Acquiring a Non-native Phonology
This page intentionally left blank
Acquiring a Non-native Phonology
Linguistic Co...
37 downloads
1081 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
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 11 York Road London SE1 7NX
80 Maiden Lane Suite 704 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
1 7
3 Researching the Nguyen Family
40
4 Acquiring a Non-native Syllable Structure
59
5 Linguistic and Task Constraints
95
6 Social Barriers
126
7 A Unified Approach
153
8 Conclusions
164
References
168
Appendices
177
Indexes
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 phonological 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 comprehensive database should be of interest to anyone concerned with the acquisition of an L2 phonology. Additionally, it should also be of
4 • ACQUIRING A NON-NATIVE PHONOLOGY
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 examined. 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 interactively 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 linguistic 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,
6 • ACQUIRING A NON-NATIVE PHONOLOGY
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.
2 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-
8 • ACQUIRING A NON-NATIVE PHONOLOGY
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 consonant 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
10 • ACQUIRING A NON-NATIVE PHONOLOGY
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
INVESTIGATING L2 PHONOLOGY • 11
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 corresponding 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, 4758). 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 processes 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
12 • ACQUIRING A NON-NATIVE PHONOLOGY
the deletion of final consonants, especially in consonant-vowelconsonant (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 modifications 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-
INVESTIGATING L2 PHONOLOGY • 13
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
14 • ACQUIRING A NON-NATIVE PHONOLOGY
1987c; Mulford and Hecht 1980). In his work on Japanese native speakers' production of English consonant clusters and final consonants. 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 developmental effect, and then strongly aspirated them, which is characteristic of stops in Icelandic. Hancin-Bhatt and Bhatt (1997), in their research on the production of English onsets and codas in monosyllabic 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 mechanisms (i.e. "developmental processes") associated with the acquisition 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 developmental 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
INVESTIGATING L2 PHONOLOGY • 15
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 developmental constraints: as the formality of the style increases, there is a decrease in errors due to transfer, whereas errors due to developmental 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; HancinBhatt 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
16 • ACQUIRING A NON-NATIVE PHONOLOGY
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 markedness 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 modifications increased as the length of the codas increased. This indicates
INVESTIGATING L2 PHONOLOGY • 17
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
18 • ACQUIRING A NON-NATIVE PHONOLOGY
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, stopfricative-fricative, nasal-fricative-fricative, nasal-stop-stop and stopstop-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.
INVESTIGATING L2 PHONOLOGY • 19
Their findings on Japanese and Korean learners of English's wordinitial 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
20 • ACQUIRING A NON-NATIVE PHONOLOGY
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 production 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.
INVESTIGATING L2 PHONOLOGY • 21
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 longitudinal study of two Vietnamese adolescents, for example, examined whether closed syllables would be preferred in the English interlanguage 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
22 • ACQUIRING A NON-NATIVE PHONOLOGY
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 production, 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;
INVESTIGATING L2 PHONOLOGY • 23
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
24 • ACQUIRING A NON-NATIVE PHONOLOGY
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 nonstandard 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
INVESTIGATING L2 PHONOLOGY • 25
to signal their identity. Gatbonton (1975) for example, found that the production of interdental fricatives by French-Canadian learners of English varied based on the learners' self-identification as either nationalistic and pro-French or non-nationalistic and proEnglish, 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
26 • ACQUIRING A NON-NATIVE PHONOLOGY
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.
INVESTIGATING L2 PHONOLOGY • 27
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 community speech norms by Vietnamese and Cambodian immigrants to the US. The researchers examined the participants' use of the phonological 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 /irj/, which is present), would indicate a greater integrating of participants into their respective speech communities. The participants 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.
28 • ACQUIRING A NON-NATIVE PHONOLOGY
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 nonaccentedness 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
INVESTIGATING L2 PHONOLOGY • 29
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
30 • ACQUIRING A NON-NATIVE PHONOLOGY
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 motivation. 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
INVESTIGATING L2 PHONOLOGY • 31
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
32 • ACQUIRING A NON-NATIVE PHONOLOGY
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 determining 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 operationally), 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
INVESTIGATING L2 PHONOLOGY
• 33
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 environment 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
34 • ACQUIRING A NON-NATIVE PHONOLOGY
Japanese learners of English were twice as accurate in their production 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);
INVESTIGATING L2 PHONOLOGY • 35
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 wordmedially 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 Bilabial Labio- Dental Alveolar Palato- Palatal Velar Glottal Place of alveolar dental Articulation Manner of Articulation Stops voiceless voiced Fricatives voiceless voiced
k g
t d
P b f V
6 5
s z
Affricates voiceless voiced
h
I 3 tj d3
1
Lateral Nasals
m
n
Approximants
w
i
From Jensen (1993)
q y
36
• ACQUIRING A NON-NATIVE PHONOLOGY
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/*, III/ *, /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 threeconsonant 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/, Ips/, /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/, /Id3/, /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 threeconsonant 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 III, /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 threeconsonant 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.
INVESTIGATING L2 PHONOLOGY • 37 Table 1.2 An overview of Vietnamese consonants Place of Bilabial Labio- Dental Retroflex Alveolar Alveo- Velar Glottal Articulation dental Palatal Manner of Articulation Plosive plain aspirated voiced
t ih d
b
Fricatives voiceless voiced
f
§
c
;
h
Y
1 m
n
q
Jl r
Trill Semivowels
X
V
Lateral Nasals
k
t
w
j
From Santry (1997) and Nguyen (1970)
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, /th/ 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 [kp], which means it has a simultaneous bilabial closure; in final position after lot it is produced as [kw], which means it is produced strongly labialized, i.e. with rounding of the lips.
38 • ACQUIRING A NON-NATIVE PHONOLOGY
In final position after hi, it is weakly labialized, and produced as [kw]. After all other vowels, it is unreleased and produced as [k1]. 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 [mn] 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 3 with O representing the Onset, C as Consonant, W as semi-vowel, N as
INVESTIGATING L2 PHONOLOGY • 39
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 semivowel, 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, lowrising 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) articulation of a sound/word. This practice will be employed throughout this book.
3
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
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 participants' 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
42 • ACQUIRING A NON-NATIVE PHONOLOGY
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 participant 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 twosyllable 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
RESEARCHING THE NGUYEN FAMILY • 43
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 corresponded 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) absence1 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
44 • ACQUIRING A NON-NATIVE PHONOLOGY
(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
RESEARCHING THE NGUYEN FAMILY • 45
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 environment 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/).
46 • ACQUIRING A NON-NATIVE PHONOLOGY
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, analysis, 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-
RESEARCHING THE NGUYEN FAMILY • 47
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 throughout 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' questionnaire 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 participants. 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)
48 • ACQUIRING A NON-NATIVE PHONOLOGY
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 information 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, approximately 455,000 south-eastern Asian refugees settled in the United
RESEARCHING THE NGUYEN FAMILY • 49
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
50 • ACQUIRING A NON-NATIVE PHONOLOGY
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
RESEARCHING THE NGUYEN FAMILY • 51
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 permission 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
52 • ACQUIRING A NON-NATIVE PHONOLOGY
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
RESEARCHING THE NGUYEN FAMILY • 53
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
54 • ACQUIRING A NON-NATIVE PHONOLOGY
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 pronunciation, 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 u m m . . . I was n o . . . eh... little customer... Nhi: Few... f e w . . . Anh: Few customer... they a r e . . . u m . . . spring break... and because I . . . I go home early and then and visit visit ehh... a friend ... together in class... yeah... Jette: A classmate?
