Journal of Multilingual Communication Disorders, March 2005; 3(1): 56 – 63

Phonological cross-linguistic effects in bilingual Spanish – English speaking children

LEAH FABIANO & BRIAN GOLDSTEIN Temple University, Department of Communication Sciences, Philadelphia, PA 19122, USA

Abstract Paradis (2001) posed the Interactional Dual Systems Model of language representation in which bilingual children possess two separate phonological systems that have mutual influence. Frequency of phonological cross-linguistic effects has been cited as evidence for determining phonological representation. The purpose of this study was to examine the frequency and types of phonological cross-linguistic effects that occur over time in bilingual children. Three female typically developing, bilingual Spanish – English speaking children, ages 5;0, 6;2, and 7;0 participated in the study. Single word, conversation, and narrative samples were examined for each child to determine the phonological makeup of each participant’s speech. Two analyses were performed: (1) analysis of phonemic cross-linguistic effects and (2) analysis of syllabic cross-linguistic effects. There were eight occurrences of cross-linguistic effects across the three children. Overall, the frequency of cross-linguistic effects was low, supporting the Interactional Dual Systems Model in which a child’s two phonological systems are separate, but non-autonomous.

Keywords: Bilingual, phonology, representation, Spanish

Background Two general theories on phonological representation in bilingual children have been debated by researchers. First, authors such as Schnitzer and Krasinski (1994) and Vogel (1975) posit that bilingual children begin with a single phonological system that later separates into two systems (i.e., the Unitary Language System Hypothesis (ULS)). Second, researchers such as Paradis (2001) and Keshavarz and Ingram (2002) propose a Dual Systems Model (DSM), in which bilingual children develop separate phonological systems for each language. It has been suggested by Paradis (2001) that bilingual children might have two separate phonological systems with mutual influence. The issue of phonological cross-linguistic effects is closely related to the discussion of these models in that the frequency of cross-linguistic effects has provided evidence for one model over the other.

Correspondence: Leah Fabiano, Temple University, Department of Communication Sciences, 110 Weiss Hall, Philadelphia, PA 19122, USA. Tel: 1 267 243 3277. Fax: 1 215 204 5954. E-mail: [email protected] Received for publication 21 June 2004. Accepted 16 November 2004. ISSN 1476-9670 print/ISSN 1476-9689 online # 2005 Taylor & Francis Group Ltd DOI: 10.1080/14769670400027316

Phonological cross-linguistic effects

57

Cross-linguistic effects Phonological cross-linguistic effects are defined as the influence that one language has on the other in bilingual individuals; these effects can occur from L1 to L2 and/or from L2 to L1 (Ellis, 1997). If effects occur from L1 to L2, they are often referred to as phonological transfer. If effects occur from L1 to L2 and from L2 to L1, they are referred to as phonological cross-linguistic effects (the focus of the current study). Frequency of phonological cross-linguistic effects has been cited, for a number of reasons, as one key piece of evidence for determining phonological representation. First, bilingual children could have one unitary system that separates into two separate systems as age increases if children exhibit a high frequency of cross-linguistic effects early in development that decreases over time (Schnitzer, & Krasinski, 1994). These authors found that, for the child in their study, English had a strong effect on Spanish, as English phonemes were often transferred into Spanish words; however, the frequency of transfer decreased as the child got older. Second, bilingual children could have two separate systems for phonological representation as evidenced by a low frequency of cross-linguistic effects (it is possible, of course, that some bilingual children would not show any evidence of cross-linguistic effects) (Keshavarz, & Ingram, 2002). Keshavarz and Ingram (2002) found that a few tokens of segmental or suprasegmental interference occurred when the child in their study was in a linguistic environment in which only one of his languages predominated, instead of his typical environment of half English input, half Farsi input (e.g., According to the authors, this input is defined as one parent, one language except for when his parents had English-speaking friends over to the house and both parents spoke in English). These results indicated an overlap, although a minimal one, of phonological elements. Finally, bilingual children could have two separate systems with systematic, mutual influence due to a systematic pattern of phonological cross-linguistic effects between the two languages (Johnson, & Lancaster, 1998; Paradis, 2001). Johnson and Lancaster (1998) studied a Norwegian – English bilingual child who organized each language’s characteristics as specified by each of the two ambient languages, but also demonstrated phonemic cross-linguistic effects (e.g., /C ¸ / found in English productions and /:/ found in Norwegian productions), demonstrating some interaction between the two languages. Paradis (2001) posed the idea of an Interactional Dual Systems Model after observing that the stress patterns of French influenced initial syllable stress patterns of English, while heavy versus light production of weak syllables was identical to that of monolingual children in each language. Her conclusion was that the children seemed to be functioning with two separate phonological systems, but that those systems were non-autonomous. That is, the phonological cross-linguistic effects that occurred were not due to the amount of input of those languages in the child’s environment, but rather were a result of a systematic interaction that created a more efficient system. There are a number of limitations to previous studies in this area that motivate the current investigation. Previous studies generally have not focused systematically on phonological cross-linguistic effects and mostly have been single case studies. The purpose of this study was to examine the frequency and type of phonological cross-linguistic effects that occur over time in bilingual Spanish – English speaking children. We predicted that bilingual children would exhibit a low frequency of cross-linguistic effects, and that the frequency of those effects would remain stable over time, thereby supporting an Interactional Dual Systems Model of phonological representation in which bilinguals have separate but non-autonomous phonological systems (Keshavarz, & Ingram, 2002; Paradis, 2001).

