Loss of Gender Agreement in
L1 Attrition: Preliminary Results
Raquel T. Anderson
Indiana University
Abstract
The purpose of the present investigation was to obtain preliminary data on the effects of first language (L1) loss on the Spanish of bilingual children, particularly with respect to noun phrase (NP) gender agreement. Two children, bilingual in English and Spanish, were followed longitudinally and their use of gender agreement was examined. Data on incidence and type of NP gender agreement errors were obtained. Results of the investigation suggest that Spanish gender agreement morphology may be vulnerable to language loss. Nevertheless, differences in relative impact of L1 loss on gender agreement varied among the children. In contrast to previous research with Spanish-speaking children who are learning English as a second language and who have language impairment, it appears that gender errors are the result of the language contact situation, and not because of language learning deficits. Possible reasons for the observed patterns and directions for future research are presented.
In many countries, minority languages are in a precarious situation. Their survival, specifically across generations, is tenuous at best. This is especially the case in the United States, where research on language use among immigrant populations suggests a shift from the minority language to the majority language (Fillmore, 1991; Harlan, 1991; Hudson-Edwards & Bills, 1982; Veltman, 1988). This language shift has been reported in enclaves where residents are mainly of a particular ethnic group that is associated with a particular language (Hudson-Edwards & Bills, 1982; Martínez, 1993; Silva-Corvalán, 1991). For example, in a comprehensive longitudinal study conducted by Zentella (1997) on a small community of Puerto Rican families in New York City, the movement from Spanish usage in the home to the use of English for communication across domains was observed in later generations. English became a dominant language of use in the home and of parent-child interaction. This shift from minority language to majority language has also been evidenced in children who are isolated from other speakers of the minority language, and often times more dramatically, as there is minimal or no support outside the home for the minority language (Kravin, 1992; Orellana, 1994).
A more dramatic phenomenon that has been reported in the literature concerns the loss of L1 abilities within one generation, usually referred to as L1 loss. The phenomenon of L1 loss can be defined as one in which an individual's abilities, usually measured expressively, in his/her L1 are reduced or diminished (Kaufman & Aronoff, 1991; Orellana, 1994; Pfaff, 1991; Schiff-Myers, 1992; Turian & Altenberg, 1991). It has been observed in children in language contact situations that do not foster language maintenance. Even in instances where there is a conscious attempt by parents to foster the L1, as in enrolling the children in special programs where the home language is taught, diminished abilities in the language have been noted (Kaufman, 1998). Children in particular are especially vulnerable to L1 loss, as often they have an "integrative orientation" (Kaufman, 1998, p. 409) that moves them toward the majority language. Thus, affective factors interact with the sociolinguistic realities in which they are developing two languages, resulting in loss of skill in their first language.
Often, L1 loss is evidenced by changes in the grammatical and lexical aspects of the language. For example, in a recent study on Hebrew skills of children living in the United States and whose parents were native speakers of the language, Kaufman (1998) reported a significant change in the children's Hebrew lexicon. In narrative tasks, she found a reduction in lexical diversity, with the children evidencing overuse and overgeneralization of lexical items, especially nouns. Other studies have reported significant changes in the children's use of L1 grammar, usually resulting in a reduction of inflectional morphology (Anderson, 1999; Bayley, Alvarez-Calderón, & Schechter, 1998), as well as transfer of L2 syntactic structure for semantically similar constructions (Seliger, 1989). What is interesting about these patterns is that some have also been reported for other types of language learners, in particular, children who have been identified as specific language impaired (SLI). This is especially apparent in inflectional morphology.
Children with SLI can be characterized as having language learning deficits in the absence of any hearing loss, frank neurological and emotional impairment, and cognitive deficits (Craig, 1996). A particularly salient characteristic of these children is the substantial problems they encounter in using inflectional morphology, as compared to age-matched and language-matched peers (Connell & Stone, 1992; Rice & Wexler, 1996). This pattern has been noted in children who are monolingual speakers of various languages, including English (Rice & Wexler, 1996), Swedish (Hansson & Nettelbladt, 1995), Hebrew (Dromi, Leonard, & Shteiman, 1993), and Italian (Bortolini, Caselli, & Leonard, 1997; Leonard, Sabbadini, Volterra, & Leonard, 1988). It has also been observed in Spanish-speaking children in the United States who are learning English as a second language (Restrepo, 1997, 1998; Restrepo, Chasteen-Bustelo, & Mattute, 1996).
Recent research describing the language skills of Spanish-speaking children who have been identified as specific language impaired, and who are learning English as a second language, suggests certain production patterns that may be used to distinguish typical from atypical Spanish-speakers. Restrepo and colleagues (Restrepo, 1997; Restrepo, Chasteen-Bustelo, & Mattute, 1996) have studied the expressive language skills of Spanish-speaking children. Although primarily Spanish-speaking, all children were learning English as a second language. An analysis of the children's spontaneous speech revealed that the aspect most frequently compromised was noun phrase agreement (NP) morphology. More specifically, gender agreement in the use of articles tended to be affected (Restrepo, 1997). As a result of these findings, Restrepo has suggested that gender agreement errors may be used as linguistic markers for identifying Spanish-speaking children with language disabilities.