RESEARCHING THE NGUYEN FAMILY • 55
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: C u t . . . [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 e h . . . 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
56 • ACQUIRING A NON-NATIVE PHONOLOGY
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 m e . . . tell u s . . . tell m e . . . any idea f r o m . . . from film yeah... yeah... a n d . . . I watch... and Saturday... Saturday yesterday yesterday I watched... videotape titled Virus II Jette: And what will you do today? Anh: M m m . . . 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 communicating their ideas, and were able to speak without as many hesitations and pauses. Overall, Nhi had greater grammatical accuracy 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 communicated more in single words (nouns, main verbs) than in complete sentences, unlike Nhi, who was able to communicate in longer phrases and clauses.
RESEARCHING THE NGUYEN FAMILY • 57
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 m e . . . American life is very stress very stress... and very hard very hard... and and... we a r e . . . 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 o b . . . b u t . . . h e . . . my husband has a j o b . . . 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 like... I like the live here Jette: Why? Nhi: Because... in Vietnam... I always... I always confuse... confuse about communism politics politics yeah always about politics yeah... I I teach... I I taught... always controlled always was controlled... a n d . . . don't have... free time t o . . . relax... and in here... when I have j o b . . . I take money... small amount of money I feel... I feel... my life in here... enough... don't worry... and... everybody around in Tucson... always friendly yeah... and always help m e . . . 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.
58 • ACQUIRING A NON-NATIVE PHONOLOGY
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.
4 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 participant, as is illustrated in Table 4. 1 below. As indicated in Table 4. 1, production accuracy was consistent across time, length and participant: 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 o f 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;
60 • ACQUIRING A NON-NATIVE PHONOLOGY Table 4. 1 Onset production by length, participant and time Time: Participant:
Time 1
Time 2 Nhi Both
Time 3 Nhi Both
Ann
Nhi
Both
790 91% 11 1% 0 0% 68 8% 869
720 90% 10 1% 0 0% 73 9% 803
1510 90% 21 1% 0 0% 141 8% 1672
580 94% 3 0% 0 0% 35 6% 618
1417 93% 3 0% 0 0% 99 7% 1519
1407 968 2375 89% 89% 89% 0 0 6 0% 0% 0% 0 0 0 0% 0% 0% 283 165 118 11% 11% 11% 1578 1086 2664
Feature Change Total
70 69% 28 27% 0 0% 4 4% 102
58 60% 30 31% 8 8% 1 1% 97
CC = 785 44 58 128 64% 59% 56% 20 58 16 29% 16% 26% 0 3 8 0% 4% 3% 14 21 5 21% 18% 3% 98 78 199
102 58% 36 20% 3 2% 35 20% 176
162 63% 74 29% 7 3% 15 6% 258
115 76% 19 13% 4 3% 14 9% 152
277 68% 93 23% 11 3% 29 7% 410
Targetlike Absent
0 0% 11
0 0% 5
100% 0 0% 0 0%
100% 0 0% 0 0% 5
CCC = 33 3 60% 2 100% 40% 0 0 0% 0% 0 0 0% 0% 5 16
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
Targetlike Absent Epenthesis Feature Change Total
Targetlike Absent Epenthesis
Epenthesis Feature Change Total
11
Anh
C = 5855 837 93% 0 0% 0 0% 64 7% 901
0 0% 16
0 0% 0 0% 0 0% 0 0% 0
Anh
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:
Targetlike Absent Epenthesis Feature Change Total
Anh
860 88% 50 5% 0 0% 72 7% 982
Time 1 Nhi Both
778 86% 45 5% 8 0% 74 8% 905
Anh
Time 2 Nhi Both
ALL = 6673 1638 898 624 87% 89% 90% 18 95 23 2% 5% 3% 3 0 8 0% 0% 0% 146 85 49 8% 7% 8% 1887 1004 696
1522 90% 41 2% 3 0% 134 8% 1700
Anh
Time 3 Nhi Both
1577 86% 81 4% 7 0% 180 10% 1845
1085 2662 87% 86% 20 101 3% 2% 4 11 0% 0% 132 312 11% 10% 1241 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 production type to find the best goodness-of-fit; while the factor groups found not to be significant were eliminated from further statistical
62 • ACQUIRING A NON-NATIVE PHONOLOGY Table 4. 2 VARBRUL analysis of syllable onsets Target-like Production
Absence
Epenthesis
Feature Change
Total
n = 5822
n = 237 4%
n = 22 0%
n = 592 9%
N = 6673
1. Preceding linguistic environment Pause . 47tns . 48tns** P** 66 1608 # 28% 28% %
ko 0 0%
. 53fns 203 28%
1877 28%
Vowel P{ # %
. 52ns 2932 50%
. 47tns 103 50%
. 49 14 50%
. 46ns 261 50%
3310 50%
Consonant P* # %
. 48|ns 1282 22%
. 58ns 68 22%
. 53 8 28%
. 53|ns 128 22%
1486 22%
ko 22 100%
. 55 558 85%
5690 85%
Factor Group
87%
2. Following linguistic environment Vowel . 53tns . 46 $ 214 4896 # 85% 85% % Diphthong P* # %
. 68 514 8%
. 53tns 19 8%
ko 0 0%
. 24 18 8%
551 8%
%
. 76 412 6%
. 36ns 4 6%
ko 0 0%
. 27 16 6%
432 6%
3. Participant Anh P* # %
. 50ns 3335 57%
. 53ns 149 57%
. 56 12 55%
. 50ns 337 57%
3833 57%
Nhi P* # %
. 51ns 2487 43%
. 46ns 88 43%
. 45 10 45%
. 50ns 255 43%
2840 43%
. 48fns 5302 88%
. 37 30 12%
ko 0 0%
. 51ns 523 88%
5855 88%
/D/
$ #
4. Length C
Pf
# %
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 63 continued
Target-like Production
Absence
Epenthesis
Feature Change
Total
n = 5822 87%
n = 237 4%
n = 22 0%
n = 592 9%
N = 6673
. 55ns
. 98| 187 88%
ko 22
. 47ns
507 12%
. 48-fns
. 98f
13 0%
20 0%
f
. 47
. 44ns
#
4117
%
72%
168 72%
Factor Group
CC / # %
100%
69 12%
785 12%
ko 0 0%
ko 0 0%
33 0%
. 46 13 59%
. 55 482 72%
4780
. 62 9 41%
. 37 110 28%
1893
ccc
y#
% 5. Stress
Stressed
72%
Unstressed
P' #
. 59
. 64ns
1705
%
28%
69 28%
28%
6. Time TzVwe / />'' # %
. 48tns 1638
. 64 95 28%
. 56f
. 46|ns
8 36%
146 28%
. 43 41 25%
. 34 3 14%
. 46fns
. 56f
. 54ns
46%
. 45 101 46%
11 50%
312 46%
. 006 8. 3921 1. 32986
. 008 7. 4049 1. 2241
. 004 28. 6882 1. 1953
. 082 5. 8965 . 9828
28%
1887 28%
TzVwe 2
#
. 55ns 1522
%
25%
1*
134 25%
1700 25%
Time 3 P' # %
Input Weight Total X2 X2 Cell df*** Significance
. 48tns 2662
3
3
5
3
p <. 05
p <. 06
p <. 001
p <. 25
3086 46%
*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, f These factors were collapsed and run as one factor due to similar input probabilities in earlier runs.