58

L. Fabiano & B. Goldstein

Method Three typically developing, Spanish – English bilingual, female children, ages 5;0, 6;2, and 7;0 participated in this quasi-longitudinal study. All children were speakers of Puerto Rican Spanish, Northeastern American English, and were able to perform all tasks in the study in both Spanish and English. A bilingual graduate student in communication sciences collected the data in both languages from each child in one session. In all cases, the examiner began data collection in Spanish and then collected English samples. Language history and child background All information regarding language history was obtained from parent or clinician report. The parent interview for the youngest child (C1) was unavailable so preschool language history is unknown; however, what is known is that she was exposed to English for at least 5 months after entering Kindergarten, there was no clinician concern regarding speech or language delays or disorders, and she was able to complete all tasks in both English and Spanish at a moderate to high proficiency level. The parent of the middle child (C2) reported that she and other family members have spoken both English and Spanish in the home since C2’s birth. The parent interview includes two proficiency scales: one examining language ability and the second measuring language use. On the language ability proficiency scale of 0 – 4 (i.e., 0 = cannot speak the indicated language and 4 = native-like ability in the indicated language), C2’s parent gave her a score of 3 in Spanish and 4 in English. The same scale was used to evaluate C2’s use of English and Spanish, and her parent rated her language use as 3 in Spanish and 4 in English. The parent of the oldest child (C3) reported that she and the child’s father spoke only Spanish in the home from the time C3 was born until she was 2 years old. After that point in time, both English and Spanish were spoken in the home. On the language ability proficiency scale of 0 – 4 (i.e., 0 = cannot speak the indicated language and 4 = native-like ability in the indicated language), C3’s parent gave her a score of 4 in both Spanish and English. The same scale was used to evaluate C3’s language use in English and Spanish and her parent rated her spoken language abilities as 4 in both languages as well. According to parent or clinician report, the children had no speech, language, cognitive, or neurological deficits. Overall Percentage of Consonants Correct (PCC) (Shriberg, & Kwiatkowski, 1982) for each child in both Spanish and English is displayed in Table I. All three children were labeled as bilingual based on a measure of grammaticality using conversation and narrative samples (Restrepo, 1998). Speech samples were recorded and digitized for playback on a computer. The connected speech samples were then aggregated for each child to increase the overall number of utterances for analysis and to equalize the

Table I. Age, percentage of ungrammatical utterances, and percent consonants correct (PCC) for each participant.