Previous studies of children with specific language impairment and who are learning languages that require gender agreement in the NP (i.e., article and adjective, if used, must agree with the noun) have failed to report such a pattern in the children's spontaneous speech. Recent research on monolingual Italian-speaking children with SLI has consistently indicated that these children do not experience difficulty in using NP gender agreement (Bortoloni, et al., 1997; Leonard, et al., 1988). In addition, previous descriptions of the skills of Spanish-speaking children with atypical and typical language skills have failed to pinpoint difficulties in the children's acquisition of gender agreement rules (Anderson, 1995; Merino, 1983). Gender agreement morphology is established quite early in monolingual Spanish-speaking children with typical language skills, most accounts suggesting that by three years of age NP gender agreement is in place (Hernández-Pina, 1984; Martínez & Shatz, 1996). Furthermore, anecdotal evidence obtained by this author from Spanish-speaking clinicians, as well her extensive clinical experience with monolingual language impaired Spanish-speaking children, suggest that NP gender agreement is not impaired in these children.
The discrepancy in the research on children with SLI and use of NP gender agreement rules may lie in the particular sociolinguistic realities of the children. Restrepo et al.'s (1996) subjects, although Spanish speakers, were in a bilingual context. The other children were in monolingual environments. More specifically, the Spanish-speaking children who evidenced significant deficits in NP gender agreement were in a situation where they spoke as an L1 a minority language that, as previous research on language shift in the United States suggests, is vulnerable to loss (Gutiérrez, 1990; Harlan, 1991; Ocampo, 1990). In fact, earlier research by Brisk (1974) suggests that in language contact situations, NP gender agreement may be later acquired. It is plausible, then, that the children studied by Restrepo et al. (1996) showed deficits in gender agreement, not because this is an inherent problem of Spanish-speaking children with language impairment, but because it is the result of sociolinguistic factors that impact L1 performance in general.
Although various Spanish linguists have attempted to establish rules for ascribing grammatical gender to Spanish (Klein, 1989; Natalicio, 1983; Roca, 1989; Teschner, 1987), the reality is that gender is arbitrary. This does not mean that there are no phonological characteristics of the nouns that correlate with grammatical gender. For example, nouns ending with -a tend to be feminine (e.g., la [fem] casa, "the house"), and nouns ending with -o are often masculine (e.g., el [masc] pelo, "the hair"). Nevertheless, the exceptions abound in the language (e.g., el [masc] problema, "the problem"; la [fem] mano, "the hand"), making it difficult to establish an efficient system for ascribing grammatical gender to nouns. Furthermore, many nouns do not end in a or o, and similar word endings can be masculine or feminine (e.g., el [masc] café, "the coffee"; la [fem] torre, "the tower"; el [masc] avión, "the airplane"; la [fem] canción, "the song"). Because of this, it has been suggested that lexical acquisition includes learning the noun's corresponding grammatical gender; that is, grammatical gender is a feature of the noun (Bybee, 1985; Gariano, 1984). It is this characteristic of grammatical gender that makes it susceptible to loss.
One of the immediate consequences of a minority-majority language contact is that the domains of use of the minority language will be restricted. For example, in the United States, use of Spanish is often confined to home situations predominantly in adult-child interactions. As English becomes the language of instruction and peer interaction (Fillmore, 1991; Kravin, 1992), the amount of time the child spends listening and speaking the minority language continues to be reduced. This reduction, in turn, impacts L1 lexical skill.
Lexical knowledge is particularly vulnerable in situations where language loss is occurring (Gal, 1989; Weltens & Grendel, 1993). This has especially been noted at the productive or expressive level. It has been observed that diminished exposure to a language negatively impacts vocabulary use (Kuhberg, 1992; Schaufeli, 1992; Weltens & Grendel, 1993). A reduction in L1 input, then, would result in diminished use of L1 lexical items. According to Bybee (1985, 1995), frequency of use of a particular lexical item or morphological form impacts the speaker's ability to access that form. She further states that specific features of lexical items are more vulnerable to changes in frequency of occurrence. These features are those that are peripheral to the lexeme's meaning. Being an arbitrary assignment, grammatical gender can be considered a marginal feature of a noun, and thus more susceptible to loss. The end result would be that in situations of L1 loss, the grammatical gender of Spanish nouns will be affected, regardless of an individual's language learning skill. In other words, Spanish NP gender agreement errors are characteristic of Spanish speakers in a language contact situation where Spanish is the minority language, and not due to particular difficulties in language learning per se. If this were the case, Spanish-speaking children with typical language skills and who are experiencing L1 loss should present a pattern where NP gender agreement errors are produced and where these errors will increase with greater exposure to the L2, with a concomitant reduction in use of the L1 Spanish.
The purpose of the present investigation was to evaluate NP gender agreement morphology in Spanish-English-speaking children who are in a language minority situation. Longitudinal spontaneous speech data on two children's use of gender agreement morphemes in Spanish noun-phrases were obtained. The data are tentative in nature, as the focus of the investigation was to discern possible typical L1 loss patterns in Spanish-speaking children. For this investigation, the following research questions were posed: First, is NP gender agreement morphology vulnerable to language loss in typical language learners? Second, Are the observed patterns in gender agreement similar to those reported for Spanish-speaking children in a bilingual environment who have been diagnosed as SLI?
