Abstract
The article by Eichler, Jansen, and Müller presents a genuine and impressive contribution to the study of bilingual first language acquisition in general and to the analysis of gender acquisition in particular. Analyzing the transparency of the gender systems in different language pairs (French–German, Italian–German, Spanish–German, and Italian–French) and looking at different acquisition paths (i.e. balanced and unbalanced acquisition), the authors claim that not only gender rules but also individual strategies in the number of noun types acquired affect the accuracy of gender assignment in child speech. Moreover, the study does not corroborate (Eichler, 2012), as happened in former studies, that unbalanced language acquisition tends to result in not acquiring specific linguistic aspects (here: gender; see however the discussion on neuter in German in section 3.3). It is rather claimed that the low amount of production (here: number of noun types in the weaker language)—and maybe of input—might lead to different, incomplete analyses (here: to a low accuracy in gender assignment).
The authors provide a good overview of previous studies on the acquisition of gender in monolingual and bilingual contexts, arguing that Romance languages are more transparent and thus easier to acquire when it comes to gender assignment. In fact, the reliability of gender cues is shown to differ between the languages under discussion, Spanish and Italian having the most transparent rules, followed by French and, finally, German (see also Eichler, 2012). It seems that gender marking (and whether case and number are also marked and thus co-occur with gender in determiner phrases) affects the acquisition of gender as well.
In sum, the present article suggests that it is worth having a closer look at the individual development of the two languages in bilinguals (in this study, the production of noun types in each child has been explored), for it might explain delays in acquisition despite the transparency of the gender system. A reduced input, and therefore a reduced lexicon, seems to affect the gender accuracy rates.
Future studies on both monolingual and bilingual development in different acquisition contexts, including languages with no gender (e.g. Turkish) and those without articles, which mark gender on nouns (Arabic languages, Russian), might promote the idea that not only the languages involved can predict inaccurate gender assignment but also the acquisition context—primarily the amount of noun types acquired and uttered.
Eicher et al.’s results are helpful to shed light on further research issues, such as the incomplete acquisition of languages in the context of migration. As Montrul (2008) has shown in her book, in which former and new studies are discussed, different results will be reached at the end state of acquisition depending on the acquisition settings. Early bilingualism in children might lead to the successful mastery of both linguistic systems but could also result in the mastery of only one and to the incomplete acquisition of the other system. This is more likely to occur in the well-studied strategy “one person-one language” than in other situations because only one parent is able to support the minority language. Acquiring two languages in the context of migration on the other hand—according to Montrul—might result in the successful acquisition of both languages given that both parents provide input in the minority language. These findings are very interesting and have a huge sociopolitical impact: not only do they revalue the acquisition setting “one language at home–one language outside” (for an overview of multilingual acquisition settings, see Romaine, 1995) and thus the importance of supporting heritage languages. These results also encourage educational programs such as the dual-language one and thus enforce bilingual programs.
As, for example, Mueller Gathercole (2002) has shown studying Spanish–English children in the United States, less gender agreement errors are stated in Spanish in those children who have the greatest amount of input in Spanish. This is the case for the children who speak Spanish at home and attend a dual-immersion program, whereas the bilinguals who only speak English at home do worse in Spanish, which is not surprising. The most interesting finding is that the bilinguals who speak English and Spanish in their families show more gender agreement errors than those who are exposed to Spanish only at home. Hence, the family strategy “one person-one language” seems to be counterproductive in some way, in that it could lead to a less successful—and thus to an incomplete—acquisition of the minority language because the two languages are acquired simultaneously and compete in the daily input of the bilingual children.
At this point, some clarification is inevitable: it is for sure not my (and neither Montrul’s nor Mueller Gathercole’s) intention to devaluate bilingual education, saying that it does not lead to a successful mastery of both languages or to plead for a time-on-task approach and suggesting not to focus on the minority language but to exclusively concentrate on the acquisition of the majority language. The main issue I want to highlight here is that extralinguistic factors play a crucial role in a child’s language acquisition. Thus, research should elaborate on these factors in order to support bilingual education. In very general terms, the main challenge for bilingual education is to ensure that both languages play an important role in the child’s daily life—a task that seems easier to master when both parents speak the same language (no conflict of input time at home) and when schools encourage and foster bilingual education (the status of both languages is enhanced).
