acquire a non-native phonology

8/11/2019 Acquire a Non-native Phonology

1/204


2/204

Acquiring a Non native Phonology


3/204

his page intentionally left blank


4/204

Acquiringa

Non native Phonology

Linguistic Constraints and Social arriers

Jette

G

Hansen

ont nuum


5/204

Continuum

The

Tower Building

80

Maiden Lane

11 York Road Suite 704

London SE1 7NX New York, NY 10038

USA

First

published 2006.

Jette

G.

Hansen 2006

ll

rights reserved.

No

part

of

this publication

may be

reproduced

ortransmittedin

anyform or by anymeans, electronicormechanical, including photocopying,

recording,

or any

information storage

or

retrieval system, without prior permission

in

writingfromthe publishers.

BritishLibraryCataloguing in Publication

Data

catalogue recordforthis bookisavailablefrom theBritish 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


6/204

ont nts

1 Introduction

2 InvestigatingL2Phonology

3 ResearchingtheNguyen Family

4

Acquiring

a

Non nativeSyllableStructure

5 Linguistic and Task Constraints

6 Social Barriers

7 AUnifiedApproach

8

Conclusions

eferen es

ppendices

ndexes

1

7

40

59

95

126

153

164

168

177

194


7/204

his page intentionally left blank


8/204

ntroduction

In this introductory chapter to the book. Acquiring

Non-native

Phonology:

Linguistic

Constraints

and

SocialBarriers

the

purpose

of the

book,

as

well

as a

brief introduction

to the

study

and its

background,

is

presented. Finally,anoverviewof thestructureof thebookisgiven.

Purpose

of the

book

The followingvolumeis amonograph studyof 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,

linguisticfactors, suchastransferandmarkedness,andsocial factors,

such

asopportunities 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

relativelycomplex phonology, withawide rangeofconsonants ableto

occupy both syllable-initial onset) andsyllable-final coda)

position,


9/204

2

ACQUIRING A NON-NATIVE PHONOLOGY

aswell 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 typesand lengths of consonants

in

onsets and codas. L2 learners of English whose LI is simpler, such

asVietnamese, 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 typesof modification processes learners favour Onereason

for thelack ofconsensus is the

fact

that the

majority

of theresearch in

L2phonology hasbeen one-off, rather than longitudinal, with studies

being done with learnersof different ages andlanguage levels, making

comparisons among studies difficult Therefore there

is a

lack

of

knowledge about

how

processes interact

and

change over

a

longer

periodoflanguage 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 hasfocusedon 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

alsosuggested that the socially structured and reinforced gender roles

of

boththe LI and the L2culturemayhavean

effect

onopportunities

forL2development. Additionally, work byFlege andcolleagues (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

extentof LI use; that is,individuals may be more likely toretain an

accent in L2 if they have greater use of LI, whereas greater use of L2

maynot minimize anaccent 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 acquiringan LI phonology (i.e. child LI phono-

logical

acquisition) (e.g. Leopold 1939; Macken 1979)

are

considered

to be

classics

in the field,

there have been

few

longitudinal

or

in-depth

analyses

of individual learners acquiring an L2 phonology. Yet,


10/204

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,

butalsodue to itslongitudinal

nature,

and finally, thecomprehensive

and

detailed analysis

of the

data

due to a

focus

on one family.As

such,this bookfills aunique voidin the SLAliterature- thatis, 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.Thisbook should be of interest to teachers

since it

will

provide them with concrete examples of SLA processes

viaadetailed discussionofactual language learners, and bymaking

theory and research accessible and meaningful to their needs via

specific

examplesand 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 secondaudience,researchers andteachers interestedin L2

phonology and pronunciation teaching in specific, this monograph

provides an in-depth analysis and

discussion

of the developmental

processesinacquiringanon-native sound system- presentingarich

baseof L2phonological data acrossalongitudinal timefram e -which

hasnotpreviously been presentedin theliterature.Thiscomprehen-

sive

database should

be of

interest

to

anyone concerned with

the

acquisition

of an L2

phonology. Additionally,

it

should

also

be of


11/204

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

oflearners,

as

well

as

specific details about

why

theseproductions

and

modificationstake place.