Subjects
The children who participated in this longitudinal project, Beatriz and Victoria, were two typical Spanish-English-speaking children of Puerto Rican descent. At the beginning of data collection, Beatriz was 6 years 7 months and Victoria was 4 years 7 months old. At the last taping session, they were 8 years 5 months and 6 years 5 months old, respectively. The children were siblings who lived with their parents, both of whom were native speakers of Puerto Rican Spanish. Both parents were college educated. The mother had native-like fluency in English, while the father spoke with a Spanish accent. The parents spoke Spanish at home with each other and with their children. The only exception occurred in instances where there were visitors who were monolingual English speakers. Although the parents were consistent in using Spanish when interacting with their daughters, the children were allowed to respond in English. During the initial tapings, Beatriz and Victoria spoke both Spanish and English to each other while interacting during play. Midway into the study, English was the only language used during play, the exception being when one of the parents was an active participant. The children's playmates included monolingual English-speaking children and English- speaking children who also spoke other languages (e.g., Korean, Chinese).
The family moved from Puerto Rico to a large urban area in the United States when Beatriz was 3 years 6 months old and Victoria was 1 year 6 months old. Prior to that time, interactions with the children in and outside the home environment were in Spanish. In the United States, Beatriz was enrolled in a preschool program where only English was spoken. Victoria attended the same preschool at 2 years 6 months of age. Unlike Beatriz's teacher, Victoria's teacher was bilingual and often interacted in Spanish with her. At the ages of 5 years 6 months and 3 years 6 months, the family moved to a small city in the Midwest where they are presently residing. Both children continued attending schools/preschools where English was the primary language. During the time when the data were collected, the children were enrolled in an elementary school with a high percentage of children from homes in which languages other than English were spoken. Most of the children were speakers of Asian languages. Interactions with schoolmates and teachers were in English. The only instance when Spanish was used for instruction was in a Spanish as a foreign language class that was taught twice a week and in which Beatriz was enrolled while in the second grade. Victoria was not exposed to this instruction. Both children acquired literacy skills in English. The older child, although limited, had literacy skills in Spanish, while the younger child was beginning to learn to read in Spanish. The parents were involved in the development of the children's Spanish literacy skills.
Typical language development was established from parental information. According to both parents, one of whom is a certified speech-language pathologist, Beatriz and Victoria have typical language skills. There were no concerns regarding speech, language, and hearing abilities. Both children are at grade appropriate levels in English reading and writing skills and have consistently performed in an above average manner in school, as evidenced by teacher and grade reports.
Procedures
Victoria and Beatriz were videotaped during child-adult dyads with their mother. The children were told by their mother that during the taping session, they were expected to speak Spanish. Each videotaped interaction lasted approximately 30 minutes. These sessions were conducted in quiet rooms at the investigator's work site. A variety of toys and materials were present in the room. These were chosen prior to the taping session by the children and their mother. During each data collection period the same materials were available in the room for both children. Nevertheless, the children were permitted to choose the particular toys with which they wanted to play. Child-initiated choice was established so as to maintain more naturalistic interaction between the children and their mother, as this was the typical pattern of interaction at home.
The adult-child Spanish interactions were video recorded every one to two months during a period of 22 months, resulting in a total of 12 recordings for each child. A research assistant with knowledge of Spanish orthographically transcribed the productions of the children and the adult. All productions made by the children were transcribed, including instances of English use and intra-sentential codeswitching (e.g., a hand está allí, "a hand is there"). After the initial transcriptions were obtained, the investigator, who is a native Puerto Rican Spanish-speaker and is very familiar with the children's speech patterns, checked each one for accuracy and made the necessary corrections in both the child and adult transcriptions. Victoria's transcriptions varied in length from 146 to 361 utterances. The number of utterances produced by Beatriz during the taping sessions varied from 184 to 328.
For each taping session, all noun phrases produced by each child were transcribed. Because the purpose of the present investigation was to analyze the children's use of agreement morphemes in Spanish NPs, only NPs that included an article and noun (e.g., una casa, "a house"), an article, noun, and adjective (e.g., una niña nueva, "a new girl"), and an adjective and noun (e.g., casa chiquita, "small house"), were included in the gender agreement corpus for each child. NPs that contained English words were included if they did not compromise evaluating agreement morphology. For example, a NP such as los ojos blue was included, as agreement between the article los and the noun ojos could be established. Similarly, if noun-adjective agreement could be assessed, such as in a niña chiquita, the phrase was incorporated into the NP corpora.
Analysis Procedures
General Linguistic Complexity Measures
Two general complexity measures were obtained: (a) mean length of utterance (MLU), and (b) use of Spanish nouns. Both were established for each of the children's transcripts. Although MLU has been used with children who are younger than Beatriz and Victoria, the measures were used to assess change in morphological complexity across taping sessions, and not to contrast the children's performance to those of monolingual controls. Procedures for calculating MLU followed those suggested by Linares (1983) for Spanish-speaking children. Only utterances that were completely in Spanish were included in the analysis. Utterances were identified following Miller's (1981) guidelines. The total number of Spanish nouns and the number of different Spanish nouns were determined. All Spanish nouns were tallied, regardless of the context in which they occurred (i.e., codeswitched and Spanish-only utterances).
