus.
Even though she is taking classes at Pima, she does not talk with too
many of her classmates as they are mostly Mexican and tend to speak
in Spanish together before and after class. She is also very busy and
does not have time to talk with them after class, as she must
immediately go to work. Instead, as she says, because most of her
friends are Vietnamese work mates, 'I I talk in Vietnamese many. '
In the autumn, Nhi also enrols in classes at Pima, taking not only
two ESL classes (a beginner reading class and an English on the job
class), but also two content courses: Electrical drafting and com-
puters. As he says, 'Everyday I am happy... because... everyday I... 1
go to school. '
Nhi still spends a lot of time talking with co-workers, and they tend
to discuss fairly complex issues, such as one friend's marital problems
as well as another's husband's alcoholism. His classes continue to go
well, except he decides to drop the electrical drafting class since
'because because... in in in... electric electric class no speak English
only read... reading. ' However, his computer science class is going
SOCIAL BARRIERS 135
well, though when he first registered for the class, he admits he was
worried: 'when when when I when I register register... the register
the basic basic programming... computer... at Pima 11 was... very
worried... mmm... now I I feel I feel I feel ... learning good I
happy. ' Even though he appears to have miscommunications with the
teacher, he is able to communicate his language problems to the
teacher so his teacher can adjust his speech rate so Nhi can
understand:
In the computer class... the teacher... the teacher... mis...
understand me... he's handicap because I speak a little bit... 111
say I say... my teacher... I said my teacher... I understand your
your speak... but I speak slowly yeah
Being able to understand the teacher in class also has something to do
with Nhi's prior preparation - 'before before I . . . I have class... I
read book... yeah... and... then I hear... the teacher explain I
understand about... sixty percent yeah. ' Overall, his English appears
to be improving, especially listening in general and his perception of
individual sounds:
My English is improving... I can speak... I can speak more
quickly... yeah and and... 11 hear... and I hear... more clearly
... because III know... pronunciation... and I hear... who who
speak English I hear pronunciation I hear their pronunciation.
His work mates still provide him with a very supportive environ-
ment for language use, and they teach him new words as well as
correcting his mispronunciation. They also speak slowly enough for
him to be able to follow and understand their conversations, and join
in when he has something to say:
When I . . . I speak wrong... they... correct for me... example I
... I say [pa. l at ]. . . they correct [pse. l at ]. . . and sometimes
sometimes... I misunderstand... their... misunderstand... their
description yeah... 1111 ask again... they speak slowly... slowly I
understand.
In fact, he feels that he is relying less on Vietnamese to understand
English; while he was translating his ideas from Vietnamese into
English before speaking, which slowed him down, he now feels he can
think in English:
I am practising because before I speak English... I guess... I
guess my Vietnamese so I speak slowly... now III try... before
speak... speak English... I guess by English... yeah... 11 feel I
feel I f eel... I feel easy... easily improving.
His English improvement has also been noticed by his boss, who
has given him a different job assignment. Instead of pulling orders
inside the warehouse, he now directly interacts with customers, taking
orders to give to the person inside the warehouse to pull. While Nhi
appreciates the stable and supportive environment of his job, he is
also beginning to feel disenchanted with his job as he feels it prevents
him from putting as much time into studying English which he really
wants and needs to do in order to achieve his goal of going to
university to study chemistry. He has changed his mind about
becoming an electrician after visiting the university when his daughter
began there. Instead, he wants to return to college and study
chemistry again, for 3 or 4 years, to get a BSc. What he wants to do
after getting his BSc, however, is unclear: 'I don't know. . . now I
don't know... 111 only study I only study yeah. '
Although Nhi seems to have more opportunities to talk than Anh,
he perceives his opportunities to speak English as fewer than hers as
she actually interacts with more people (though her conversations are
actually fewer than Nhi's): 'because I. . . I have... a few chance to
practice English... I . . . I . . . I I have. . . I have. . . I have
conversation with my workmates... in company yeah. ' Even his social
contacts are limited outside of school and work: 'outside my my job
... 111 don't have... friends. ' Like Anh, Nhi also uses the TV as a
tool for English language learning, watching with captions, and he
understands about 50 per cent.
Interestingly, at the beginning of the autumn, Anh begins to be
more assertive in the interviews I have with them. As time passes, she
becomes more and more confident in communicating, and often
dominates the conversation, while Nhi smilingly looks on. The
talkativeness and confidence in communicating she displays with me
and continues to develop across time during the interviews contrasts
with her descriptions of her interactions in the nail shop, as well as her
interactions when she comes to my house for dinner one day in
November. All the Nguyens are invited, and during the several hour
stay at my house, Anh appears withdrawn and shy while Nhi and
Minh both freely converse with my friends and me. While Nhi also at
times defers to his daughter, who becomes the dominant voice of the
Nguyens, he still initiates conversation and responds on his own.
Anh, on the other hand, uses Minh as her translator, often telling
Minh something in Vietnamese, which her daughter then says in
SOCIAL BARRIERS 137
English. This occurs so frequently that Minh jokes about it, saying
'she uses me to talk to you. ' Interestingly, after Minh leaves the room
in order to play videogames, Ann and Nhi resume their interactive
conversational pattern that was evident during the interviews, and
Anh becomes much more confident and assertive in speaking English.
For both Anh and Nhi, the United States is still a land of
opportunity, especially if one is willing to work hard. They feel they
can achieve success here since they have adapted to the American
work ethic. As Anh says, 'I think I live in here... if if I work hard... I
have everything because... in here I don't worry... unemployed. '
They also feel peace and safety in the US, in contrast to life in
Vietnam, where they constantly worried and stressed about work and
life. Nhi says, 'and now we don't worry. . . don't worry about
anything... because everybody American Mexican... Mexican...
are friendly yeah. '
They are also constantly adjusting themselves to American life,
which they embrace openly. When asked whether it was necessary to
adapt to American culture in order to learn English, Nhi responds,
'Oh, it is too necessary. ' However, Anh and Nhi diverge in their
feelings about whether they live an American life, and whether they
view themselves as American or Vietnamese. For Nhi, it is very clear -
he lives an American life because everything he does and has is
American:
I feel I. . . I am living... I am American life... yes... because...
all all my action... and al l . . . my guess every everything is
American... yes... in my home... always... I. . . American thing
... newspaper... magazine... hear musi c. . . American...
American... no Vietnam no Vietnamese... I feel I am I am living
... American life. . . yeah.
For Anh, it's not so clear: 'I think... I not perfect American life. . .
because I confuse... confuse you American life and Vietnam yeah. '
While Nhi is also clear about his identity - he is American because his
life is American - Anh is unsure about who she is. As Nhi says, 'now
now we. . . we are... now we are. . . we feel. . . we are American...
except language... Vietnamese... because... because... my way
my way life. . . is American. ' However, Anh has a different view:
I think I different he. . . because in Vietnam I have a way l i fe. . .
enough... I. . . confident in Vietnam but when I move in here I
speak English not well and. . . sometimes in my job is difficult
because I don't understood my customer... I I I feel lonely
because my my boss... sometimes complain me... you don't talk
customer... 11... I feel I... I try very very hard... cause some
classmates... my friend almost... from Mexican... if they don't
understand they are change speak Spanish... Spanish with
together... I I don't understand yeah... yes. . . in here... a
few a few friends... I don't have free time visits... my f riend...
cause my workmates... he always change... because i f ... shop
... another shop get make a lot of money... they are go shop... I
don't like because 11... want I want... have job... long time... I
seldom feel I seldom feel American.
When asked if she wanted to be American, Anh responded, 'yes...
maybe... because I . . . I don't know everything in here. ' To her,
knowing 'everything' about life in America appears to be a
prerequisite to being an American, possibly because she compares
herself to her sisters and brothers, who are American, and have a great
deal of knowledge about life in the US.
Both Anh and Nhi, however, are critical of how other Vietnamese
live their lives in America. Many fairly recent arrivals from Vietnam
who have settled in Tucson tend to be immersed within a dense
Vietnamese community, living in the same neighbourhood and living
with several families in one home. They exist largely independent of
the larger American society, with few symbols of permanency, as if
their life in Tucson is merely transitory: they do not open bank
accounts, preferring to use money orders to pay their bills, they do
not buy new cars or furniture for their homes or new clothing, and
they find jobs in factories and Vietnamese shops, where English is not
necessary. As Nhi says, 'everything their life are... are Vietnamese...
in America... yeah... uh... they they they are many... came United
States seven year... they don't speak English. ' Anh and Nhi both
believe this is not good, for as Anh states:
If they are... live... Vietnamese way... I think they are... very
difficulty... live in here because they are not confident everything
around... they are... don't have friend in American... they are
... lif e... live together... live the same in Vietnam... I think no
good no good... because if in here... I live... a good... I live a
way of life good... American here... every good everything...
good everything... I think my f amily... have try try very hard.
While the Nguyens have a critical view of this group of Vietnamese,
the Vietnamese community views the Nguyens as perhaps too
Americanized and assimilated into American life. As Anh says,
SOCIAL BARRIERS 139
'they are surprised my family... because you are just come here one
year. . . you you bought new car. ' Because of being busy and
preferring to study and relax rather than cooking every night, Anh
also brings American foods like salads and fruit with her to work,
while her co-workers still bring Vietnamese lunches and dinners. Her
co-workers told her 'you are American' and advise her to stop
studying so much, and instead concentrate on working more hours so
she can make more money:
They told me why you you older you study every day... why you
don't work make a lot of money... I don't think... the take [of
money]... too many good... because 11 worry if if I choose live in
here... 11 must... know I must know everything in here... 111
am. . . I will study study English know everything.
Because these individuals still retain a Vietnamese lifestyle and the
Nguyens are trying hard to become American, the Nguyens, as Anh
states, prefer not to interact with them too much: 'I don't like contact
with... with them because... they are come here... long time...
they are not change no change l i f e. . . confuse American life and
Vietnamese life. '
In terms of their language skills, in December, the end of the study,
both Anh and Nhi said that while they both understood other people
fairly well, especially when they speak more slowly, they had a great
deal of difficulty being understood by others. As Nhi states, 'I begin
to work at company... I didn't hear and didn't speak... English
everybody... and now. . . I speak English everybody... in
department... in my company... and I speak... they are... they
understand. '
In fact, when he compares himself with his Vietnamese workmates,
he feels that he speaks better English than them since they do not
study English. Anh also feels her English is improving, though she is
still not confident. This is often reinforced since her customers often
have difficulty understanding her.
Both are still very aware of their linguistic difficulties. As Nhi says,
I can pronounce for each word... but when... I say a sentence...
I confuse... I confuse... I confuse about structure... so I I I
pronounce no good yeah... one word one word I pronounce good
... but when I when I speak English with. . . everybody... I I
always confuse... struc struc sentence structure... and... I forgot
I forgot pronunciation.
While Anh is highly motivated to change her way of life, working
hard to earn money and studying English so she can go back to
university, Nhi is still the optimist, who is happy just to be able to be
understood by other Americans. Both Nguyens know that it is Anh
who drives the family forward: 'I believe I believe... she will study
good... she will go. . . she very hard work very hard study... but I
don't believe me. . . because 11... I am always happy I don't worry
... I follow I follow I follow her. '
6. 2 Social barriers
As this short description of the Nguyens' social contexts of language
use indicates, the social barriers operating on their English language
development are multiple and dynamic, as well as symbiotic in
nature. Additionally, this description indicates that there are two
striking differences between Anh and Nhi when it comes to the social
data. One difference is the language opportunities available to them
outside the home, specifically in their work and school environment.
The other difference is their perceptions of what being an American
is, which is affected by multiple social factors including their language
use contexts, experiences and personalities. This in turn reinforces
their perceptions of who they themselves are, e. g. American,
Vietnamese, or both, which then affects their level of comfort and
confidence in using English. The changing social environment
around them also affects their language use opportunities, which
then also reinforces their emerging identities and feelings towards
America and their English. These issues will be discussed in more
detail below in light of prior findings on the effect of social constraints
in SLA. Rather than separating the discussion into discrete parts, the
discussion is fluid and dynamic, moving from one social constraint to
another in acknowledgment of the complex and symbiotic nature of
these types of constraints.
An important starting point for a discussion of social constraints is
the notion of social identity. As Lum (1982) states, social identity is a
complex phenomenon that involves both our own perceptions of who
we are, and also how others perceive us. Tajfel (1974) defines it as
'that part of an individual's self-concept which derives from his
knowledge of his membership of a social group (or groups) together
with the emotional significance attached to that membership' (69).
He further states that since individuals' identities are derived from in-
group memberships, an individual can choose to change membership
in a group if this membership does not satisfy those elements of their
SOCIAL BARRIERS 141
social identity which they view positively. People construct their
identity in various ways, through language, clothing, actions and
behaviours, which they believe reflect their social identity. For L2
learners, this may include but not be restricted to adopting the habits
and customs of the new culture, including clothing, eating and
working habits if they wish to develop an L2 identity, or retaining
their LI behaviours and customs if they wish to keep their LI
identity, or both under different circumstances, as social identity may
best be viewed as context-dependent, i. e. multiple and changing
across different contexts (Peirce 1995).
Nhi very much wants to be an American, and to be viewed as an
American by others. He has constructed an American identity
through American clothes, car and other material goods, the
possession of which he believes makes him American. He also
believes that others view him as American because they understand
his English and because he wears American clothes. Beginning to
smoke again is also a way to construct closer ties to his work mates, an
in-group with whom he clearly wants to be affiliated as they are
mostly American. Considering himself an American gives him
confidence in doing American things. Fortunately for Nhi, his
experiences with people in the US have also been mostly positive -
he considers most people he meets as 'friendly' - which may be their
reaction to him, as he himself is typically extremely friendly. These
interactions, which are positive in nature, give him added confidence
in using English, to the point that he tells his computer science
teacher to speak more slowly so he can understand better, to which
his teacher obliges.
