Late L2ers can acquire grammatical features that do not occur in their L1: Evidence from the effect of animacy on verb agreement in L1 Chinese.

Second language (L2) learners often have persistent difficulty with agreement between the number of the subject and the number of the verb. This study tested whether deviant L2 verb number agreement reflects maturational constraints on acquiring new grammatical features or resource limitations that impede access to the representations of L2 grammatical features. L1-Chinese undergraduate students at three age of arrival (AoA) levels were tested for online verb agreement accuracy by completing preambles in three animacy combinations: animate-inanimate [AI; e.g., The officer(s) from the station(s)], inanimate-animate [IA; e.g., The letters from the lawyer(s)], and inanimate-inanimate [II; e.g., The poster(s) from the museum(s)]. AI should be less costly to process than IA or II sequences, because animacy supports the subject in AI but competes with the subject for control of agreement in IA sequences, and is neutralized in II. Agreement accuracy was greater overall for AI than for IA or II, and although an AoA-related increase in erroneous agreement after plural subjects occurred for IA and II, there were no AoA effects for AI. Higher scores on memory tasks were associated with greater agreement accuracy, and the memory tasks significantly predicted variance in erroneous agreement when AoA was partialed out. The fact that even late learners can do verb agreement in the case of AI demonstrates that they can acquire new grammatical features. The greater difficulty with agreement in the case of IA or II than of AI, in conjunction with the results for the memory tasks, supports the resource limitation hypothesis.