64 • ACQUIRING A NON-NATIVE PHONOLOGY
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,
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 65
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
66 • ACQUIRING A NON-NATIVE PHONOLOGY Table 4. 3 Accuracy Percentiles by Onset, Time, and Participant Time: Participant:
100%
Anh
Time 1 Nhi
d, k, g, m, n, h, f, tj, w, j, y, tw, fy, si, sk, sm, vi
Anh
b, k, m, y,7 1,7 hi, J kw, pj, gi, sm, py, st "5
Time 2 Nhi
t, d, b, k, g, m, n, h, v, f, z, i, sm, sk, ti, ki, fl, hy, by, fy, sw, kw, tw,
Anh
Time 3 Nhi
t, b, k, g, m, f, v, d3, w, i, sp, kw, di, vi
b, g, m, f, v, z, w, y, ny, kw, fy, dy, si, di, sk, skw
t, d, g, n, m, h, v,
1, y, w, gj, st, ti, tw, si, sk, skw
VJ
75 - 99%
p, b, t, s, z, v, f, 1, st, tr, sp
t, d, g, n, f, v, s, J, tj, h, w, j, vy, sp
p, s, tf, w, y, 1, st, sp
d, n, s, h, 1, y, st
t, d, p, k, n, s, tf, h, i, 1, st, sp, tw, ti, sti
b, k, s, f, tf, J, fy, sp, pi, vi
50 - 74%
gJ
bl
di, pi, sti
p, z, kl, sm
f, sw, gj, bi, pi, vi
p, fc
25 - 49%
di, fi
z, dj, di, tw, fi, kl
f, vy, fi
5, 61, pi
1 - 24%
ds, 9, kl
p, ti, pi
ds, pi
f
0%
a, bi, 0i, e, a, sw, a, e, ei pi, by, dsy, ki, pi, bi, spl, sti
61, spl, sti
f, z, ds, kl 6, dj, pi, fi
8, di
tf, pi, hy, a, ki, 0i, a, ea, pi, ki, spi
sm, 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
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE
• 67
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 (x2 = 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
68 • ACQUIRING A NON-NATIVE PHONOLOGY Table 4. 4 VARBRUL analysis of target-like production by singleton onset Accurate Number Total Number Percentage Probability
/P/ M
/m/ /n/
119 382 429 351 417 100 216 213 323 24 77 7 4 98 17 386 386 313 167 260 495 518
Vowel Diphthong /o/
Following linguistic environment 4074 4606 335 348 398 406
N W N /g/ Iff M 1*1 /z/
III
/e/ /s/ AJ7
w N M M N /V
158 384 440 356 420 102 219 218 338 32 141 104 200 120 65 396 391 316 169 268 495 523
75 99 98 99 99 98 99 98 96 75 55 7 2 82 26 97 99 99 99 97 100 99
. 09 . 87 . 57 . 69 . 79 . 61 . 67 . 58 . 41 . 09 . 04 . 002 . 001 . 13 . 01 . 57 . 69 . 77 . 74 . 51 1. 00 . 78
88. 96. 98.
49 38 71
Input Weight . 97 Total ft2 60. 000 X2 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
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 69
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
70 • ACQUIRING A NON-NATIVE PHONOLOGY
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 targetlike. 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
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE
• 71
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 111. If it was /!/, the cluster was usually target-like, but with 111, 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
72 • ACQUIRING A NON-NATIVE PHONOLOGY Table 4. 5 Production of codas by length, time and participant Time: Participant: Targetlike Absent Epenthesis Feature Change 2+ Types Total
Time 1
Anh
Nhi
Both
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%
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
0 0% 22
0 0% 19 95% 0 0% 0 0%
0 0% 41 98% 0 0% 0 0%
1
1
5% 20
2% 42
Anh
Time 2 Nhi Both
C = 3703 232 219 40% 48% 194 121 34% 26% 45 49 8% 10% 92 61 16% 13% 14 12 2% 3% 1151 579 460
451 43% 315 30% 94 9% 153 15% 16 2%
Anh
Time 3 Nhi Both 356 47% 125 17% 98 13% 131 17% 43 6% 753
725 48% 285 19% 163 11% 292 19% 48 3%
1039
369 49% 160 21% 65 9% 161 21% 5 0% 760
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
0 0% 35 67% 0 0% 1 2% 16 31% 52
0 0% 5 83% 0 0% 0 0%
0 0% 29 97% 1 3% 0 0% 0 0% 30
0 0% 34 94% 1 3%
1513
CC = 607 Targetlike Absent Epenthesis Feature Change 2+ Types Total
3 4% 54 79% 3 4% 2 3% 18 25% 73
5 7% 47 64% 0 0% 0 0% 9 13% 68
CCC = 130 Targetlike Absent
100% Epenthesis Feature Change 2+ Types Total
0 0% 0 0% 0 0% 22
0 0% 16 76% 0 0% 0 0% 5 24% 21
0 0% 19 61% 0 0% 1 3% 11 35% 31
1 17% 6
0% 1 3% 36
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE
• 73
continued
Time: Participant:
Targetlike Absent Epenthesis Feature Change 2+ Types
Total
Anh 287 38% 284 37% 60 8% 104 14% 30 4% 765
Time 1 Nhi Both 287 43% 225 34% 34 5% 82 12% 43 6% 671
Anh
Time 2 Nhi Both
ALL = 4440 574 235 235 40% 35% 40% 509 257 194 35% 38% 35% 94 52 45 7% 8% 8% 186 94 62 13% 14% 11% 73 35 35 5% 5% 5% 1436 673 559
459 37% 451 37% 97 8% 156 13% 69 6%
1232
Anh 372 42% 269 30% 67 8% 163 18% 16 2% 877
Time 3 Nhi Both 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 environment (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 production 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 FactorTarget-like Absence Epenthesis Feature Group Production Change
n = 1772 n = 1442 41% 32%
2 Types of Total Production
n = 359 8%
n = 642 14%
n = 225 5%
0 0%
16 7%
N= 4440
1. Preceding linguistic environment Diphthong P2* #
%
. 49ns
. 12 23 1%
. 94 247 17%
. 20 8 2%
. 57
. 52 296 82%
. 50ns
. 51ns
95%
. 42 969 67%
593 92%
192 85%
. 24 73 4%
. 75 226 16%
. 58 55 15%
. 47ns
. 38ns
49 8%
17 8%
. 66 218 61%
. 54 278 43%
. 68ns
. 36 88 25%
. 43 248 39%
. 38ns
. 58 20 6%
. 55 33 5%
. 46ns
. 43 33 9%
. 49 83 13%
. 36ns 14 6%
567 13%
. 59 144 64%
2115
294 7%
Vowel
Pl # %
hi P* # %
1676
3726 84% 420 9%
2. Following linguistic environment Pause
. 41ns**
. 44ns
547 31%
560 39%
. 58ns
. 55ns
912 51%
601 42%
. 52ns
. 45ns
86 5%
71 5%
. 50ns
. 57ns
227 13%
210 15%
. 53 878 50%
. 45 632 44%
. 50ns
. 48ns
180 50%
281 44%
. 47 894 50%
. 55 810 56%
. 50ns
. 52ns
179 50%
361 56%