Anh on the other hand, is 'lost' - she compares herself to her
family, who are American, in most respects, but feels she does not
know everything compared with them and therefore she is not
American. Striving to fit in somewhere, to have an in-group beside
her immediate family, she then changes her clothing style to that of
her much younger Vietnamese co-workers, until she realizes that she
does not want to retain a Vietnamese lifestyle in the US. She has
difficulties developing an L2 identity, however, since she is immersed
within the LI community, from which she clearly wants to be
distanced but has no clear way of distancing herself from.
Interestingly, she is identified as American by her Vietnamese co-
workers, with whom she avoids contact in order to try to separate
herself from a Vietnamese way of life and Vietnamese identity, to
some extent. This pressure to acculturate is also an important social
constraint on L2 development and LI retention.
Schumann's (1986, 1990) 'Acculturation Model' attempts to link
social factors such as degree of acculturation with SLA. Schumann
hypothesized that social factors (social dominance, integration
pattern, enclosure, cohesiveness, size, cultural congruence, attitude,
and length of stay) are primary in affecting the degree to which the
learners acculturate into the L2 culture, and thus their level of L2
development. The ZISA project (Meisel, Clahsen and Pienemann
1981) and the Heidelberger Forschungsprojekt 'Pidgin Deutsch'
(1978) both found that social distance was a factor in L2 proficiency,
with a lesser degree of social distance promoting L2 development.
More recent work by Lybeck (2002) on American women's
acculturation into Norwegian society and their acquisition of
Norwegian, found that the learners with the strongest social networks
with Norwegians had a greater degree of Norwegian proficiency while
both those with weaker networks or stronger LI identities had a
stronger American accented Norwegian.
This study found that 'perceived' degrees of acculturation, much
like perceived social identities, affected the Nguyens' levels of
confidence in using English and in seeking opportunities to use
English. For example, Nhi perceives himself to be American and
living an American way of life. Nhi's perceived acculturation gives
him greater confidence to interact with people, which in turn gives
him greater English language use opportunities. While Anh does
perceive herself to be living an American lifestyle, she is less sure that
she is American and has acculturated as she does not believe she is
American until she 'knows everything' because she compares herself
to her brothers and sisters. As she compares herself negatively to her
family members (and even her daughter) she often shies away from
opportunities to use English, ceding them to someone more confident
(and Americanized) when they are with her. However, social identity
and acculturation are only part of a complex picture of the social
constraints operating on the Nguyens. Social interactions, as well as
the notion of 'investment' (Peirce, 1995), which encapsulates
attitude, motivation and personality, must also be considered.
Social interactions can be defined as both amount of contact and
type of contact, both of which are intricately connected with social
identity, as who one has contact with may represent the group with
which one identifies. The research in this area does indicate that
overall, a greater quality and quantity of interactions with native
speakers of the target language promotes L2 acquisition, possibly
because it leads to increased opportunities to engage in the negoti-
ation of meaning, which may facilitate L2 acquisition (Ellis 1994).
SOCIAL BARRIERS 143
Oyama (1976), for example, suggested that amount of language use
was important in the degree of accentedness for learners in her study
while Tahta, Wood and Loewenthal (1981) found that the top
predictor of non-accentedness for their participants was the use of
English at home, followed by exposure to English models outside of
school. Research on syntax such as the Heidelberger
Forschungsprojekt 'Pidgin Deutsch' (1978) and the ZISA project
(Meisel, Clahsen, and Pienemann 1981) also found that amount of
contact and degree of interaction was significant factor in L2
acquisition, with those learners having a greater extent of interactions
with native speakers of the target language achieving higher levels of
L2 proficiency than those with fewer interactions. Spada (1986) also
found that learners with qualitative better contact with native
speakers of English had higher levels of oral proficiency than those
with less contact. Furthermore, the number of native speaker friends
has also been found to be a strong predictor of L2 proficiency
(Monshi-Tousi, Hosseine-Fatima and Oiler 1980). In more recent
work, Flege, Frieda and Nozawa (1997), in research on the English of
Italian immigrants to Canada, found that LI use affected L2 accent:
those who had a higher degree of LI use were rated as having a
stronger LI accent. These findings were confirmed by Piske and
MacKay (1999); Piske, MacKay and Flege (2001); and Guion, Flege
and Loftin (2000).
An important pressure on Anh, which ultimately affects her access
to LI and L2 use opportunities, is her family in the US; they
symbolically push her and her family to become assimilated into
American life, but this pressure, as it were, creates an unrealistic
expectation for Anh in that she feels that she must 'know everything'
about America like her family in order to truly be American. On the
surface, it also appears that Anh is positioned as the 'leader' and 'go-
getter' within the family and that this enables her to have more
opportunities for L2 use. For example, Anh had several opportunities
to travel within the US, taking a ten day drive from Virginia Beach
through Florida and Texas and back to Arizona with one of her
sisters. She also went on a long weekend trip to Las Vegas with her
co-workers. In contrast, Nhi did not take any trips due to difficulties
in getting time off from work. As Nhi stated, 'she always have... many
many chance travel... I have no chance. ' Anh also had more social
opportunities; she had made friends with a Chinese student and an
Indonesian student in her ESL course during her first semester of
study at a local community college and frequently met them for social
gatherings. As Nhi often said of his wife, 'she have many friends... I
am no friends. ' Finally, at work Anh had to use English to interact
with her customers in the nail shop while Nhi primarily worked
without speaking at his job as an order filler. This is also acknow-
ledged by Nhi: 'my using English i s. . . less than her. . . because I
have few chances to practise English... '.
However, while Anh appears to be facilitated in L2 use, her
'contexts' of L2 use are in actuality more reduced than Nhi's, and the
access to various L2 use contexts both Anh and Nhi have is gendered
due to the positioning of Anh within the family as the caretaker as well
as the 'leader'. Due to the circumstances of their immigration to the
US, and the fact that they are residing in a community with Ann's
brothers and sisters, the role of caretaker gets redefined from being a
housekeeper, nurturer, wife and mother, to being the financial
caretaker for the family.
There are a number of factors that have effected this redefinition of
caretaker: first, Anh is responsible for her immediate family being in
the US as they immigrated there because her family members were
already established in the US whereas members of Nhi's family had
immigrated to Germany. Once in the US, Anh feels responsible for
her immediate family's financial well-being, especially as she realizes
that the norms of extended family financial support are different from
those in Vietnam. In Vietnam, it is common to share financial and
other resources among extended family members, whereas in the US,
this is typically not the case. As Anh finds out, her brothers and sisters
have become Americanized to the extent that they expect Anh's
family to become financially independent as soon as possible; initially,
this was difficult for Anh to understand and accept:
My culture... Asian... we are everybody I my family he always
life together... in here... I understand that everybody one family
one person... my brother and my sister... they are live in here
twenty year ago. . . they are. . . different me. . . ago two year... I
can't understand... my brother and my sister... they life.
In fact, she had begun understanding this while still in Vietnam; her
brother visited before their move and tried to prepare them for their
new life in the US. The quotes below, from both Anh and Nhi, not
only illustrate Anh's and Nhi's very different orientations to their life
in the US, especially regarding viewing life there as stressful or not,
but also Anh's concern over employment - for financial caretaking -
in contrast to Nhi's focus on the pleasures of life in the US. Both were
answering my question of 'Is American life stressful compared to
Vietnamese life?'
SOCIAL BARRIERS 145
Anh: Before we are move here... to my brother... visit my family
in... in Vietnam... he told me... American life is very stress very
stress... and very hard very hard... and and... we are... prepare
the new new life in here... when when when... we are move here
... I am worry everything... because I don't hear don't speak and
... I don't we are no j ob... but... he... my husband has a j ob...
my daughter have a job and my daughter... begin confident in the
... her school she... I study nail technician... when I... graduate
nail technician I confident... 111 feel better I have a job then I...
don't stress yeah.
Nhi: Because... in Vietnam... I always... I always confuse...
confuse about communism politics politics yeah always about
politics yeah... 11 teach... 11 taught... always controlled always
was controlled... and... don't have... free time t o... relax...
and in here... when I have j ob... I take money... small amount
of money I f eel . . . I f eel ... my life in here... enough... don't
worry... and... everybody around in Tucson... always friendly
yeah... and always help me . . . I study... when I want to study
what... I can... I can do that... for future yeah... so I... I am
happy.
Ann's realization that each family is financially independent of each
other, and that her own sisters and brothers expect Ann's family to
become financially stable and less reliant on the rest of the family,
combined with her status as the oldest sister which in Vietnamese
culture signifies responsibility for younger siblings, compels Anh to
find work as quickly as possible in one of the most accessible
workplaces for Vietnamese women in Tucson: the nail shop. As Anh
says:
I study nail technician because my [friend]... she work nails long
time... she... encourage me study nail ... I start think I don't
study nail because I don't understand everything and she encour-
age the if you study in Vietnamese you can study in here... I study
nail three months... after I try in cl ass... because I don't
understand my teacher explain everything... I always read book
and watch television and video... I study my friend come back
home... I translation dictionary everyday... after three months I
pass the test.
As Ann's statement illustrates, getting a nail technician's licence and
gaining employment in a nail shop is one of the easier (relatively
speaking) options available to Vietnamese women in Tucson because
of the pre-established networks within the Vietnamese community:
another Vietnamese nail technician helps Ann get her licence and find
employment, and even though Anh is initially unable to read and fully
understand the English texts used in the nail courses, she is able to
become certified due to the help of her Vietnamese friend.
Meanwhile, Nhi finds employment in a golf factory as an order filler,
a job that is more acceptable for men within the Vietnamese
community. Although women are employed in the factory, they are
typically American or Mexican, and not Vietnamese.
As noted previously, Anh appears to have greater opportunities for
English language use than Nhi, as she takes ESL classes at Pima, has
several friends who are not Vietnamese and with whom she must
speak English, and interacts with non-Vietnamese customers at the
nail salon all day long on a daily basis. Nhi, on the other hand, does
not appear to have many chances to use English as he does not take
any ESL classes, and speaks English mostly during his breaks at work.
However, the quantity of access to language use does not equate to
quality, and it is the quality of the interactions that facilitate L2
acquisition as more complex conversations promote the type of
interactions that are necessary for L2 acquisition to take place (Ellis
1994).
On the surface, the nail salon appears to be a site of many L2
language use opportunities - and thus many chances to develop
English skills - especially in contrast to work in a factory setting,
where Nhi only has opportunities to use English during short breaks
and the lunch hour. As stated previously, Nhi perceives that his wife
does have greater opportunities to practise her English than he does
as she has to interact with customers all day long, and as her
customers are usually American or Mexican, the interaction has to be
in English.
In actuality, however, the nail shop proves to be an unsupportive
English language use environment for Anh for a number of reasons.
First of all, the majority of Anh's customers are Mexican, and like
Anh herself, have limited English communication skills. While she
talks with her customers, her communication is restricted as she has
difficulty expressing her ideas and also understanding the English her
Mexican customers speak. Secondly, Anh also initially has an
additional difficulty: as she is new at being a nail technician, she
also experiences problems in performing her job well and has a
difficult time concentrating on her work as well as on her English at
the same time. Finally, she is also under increasing pressure from her
SOCIAL BARRIERS 147
boss to speak more English with her customers, and this pressure
fuels her anxiety, culminating in her switching nail shops after three
months in the first shop. As she says:
I have just nail technician because my owner talk talk me and why
why you don't talk customer... are you nervous... are you afraid
customer... because I am talk a little bit... speaking English with
customer... I make no good... no good... I... control machine
... yes because I worry... I nervous in talk customer.
A common theme throughout the period of the study, and across
the three nail shops Anh works in before opening her own shop, is her
sense of frustration at work because of her limited English skills. This
frustration is due partly because of the pressure from her boss to have
'small talk' with the customers. Additionally, she must talk to her
customers about nail diseases and fungi, but does not feel that she has
the vocabulary or the grammatical structures to communicate her
ideas. She forces herself to try to communicate with them, but her
experiences are not usually successful because as she states:
I talk. . . vocabulary no sentence no structure... I can't explain I
can't explain their understand nail and disease... I speak
vocabulary... no sentence no structure sentence... customer
don't understand and I don't understand them... I think it is bad
experience.
These experiences are very depressing and discouraging for Anh, and
she breaks down crying when she relates them.
Anh craves a supportive and stable work environment so she can
relax and spend her energy on improving her English and learning
more about life in the US. She switches to a second nail salon in the
hope of finding this environment, but finds that the situation is similar
to the first salon. Eventually, she finds work at a third salon - the third
in a period of nine months - but finds that although she has more
confidence in her abilities as a nail technician, she still lacks the
English skills she needs to communicate with her clients successfully.
The pressure to communicate with her customers is still present:
I I always... their name name and. . . what do you work work?
You you live around live around... yeah I ask. . . their... mmm
... family family... and children... sometimes... because in the
salon... only one me talk... talk a little customer and my owner
attention me. . . she and he. . . and they are always... complain...
complain me. . . why you don't speak English....
When she does talk with her customers, she understands only
about half: 'my my customer always ask me... 11 understand about
fifty or seventy... yeah... but I answer... a little. ' Her difficulty, she
believes, is in part because of her restricted opportunities to interact
with native speakers in English as she only communicates with
speakers of Spanish in English: 'I feel confident in my my job... yes
... mmmm... the the difficult in... with me are now pronunciation
my my. . . t he... my pronunciation... is harder... because I
always., communication with with customer Mexican... '.
She tries to speak English everyday ' but... customer Mexican...
sometime they are don't speak English... or speak a few English... I
very hard to speak... customer... Ill feel I speak... slow slow
slowly yeah... cause III want contact with Americans... I can speak
a lot of... '. Because of her work hours, she is unable to continue
attending ESL classes, and therefore feels frustrated because she has
few opportunities either in the present or in the future to learn
English.