2 ackground

to the

study

Thepresent study examinestheacquisitionofEnglishL2consonants

overnearly

one

year

and focuses on

learners

of

spoken English

who

haveentered

theUnited

Statesaround

one

year before

the

onset

of

the study. Consonants were the

focus

of the study because this is an

area

ofdifficulty formany learnersofEnglishandtherefore 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 arelatively simpler phonological structure in comparison with

English, and English onsets and codas aredifficult for these learners

to

acquire. Participants

who had

been

in the

United States

for

around

one

year

beforethe

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 partof thedata 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. fter

the

course

was

completed,

I

approached herabouther interest inparticipatingin the study, and

after

agreeing to do so, she suggested meeting in her home and the

addition of herhusband as aparticipant 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

familyhave similar

LI

backgrounds. Additionally,

the

family

provides a unique opportunity to research how the differentialsocial

networks

and

identity formations

of the

members

of the family

influenceL2acquisition, aswellaswhat varietyofEnglishistargeted

for acquisition as based on social e.g. peer) networks and in-group

identification. Communication within the

family

can also be exam-

ined. Insights into language use, language resistance and language

maintenance withinafamily arecrucialinunderstanding howsocial

forces

affect

language acquisition

and

attrition.


12/204

INTRODUCTION 5

This

study examines linguistic and social factors in order to

ascertain

how

linguistic

and

social factors individually

and

interac-

tively

constrain, both positively andnegatively, L2 acquisition. Data

on socialfactorswere collected via interviews, participants language

logs (language

use

journals),

and

observations,

and

data collection

tookplace approximately once aweek for theduration of the study.

Interviews

were alwaystape-recorded, and field notes were also taken

duringthe interviewsand 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 samplesfor 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.

3 tructure of thebook

The structure of thebook is asfollows: this

chapter

hasprovided an

overviewof the purpose of the book, the background to the study, and

the structure of the book. Chapter 2: Investigating L2 Phonology, is a

review of the literature in three major areas: linguistic constraints,

including LI transfer, developmental effects markedness and lin-

guistic environment; social factors, including gender, social identity

and extent of LI and L2 use; and variation based on linguistic, social

and task

effects.

An overview of both English and Vietnamese

phonology

is

also provided

in

Chapter

2.

Chapter

:

Researching

the

Nguyen Family, details

the

selection

of the

participants,

the

background

of the family aswell as that ofVietnamese immigrants

tothe US, aswell as how data - bothphonological 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 ofonset 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

wellaslinguistic environment, grammatical conditioningandtask

effects. Chapter

6:

Social Barriers,

first

provides

a

detailed overview

ofthesocial contexts oflanguage use for theparticipants 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,


13/204

ACQUIRING

A


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 ofsocial 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,followedby suggestions forfutureresearch. Finally, an

epilogue

to the

story

of the

Nguyens,

the

participants

in the

study,

is

given.


14/204

InvestigatingL Phonology

Thischapter provides an overview of the

major

issues and findings in

researchon L2phonology relevantto the

focus

ofthis 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 Theoreticalframeworksandresearch findings

2 LI inguistic constraints

The discussion oflinguistic constraints

will

first examine the

effect

of

LI transfer on the acquisition of an L2 phonology.This isfollowed 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

AsLeather 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-


15/204

8 ACQUIRINGA 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 themajorityof 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

ofLI transfer for syllable structures inVietnamese interlanguage in

the learners preference for closed over open syllables (open would

be a syllable ending in a vowel such as toand 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 ofsyllable position on consonant cluster

production in Vietnamese interlanguage: word-final consonant

clusters

were produced with less accuracy than word-initial con-

sonant clusters. In her study of task variation and final consonant

and consonant cluster production by a Vietnamese learner of

English, Sato (1985) also found LI transfer effects on production

accuracy.