Child C1, age 5;0 C2, age 6;2 C3, age 7;0

Percentage of ungrammatical utterances in English

Percentage of ungrammatical utterances in Spanish

English PCC

Spanish PCC

Mean PCC

7.95 1.42 1.98

0 10.52 0

90.53 96.91 94.79

82.72 85.19 96.43

86.63 91.05 95.61

Phonological cross-linguistic effects

59

length of the connected speech samples across the three children. Thus, aggregating the samples provided a more equal number of productions across the three children. The Systematic Analysis of Language Transcripts RV8 (SALT) (Miller, 2004) was used to obtain total number of utterances, total number of ungrammatical utterances, and percentage of ungrammatical utterances in each language. A criterion of 18% or less ungrammatical utterances was used to label the children as bilingual (based on Restrepo, 1998). For all three children, percentage of ungrammatical utterances was below 18% in each language indicating requisite proficiency in both languages (Table I). Specifically, percentage of ungrammatical utterances was less than 11% for each child in both languages. The single word and connected speech samples were collected in the following manner. First, single word productions were elicited using the phonology subtest of the Bilingual English-Spanish Assessment (BESA) (Pen˜a, Gutierrez-Clellen, Iglesias, Goldstein, & Bedore, in development), a phonological assessment designed to gauge children’s speech sound productions in both English and Spanish. Each target item was elicited via a spontaneous label made in reference to a photograph. Second, a 10 – 15-min conversational sample was elicited from each child in both English and Spanish. At least 25 utterances were collected in both languages from each of the children. Finally, narrative samples were elicited in English and Spanish from all three children using the wordless picture book, Frog, Where Are You? (Mayer, 1969). Analyses Each of three samples was tape-recorded, and then digitized for playback on a computer. Data from the single word, conversational, and narrative samples were transcribed and analyzed with the computerized Logical International Phonetics Program (LIPP) (Oller, & Delgado, 2000) to determine the phonological makeup of each participant’s speech. Dialectal features of both Spanish and English were taken into account and not scored as errors. Intra-judge reliability was calculated for each transcript for all three children and reached 99%. Two main analyses were performed: (1) analysis of phonemic cross-linguistic effects from L1 to L2 and from L2 to L1, and (2) analysis of syllabic cross-linguistic effects from L1 to L2 and from L2 to L1. First, the substitutions that occurred across all three speech samples were included in order to determine the type and frequency of phonemic cross-linguistic effects. Second, the frequency and types of syllabic cross-linguistic effects were examined. The method used to separate words into syllables for these analyses was determined by the phonotactic constraints of both English (Roca, & Johnson, 1999) and Spanish (Harris, & Cottam, 1985). Syllable types examined included V, CV, CVC, CCV, CCVC for both English and Spanish, and CCCV, CCCVC, CCCVCCC, CCCVCCCC for English only (there are no syllable types unique to Spanish that do not also occur in English). Once each target syllable type was determined, it was compared to the child’s production of the syllable. Differences between the syllable target and child’s production were analyzed for cross-linguistic effects between English and Spanish.

Results Phonemic cross-linguistic effects The frequency and types of bi-directional cross-linguistic effects for each child can be found on Table II. Overall, the frequency of phonemic cross-linguistic effects was low, regardless

60

L. Fabiano & B. Goldstein Table II. Characteristics of substitutions used in phonemic cross-linguistic effects.

Child (age)

Context of occurrence

Sampling condition

Substitution (number of tokens)

C1 (5;0)

Spanish

Narrative

C2 (6;2)

English Spanish Spanish Spanish

Conversation Single Word Single Word Narrative

/s/?[y] (1) /o/?[;] (1) /l/?[;] (1) /v/?[b] (1) /r/?[r] (1) /e/?[r] (1) /e/?[r] (1) 7