Analysis of Gender Agreement in Spanish NPs
Total number of NPs and total number of different nouns were calculated for each taping session. Because the number of codifiable NPs varied across sessions, percent incidence of occurrence of gender agreement errors was used for data analysis. For each incorrect production, the direction of the error was noted. Two possible types of errors could occur: (a) use of masculine form when the noun's grammatical gender was feminine (e.g., un [masc] casa [fem], "a house"), and (b) use of the feminine form when the referent's grammatical gender was masculine (e.g., una [fem] papel [masc], "a piece of paper"). Error productions were also categorized with respect to the particular NP component that was produced incorrectly. Again, there were two options: (a) incorrect gender agreement observed in the use of the article (e.g., un [masc] casa [fem] "a house"), and (b) incorrect gender agreement observed in the adjective (e.g., una casa [fem] rojo [masc], "a red house"). In all instances, percent occurrence was established. Because most of the children's codifiable NPs were of the article + noun type, analysis pertaining to where the error occurred (i.e., article and/or adjective) was not done.
Reliability
All analysis procedures were initially conducted by a non-native Spanish-speaking research assistant. Point-by-point comparison with the investigator's analysis indicated no differences in error categorization across all NPs included in the analysis; that is, there was 100% agreement in the identification of gender agreement errors.
Figure 1. Percent incidence of NP gender agreement errors
Data on percent incidence of NP gender agreement errors are presented in Figure 1. At first glance, it is evident that the children varied in frequency of occurrence of gender agreement errors. Beatriz made few errors of this kind. Throughout the taping sessions, percent incidence varied from 0% (i.e., times 3, 4, 6, and 12) to 5.8% (i.e., time 10). Victoria, on the other hand, presented a more variable pattern, with a trend toward an increase in NP gender agreement errors over time. During the initial data sets, very few gender agreement errors were observed. This was followed by increases and decreases in inaccurate gender assignment to Spanish nouns. By the end of the data collection period (i.e., times 10, 11, and 12), a shift toward an increased number of NP gender agreement errors was noted. During the last three taping sessions, the percent incidence of incorrect gender in NPs was 25% (time 10), 17.4% (time 11), and 18.2% (time 12).
Further examination of the data showed differences between the two children in the total number of NPs that required use of gender agreement morphology, including total number of different words. Across all data sets, Beatriz produced a higher number of NPs that could be evaluated for gender agreement. That is, she used article + noun or article + noun + adjective constructions more frequently than Victoria. In addition, the total number of different nouns that Beatriz produced in each session was greater than Victoria's. As a result, she had more opportunities to evidence agreement errors. Nevertheless, agreement errors were minimal. Victoria's pattern of production was one where the number of NPs that could be coded for gender agreement diminished across sessions. While during the first taping sessions, she produced 30 or more codifiable NPs, the data collected during the last three taping sessions varied from as low as 16 (times 9 and 10) to 22 (time 12). Furthermore, the total number of different nouns used during the tapings was also reduced, with as few as 15 distinct Spanish nouns within a codifiable NP produced by Victoria during the last taping session. Although a decrease in the use of NPs that necessitated the use of gender agreement morphology was noted in Victoria's corpora, the number of gender errors increased across time.
To further study the relationship between the children's use of Spanish nouns and the incidence of incorrect gender assignment across time, the correlation between number of errors and the total number of Spanish nouns was established for each child. Pearson product correlations indicate a significant negative correlation ( r = -.634; p < .028 ) between number of words and incidence of gender errors in Victoria's data; that is, the highest incidence of errors were observed when there were less codifiable Spanish noun phrases in the corpora. Although the trend was similar to what was observed in Victoria's data, results from the analysis of Beatriz's data was nonsignificant (r = -.538; p > .07).
Figure 2. MU scores for Victoria and Beatriz
In consonance with an increase of gender agreement in NPs, Victoria displayed a gradual reduction in general morphological complexity, as assessed by MLU. This can be readily observed in Figure 2. During the initial taping sessions (i.e., times 1 through 7), Victoria's MLU scores ranged from as low as 4.91 (time 3) to as high as 6.13 (time 1). By the last sessions, her MLU scores varied from 3.23 (time 11) to 4.15 (time 9). In fact, the lowest MLU scores were those of the last three data points. These lower MLU scores correspond to a higher incidence of gender agreement errors. As with Victoria, Beatriz presented a pattern in which MLU scores tended to decrease across tapings. For example, MLU scores obtained from the initial tapings fluctuated between 8.76 (time 4) and 6.12 (time 2). Analysis of later transcripts yielded MLU scores in the range of 4.64 (time 10) and 5.69 (time 9). Nevertheless, as can be observed in Figure 2, this pattern was reversed during the last taping session, where she obtained an MLU score of 7.22. For both children the correlation between MLU scores and incidence of gender errors was negatively skewed and significant (Victoria - r = -.720, p < .01; Beatriz - r = -.650, p < .023 ).
Direction of Gender Errors
Figure 3. Beatriz: Percent incidence of errors per gender type
The direction of the gender errors evidenced in Victoria's and Beatriz's productions across time was analyzed. For each taping session, the number of target Spanish nouns that were masculine and the number of nouns that were feminine were established. For each grammatical gender type, percent incidence of error was determined. The results are summarized in figures 3 and 4. As can be observed in figure 3, Beatriz presented a pattern where the incidence of feminine-to-masculine and masculine-to-feminine errors fluctuated across time. No definite movement toward a specific gender error direction was noted. As expected from the general incidence data, few errors were produced across taping sessions.