Nhi, on the other hand, has an extremely supportive English
language use environment at work, and creates one for himself at
college as well, once he begins taking classes. His relaxed attitude and
constant smile make him very approachable, and he soon makes
friends with whom he can practise his English. Most of these friends
are native speakers of English, and those who are not are proficient in
English, to a certain extent. While he does not speak English for a
long time everyday, his interactions are more complex than Anh's,
and his co-workers provide him with a supportive scaffold, giving him
the vocabulary he needs, as well as correcting him when he makes a
mistake. These collaborative interactions have also been claimed by
some researchers (e. g. Hatch 1978; Peck 1978; Wagner-Gough
1975) to aid in SLA as it is through this that a learner 'learns how to
interact verbally, and out of this interaction syntactic structures are
developed' (Hatch 1978: 404). Interestingly, Anh does not appear to
participate in such collaborative interactions; instead, as described
above, when miscommunications occur, either conversation stops or
a more confident speaker of English takes over the interaction. Nhi,
on the other hand, often engages in these types of interactions with his
co-workers, and they discuss complex topics such as alcoholism and
marital problems and discussions change day-to-day, exposing Nhi to
a rich array of vocabulary and grammatical structures. Additionally,
across time, Nhi's opportunities to use English expand, rather than
becoming more restricted, contrary to Anh's. His new co-worker is
extremely talkative, giving Nhi more chances to use English. He
SOCIAL BARRIERS 149
begins taking ESL classes and eventually takes a computer science
class with native speakers of English, where he must not only interact
with other students but also negotiate meaning with the instructor,
which he is not afraid to do. In fact, he feels that he is relying less on
Vietnamese to understand English; while he was translating his ideas
from Vietnamese into English before speaking, which slowed him
down, he now feels he can think in English: 'I am practicing because
before I speak English... I guess... I guess my Vietnamese so I speak
slowly... now I 11 try. . . speak English... I guess by English...
yeah... II feel I feel I feel. . . I feel easy... easily improving. ' His
English improvement has also been noticed by his boss, who has
given him a different job assignment nine months later, at a time
when Anh is switching to her third nail salon. Instead of pulling
orders inside the warehouse, he now directly interacts with custom-
ers, taking orders to give to the person inside the warehouse to pull.
Additionally, he views the move of Anh's sister to next door as an
opportunity to learn English from his nieces.
These two very different communities of language use brings up
another issue - the effect of the LI and L2 community on language
learners' opportunities to use the target language. Prior research by
Peirce (1995) has found that the L2 community limited the
interactions learners had with members of that community.
However, Goldstein (1995) found it was the LI community that
limited the opportunities the learners had to use English. Nhi and
Anh were immersed within two different communities; while Nhi was
immersed within a predominant L2 community both at college and at
work, Anh was immersed within the LI community at work and a
fairly Mexican community at college. Unlike Peirce (1995), this study
found that the L2 community enabled Nhi to create and have more
L2 use opportunities both at college and at work. Anh, however, was
immersed in the LI community but was hindered in English language
use even though on the surface it appeared that this environment
would actually promote English use. The manner in which these
opportunities were promoted, i. e. with some pressure by the boss,
actually debilitated the communication, as Anh constantly felt she
was being observed by her boss. Additionally, the communication
within the L2 community was not with native speakers of English but
with a third language community, speakers of Spanish, whom Anh
had difficulty understanding. This third community also formed the
environment at college and so even in a supposedly nurturing L2
language use environment, Anh was distanced as the other members
of this community all belonged to another in-group to which she did
not belong and could not join as they spoke a different language.
Anh's immersion within the LI community also extends to her social
life, which consists mostly of her family. She becomes even more
immersed within this community across time as her contact with
English speakers decreases and her sister moves next door.
This brings up the notion of 'investment, ' which Peirce (1995)
coined in order to explain the complex interaction between the
language learner and the learning environment. Building on notions
of motivation and attitude, as well as personality, Peirce (1995)
argues that past research has not adequately examined why learners
sometimes appear to be motivated and extroverted while at other
times anxious, introverted and unmotivated, as prior research on
attitudes, motivation and personality have operationalized these
constructs as stable, rather than dynamic across social contexts.
Thus, while both Anh and Nhi have positive attitudes towards
members of the target culture, which has been found to promote
learning, and both express they are highly motivated to learn English,
these constructs do not explain why Anh and Nhi acquired English to
such differing extents. Why does Anh not create more language use
opportunities when she is constantly interacting with non-
Vietnamese? One could argue that different personality traits could
affect communication styles. For example, Anh is typically intro-
verted and anxious while Nhi is outgoing and more confident, which
supports prior research findings (Strong 1983) that extroverted
learners have an advantage in acquiring basic interpersonal commu-
nication skills (BIGS) over introverted learners. Additionally, vari-
ables such as talkativeness, responsiveness and gregariousness, all
associated with extroversion, have been linked to higher levels of L2
proficiency (Strong 1983).
However, these traits do not appear to be fixed for Anh; in her work
environment, Anh appears to be very introverted and almost
unmotivated to use English as she still struggles with making 'small
talk' with her customers in December, even as her English skills have
become much better. A very different side of Anh emerges in our
interviews, however. In this setting, Anh is not only confident, but she
is an extremely capable communicator who expresses her ideas freely,
with emotion and clarity. It is obvious that Anh is neither introverted
or extroverted, and neither motivated nor lacking motivation, but all
of them, depending on the situation of language use. She does not
invest heavily in English language use in her work environment, even
though she wants to, because of the difficulties in communicating
with her customers, and the fact that the communication does not
SOCIAL BARRIERS 151
help her learn English, of which she is very aware. Having difficulties
conversing both in English and operating the equipment, Anh instead
invests largely in becoming a better nail technician, which she feels
can win her steady customers who will return and ask for her because
of her nail abilities rather than her English proficiency. It is not that
she does not want to speak English with them, but rather that she
perceives that given the choice of investing in her nail or her English
skills, having difficulties doing both at once, it is more efficient to
invest in her nail skills as her interactants are non-native speakers as
well, and therefore unlikely to help her learn English. It is also
important for her to become highly proficient at being a nail
technician as soon as possible in order to be a caretaker for the family.
In contrast, her interviews with me are perceived by her to be the
only opportunity she has to use English in a supportive and relaxed
atmosphere. In fact, Nhi almost 'cedes' this opportunity to her, as he
is not as anxious as he has supportive atmosphere at work and school,
whereas for Anh, this is her only chance to practise English. As such,
Nhi noticeably lets Anh take the lead in initiating conversation, and
typically only speaks when directly addressed or responding to an
issue Anh has brought up. If the question is for both of them, Anh
does most of the talking. These interviews, however, are not just an
opportunity to use English for as time passes, the subjects become
more and more personal, addressing her struggles at work, in the US,
and learning English. It seems that Anh and Nhi both view me as a
'confessor' - i. e. a non-judgmental person to whom they can confide
their struggles in life without repercussions. This is especially
important for Anh, who has no one else to confide in except for
Nhi. Additionally, she does not discuss these issues with her family
since she is under pressure, both from them and herself, to adjust to
life in the US as quickly as possible and become independent as is
typical of extended families in the US. Discussing these issues with
them would indicate that they were not succeeding.
In summary, these social constraints are difficult to separate from
one another as they interact with one another while also differing
across time and place. From this research, it seems clear that
supportive social contexts promote the learners' investment in
English language use, and that unsupportive contexts may restrict
learners' opportunities to engage in meaningful communication even
though English language use appears to be a necessity in these
contexts. Additionally, perceptions of levels of acculturation and
social identity may not directly affect SLA, but may affect learners'
levels of confidence in using the target language, which may then lead
the learners to actively seek more opportunities to use English.
Confidence, of course, can be facilitated or debilitated depending on
the context of language use, and therefore is dynamic as well.
This brings up an important question: How do these social
constraints affect the Nguyens' development of English syllable
margins? This question will be addressed in the following chapter.
7
A Unified Approach
This chapter synthesizes the findings from the previous chapters in
several ways. First, stages in onset and coda acquisition are suggested
and described. Secondly, interactions among social and linguistic
constraints are explored. Finally, the effects of social barriers on the
participants' acquisition of L2 English syllable onsets and codas are
discussed.
7. 1 Stages in onset and coda acquisition
It does appear that a developmental sequence, constrained by LI
transfer effects, developmental effects and markedness, can be
established. The developmental sequence is similar for both partici-
pants, and therefore may tentatively be generalized to other learners
of English who speak Vietnamese as a native language. In addition, all
learners of English, regardless of native language, may be constrained
by similar developmental and markedness effects in the acquisition of
an L2 phonology; however, this developmental sequence may not be
generalizable to all L2 learners of English as the initial constraints on
the development of L2 syllable onsets and codas appear to be
primarily LI transfer effects, which differ across learners of different
Lls according to the phonotactics of the LI.
In the first stage of development, it appears that the participants
make equivalence classifications between Vietnamese and English.
Those consonants that are similar in type and in position, i. e. they are
allowed syllable-initially or finally in Vietnamese as well as in English,
are transferred. These consonants are the nasals and voiceless stops for
coda position and nasals, voiceless and voiced stops, voiceless
fricatives as well as /v/, and the liquids, in onset position. Those
consonants that are similar in type but different in allowable syllable
position, such as the fricatives and voiced stops, which only exist
syllable-initially in Vietnamese, are also transferred into the learners'
emerging L2 repertoire. While produced correctly in onset position,
these consonants undergo modification in production (e. g. epenthesis
and devoicing) in order to be modified towards an acceptable LI
syllable structure. However, all the consonants, whether transferred
from the LI into the L2 based on equivalence classifications or
modified towards the LI, are subject to the constraints of develop-
mental effects in terms of the sequence of acquisition. Front
consonants are acquired before back consonants, and voiceless
consonants are developed before voiced consonants. In addition, the
interdental fricatives, palatal fricatives and affricates are developed
relatively late. Markedness effects also affect acquisition in terms of
length, type, length and type and sonority. Generally, the more
unmarked onsets and codas, both in terms of length (shorter >
longer), type (e. g. stop-fricative > stop-stop, fricative-fricative),
length and type (shorter > longer with similar constituents) and
sonority (margins meeting UCSS > margins violating UCSS) would
be acquired before the comparatively more marked onsets and codas.
At this point, the emerging L2 repertoire would begin including final
voiceless stops, nasals and fricatives (with epenthesis in coda position).
In the second stage of development, those codas and consonants
that are not in the LI, such as the affricates, and some of the
fricatives, as well as the CC, CCC and CCCC structures, are
produced in light of the transferred LI consonants, which now
comprise the learners' emerging L2 repertoire due to transfer of these
consonants in stage 1. Some of these consonants are devoiced or
stopped in order to be produced as voiceless stops, which are
allowable syllable-finally in Vietnamese, while the longer onsets and
codas may undergo two types of modifications, e. g. epenthesis and
absence, in order to approximate an LI syllable structure. Again, all
these singleton consonants and consonant clusters would be subject
to the constraints of both developmental and markedness effects in
terms of the sequence of development. Additionally, the consonants
at this point are typically modified towards the emerging L2
repertoire, rather than towards the LI. However, LI transfer is still
a constraint, especially for Anh, who has greater LI retention. The
participant's emerging L2 phonologies at this point include the onsets
and codas from stage 1, plus the voiceless affricate, and the more
difficult fricatives are also beginning to emerge. Furthermore, some
CC and CCC onsets and codas are beginning to emerge, especially
those containing the nasals, voiceless stops and fricatives, which have
emerged as singletons and can be modified towards an acceptable LI
syllable structure through absence and/or epenthesis.
At stage 3, the participants make fewer equivalence classifications
between Vietnamese and English as LI transfer effects appear to
A UNIFIED APPROACH 155
decrease. Instead, onsets and codas are initially modified towards the
already emerging structures that represent the participants' existing
L2 phonologies. The more marked consonants begin emerging, as do
longer margins. There appear to be three developmental levels within
this stage: for some consonants, LI transfer effects are still dominant.
This means that the longer margins and more difficult singletons are
still modified towards an LI phonology via absence, epenthesis and
feature change. This is especially true for Anh. The other level is
modification towards the emerging L2 repertoire, i. e. building on the
previously acquired margins. This is especially prevalent for Anh. The
third level is correct production, usually of those margins containing
the voiceless stops and nasals, as well as voiceless affricates and
fricatives, which appear to have been acquired at this point.
Developmental as well as markedness effects would also still constrain
the development of onsets and codas at this stage.
Finally, stage 4, which the participants did not reach at the end of
this study, would include the development of almost all the onsets
and codas, including the emergence of the CCCC codas. This stage is
also constrained by developmental and markedness effects. The result
of this stage would be the approximation of a native speaker-like
phonology, which may still include some errors. This stage is
extremely difficult to reach, and the participants may never reach
this level of development.
It is clear that Nhi, at least, has entered stage 3 by the end of the
study. He is able to produce most single consonant onsets and codas,
as well as many of the CC and CCC onsets and codas, especially
those containing the nasals, voiceless stops, fricatives and affricates,
and clearly builds upon his emerging L2 phonological repertoire
rather than being constrained by LI retention. Anh, on the other
hand, appears to be in a transitional stage as she is still struggling with
longer structures, which marks stage 2, while being able to produce
some of the longer onsets and codas, which marks stage 3.
It is unclear whether either participant will ever reach stage 4. As
they are adult learners and as Vietnamese has relatively simple
margins compared with English, which makes it more difficult to
acquire the fairly complex syllable structures of English, it is doubtful
whether either of them will ever approximate a native speaker in
phonology. However, it is quite possible that both Anh and Nhi will
reach a transitional stage between stage 3 and 4, wherein they have
acquired many features of English phonology, but where the CCCC
coda, for example, along with the interdental fricatives, are still
marked by modifications.
7. 2 Interactions among linguistic and social constraints
It is possible to establish a tentative framework for the interaction
among primary linguistic constraints, which affect acquisition and
developmental sequences, secondary linguistic constraints, which
affect production, and social constraints, which affect rate of
development and extent and duration of LI transfer effects.
Primary linguistic constraints, which include the effects of LI
transfer, developmental processes, and markedness, constrain the
sequence of development of syllable margins both within and across
time. This is initially via the equivalence classifications the learners
make between their LI and their perceptions of the sounds in the L2.
Additionally, primary linguistic constraints affect how consonants are
modified based on LI transfer, developmental and markedness
effects, which change across time. Initially, consonants are modified
towards the LI as these effects may be predominant while as the
learners move through the stages of development, these effects lessen
as they begin building an L2 phonology. At this point, developmental
and markedness effects increase, especially for those codas that are
not similar to those that exist in the learners' LI.
