Gender Agreement in Spanish Heritage and L2 Grammars: When Age Meets Task and Context of Acquisition
Silvina Montrul
(with Dan Thornhill, Rebecca Foote, Silvia Perpiñán and Susana Vidal)
4PM, Thursday, March 2, 2006
Lucy Ellis Lounge, FLB
In this study, I compare the linguistic behavior of L2 learners (late bilinguals) and heritage speakers (early bilinguals) of low-to-intermediate proficiency in Spanish in the area of gender agreement in DPs (la casa nueva ‘the-fem house-fem new-fem’, el auto viejo ‘the-masc car-masc old-masc’). While normally developing monolingual Spanish speaking children acquire gender marking and agreement completely by age 4 (with 100% accuracy), gender marking and agreement cause persistent difficulty to adult L2 learners of gender-marking languages at all proficiency levels (Carroll 1989, Franceschina 2001, Hawkins and Franceschina 2004), especially if their L1 does not mark gender. According to some current theories of L2 competence, incomplete L2 acquisition in this area (and others) is consistent with the unavailability of aspects of Universal Grammar due to age (Bley-Vroman 1989, Meisel 1997, Hawkins and Chan 1997) or a critical period. The problem with this explanation is that gender agreement also appears to be acquired incompletely by heritage speakers, who were exposed to Spanish at home in childhood (before age 5), attended school in English, and have stronger command of English. If L2 learners cannot attain native-like knowledge due to age, one would expect heritage speakers to be better than L2 learners in this domain.
To date, the majority of studies testing gender agreement in L2 acquisition have mostly relied on oral production data. By contrast, in this study I used multiple measures to test recognition, interpretation and production of gender morphology. Nineteen monolingual Spanish speakers, 22 L2 learners, and 22 early bilingual speakers of low-intermediate proficiency (determined by a proficiency test) completed a written interpretation task with pictures (IT), a written morphology recognition task (MRT), and an oral picture naming task (PNT). The IT and the MRT tested comprehension and recognition of gender morphology, whereas the PNT tested production of correct gender agreement between nouns, determiners and adjectives.
Theories of L2 competence which assume unavailability of gender features predict poor performance across a variety of tasks (Carroll 1989, White et al. 2004). Results showed differences between L2 learners and heritage speakers. Contrary to what age-based theories predict, the L2 learners were significantly better than the heritage speakers in the two written tasks. However, there was a significant advantage for heritage speakers in the oral task, in accordance with the age explanation. Clearly, both types of bilinguals have developed some kind of linguistic knowledge of gender in Spanish, but the comprehension-production dissociation found with the L2 learners and the heritage speakers suggests that the nature of knowledge might be different (explicit/metalinguistic vs. implicit) not only due to age but also to nature of input and mode of acquisition (naturalistic vs. instructed). These results challenge current generative theories of L2 competence, which do not take into account whether context of acquisition and type of input (oral vs. written) lead to different types of linguistic knowledge (Paradis 2004).