I want to address another outcome of Eichler et al.’s work: stating a correlation between the path of acquisition (i.e. determining a stronger and a weaker language in early development) and the successful final state of acquisition might have a huge impact on both parents’ guidebooks and nursery school educators’ training. As a matter of fact, we know from several studies on the early second language acquisition of German that—whereas some aspects of language are acquired relatively fast and in a monolingual fashion, for example, word order (see, among others, Thoma & Tracy, 2006; Tracy, 2007)—inflection in the determiner phrase causes serious problems for young children who are only exposed to German from approximately age 3 onward. Following Eichler et al.’s proposal, not only the languages under discussion can explain delays or problems in acquisition but also the amount of input the children are exposed to, which in turn results in whether their languages can be attributed to be “stronger” or “weaker” (for the criteria used in order to accurately measure language imbalance, please see Eichler et al., section 3.3, as well as Cantone, Kupisch, Müller, & Schmitz, 2008). Hence, mere contrastive analyses, as sometimes applied in the field of education or descriptive linguistics (predicting negative transfer on a theoretical ground for certain language pairs), are not sufficient. In contrast, more detailed analyses on the development of the two languages in young bilinguals are necessary in order to predict which language might be considered the weaker and to annul the differences in language input as early as possible.
Eichler et al.’s approach (as well as latest theoretical approaches that assign an important role to the lexicon) raises some questions with respect to the well-known topics such as “How much input is enough for ensuring the acquisition of a language?” “What do we know about the vocabulary size in bilinguals?” and “How is it possible that schools still treat bilingual children like monolinguals and therefore expect their vocabulary to be of the same size and to comprise exactly the same words as monolinguals?”
With respect to the first question, we surely need more studies on the precise amount of input in bilingual acquisition. Given that our modern society facilitates all sorts of technical communication, one might think about more detailed ways of collecting data than the actual established “every fortnight” recordings. The second point addresses the analysis of the bilinguals’ two lexicons: it is reasonable to assume that bilinguals presumably have a bigger lexicon than monolinguals given that they are exposed to two languages. Let us assume that a monolingual child disposes of approximately 6000 words at age 6, and let us further assume that we analyzed the content of the lexicons in a German–Spanish bilingual child called Luna, raised in Germany. We could expect Luna to productively use—let us say—9000 words. Nevertheless, we have to divide the 9000 words in two lexicons; in other words, if Luna had a balanced language acquisition, she might possess 4500 words in each language, whereas a child with an unbalanced development (let us call him Juan) would have 6000 words in one and 3000 in the other language at his disposal. Now imagine Juan, who lives in Germany, knows 6000 in the minority language Spanish and only 3000 in German, the majority/school language in a monolingual setting: no one will be interested in the age-based development in his Spanish but only in the less developed school language German! Even in Luna’s case, if we were able to count each word of her vocabulary, she also would lack 1500 in German compared to her monolingual peers. It is consensus that the school should be able to adjust the children’s vocabulary “lack” in German, and it is reasonable to support the less developed language at this point, in order to make sure that the conditions favor the child’s educational achievement. But what about the well-developed minority language? At the present time, too many bilingual children do not experience that their other first language is accepted and integrated into school (see Cantone, 2011).
Another issue in further studies on the quantity of input needed might include the analysis of the quality of input of the main caretakers, given that in both strategies—the “one person-one language” and the “one language at home-one language outside”—different topics and habits in the single languages might cause an unequal amount of noun types. For the sake of a quick, unproblematic and successful acquisition, caretakers might thus be instructed to provide a high variation of noun types in the input.
To conclude, there is still a long way to go to assure the best conditions for a successful bilingual acquisition, and there is further need for serious and continuative studies in this area.