Secondary constraints such as data type and linguistic environment
appear to affect production accuracy. In fact, production effects
follow a fairly predictable pattern: the participants tend to make
orthographic mistakes in the reading data, mispronouncing words
based on orthographic miscues. However, these types of errors are
rare in the interview data, which in turn promotes epenthesis due to
the frequent pausing. Additionally, this speech style promotes
absence. Interestingly, the absence that is promoted by the interview
data follows a different pattern from that in the reading data even for
the same coda, indicating that these are two different types of effects.
In the reading data, for example, as a CC or CCC coda is beginning
to be acquired, the absence patterns may be explained by LI transfer
effects or markedness. Therefore, absence is coda specific, and may
occur for the initial, medial or final consonant. However, in the
absence data for the interview data, for those codas which are
beginning to be produced accurately in the reading data, absence is
always of the final consonant in the coda, even if when the coda was
begin developed, as indicated by the reading data, absence was of the
initial or medial consonant in the coda.
Another secondary constraint which has a strong effect on
production regardless of data type is the effect of linguistic
environment, which although it is a constraint based on the effects
of LI transfer, is considered secondary since it appears to be a rule
that is activated after the primary constraints since some codas are
produced accurately after a vowel, but then are absent after a
diphthong. Therefore, the codas themselves may be acquired, but the
preceding linguistic environment affects the production accuracy of
the coda.
Finally, social barriers in terms of the language use environment
provides opportunities for and investment in meaningful L2 use,
which may be necessary to push learners through the stages of
development and minimize LI retention. Social constraints also
provide learners with the rich input necessary for L2 development,
without which they may not move through the stages of L2
acquisition, as well as potentially increasing the effect and duration
of LI transfer effects, for example, LI retention, if the language use
environments of the learner are primarily LI.
In summary, social barriers appear to affect the rate of development
across time and across stages. The stages of development appear to be
constrained by linguistic effects. Finally, production across all the
stages may be constrained by secondary effects such as linguistic
environment and data type, which may not be indicative of actual
acquisition of the syllable margins, but rather a temporary effect.
7. 3 The effect of social barriers on L2 Acquisition
The social barriers operating on the English language development of
the participants in this study are intricately interconnected as well as
multidimensional in nature. At the nexus of these constraints are the
social environments of language use. These language use environ-
ments may be: a) supportive of L2 use; b) unsupportive of L2 use; or
c) predominantly LI use contexts. The participants encounter these
environments across work, home, college and social situations, and
their reactions to these environments based on social identity, gender
roles, acculturation and their levels of investment in L1/L2 language
use are also important factors in how social constraints affect SLA.
Based on the findings from this study, it appears that a supportive
L2 use atmosphere, such as the one Nhi has at work and at college,
provides increased opportunities to use the L2, and facilitates the
learner's investment in L2 use. Both of these may lead to greater
opportunities for complex interactions. This then could lead to
increased negotiation of meaning in the L2, which is hypothesized to
lead the development of the L2, in this case, the development of
syllable onsets and codas.
In contrast, an unsupportive L2 atmosphere, such as the one Anh
has at work and at college, appears to minimize the opportunities
learners have to use the target language. The environment itself as
well as the few opportunities to use the L2 may in turn lead the
learner to minimize their investment in L2 use, as the learner
perceives that it is anxiety provoking, and/or not beneficial, which is
Anh's case. Instead, the learner may invest in other things. This is
illustrated by Anh's investment in developing her nail technician
skills over her English skills in her work environment as she
perceives that her investment in the former is more beneficial in the
long run.
Finally, the effect of an LI environment on L2 development, such
as Anh and Nhi's environments at home and Anh's environment at
work when there are no customers present, does not provide learners
with many opportunities for L2 use, and this in turn decreases the
learner's investment in L2. This environment inhibits L2 develop-
ment; instead, LI use is promoted, as is the learners' investment in
LI use, which in turn facilitates LI retention rather than L2
development.
The participants' reactions to these environments are also import-
ant. Social identity, levels of perceived acculturation and investment
in language use, are all important social constraints that affect how
the participants react to their various social environments of language
use. For example, Nhi perceives himself to be American, and believes
he is viewed as an American by others because he possesses American
goods (e. g. car, clothes, etc. ). His view that he has an American social
identity helps him gain confidence in forging in-group memberships
with Americans, which provide him with more opportunities for
English language use. Anh, on the other hand, is confused about her
social identity. She works with many Vietnamese who have retained
their LI identities within the L2 culture. She does not want to be like
them, and in comparison to them perceives herself and is perceived by
them as being more Americanized. Though more Americanized than
her co-workers, she is not 'American' in comparison with her brothers
and sisters, and even her daughter, and she loses opportunities to use
English as she cedes these opportunities to the more 'American'
speaker due to her lack of confidence in herself as a speaker of
English. Perceived level of acculturation is an important component
of Anh's perception of her identity, as it also is for Nhi. However, this
is in a contrastive manner to Anh as Nhi's perceived acculturation
gives him greater confidence to interact with people, which in turn
gives him greater English language use opportunities.
What Anh and Nhi do have in common is their positive attitudes
towards members of the target culture, and high levels of motivation
to learn English. Therefore, these constructs alone do not help
explain the different reactions to their social environments of
language use since Anh does not appear to take advantage of the
chances she has to speak English all day at work while Nhi appears to
actively seek and exploit any opportunities for L2 use. One possible
explanation is personality, since Anh is typically introverted and
anxious, while Nhi is outgoing and more confident, the latter of which
has been linked to higher levels of L2 proficiency (Strong 1983).
However, as the study indicated, Anh only appeared to be introverted
in some social situations, such as work, and not in her interviews with
me. Investment, a term coined by Peirce (1995), in L2 use may be a
better explanation of the differential reactions to supportive and
unsupportive language use environments illustrated by the learners.
For example, Anh does not invest heavily in English language use in
her work environment, even though she wants to, because of the
difficulties in communicating with her customers, and the fact that
the communication does not help her learn English, of which she is
very aware. Additionally, she is more invested in being the financial
caretaker for the family and therefore in being a good nail technician.
In contrast, her interviews with me are perceived to be the only
opportunity she has to use English in a supportive and relaxed
atmosphere and she invests heavily in L2 use in this environment.
Nhi, on the other hand, invests in language use in his work
environment, which is very supportive of L2 use, and at college,
which in turn provides him with increased opportunities for L2 use.
When examined integratively, it becomes apparent that having a
supportive L2 use environment may be a crucial, though not in itself
sufficient, stimulus for L2 acquisition to take place. Just as important
are the participants' reactions to the language use environment -
accepting, as in Ann's case, or exploitive, as in Nhi's case. These
reactions are tied to social identity and perceived levels of accultura-
tion as well as the perceived gains from investing time and energy, etc.
in L2 use. Obviously, Nhi's social environments facilitate investment
in L2 use since they are mostly supportive in nature. Even when faced
with an LI use environment when his sister-in-law moves in next
door, Nhi's reaction is to exploit the opportunity by practising his
English with his English-speaking nieces. The majority of Aim's social
environments are supportive for L2 use on the surface - working with
English speaking clients and taking ESL classes. However, in
actuality, these environments are unsupportive for L2 use since the
interactants also have difficulties with English (both at college and
work), and her classmates often rely on Spanish language use among
themselves and with the teacher in order to understand the class
lessons. Therefore, her investment in L2 use in these environments is
low, which then decreases her opportunities for engaging in mean-
ingful communication. These very different social environments and
reactions to the social environments help explain Anh's and Nhi's
differential levels of L2 development across and within time.
The language use environments examined above help to explain
the different rates of development of syllable onsets and codas by Anh
and Nhi across time. As discussed earlier, not only did Nhi appear to
acquire syllable onsets and codas at a faster rate than Anh, but in his
development of these structures, Nhi appeared to have less LI
retention, and was able to build upon his emerging L2 repertoire to a
greater extent than Anh. In other words, not only did he move
through the stages of development at a faster rate but within each
stage he also more closely approximated target-like onset and coda
production than Anh.
As these two language learners have similar LI and English
language backgrounds, are similar in age, and came to the US at the
same time, it is important to examine social constraints in order to
understand their differential levels of L2 achievement both within and
across time as the participants appear to be constrained by the same
linguistic effects but at different rates across time.
In March, time 1, Anh's social contexts of language use are work,
home, college, her interviews with me and social opportunities with
classmates. While a number of these are supportive for L2 use, such
as college and social interactions, these L2 use opportunities are with
other non-native speakers of English who have limited English skills,
which may not provide her with the input necessary to push her L2
development. However, her interviews with me are very supportive in
L2 use and do push her to negotiate and clarify meaning. At work, she
has two types of language environments, LI use with her co-workers
and her boss, and L2 with her customers. Although her interactions
with customers provides her with an opportunity to practise the target
language, the language use environment is actually unsupportive due
to pressure from her boss, anxiety about work and limited ability to
communicate with her Spanish-speaking customers. In terms of her
linguistic development, at this time she appears to be at stage 1 of
development.
At time 2, in August, her opportunities to use English have become
more limited. She is no longer at college, and her social connections
to her classmates have been broken, replaced by Vietnamese co-
workers and her family, as her sister moves next door. The other
language opportunities at home, in the interviews and at work remain
the same. By time 2, Anh is still in stage 1 in her language
development, while moving towards stage 2. Her development of
English syllable margins is also constrained by LI retention. The fact
that she is still in stage 1 could be due to her restricted opportunities
to engage in meaningful interaction in the L2, and the tight LI social
structures around her may promote LI retention, both of which may
minimize her opportunities for L2 development.
This only changes slightly by time 3. Her home, work and interview
language use environments remain the same. Her social connections
also stay the same: Vietnamese co-workers or family members. While
she does begin college again, Anh's L2 use opportunities do not
improve as this is an unsupportive L2 use environment: her language
learning is non-interactive via a videocassette tape and most of her
classmates are Spanish-speakers and she has a difficult time commu-
nicating with them. Her language development is relatively limited as
well, as she remains at stage 2, only beginning to move into stage 3.
LI retention is decreasing, but still constrains her development of
English syllable margins along with her limited opportunities for
English language use.
It is clear that Nhi's language use environments differ from Anh's,
which may be a reason for their differing rates of development of
syllable margins across time. At time 1, in March, his home and social
language use opportunities are primarily in Vietnamese and he is not
enrolled in college. Like Anh, his interviews with me are also in a
supportive L2 use environment. However, unlike Anh, his work is an
extremely supportive L2 use environment, wherein he receives input,
feedback and has meaningful interactions that push him to negotiate
meaning. At time 1, he is also in stage 1 of L2 development, for the
most part, though he appears to be in the initial stages of stage 2.
By time 2, his L2 use opportunities have expanded, contrary to
Anh's. Nhi has begun taking ESL classes, and his classroom
environment is supportive for L2 use. In addition, a new American
co-worker at his job provides him with even more opportunities for
English language use as she is extremely outgoing and interactive with
him. While his home language use opportunities remain primarily
Vietnamese, his social opportunities expand to include English
language practice, as he views the move of his sister-in-law, along
with his two native English speaking nieces, as giving him more
chances to speak English. Additionally, his interviews with me remain
supportive of L2 use. It is not surprising then to see that he has
reached another stage of syllable margin development, and is at the
beginning of stage 3. Additionally, as he is not surrounded by as many
LI environments in comparison with Anh, it is not surprising that his
LI retention is not as great as hers.
These language use environments, which are primarily supportive
of L2 use, remain stable across time, and by time 3, December, Nhi
has reached stage 3 of syllable margin development. LI transfer
effects continue to decrease, while L2 development increases across
time for Nhi.
By the end of the study, Anh and Nhi have different perspectives on
their English language abilities and how their ability has changed over
time. As Nhi states:
I begin to work at company... I didn't hear and didn't speak...
English everybody... and now... I speak English everybody in
my department... in my company... and I speak they are
understand... they understood I speak... because study English
... with work mates...
Nhi perceives that his English has improved significantly across time
because when he began working in the factory, he was unable to
communicate in English and he now understands and is understood
by everybody. He credits his co-workers with helping him - in fact, he
states he studies English with them.
In contrast, while Ann's listening skills have improved, she still feels
that her speaking skills, especially in the area of pronunciation, are
weak. As she states:
Now I think I can hear... I can hear my customer speak... I
understand... fifty or sixty percent... my customer talk long
sentences... sometimes I don't understand... sometime I talk
with customer understood but I try because not confident...
because my customer Mexican... few customer they speak English
very well... almost they speak English very poor. . . they are
always speak short sentences yes I understood but I answer I
answer sometime... they don't understand... they don't
understood because I pronunciation very bad.
A loglinear analysis was also conducted in order to determine whether
the differences in accurate production between the two participants
were significant, and whether differences in production by length and
time were significant; it was found that all three factors, participant,
time and margin length, were statistically significant at p <. 025
(df = 4, total = 11. 4385).
By examining the interaction among social and linguistic con-
straints within and across time, it is possible to explain, in part, why
Anh and Nhi appear to acquire English syllable margins to such
differing extents. First of all, Ann's opportunities for meaningful
communication in the L2 are extremely limited at each time, and
become more limited across time. She has few opportunities to
engage in meaningful exchanges which could push her to modify her
output through the clarification and negotiation of meaning, as well
as providing her with the rich input which may be necessary for L2
development (Ellis 1994). While Anh does, on the whole, acquire
some of the L2 across time, her development is much slower than
Nhi's, and characterized by more LI retention than Nhi's, possibly as
her LI interactions and environments are qualitatively and quantita-
tively greater than Nhi's. In comparison, Nhi has opportunities both
at college and at work to engage in the type of communicative
behaviours that may push him in L2 acquisition, and overall, has few
opportunities for LI use, which may in part explain his lesser degree
of LI retention than Anh. Therefore, by examining the linguistic
findings in light of each participants' social contexts of language use,
which take into consideration both investment in L1/L2 use and
opportunities for language development, the differential rates and
sequences of development between the two participants may be
explained, in part.