16/204


17/204

1

ACQUIRING

A


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 inKorean, andproduction 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 onwhich

LI rules tend totransfer, with those rulesthatsimplify pronunciation

transferring more easily than morphologically restricted rules

(Altenberg and Vago 1987; Broselow 1987; Hammarberg 1997;

Vagoand

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, Dutchor

Japanese children,

every

description based on

careful

observation

repeatedly confirms the strikingfactthat 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


18/204

INVESTIGATINGL2PHONOLOGY 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/, aprocess knownas 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:

frontfricativesare

acquired prior

to

back

fricativesand the

acquisition

ofback 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

areusually 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 hypercorrectionof the twophonemes in favour of/k/.

Therefore, when acquired,/k/ at first merges with /t/ and only later

develops

into a separate phoneme. Affricates also called half-stop

consonants,

are

acquired only

after the

acquisition

of the

corres-

ponding fricative, and before acquiring the affricate a child may

substitute the sound with either its corresponding

fricative

or stop.

The latest phonemic acquisitions for the child are those oppositions

that rarely occurin thelanguagesof theworld, e.g. nasal vowelsor the

second of one of the two liquids (either III or /r/) (Jakobson 1968, 47-

58).

In

development

of

English

as an LI,

children often substitute

/w/

for/i/.

1

Other developmental processes include assimilation processes such

asconsonant harmony, wherein consonants within

the

same word

are

assimilated. Typically, velars and labials are most resistant to

assimilation while dentals are the most frequently assimilated

consonants (Macken and

Ferguson 1981).

Other assimilation pro-

cesses include consonant voicing

in

voiced environments

and

devoicingin 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


19/204

12 ACQUIRING A NON-NATIVE PHONOLOGY

the deletion of final consonants, especially in consonant-vowel-

consonant (CVC)syllable structures. Ingram notes that the first

consonants typically acquired

in final

position

are

velars.

In

terms

of

the

deletion

of

unstressed syllables,

in the

initial stages

of

language

acquisition the first words acquired are monosyllabic, and when

bisyllabic words are

used,

the unstressed syllableis 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 isreduced to aschwa), and secondly, all unstressed syllables

existing in three syllable words are deleted. In the third stage, medial

stressedsyllables

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 (Ingram1976: 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 firststageismarkedby thedeletionof theentire 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 aglide, for example. Stagefourmarks thechild 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).Intwo-consonant final clusters, stops may be deleted,

especially

in

nasal-stop sequences.

Research

on L2 phonological acquisition has examined the extent

to which L2 developmental processes(the sequences and modifica-

tionsan 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) learningESL. 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-


20/204


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 alsodid not examinetheeffect 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 ofword-final obstruent devoicing is

also

common

for L2

learners

of

English whose

first

languages

do not

haveword-final

voiced

(or

voiceless) obstruents, such

as

speakers

of

Mandarin Chinese (Flegeand Davidian 1984), and Icelandic (Hecht

and Mulford 1982; MulfordandHecht 1980).

2 1 1 3 nteraction between transferanddevelopmental 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;

ajor


21/204

14

ACQUIRING

A


1987c; Mulford and Hecht 1980). In hisworkonJapanese native

speakers' production

of

English consonant clusters

and final

conson-

ants.

Major (1987c) found that both transfer and developmental

effects applied todevoicing. Flegeand Davidian (1984)also found

thatbothLI 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

finalstops, whilebothdevelopmental and LI transfer effects were

foundto

affect

learners' devoicingof finalstops. Overall, therewas a

higher frequencyofdevoicing than deletionfor finalstops.

Mulford

and

Hecht (1980)

and

Hecht

and

Mulford (1982)

researched a6-yearoldIcelandic child'sacquisition ofEnglish 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 devoicedfinalstops, anEnglish LI devel-

opmental

effect

and then strongly aspirated them,which is charac-

teristic

of

stops

in

Icelandic. Hancin-Bhatt

and

Bhatt (1997),

in

their

research

on the

production

of

English onsets

and

codas

in

monosyl-

labicwords by native speakersofJapanese and Spanish, found that LI

transfer had asignificant

effect

on L2 syllable structure production,

affectingboth 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 similartoonsetsintheirLI. In theabsenceofpositive transfer,

however, developmental effects were evidencedtohavean effect on

production.