. 42 81 36%
2325
þi
. 59
# %
1733
. 59 348 97%
. 60 627 98%
. 42 105 47%
3703
98%
. 40 890 62%
. 09 39 2%
. 87 442 31%
. 16 10 3%
. 14 14 2%
. 84 102 45%
P* # %
158 70%
1761 40%
Consonant
þi # %
405 20%
1894 43%
Glide
P* # % Vowel
þi # %
8 4%
218 5%
3. Participant Nhi Pl # % Anh
þi # %
48%
52%
4. Length C
cc p*# %
83% 607 14%
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 75 continued
Factor Group
Target-like Absence Epenthesis Production
Feature Change
2 Types of Total Production
n = 1772 41%
n = 1442 32%
n = 359 8%
n = 642 14%
n = 225 5%
N= 4440
0 0%
. 95 110 7%
. 07 1 0%
. 06 1 0%
. 79 18 8%
130 3%
ccc
p*#
%
5. Grammatical conditioning Mono-morphemic P*
1727 97%
1298 90%
. 51ns 353 98%
. 51ns 631 98%
. 49ns 185 82%
4194 94%
. 55ns 45 3%
. 56ns 144 10%
. 37ns 6 2%
. 49ns 11 2%
. 63ns 40 18%
246 6%
y# %[7ws£res.sec?
. 51ns 1311 74%
. 50ns 1149 80%
. 62 341 95%
. 44 362 56%
. 59ns 196 87%
3359 76%
P* # %
. 48ns 461 26%
. 50ns 293 20%
. 18 18 5%
. 69 280 44%
. 42ns 29 13%
1081 24%
. 52ns 574 32%
. 54ns 509 35%
. 47f 94 26%
. 47tns 186 29%
. 49ns 73 32%
1436
. 47ns 459 26%
. 55ns 451 31%
. 47f 97 27%
. 47fns 156 24%
. 50ns 69 31%
1232 28%
. 51ns 739 42%
. 43ns 482 33%
. 55 168 47%
. 55ns 300 47%
. 51ns 83 37%
1772 40%
. 37 11. 2648 . 9387 4 p <. 01
. 30 12. 2906 . 7230 5 p <. 025
. 04 37. 7122 0. 5312 9 p <. 0001
. 11 24. 5871 1. 0690 6 p <. 0005
. 03 5. 8358 . 9726 3 p <. 025
#
.
5
0
% Bi-morphemic P* #
%
n
s
.
4
4
n
s
n
6. Stress Stressed
7. Time Time 1 t? #
% Time 2 Pf # % Time 3
Pf
#
% Input Weight Total ft2 X2 Cell dP** Significance
32%
*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.
76 • ACQUIRING A NON-NATIVE PHONOLOGY
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 targetlike. Three factor groups, preceding linguistic environment, participant 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 environment, 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 (p1. 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
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 77
(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
Table 4. 7 Accuracy Percentiles by Coda, Task, Time, and Participant Time Participant Anh
100% 75 - 99% 50 - 74% 25 - 49% 1 - 24%
0%
100%
75 - 99% 50 - 74%
Time 1
Time 2
Nhi
Anh
Time 3
Nhi
Interview Coda Data p, f, nz, jz m, q m, n, f, nz P t, m, n s, Iz n, m, q, s, Jz t, m, n, q, s, nz k, n, s t, d, v, j, 1 t, k, j, 1 k, d, j, 1, ts k, v, j, 1 b, d, z, J, ks, nt, jk d, J, v, z, ks J, ks v, J, z, 0, Jk b, nds, ntj, nd, nt, st, 0, ds, ntj, nd, st, qk, b, g, tj, 0, ntj, nd, nt, d, z, tj, 0, ntj, nd, st, qk, mz, Jt, It, zd, Id, ks, mz, it, zd, dz, jn, bz, st, mz, Jt, ts, nz, jk, JtJ, mz, ts, sk, jn, kt, nt, Iv, kt, sk, pt, Jd, ts, 5z, jst, vd, sk, J0, Iz, qk, jld, zd, Jt, nts, jkt, jnd, dz, jst, ndz, nts, its, kst, ndz, nts jst, ndz, nts, kst, jkt, nt0, ndz nte jlz, nt6, jnd
f, jz, jd
p, JZ
Iz, n0, ns, mt, ntj, jm6 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 t, k, m, n, rj p, d, f, s, 9, Jt, ps, j6, 16 k, s, z, v, tj, 3, j, jl, It, nt, bz, mz, dz, st, ts, ks, kts
f, jl
Anh
p, Id n, m t, k, q, f, s, tj, ts k, v d, v, ks, nz, jn, jz d, z, 1, j, jk, nt z, j, jd 0, ds, tj, ntj, nd, st, b, g, J, ds, 0, If, nd, nt, mz, nz, sk, j0, JtJ, jz, st, qk, jk, jt, Ip, pt, J0, ft, jn, jd, Id, Iv, Jt, ks, Iv, Iz, nds, jst, ndz, jst, ndz, nts, nt0 kst, Idz, jdz, jnd, nts,
P, J
t, n, q g, m, f, s, ts
nt0
Word List Coda Data mf, ns, Ip, ntj, nst 3, Ip, JtJ, Ik, ps, bz, mt, p, Ip, ps, ItJ, mf, ns, st, ns, qz, sk, jmz, nst, mt, mp, qd, ntj, n0, jst, qst nst, qst
p, t, k, m, n, qk p, t, k, n, m qk q, f, st, 10, 1m, J0, JtJ, k, q, z, st, jz, Iz, ks, nz, 1J, md, sp, Ip, ft, mf, Is, sk It, vz, mp, bz, mz, jkt,
t, n, m, f, qk k, q, 1, 0, js, J0, 10, JtJ, nd3, 1J, jb, bz, Is, If, jn,
Ipt
Iks 25 - 49% 1 - 24%
b, z, 1, st, If, sk, 1J, Ip, p, d, 1, jf, mp, vz, jm, t0, jf, bz, nz, zd, an, jz, 1m, Jt, JZ, kst, Jm0, jpt pt, qk, nst, jst b, v, 0, tj, id, mz, nt b, If, tjt, jnz
Nhi
s, 3, j, 1, If, ps, fs, bd, d, s, v, j, 1, In, J, Jt, st, b, d, s, tj, 3, vz, In, mz, 1J, mt, ItJ, mz, jz, jn, jnz, jdz, dst nz, sk, st, ts, jnz, jmz mps, qst, qks b, f, If, sks, qks d, tj, z, v, nz 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
continued Time 1 Time Participant Anh 0%
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, IfGs, jldz
Time3
Time 2 Nhi
Anh
Nhi
Anh
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
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
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
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
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
80 • ACQUIRING A NON-NATIVE PHONOLOGY
/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 participants, 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 targetlike 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 composed 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
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE
• 81
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 percentages 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 statistically 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
Ibl
Itl
Idl
/k/
Isl
Izl
/fif
M
Id
/e/
Iml Inl
Idl
III
Coda
/p/
Both P1' # n % Anh
. 90 30 34 88%
. 12 . 58 204 2 377 19 11% 54%
. 21 . 51 . 32 . 64 4 24 129 93 126 272 159 15 19% 47% 27% 58%