8
Conclusions
8. 1 The acquisition of an L2 Phonology
The interaction among social and linguistic constraints is complex
and dynamic. Social constraints affect learners' levels of investment in
L2 use as well as opportunities for L2 use, which are necessary in
order for learners to initiate and utilize the opportunities to engage in
meaningful interactions in the L2, which may provide chances to
negotiate meaning. Social constraints also affect rate of development
by providing supportive or unsupportive (or LI only) environments
for language use. Additionally, environments that are LI rich and L2
poor may promote greater LI retention and thus increase the effect
and duration of LI transfer effects. It is important to note that social
constraints are dynamic and change across time and space, and are
different among learners even in similar situations, as is exemplified
by the different cases of Ann and Nhi. Therefore, while the type of
social constraints operating on language development may be similar
across learners, individual learners' social situations of language use
and their reactions to these situations are not generalizable.
Linguistic constraints appear to be similar across learners from the
same LI backgrounds as well as social situations, to some extent.
Overall, it may be possible to able to generalize these linguistic
constraints across learners from the same LI backgrounds and in
general, to most L2 learners, though more research is necessary.
Linguistic constraints are also interconnected and do change across
time. LI transfer appears to be the primary linguistic constraint in the
initial stages of SLA - LI transfer is facilitative for SLA via
equivalence classifications between the LI and the L2, and the
transfer of LI sounds into the L2 when similarities exist. However, LI
transfer effects can also be debilitating through the transfer of LI
phonotactics that do not exist in the L2, and the difficulties of
acquiring more marked L2 structures that are not in the LI.
Developmental effects constrain acquisition orders in terms of place
CONCLUSIONS 165
and manner of articulation (front before back consonants, voiceless
before voiced consonants). Markedness also constrains acquisition
orders in terms of length, type and sonority. All these linguistic
constraints are primary and can affect the development and produc-
tion of a single segment simultaneously. Secondary constraints such
as linguistic environment and data type also affect production.
In summary, the acquisition of English syllable margins by native
speakers of Vietnamese is a complex process which is affected by
multiple constraints, primary and secondary linguistic constraints as
well as social constraints, which change across time. Furthermore,
while these constraints have similar effects across learners in terms of
the sequence of development of syllable onsets and codas, they also
differ across learners and time because of different social contexts of
language use, and different levels of investment in and opportunities
for L2 use. Therefore, it is not possible to disentangle these
constraints from one another, as prior research has often done in
investigating only one facet of SLA such as LI transfer or
markedness. Rather, these constraints should be examined simultan-
eously and interconnectedly so that a multidimensional perspective
on SLA can be developed, which is what this research project has
sought to do.
8. 2 Limitations to the study
There are several limitations to this study: first, only data on syllable
onsets and codas were analysed, which does not represent L2
phonological acquisition as a whole. However, syllable margins are an
important unit of analysis, and are a problematic area of acquisition
for many non-native speakers of English, especially speakers of
Vietnamese. Additionally, onset and coda production affects com-
municative efficacy and thus research in this area of phonology has
important pedagogical implications.
A second limitation of this study is that it is confined to speakers of
Vietnamese. However, many L2 phonological processes are similar
cross-linguistically, and thus information gathered on Vietnamese
speakers provides important insights into the process of L2 phono-
logical acquisition.
A third limitation of this study is the small sample size of only two
participants. This is not seen as a limitation within L2 phonology
research, however, where sample sizes are often very small, especially
for longitudinal research. A more important consideration in L2
phonological research is the number of tokens gathered from each
participant. Since this project was longitudinal, lasting 10 months.,
approximately 16, 000 individual pieces of phonological data were
analysed in this study, which is a robust data set, enabling greater
levels of analysis and interpretability.
8. 3 Suggestions for future research
Based on the findings from this research project, the following
suggestions for further research are made:
1. A research project in longer duration, for example several years,
etc. would make it possible to further examine the sequence of
development as well as to confirm the findings from this study.
This type of project would also help to determine the eventual
proficiency level the learners may attain, and which social and
linguistic constraints facilitated the development to this level
and prevented the learners from attaining native-like profi-
ciency, if they did not reach this level.
2. In addition, it would be beneficial to conduct research with
learners who have just arrived within several months of the onset
of the study so the early stages of acquisition can be documented
with more detail than was possible in this study.
3. Additionally, a research study that was ethnographic in meth-
odology and included observations of the participants in the
different social environments of language use would facilitate
understanding of the complex notion of 'investment' - why
learners invest in English language use in some situations and
not in others, and to what extent they invest. Furthermore, this
type of research project would provide expanded insight into the
learners' language use opportunities, how they react to these
opportunities, and in what ways and to what extent the learners'
social contexts at home, work, college, etc. are unsupportive or
supportive of L2 use. This type of research project could also
gather linguistic data on the learners' communicative inter-
actions, which is necessary in understanding the scaffolding and
negotiations of meaning that take place, both of which may
facilitate SLA, during the learners' interactions in English.
4. It is also suggested that this research project be repeated with
more learners of Vietnamese in order to substantiate the
findings from this study and to further elucidate on how
linguistic constraints, especially LI transfer effects, constrain
the development of L2 syllable margins.
CONCLUSIONS 167
5. Moreover, it is suggested that this research project be repeated
with learners of other language backgrounds in order to make
comparisons between the linguistic and social constraints which
affected the learners of this study with those which learners from
other linguistic and cultural backgrounds may face.
6. This study only examined one area of L2 development - syllable
margins. There are many other areas of L2 phonological
development that need further investigation through a longitu-
dinal research study which examines both linguistic and social
constraints, such as vowels and intonation.
8. 4 Epilogue
This study was completed in 1999. Since that time, I have remained
in contact with the Nguyen family, and have kept up with the major
events in their lives. The daughter, Minh, graduated from the
University of Arizona with a double-major. In part because of her
frustrations about her English language skills, and the pressure put on
her by the bosses of all three nail shops she worked at, and the
pressure she puts on herself to be the financial caretaker of the family,
Anh opened her own nail salon in 2000, even though this makes it
even more difficult for her to pursue her dream of attending college so
she can become a computer programmer. However, she sublimates
this desire to her family's well-being as she perceives that financial
caretaking is the primary and immediate goal even at the expense of
her own future. Nhi began working four days a week, ten hours a day,
at the golf factory, so that he could work part-time in the nail salon.
He hoped to eventually get his own nail technician licence so he could
work in the family business - for Anh - full time.
References
Abrahamsson, N. (2001), 'Universal constraints on L2 coda production', in
N. Abrahamsson, Acquiring L2 Syllable Margins. Stockholm, Sweden:
Centre for Research on Bilingualism, pp. 1-38.
Adamson, H. D. and Regan, V. M. (1991), 'The acquisition of community
speech norms by Asian immigrants learning English as a second
language'. Studies in Second Language Acquisition, 13, 1-22.
Altenberg, E. P. and Vago, R. M. (1987), 'Theoretical implications of an
error analysis of second language phonology production', in G. loup and
S. H. Weinberger (eds) Interlanguage Phonology: The Acquisition of a
Second Language Sound System. Cambridge, MA: Newbury House
Publishers, pp. 148-64.
Anderson, J. (1987), 'The markedness differential hypothesis and syllable
structure difficulty', in G. loup and S. H. Weinberger (eds), Interlanguage
Phonology: The Acquisition of a Second Language Sound System.
Cambridge, MA: Newbury House Publishers, pp. 279-91.
Bayley, R. (1996), 'Competing constraints on variation in the speech of
adult Chinese learners of English', in R. Bayley and D. R. Preston (eds),
Second Language Acquisition and Linguistic Variation. Amsterdam: John
Benjamins Publishing Company, pp. 98-120.
Beebe, L. M. (1980), 'Sociolinguistic variation and style shifting in second
language acquisition'. Language Learning, 30, 433-41.
Beebe, L. M. (1985), 'Input: Choosing the right stuff, in S. M. Gass and C.
G. Madden (eds), Input in Second Language Acquisition. Rowley, MA:
Newbury House Publishers Inc., pp. 404-14.
Beebe, L. M. and Zuengler, J. (1983), 'Accommodation theory: An
explanation for style shifting in second language dialects', in N.
Wolfson and E. Judd (eds), Sociolinguistics and Language Acquisition.
Rowley, MA: Newbury House Publishers, pp. 195-213.
Benson, B. (1988), 'Universal preference for the open syllables as an
independent process in interlanguage phonology'. Language Learning, 38,
(2), 221-35.
Blackledge, A. (2001), 'Complex positionings: Women negotiating identity
and power in a minority urban setting', in A. Pavlenko, A. Blackledge, I.
Filler, and M. Teutsch-Dwyer (eds), Multilingualism, Second Language
Learning, and Gender. Berlin: Mouton de Gruyter, pp. 53-75.
REFERENCES 169
Broselow, E. (1987), 'Non-obvious transfer: On predicting epenthesis
errors', in G. loup and S. H. Weinberger (eds), Interlanguage Phonology:
The Acquisition of a Second Language Sound System. Cambridge, MA:
Newbury House Publishers, pp. 292-304.
Broselow, E. and Finer, D. (1991), 'Parameter setting in second language
phonology and syntax'. Second Language Research, 7, (1), 3559.
Cancino, H., Rosansky, E. and Schumann, J. (1978), 'The acquisition of
English negatives and interrogatives by native Spanish speakers', in E.
Hatch (ed. ), Second Language Acquisition. Rowley, MA: Newbury House.
Carlisle, R. S. (1991), 'The influence of syllable structure universals on the
variability of interlanguage phonology', in A. D. Volpe (ed. ), The
Seventeenth LACUS Forum, 1990. Lake Bluff, IL: Linguistic Association
of Canada and the United States.
Carlisle, R. S. (1994), 'Markedness and environment as internal constraints
on the variability of interlanguage phonology', in M. Yavas (ed. ), First and
Second Language Phonology. San Diego, CA: Singular Publishing Group,
Inc, pp. 223-49.
Carlisle, R. S. (1997), 'The modification of onsets in a markedness
relationship: Testing the interlanguage structural conformity hypothesis'.
Language Learning, 47, (2), 327-61.
Carlisle, R. S. (1998), 'The acquisition of onsets in markedness relation-
ships: A longitudinal study'. Studies in Second Language Acquisition, 20,
245-60.
Celce-Murcia, M., Brinton, D. M. and Goodwin, J. M. (1996), Teaching
Pronunciation: A Reference for Teachers of English to Speakers of Other
Languages. Cambridge: Cambridge University Press.
Gumming, A. and Gill, J. (1992), 'Motivation or accessibility? Factors
permitting Indo-Canadian women to pursue ESL literacy instruction', in
B. Burnaby and A. Gumming (eds), Socio-Political Aspects of ESL
Education in Canada. Toronto: OISE Press, pp. 241-52.
Dauer, R. M. (1992), Accurate English: A Complete Course in Pronunciation.
Harlow: Pearson Longman.
Day, R. (1985), 'The use of target language in context and second language
proficiency', in S. Gass and C. Madden (eds), Input in Second Language
Acquisition. Rowley, MA: Newbury House.
Dickerson, L. J. (1975), 'The learner's interlanguage as a system of variable
rules'. TESOL Quarterly, 9, (4), 401-7.
Dickerson, L. J. and Dickerson, W. (1977), 'Interlanguage phonology:
Current research and future directions', in S. P. Corder and E. Roulet
(eds), Actes du 5 erne Colloque de Linguistique Apliquee de Neuchdtel: The
Notions of Simplification, Interlanguages and Pidgins and their Relation to
Second Language Acquisition. Geneva: Droz, pp. 1829.
Dowd, J., Zuengler, J. and Berkowitz, D. (1990), 'L2 social marking:
Research issues'. Applied Linguistics, 11, (1), 16-29.
Eckman, F. R. (1987), 'The reduction of word-final consonant clusters in
interlanguage', in A. James and J. Leather (eds), Sound Patterns in Second
Language Acquisition. Dordrecht, Holland: Foris Publications, pp. 143-
62.
Eckman, F. R. (1991), 'The structural conformity hypothesis and the
acquisition of consonant clusters in the interlanguage of ESL learners'.
Studies in Second Language Acquisition, 13, 23-37.
Eckman, F. R. and Iverson, G. K. (1993), 'Sonority and markedness among
onset clusters in the interlanguage of ESL learners'. Second Language
Research, 9, 3, 234-52.
Edge, B. A. (1991), 'The production of word-final voiced obstruents in
English by LI speakers of Japanese and Cantonese'. Studies in Second
Language Acquisition, 13, 377-93.
Ehrlich, S. (1997), 'Gender as social practice: Implications for second
language acquisition'. Studies in Second Language Acquisition, 19, 421-46.
Ehrlich, S. (2001), 'Gendering the 'learner': Sexual harassment and second
language acquisition', in A. Pavlenko, A. Blackledge, I. Filler and M.
Teutsch-Dwyer (eds), Multilingualism, second language learning, and
gender. Berlin: Mouton de Gruyter, pp. 103-29.
Ellis, R. (1994), The Study of Second Language Acquisition. Oxford: Oxford
University Press.
Flege, J. E. and Davidian, R. D. (1984), 'Transfer and developmental
processes in adult foreign language speech production'. Applied
Psycholinguistics, 5, 323-47.
Flege, J. E., Frieda, E. M. and Nozawa, T. (1997), 'Amount of native-
language (LI) use affects the pronunciation of an L2'. Journal of Phonetics,
25, 169-86.
Flege, J. E., Munro, M. J. and MacKay, I. R. A. (1996), 'Factors affecting
the production of word-initial consonants in a second language', in R.
Bayley and R. Preston (eds), Second Language Acquisition and Linguistic
Variation. Amsterdam: John Benjamins Publishing Company, pp. 47-73.
Gal, S. (1991), 'Between speech and silence: The problematics of research
on language and gender', in M. Di Leonardo (ed. ), Gender at the cross-
roads of knowledge. Berkeley: University of California Press, pp. 175-203.
Gardner, R. (1980), 'On the validity of affective variables in second
language acquisition: Conceptual, contextual and statistical consider-
ations'. Language Learning, 30, 255-70.
Gardner, R. and Lambert, W. (1972), Attitudes and Motivation in Language
Learning. Rowley, MA: Newbury House.