It has

also been argued that

LI

transfer

is

more prominent

in the

earlystages of acquisition. Leather and James (1991) state that:

it would appearthat,as acquisition proceeds, the influence of the

LI and the

mechanism

of

transfer give

way

gradually

to other

influences that shape developing L2 speech such as the mechan-

isms (i.e. developmental processes ) associated withthe acqui-

sition

of the

mother tongue, (original parenthetical aside, 326)

Major (1987a) developed the Ontogeny Model2 to explain how

developmental

and

transfer

effects

interact

in L2

phonological

acquisition. Transfer

effects

predominate in the early stagesof L2

phonological acquisition, and then gradually decrease, while devel-

opmental

effects

increase

in the

middle stages

of

acquisition,

and

then

gradually

decrease. In terms of phonological similarity of the LI and

the L2,

Major

postulates that

for

similar phenomena, transfer

effects


22/204

INVESTIGATINGL2 PHONOLOGY 15

will

be more common but for phenomena that are dissimilar,

developmental

effects may

play

a

greater role

in

acquisition

and

production.

Major

furtherstates that production indifferent speaking

styles

is

affected

differently

by

transfer

from

the LI and

develop-

mental constraints: as the formality of the style increases, there is a

decrease in errors due to transfer, whereas errors due to develop-

mental effects increaseand 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 elicitdifferent styles (in this

case,

different levels of formality). Results indicate a

trend,

which was

not

statistically significant,

of

advanced learners having more

instancesoferrors 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 testedtheOntogeny Model in his study ofnative

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 ofthis study wasonly

four

weeks, it is

difficult to

determine whether

the

data

is

truly able

to

show change

between

the effects of

transfer

and

developmental

constraints.

In summary, research (e.g. Flege and Davidian 1984; Hancin-

Bhatt and Bhatt 1997; Hecht and Mulford 1980; Major 1987c;

Mulford and

Hecht

1982) indicates that transfer and developmental

effects may

interact

in L2

acquisition, with

LI

transfer dominant

in

the

early stages

of

acquisition

and

developmental

effects

increasing

as

LI transfer

effects

decrease. Both mayalso

affect

theproduction and

acquisition

of a

single segment;

it has

also been found that while

LI

transfer may

affectwhich sounds

are

difficult developmental effects

may

affectsubstitutions. Finally, it is possible that positive LI transfer

effects m ay override developmental

effects.

2 4

arkedness

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 onissues such asnaturalness, salienceand

ease

of

articulation

in

terms

of

phonological features.

SLA

researchers


23/204

6

ACQUIRING

A


have employed several definitionsof markedness inorder 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 thesewillbe examined in turn.

a. A pre ference for a universal syllable structure

Onearea of markedness research is theuniversal preference for a CV

syllablestructure, 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

learnersofEnglish has found that LI transfereffects 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

Anotherway researchers have employed markedness in SLA research

is to definemarkedness inonsets/codas bylength. Alllanguages have

a core consonant-vowel (CV) syllable structure (James 1988: 5;

Spencer 1996: 82).

CV

syllables

are

thus

the

least marked syllable

structure in languages, with increasing length increasing the marked-

ness of the

syllable structure (Carlisle 1997). Although

the CVC

syllable

does

not

occur

in all of the

world s languages,

it

does occur

frequently

and is

therefore also considered universally unmarked

(Spencer 1996: 82). Additionally, the existence of codas or onsets

witha length of A implies the presence ofcodas/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

exceptiontothis 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 ofword-final codas,

found

that the number of

modifica-

tions increased as the length of the codas increased.

This

indicates


24/204

INVESTIGATING L2PHONOLOGY 17

that markedness had an effect on the number of modifications.

Anderson s (1987) research

on

word-final consonants

and

consonant

clusters also foundthat 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

themore marked voiced

obstruents.

Carlisle (1997) also found

that shorter onsets werenotmodifiedasfrequently asmore marked

onsets.

c .

