phonological acquisition in bilingual spanish–english speaking children

Upload: nicolelissettecarvacho

Post on 01-Mar-2018

229 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    1/20

    Phonological Acquisition in BilingualSpanishEnglish Speaking Children

    Purpose:In this study, the authors aimed to determine how between-languageinteraction contributes to phonological acquisition in bilingual SpanishEnglishspeaking children.Method:A total of 24 typically developing children, ages 3;0 (years;months) to4;0, were included in this study: 8 bilingual SpanishEnglish speaking children,8 monolingual Spanish speakers, and 8 monolingual English speakers. Single wordand connected speech samples were obtained for each child. This study examinedinteraction between the two languages of bilingual children during phonological

    acquisition through the measurement of (a) transfer(the frequency and types ofphonological transfer present in the speech of bilingual children); (b) deceleration(a slower rate of acquisition for bilinguals as compared with monolinguals); and(c) acceleration (a faster rate of acquisition for bilinguals as compared withmonolinguals.Results:Findings demonstrated that (a) transfer was evident in the productions ofbilingual children, (b) differences were found in accuracy between monolingualand bilingual children, and (c) sound frequency did not predict differential accuracyof either phonetically similar sounds between languages or phonetically dissimilarsounds specific to Spanish or English.Implications:The results from this study indicate that transfer, deceleration, and apossible variation of the acceleration hypothesis occur in bilingual phonologicalacquisition. Evidence was found for separation and interaction between the bilingual

    children

    s 2 languages (J. Paradis & F. Genesee, 1996).KEY WORDS: phonology, interaction, bilingual, Spanish

    S ome studies have found that bilingual children begin with a singlelanguage system that gradually separates into two autonomoussystems (e.g., Leopold, 1970; Schnitzer & Krasinski, 1994; Vogel,1975). This orientation to bilingual language acquisition is referred to as

    the unitary system model (USM; e.g., Bhatia & Ritchie, 1999; Genesee,

    1989). In contrast, thedual systems model maintains that bilingual chil-

    dren develop separate language systems for each language andthat thesesystems do not interact (e.g., Keshavarz & Ingram, 2002). A third possi-

    bility is that the two systems of bilingual children do interact (Paradis,

    2001; Paradis & Genesee, 1996). To account for interaction between lan-

    guages, Paradis and Genesee (1996) proposed a series of interdependence

    hypotheses in an attempt to determine how bilingual children acquire

    their two language systems.Interdependence has been defined as transfer,

    deceleration, and accelerationin bilingual language acquisition(Paradis &

    Genesee, 1996). Hereafter, this notion of interdependence in acquisition

    will be referred to as interaction. The present study investigates inter-

    action specifically in the area of phonological acquisition.

    Leah Fabiano-SmithState University of New York at New Paltz

    Brian A. GoldsteinTemple University, Philadelphia, PA

    Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010 DAmerican Speech-Language-Hearing Association160

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    2/20

    TransferParadis and Genesee (1996) hypothesized that con-

    sonants and/or vowels that are specific to one language

    will transfer to productions of the other language. This

    is known as segmental transferan example would be

    when a bilingual SpanishEnglish speaking child uses

    the approximant /a/, a sound specific to English, in theproduction of a Spanish word (e.g., carro /karo/ Y

    [kaao]). Transfer has been found to occur in a bi-directional

    mannerthat is, from English to Spanish as well as from

    Spanish to English. Researchers who investigate bi-

    lingualism have argued that bilingual childrens rates

    of transfer may be indicative of interaction between their

    two languages (Keshavarz & Ingram, 2002; Paradis,

    2001). A number of studies have examined this issue.

    For example, in a study on one of their SpanishEnglish

    speaking sons (from ages 1;6 [years;months] to 4;6),

    Schnitzer and Krasinski (1996) found a low frequency

    (i.e., few tokens) of phonological transfer between their

    childs two languages, indicating a low level of between-

    language interaction. Keshavarz and Ingram (2002), in

    their case study of a FarsiEnglish bilingual child, found

    some specific examples of transfer, although it was not

    widespread from either English to Farsi or vice versa.

    Thus, they concluded that the evidenced transfer was in-

    dicative of two language systems that were not entirely

    separate from one another. In another study, Fabiano

    and Goldstein (2004a) compared two typically develop-

    ing (TD) 4-year-old SpanishEnglish speaking children

    (one who was undergoing bilingual first language acqui-

    sition (BFLA) and one who was a sequential bilingual

    child) and two 4-year-old SpanishEnglish speaking chil-dren with phonological disorders (PDs; one who was un-

    dergoing BFLA and one who was a sequential bilingual

    child). Three occurrences of segmental transfer were

    found, with all occurrences being produced by the 2 bi-

    lingual children withphonological disorders. In a follow-up

    study, Fabiano and Goldstein (2004b) examined 2 ad-

    ditional bilingual children undergoing BFLA (one TD

    and one with a PD) following the same methodology as

    discussed in their 2004a study. They found one occur-

    rence of segmental transfer in the child who presented

    with a PD yet no instances of either prosodic or syllabic

    transfer. Paradis (2001) compared 17 FrenchEnglishspeaking children undergoing BFLA, ages 2335 months,

    to 18 French monolingual and 18 English monolingual

    speakers on sound and syllable production. In the bi-

    lingual childrens productions, she found that the stress

    patterns of French influenced initial syllable stress

    patterns of English; however, the bilingual childrens

    stress patterns on weak syllables were identical to the

    patterns of monolingual children. Paradiss (2001) con-

    clusion was that for many of the bilingual children, in-

    teraction was occurring infrequently between their two

    languages.

    Past studies examining transfer in bilingual chil-

    dren indicated that at least some transfer exists from

    one language to the other. Previous studies, however

    with the exception of Paradis (2001)have yet to exam-

    ine patterns of between-language transfer across groups

    of bilingual speakers. A multiple-subject design is needed

    to examine patterns of phonological acquisition across, as

    well as within, bilingual children.

    DecelerationFortheir second hypothesis, deceleration, Paradisand

    Genesee (1996) predicted that aspects of phonological de-

    velopment emerge at a slower rate in bilingual children

    than in monolingual children.1 The premise behind thede-

    celeration hypothesis is that interaction between the two

    languages of bilinguals interferes with acquisition and

    thus results in poorer linguistic skills in bilinguals com-

    pared with monolinguals. Some evidence for this hy-

    pothesis has been found in studies comparing bilingual

    children withmonolingual children in the domains of syn-

    tax and morphology (e.g., Swain, 1972; Vihman, 1982)

    and, more recently, phonology (Gildersleeve-Neumann,

    Kester, Davis, & Pea, 2008).

    Gildersleeve-Neumann et al. (2008) examined the

    English phonological skills of TD, bilingual English

    Spanish speaking 3-year-olds. The results revealed that

    although the bilingual children exhibited phonetic in-

    ventories in English that fell within the normal range for

    their monolingual English-speaking peers, they also dem-

    onstrated lower accuracy than monolinguals. Therefore, as

    compared with their English-speaking peers, the bilingual

    children in their study demonstrated a slower rate ofdevelopment, as measured by accuracy of production.

    Goldstein and Washington (2001) compared 4-year-old

    SpanishEnglish bilinguals to their age-matched mono-

    lingual peers in both Spanish and English. The bilingual

    children in their study were much less accurate than

    monolingual speakers on three Spanish sound classes

    (spirants, flap, and trill) but exhibited skills within the

    normal range formonolingual speakers on all other sound

    classes. Therefore, there is some evidence in previous

    work that the rate of acquisition for bilinguals might be

    slower when compared with their monolingual counter-

    parts, at least on measures of accuracy.

    AccelerationFor their third hypothesis, Paradis and Genesee

    (1996) posit that certain properties in the grammar2 of

    1For their second hypothesis, Paradis and Genesee (1996) used the term

    delayto refer to cases in which language abilities acquired by monolingual

    children are acquired at a later age by bilingual children. Although Paradis

    andGenesee (1996)did notintend to suggestimpairment in their discussionof

    bilingual acquisition, the termdelaymight connote some level of impairment.

    Thus, here, thedelayhypothesis is termed deceleration.2The termgrammaris used here in the broad sense to include phonology.

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 161

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    3/20

    bilingual children emerge at a faster rate in bilingual

    children relative to age-matched monolinguals. The prem-

    ise behind the acceleration hypothesis is that interaction

    between the two languages of bilingual children aids in

    the acquisition process and thus results in superior lin-

    guistic skills in bilinguals compared with monolinguals.

    A number of studies have shown support for this hypoth-

    esis as well. Specifically, Tracy (1995) and Gawlitzek-Maiwald and Tracy (1996) examined GermanEnglish

    bilingual children and observed that certain structural

    properties of German allow for faster acquisition of some

    structural properties in English. For example, they found

    that one of the bilingual children in their study, Hannah,

    acquired English infinitival phrase structure at a faster

    rate than did age-matched monolingual English-speaking

    children. Similarly, Kehoe, Trujillo, and Lle (2001) and

    Lle, Kuchenbrandt, Kehoe, and Trujillo (2003) exam-

    ined interaction between the two languages of Spanish

    German bilingual children. They found that acquisition

    of German resulted in faster acquisition of coda conso-nants in the Spanish productions of bilingual children as

    compared with theirmonolingual Spanish-speakingpeers.