C3 (7;0) Total

of sample type. That is, sampling condition did not seem to have an effect on the use of cross-linguistic effects, as phonemic cross-linguistic effects occurred at least once in each sampling condition. There were seven occurrences of phonemic cross-linguistic effects overall across the three children. More specifically, the youngest child in the study had 379 opportunities for phonemic cross-linguistic effects and demonstrated only three occurrences (0.8%), the intermediate child had 512 opportunities exhibited only three occurrences (0.6%), and the oldest child had 378 opportunities and demonstrated only one occurrence of phonemic cross-linguistic effects (0.3%). Further analysis indicated that bi-directional cross-linguistic effects did occur, and occurred at a higher rate from L2 to L1 (English effects on Spanish). Of the seven instances of phonemic cross-linguistic effects, only one occurred from Spanish to English (C2: /v/?[b]), while the remaining tokens were a result of English influence on Spanish (e.g., /s/?[y], /o/?[;], /r/?[r]). Syllabic cross-linguistic effects Overall, syllable structure was appropriately maintained by all children in samples from both languages. Across all three children, there was only one occurrence of syllabic crosslinguistic effects. The youngest child (C1) exhibited one exemplar of epenthesis (e.g., addition of the vowel /e/ before the phoneme /s/ in an English word: /skutE/?[eskutE]) in her English conversation sample (i.e., transfer from L1 to L2). There were no other tokens of syllabic cross-linguistic effects. Discussion The purpose of this study was to examine the frequency and types of phonological crosslinguistic effects in three bilingual Spanish – English speaking children. It was predicted that the oldest child would demonstrate fewer cross-linguistic effects than the youngest child due to increased experience with both languages, but that occurrences of cross-linguistic influence would remain relatively stable over time. Overall, a decrease in phonological crosslinguistic effects was observed across the three children. The youngest child exhibited three instances of phonemic cross-linguistic effects, and one token of syllabic cross-linguistic effects, the intermediate child produced three instances of phonemic cross-linguistic effects and no tokens of syllabic cross-linguistic effects, and the oldest child demonstrated one exemplar of phonemic cross-linguistic effects and no instances of syllabic cross-linguistic

Phonological cross-linguistic effects

61

effects. Overall, the intermediate child differed slightly from the youngest (in syllabic crosslinguistic effects) and oldest child (in number of occurrences of phonemic cross-linguistic effects). Because there were so few occurrences of cross-linguistic effects, it is difficult to determine if the frequency of occurrence truly decreases as children mature. It seems more likely that the relatively low frequency of phonological cross-linguistic effects remains stable over time from ages 5;0 to 7;0. It is difficult to determine if these slight differences are due to chronological or maturational factors or if they are a result of the three children selected, despite matches on gender and grammaticality. Finally, it is also possible to infer that phonological cross-linguistic effects are not influenced by maturity or experience with both languages, but that they are a characteristic of bilingualism itself. Paradis (2001) noted that bilingual children exhibit ‘‘interlanguage structural ambiguity’’ on some features of each language (p. 34). That is, bilingual children demonstrate linguistic characteristics that differ from monolingual speakers of either language, therefore cross-linguistic effects could possibly be an efficiency mechanism found in speakers of more than one language that is simply not required in monolingual speakers. The results of this study demonstrated that bilingual children exhibit phonological differentiation as well as the borrowing of elements from one language for use in the other. These findings support the Interactional Dual Systems Model in that the phonological systems in these three children were separate, but non-autonomous. The children in this study did not appear to be functioning from one unitary system due to the limited number of phonological cross-linguistic effects that occurred across the three ages, thus clearly distinguishing phonological elements that belong to either English or Spanish. This finding demonstrates that although the children were showing a definite distinction of phonological elements, they were also utilizing elements specific to one language in their other language. In conclusion, the results of this study support the hypothesis based on The Interactional Dual Systems Model in that the children demonstrated separation of phonological elements of each language, but also demonstrated cross-linguistic effects, or interaction, between Spanish and English. There are a number of limitations to the current study that should be avoided in future research. First, a case study methodology was used, limiting the generalizability of the results. In the future, a larger sample of children needs to be included in order to generalize the findings. Second, the three children were relatively old and had at least 2 years experience with English. Including younger children and/or those with less English experience may yield a higher frequency of cross-linguistic effects. Third, the tasks themselves may have limited the frequency of cross-linguistic effects. The data were collected for this study in the children’s school, a primarily English-speaking environment. In addition, the tasks required of them were school-based (e.g., picture-naming and storytelling). Thus, the children may have been performing tasks that were more familiar in English but less familiar in Spanish. This issue of context is one possibility for the uneven amount of cross-linguistic effects that was exhibited in the study’s results. Finally, the frequency of cross-linguistic effects in the children’s outputs may be correlated to their frequency in the input. In the current study, only the children’s output was examined. In the future, it might be useful to examine the input the children are receiving from their parents in order to correlate frequency of cross-linguistic effects in the input and output. Clinical implications It is somewhat difficult to make clinical recommendations from these findings given the overall low frequency of cross-linguistic effects. However, the type, amount, and direction of