. 12 . 64 . 33 . 10 . 10 . 03 . 77 . 82 . 72 . 31 . 33 62 12 1 209 552 103 180 28 37 12 62 0 75 273 61 210 75 19 41 282 700 150 666 212 10% 61% 30% 16% 16% 0% 2% 74% 79% 69% 27% 29%
17 21 81%
0 7 0%
4 8 50 65 57 131 12 115 14% 38% 33% 57%
1 8 45 97 14 18 18 8 91 311 0 31 135 27 90 29 29 70 397 113 3 19 134 397 10% 67% 20% 28% 28% 0% 5% 68% 78% 64% 24% 27%
13 13 100%
2 16 79 100 141 172 12 69 17% 58% 23% 56%
# n % Nhi # n %
104 199 52%
/g/
0 3 0%
28 44 64%
44 14 19 138 34 120 10% 56% 37%
/JV
4 46 9%
/tf/
4 46 9%
W
0 118 241 58 83 0 31 16 22 148 303 80 269 99 0% 0% 80% 80% 73% 31% 31%
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 83
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 primarily 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 wordspecific. 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/.
84 • ACQUIRING A NON-NATIVE PHONOLOGY
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 coarticulated 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,
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 85
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 (pl . 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 production (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 (pl. 78) were strongly favoured; for the voiceless fricative /JY 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 Jt 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 Jt/ 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
86 • ACQUIRING A NON-NATIVE PHONOLOGY
Table 4. 9 VARBRUL analysis of production by singleton coda Factors T-L Abs.
Ep.
Fc.
2T
/Pi
(n: 34)
y#
%
/tk/ (n: 642) y # % /mn/ (n: 982) y # %
. 53 30 88%
0
. 36 4 12%
0
0
. 75 . 30 . 52 761 114 69 78% 12% 7%
. 77 103 69%
. 06 1 1%
/bdg/ (n: 160) P* # %
. 20 36 4%
. 35 . 53 39 30 19% 24%
. 16 . 82 1 45 1% 30%
0
%
. 60 . 42 . 61 26 130 51 59% 23% 12%
. 29 13 6%
/2V/
(n: 483)
þi # ^ %
. 35 . 78 . 85 . 27 90 226 122 17 19% 47% 25% 4%
/jtjy
(n: 87) P' # %
Input weight:. 012 Total x2: 20. 1147 X2/cell:. 8381 df: 5; p <. 0001
. 39 13 15%
. 34 4 5%
Following linguistic environment: Pause:. 40, Vowel:. 36, Glide:. 52, Consonant:. 62
Grammatical conditioning: Monomorphemic:. 48 . 32 . 86 . 73 Bimorphemic:. 97 7 5 79 3% 49% 4%
/sf/ (n: 220)
y#
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
Preceding linguistic environment: . 05 Vowel:. 56, Diphthong:. 83, 2 /o/:. 04 0%
(n: 150)
%
Statistical significance
Time: Time 1:. 55, Time 2:. 48, . 58 . 61 . 69 . 19 . 51 Time 3:. 47 12 16 332 186 96 52% 29% 15% 2% 2%
to/
y#
Other significant factor groups
0
Preceding linguistic environment: Vowel:. 50, Diphthong:. 85, /o/:. 25; Time: Time 1:. 42, Time 2:. 44, Time 3:. 61
Grammatical conditioning: Monomorphemic:. 49 . 78 Bimorphemic:. 80 28 Participant: Anh:. 46 Nhi: 6% . 55
Preceding linguistic environment: Vowel:. 54, /o/: . 21 . 90 . 94 . 20 1 12 Time: Time 1:. 58, Time 2: 57 1% 66% 14% . 63, Time 3:. 35
Input weight:. 15 Total x2: 11. 6052 X2/cell:. 7737 df: 6; p. <. 05 Input weight:. 176 Total x2: 13. 3545 X2/cell:. 9538 df: 6; p <. 025 Input weight:. 176 Total x2: 13. 3545 X2/cell:. 9538 df: 6; p <. 025 Input weight:. 025 Total x2: 9. 9535 X2/cell:. 9953 df: 7; p <. 05 Input weight:. 047 Total x2: 30. 8839 X2/cell:. 9359 df: 7; p <. 0001 Input weight:. 076 Total x2: 11. 6202 X2/cell:. 6835 df: 6, p <. 05 Input weight:. 009 Total x2: 21. 6941 X2/cell:. 7231 df: 7; p <. 001
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 87 continued
Factors T-L Abs. Ep.
/e
(n: 60)
Fc.
2T Other significant factor groups
Following linguistic environment: . 50 . 97* . 97* Pause:. 33, Glide:. 51, 1 35 Consonant:. 59, Vowel: . 67 19 2% 32% 58%
P' # %
. 17 . 61 1 4 2% 7%
/j I/ (n: 878)
Preceding linguistic environment: . 38 . 65 . 19* . 82 . 19* Diphthong:. 08, Vowel: . 53, 242 261 23 347 5 hi:. 67 28% 30% 3% 40% 1%
y # %
Statistical significance Input weight:. 003 Total x2: 13. 3000 5C2/cell:. 8313 df: 6; p <. 025 Input weight:. 19 Total i2: 9. 7433 x2/cell:. 8857 df: 5; p <. 05
T-L: Target-like, Abs: Absence, Ep.: Epenthesis, Fc.: Feature change, 2T: Two types of production. *Run as one factor to achieve a better model of variation due to similar input probabilities in earlier runs.
production was word-specific: as [t]: expect, correct', as [k]: connect; absence of both: correct. At time 2, all the clusters were bimorphemic and [k] was produced. Stop-Fricative: (/ks ts bz/). 68 per cent (49/72) had one member absent and 15 per cent (11) were target-like; /bz/ was produced as [ps] in both cases. If one member was absent, it was the fricative. Bimorphemic clusters had a higher accuracy percentage (50 per cent) than monomorphemic clusters (10 per cent). Fricative-Fricative: (/5z/). There were only 5 tokens of this cluster, all bimorphemic and with [z] produced for the entire cluster. Liquid-Liquid: (Iill}. In the only example of this coda, only [1] was produced. Liquid-Fricative: (/If 16 Iv iz Iz JtJ/). 24 per cent (9/38) were target-like, 39 per cent (15) had one member absent (the fricative in every case), 16 per cent (6) were produced as different consonants (e. g. [jt] for /iQf) and 21 per cent (8) had absence and epenthesis following a tense vowel or absence and feature change. The highest accuracy percentages were for /iz/ - 50 per cent (9/18) were target-like. Liquid-Nasal: (/in/). Eight of the 10 tokens of this coda were produced
88 • ACQUIRING A NON-NATIVE PHONOLOGY
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 targetlike 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).
