Transfer and semantic universals in the L2 acquisition of the English article system by child L2 learners

ABSTRACT

L2 learners’ problems with English articles have been linked to learners’ L1 and their access to universal semantic features (e.g., definiteness and specificity). Studies suggest that L2 adults rely on their L1 knowledge, while child L2 learners rely more on their access to semantic universals. The present study investigates whether child L2 learners access features like adult L2 learners (i.e., reassembling the features of the L1) or as L1 acquirers (i.e., choosing from the inventory of universal features). We evaluated the acquisition of English articles in 30 Spanish-speaking children learning English as their L2 and found that they transfer knowledge from their L1 much in the same way as do adults. However, the error analysis also suggests that article acquisition in child L2 learners is restricted by Universal Grammar resembling L1 acquisition. In conclusion, we propose that article acquisition in child L2 learners patterns with both adult L2 and L1 acquisition.

Notes

¹ We assume that access to Universal Grammar and explicit learning mechanisms can coexist in the adult learners’ interlanguage and, potentially, in child L2 learners’. As Ionin, Zubizarreta & Philoppov (2009) showed, adults have access to universal principles and at the same time use explicit learning strategies when building their grammar.

² A point to consider is that the preference for specificity rather than definiteness could also be related to a more general problem having to do with an issue of interface delay; in our case, it could be due to a difficulty to integrate information at the syntax-discourse interface. This is a well-documented phenomenon in L1 acquisition (Avrutin 1994; Thorton & Wexler 1999; Grinstead 2004), bilingual L1 acquisition (Serratrice, Sorace & Paoli 2004; Sorace & Serratrice 2009), and L2 acquisition (Sorace & Filiaci 2006; Belletti, Bennati & Sorace 2007). Particularly relevant in this area have been the works of Sorace and colleagues that led to the proposal of the Interface Hypothesis (Sorace & Filiaci 2006). This hypothesis is based on a modular view of the language architecture (Fodor 1983). Two types of interfaces are distinguished on this model: internal and external interfaces. The former are those in which linguistic modules interface with each other (e.g., syntax-semantics), whereas the latter are those in which linguistic modules interface with cognitive modules outside of the language faculty (e.g., syntax-semantics). The external interfaces are the ones that have been predicted to cause acquisitional delays and challenges. Although it is feasible that a delay at the syntax-pragmatics interface could cause the pattern of error that we see in this study, confirming this hypothesis would require further testing. We would need to test the same children on a purely syntactic property and on pragmatic concepts independent of syntax. If children showed adultlike behavior in a narrow syntactic property and show that they understand pragmatic concepts (irrespective of syntax), then we could assume that these children have acquired syntactic and pragmatic properties but still have trouble integrating their knowledge of syntax with information that lies out of the language proper.

³ Although Ionin’s (2003) Fluctuation Hypothesis is expressed in terms of parameter resetting, in this article, we will interpret our results in terms of features like Hawkins et al. (2006). If children are able to reset the Article Choice Parameter (in Ionin’s terms), we will take it as evidence that they are able to reassemble the features from their L1 in a way that matches the feature configurations of the L2 (Lardiere 2004). We will refer to this as access to the features of [+/–specificity] and [+/–definiteness].

⁴ Given the high accuracy of the children tested in this experiment, it is possible that they had already established the article paradigm in their L2 grammar. Therefore, a longitudinal study or testing participants with varied level of skill in the L2 would be necessary to confirm our claim.

⁵ However, this group also included the youngest children. In early immersion programs it is common for those children with the least amount of exposure to be also the youngest ones because they have started school more recently than the oldest children who were at higher grade levels. As a consequence, the children with least exposure might have lower reading skills than those at higher grades whose reading skills might be further developed. Therefore we are not able to confirm if poorer performance in this group was due to them being affected by the specificity of the noun or whether their reading skills played a role. In addition this group was rather small (n = 6), and generalizations are difficult to draw.

⁶ None of the children was attending the school on a scholarship.