    The authors concluded that interaction between German

    and Spanish led to acceleration in bilingual phonological

    acquisition. These findings demonstrated that bilingual

    children may, at times, show a faster rate of acquisition

    for some language constructs when compared with their

    monolingual peers.

    A Variation of Acceleration

    Previous studies have found evidence of transfer,deceleration, and acceleration in bilingual acquisition.

    However, there is no hypothesis that accounts for re-

    sults of studies that have found bilingual children to

    demonstrate rates of acquisition that fall within the

    normal range for monolingual speakers of both languages

    (DeHouwer, 1995; Goldstein, Fabiano, & Washington,

    2005; Goldstein & Washington, 2001 [for phonology];

    Nicoladis, 1994; Padilla & Liebman, 1975 [for morpho-

    syntax]; Paradis & Genesee, 1996). Demonstrating such

    a rate of acquisition at a given point in development

    might be indicative of interaction between a bilingual

    childs two languages. Perhaps one language might beaiding in the acquisition of the other, allowing for a rate

    of acquisition in bilinguals that falls within the normal

    range for monolingual children of the same age. This

    possibility could provide evidence for a variation of the

    acceleration hypothesis.

    For instance, Paradis and Genesee (1996) examined

    the syntactic skills of three bilingual FrenchEnglish

    speaking children undergoing BFLA (ages 1;11-2;2). Spe-

    cifically, the authors examined functional categories,

    as this construct is acquired earlier in French than in

    English. They found that the bilingual children in their

    study were acquiring the targeted aspect of syntax at the

    same rate as that of their monolingual peers. In another

    study, Goldstein and Washington (2001) examined the

    English and Spanish phonological skills of TD, 4-year-

    old bilingual children and found no significant differ-

    ences between bilingual and monolingual children on

    percent correct consonants, percent correct consonants

    for sound classes, or percentage of occurrence of pho-nological patterns. The results of their study indicated

    that the overall phonological skills of bilingual 4-year-

    olds were similar to those of monolingual children in

    both languages. Similarly, Goldstein et al. (2005) exam-

    ined the phonological skills of TD 5-year-old bilingual

    SpanishEnglish speaking children and compared them

    with their age-matched monolingual peers. The authors

    found no significant difference between bilingual and

    monolingual children on segmental accuracy, syllabic

    accuracy, or percentage of occurrence of phonological

    patterns. These findings indicate that bilingual children

    are maintaining a rate of acquisition that is similar tothat of their monolingual peers. Thus, the relative load

    of two inputs is not resulting in acquisition for bilinguals

    that occurs at a slower rate, nor is one language aiding

    the other to the extent that acquisition is occurring at a

    faster rate. Rather, interaction between the two languages

    of bilingual children leads to a rate of development that

    is within the normal range of their monolingual peers.

    It is important to highlight that bilingual children

    can exhibit characteristics of transfer, deceleration, and

    a variation of acceleration simultaneously. For exam-

    ple, Fabiano-Smith and Barlow (2009) examined 8 bilin-

    gual Spanish

    English speaking children (ages 3;0-4;0)and compared them to their monolingual, age-matched

    peers. An accuracy analysis and a typological analysis of

    phonetic inventories were performed in order to identify

    instances of transfer, deceleration, and acceleration. They

    found that even though the bilingual children demon-

    strated lower accuracy of production when compared

    with monolinguals, their phonetic inventories were just

    as complex in both languagesas the phonetic inventories

    of monolingual speakers of either language. The bilin-

    gual children acquired two complex phonetic inventories

    at the same rate that the monolingual children acquired

    only one (i.e., a variation of acceleration). The authorsalso found instances of transfer in the productions of

    many of the bilingual children. These results indicate

    that examining one phonological construct may provide

    evidence for deceleration, but examining other phono-

    logical constructs may provide evidence for a variation

    of acceleration and/or transfer. Thus, it is possible for

    transfer, deceleration, and a variation of acceleration to

    exist in development simultaneously.

    It is necessary, then, to determine what bilingual

    children implicitly knowabout the sounds in the two

    languages they are acquiring (Watson, 1991). It may be

    162 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    4/20

    possible to provide evidence for Paradis and Genesees

    (1996) hypotheses by determining how accurately bi-

    lingual SpanishEnglish speaking children produce

    sounds that are phonetically similar between their two

    languages (referred to here as shared sounds) versus

    sounds that are phonetically dissimilar and language spe-

    cific (referred to here as unsharedsounds; see Table 1).

    The examination of shared and unshared sounds couldalso provide evidence for how bilingual children achieve

    rates of acquisition that fall within the normal range

    for monolingual children (i.e., a variation of the accel-

    eration hypothesis).

    Phonetic SimilarityOur notion of shared versus unshared sound accu-

    racy is adapted from the literature on the relevance of

    phonetic similarity on second language (L2) phonolog-

    ical acquisition (Flege, 1981, 1987). L2 learners have

    little difficulty in the perceptual categorization of second-

    language sounds that are phonetically similar to those in

    their native language (L1; Flege, 1981). L2 learners often

    perceive L2 speech sounds in terms of their L1 phonemic

    categories (Flege, 1987). This may account for why speak-

    ers extend the production of already familiar sounds into

    new (L2) phonetic contexts. Thus, there seems to be a sys-

    tematic interaction between the two languages of the L2

    speaker, reminiscent of Paradis and Genesees (1996)

    notionof interaction between the two language systemsof

    bilingual children.

    Fleges (1981) notion of sound categorization in adult

    L2 acquisition could be occurring in bilingual phonolog-ical acquisition in a similar way. Just as Flege posits for

    L2 learners, we hypothesize that bilingual children per-

    ceive phonetically similar sounds as common between

    their two languages and categorize them into the same

    phonemic category despite their fine phonetic distinc-

    tions (e.g., aspiration on stops). The result of this cate-

    gorization, according to our prediction, is that sounds

    that are phonetically similar between the two languages

    are more quickly accessed and therefore extended intothe phonetic contexts of both languages. Bilingual chil-

    dren would then have more production experience with

    these phonetically similar (i.e., shared) sounds than with

    phonetically dissimilar (i.e., unshared) sounds, thereby

    influencing accuracy. Bilingual children may reduce the

    use of allophonic variants of a shared phoneme, but the

    essential features of that phoneme remain intact (i.e.,

    voicing, place, and manner). That is, another phoneme

    is not substituted for the target sound, which would

    thereby result in a production error. Rather, bilingual

    children may use the same sound in both language con-

    texts because doing so does not interfere with the sounds

    ability to mark distinctions between words (e.g., /p/ in

    both English and Spanish carries the same function).

    This, in turn, may lead to an increase in rate of ac-

    quisition of these phonetically similar sounds. Bilingual

    children could be using one sound in productions of both

    languages due to interaction between their two lan-

    guages. The interaction occurring between the bilingual

    childs two languages that allows for quicker access of

    phonetically similar sounds could lead to more experi-

    ence in production and an increased rate of acquisition

    (i.e., a variation of acceleration that allows bilinguals to

    demonstrate a rate of acquisition that falls within thenormal range of monolinguals).

    Table 1.Shared and unshared sounds between English and Spanish.

    Sound classes Shared sounds Unshared sounds specific to EnglishUnshared sounds

    specific to Spanish

    Plosives /p/, /b/, /t/, /d/, /k/, /g/Nasals /m/, /n/ nasal /:/ //Fricatives /f/, /s/, /8/* /v/, /Z/, /z/, /S/, /q/, /h/Spirants* [b], [S]Affricate /^/ /u/

    Lateral liquid /l/Nonlateral liquid /a/Glides /w/, /j/**Flap /R/Trill /r/

    *The voiced interdental fricative /8/ is phonemic in English but allophonic in Spanish. It is being analyzed assharedbecause phonologists are divided over which group of soundsspirants or voiced stopsis theunderlying form. For example, Barlow (2003) argues that the spirant [8] in Spanish may actually be theunderlying form and not the phonetic realization of /d/; thus, it was included in the shared sound category.**Glides were analyzed as part of the vowel nucleus for Spanish (Harris, 1993); thus, they were not included inanalyses of accuracy for either English or Spanish.

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 163

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    5/20

    This concept of sound categorization for L2 speakers

    has been generalized into the context of bilingual chil-

    dren in previous studies examining the accuracy of shared

    and unshared sounds. Goldstein, Fabiano, and Iglesias

    (2003) examined 5 TD sequential bilingual Spanish

    English speaking children (children whose L1 is Spanish

    and began exposure to English at age 3;0) and 5 sequen-

    tial bilingual children with PD (ages 4;0-7;0).They foundhigher accuracy (a) on shared sounds and syllable types

    than on unshared sounds and syllable types and (b) on

    manner classes that possessed more shared sounds than

    on mannerclasses that possessed more unshared sounds.

    In addition, high accuracy on shared sounds was found

    independent of the effect of developmental sequence

    that is, bilingual children were not more accurate on

    shared sounds simply because these sounds tended to be

    acquired earlier than unshared sounds. This result was

    found across the two languages and language ability

    groups.