Gatbonton, E. (1975), 'Systematic Variations in Second Language Speech:
A Sociolinguistic Study' (unpublished doctoral dissertation, McGill
University).
Gatbonton, E. (1978), 'Patterned phonetic variation in second language
speech: A gradual diffusion model'. Canadian Modern Language Review,
34, 335-47.
Giles, H. and Byrne, J. L. (1982), 'An intergroup approach to second
REFERENCES 171
language acquisition'. Journal of Multilingual and Multicultural
Development, 3, 17-40.
Goldstein, T. (1995), 'Nobody is talking bad', in K. Hall and M. Bucholtz
(eds), Gender Articulated: Language and the Socially Constructed Self. New
York: Routledge.
Goldstein, T. (2001), 'Researching women's language practices in multi-
lingual workplaces', in A. Pavlenko, A. Blackledge, I. Filler and M.
Teutsch-Dwyer (eds), Multilingualism, Second Language Learning, and
Gender. Berlin: Mouton de Gruyter, pp. 77-101.
Greenberg, J. (1978), Language Universal. The Hague: Mouton Publishers.
Greenberg, J. H., Ferguson, C. and Moravcsik, E. (1978), Universals of
Human Language (Vol. 2). Stanford, CA: Stanford University Press.
Guion, S. G., Flege, J. E. and Loftin, J. D. (2000), 'The effect of LI use on
pronunciation in Quichua-Spanish bilinguals'. Journal of Phonetics, 28,
27-42.
Guy, G. R. (1980), 'Variation in the group and the individual: The case of
final stop deletion', in Locating Language in Time and Space. Academic
Press.
Hammarberg, B. (1997), 'Conditions on transfer in phonology', in A. James
and J. Leather (eds), Second-Language Speech: Structure and Process.
Berlin: Mouton de Gruyter.
Hammond, M. (1999), The Phonology of English: A Prosodic Optimality-
Theoretic Approach. Oxford: Oxford University Press.
Hancin-Bhatt, B. and Bhatt, R. M. (1997), 'Optimal L2 syllables:
Interactions of transfer and developmental effects'. Studies in Second
Language Acquisition, 19, 331-78.
Hansen, J. G. (2001), 'Linguistic constraints on the acquisition of English
syllable codas by native speakers of Mandarin Chinese'. Applied
Linguistics, 22, 338-65.
Hansen, J. G. (2004), 'Developmental sequences in the acquisition of L2
syllable codas: A preliminary study'. Studies in Second Language
Acquisition, 26, 85-124.
Hardison, D. M. (1996), 'Bimodal speech perception by native and
nonnative speakers of English: Factors influencing the McGurk Effect'.
Hatch, E. (1978), Second Language Acquisition. Rowley, MA: Newbury
House.
Hecht, B. F. and Mulford, R. (1982), 'The acquisition of a second language
phonology: Interaction of transfer and developmental factors'. Applied
Linguistics, 3, 313-28.
Heidelberger Forschungsprojekt 'Pidgin Deutsch' (1978), 'The acquisition
of German syntax by foreign migrant workers', in D. Sankoff (ed. ),
Linguistic Variation: Models and Methods. New York: Academic Press.
Herbert, R. K. (1986), Language Universals, Markedness Theory, and Natural
Phonetic Processes. Berlin: Mouton de Gruyter.
Hieke, A. E. (1987), 'Absorption and fluency in native and non-native
casual English speech', in A. James and J. Leather (eds), Sound Patterns in
Second Language Acquisition. Dordrecht, Holland: Foris Publications,
pp. 41-58.
Hodne, B. (1985), 'Yet another look at interlanguage phonology: The
modification of English syllable structure by native speakers of Polish'.
Ingram, D. (1976). Phonological Disability in Children. New York: Elsevier.
Jakobson, R. (1968), Child Language Aphasia and Phonological Universals.
The Hague: Mouton.
James, A. R. (1988), The Acquisition of a Second Language Phonology: A
Linguistic Theory of Developing Sound Structures. Tubingen, The
Netherlands: Gunter Narr Verlag.
Jensen, J. T. (1993), Current Issues in Linguistic Theory, 99: English Phonology.
Amsterdam: John Benjamins Publishing Company.
Jones, D. (1997), English Pronouncing Dictionary (15th ed. ). Cambridge:
Cambridge University Press.
Kahn, D. (1976), 'Syllable-Based Generalizations in English Phonology'
(unpublished doctoral dissertation, MIT).
Kouritzen, S. (2000), 'Immigrant mothers redefine access to ESL classes:
Contradiction and ambivalence'. Journal of Multilingual and Multicultural
Labov, W. (1969), 'The study of language in its social context'. Studium
Generale, 23, 30-87.
Lambert, W. E. (1974), 'Culture and language as factors in learning and
education', in F. E. Aboud and R. D. Meade (eds), The Fifth Western
Symposium on Learning. Bellingham, WA: Western Washington State
College, pp. 91-122.
Lanoue, G. (1991), 'Language loss, language gain: Cultural camouflage and
social change among the Sekani of Northern British Columbia'. Language
in Society, 20, 87-115.
Leather, J. and James, A. (1991), 'The acquisition of second language
speech'. Studies in Second Language Acquisition, 13, 305-41.
Leopold, W. F. (1939), Speech Development of a Bilingual Child, 1:
Vocabulary Growth in the First Two Years. Evanston: Northwestern
University Press.
LePage, R. B. and Tabouret-Keller, A. (1985), Acts of Identity: Creole-Based
Approaches to Language and Ethnicity. Cambridge: Cambridge University
Press.
Losey, K. M. (1995), 'Gender and ethnicity as factors in the development of
verbal skills in bilingual Mexican American women'. TESOL Quarterly,
29, 635-61.
Lukmani, Y. (1972), 'Motivation to learn and language proficiency'.
Language Learning, 22, 261-73.
Lum, J. (1982), 'Marginality and multiculturalism: Another look at
REFERENCES 173
bilingual/bicultural education', in L. Samovar and R. Porter (eds), Culture
Learning Report Vol. 5(1). Honolulu: East-West Center.
Lybeck, K. (2002), 'Cultural identification and second language pronun-
ciation of Americans in Norway'. The Modern Language Journal, 86, 174-
91.
Macken, M. A. (1979), 'Developmental reorganization of phonology: A
hierarchy of basic units of acquisition'. Lingua, 49, 219-53.
Macken, M. A. and Ferguson, C. A. (1981), 'Phonological universals in
language acquisition', in H. Winitz (ed. ), Native Language and Foreign
Language Acquisition. New York: The New York Academy of Sciences,
pp. 110-29.
Major, R. C. (1986), 'The Ontogeny Model: Evidence from L2 acquisition
of Spanish r
1
. Language Learning, 36, 4, 453504.
Major, R. C. (1987a), 'A model for interlanguage phonology', in G. loup
and S. H. Weinberger (eds), Interlanguage Phonology: The Acquisition of a
Second Language Sound System. Cambridge, MA: Newbury House
Publishers, pp. 101-24.
Major, R. C. (1987b), 'The natural phonology of second language
acquisition', in A. James and J. Leather (eds), Sound Patterns in Second
Language Acquisition. Dordrecht, Holland: Fortis Publications, pp. 206-
24.
Major, R. C. (1987c), 'Variation in second language phonology'. ERIC
Reproduction Service Document No. ED 310 622.
Major, R. C. (1994), 'Chronological and stylistic aspects of second language
acquisition of consonant clusters'. Language Learning, 44, (4), 655-80.
Major, R. C. (1996), 'Markedness in second language acquisition of
consonant clusters', in R. Bayley and R. Preston (eds), Second Language
Acquisition and Linguistic Variation. Amsterdam: John Benjamins
Publishing Company, pp. 7596.
Major, R. C. (2001), Foreign Accent: The Ontogeny and Phytogeny of Second
Language Phonology. Lawrence Erlbaum Associates.
Major, R. C. and Faudree, M. C. (1996), 'Markedness universals and the
acquisition of voicing contrasts by Korean speakers of English'. Studies in
Second Language Acquisition, 18, 69-90.
Meisel, J., Clahsen, H. and Pienemann, M. (1981), 'On determining
developmental stages in natural second language acquisition'. Studies in
Milroy, L. (1987), Observing and Analysing Natural Language. Oxford: Basil
Blackwell.
Monshi-Tousi, M., Hosseine-Fatima, A. and Oiler, J. (1980), 'English
proficiency and factors in its attainment: A case study of Iranians in the
United States'. TESOL Quarterly, 14, (3), 365-71.
Mulford, R. and Hecht, B. F. (1980), 'Learning to speak without an accent:
Acquisition of second language phonology'. Papers and Reports on Child
Language Development, 18, 16-74.
Nguyen, D. L. (1970), Vietnamese Pronunciation. University of Hawaii Press.
Norton, B. (2000), Identity and Language Learning. Harlow: Longman.
Odlin, T. (1989), Language Transfer: Cross-Linguistic Influence in Language
Learning. Cambridge: Cambridge University Press.
Ohala, D. (1994), 'Sonority-Driven Cluster Reduction' (unpublished
manuscript, University of Arizona).
Ohara, Y. (2001), 'Finding one's voice in Japanese: A study of the pitch
levels of L2 users', in A. Pavlenko, A. Blackledge, I. Filler, and M.
Teutsch-Dwyer (eds), Mukilingualism, Second Language Learning, and
Oiler, J., Baca, L. and Vigil, F. (1977), 'Attitudes and attained proficiency in
ESL: A sociolinguistic study of Mexican-Americans in the Southwest'.
TESOL Quarterly, (11), 173-83.
Olmsted, D. L. (1971), 'Out of the mouth of babes'. Janua Linguarum, 117.
The Hague: Mouton.
Osburne, A. G. (1996), 'Final cluster reduction in English L2 speech: A case
study of a Vietnamese speaker'. Applied Linguistics, 17, (2), 164-81.
Owens, R. E., Jr. (1996), Language Development: An Introduction (4th ed).
Boston, MA: Allyn and Bacon.
Oyama, S. (1976), 'A sensitive period for the acquisition of a nonnative
phonological system'. Journal of Psycholinguistic Research, 5, (3), 261-83.
Peck, S. (1978), 'Child-child discourse in second language acquisition' in E.
Hatch (ed. ), Second Language Acquisition. Rowley, MA: Newbury House.
Peirce, B. N. (1993), 'Language Learning, Social Identity, and Immigrant
Women' (unpublished doctoral dissertation, University of Toronto).
Peirce, B. N. (1995), 'Social identity, investment, and language learning'.
TESOL Quarterly, 29, (1), 9-31.
Pennington, M. C. (1994), 'Recent research in L2 phonology: Implications
for practice', in J. Morley (ed. ), Pronunciation Pedagogy and Theory: New
Views, New Directions. Alexandria, VA: TESOL.
Piper, T. (1984), 'Observations on the second-language acquisition of the
English sound system'. The Canadian Modern Language Review, 40, (5),
542-51.
Piske, T. and MacKay, I. (1999), 'Age and LI use affects on degree of
foreign accent in English'. Proceedings of the 14
th
International Congress of
Phonetic Sciences, San Francisco, 1-7 August 1999. 1433-6.
Piske, T., MacKay, I. R. A. and Flege, J. E. (2001), 'Factors affecting degree
of foreign accent in an L2: A review'. Journal of Phonetics, 29, 191-215.
Preston, D. (1989), Sociolinguistics and second language acquisition. Oxford:
Blackwell.
Preston, D. (1996), 'Variationist perspectives on second language acquisi-
tion', in R. Bayley and R. Preston (eds), Second Language Acquisition and
Linguistic Variation. Amsterdam: John Benjamins Publishing Company,
pp. 1-45.
Rand, D. and Sankoff, D. (1990), GoldVarb Version 2: A Variable Rule
REFERENCES 175
Application Program for the Macintosh. Centre de recherches mathema-
tiques: University of Montreal.
Rutledge, P. J. (1992), The Vietnamese Experience in America. Bloomington,
IN: Indiana University Press.
St. Martin, G. M. (1980), 'English language acquisition: The effects of
living with an American family'. TESOL Quarterly, 14, (3), 388-90.
Santry, P. A. (1997), The Way South Vietnamese Pronounce English. Frankfurt
am Main: Hector.
Sato, C. J. (1984), 'Phonological processes in second language acquisition:
Another look at interlanguage syllable structure'. Language Learning, 34,
(4), 43-57.
Sato, C. J. (1985), 'Task variation in interlanguage phonology', in S. M.
Gass and C. G. Madden (eds), Input in Second Language Acquisition.
Cambridge, MA: Newbury House.
Saunders, N. (1987), 'Morphophonemic variations in clusters in Japanese
English'. Language Learning, 37, 247-72.
Schmidt, R. W. (1974), 'Sociostylistic Variation on Spoken Egyptian
Arabic: A Re-Examination of the Concept of Diglossia' (unpublished
doctoral dissertation, Brown University).
Schmidt, R. W. (1983), 'Interaction, acculturation, and the acquisition of
communicative competence: A case study of an adult', in N. Wolfson and
E. Judd (eds), Sociolinguistics and Language Teaching. Rowley, MA:
Newbury House.
Schmidt, R. W. (1987), 'Sociolinguistic variation and language transfer in
phonology', in G. loup and S. H. Weinberger (eds), Interlanguage
Schumann, J. (1986), 'Research on the acculturation model for second
language acquisition'. Journal of Multilingual and Multicultural
Schumann, J. (1990), 'Extending the scope of the acculturation/pidginiza-
tion model to include cognition'. TESOL Quarterly, 24, 667-84.
Siegal, M. (1996), "The role of learner subjectivity in second language
sociolinguistic competency: Western women learning Japanese'. Applied
Linguistics, 17, (3), 356-82.
Skaer, P. M. (1984), 'Language sound systems and second language
acquisition'. ERIC Reproduction Document Service No. ED 248 712.
Spada, N. (1986), 'The interaction between type of contact and type of
instruction: Some effects on the L2 proficiency of adult learners'. Studies
in Second Language Acquisition, 8, 181-200.
