Session 5G

Based on a comparison between previous studies on different types of Germanic learners of Spanish, it has been argued that the way in which temporal-aspectual features are represented in the learners’ L1 is decisive when it comes to acquiring the Spanish past tenses (see e.g. Domínguez et al. 2017, Slabakova & Montrul 2008 for comments on the feature reassembly). Interestingly, although with respect to the use of past tenses all Germanic languages differ significantly from the Spanish target-system, not all learners are reported to behave alike (McManus 2015, Authors in press). For example, whereas Dutch and English-speaking learners can benefit from the progressive tense of their L1, German learners do not have that option and tend to rely only on adverbials (Diaubalick & Guijarro-Fuentes 2019). In a similar vein, Dutch and English-speaking learners also behave differently (González & Quintana Hernández 2018). In Spanish, aspect has to be marked grammatically when speaking about past events (Zagona 2007). Whereas the Preterit (jugué ‘I played’) conveys perfective contexts, the Imperfect (jugaba ‘I was playing’/’I would play’) encodes imperfectivity. English and Dutch have a grammaticalized progressive tense, but only in English its use is obligatory in progressive contexts (Ebert 1996). German, in contrast, does not mark grammatical aspect at all (Heinold 2015). In this talk, we intend to build up on the recently presented claim that Dutch and German learners encounter different kinds of challenges. Based on the results drawn from a Grammaticality Judgment Task and a Sentence Completion Task, we have tested 70 Dutch and German learners as compared to 20 native speakers. Whereas both learner groups encounter difficulties, they do so in a different way: in those contexts, where grammatical and lexical aspect seem to be contradicting, Germans judge sentence based on temporal markers while Dutch learners take inherent aspectual information into consideration. In the Sentence Completion Task, conversely, this was not confirmed and both groups are easily distracted by misguiding markers. Although findings have to be confirmed by a larger sample size, the results show two things: first, Germanic learners are no homogeneous group. Second, there seems to be a task effect, since in production, Dutch learners are distracted by pedagogical rules (congruent with the Competing Systems Hypothesis, Rothman 2008). Importantly, such a distraction can only appear when the underlying aspectual features are acquired already, which seems to be doubtful for the German group.

Several theoretical proposals have claimed that L2 learners process morphosyntactic information (e.g. agreement) differently than native speakers. Within these accounts, the shallow structure hypothesis (Clahsen & Felser, 2006, 2017) maintains that late L2 learners compute less detailed morphosyntactic relations and that proficiency in the L2 does not modulate sensitivity to morphosyntactic discord relations. Previous studies have shown that the processing of agreement violations within constituents is a daunting task for L2 learners, as beginning and intermediate English L2 learners of Spanish show no sensitivity to the aforementioned relationship (Barber & Carreiras, 2005; Sagarra & Herschensohn, 2010). Some experimental investigations have tried to investigate structural distance (i.e. agreement across constituents) effects on gender agreement processing (Keating, 2009), although the experimental sentences also contained linear distance, thus resulting in confounded results. Thus, it remains unclear whether processing of agreement across constituents is more costly than the processing of agreement within a constituent, and whether late L2 learners can gain sensitivity to this hierarchically complex linguistic structure. The present study,  examines whether working memory and locality (i.e., agreement within or across constituents) modulate L2 gender agreement processing in beginner and advanced L2 learners and Spanish monolinguals. Participants performed a  self-paced reading task containing gender agreement and disagreement relations. Half of the sentences were engaged in structural distance relations whereas the other half were not. Also, participants completed a verbal WM updating task. Reaction times were analyzed for both the structural distance and non-structural distance variables. Native speakers took significantly more time to read gender violations than gender agreement relations both within and across constituents. Beginner learners, on the other hand, were not sensitive to violations at any position. Finally, advanced learners took more time to read gender agreement violations than agreement relations within a constituent overall, while those with higher WM were also sensitive to violations across constituents. The results suggest that structural distance, proficiency and working memory capacity modulate the processing of hierarchical relations. Sufficient linguistic experience and enough cognitive resources may thus allow L2 speakers to compute hierarchically complex structures, thus contradicting models of deficient processing in the L2 (Clahsen & Felser, 2006, 2018).