What does the acquisition of the determiner phrase features reveal about bilingualism and L2 acquisition? Some introductory remarks

Overview of the contributions to this volume

The first two articles deal with child bilingualism. In particular, the study by Eichler, Jansen and Müller evaluates gender acquisition within DPs adopting Abney’s (1987) analysis comparing 2 monolingual German children and 17 bilingual children acquiring either a Romance language (i.e. French, Spanish, Italian) and German or two Romance languages simultaneously at different ages. The authors further compared the two languages within the bilingual children to one another with respect to the acquisition of gender. The influence of different variables on gender acquisition is also under scrutiny: namely, language dominance, transparency of gender marking and/or reliability of gender cues in the respective languages are hypothesized to influence the speed or delay of its acquisition. According to their main findings, bilingual children can gain the gender systems in both languages just as monolinguals do and bilingualism per se does not have a delaying effect contrary to what previous studies have claimed. In the case of the bilingual as well as in monolingual children, the acquisition of gender in German seems to be most problematic. French, on the other hand, embodies only slightly more problems than the other two Romance languages (i.e. Spanish and Italian), with the two-gender systems in both Spanish and Italian being acquired with ease. Eichler, Jansen and Müller came to the conclusions that, in both German and French, adult phonological gender rules are typified by their rather low validity. In addition, they claimed that the lower accuracy with German gender proves that the way in which gender is marked in the languages influences its acquisition; besides, in German (but not in French), gender marking is entangled with case and number markings. Although Eichler, Jansen and Müller’s results seem to indicate that gender accuracy depends on the language acquired, some children do not fit into the expected pattern of gender acquisition with respect to the investigated languages. More importantly, all of these children show a language imbalance. Hence, the authors claim that the children’s gender accuracy can be calculated on the basis of the language acquired, but language dominance can distort this ranking. Moreover, it is not the case that gender is delayed in the weak language of the bilingual children; language imbalance has the effect that the child may not tend towards the predicted side of an accuracy ranking (for the weaker language) (Müller & Pillunat, 2008). A further important finding of this study is that accuracy on neuter gender is lower in bilingual Germans than in monolingual Germans, hinting that the fact that the bilingual children acquire a two-gender system simultaneously with a three-gender system has a delaying effect for target-like neuter marking (Müller, 1995).

The next article by Larrañaga and Guijarro-Fuentes investigates the acquisition of gender features by two Basque–Spanish bilingual children compared to a Spanish monolingual child. Basque is a language that lacks gender features and nominal agreement (Artiagoitia, 2002), whereas Spanish classifies nouns into two classes, that is, masculine and feminine, and has DP internal agreement (Abney, 1987; Bernstein, 1993, 2001). The internal architecture of Basque and Spanish differs in two crucial ways; the presence or absence of agreement and the presence or absence of the syntactic projection ClassP (Picallo, 2008). Hence, the acquisition of gender would enlighten us with regard to the internal architecture of the DP. Given the fact that the studies on gender acquisition by Spanish monolinguals or bilinguals of any combination are not very numerous with the exception of, for example, the study by Barreña (1997), which is the only study that has previously looked at Basque–Spanish bilingual gender acquisition, one of the main objectives of this study is to provide an in-depth description of gender development and a thorough analysis of gender errors by a monolingual Spanish child from Aguirre’s corpus (http://childes.psy.cmu.edu) and two Basque–Spanish bilinguals from the Busde Corpus under investigation, which are both naturalistic Corpora. The three children were matched according to their mean length of utterances (MLUs). According to the main results from the data set, the masculine form is not the default gender contrary to previous research, neither for the monolingual child analysed nor for bilinguals; on the contrary, both groups tend to overgeneralize both masculine and feminine. In addition, none of the children solely employs the masculine with all nouns at a first stage to converge to target grammar in a subsequent stage. While Basque–Spanish bilinguals use masculine determiners with feminine nouns in the majority of linguistic contexts from the outset of language acquisition, the monolingual child acts the opposite way. More importantly, bilingual children need more time to acquire the niceties of Spanish gender. In other words, they make gender errors even at advanced stages of development, whereas the monolingual Spanish child studied in the present article presents a target-like gender performance from early on. Moreover, the analysed data illustrate that Basque ultimately influences Spanish resulting in language delay as a consequence of the internal architecture of the DP and to a lesser extent as a result of the surface overlap between Spanish and Basque (Müller & Hulk, 2001). Nevertheless, the authors’ interpretation is cautious because of the scarcity of such examples and the limited corpus available.