62

L. Fabiano & B. Goldstein

cross-linguistic effects have clinical implications for speech-language pathologists (SLP), especially in the evaluation and diagnosis of phonological disorders. Instances of phonological cross-linguistic effects could be mistaken for substitution (i.e., true) errors, causing a child to appear less phonologically mature. Thus, if children exhibit a high frequency of cross-linguistic effects, there is the possibility that such children might be inappropriately labeled as phonologically disordered. Goldstein and Iglesias (2001) found such an effect when the dialect features of Spanish were not accounted for. The findings from this study also indicate that it is important take into account the context in which bilingual children use each or both of their languages. This includes determining where and with whom the child interacts. SLPs should prepare for the possibility that collecting a Spanish speech sample in a primarily English-speaking environment, within an Anglo-based school culture, will result in a higher frequency of L2 to L1 cross-linguistic effects than L1 to L2 cross-linguistic effects. Thus, SLPs should be aware of how context affects assessment and diagnosis. Doing so will result in the valid and reliable assessment of phonological disorders in bilingual speakers. Acknowledgements Funded in part by National Institutes of Health (NIH) Contract# N01-DC-8-2100. References Ellis, R. (1997). SECOND language acquisition. Oxford, UK: Oxford University Press. Goldstein, B., & Iglesias, A. (2001). The effect of dialect on phonological analysis: Evidence from Spanish-speaking children. American Journal of Speech-Language Pathology, 10, 394 – 406. Harris, J., & Cottam, P. (1985). Phonetic features and phonological features in speech assessment. British Journal of Disorders of Communication, 20, 61 – 74. Johnson, C., & Lancaster, P. (1998). The development of more than one phonology: A case study of a NorwegianEnglish bilingual child. International Journal of Bilingualism, 2, 265 – 300. Keshavarz, M., & Ingram, D. (2002). The early phonological development of a Farsi English bilingual child. International Journal of Bilingualism, 6, 255 – 269. Mayer, M. (1969). Frog, where are you? New York: Dial Books. Puffin Books, USA. Miller, J. (2004). The systematic analysis of language transcripts RV8 (SALT). Copyright University of WisconsinMadison. Oller, D., & Delgado, R. (2000). Logical international phonetics program (LIPP). Miami: Intelligent Hearing Systems. Paradis, J. (2001). Do bilingual two-year-olds have separate phonological systems? The International Journal of Bilingualism, 5(1), 19 – 38. Pen˜a, E., Gutierrez-Clellen, V., Iglesias, A., Goldstein, B., & Before, L. et al. (in development). Bilingual English Spanish assessment. Unpublished assessment tool. Restrepo, A. (1998). Identifiers of predominantly Spanish-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research, 41, 1398 – 1411. Roca, I., & Johnson, W. (1999). A course in phonology. Malden, MA: Blackwell Publishers. Schnitzer, M., & Krasinski, E. (1994). The development of segmental phonological production in a bilingual child. Journal of Child Language, 21, 585 – 622. Shriberg, L., & Kwiatkowski, J. (1982). Phonological disorders III: A procedure for assessing severity of involvement. Journal of Speech and Hearing Disorders, 47, 256 – 270. Vogel, I. (1975). One system or two: an analysis of a two-year-old Romanian-English bilingual’s phonology. Papers and Reports on Child Language Development, 9, 43 – 62.

Biographical Notes Leah Fabiano, M.S. is a doctoral student in the Department of Communication Sciences at Temple University. She holds the Certificate of Clinical Competence (CCC) in speech-

Phonological cross-linguistic effects

63

language pathology. Her research interests are phonological development and disorders in bilingual Spanish – English speaking children. Brian A. Goldstein, Ph.D., is an Associate Professor in the Department of Communication Sciences at Temple University. He holds the Certificate of Clinical Competence (CCC) in speech-language pathology. His research interests are phonological development and disorders in Spanish-speaking children. He is also the Editor of Language, Speech, and Hearing Services in the Schools (2004 – 2006).