Markedness defined

by

type

o f

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,

aremore marked inrelationtoconsonants 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+ fric tive (/ts/ as in hats ) and istherefore more marked; fric tive

+

fric tive

(/fs/

as in h o o f s )

impliesfricative

+

stop

and

stop

+

fric tive

and istherefore more marked; stop-fricative ispreferredinonsets 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 consonantsfrommost to least marked: final >

medial > initial. In research on native speakers of Vietnamese


25/204

18

ACQUIRINGANON-NATIVE PHONOLOGY

acquiring

English,

Hansen (2004) found

that

learners had lower

accuracyratings 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 o f 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 doesnotexist between

a three-member and a two-member cluster based on length alone -

the twoelementsof thetwo-member coda must also comprise part of

the

three-member coda.

Thus,

/illwould

be

unmarked

in

relation

to

/ild/ w he rea s A n/ w ould not. H ans en 20 04) exam ined ac qu isition of

codas

by

length

and

type,

and

found that

the

three member codas

learners

had

most

difficulty

with were liquid-fricative-fricative, stop-

fricative-fricative, nasal-fricative-fricative, nasal-stop-stop and stop-

stop-fricative,

all comprising the more marked structures.

Eckman (1991) examined

the

acquisition (defined

as

correct usage

80 per cent of the time) of two- and three-member onsets and codas

by

native speakers

of

Japanese, Cantonese

and

Korean,

as

well

as how

markedness relationships influenced acquisition. Eckman

found

that

for

almostall the cases(98 per cent), if themore marked structure

(e.g. three-member codas) were acquired by the learner, the

subsequent less marked structure (i.e. a related two-member coda)

hadalready 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 inrelation to other

segments.

The

researchers posit that clusters closer

in

sonority

are

more marked than clusters wherein consonants have

a

wider sonority

distance.Thesonority hierarchy, from themost sonorousto theleast,

is

as

follows

vowels

>

glides

>

laterals

>

nasals

>

fricatives

>

stops.


26/204

INVESTIGATING

L2

PHONOLOGY

19

Their findings on Japanese and Korean learners of English s word-

initial consonant cluster production supportedtheirhypothesis

as the

participants

in

their study produced less marked onsets more

accurately

than the more marked onsets.

Another approachtosonority isresearchthat 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). ResearchbyTropf (1987) andCarlisle (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

wereoftenmodified

to

conform

toUCSS via

absence, feature

change(liquid-stop-fricativeandnasal-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

modifiedwere changed into CC codas

that

met theUCSS.

In summary, research has consistently found that L2 learners

modify marked syllable structures more frequently than they modify

unmarked syllable structures asbased 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 aremodified in favour ofshorter, orless 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 ofEnglish syllable structures, found that the

preceding vocalic contextwas a

factor

in LI transfer processes, for

example,single consonants were

always

deleted

after

diphthongs, and


27/204

20

ACQUIRING

A


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, wasmost commonly deleted post-vocalically in

both

final clusters and in word-medial sequences, while

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 finalobstruents with the produc-

tionofnative speakers since native speakers of English often devoice

finalobstruents before pauses orbeforeavoiceless sound. Portions of

her data revealed similar processes ofvowelepenthesis by both native

speakers

and

Japanese speakers

of

English: after word-final voiced

stops. Carlisle (1994) re-examined Tarone s (1980) data andfound

that over

50 per

cent

of the

instances

of

epenthesis

after

word-final

consonants occurredbefore 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 morelikelyto 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 ofthis type ofcluster

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 significantlyto 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.


28/204

INVESTIGATING

L2 PHONOLOGY 21

2 1 1 6 The

acquisition

of

English

phonologyby speakers of

Vietnamese

There have been several previous studies on Vietnamese learners

production of syllable onsets and codas. Sato (1984), in a

longitu-

dinal study of two Vietnamese adolescents, for example, examined

whether closed syllables would

be

preferred

in the

English

inter-

language ofVietnamese learners, basingher 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

confirmedboth

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

wasfavoured over deletion, epenthesis andfeature 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

ofEnglish, Sato (1985) also found LI transfer affected production

accuracy.

Benson (1988) also examined theuniversal preferencefor an open

syllablestructure

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 theuniversal preferencefor an open-syllable structurehad

alesser

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 questionedSato s (1984) conclusions

that learners reductions

of

clusters

by one

member

is due to LI

transferexclusively, since

it

also

is

evidence

of a

universal preference

for anopen syllable.