    Frequency of Occurrenceof Shared Sounds

    It is possible that, as with L2 speakers, bilingual

    children are perceiving equivalence (Flege, 1981) for

    overlapping phonological properties of their two lan-

    guages and thus are establishing a single phonemic cat-

    egory for particular sounds that occur in both languages

    (i.e., shared sounds). At the same time, they may main-

    tain separate phonemic categories for those that only

    occur in one language (i.e., unshared sounds). That is,

    bilingual children may be more accurate on sharedsounds because these sounds are used in both language

    contexts. This differential accuracy between shared and

    unshared sounds may be evidence for and may account

    for how bilingual children demonstrate rates of acquisi-

    tion that are within the normal range for monolingual

    peers in both languages. Sound accuracy, however, might

    be predicted by frequency of occurrence of shared sounds.

    Although the influence of developmental sequence did

    not influence high accuracy of shared sound production

    (Goldstein et al., 2003), it is possible that frequency of

    occurrence of shared sounds does serve as a predictor.

    The investigation of frequency was incorporatedinto the current study in an attempt to examine the

    effect of frequency of occurrence of shared sounds on the

    accuracy of production of shared sounds. Sounds that

    are frequently occurring in a language are thought to be

    highly unmarked sounds (Cysouw, 2003). High frequency,

    in other words, is related to low complexity (Greenberg,

    1966; Jakobson, 1941, 1968; Trubetzkoy, 1939). Previous

    studies have found that frequently occurring sounds are

    produced with higher accuracy than sounds that are sig-

    nificantly less frequent. Indeed, Kirk and Demuth (2003)

    reported that English-learning children acquired the

    more frequent coda clusters before the less frequent on-

    set clusters. In addition, high frequency stop + /s, z/ clus-

    ters were acquired before other, less frequent phoneme

    combinations.

    At issue is whether high accuracy of shared sounds

    in bilingual speakers is a result of frequency of occur-

    rence of those sounds in a language, given that speech

    sound accuracy and frequency are related, at least cross-linguistically (Demuth, 2001; Levelt, 2000; Roark &

    Demuth, 2000; Stemberger & MacWhinney, 1986).

    Sounds in the shared category could be produced more

    accurately than sounds in the unshared category simply

    because they occur more frequently in thebilingualstwo

    languages and not because of interaction between the

    bilinguals two languages. If frequency of occurrence is a

    positive predictor of accuracy, then it will not be neces-

    sary to invoke notions of interaction to account for the

    more accurate productions of shared over unshared

    sounds. Thus, the predictive capability of frequency of

    occurrence of sounds in each language on accuracy ofshared sounds was examined in the present study.

    PurposeThe purpose of this study was to determine how

    between-language interaction contributes to phonolog-

    ical acquisition in bilingual SpanishEnglish speak-

    ing children. Three research questions motivated this

    investigation.

    Research Question 1: Will bilingual children demon-strate evidence of interaction between their two languages

    (i.e., transfer, deceleration, and acceleration) as predicted

    by Paradis and Genesee (1996)?The type and frequency

    of segmental transfer between the two languages of these

    bilingual children will illustrate if, and to what extent,

    this type of interaction occurs. Furthermore, lower ac-

    curacy in the productions of bilinguals compared with

    monolinguals will provide evidence for a slower rate of

    acquisition in bilingual children (i.e., deceleration). Con-

    versely, significantly higher accuracy in the productions

    of bilinguals as compared to monolinguals will provide

    evidence for a faster rate of acquisition in bilingual chil-

    dren (i.e., acceleration).

    Research Question 2: Will bilingual children demon-

    strate evidence for a variation of the hypothesis of accel-

    eration due to phonetic similarity (Flege, 1981)? Higher

    accuracy of bilingual childrens productions on shared

    sounds than on unshared sounds, paired with the ab-

    sence of this pattern in monolingual children, could pro-

    vide evidence for between-language interaction and a

    variation of the hypothesis of acceleration.

    Research Question 3: Does sound frequency of oc-

    currence predict high accuracy of shared sounds in the

    164 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    6/20

    productions of bilinguals? Frequency of occurrence of

    sounds in a language could motivate high accuracy of

    shared sounds. If frequency of occurrence predicts high

    accuracy of shared sounds, then between-language in-

    teraction could not account for the higher accuracy of

    shared sounds as compared with unshared sounds.

    MethodParticipants

    Twenty-four TD children (ages 3;04;0) were included

    in the present study. Children in the United States who

    participated in the study were enrolled in a Head Start

    program for which there is an income requirement; thus,

    socioeconomic status (SES) was controlled for, despite the

    fact that maternal education differed. Six of the 8 children

    recorded in Mexico were from low-income households; the

    2 remaining participants were the children of university

    instructors (see the Results section for further details on

    the childrens SES and phonological skills).

    The children were categorized into three groups

    based on language history: (a) eight bilingual Spanish

    English speaking children (mean age = 3;6; range =

    3;04;0); (b) eight monolingual Spanish speakers (mean

    age = 3;4; range = 3;24;0); and (c) eight monolingual

    English speakers (mean age = 3;3; range = 3;03;11). A

    KruskalWallis nonparametric test (Kruskal & Wallis,

    1952) indicated no significant difference between the

    groups based on age, c2(2,N= 24) = 3.55,p = .169. An ex-

    tensive parent and/or teacher report was used to deter-

    mine each child

    s language status (i.e., monolingual orbilingual) and phonological ability (i.e., to ensurethat all

    children were TD with no speech, language, cognitive, or

    neurological deficits).

    Bilingual participants. Demographic characteris-

    tics of the bilingual participants can be found in Table 2.

    The bilingual children were speakers of Puerto Rican

    and Dominican Spanish. This study took place during

    May and June, and the children had been enrolled in a

    bilingual Head Start program since the beginning of the

    school year. Percent input and output were determined

    for both languages through parent report (after Restrepo,

    1998). All bilingual participants received at least 20%

    input in both languages andproduced at least 20%outputin both languages. This criterion follows from previous

    work, which has shown that children need at least 20%

    exposure in order to use the target language (Pearson,

    Fernandez, Lewedeg, & Oller, 1997). The first author

    asked parents to describe his or her childs schedule on a

    typical day. The childs activities, the interlocutor(s) in-

    volved, and the language typically used duringthat activ-

    ity were recorded. The number of hours of exposure to

    each language per day, Monday through Friday, was de-

    termined (i.e., input). The same method was used to de-

    rive the number of hours the child used each language

    duringthe workweek (i.e., output). Similarquestions wereasked regarding the childs weekend schedule. Overall

    percentages of language use were calculated separately

    by multiplying the number of hours of exposure (input)

    or use (output) by 100, then dividing that number by the

    total number of hours in the week. In addition, parents

    rated their childrens proficiency in both English and

    Spanish on a scale from 0 (child could not speak the indi-

    cated language at all) to 4 (child had native-like profi-

    ciency in the language; Pea, Bedore, & Rapazzo, 2003;

    Pea, Bedore, & Zlatic-Giunta, 2002).All childrenincluded

    in the present study were rated as either 3 or 4 by their

    parents in both English and Spanish, indicating native ornear native-like competence in the indicated languages.

    North Philadelphia, where the bilingual data were

    collected, is a bilingual, Puerto Rican community that has

    maintained the use of both English and Spanish for more

    than 40 years. All children living in this community have

    Table 2.Demographic data on bilingual participants.

    ChildID

    CA (M= 3;5)[years;months] Gender

    Motherseducation

    Percent inputSpanish

    Percent outputSpanish

    Percent inputEnglish

    Percent outputEnglish

    ProficiencySpanish

    ProficiencyEnglish

    B01 3;8 Male High school 63 63 37 37 4 3B02 3;5 Male High school 75 75 25 25 4 3B03 3;8 Male Some

    university58 30 42 70 3 3

    B04 3;5 Male Bachelor sdegree

    50 20 50 80 3 4

    B05 3;8 Female High school 40 20 60 80 3 4B06 3;4 Male Some

    university70 20 30 80 3 4

    B07 3;11 Male High school 40 20 60 80 3 4B08 3;5 Male Some

    university80 50 20 50 3 3

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 165

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    7/20

    some English exposure before entering school, even if

    Spanish is theonly language spoken in the home. The na-

    ture of the community and the large amount of English

    input that the children receive within the school system

    could account for rapid acquisition of English upon enter-

    ing preschool. Parent report was also used to determine

    the length of exposure to each language for all bilingual

    speakers. Each child was determined to have had mostlySpanish input and output in the home up to age 3;0, after

    which English exposure began at preschool, or they re-

    ceived exposure to both languages in the home from birth.

    Thus, all bilingual children were categorized as early

    bilinguals(Genesee, Paradis, & Crago, 2004).

    Monolingual participants.Demographic characteris-

    tics of the monolingual Spanish and monolingual English

    participants are found in Table 3. Datafrom 8 monolingual

    Spanish speakers were collected in Quertaro, Mexico,

    and data from 8 monolingual English speakers were

    collected in Philadelphia, Pennsylvania. Children in

    each monolingual group had no input or output in anylanguage other than their native language. To be in-

    cluded in the study, the monolingual children s pro-

    ficiency rating (obtained through parent report) in

    their language had to be either 3 or 4. Monolingual

    children were included in this study in order to obtain

    developmental information for each language and to

    compare the accuracy of sound production and the sub-

    stitution patterns between monolingual and bilingual

    speakers.