Figure 4. Victoria: Percent incidence of errors per grammatical gender
A different pattern emerges when Victoria's error data are analyzed. The pattern noted was one where both feminine-to-masculine and masculine-to- feminine errors were present during the first half of the taping sessions. The second half suggests a movement toward feminine-to-masculine errors. As evidenced in Figure 4, for the latter part of the taping sessions the percent incidence of incorrect gender assignment in the production of feminine nouns ranged from as low as 16.7% (times 7 and 9) to as high as 50% (times 10 and 11). No errors in the production of masculine forms were noted during five of the last six taping sessions. During the last taping, incorrect gender assignment was observed in the production of 37.5% of the feminine nouns, while 7.1% of the masculine nouns were incorrect. The observed pattern suggests that Victoria was moving toward the use of the default form (i.e., masculine) for all targets.
Other Production Patterns Observed in the Use of NPs With Spanish Nouns
To address the possibility that the children were evidencing other production patterns that differed from those established for the production of Spanish noun phrases with articles, further analyses were performed. In particular, incidence of omissions of articles when these were obligatory (e.g., Aquí está casa/Aquí está la casa), and the use of an English definite and indefinite article with a Spanish noun (e.g., the casa - a casa) were explored. Across all sessions, both Victoria and Beatriz failed to evidence omission of a feminine or masculine article though omissions occurred for the neutral article lo. In addition, the use of an English article with a Spanish noun was not observed in either of the children. All error patterns noted in the children's production of Spanish article + noun and article + noun + adjective NPs concerned gender agreement.
Monolingual Data
Because the incidence of gender errors in the children was never higher than 20%, and because Beatriz demonstrated a small percentage of errors in her corpora (i.e., no higher than 6%), the spontaneous speech data from 24 monolingual, Spanish-speaking children living in Puerto Rico were analyzed for the presence of NP gender agreement errors. The children were subjects of a previous study by the author on pronoun development in typical learners (Anderson 1998a, 1998b), and were between the ages of 2 years 0 months and 3 years 11 months. Eight of the children evidenced gender errors, the incidence ranging from 1.4% to 5.7%. All these children were between the ages of 2 years 1 month and 2 years and 11 months. No 3-year-old child produced incorrect gender agreement, thus suggesting, as in previous research with monolingual Spanish speakers, that NP gender agreement is acquired early.
The purpose of the present investigation was to obtain initial information on bilingual Spanish English-speaking children's use of NP gender agreement morphology across time. Of particular interest was to examine how L1 loss may impact this aspect of Spanish inflectional morphology, and how the patterns observed in children who are experiencing L1 loss coincide with those observed in previous reports. Results from analyzing the NP data of two children suggest that NP gender agreement morphology may be vulnerable to language loss, but that this may vary among children. It was observed that one child experienced an increase of agreement errors in her spontaneous production of Spanish across time, while the other evidenced minimal errors. These initial data thus suggest that a reduction in Spanish input with a resulting decrease in the productive use of the language may impact children's use of NP gender agreement morphology. They further posit the possibility that the contradictory data on gender agreement errors in children with SLI may be due to differences in linguistic environment (i.e., monolingual versus bilingual). These differences impact the relative frequency of input and productive use of the L1, which in turn may result in weaker lexical representations or difficulties in accessing the target lexical form (Bybee, 1985, 1995). The end result will be that those aspects of the lexicon that do not impact meaning, such as grammatical gender, will be more difficult to access or may be lost. A discussion of how the present data compare to previous data in the area of NP gender agreement, as well as how child-specific factors may influence productive skill in the L1, follows.
Patterns of NP Gender Agreement and Comparison with Previous Research
The results of the present investigation suggest that gender agreement may be vulnerable to language loss. Furthermore, the direction of error, especially in the child who evidenced the most significant loss across time, appears to be one where the default masculine gender is used in instances where the noun has a feminine grammatical gender. Interestingly, data on adult Spanish-English transitional bilinguals reported by Lipski (1993) corresponds to what was observed in the present corpora. Lipski defines transitional bilinguals as individuals who demonstrate passive abilities (i.e., comprehension) as similar to those of native speakers of the target language but whose productive (i.e., speaking) skills are limited. These individuals usually learned the target language during childhood but, due to the minority status of the language, evidenced L1 loss. Lipski reports that NP gender agreement errors were quite frequent in the adult transitional bilingual's corpora. Specific direction of the error patterns was not given, though all examples provided suggest that it is toward the default masculine form.
The similarities observed in the adult data reported by Lipski (1993) and the present data support the hypothesis that reduction in use as well as input negatively impacts L1 grammatical skill. The question arises as to why NP gender agreement may be affected by loss. A plausible explanation is that the reduction in L1 input and output limits an individual's access to the L1's lexicon. As lexical loss is a common and early consequence of L1 loss (Schaufeli, 1992; Weltens & Grendel, 1993), it can be expected that certain features of the lexicon may be especially vulnerable. According to Bybee (1985, 1995), the various features that define a term, including descriptors such as tense, aspect, person, and gender, vary on how strongly they are affixed to the meaning or conceptualization of the lexeme/word. Those that are not strongly fixed to the word's meaning are more vulnerable to the effects of frequency in the input, as well as the effects of use. This characteristic of lexical representations will predict, then, that grammatical gender will be impacted by these changes in input and output, a prediction that is supported by the present data. The fact that Victoria evidenced a downward trend in the number of nouns used suggests that lexical loss may have been occurring as well. The negative correlation between number of words and incidence of errors observed in the children's corpora further support the presence of lexical loss across time.