Drawing on previous studies (Sagarra & Herschensohn, 2010, 2011), Sagarra and Herschensohn sought to examine current approaches to grammatical representation and processing of L2 gender and number agreement with particular focus on Spanish DPs; such proposals claim incomplete acquisition and computation of L2 grammatical features (Hawkins & Hattori, 2006; Tsimpli & Dimitrakopoulou, 2007), whereas others advocate native-like representation and computation of new grammatical values (see Bruhn de Garavito & White, 2002; White, Valenzuela, Koslowska-Macgregor, & Leung, 2004, among others). Testing L2 English learners of Spanish at two different proficiency levels – beginning (n = 69) and intermediate (n = 64) – the authors assess comprehension data with a moving window paradigm (i.e. self-paced reading) together with a grammaticality judgment task. In particular, they investigate whether late learners whose L1 does not possess gender (i.e. English) can indeed attain native-like usage patterns of sensitivity to grammatical gender and number (features that exists in English) agreement violations compared to a group of 63 Spanish monolinguals. In this study, the authors also included other additional variables related to proficiency level (i.e. beginners vs. intermediates) and processing cost (namely, animate vs. inanimate nouns; gender vs. number agreement) into their two experiments. According to their data from moving window and grammaticality judgment tasks, intermediate learners, but not beginner learners, show similar qualitative reactions to Spanish monolinguals with regard to gender and number concord/discord distinctions, substantiating the importance and relevance of proficiency while suggesting native-like processing by L2 learners. They also prove that both grammatical (gender, number) and semantic (animacy) features differentially impact concord processing for both natives and L2 learners and that working memory plays a role in developing L2 processing skills.

Concluding the articles on L2 acquisition, the study by Cuza, Guijarro-Fuentes, Pires and Rothman explores the extent to which 16 advanced English-native L2 learners of Spanish come to gain restrictions on bare plural (BP) pre-verbal subjects compared to a group of 10 native speakers. In order to accomplish the main goal of this study, the authors consider L2 knowledge of available semantic readings of BPs and definite plurals (DefPs) in L2 Spanish where [+specific] and [+generic] interpretations are mapped differently when compared to English. Presupposing L1 transfer (Schwartz & Sprouse, 1996; White, 2003), and bearing in mind the subset/superset relationship of the two grammars (e.g. Manzini & Wexler, 1987) as stated by the Nominal Mapping Parameter (Chierchia, 1998), the learning task in this semantic domain does not seem to be a straightforward one. Target acquisition entails grammatical expansion and retraction; namely, Spanish DefP subjects imply the addition of an L1-unavailable reading [+generic] while a loss of an L1-available reading for pre-verbal subject BPs [+generic] is required. The data from a 24-item Context Felicitousness Task adapted from Slabakova’s (2006) previous study demonstrates that the 16 advanced learners of Spanish overcome this subset/superset learnability problem and, more importantly, that L2 learners of Spanish can attain semantic features despite the differences between their L1 and target grammar, which supports previous findings (e.g. Ionin, Montrul, & Crivos, 2009).

To sum up, this Special Issue is made of four original research articles unified by the same theoretical perspective but with diverse research objectives and large language pair combinations, with a broad range of participants (from beginners to advanced adult L2 learners to simultaneous bilingual children) and large-sized samples (e.g. 60+ participants in some of the articles included herein), and a range of methodologies such as on-line and off-line modalities which, I believe, ultimately provides a forum for further dialogue between researchers working on these diverse, but, at the same time, similar projects. Findings from all the articles would allow the pursuit of individually refined questions on these specific phenomena and would undoubtedly add significantly to overarching theoretical questions currently being discussed (e.g. the role of previous linguistic knowledge, the relative complexity for acquisition of particular structures, the role interfaces play in delays or possible non-convergence and so on). In sum, findings from this diverse group of studies will ultimately lead to improve inquiry within the same theoretical framework and will further our knowledge of the process of language acquisition and language use in general but, more importantly, will contribute to our understanding of how syntactic and semantic properties are acquired.