    Data CollectionBoth single word and connected speech samples

    were collected from each child for analysis. Single word

    samples were used for phonological analyses. Connected

    speech samples were used to provide background in-

    formation on the childrens language skills. If a sound

    production was not found in the single word sample, con-

    nected speech aided in the phonetic inventory analy-

    sis. Each bilingual child was recorded in Spanish and

    English, and each monolingual child was recorded in

    his or her respective language.

    The phonology subtest of the Bilingual English Span-

    ish Assessment (BESA; Pea, Gutirrez-Clellen, Iglesias,

    Goldstein, & Bedore, 2005), was used to elicit sounds

    in single words. The assessment contains 31 separate

    target items for English and 28 separate target items

    for Spanish. This assessment has been used previously

    with bilingual children (e.g., Goldstein & Washington,

    2001; Goldstein et al., 2005). Each target item was elic-

    ited via a spontaneous label made in reference to a

    photograph. If the child did not label the photograph

    spontaneously, the function of the item was provided to

    thechild. If thechildstill did not label theitem, delayed

    imitation was used because of the negligible difference

    between spontaneous and imitative forms (Goldstein,

    Fabiano, & Iglesias, 2004).

    A pre-determined set of toys whose names targeted

    English andSpanish consonant soundswas used to elicit

    a conversational speech sample from each bilingual child

    in English and Spanish and from each monolingual child

    in English or Spanish. Each single word and connected

    speech sample was recorded using The Presenter wire-

    less lapel microphone, transmitter (Model T1-CL), and

    receiver (Model T3-CL; Shure, Inc., Niles, IL) with inputinto a Dell Latitude 100L computer using a Creative

    Labs Sound Blaster Audigy 2-Z5, 24-bit sound card.

    Table 3.Demographic data on monolingual participants.

    Child ID CA Language Gender Mother s education Parent concern Clinician concern

    S 01 4;0 Spanish Male Some high school No NoS 02 3;3 Spanish Male Master s degree No NoS 03 3;3 Spanish Male High school diploma No NoS 04 3;10 Spanish Male Some high school No No

    S 05 3;2 Spanish Female Bachelor s degree No NoS 06 3;6 Spanish Female Master s degree No NoS 07 3;4 Spanish Female Some high school No NoS 08 3;4 Spanish Male Some high school No NoE 01 3;3 English Female Master s degree No NoE 02 3;1 English Female Medical degree No NoE 03 3;11 English Female Master s degree No NoE 04 3;8 English Male High school diploma No NoE 05 3;0 English Male Bachelor s degree No NoE 06 3;0 English Male Bachelor s degree No NoE 07 3;1 English Male Bachelor s degree No NoE 08 3;7 English Male Master s degree No No

    166 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    8/20

    AnalysesData from the single word samples were phoneti-

    cally transcribed with diacritics using the International

    Phonetic Alphabet (IPA). Dialectal features of Puerto

    Rican, Mexican, and Dominican Spanish (e.g., in Span-

    ish, [x] for /r/ in perro) and English (e.g., [n] for /:/ in

    going

    ) consonants were taken into account and were

    not scored as errors.

    Reliability of TranscriptionSingle word samples. Reliability of transcription

    was performed on the single word samples between two

    primary transcribers (bilingual graduate students in

    speech-language pathology [SLP] at Temple University,

    Philadelphia, Pennsylvana) and the first author of this

    study, who is an EnglishSpanish bilingual. Both the

    first author and the bilingual graduate students were

    trained in narrow transcription using the IPA. The twobilingual graduate students phonetically transcribed all

    of the single word samples. The first author then per-

    formed interjudge reliability on those transcriptions; in

    addition, the original transcribers performed intrajudge

    reliability on their own transcriptions. Since three judges

    were involved in the reliability process, when a disagree-

    ment occurred between two of the judges, the third judge

    was called in to make a decision. The decisions made by

    the third judge were accepted in the final transcriptions.

    Intrajudge and interjudge reliability of IPA narrow tran-

    scription was calculated for 100% of the Spanish and

    English target words on the single word assessment forall of the children.

    For the Spanish monolingual group, reliability

    reached 99.16% and 98.74% for intra- and interjudge

    reliability, respectively, for the single word samples. For

    the English monolingual group, reliability reached 98.7%

    and 96.94% for intra- and interjudge reliability, respec-

    tively. For the Spanish samples of the bilingual chil-

    dren, reliability reached 99.14% and 97.48% for intra- and

    interjudge reliability, respectively. Finally, for the En-

    glish samplesof the bilingual children, reliability reached

    98.61% and 95.67% for intra- and interjudge reliability,

    respectively.

    Because the reliability rates were higher than those

    reported in similar studies, we conducted an additional

    reliability check on 10% of the sample to be certain that

    our estimateswere correct. A bilingual SpanishEnglish

    speaking student in SLP at San Diego State University

    (who was unfamiliar with this particular single word

    probe and who was trained in IPA narrow transcription)

    performedinterjudge reliability. Reliability of transcrip-tion performed by this additional judge reached 98.44%

    for the monolingual Spanish samples (10 disagreements

    out of 640 opportunities), 99.54% for the monolingual

    English samples (3 disagreements out of 651 opportuni-

    ties), 98.59% for the Spanish productions of bilinguals

    (9 disagreements out of 634 opportunities), and 98.47%

    for the English productions of bilinguals (10 disagree-

    ments out of 650 opportunities).

    Connected speech samples. The connected speech

    samples were used to provide background information

    on each childs language skills and to obtain three oral

    language measures: mean length of utterance in mor-phemes (MLUm); number of different words (NDW),

    and total number of utterances (TNU; see Table 4). Data

    from the connected speech samples were transcribed or-

    thographically into communication units (C-units). The

    language transcripts were analyzed using the System-

    atic Analysis of Language Transcripts (SALT [Beta V9

    program]; Miller & Iglesias, 2006) to obtain grammat-

    icality measures. Interjudge reliability of orthographic

    transcription was performed between a bilingual SLP

    graduate student and the first author. Interjudge relia-

    bility was calculated on 100% of the utterances in the

    English and Spanish connected speech samples for allchildren. Reliability of transcription reached 93.67% for

    the monolingual Spanish speakers, 98.27% for the mono-

    lingual English speakers, 98.06% for the Spanish produc-

    tions of bilinguals, and 96.39% for the English productions

    of bilinguals.

    The results from analyses examining the childrens

    connected speech samples showed that their oral lan-

    guage skills were commensurate with published norms

    for monolingual English (Bland-Stewart & Fitzgerald,

    2001) and monolingual Spanish speakers (Bedore, 2005).

    The results are also commensurate with norms accom-

    panying the Bilingual Spanish

    English SALT (Miller &

    Table 4. Means and standard deviations for grammatical characteristics of monolingual and bilingual children.

    Language GroupsMean length of

    utterance (MLUm)Number of different

    words (NDW)Total number ofutterances (TNU)

    Monolingual Spanish speakers 2.38 (SD= 0.61) 91.25 (SD= 41.75) 100.37 (SD= 53.07)Spanish productions of bilinguals 2.25 (SD= 0.62) 103.75 (SD= 46.27) 135.25 (SD= 65.13)Monolingual English speakers 3.57 (SD= 0.75) 221.62 (SD= 58.49) 216 (SD= 68.27)English productions of bilinguals 2.70 (SD= 0.91) 123.62 (SD= 58.47) 144.12 (SD= 61.09)

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 167

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    9/20

    Iglesias, 2006) program for both Spanish and English

    productions of the bilingual children. All children scored in

    the typical range for grammatical characteristics for their

    age and language status (i.e., monolingual or bilingual).

    Phonetic Analyses

    Analyses of the single word samples for all threegroupsmonolingual Spanish-speakers, monolingual

    English speakers, and bilingual SpanishEnglish

    speakerswere performed using Logical International

    Phonetic Programs (LIPP; Oller & Delgado, 2000) to de-

    termine the phonological skills of each child. A phonetic

    analysis was performed on the childrens productions of

    segments overall. Specifically, phonetic inventories were

    constructed to determine what sounds each child had

    acquired. If a child produced a particular sound two or

    more times, regardless of whether it occurred as a cor-

    rect production in a word or as a substitute in another

    word, that child was said to possess that sound in his or

    her inventory (Elbert & Gierut, 1986). If a target did not

    occur at least twice, that sound was not included as part

    of the childs inventory.

    A substitution error analysis was performed on the

    productions of the bilingual speakers to determine what

    sounds the children were using as substitutes. Given our

    interest in transfer, if bilingual children used language-

    specific soundsin the productions of their other language,

    those substitutions were not counted as errors. To that

    end, two analyses were completed (see next paragraph).

    Context-free inventories were obtained from LIPP

    for each child, in each language. Each sound target, thenumber of times that target occurred, and the sounds

    produced for that target were listed for each inventory.

    Both correct productions and substitution errors were

    recorded. First, an analysis determining the quantity

    (i.e., frequency) of phonological transfer was completed.

    Instances of phonological transfer were recorded when

    an unshared, or language-specific, sound appeared in

    the other language (e.g., the Spanish /r/ substituting for

    the English liquid /a/ in an English production) or when

    a fine phonetic distinction specific to one language was

    found in the production of the other language (e.g., aspi-

    ration of /p/ in a Spanish production). Next, an analysis

    examining the quality (i.e., type) of transfer was per-

    formed. Instances of transfer were examined for (a) over-

    all patterns or preferences that the children were using

    (i.e., whether they used certain language-specific sounds

    as substitutes and not others), (b) the language-specific

    sounds used as substitutes, and (c) the direction of trans-

    fer (i.e., Spanish into English or vice versa).