Spencer, A. (1996), Phonology: Theory and Description. Oxford: Blackwell
Publishers.
Strauss, A. and Corbin, J. (1990), Basics of Qualitative Research: Grounded
Theory Procedures and Techniques. Newbury Park: Sage Publications.
Strong, M. (1983), 'Social styles and second language acquisition of
Spanish-speaking kindergarteners'. TESOL Quarterly, 17, 241-58.
Strong, M. (1984), 'Integrative motivation: Cause or result of successful
second language acquisition?'. Language Learning, 34, (3), 1-14.
Tahta, S., Wood, M. and Loewenthal, K. (1981), 'Foreign accents: Factors
relating to transfer of accent from the first language to a second language'.
Language and Speech, 24, (3), 265-72.
Tajfel, H. (1974), 'Social identity and intergroup behavior'. Social Science
Information, 13, 65-93.
Tarone, E. (1980), 'Some influences on the syllable structure of
interlanguage phonology'. IRAL, XVIII, (2), 138-52.
Tarone, E. (1987), 'Some influences on the syllable structure of
interlanguage phonology', in G. loup and S. H. Weinberger (eds),
Interlanguage Phonology: The Acquisition of a Second Language Sound
System. Cambridge, MA: Newbury House Publishers, pp. 233-47.
Teutsch-Dwyer, M. (2001), '(Re)constructing masculinity in a new
linguistic reality', in A. Pavlenko, A. Blackledge, I. Filler and M.
Teutsch-Dwyer (eds), Multilingualism, Second Language Learning, and
Tropf, H. (1987), 'Sonority as a variable factor in second language
phonology', in A. James and J. Leather (eds), Sounds Patterns in Second
Language Acquisition. Dordrecht, Holland: Foris Publications, pp. 173-
91.
Vago, R. M. and Altenberg, E. (1977), 'A study of English second language
phonology'. ERIC Reproduction Service Document ED 161 282.
Wagner-Gough, J. (1975), 'Comparative studies in second language
learning'. CAL-ERIC/CLL Series on language and linguistics, 26.
Weinberger, S. H. (1987), 'The influence of linguistic context on syllable
simplification', in G. loup and S. H. Weinberger (eds), Interlanguage
Yavas, M. (1997), 'The effects of vowel height and place of articulation in
interlanguage final stop devoicing'. IRAL, 35, 115-25.
Young, R. (1988), 'Variation and the interlanguage hypothesis'. Studies in
Young, R. and Bayley, R. (1996), 'VARBRUL analysis for second language
acquisition research', in R. Bayley and D. R. Preston (eds), Second
Language Acquisition and Linguistic Variation. Amsterdam: John
Benjamins Publishing Company, pp. 253-306.
Appendix A
Word List
A. Single codas:
/g/
/k/
/t/
/d/
/p/
Ibl
/ s /
Izl
Ate/
/a/
/I/
/tj /
/e/
hrl
/3/
/ff
/m/
/n/
/q/
/j /
/!/
dog
duke, music, thwack, snake, check
schmidt, straight, squat, quiet, throat, that
good
schlep, sheep
flub
less
Tuesday, noise
language
breathe
splash
match
breath, mouth
cave
garage (alternately pronounced with a /ds/ - either pronunci-
ation was considered correct)
hoof
vroom
onion, gwen, hundredth, human, prawn, brown,
language, spring
popular, future, pure, hear, yourself, other
beautiful, j ewel, scowl
B. Two-member codas:
/sp/
/sk/
/qk/
/ids/
/ik/
/ib/
/if
/j j /
grasp
risk
think
merge
fork
curb
dwarf, smurf
porsche
/ I g /
/ in/
liml
/ jp/
/ jl/
/ jtj/
/ ntj/
/ nd3/
3
/
/ mtf
/ mp/
/ I n/
/ 1m/
/ lp/
/ lb/
/ lv/
/ lz/
/ I ds/
/ I k/
/ itj/
/ ij/
/ I d/
/ I k/
I \ S I
/ fB /
/ 10/
/ de /
/ n0/
/ te /
/ j0/
/ JS/
/ ps/
/ I s/
/ fs/
/ ns/
/ 0s/
/ ts/
/ ks/
/ g z/
/ nz/
/ qz/
/ bz/
/ 5z/
burg
turn
worm
harp
g irl
church
lunch
lung e
humph
g rump
kiln
elm
he lp
alb
shelve, twe lve
frills
bulg e
milk
be lch
welsh
fie ld
milk
yourse lf
fifth
he alth
hundre dth
month
eig hth
fourth
pierce
caps, schnaps
pulse
hofs
mons
cloths, smiths
nig hts
bikes
dog s
g uns
sing s
robs
clothe s
WORD LIST 179
/JZ/
/mz/
/vz/
/dz/
/bd/
/zd/
/r)d/
/dsd/
/gd/
/5d/
/Id/
/vd/
/jd/
/nd/
/st/
/nt/
/it/
/It/
/0t/
/It/
/pt/
/tjt/
/mt/
/ft/
/kt/
cars
times
gloves
needs
robbed
pleased
hanged
judged
jogged
bathed
smelled
loved
devoured
behind
moist
point
start
salt
toothed
wished
script
touched
amt
swift
hiked
C. Three-member codas:
/dst/
/t6s/
/sts/
/sks/
/mpt/
/mps/
/ndz/
/nst/
/nts/
/ntjt/
/qst/
/qkt/
/qks/
/imO/
/jmz/
amidst
eighths
lasts
flasks
crammed
' bumps
wands
canst
months, since
lunched
amongst
linked
thanks, sphinx
warmth
worms
/mz/
/jdz/
/ild/
/jpt/
/JSt/
/ikt/
/iGs/
/its/
/ibz/
/jps/
/Ikt/
/Iks/
/IfB/
/Idz/
/Ipt/
/kst/
turns
words
world
harped
worst
worked
worths
arts
serbs
harps
milked
milks
twelfth
molds
helped
taxed
D. Four-member codas:
/jldz/
/If0s/
/kstG/
worlds
twelfths
sixth
Appendix B
Reading Passage
Accent Inventory (Dauer 1992)
When a student from another country comes to study in the United
States, he has to find out for himself the answers to many questions
... and he has many problems to think about. Where should he live?
Would it be better if he looked for a private room off campus or if he
stayed in a dormitory? Should he spend all of his time just studying.
Shouldn't he try to take advantage of the many social and cultural
activities which are offered? At first... it is not easy for him to be
casual in dress, informal in manner, and confident in speech. Little by
little he learns what kind of clothing is usually worn here to be
casually dressed for classes. He also learns to choose the language and
customs that are appropriate for informal situations. Finally he begins
to feel sure of himself. But let me tell you, my friend, this long awaited
feeling doesn't develop suddenly, does it? All of this takes will power.
Appendix C
Initial Questionnaire
I. Biographical information:
1. Name: E-mail: _
Address:
Home phone number: Office phone number:.
2. Age: Sex: MALE FEMALE
3. How long have you been in the United States?
4. Have you lived any other place than Tucson in the United States? YES NO
5. If yes, where and for how long?
6. Why did you come to the US?
7. Did you have any family in Tucson before you arrived? YES NO
8. If yes, who?
9. Did you have any friends in Tucson before you arrived? YES NO
10. If yes, who?
11. Language background:
1. Have you taken the TOEFL test? YES NO
2. If yes, what is your most recent score?
INITIAL QUESTIONNAIRE 183
3. Have you taken a spoken English test such as the TSE or SPEAK test? YES NO
4. If yes, what is your most recent score?
5. What English classes have you taken in the United States (please include ALL ESL
classes you have taken):
6. When did you take them?.
7. Did you study English in Vietnam? YES NO (please circle)
8. If yes, for how long did you study English?
9. When you studied English, did you: (please circle as many as apply):
Translate Learn about American culture
Listen to tapes Study grammatical rules
Write compositions Speak English with native speakers
Read English books Have English dialogues with classmates
Other:
10. Did you learn spoken English in Vietnam? YES NO
11. If yes, did you learn British English or American English?
12. For how long did you learn spoken English in Vietnam?
13. Did you have any opportunities to speak English in Vietnam? YES NO
14. If yes, with whom did you speak English and how often?
15. Comparing yourself to native speakers of English, how well do you think you
speak English? (Please circle)
About the same A little worse Much worse
16. Comparing yourself to other Vietnamese people studying English, how well do
you think you speak English? (please circle)
Much better Better About the same A little worse Much worse
17. Please rate your own proficiency in English in the following areas:
5=Excellent
Grammar
Listening comprehension
Reading comprehension
Writing
Pronunciation
Fluency
Vocabulary
5
5
5
5
5
5
5
Knowledge of American culture 5
4=Good
4
4
4
4
4
4
4
4
3= Average
3
3
3
3
3
3
3
3
2=Poor
2
2
2
2
2
2
2
2
l=Very
Poor
1
1
1
1
1
1
1
1
18. Do you feel different when you speak English than when you speak Vietnamese?
Please explain:
19. Does the way of life in the US make it easy or hard for you to learn English?
Please explain:
III. Language contact:
Please circle your answers to these questions:
b.
9. List the three English-speaking Americans that you speak English with the most.
What is their relationship to you? (For example: teacher, friend, neighbor, bus
driver). How much time do you spend with them each day?
Name of American Relationship Amount of time each day
b.
All
1. How many people in your neighborhood
speak English as a native language: 1
2. How many people in your neighborhood
speak Vietnamese as a native language: 1
3. How many of your friends speak Vietnamese
as a native language: 1
4. How many of your friends speak English
as a native language: 1
5. If you are in school, how many of your
classmates speak English as a
native language? 1
6. How many of your classmates speak
Vietnamese as a native language? 1
7. How many of your workmates
speak English as a native language? 1
Most
2
2
2
2
2
2
2
Some
3
3
3
3
3
3
3
A few
4
4
4
4
4
4
4
None
5
5
5
5
5
5
5
a.
c.
a.
c.
8. List the three friends in Tucson that you see the most every day.
Name of friend
Language used amount of time erach day
IV. English language use:
a) your husband/wife
b) your children
c) your relatives
d) friends
e) neighbors
f) classmates
g) workmates
h) shop/bank
employees
i) others
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
4. Circle the average number of hours each day you watch television in Vietnamese.
0 1/2 1 11/2 2 21/2 3 31/2 4 41/2
5 51/2 6 61/2 7 71/2 8 81/2 9 over 9
5. Circle the average number of hours each day you watch television in English.
0 1/2 1 11/2 2 21/2 3 31/2 4 41/2
5 51/2 6 61/2 7 71/2 8 81/2 9 over 9
1. How often do you
speak English in your
home?
2. How often do you
speak English outside
your home?
3. How often do you
speak English to:
Every
day
1
1
1
2 or 3
times a week
2
2
2
Once
a week
3
3
3
Once or twice
a month
4
4
4
Never
5
5
5
6. Are newspapers, magazines, or books available in your native language? YES NO
If yes, circle the average number of hours you spend reading them each day.
9. What kinds of English TV programs do you watch?_
10. What kinds of things do you write in English?
11. What do you think has helped you the most to learn English?,
D. Language and culture
1. Please reply to the following questions by answering YES! if you strongly agree, yes
if you agree, no if you do not agree, and NO! if you strongly disagree.
a) Would you like more chances
to speak English?
b) Would you like more chances
to write English?
c) Do men have more chances to
speak English than women?
d) Do children have more chances
to speak English than adults?
e) Would you like more English
speaking friends?
YES!
1
1
1
1
1
1
yes
2
2
2
2
2
2
no
3
3
3
3
3
3
NO!
4
4
4
4
4
4
0 1/2
5 51/2
1
6
11/2
61/2
2
7
21/2
71/2
3
8
31/2
81/2
4
9
41/2
over 9
0
5
1/2
51/2
1
6
11/2
61/2
2
7
21/2
71/2
3
8
31/2
81/2
4
9
41/2
over 9
7. Circle the average number of hours each day you read newspapers, books, or
magazines in English.
8. What kinds of radio programs do you listen to?_
f) Are Americans helpful when
you speak English?
g) Do you have to speak English to
do well in the US
h) Do you have to speak English to
do well at Pima?
i) Would you do better at Pima
if you spoke English better?
j) Does your daughter speak better
English than you?
k) Does your husband/wife
speak better English than you?
1) Will you lose contact with
your children if you do not
learn English?
m) Are you happy you came
to the US?
YES!
1
1
1
1
1
1
1
yes
2
2
2
2
2
2
2
no
3
3
3
3
3
3
3
NO!
4
4
4
4
4
4
4
2. Please circle your answers to these questions.
a) Have you ever had a good
experience at school because you
are an international student who
speaks English as a second language?
b)Have you ever had a bad
experience at school because you
are an international student who
speaks English as a second language?
c)Have you ever had a good
experience in the community
because you are Vietnamese and
speak English as a second language?
d)Have you ever had a bad
experience in the community
because you are Vietnamese and
speak English as a second language?
Many
times
A few
times
One
time
Never
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
3. Do you think people treat you differently because you are Vietnamese and not a
native speaker of English? Please explain:
4. Do you feel you need to adjust to the American way of life in order to be successful
in the US?
Please explain:
5. Have you changed your lifestyle/appearance since you came to the US? Why?
Please explain:
THANK YOU VERY MUCH FOR YOUR HELP!
Appendix D
Sample follow-up interview
questions
1. Since our last meeting, how often have you spoken English in
your home?
2. How often have you spoken English outside your home?
3. With whom have you spoken English? For how long?
4. Do you speak English or Vietnamese more often at school? At
home?
5. With whom do you usually speak Vietnamese?
6 How often have you watched TV in English?
7. In Vietnamese?
8. What is helping you the most to learn English?
9. Do you have many chances to speak English? Why or why not?
10. Would you like more chances to speak English? Why or why
not?
11. How can you get more chances to speak English?
12. Do you think your English has improved since...
13. If yes, how has it improved?
14. What has helped the most in improving it?
SAMPLE FOLLOW-UP INTERVIEW QUESTIONS 191
15. If not, why don't you think your English has improved?
16. Are you adjusting to the American way of life?
17. If yes, in what way?
18. If no, why not?
19. Do you think it is necessary to adjust the American way of life
in order to learn English or be successful in the US? Why or
why not?