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 89
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 disfavoured 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
90 • ACQUIRING A NON-NATIVE PHONOLOGY
absence at. 56 while bimorphemic codas strongly disfavoured absence at. 28. For CCC codas, grammatical conditioning was also statistically 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 targetlike while 43 per cent (13/30) of the past tense marker /d/ were targetlike, 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
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 91
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 production, 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 environment 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
92 • ACQUIRING A NON-NATIVE PHONOLOGY
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 / m n t k b d g s f h w y j 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 if) 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,
ACQUIRING A NON-NATIVE SYLLABLE STRUCTURE • 93 Table 4. 10 Hierarchies in target-like production for singleton codas Most target-like
Voiceless stop Nasals
P n> m> q
Fairly target-like
Voiceless fricatives
s, f
Somewhat target-like
Voiceless stops
t> k
Poorly produced
Liquids Voiced fricative Voiceless affricate
V
Very poorly produced
Unclear
Voiced fricative Voiceless fricative Voiced stops Affricates Voiced fricatives
u
tj z
J> e b, d, g tf> d3 3, 9
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 developmental 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]
94 • ACQUIRING A NON-NATIVE PHONOLOGY Table 4. 11 Onset and coda cluster production patterns Onsets
Codas
#FS -» #F #FG -» #F #FL (/I/) -* #L #FL (/i/ -> #¥ #SL -> #L #SG -> #S
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#
#FSL -» #F, #SL, #FS
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.
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 represented 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
96 • ACQUIRING A NON-NATIVE PHONOLOGY
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 participants 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, comparing 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 targetlike 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
98 • ACQUIRING A NON-NATIVE PHONOLOGY
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' phonologies 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
LINGUISTIC AND TASK CONSTRAINTS
• 99
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 classification 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.
100 • ACQUIRING A NON-NATIVE PHONOLOGY
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 participants 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
LINGUISTIC AND TASK CONSTRAINTS
• 101
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 syllableinitial 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].
102 • ACQUIRING A NON-NATIVE PHONOLOGY
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 diphthongs 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.
LINGUISTIC AND TASK CONSTRAINTS • 103
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 modifications 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
104 • ACQUIRING A NON-NATIVE PHONOLOGY
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
LINGUISTIC AND TASK CONSTRAINTS • 105
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 repertoires 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
106 • ACQUIRING A NON-NATIVE PHONOLOGY
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
LINGUISTIC AND TASK CONSTRAINTS • 107
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
108 • ACQUIRING A NON-NATIVE PHONOLOGY
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.
LINGUISTIC AND TASK CONSTRAINTS • 109 Table 5. 1 Onset and coda accuracy percentages by time, length and participant Time 1
Time Part. C onset CC onset CCC onset C coda CC coda CCC coda
Anh 91% 69% 0% 45% 5% 0%
Nhi 90% 60% 0% 53% 9% 0%
Time 2 Botii 90% 64% 0% 48% 7% 0%
Anh 93% 59% 60% 40% 4% 0%
Nhi 94% 56% 0% 48% 7% 0%
Time 3 Both 93% 58% 60% 43% 6% 0%
Anh 89% 63% 89% 49% 2% 0%
Nhi 89% 76% 67% 47% 11% 0%
Both 89% 68% 83% 48% 6% 0%
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-
110 • ACQUIRING A NON-NATIVE PHONOLOGY
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
LINGUISTIC AND TASK CONSTRAINTS • 111
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 / m n t k b d g s f h w y 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 w y n m s 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
112 • ACQUIRING A NON-NATIVE PHONOLOGY
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, III 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 twomember 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.
LINGUISTIC AND TASK CONSTRAINTS • 113
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 consonants 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
114 • ACQUIRING A NON-NATIVE PHONOLOGY
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 significant 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
LINGUISTIC AND TASK CONSTRAINTS • 115
(1996) and Benson (1988) found that Vietnamese speakers frequently 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 targetlike 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
116 • ACQUIRING A NON-NATIVE PHONOLOGY
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 environments 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 environment 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.
LINGUISTIC AND TASK CONSTRAINTS • 117
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];
118 • ACQUIRING A NON-NATIVE PHONOLOGY
/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 correspondence 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 accuracy, 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 Ushaped 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
LINGUISTIC AND TASK CONSTRAINTS • 119
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
120 • ACQUIRING A NON-NATIVE PHONOLOGY
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 CelceMurcia, 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 markedness 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
LINGUISTIC AND TASK CONSTRAINTS • 121
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
122 • ACQUIRING A NON-NATIVE PHONOLOGY
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
LINGUISTIC AND TASK CONSTRAINTS • 123
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 syllableinitial 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
124 • ACQUIRING A NON-NATIVE PHONOLOGY
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
LINGUISTIC AND TASK CONSTRAINTS
• 125
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 modification 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 performing 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
128 • ACQUIRING A NON-NATIVE PHONOLOGY
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 nonVietnamese 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 coworker 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 'happygo-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... m y . . . 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 comprehension; 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 understands 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:
130 • ACQUIRING A NON-NATIVE PHONOLOGY
'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 m e . . . y e a h . . . 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 coworkers 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... workfriends,' 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 factory... my group have... a new friend a new friend... woman... she she talk all the way. ' He credits her talkativeness to his English improvement: 'III speak English with my my workmates better... better y e s . . . 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
132 • ACQUIRING A NON-NATIVE PHONOLOGY
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 d a y . . . 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 simultaneously, 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 m e . . . 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 a n d . . . 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 m e . . . she and h e . . . and they are always... complain... complain m e . . . why you don't speak English... When she does talk with her customers, she understands only about half: 'my my customer always ask m e . . . 11 understand about fifty or seventy... y e a h . . . 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 job... yes... mmmm... the the difficult i n . . . with me are now pronunciation my m y . . . t h e . . . my pronunciation... is harder... because I always... communication
134 • ACQUIRING A NON-NATIVE PHONOLOGY
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: B u t . . . 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 o f . . . 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... y e s . . . 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 computers. 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... m m m . . . now I I feel I feel I f e e l . . . 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 m e . . . 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... a n d . . . 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 a n d . . . 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 environment 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 m e . . . example I ... I say [pa. l a t ] . . . they correct [pse. l a t ] . . . 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
136 • ACQUIRING A NON-NATIVE PHONOLOGY
speak... speak English... I guess by English... yeah... 11 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. 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 h a v e . . . I h a v e . . . 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 w o r r y . . . 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 a l l . . . my guess every everything is American... yes... in my home... always... I... American thing ... newspaper... magazine... hear m u s i c . . . American... American... no Vietnam no Vietnamese... I feel I am I am living ... American l i f e . . . yeah. For Anh, it's not so clear: 'I think... I not perfect American l i f e . . . 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 w e . . . we are... now we a r e . . . we f e e l . . . 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 h e . . . because in Vietnam I have a way l i f e . . . enough... I . . . confident in Vietnam but when I move in here I speak English not well a n d . . . sometimes in my job is difficult because I don't understood my customer... I I I feel lonely
138 • ACQUIRING A NON-NATIVE PHONOLOGY
because my my boss... sometimes complain m e . . . 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 y e a h . . . y e s . . . in here... a few a few friends... I don't have free time visits... my friend... 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 w a y . . . I think they are... very difficulty... live in here because they are not confident everything around... they a r e . . . don't have friend in American... they are ... life... 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 family... 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 y e a r . . . 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 a m . . . 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 n o w . . . 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.