    Relational AnalysesOverall accuracy and accuracy by manner class.

    Through LIPP, measures of overall percentage of

    consonants correct (PCC; Shriberg, Austin, Lewis,

    McSweeney, & Wilson, 1997) and PCC for each manner

    class were obtained. The following manner classes were

    analyzed: stops, nasals, fricatives,affricates, liquids, glides,

    flap (Spanish only), and trill (Spanish only). Overall ac-

    curacy and accuracy of sounds organized by manner class

    were calculated for each language group (i.e., bilingual

    Spanish productions,bilingual English productions, mono-lingual Spanish productions, and monolingual English

    productions) to determine if bilingual children were

    demonstrating evidence of deceleration and/or a varia-

    tion of acceleration.

    Accuracy of shared and unsharedsounds. A separate

    analysis was performed to examine accuracy of shared

    and unshared sounds. For this analysis,the LIPP output

    of accuracy for each individual sound was organized into

    shared and unshared categories for English and Span-

    ish. Mean accuracy and standard deviations were then

    calculated for these categories. The productions of both

    bilinguals and monolinguals were analyzed in the sameway to determineif differential accuracybetween shared

    and unshared sounds was (a) characteristic of phono-

    logical skills in bilinguals or (b) characteristic of all chil-

    drens phonological skills, regardless of language status

    (i.e., bilingual or monolingual).

    For these analyses, shared and unshared sounds

    were categorized at the phonemic rather than phonetic

    level, as described previously in the section on phonetic

    similarity (Flege, 1981, 1987). To determine if the differ-

    ence between the accuracy of shared and unshared sounds

    was significant, a nonparametric MannWhitney test

    (Mann & Whitney, 1947) was performed for each of thethree groups of speakers using the following related

    samples: (a) PCC of shared sounds and (b) PCC of un-

    shared sounds. These analyses were performed on En-

    glish and Spanish samples separately.

    Analysis of FrequencyA mixed-effects regression analysis was performed

    to determine if the frequency of occurrence of sounds in

    each language had predictive capability on the accuracy

    of shared sounds. This analysis was performed separately

    on English and Spanish data. The first step in this se-ries of analyses was to obtain descriptive information for

    each participant, thus means and standard deviationsfor

    PCC were obtained. A nonparametric alternative to the

    analysis of variance (ANOVA), the KruskalWallis test

    (Kruskal & Wallis, 1952; and a post hoc Tamahane T2, if

    necessary), examining PCC by participant was then

    performed to determine if the data from all 8 bilingual

    participants were similar and could be collapsed. Fre-

    quency of occurrence values were taken from Shriberg

    and Kent (1995) for English and Wilson (1984) for Span-

    ish, which were used as the frequency measure in the

    168 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    10/20

    statistical analyses. The term frequency here refers to

    token frequency, or thenumber of times thesound occurs

    in adult productions, representing the childrens input.

    The studies from which our frequency values were ob-

    tained assigned percent occurrence values to each sound

    to represent that sounds frequency in the language. For

    example, /s/ was assigned a frequency value of 7.88 in

    English and a value of 16.69 in Spanish. The higher thevalue, the more frequent the sound in that language.

    Finally, to determine if frequency is a significant pre-

    dictor of PCC, a general model was used to analyze the

    data through a univariate ANOVA. Because the data for

    these analyses were taken from the single word test, lim-

    ited opportunities were available for some low-frequency

    sounds (e.g., /8/ and / /). Therefore, the data were ana-

    lyzed in the mixed model whilerestricting the analyses to

    only sounds that had six or more opportunities for pro-

    duction. High-frequency and low-frequency sounds were

    equally represented in the analysis. English frequency

    values ranged from 3.07 to 11.49 (x = 7.30,SD = 3.08), andSpanish frequency values ranged from 3.85 to 16.69 (x=

    9.85,SD= 4.51).

    The outcome measure was PCC at the sound level.

    Frequency was the independent measure, and partici-

    pant was the random effect variable. The model took

    into account the correlations within each participant.

    Model assumptions were checked, and no transforma-

    tions were necessary. By taking into consideration the

    relationships between all of the variables, the predictive

    capability of frequency of occurrence of sounds on the ac-curacy of shared sounds can be interpreted more accu-

    rately than with a single correlation.

    ResultsPhonetic Analyses

    Consonant inventories. Results of the analysis ex-

    amining consonant inventories can be found in Table 5.

    The independent analysis showed that all bilingual chil-

    dren demonstrated phonetic inventories typical of their

    chronological age. Many of the bilingual children didnot produce later-developing Spanish sounds such as the

    flap, trill, and spirant [8], which is typical for monolingual

    Table 5.Sounds not included in the phonetic inventories of the children.

    Child ID Produced only once Not produced at all

    Monolingual Spanish speakersS 06 /, x/S 07 /g,, r/

    Monolingual English speakersE 01 /8/ /q/E 02 /v/ /8/E 03 /S/E 04 /q/ /8, u/E 05 /q, 8, a/E 06 /8, u/E 07 /q, 8/E 08 /q/

    Bilingual speakers: Spanish productionsB 02 /x/, [8, S], clustersB 03 //, [b], [S] /x/, [8], /r/B 04 [S] /r/B 05 // /r/

    B 06 /x/ [8

    ], /r/B 07 [b] [8], /R, r/B 08 //, /r/

    Bilingual speakers: English productionsB 01 /^, v, z, 8, j/ /:/B 02 /q, 8, u/B 03 /q, u, 8, :/B 04 /z, u/ /8, v/B 05 /8, v/B 06 /S, u/ /v/B 07 /S, u/ /q/B 08 /j/ /q, z, 8/

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 169

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    11/20

    Spanish speakers of this age (Acevedo, 1993). In English,

    the bilingual group followed a similar pattern as that

    of monolingual English speakers, as many did not pro-

    duce later-developing sounds such as /q/ and /8/ (Shriberg,

    1993).

    Transfer. Results of the analysis examining transfercan be found in Table 6. Two of the 8 bilingual children

    (25%) produced instances of transfer (nine instances

    between the 2 children). They tended to modify low-level

    phonetic rules such as the production of an English as-

    pirated stop phoneme as unaspirated (e.g., /khA/Y

    [ k=A]), as is the rule in Spanish.

    Consonant AccuracyResults of the analyses examining consonant accu-

    racy can be found in Tables 7 and 8. Results showed that

    mean overall PCC was greater than 70% for all groups.

    Overall accuracy and accuracy by manner class. The

    results for the analysis examining PCC by manner class

    can be found in Table 7. To determine if bilingual chil-

    dren demonstrated deceleration, acceleration, and/or a

    variation of acceleration, a MannWhitney test was per-

    formed using overall PCC as the dependent measure.

    Language group (i.e., monolingual vs. bilingual) served

    as the independent measure. Results indicated thatmono-

    lingual children were significantly more accurate than

    bilingual children for Spanish (z = 1.99, p = .046) butnot for English (z= 1.73,p = .083). These findings are

    suggestive of deceleration in the Spanish productions of

    bilinguals. Subsequently, the accuracy of the bilingual

    children was compared with the accuracy of monolingual

    children from a low SES to eliminate any potential con-

    founds (see Figure 1). It was determined that consonant

    accuracy was lower in Spanish for all bilingual children

    compared with the most accurate monolingual Spanish-

    speaking child from a low SES background. Furthermore

    consonant accuracy was lower for 5 of the 8 bilingual

    children in English compared with the most accurate

    monolingual English-speaking child from a low SES back-ground. Overall, bilingual children demonstrated lower

    consonant accuracy than their monolingual peers from low

    SES backgrounds, suggesting deceleration in bilingual

    acquisition (Paradis & Genesee, 1996).

    Table 6.Instances of transfer in the productions of bilingual SpanishEnglish speaking children.

    Child ID Language context Target Production Substitute used Number of occurrences

    B 06 Spanish /negRo/ [negao] [a] 1Spanish /bisikleta/ [bikisikleqa] [q] 1English /t helfon/ [t =elfon] Unaspirated voiceless stop 1English /k ha/ [k=a] Unaspirated voiceless stop 1

    English /t host/ [t =ost] Unaspirated voiceless stop 1B 08 Spanish /seoR/ [toR] [] 1

    Spanish /gajeta/ [gaZeZa] [Z] 2English /k ha/ [k=a] Unaspirated voiceless stop 1

    TOTAL 9

    Table 7.Means and standard deviations for percentage of consonants correct (PCC) by manner class.