Although the data gathered in this study, especially that of Victoria, correspond with what has been reported in the literature on L1 loss in typical learners (Lipski, 1993), they also pattern what has been noted in the spontaneous speech of Spanish-speaking children living in the United States and who have been diagnosed as having atypical language skills (i.e., SLI) (Restrepo et al., 1996). Because this has not been reported in previous research with monolingual children with SLI who have as a first language similar NP gender agreement rules (article-noun agreement), such as Italian (Bortoloni et al., 1997), and has not been noted in this investigator's extensive work with monolingual Spanish-speaking children with atypical language skills, it appears that the deficits noted result from the bilingual context and not from the atypical learning patterns per se. The fact that in Restrepo et al.'s (1996) study the SLI children evidenced a higher frequency of NP gender agreement errors, specifically in article use, than did typical Spanish-speaking peers in similar (i.e., bilingual) environments, may result from the possibility that children with SLI may be more vulnerable to L1 loss/attrition than typical learners. This, in turn, will result in a higher incidence of NP gender errors as compared to matched peers within the same linguistic environment. Comprehensive longitudinal data on L1 skills in both child populations is necessary to ascertain the relative effect that the particular language contact situation has on each group of children.
When contrasting the error data obtained from Victoria and Beatriz's productions to that of Restrepo et al. (1996), two important differences emerge. In Restrepo's data (Restrepo, personal communication, November 20, 1998), the gender errors occurred more frequently in the production of the masculine article el. The article la was used in place of the article el (e.g., la carro/el carro, "the car"), the direction being from masculine to feminine. The present data suggest the opposite error direction, one that is from feminine to masculine. The former error pattern corresponds to what has been observed in the early stages of article acquisition in monolingual Spanish-speaking children (Clark, 1985). The latter correspond to what has been observed in the adult L1 loss literature (Lipski, 1993). A second difference was in the incidence of other errors, in particular, omissions. The children with SLI studied by Restrepo (personal communication, November 20, 1998) showed omission of articles, both definite and indefinite. Interestingly, the data from Victoria and Beatriz showed very limited use of this error pattern, and was usually observed with the neutral article lo. Article omission has been noted as a salient characteristic of children with SLI (Bortolini et al., 1997). These two differences suggest that perhaps what distinguishes typical from atypical learners in a minority language situation is not the amount of gender agreement errors, but the type of errors produced.
Both Restrepo's and the present data suggest variability among children in the incidence of NP gender agreement errors. In the present investigation, Beatriz showed few gender agreement errors, while Victoria demonstrated an increase of these with increased exposure to the L2. The children with SLI studied by Restrepo also varied in a similar manner, with the number of errors ranging from 1 to 64 in the children's language sample. The typical language learners in her study demonstrated an incidence of error productions that ranged from 0 to 8 (Restrepo, personal communication, November 20, 1998). This variability in performance warns us from utilizing frequency of occurrence in the corpora as a marker of true disability and not L1 loss.
Individual Differences in Performance
The fact that there were differences between the two children with respect to relative frequency and error patterns in the use of NP gender agreement suggest that child-specific factors impact this aspect of the grammar under situations of L1 loss. Clinically and educationally, it points to the need for obtaining a careful history of language use when assessing language skills in children who may be experiencing L1 loss. For Victoria and Beatriz, differences in initial Spanish skill prior to being exposed to English and relative use of the L1 may explain the variable performance observed.
The children differed in age of exposure to English. In fact, if McLaughlin's (1984) age division is followed, Victoria could be described as a simultaneous bilingual, while Beatriz would be identified as a sequential bilingual. This age difference resulted in each child being in contact with English at different levels of Spanish skill. It is plausible that because Victoria was exposed to English at an earlier age, acquisition of NP gender agreement morphology in Spanish was still occurring. In addition, earlier consistent exposure to English results in an earlier reduction of Spanish input, which in turn can affect lexical development, as this is highly dependent on experience (Schaufeli, 1992; Umbel & Oller, 1995). The consequence would be that the child's system was thus more vulnerable to loss. The data do suggest correct use of gender morphology during the initial taping sessions, thus this cannot be considered a case of arrested development, but of loss. Nevertheless, "lower" skills in the L1 at the moment of introduction of the L2 may be a variable that impacts the rate of L1 loss. Although Victoria's Spanish skills as assessed by her parents prior to beginning the data collection procedure were commensurate to those of monolingual children her age, the fact remains that she was exposed to English at an earlier age. In addition, unlike Beatriz, she did not have the opportunity for consistent peer interaction in the L1. These factors may have made her more vulnerable to the effects of a minority language contact situation on L1 skills.
Relative use of Spanish during interactions with the parents and other Spanish-speaking adults differed across the children. From parental input it was noted that Beatriz used Spanish while interacting with Spanish-speaking adults more frequently than Victoria. During the time when the data were collected, the parents reported that Beatriz frequently verbalized that she enjoyed speaking Spanish and that she wanted to develop Spanish literacy skills. On many occasions she also indicated her desire to return to Puerto Rico. Victoria, on the other hand, during the data gathering period showed an increased reticence to speak in Spanish both to her parents and to other Spanish-speaking adults. Although this may be associated to differences in relative Spanish skill, it resulted in lower frequency of Spanish use by Victoria, as compared to her sister. Following Bybee's (1985, 1995) model for morphological use, reduced use of a form impacts its access. As a result, this reticence to speak Spanish impacted her opportunities to use the language productively, thus possibly moving toward what Lipski (1993) has described as transitional bilingualism.