140 • ACQUIRING A NON-NATIVE PHONOLOGY
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 g o . . . she very hard work very hard study... but I don't believe m e . . . 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 ingroup 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 coworkers, 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.
142 • ACQUIRING A NON-NATIVE PHONOLOGY
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 negotiation 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 'gogetter' 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
144 • ACQUIRING A NON-NATIVE PHONOLOGY
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 acknowledged by Nhi: 'my using English i s . . . less than h e r . . . 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 m e . . . 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 - forfinancialcaretaking 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 o b . . . b u t . . . he... my husband has a j o b . . . 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... a n d . . . don't have... free time t o . . . relax... and in here... when I have j o b . . . I take money... small amount of money I f e e l . . . I f e e l . . . my life in here... enough... don't worry... a n d . . . everybody around in Tucson... always friendly yeah... and always help m e . . . 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... s h e . . . encourage me study nail... I start think I don't study nail because I don't understand everything and she encourage the if you study in Vietnamese you can study in here... I study nail three months... after I try in class... 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
146 • ACQUIRING A NON-NATIVE PHONOLOGY
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 a n d . . . 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 h e . . . and they are always... complain... complain m e . . . why you don't speak English....
148 • ACQUIRING A NON-NATIVE PHONOLOGY
When she does talk with her customers, she understands only about half: 'my my customer always ask m e . . . 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 j o b . . . yes ... mmmm... the the difficult i n . . . with me are now pronunciation my m y . . . t h e . . . 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 o f . . . '. 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 t r y . . . 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 customers, 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
150 • ACQUIRING A NON-NATIVE PHONOLOGY
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 nonVietnamese? One could argue that different personality traits could affect communication styles. For example, Anh is typically introverted 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 communication skills (BIGS) over introverted learners. Additionally, variables 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
152 • ACQUIRING A NON-NATIVE PHONOLOGY
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 participants, 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
154 • ACQUIRING A NON-NATIVE PHONOLOGY
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 developmental 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.
156 • ACQUIRING A NON-NATIVE PHONOLOGY
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
A UNIFIED APPROACH • 157
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 environments 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.
158 • ACQUIRING A NON-NATIVE PHONOLOGY
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 development; 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 important. 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.
A UNIFIED APPROACH • 159
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 acculturation 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
160 • ACQUIRING A NON-NATIVE PHONOLOGY
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 meaningful 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
A UNIFIED APPROACH • 161
to her classmates have been broken, replaced by Vietnamese coworkers 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 communicating 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
162 • ACQUIRING A NON-NATIVE PHONOLOGY
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 p o o r . . . 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,
A UNIFIED APPROACH • 163
time and margin length, were statistically significant at p <. 025 (df = 4, total £ = 11. 4385). By examining the interaction among social and linguistic constraints 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 quantitatively 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 production 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 simultaneously 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 communicative 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 phonological 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
166 • ACQUIRING A NON-NATIVE PHONOLOGY
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 proficiency, 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 methodology 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 interactions, 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 longitudinal 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), 35—59. 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 relationships: 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. 18—29. 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
170 • ACQUIRING A NON-NATIVE PHONOLOGY interlanguage', in A. James and J. Leather (eds), Sound Patterns in Second Language Acquisition. Dordrecht, Holland: Foris Publications, pp. 14362. 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 nativelanguage (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 crossroads 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 considerations'. 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 multilingual 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 OptimalityTheoretic 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'. Language Learning, 46, (1), 3-73. 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.
172 • ACQUIRING A NON-NATIVE PHONOLOGY 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'. Language Learning, 35, (3), 405-17. 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 Development, 21, 14-32. 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 pronunciation of Americans in Norway'. The Modern Language Journal, 86, 17491. 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 r1. Language Learning, 36, 4, 453—504. 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. 20624. 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. 75—96. 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 Second Language Acquisition, 3, 109-35. 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.