Osburne s (1996) research confirmed previous studies,as shealso

foundthat

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


29/204

22

ACQUIRING

A


arenot allowed in syllable final position in Vietnamese). Osburne s

research

also

confirms Benson s (1988)

findings

that native language

influences

onsyllable structure production interms 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 isrealized as avowel+ glide, rather than a single segment

asit is in English. Therefore, this syllable isclosed, and anyfollowing

consonants would violate Vietnamese

syllable

structure, and thus be

optimally

deleted. Additionally, Osburne

found

that

for her

subject,

i wasnever realized inEnglish 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, thereissome speculation that apostvocalic 111should

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 l t wasconsidered 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

likelyto be

reduced. This confirms research by Young (1988) on English tense

markingby native speakers

of

English- redundancy in marking which

make the past tense ending more salient werefound to increase the

likelihoodof the past tense endings being produced. Finally, Osburne

alsofound

that

the final

cluster /ks/, when representing

the

letter

xin

orthography, was not reduced, possibly due to its orthographic

salience.

Overall, the

research

on

Vietnamese learners

of

English indicates

that LI transfer playsa significant role insyllable structure produc-

tion, and that linguistic environment, especially preceding vocalic

environment, plays a great role in consonant deletion, providing

additional evidence

for the

primacy

of LI

transfer

in the

production

of

syllable codas

for

native speakers

of

Vietnamese (cf. Benson 1988;


30/204

INVESTIGATING

L2

PHONOLOGY

23

Osburne 1996; Sato 1984, 1985). Additionally., salience of past tense

marking

and

orthography

may

also

be a f ctor in

terms

of

which

consonant clusters are typically not reduced (Osburne 1996).

2 2 ocial

barriers

The discussion of social barriers in L2 acquisition, and specific llythe

acquisition

of an L2

phonology, starts with

a

discussion

of

social

identity.

Gender,as one

aspect

of

identity,

isthen

addressed.

The

social contextof LI and L2use,in the

form

ofextentofusageof the

LI and the L2, is

then

explored. Finally,

ffective

variables, namely

attitudes, motivation and investment, are explored.

2.1.2.1 ocial 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 thatone has to be...' (386).

Language

and

identity

are

extricably intertwined. Giles

and

Byrne

(1982) state that 'Accent servesas apow erfulsymbolofethnicityand

'psychological distinctiveness' (as quoted by Leather and James

1991: 310). In

fact,

linguistic behavior is 'a series of

acts

ofidentity

in

which people reveal both their personal identity

and

their search

for

social

roles' (LePage

and

Tabouret-Keller 1985: 14).Theseacts

can

be both conscious and unconscious, and resisting specific language

featuresis asimportantinmarking one's social identityas 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

identifythemselves with

particular sociocultural groups' (Leather

and

James, 1991: 310).

In a

summary ofresearch on L2 social markingsviaphonology, Dowd,

Zuengler and Berkowitz (1990) state that social markings can occur

at any age

group

and

level

of L2

acquisition, despite

a

limited


31/204

24

ACQUIRING

A


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 inwhich weneed agreater awareness ofsocial

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 astheir 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 togive 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 studyofCantonese speakers production of

English obstruents states that the reason why some Cantonese

learners exhibit non-target variantsintheir obstruent production may

be

because they

are

targeting

a

nativized variety

of

English, namely

Hong Kong English, which exhibits these variants, and therefore

analysingthis data forinterlanguage (IL) rules andprocessesis, as she

states,

a

questionably valid procedure (391).

She

further states that,

for

many phonetic features, nativized varieties and

other

non-

standard Englishes

offer

different targets than those

of

what

has

been

calledStandard English (391).

Similarly, Hecht and Mulford (1982), in their research on an

Icelandic child s acquisitionofEnglish

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

exposedto 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


32/204


to signal their identity. Gatbonton (1975) for example, found that

the

production

of

interdental

fric tives by

French-Canadian learn-

ers of

English varied based

on the

learners self-identification

as

either nationalistic

and

pro-French

or

non-nationalistic

and

pro-

English,

with

the

latter employing greater English

L2

dental

fricative usage.