20. What are the biggest appearance/lifestyle changes you have
made since moving to the US? Why did you make these
changes? Have they helped you make American friends or
improve your English?
21. What are your strengths in English?
22. What are your weaknesses in English?
23. Is it easy/difficult to speak/read/write/listen to English? Why or
why not?
24. How do you study English and what English do you study?
25. What friends, American or Vietnamese, do you talk with most
with now?
26. When you watch TV, how much do you understand with
captions? Without?
27. In meetings, work talk, customers, etc., how much do you
understand?
28. When you talk with me, how much do you understand?
29. Do you still watch TV with captions? Why or why not? How
much do you understand with/without captions?
30. What is/has helped you the most in learning English?
31. Who are your closest friends now? Are they American or
Vietnamese or other?
32. How often do you rent videos? How much of them do you
understand? What kinds of videos are they?
33. Do you get the newspaper? A US one? When/how often? How
often do you read it? How much do you understand?
34. Vietnamese newspaper?
35. You said that you wanted to go to the theater together every
week... are you? What do you go to watch? How much do you
understand?
36. Do you have an American or Vietnamese life? Why?
37. Do you still go to church every Sunday? In English? How much
of the sermon do you understand? Are the other people going
to church American or Vietnamese? Is the priest American or
Vietnamese (i. e., is it a Vietnamese church?)
38. What are the biggest lifestyle changes you have made since
coming to the US?
39. What do you miss the most/least about your life in Vietnam?
40. What do you like most/least about your life in the US?
41. What are your future plans? Goals?
42. What do you like the most about Americans?
43. Do you feel you are American or Vietnamese? When? Why?
Appendix E
Language Log
Minutes
per day:
Speaking
English
at home
Speaking
Vietnam,
at home
Speaking
English
with friends
Speaking
Viet,
with friends
Speaking
English
at school
Speaking
Viet,
at school
Speaking
English
at work
Speaking
Viet,
at work
Speaking
English
in shops
Watching
TV in
English
English
Reading
Vietnam.
Reading
Studying
English
Sunday Monday Tuesday Wed. Thursday Friday Saturday
Index
Abrahamsson, N. 20, 114, 168
absence, definition of 18, 44
accent (degree of) 26, 143
acculturation 25, 142, 158-9
Acculturation Model 142
acquisition, definition of 18, 44
acquisition orders 65-7, 77-81, 90-4,
95-114
see also developmental patterns
Adamson, H. D. 2, 27, 31, 33, 44,
168
age of L2 acquisition 33
age of arrival 28
Altenberg, E. P. 7, 9-10, 122, 168,
176
amount of L1/L2 use 2, 28, 143
see also extent of L1/L2 use
Anderson, J. 16, 17, 19, 20, 108,
124, 168
assimilation
consonant 10, 12-13, 116, 121
cultural 143-6
attitude 29-32, 132, 159
Baca, L. 30, 174
basic interpersonal communication
skills 150
Bayley, R. 20, 28-9, 31, 33, 44, 45,
168, 176
Beebe, L. M. 2, 24, 31, 33, 34, 168
Benson, B 8, 9, 10, 16, 19-20, 21,
22, 33, 95, 100, 107-8, 115, 122,
168
Berkowitz, D. 2, 23^, 169
Bhatt, R. M. 8, 13, 14, 15, 19, 105,
122, 171
Blackledge, A. 25, 168
Brinton, D. M. 120, 169
Broselow, E. 8, 9, 10, 16, 18-19,
108, 122, 124, 169
Byrne, J. L. 29, 170-1
Capability Continuum Paradox 33
caretaker identity 144, 151, 159
Carlisle, R. S. 16, 17, 18-19, 20, 44,
108, 111, 112, 113, 124, 169
Celce-Murcia, M. 120, 169
child LI phonological development 2,
10, 105-7
see also developmental effects
Clahsen, H. 31, 142, 143, 173
codas, production of 71-90, 90-4
by length and participant 71-7
individual codas 77-81, 81-90
and onset 90-4
collaborative interactions 148
community, effect of L1/L2 29, 149-
52
Corbin, J. 46, 48, 175
Gumming, A. 2, 25, 31, 169
data
analysis
collection
phonological
social
Dauer, R. M.
Davidian, R. D.
41-8
41-8
41-6
46-8
42-3, 169
8, 9, 13, 14, 15,
105, 122, 123, 170
Day, R. 47, 169
descriptive statistics 44, 59-61, 71-3
developmental effects 10-15, 105-7,
123-5, 153-7, 164-5
see also child LI phonological
development
patterns 94-114, 120-5, 153-6
see also acquisition orders
processes see developmental effects
sequence see developmental patterns
Dickerson, L. J. 32, 33, 34, 118, 119,
169
Dickerson, W. 33, 34, 118, 119, 169
Dowd, J. 2, 23-4, 169
Cancino, H. 44, 169
Eckman, F. R. 16, 17, 18, 19, 44,
108, 111, 112, 124, 169, 170
Edge, B. A. 20, 24, 31, 33, 116, 170
Ehrlich, S. 25, 170
Ellis, R. 13, 31, 142, 163, 170
English phonology 35-6
epenthesis, definition of 43
extent of L1/L2 use 2, 28-9, 143,
146-52
see also amount of L1/L2 use, social
interactions
extroversion 150
factor groups 45-6
Faudree, M. C. 8, 10, 44, 122, 173
feature change, definition of 43
Ferguson, C. 11, 17, 171, 173
Finer, D. 16, 18-19, 108, 124, 169
Flege, J. E. 2, 8, 9, 13, 14, 15, 26,
29, 105, 122, 123, 143, 170, 171,
174
Frieda, E. M. 2, 26, 29, 143, 170
Gal, S. 25, 170
Gardner, R. 30, 31, 32, 170
Gatbonton, E. 25, 33, 118, 119, 170
gender 2, 25-8, 31, 33
Giles, H. 29, 170-1
Gill, J. 2, 25, 31, 169
Goldstein, T. 2, 25, 29, 31, 149, 171
Goldvarb 45
Goodwin, J. M. 120, 169
grammatical conditioning 20, 22, 32,
89-90, 117-18, 121, 122, 124
Greenberg, J. H. 16, 17, 109, 171
Grounded Theory 46, 48
Guion, S. G. 2, 29, 143, 171
Guy, G. R. 45, 171
Hammerberg, B. 10, 171
Hammond, M. 35, 171
Hancin-Bhatt, B. 8, 13, 14, 15, 19,
105, 122, 171
Hansen, J. G. 8, 16, 17-18, 19, 32-3,
105, 108, 114, 171
Hardison, D. M. 98, 171
Hatch, E. 148, 171
Hecht, B. F. 13, 14, 15, 24, 31, 105,
122, 171, 173
Herbert, R. K. 19, 113, 171
Heidelberger Forschungsprojekt 'Pidgin
Deutsch' 31, 142, 143, 171
Hieke, A. K. 12, 13, 24, 172
Hodne, B. 8, 9-10, 122, 172
Hong Kong English 24
Hosseine-Fatemi, A. 29, 31, 143, 173
inferential statistics see VARBRUL
Ingram, D. 11-12, 172
input probability, definition of 44-5
instrumental motivation 30, 31
integration 25
integrative motivation 30, 31, 33
introversion 150, 159
investment 29-32, 142, 150-1, 157,
159-60
IPA transcription 39, 58
Iverson, G. K. 18, 19, 44, 170
Jakobson, R. 10-11, 106, 123, 172
James, A. R. 7, 10, 14-16, 23, 172
Jensen, J. T. 35, 36, 172
Jones, D. 42, 172
judgement sampling 40
Kahn, D.
Kouritzen, S.
99, 172
25, 31, 172
L1/L2 identity see social identity
LI transfer s ee transfer
Labov, W. 33, 172
Labov's Observer's Paradox 33
Lambert, W. E. 30, 31, 32, 170, 172
language use environment/
opportunities 126-40, 146-52,
157-63
Lanoue, G. 30, 172
Leather, J. 7, 14, 23, 172
length of stay 28
Leopold, W. F. 2, 172
LePage, R. B. 23, 172
linguistic
environment 19-20, 22, 90-1, 99-
100, 102-3, 114-17, 121, 122,
156-7
variation 32-3
see also linguistic environment
Loftin, J. D. 2, 29, 143, 171
loglinear analysis 44, 162-3
Losey, K. M. 25, 172
Loewenthal, K. 28, 143, 176
Lukmani, Y. 30, 32, 172
Lum, J. 23, 140, 172-3
Lybeck, K. 25, 31, 142, 173
MacKay, I. R. A. 2, 26, 29, 33, 143,
170, 174
Macken, M. A. 2, 11, 173
Major, R. C. 8, 9, 10, 13-15, 17,
iNDEX . 195
33, 34, 39, 44, 100, 105, 113, 116,
118, 119, 122, 123, 173
markedness 2, 15-19, 107-14, 123-5,
153-7, 165
defined by length 16-17, 108-9, 154
denned by type 17-18, 109-11, 154
defined by length and type 18, 111-12,
154
defined by sonority 18-19, 113-14,
154
see also sonority
Meisel, J. 31, 142, 143, 173
Milroy, L. 40, 173
Monshi-Tousi, M. 29, 31, 143, 173
Moravcsik, E. 17, 171
motivation 29-32, 33, 150, 159
instrumental 30, 31
integrative 30, 31, 33
Mulford, R. 13, 14, 15, 24, 31, 105,
123, 171, 173
Munro, M. J. 33, 170
negotiation of meaning 142, 149,
157-8, 159, 160, 163
Nguyen, D. L. 37, 39, 95, 98, 116,
174
Nguyen family, description of 49-54
Norton, B. 2, 174
see also Peirce, B. N.
Nozawa, T. 2, 26, 29, 143, 170
Observer's Paradox 33
Odlin, T. 7, 8, 9, 174
Ohala, D. 107, 174
Ohara, Y. 27, 31, 174
Oiler, J. 29, 30, 31, 143, 173, 174
Olmsted, D. L. 107, 174
onsets, production of 60-71, 90-4
by length and participant 60-71
individual onsets 65-71
and codas 90-4
Ontogeny Model 14-15, 39
Osburne, A. G. 9, 16, 20, 21-2, 23,
33, 45, 95, 99, 100, 103, 107-8,
114-15, 174
Owens, R. E. jr. 106, 174
Oyama, S. 28, 31, 143, 174
participant selection 40-1
Peck, S. 148, 174
Peirce, B. N. 29, 30-1, 32, 47, 141,
142, 149, 150, 159, 174
see also Norton, B.
Pennington, M. C. 23, 26, 31, 174
personality traits 150, 159
phonological environment see linguistic
environment
phonological marker 23-4
Pienemann, M. 31, 142, 143, 173
Piper, T. 12-13, 105, 123, 174
Piske, T. 2, 26, 29, 143, 174
Preston, D. 32, 174
Rand, D.
Regan, V.
Rosansky, E.
Rutledge, P. J.
45, 174-5
2, 27, 31, 33, 44, 168
44, 169
48, 49, 174
St. Martin, G. M. 31, 175
Sankoff, D. 45, 174-5
Santry, P. A. 37-9, 100, 115, 175
Sato, C. J. 8, 10, 16, 17, 19, 21, 23,
33, 34, 95, 107-8, 118, 122, 124,
175
Saunders, N. 20, 175
scaffolding 148
Schmidt, R. W. 2, 33, 34, 175
Schumann, J. 44, 142, 169, 175
selection of participants 40-1
setting of the study 40-1
Siegal, M. 26-7, 31, 175
Skaer, P. M. 8, 175
social
barriers 23-32, 140-52, 157-63,
164
constraints 23-32, 140-52, 157-
63, 164
contexts of language use 126-40,
144
distance 142
identity 4, 23-5, 26-7, 29, 31, 32,
140-2, 158
interactions 28-9, 31, 32, 126-40,
142-4, 146-52, 157-63
marking 23-4
networks 4, 25, 28-9, 31, 146
see also social interactions
practice 25
variation 323
sonority 18-19, 22, 107, 113-14
see also markedness
Spada, N. 31, 143, 175
Spencer, A. 16, 35-6, 175
stages in acquisition 153-6
Strauss, A. 46, 48, 175
Strong, M. 30, 150, 159, 175-6
style variation see task variation
Tabouret-Keller, A.
Tahta, S.
23, 172
28, 143, 176
INDEX 197
Tajfel, H. 140, 176
Tarone, E. 8, 10, 20, 33, 116, 122,
176
Tarone's Capability Continuum
Paradox 33
task variation 8, 33-5, 92-3, 118-20,
156
Teutsch-Dwyer, M. 25, 176
transfer 2, 7-10, 13-15, 19, 21-2,
95-105, 108, 114-15, 120-5, 153-7,
164
Tropf, H. 19, 113, 124, 176
Universal Canonical Syllable Structure
19, 20, 113-14, 124, 154
universal syllable structure preference
16, 21, 107-8, 123
U-shaped curve 13, 118-19
Vago, R. M. 8, 9, 10, 122, 168, 176
VARBRUL
analysis of data 61-5, 66-8, 73-7,
81, 84-7, 89, 116
description of 28, 44-6
variation 8, 32-5, 118-120, 156
linguistic 32-3
see also linguistic environment
social 32-3
task 8, 33-5, 118-20, 156
Vietnamese
people in America 48-9
phonology 2, 4, 37-9, 45, 95,
114-15
Vigil, F. 30, 174
Wagner-Gough, J. 148, 176
Weinberger, S. H. 8, 16-17, 19, 20,
108, 122, 124, 176
Wood, M. 28, 143, 176
Yavas, M.
Young, R.
Zisa Project
Zuengler, J.
33, 176
22, 44, 176
142, 143
23-4, 33, 168, 169

Acquire A Non-Native Phonology

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Acquire A Non-Native Phonology

Enviado por

Direitos autorais:

Formatos disponíveis

Acquiring a Non-native Phonology

This page intentionally left blank

Você também pode gostar