    LanguageGroups Stops Nasals Fricatives Affricates Liquids Glides Flap Trill

    MonolingualSpanish

    speakers

    82.54

    (SD= 7.53)

    91.80

    (SD= 7.8)

    79.75

    (SD= 9.1)

    81.25

    (SD= 37.2)

    68.75

    (SD= 29.88)

    100

    (SD= 0)

    33.33

    (SD= 17.81)

    37.5

    (SD= 34.21)Monolingual

    Englishspeakers

    88.35(SD= 6.23)

    91.91(SD= 6.98)

    84.43(SD= 12.85)

    75(SD= 34.5)

    70.79(SD= 25.64)

    87.5(SD= 24.8) N/A N/A

    Spanishproductionsof bilingualspeakers

    77.40(SD= 9.51)

    94.44(SD= 8.39)

    66.47(SD= 13.22)

    100(SD= 0)

    66.98(SD= 25.13)

    81.25(SD= 21.67)

    25.09(SD= 18.9)

    4.1(SD= 11.78)

    Englishproductionsof bilingualspeakers

    73.31(SD= 14.05)

    82.35(SD= 13.33)

    63.45(SD= 16.29)

    79.16(SD= 24.8)

    67.5(SD= 23.45)

    83.33(SD= 17.81) N/A N/A

    170 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    12/20

    Table 8.Overall PCC, and PCC for accuracy of shared and unshared sounds.*

    Monolingual Spanish speakers

    Child ID Overall PCC PCC shared PCC unshared

    S 01 78.16 84.21 65.21S 02 73.26 82.0 39.13S 03 68.24 69.64 56.52S 04 82.35 84.21 73.91S 05 68.24 82.45 30.43S 06 80.0 91.37 47.82S 07 72.94 82.45 47.82S 08 69.51 78.18 47.82M (SD) 75.58 (5.49) 81.69 (6.15) 51.08 (13.89)

    Monolingual English speakers

    Child ID Overall PCC PCC shared PCC unshared

    E 01 77.14 87.83 75.00E 02 76.42 80.24 80.00E 03 98.11 100 55.00E 04 75.47 82.71 90.00E 05 84.76 83.75 90.00E 06 81.90 83.75 95.00E 07 93.33 93.75 50.00E 08 85.71 90.00 68.18M (SD) 84.10 (8.20) 87.75 (6.60) 75.39 (16.67)

    Bilingual speakers (Spanish PCC data)

    Child ID Overall PCCSpanish PCC sharedSpanish PCC unsharedSpanish

    B 01 76.54 86.79 47.82

    B 02 66.27 75.00 33.33B 03 58.14 66.66 23.80B 04 69.77 84.74 31.81B 05 69.77 77.96 40.90B 06 69.41 77.58 50.00B 07 59.3 72.41 26.08B 08 56.98 66.10 31.81M (SD) 65.77 (6.95) 75.90 (7.54) 35.69 (9.62)

    Bilingual speakers (English PCC data)

    Child ID Overall PCCEnglish PCC sharedEnglish PCC unsharedEnglish

    B 01 72.38 68.75 85.00B 02 57.55 70.00 36.36

    B 03 63.81 68.75 50.00B 04 81.90 87.5 70.00B 05 86.67 90.00 75.00B 06 88.57 92.50 85.00B 07 70.48 77.5 50.00B 08 57.14 62.5 45.00M (SD) 72.31 (12.45) 77.18 (11.43) 62.04 (18.99)

    *The mixed effects regression analysis utilized data from those sounds with six or more opportunities for production.This table reflects data for all sounds.

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 171

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    13/20

    To determine if monolingual and bilingual children

    differed in the same way on the accuracy of manner

    classes, a MannWhitney test was performed for each

    manner class, comparing bilingual to monolingual pro-

    ductions. For Spanish, monolinguals demonstrated sig-nificantly higher accuracy than bilingual children on the

    following manner classes: trill (z= 2.43,p= .015), frica-

    tives (z= 1.99,p= .046), and glides (z= 2.21,p= .027).

    For English, monolinguals demonstrated significantly

    higher accuracy than bilingual children on stops (z =

    2.05, p = .04) and fricatives (z = 2.62, p = .00) only.

    Overall, some evidence was found for deceleration in

    the bilinguals; however, in general, monolingual and

    bilingual children differed only on a minority of man-

    ner classes.

    Accuracy of shared and unshared sounds. The dis-

    tinction betweenaccuracy on shared and unshared soundswas subsequently examined in a separate analysis, the

    results of which can be found in Table 8. To determine if

    the difference between shared and unshared sound pro-

    duction was significantly different within each language

    group, a Wilcoxon signed ranks test and effect size anal-

    ysis (Cohen, 1988) were performed for each of the four

    separate sets of data. Effect sizes were calculated using

    Cohens dwith interpretation using the following guide-

    lines extant in the SLP literature (e.g., Fiestas & Pea,

    2004; 00.10 = negligible; 0.100.25 = small; 0.250.50 =

    moderate; 0.500.80 = large; 0.801.00+ = very large).

    For the monolingual Spanish group, a significant

    difference was found between PCC of shared and un-

    shared sounds (z= 2.52,p = .012) with a very large ef-

    fect size (d= 2.86) indicating a large difference between

    the mean PCC of shared and unshared sounds. A sig-nificant difference was also found for the Spanish of bi-

    linguals between PCC of shared and unshared sounds

    (z = 2.52,p = .012) with a very large effect size (d = 4.65).

    For the monolingual English group, no significant differ-

    ence was found between PCC of shared and unshared

    sounds (z= 1.82,p = .069) with a very large effect size

    (d= 0.97). For the English of bilinguals, however, a sig-

    nificant difference was found between PCC of shared

    and unshared sounds (z = 2.10, p = .035) with a very

    large effect size (d = 0.96). Overall, bilingual children

    demonstrated significantly higher accuracy on shared

    sounds than on unshared sounds; however, so did mono-

    lingual Spanish-speaking children. Upon further anal-

    ysis, significant differential accuracy in the monolingual

    Spanish-speaking group was the result of very low accu-

    racy on flap (33.33%) and trill (37.5%) only.

    FrequencyBecause bilinguals (and the monolingual Spanish-

    speakers) demonstrated significantly higher accuracy on

    shared sounds than on unshared sounds, the following

    analyses were performed to determine if the complexity

    Figure 1. Comparison of bilingual children to low socioeconomic status (SES) monolingual children onconsonant accuracy.*

    *Consonant accuracy for each participant is plotted by Child ID. For bilingual children, the inclusion ofEin the Child ID indicates an English production, and the inclusion of Sindicates a Spanish production.

    172 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    14/20

    parameter of frequency of occurrence of sounds in each

    language was motivating high accuracy of shared sounds.

    These analyses were performedseparately on the English

    and Spanish data.

    Spanish productions. The KruskalWallis test per-

    formed on PCC by participant yielded a nonsignificant

    difference (c2 = 6.55,p = .477,df= 7), indicating that the

    participants were not significantly different from one

    another on PCC. In order to assess whether frequency

    was a significant predictor of PCC, a general linear mixed

    model was used with participant as a random effect and

    frequency as a fixed effect. The results of this analysis in-

    dicated that neither frequency,F(5, 27) = 1.32, p = .284,

    nor participant,F(7, 27) = 1.40,p = .243, was a significant

    predictor of accuracy. Thus, frequency of occurrence of

    sounds in Spanish was not a significant predictor of PCC

    in the Spanish productions of bilingual children.

    English productions.The KruskalWallis test per-

    formed on PCC by participant yielded a significant dif-ference between PCC and participant (c2 = 19.68, p =

    .006, df = 7). Because the nonparametric Kruskal

    Wallis does not have a post hoc test associated with it,

    a one-way ANOVA and post hoc test used for unequal

    variancesthe Tamahane T2was used to examine sig-

    nificant differences between participants on PCC. The

    Tamahane T2 test indicated that participants B07 and

    B08 differed significantly on PCC (p = .027). Further

    analysis of these 2 participants indicated that child B07

    demonstrated the highest percent accuracy on produc-

    tion of shared sounds (92.5%), and B08 demonstrated the

    lowest percent accuracy on production of shared sounds(62.5%). Both children demonstratedtypical phonological

    characteristics for English-speaking 3-year-olds; thus,

    data from all 8 children were used for analysis.

    To assess whether frequency of occurrence of sounds

    was a significant predictor of PCC, a general linear mixed

    model was used with participant as a random effect and

    frequency of occurrence as a fixed factor. As in the Span-

    ish analysis, the data were analyzed in the mixed model

    while restricting the analysis to only those sounds that

    had six or more opportunities for production. Frequency

    of occurrence of sounds was not found to be a significant

    predictor of accuracy, F(3, 21) = 1.68, p = .200; however,

    participant was found to be a significant predictor of ac-

    curacy,F(7, 21) = 3.15, p = .019. Therefore, frequency of

    occurrence of sounds in English did not predict high

    accuracy of shared sounds in the English productions of

    bilinguals. The significance of participant was driven by

    the difference in PCC between the child with the highest

    PCC (Participant B07) and the child with the lowest PCC

    (Participant B08). It should be noted, however, that the

    PCC for these children still fell into the typical range for

    3-year-olds.

    DiscussionThe purpose of this study was to examine how in-

    teraction between the two languages of bilinguals con-

    tributes to bilingual phonological acquisition. It was

    predicted that bilingual children would demonstrate in-

    teraction between their two phonological systems, thus

    supporting the hypotheses of transfer, deceleration, ac-celeration (Paradis & Genesee, 1996), and /or a variation

    of the hypothesis of acceleration. Evidence for these hy-

    potheses was examined through (a) transfer, (b) over-

    all consonant accuracy and accuracy by manner class,

    (c) differential accuracy on the production of shared

    and unshared sounds, and (d) the predictive capability

    of frequency of occurrence of sounds in each language

    on the accuracy of shared sounds.