Conclusions and Directions for Future Research
Because of the nature of the data, the results obtained from the present investigation need to be seen as tentative in nature. The data support Bybee's (1985, 1995) conceptualization of lexical and morphological representations and how frequency may impact productive use of various grammatical markers. They also suggest that NP gender agreement errors may not be indicative of language learning deficits per se, but of typical patterns associated with L1 loss, and that perhaps what differentiates typical from atypical learners are the types of errors. Future research should focus on developing specific procedures to assess productive use of gender agreement morphology in both monolingual Spanish and bilingual children with various levels of Spanish proficiency and with various language histories. Included in these groups should be children who evidence both typical and atypical language skills.
Anderson, R. (1995). Spanish morphological and syntactic development. In H. Kayser (Ed.), Bilingual speech-language pathology: A Hispanic focus (pp. 41-74). San Diego, CA: Singular.
Anderson, R. (1998a). The development of grammatical case distinctions in the use of personal pronouns in Spanish-speaking preschoolers. Journal of Speech, Language and Hearing Research, 41, 394-406.
Anderson, R. (1998b). The use of reflexive constructions by Spanish-speaking children: Differences across functions. Applied Psycholinguistics, 19, 489-512.
Anderson, R. (1999). First language loss: A case study of a bilingual child's productive skills in her first language. Communication Disorders Quarterly, 21, 4-16.
Bayley, R., Alvarez-Calderón, A., & Schechter, S. R. (1998). Tense and aspect in Mexican-origin children's Spanish narratives. In E. V. Clark (Ed.), The proceedings of the twenty-ninth annual child language research forum (pp. 221-230). Stanford, CA: Center for the Study of Language and Information.
Bortolini, U., Caselli, M. C., & Leonard, L. B. (1997). Grammatical deficits in Italian-speaking children with specific language impairment. Journal of Speech, Language and Hearing Research, 40, 809-820.
Brisk, M. E. (1974). New Mexican Spanish syntax. In G. D. Bills (Ed.), Southwest area linguistics. San Diego, CA: Institute for Cultural Pluralism, San Diego State University.
Bybee, J. (1985). Morphology: A study of the relation between meaning and form. Amsterdam: John Benjamins.
Bybee, J. (1995). Regular morphology and the lexicon. Language and Cognitive Processes, 10, 425-455.
Clark, E. V. (1985). The acquisition of Romance, with special reference to French. In D. E. Slobin (Ed.), The crosslinguistic study of language acquisition: Vol. 1. The data (pp. 687-782).
Connell, P. J., & Stone, C. A. (1992). Morpheme learning of children with specific language impairment under controlled conditions. Journal of Speech and Hearing Research, 35, 844-852.
Craig, H. K. (1996). Specific language impairment: A changing role for normal developmental language theories during the 1990s. In M. D. Smith & J. S. Damico (Eds.), Childhood communication disorders (pp. 141-156). New York: Thieme.
Dromi, E., Leonard, L. B., & Shteiman, M. (1993). The grammatical morphology of Hebrew-speaking children with specific learning impairment: Some competing hypotheses. Journal of Speech and Hearing Research, 36, 760-771.
Fillmore, L. W. (1991). When learning a second language means losing the first. Early Childhood Research Quarterly, 6, 323-346.
Gal, S. (1989). Lexical innovation and loss: The use and value of restricted Hungarian. In N. C. Dorian (Ed.), Investigating obsolescence: Studies in language contraction and death (pp. 31-331). Cambridge, MA: Cambridge University Press.
Gariano, C. (1984). El aprendizaje del género en español (The learning of gender in Spanish). Hispania, 67, 609-613.
Gutiérrez, M. (1990). Sobre el mantenimiento de las cláusulas subordinadas en el español de Los Angeles (The maintenance of subordinate class in Los Angeles' Spanish). In J. J. Bergen (Ed.), Spanish in the United States: Sociolinguistic issues (pp. 31-38). Washington, DC: Georgetown University Press.
Hansson, K., & Nettelbladt, U. (1995). Grammatical characteristics of Swedish children with SLI. Journal of Speech and Hearing Research, 38, 589-599.
Harlan, J. (1991). Bilingualism in the United States. NY: Watts.
Hernández-Pina, F. (1984). Teorías psico-sociolingüísticas y su aplicación a la adquisición del español como lengua materna (Psycho-sociolinguistic theories and their application to the study of the acquisition of Spanish as a first language). Madrid: Siglo XXI.
Hudson-Edwards, A., & Bills, G. D. (1982). Intergenerational shift in an Albuquerque barrio. In J. Amastae & L. Elías Olivares (Eds.), Spanish in the United States: Sociolinguistic aspects. Cambridge, MA: Cambridge University Press.
Kaufman, D. (1998). Children's assimilatory patterns and L1 attrition. BUCLD 22 Proceedings, 409-420.
Kaufman, D., & Aronoff, M. (1991). Morphological disintegration and reconstruction in first language attrition. In H. W. Seliger & R. M. Vago (Eds.), First language attrition (pp. 175-188). Boston, MA: Cambridge University Press.
Klein, P. W. (1989). Spanish "gender" vowels and lexical representation. Hispanic Linguistics, 3, 147-162.
Kravin, H. (1992). Erosion of a language in bilingual development. Journal of Multilingual and Multicultural Development, 13, 307-325.