The

learners level

of

integration

and

acculturation into

the L2

culture may also ffect the degree towhich certain L2 variants are

produced,

as Lybeck (2002) found in her research on the acquisition

ofNorwegian byAmerican 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

than

Norwegian /r/.

2.1.2.2

ender

Social

identity is

lso

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

ininteraction 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 m jority 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

owthese actions, activities,

and behaviours areencouraged differently for women andmen, 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

differenti l

access to L2 use and


33/204

26 ACQUIRINGANON-NATIVE PHONOLOGY

development opportunities

on the

cquisition

of an L2

although recent

research in the acquisition of an L2 phonology has begun suggesting

that thedegreeofaccent in the L2 may beinfluencedbyextentof LI

use- the greaterthe LI use, thegreaterthe degreeofforeign accent

cf. Flege, Frieda

and

Nozawa 1997; Piske, MacKay

and

Flege

2001).

Learners

may also resist

acquiring,

and

using,

specific linguistic

formsbecause

it

conflictswith their

LI

identities, particularly gender

roles

in the LI

culture

and the L2

culture.

This is

important

for SLA

researchers andteachersto 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 signaledby a certain accent sometimes serves as a

psychosocial constraint pressuringthenormativenot to speakin 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

ofwhite women, all native speakers of English, studying Japanese in

Japan. Although this is focusing on syntax, the study provides

interestingexamples

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

womenreported

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.Forexample,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

ofhonorific languageandused thepolite statement I

think

inplace

of honorifics. Another woman also avoided the use of honorifics,

instead

using

the

modal auxiliary desho

tosoftenher

utterances

in a

more feminine manner.

A

third woman,

who

inaccurately believed

that sentential-final particles functioned only in women s language.


34/204

INVESTIGATING L2 PHONOLOGY 27

avoided usingpragmatic particles altogether.

All

threewomen were

conscious

of

their language choices,

but

avoided using 'correct'

Japanese formsin situations when they knew theyshould,because it

conflicted withtheirLI identity.

Ohara's (2001) work

on the

acquisition

of

intonation

by

American

women learning Japanese also illustrates how learners may resist

using certain L2 featuresifthey 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

voiceand '..

.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, andEnglish (LI) - Japanese (L2) bilinguals were aware

of these norms and willing to employ them to perform Japanese

gender.

She

foundthatwhile

the

beginner learnerswere unaware

of

the

norms, theJapanese (LI)-English (L2) bilinguals were awareof

the norms and employed them to perform gender. However, the

English (LI) -Japanese(L2) bilinguals were awareof thenormsbut

several

(2 out of 5) ofthem were unwilling to employa high pitch

because they perceived 'that the voice pitch levels they employed

correlated neatly with their attitude toward

the

kinds

of

images

typically associated with Japanese women' (Ohara 2001: 242)and an

identity they

did not

accept.

Finally, gender has also been linked to the variety targeted.

Adamson

and

Regan (1991) investigated

the

acquisition

of

commu-

nityspeech normsbyVietnameseandCambodian immigrantsto the

US. The

researchers examined

the

participants'

use of the

phono-

logical variable'ing', specifically focusing on two variants, /irj/and

/in/, hypothesizing that greater use of/in/, the non-prestige variant

and

a

variant

not

present

in the

participants'

LI

phonology (versus

/ i r j /, w h i c h i s p r e s e n t , w o u l d in d i c a t e a g r e a t e r i n t e g r a t in g o f

participants

into

their respective speech communities. The partici-

pants were all

from

either Philadelphia or Washington DC, where /in/

functioned as a

non-prestige variant

of

'-ing'.

The

researchersfound

menused

/in/ more frequently

than

women,

a pattern that was similar to that found for native speakers in these

particular

contexts. Theresearchers state that these findings suggest

thattheparticipants aretargeting differentgroupsofspeakers, i.e.the

non-native women

are

targeting

the

language

of

nati

acquire a non-native phonology

Documents