    TransferOverall, 25% of the bilingual children demonstrated

    a low frequency of bi-directional transfer. These results

    demonstrate that for the most part, bilingual children

    maintain separation between their two phonological sys-

    tems. Thus, frequency of transfer provided little evi-

    dence for interaction between the bilingual childrens

    two languages. It seems that these instances of transfer

    are not random errors, however. The 2 bilingual children

    who demonstrated transfer exhibited systematic modi-

    fication of low-level phonetic characteristics. Specifi-

    cally, stop consonants were deaspirated in some English

    productions. Thus, the bilingual children maintained

    separation for the majority of their productions, andbetween-language interaction at this level is rare, sup-

    porting Paradis and Genesee (1996).

    This finding also informs Fleges (1981) notion of

    phonetic similarity in that bilingual children, at times,

    abandon fine phonetic distinctions and use phonetically

    similar sounds in both languages. For example, the shared

    stops /k/ and /t/ were sometimes produced without as-

    piration in English, demonstrating that bilingual children

    can use the Spanish stops /k/ and /t/ without sacrificing

    meaning in their English productions. This finding sup-

    ports Flege (1980), who states, [L2] learners frequently

    produce a range of different phonetic variants (including

    the correct realization) for a single L2 phoneme(p. 117).

    DecelerationThe deceleration hypothesis predicts that the rate

    of phonological development in bilingual children is

    slower than that of monolingual children. Overall con-

    sonant accuracy in Spanish was significantly higher in

    monolinguals than in bilinguals; however, there was

    no significant difference on that construct in English

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 173

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    15/20

    between bilingual and monolingual children. When ex-

    amined by manner class, bilinguals and monolinguals

    differed in accuracy on only a few manner classes. Mono-

    lingual Spanish speakers were significantly more accu-

    rate than bilingual children on their productions of the

    trill, fricatives, and glides. Monolingual English-speaking

    children demonstrated significantly higher accuracy than

    bilingual children on stops and fricatives. Therefore, theseresults provide some evidence for Paradis and Genesees

    (1996) hypothesis of deceleration. Interestingly, the de-

    celeration that occurred was not identical across the two

    languages. Specifically, the bilinguals showed a slower

    rate of acquisition for glides in Spanish but not in En-

    glish when compared with monolinguals. Similarly, bi-

    linguals showed a slower rate of acquisition for stops

    only in English, yet they showed a slower rate of ac-

    quisition for fricatives in both languages. It appears

    that between the ages of 3;0 and 4;0, some, but not all,

    phonological skills in bilingual children are acquired at

    a slightly slower rate than for monolingual children ofthe same age. It is important to note that these findings

    were independent of the influence of SES (see Figure 1).

    The majority of bilingual children demonstrated lower

    consonant accuracy in both languages than their mono-

    lingual peers from low SES backgrounds.

    It is important to point out, however, that the bi-

    lingual children in the present study performed within

    the typical range for their chronological age, even though

    they demonstrated lower accuracy than the monolin-

    guals. In Spanish, 5 of the 8 bilingual children demon-

    strated overall consonant accuracy greater than 66%.

    Jimnez (1987) found that monolingual Spanish-speakingchildren between the ages of 3;0 and 4;0 produce about

    half of their consonants accurately. In English, 5 of the

    8 bilingual children demonstrated overall consonant ac-

    curacy greater than 70%, classifying them in the typical

    range for monolingual English speakers their age (Shriberg

    & Kwiatkowski, 1982). In addition, sounds produced in

    error by these bilingual participants were age appropri-

    ate. In their Spanish productions, these bilingual chil-

    dren demonstrated errors on flap and trill, errors also

    common in monolingual Spanish-speaking preschoolers

    (Acevedo, 1993). In English, these bilingual children

    demonstrated low accuracy on fricatives, which is alsocommon among monolinguals of the same age (Bauman-

    Waengler, 2008). These findings are consistent with pre-

    vious studies examining bilingual language acquisition

    reporting that bilingual children perform within the typ-

    ical range of age-matched monolinguals for grammatical

    development (e.g., DeHouwer, 1990; Nicoladis, 1994;

    Padilla & Liebman, 1975). Overall, lower accuracy in bi-

    lingual production does not indicate delayed or disordered

    acquisition in a clinical sense; rather, at some points in

    development, bilingual children may exhibit a slower rate

    of acquisition; however, that rate still falls within the nor-

    mal range for age-matched monolingual children.

    AccelerationThe hypothesis of acceleration posits that certain

    properties in the grammar of bilingual children emerge

    at a faster rate than for monolingual children of the sameage. The results of this study did not provide evidence for

    Paradis and Genesees (1996) hypothesis of acceleration

    that is, these bilinguals did not demonstrate a faster rate

    of acquisition when compared with monolinguals on over-

    all accuracy or accuracy by manner class. The findings

    of this study support previous studies that have found

    evidence of a slower rate of acquisition for bilingual chil-

    dren (Gildersleeve-Neumann et al., 2008; Swain, 1972;

    Vihman, 1982) and those that have found bilinguals

    to demonstrate a rate of acquisition within the normal

    range for monolinguals (Goldstein et al., 2005; Goldstein

    & Washington, 2001) but not those that have foundbilingual children to be acquiring skills at a faster rate

    than their monolingual peers (Kehoe et al., 2001; Lle

    et al., 2003).

    The bilingual children in the present study did not

    demonstrate evidence of acceleration according to its

    strict definition. That is, they did not demonstrate a faster

    rate of acquisition than that of their monolingual peers;

    rather, they demonstrated phonological skills that fall

    withinthe typical range for their monolingual peers.It is

    possible that this finding is evidence of a variation of the

    acceleration hypothesis. Perhaps interaction between

    the two languages of the bilingual child results in a sys-tem in which one language is aiding in acquisition of the

    other, allowing bilingual children to acquire phonological

    skills that fall within the normal range for monolinguals

    in both languages.It is also possible that deceleration and

    acceleration are occurring simultaneously. More specifi-

    cally, these bilingual children demonstrated lower accu-

    racy overall than monolingual children; however, these

    bilingual children demonstrated accuracy that was still

    within the typical range for that of monolinguals. There-

    fore, bilingual phonological development could be slower

    than that of monolingual children (i.e., deceleration) be-

    cause of the relative load of two inputs; however, in-

    teraction between the two phonological systems could

    counteract that load, maintaining a similar level of pho-

    nological skill (i.e., a type of acceleration) in both lan-

    guages in comparison to monolinguals.

    Phonetic Similarity and Shared SoundsIt was hypothesized in the present study that bi-

    lingual children might be using phonetically similar

    sounds (i.e., shared sounds) between their two languages

    to aid in the rate of acquisition. This phonetic similarity

    174 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    16/20

    between languages could help bilingual children main-

    tain phonological skills that fall within the normalrange

    for their monolingual peers in both languages, possibly

    leading to a variation of the acceleration hypothesis.

    That is, properties of one language could possibly aid in

    the rate of acquisition of the other language.

    In the Spanish and English productions of the bi-

    lingual group, a significant difference was found betweenthe accuracy of shared and unshared sounds (with a large

    effect size), thus appearing to support our hypothesis.

    However, a significant difference was found for the ac-

    curacy of shared and unshared sounds in the mono-

    lingual Spanish group as well. Further examination of

    the data within theSpanishmonolingual group and the

    Spanish productions of the bilingual group indicated

    that the accuracy of these bilinguals did, in fact, differ

    from those of monolingualsbut not in the way that

    was expected.

    First, the monolingual Spanish-speaking group ex-

    hibited a higher standard deviation for PCC of unshared

    sounds than did the bilingual children, indicating var-

    iation in accuracy of production of unshared sounds.

    Bilingual speakers, however, consistently demonstrated

    low accuracy on unshared sounds. That is, bilingual chil-

    dren exhibited low accuracy on all unshared sounds, and

    this pattern was observed across children. Second, the

    large difference exhibited by the monolingual Spanish-

    speaking group between accuracy on shared andunshared

    sounds was due to performance on two specific sounds,

    /R/ and /r/. The range of percent accuracy for unshared

    sounds in the monolingual Spanish-speaking group was

    between 30% and 73%. The range of percent accuracyfor unshared sounds in the Spanish productions of bi-

    linguals was between 23% and 50%, which is consider-

    ably lower. The bilingual group demonstrated somewhat

    equal accuracy across unshared sounds in their Spanish

    and English productions, whereas the monolingual speak-

    ers demonstrated difficulties with specific unshared

    sounds. It should be noted, however, that /R/ and /r/ are

    frequently occurring sounds in Spanish. Therefore, it is

    possible that low accuracy on these two sounds con-

    tributes to the significant effects in the unshared group

    for both bilinguals and monolinguals in Spanish. How-

    ever, it was observed in this study that frequency did notpredict sound accuracy; thus, this issue should be inves-

    tigated further in future studies.