Kuhberg, H. (1992). Longitudinal L2-attrition versus L2-acquisition, in three Turkish children: empirical findings. Second Language Research, 8, 138-154.
Leonard, L. B., Sabbadini, L., Volterra, V., & Leonard, J. S. (1988). Some influences on the grammar of English- and Italian-speaking children with specific language impairment. Applied Psycholinguistics, 9, 39-57.
Linares, N. (1983). Rules for calculating mean length of utterance in morphemes in Spanish. In J. G. Erickson & D. O. Omark (Eds.), Communication assessment of the bilingual, bicultural child: Issues and guidelines (pp. 291-295). Baltimore, MD: University Park Press.
Lipski, J. M. (1993). Creolid phenomena in the Spanish of transitional bilinguals. In A. Roca & J. M. Lipski (Eds.), Spanish in the United States: Linguistic contact and diversity (pp. 155-182). New York: Mouton de Gruyter.
Martínez, E. A. (1993). Morphosyntactic erosion between two generational groups of Spanish-speakers in the United States. New York: Peter Lang.
Martínez, I. M., & Shatz, M. (1996). Linguistic influences on categorization in preschool children: A cross-linguistic study. Journal of Child Language, 23, 529-545.
McLaughlin, B. (1984). Second language acquisition in childhood: Vol. 1. Preschool children. Hillsdale, NJ: Erlbaum.
Merino, B. J. (1983). Language development in normal and language handicapped Spanish-speaking children. Hispanic Journal of Behavioral Sciences, 5, 379-400.
Miller, J. F. (1981). Assessing language production in children: Experimental procedures. Boston, MA: Allyn & Bacon.
Natalicio, D. (1983). Native speaker intuitions as a basis for determining noun gender rules in Spanish. Southwest Journal of Linguistics, 6, 49-55.
Ocampo, F. (1990). El subjuntivo en tres generaciones de hablantes bilingües (The subjunctive in three generations of bilingual speakers). In J. J. Bergen (Ed.), Spanish in the United States: Sociolinguistic issues (pp. 39-48). Washington, DC: Georgetown University Press.
Orellana, M. F. (1994, April). Superhuman forces: Young children's English language acquisition and Spanish language loss. Paper presented at the Annual Meeting of the American Educational Research Association, New Orleans, LA.
Pfaff, C. W. (1991). Turkish contact with German: Language maintenance and loss among immigrant children in Berlin (West). International Journal of the Sociology of Language, 90, 97-129.
Restrepo, M. A. (1997). Grammatical errors in Spanish-speaking children with language impairment who are learning English as a second language. Paper presented at the International Symposium on Bilingualism, Newcastle upon Thyme, England.
Restrepo, M. A. (1998). Identifiers of predominantly Spanish-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research, 41, 1398-1411.
Restrepo, M. A., Chasteen-Bustelo, W., & Mattute, C. (1996). Article use in Spanish-speaking children with normal language and language impairment. Poster session presented at the annual convention of the American Speech-Language-Hearing Association, Seattle, WA.
Rice, M. L., & Wexler, K. (1996). Toward tense as a clinical marker of specific language impairment in English-speaking children. Journal of Speech and Hearing Research, 39, 1239-1257.
Roca, I. M. (1989). The organization of grammatical gender. Transactions of the Philological Society, 87, 1-32.
Schaufeli, A. (1992). A domain approach to the Turkish vocabulary of bilingual Turkish children in the Netherlands. In W. Fase, K. Jaspaert, & S. Kroon (Eds.), Maintenance and loss of minority languages (pp. 117-135). Amsterdam: John Benjamins.
Schiff-Myers, N. B. (1992). Considering arrested language development and language loss in the assessment of second language learners. Language, Speech, and Hearing Services in Schools, 25, 156-164.
Seliger, H. (1989). Deterioration and creativity in childhood bilingualism. In K. Hyltenstam & L. K. Obler (Eds.), Bilingualism across the lifespan: Aspects of acquisition, maturity and loss (pp. 185-201). Cambridge, MA: Cambridge University Press.
Silva-Corvalán, C. (1991). Spanish language attrition in a contact situation with English. In H. W. Seliger & R. M. Vago (Eds.), First language attrition (pp. 151-171). Boston, MA: Cambridge University Press.
Teschner, R. V. (1987). Noun gender categories in Spanish and French: Form-based analyses and comparisons. In L. Studerus (Ed), Current trends and issues in Hispanic linguistics (pp. 81-107). Dallas, TX: The Summer Institute of Linguistics.
Turian, D., & Altenberg, E. P. (1991). Compensatory strategies of child first language attrition. In H. W. Seliger & R. M. Vago (Eds.), First language attrition (pp. 207-226). Boston, MA: Cambridge University Press.
Umbel, V. M., & Oller, D. K. (1995). Developmental changes in receptive vocabulary in Hispanic bilingual school children. In B. Harley (Ed.), Lexical issues in language learning (pp. 59-80). Amsterdam: John Benjamins.
Veltman, C. (1988). The future of the Spanish language in the United States. NY: Hispanic Policy Development Project.
Weltens, B., & Grendel, M. (1993). Attrition of vocabulary knowledge. In R. Schreuder & B. Weltens (Eds.), The bilingual lexicon (pp. 135-156). Amsterdam: John Benjamins.
Zentella, A. C. (1997). Growing up bilingual. Malden, MA: Blackwell.