Future directions

Let us take, for instance, the implicit claim made in the article by Eichler, Jansen and Müller that individual differences/strategies in the number of noun types acquired affects gender accuracy in child speech, implying that the lower the number of types acquired, the better the child masters gender assignment. In my view, this smart and innovative argument could turn into a reliable factor for future studies of the lexicon of language balance and bilingual acquisition (and potentially early and late L2 acquisition) with a large number of participants where the individual differences in the amount of types in child speech are also taken into account. The findings by Larrañaga and Guijarro-Fuentes also raised a fashionable question and added to our knowledge that language distance/typology affects acquisition; that is, bilinguals with two languages that have conflicting systems or which lack parallel systems across their two languages experience different developmental patterns. This article is a nice addition to some existent ones on the discussion regarding acceleration versus potential delay of acquisition of certain structures (e.g. Paradis & Genesee, 1996). However, it remains to be investigated whether for some gendered languages with less formal form-function mappings, the semantic component is a crucial component of the acquisition process, especially at the early ages. What is interesting to note from Larrañaga and Guijarro-Fuentes’s data set is that in some cases features do not ‘spread’ to the whole DP. Hence, something has to be said about the nature of ‘concord’ or ‘agreement’ in phi features in the DP at a more theoretical level. A future study needs to address the status of gender with respect to logical form (LF) and syntax. In that respect, one could entertain the possibility of distinguishing between the notions of ‘valued/unvalued feature’ and ‘interpretable/uninterpretable feature’ as, for instance, in Pesetsky and Torrego (2007). As for the interpretation of their main findings, there is also the question of whether the problem is purely a processing one: that is, no problems with syntactic competence but problems with external performance components, a hypothesis gaining ground in current L2 acquisition studies. This assumption leads us to Sagarra and Herschensohn’s study and the necessity of future studies that could provide evidence for or against the accessibility/deficit model, as both predict some development at the early stages (see Liszka, 2009, for an overview of this line of thought). More work is also needed about the role of working memory in both acquisition and processing. It is still unclear (a) whether differences in working memory span would affect learners’ acquisition of certain grammatical features like gender marking but not others and (b) whether working memory would facilitate sensitivity to gender disagreement. More importantly, a link with the literature on speed/accuracy trade-offs needs to be made. However, what is clear is that L2 adult learners can acquire grammatical features (uninterpretable features) absent in their L1 contrary to some current L2 hypothesis (e.g. Interpretability Hypothesis). Moving from the acquisition of syntactic features to transfer knowledge and learnability in the realm of second language semantics (Gabriele, 2005, 2009), something left unexplored by Cuza, Guijarro-Fuentes, Pires and Rothman is the learning of the morphosyntactic component. Following Chierchia (1998), there is a morphosyntactic trigger for converging on the correct semantics. Importantly, another untouched aspect is the role of instruction in the learners’ success, that is, since the morphological paradigms are taught, how will this inform the acquisition of semantics? In relation to that, it would be desirable in future studies to explore whether the findings are due to linguistic knowledge or to some kind of metalinguistic strategies in the development of these interpretive properties. One other immediate question is as follows: is universal grammar (UG) accessibility excluded if convergence with the L2 is not achieved? This was the view put forward by Bley-Vroman (1990), among others. A recent debate, however, discussed the possibility of UG accessibility even if the outcome of the learning process is not convergence with the L2. For instance, Sorace (1993) talks about divergent or incomplete grammars in L2 acquisition. Lack of convergence with the L2 and its interpretation constitutes the focus of the discussion on representational deficit accounts (Hawkins & Hattori, 2006; Tsimpli & Dimitrakopoulou, 2007) versus computational problems (Lardiere, 1998, 2000; Prévost & White, 2000, White, 2003).

Thus, many questions and answers are unveiled, but many others remain open for future research in the DP domain and beyond including, but not limited to, the quest for symmetry in terms of acquiring the internal makeup structure of the DP and other phrases of different categorical types (e.g. verb phrase (VP), prepositional phrase (PP) and so on). This search for symmetry in generative acquisition can throw out interesting results, as it has been the case in theoretical terms (Abney, 1987). Given the overall findings of the studies included herein, the overall aim of the present volume was to further our insight into the general questions as to whether and how certain elements encoded in the nominal domain are acquired. All four articles offer interesting views and results.