    In addition to the findings from the productions of

    the monolingual Spanish speakers, large effect sizes were

    found between shared and unshared sounds for both bi-

    linguals and monolinguals, calling into question if dif-

    ferential accuracy of shared and unshared sounds is

    evidence of interaction between the two languages of

    bilinguals or if monolingual children also demonstrate

    this pattern. However, further analysis of shared and

    unshared sounds uncovered more differences between

    monolingual and bilingual speakers. First, these bi-

    lingual children demonstrated a statistically significant

    difference between accuracy of shared and unshared

    sounds in both languages, and this difference was not sta-

    tistically significant for monolingual English-speaking

    children. Second, when mean percent accuracy of shared

    sounds was compared across bilinguals (75.9% for Span-ish; 77.18% for English) and monolinguals (81.69% for

    Spanish; 87.75% for English), there was an accuracy dif-

    ference between bilinguals and monolinguals of 10.6%

    or less for both languages. More specifically, bilingual

    children might produce shared sounds with less accuracy

    than monolingual children, but they are within 10 per-

    centage points of their monolingual peers in both lan-

    guages. Finally, the standard deviation for bilingual

    productions on the accuracy of shared sounds is much

    smaller (SD= 7.54 for Spanish; SD = 11.43 for English)

    than for unshared sounds (SD= 9.62 for Spanish;SD =

    18.99 for English), indicating that there is less varia-tion in the production of shared sounds than unshared

    sounds, possibly due to phonetic similarity (Flege, 1981).

    FrequencyIt was predicted that frequency of occurrence of

    sounds in each language might predict high accuracy

    of shared sounds. If frequency had been found to pre-

    dict high accuracy of shared sounds, high accuracy of

    shared sounds would not provide evidence for between-

    language interaction (Paradis & Genesee, 1996). High

    accuracy of shared sounds, therefore, could not be at-tributed exclusively to interaction between the bilingual

    childs two languages.

    Results showed that frequency did not predict the

    accuracy of shared sounds. This finding is contrary to

    previous studies that have found that frequently oc-

    curring sounds are produced with higher accuracy than

    sounds that are significantly less frequent (e.g., Kirk &

    Demuth, 2003). It is possible that the parameter of fre-

    quency is not the driving force behind high accuracy of

    shared sounds, but other parameters of complexity, yet

    to be studied, influence sound accuracy. It is also pos-

    sible, as hypothesized in the present study, that highaccuracy of shared sounds is a result of interaction be-

    tween the two language systems of bilingual children.

    Limitations and Future DirectionsThese results indicate that it is possible to examine

    and compare phonological structures within the bilingual

    child for evidence of interaction between his or her two

    languages. A variation of the hypothesis of acceleration

    could possibly be added to complement Paradis and

    Fabiano-Smith et al.:Phonological Acquisition in Bilingual Children 175

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    17/20

    Genesees (1996) definition of this hypothesis as an ad-

    ditional way to provide evidence for how bilingual chil-

    dren acquire two languages. It is also possible that

    phonetically similar sounds are not the only phonologi-

    cal construct that bilingual children categorize as shared

    between their two languages. It is possible that bilingual

    children treat the factors related to linguistic complexity,

    or rules, of each language in much the same way. Thefindings from the present study provide some evidence

    for the investigation of factors such as these in future

    research.

    The major limitation of the present study is the rel-

    atively small number of participants included for statis-

    tical analysis. A relatively small number of participants

    reduces power; thus, statistical outcomes are suscep-

    tible to Type II error. There is inherent difficulty in

    matching bilingual participants on chronological age,

    input in each language, and output in each language as

    well as in assessing 3-year-old bilingual children (de-

    pending on the ambient language environment) thatcan produce language samples in both of their languages.

    In an attempt to create a homogeneous sample out of a

    heterogeneous population, the number of participants

    included in the present study was relatively limited.

    In future studies, it might be possible to use param-

    eters of complexity to predict on which elements in-

    teraction and separation will be observed between a

    bilingual childs two languages. Frequency of occurrence

    of soundsin a languageis often viewed as a factor related

    to linguistic complexity, such that sounds that occur

    frequently are viewed as less complex than those that

    occur frequently (Greenberg, 1966; Trubetzkoy, 1939).Because frequency was not found to be a predictor of

    high accuracy of shared sounds, additional parameters

    of complexity should be examined to determine at what

    level between-language interaction is occurring in bi-

    lingual children.

    ConclusionRelatively little is known about how typically devel-

    oping bilingual children acquire their phonological skills

    and even less about if and how the two languages of

    bilinguals interact. The findings of this study indicatethat although bilingual children demonstrate separa-

    tion between their two phonological systems, those sys-

    tems interact to aid in rate of acquisition. Although we

    found that bilingual children demonstrated a slower rate

    of acquisition than their monolingual peers on some mea-

    sures (e.g., phonological accuracy), these skills in the

    bilingual children were within the normal range of those

    for monolingual children in both English and Spanish. A

    similar rate of acquisition between bilinguals and mono-

    linguals occurred as well (e.g., on phonetic inventories).

    Finally, we found that sound frequency did not predict

    accuracy of either phonetically similar sounds between

    languages or phonetically dissimilar sounds specific to

    Spanish or English. The findings on rate of acquisition

    aid in our understanding of the interaction between

    (a) the languages of bilingual children and (b) the devel-

    opmental similarities and differences between mono-

    lingual and bilingual children.

    Acknowledgments

    We would like to express our gratitude to the children and

    families who participated in this project, both in the United

    States and Mexico. We thank Ferenc Bunta for assistance in

    the selection of recording equipment; Alexandra Hanlon for

    consultation on statistical analyses; and Jessica Barlow, Sonja

    Pruitt, Skott Freedman, Aquiles Iglesias, MeganDunn Davison

    and Ral Rojas for comments on earlier versions of the article.

    We express deep appreciation to Donna Jackson Maldonado,

    Rosa Patricia Brcenas Acosta, and Martha Beatrz Soto

    Martnez at the Universidad Autnoma de Quertaro in

    Mexico for their many efforts in the attainment of monolingualparticipants. Finally, we thank the following students who

    performed phonetic and orthographic transcription of the data

    and participated in analyses of reliability: Jenny Lange,

    Monica Krewson, Vanessa Gonzlez, Andrea Fisher, and

    Roxanna Palma.

    References

    Acevedo, M. A. (1993). Development of Spanish consonants inpreschool children.Journal of Childhood Communication

    Disorders, 15,915.

    Barlow, J.(2003). The stop-spirant alternation in Spanish:

    Converging evidence from a fortition account. SouthwestJournal of Linguistics, 22, 5186.

    Bauman-Waengler, J. (2008).Articulatory and phonologi-cal impairments: A clinical focus(3rd ed.). Boston: Allyn& Bacon.

    Bedore, L.(2005). Morphosyntactic development. InB. Goldstein (Ed.),Bilingual language development and dis-orders in Spanish-English speakers (pp. 163185). Baltimore:Brookes.

    Bhatia, T., & Ritchie, W. (1999). The bilingual child: Someissues and perspectives. In W. C. Ritchie & T. K. Bhatia (Eds.),

    Handbook of child language acquisition (pp. 569643).San Diego: Academic Press.

    Bland-Stewart, L. M., & Fitzgerald, S. M. (2001). Use ofBrowns 14 grammatical morphemes by bilingual Hispanicpreschoolers: A pilot study.Communication Disorders Quar-terly, 22(4), 171186.

    Cohen, J. (1988). Statistical power analysis for the behavioralsciences. Hillsdale, NJ: Erlbaum.

    Cysouw, M. (2003).Typology: Interpreting world-wide pat-terns of variation. LOT Summerschool, Tilburg. Berlin,Germany. Retrieved March 9, 2007, from http://email.eva.mpg.de/~cysouw/pdf/cysouwLOT4.pdf.

    DeHouwer, A. (1995). Bilingual language acquisition. InP. Fletcher & B. MacWhinney (Eds.), The handbook of childlanguage(pp. 219250). Malden, MA: Blackwell.

    176 Journal of Speech, Language, and Hearing Research Vol. 53 160178 February 2010

  • 7/25/2019 Phonological Acquisition in Bilingual SpanishEnglish Speaking Children

    18/20

    Demuth, K. (2001). Prosodic constraints on morphological de-velopment. In J. Weissenborn & B. Hohle (Eds.),Approachesto bootstrapping: Phonological, syntactic, and neurophysio-logical aspects of early language acquisition (pp. 321).

    Amsterdam: John Benjamins.

    Elbert, M., & Gierut, J. (1986).The handbook of clinicalphonology. Austin, TX: Pro-Ed.

    Fabiano-Smith, L., & Barlow, J. (2009). Typological varia-

    tion in the phonetic inventories of bilingual Spanish-Englishspeaking children.International Journal of Bilingual Educa-tion and Bilingualism.doi: 10.1080/13670050902783528.

    Fabiano, L., & Goldstein, B. (2005). Phonological transfer inbilingual Spanish-English-speaking children.Journal of

    Multilingual Communication Disorders, 3(1), 5663.

    Fabiano, L., & Goldstein, B. (2004a, May).Phonologicalrepresentation in simultaneous and sequential bilingual

    Spanish-English speaking children. Seminar presented atthe Child Phonology Conference, Tempe, AZ.

    Fabiano, L., & Goldstein, B. (2004b, June).Phonologicalrepresentation in simultaneous bilingual Spanish-Englishspeaking children: Two case studies. Poster presented at

    the Symposium of Research on Child Language Disorders,Madison, WI.

    Fiestas, C., & Pea, E. (2004). Narrative discourse in bi-