174 • ACQUIRING A NON-NATIVE PHONOLOGY 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 Gender. Berlin: Mouton de Gruyter, pp. 231-56. 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 14th 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 acquisition', 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 mathematiques: 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 Phonology: The Acquisition of a Second Language Sound System. Cambridge, MA: Newbury House Publishers, pp. 365-77. Schumann, J. (1986), 'Research on the acculturation model for second language acquisition'. Journal of Multilingual and Multicultural Development, 7, 379-92. Schumann, J. (1990), 'Extending the scope of the acculturation/pidginization 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
176 • ACQUIRING A NON-NATIVE PHONOLOGY 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 Gender. Berlin: Mouton de Gruyter, pp. 175-98. 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. 17391. 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 Phonology: The Acquisition of a Second Language Sound System. Cambridge, MA: Newbury House Publishers, pp. 401-16. 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 Second Language Acquisition, 10, 281-302. 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 pronunciation was considered correct) hoof vroom onion, gwen, hundredth, human, prawn, brown, language, spring popular, future, pure, hear, yourself, other beautiful, jewel, scowl
B. Two-member codas: /sp/ grasp /sk/ risk /qk/ think /ids/ merge /ik/ fork /ib/ curb /if dwarf, smurf /jj/ porsche
178
/Ig/
• ACQUIRING A NON-NATIVE PHONOLOGY
burg
/in/ turn liml worm harp girl church lunch lunge humph grump kiln elm help alb shelve, twelve frills bulge milk belch welsh /Id/ field /Ik/ milk I\SI yourself /fB/ fifth /10/ health /de/ hundredth /n0/ month /te/ eighth /j0/ fourth /JS/ pierce /ps/ caps, schnaps /Is/ pulse /fs/ hofs /ns/ mons /0s/ cloths, smiths /ts/ nights /ks/ bikes /gz/ dogs /nz/ guns /qz/ sings /bz/ robs /5z/ clothes
/jp/ /jl/ /jtj/ /ntj/ /nd3/ 3/ /mtf /mp/ /In/ /1m/ /lp/ /lb/ /lv/ /lz/ /Ids/ /Ik/ /itj/
/ij/
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/ amidst /t6s/ eighths /sts/ lasts /sks/ flasks /mpt/ crammed /mps/' bumps /ndz/ wands /nst/ canst /nts/ months, since /ntjt/ lunched /qst/ amongst /qkt/ linked /qks/ thanks, sphinx /imO/ warmth /jmz/ worms
180 • ACQUIRING A NON-NATIVE PHONOLOGY
/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/ worlds /If0s/ twelfths /kstG/ 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
184 • ACQUIRING A NON-NATIVE PHONOLOGY 17. Please rate your own proficiency in English in the following areas: 5=Excellent 4=Good 3= Average
2=Poor l=Very Poor
Grammar
5
4
3
2
1
Listening comprehension
5
4
3
2
1
Reading comprehension
5
4
3
2
1
Writing
5
4
3
2
1
Pronunciation
5
4
3
2
1
Fluency
5
4
3
2
1
Vocabulary
5
4
3
2
1
Knowledge of American culture 5
4
3
2
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:
INITIAL QUESTIONNAIRE • 185 III. Language contact: Please circle your answers to these questions: All Most
Some A few None
1. How many people in your neighborhood speak English as a native language:
1
2
3
4
5
2. How many people in your neighborhood speak Vietnamese as a native language:
1
2
3
4
5
3. How many of your friends speak Vietnamese as a native language: 1
2
3
4
5
4. How many of your friends speak English as a native language:
1
2
3
4
5
5. If you are in school, how many of your classmates speak English as a native language?
1
2
3
4
5
6. How many of your classmates speak Vietnamese as a native language?
1
2
3
4
5
7. How many of your workmates speak English as a native language?
1
2
3
4
5
8. List the three friends in Tucson that you see the most every day. Name of friend
Language used
amount of time erach day
a. b. c.
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 a. b. c.
186 • ACQUIRING A NON-NATIVE PHONOLOGY IV. English language use: Every day 1
2 or 3 Once times a week a week 2 3
Once or twice a month 4
Never 5
1. How often do you speak English in your home?
1
2
3
4
5
2. How often do you speak English outside your home?
1
2
3
4
5
3. How often do you speak English to: a) your husband/wife
1
2
3
4
5
b) your children
1
2
3
4
5
c) your relatives
1
2
3
4
5
d) friends
1
2
3
4
5
e) neighbors
1
2
3
4
5
f) classmates
1
2
3
4
5
g) workmates
1
2
3
4
5
h) shop/bank employees
1
2
3
4
5
i) others
1
2
3
4
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
INITIAL QUESTIONNAIRE • 187 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. 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
7. Circle the average number of hours each day you read newspapers, books, or magazines 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
8. What kinds of radio programs do you listen to?_ 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. YES! 1
yes 2
no 3
NO! 4
a) Would you like more chances to speak English?
1
2
3
4
b) Would you like more chances to write English?
1
2
3
4
c) Do men have more chances to speak English than women?
1
2
3
4
d) Do children have more chances to speak English than adults?
1
2
3
4
e) Would you like more English speaking friends?
1
2
3
4
188 • ACQUIRING A NON-NATIVE PHONOLOGY YES! 1
yes 2
no 3
NO! 4
f) Are Americans helpful when you speak English?
1
2
3
4
g) Do you have to speak English to do well in the US
1
2
3
4
h) Do you have to speak English to do well at Pima?
1
2
3
4
i) Would you do better at Pima if you spoke English better?
1
2
3
4
j) Does your daughter speak better English than you?
1
2
3
4
k) Does your husband/wife speak better English than you?
1
2
3
4
1) Will you lose contact with your children if you do not learn English? m) Are you happy you came to the US? 2. Please circle your answers to these questions. Many times
A few times
One time
a) Have you ever had a good experience at school because you are an international student who speaks English as a second language?
1
2
3
4
b)Have you ever had a bad experience at school because you are an international student who speaks English as a second language?
1
2
3
4
c)Have you ever had a good experience in the community because you are Vietnamese and speak English as a second language?
1
2
3
4
d)Have you ever had a bad experience in the community because you are Vietnamese and speak English as a second language?
1
2
3
Never
4
INITIAL QUESTIONNAIRE • 189 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?
192 • ACQUIRING A NON-NATIVE PHONOLOGY
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 Sunday Monday Tuesday 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
Wed.
Thursday Friday Saturday
Index 20, 114, 168 Abrahamsson, N. absence, definition of 18, 44 26, 143 accent (degree of) acculturation 25, 142, 158-9 Acculturation Model 142 18, 44 acquisition, definition of 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 Cancino, H.
44, 169
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 77-81, 81-90 individual codas and onset 90-4 collaborative interactions 148 community, effect of L1/L2 29, 14952 46, 48, 175 Corbin, J. 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
iNDEX . 195 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. Hammond, M. Hancin-Bhatt, B.
10, 171 35, 171 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 see 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, 99100, 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. Macken, M. A. Major, R. C.
2, 26, 29, 33, 143, 170, 174 2, 11, 173 8, 9, 10, 13-15, 17,
196 • ACQUIRING A NON-NATIVE PHONOLOGY
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 65-71 individual onsets 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 23-4 phonological marker 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 20, 175 Saunders, N. scaffolding 148 2, 33, 34, 175 Schmidt, R. W. 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, 15763, 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 32—3 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 description of 28, 44-6 140, 176 8, 32-5, 118-120, 156 8, 10, 20, 33, 116, 122, variation linguistic 32-3 176 see also linguistic environment Tarone's Capability Continuum social 32-3 Paradox 33 task 8, 33-5, 118-20, 156 task variation 8, 33-5, 92-3, 118-20, 156 Vietnamese Teutsch-Dwyer, M. 25, 176 people in America 48-9 phonology 2, 4, 37-9, 45, 95, transfer 2, 7-10, 13-15, 19, 21-2, 114-15 95-105, 108, 114-15, 120-5, 153-7, 164 Vigil, F. 30, 174 Tropf, H. 19, 113, 124, 176 Wagner-Gough, J. 148, 176 Weinberger, S. H. 8, 16-17, 19, 20, Universal Canonical Syllable Structure 108, 122, 124, 176 19, 20, 113-14, 124, 154 universal syllable structure preference Wood, M. 28, 143, 176 16, 21, 107-8, 123 U-shaped curve 13, 118-19 Yavas, M. 33, 176 Young, R. 22, 44, 176 Vago, R. M. 8, 9, 10, 122, 168, 176 142, 143 VARBRUL Zisa Project analysis of data 61-5, 66-8, 73-7, Zuengler, J. 23-4, 33, 168, 169 81, 84-7, 89, 116 Tajfel, H. Tarone, E.