In this article, I will present partial results of three experiments that deal with the second language (L2) acquisition of the same phenomenon: Clitic Left Dislocation (CLLD) in Spanish and Bulgarian by native English speakers. This is the phenomenon where an object that is previously mentioned in the discourse (a Topic) moves to the beginning of the sentence, and is doubled by a clitic agreeing with it in person, number, and gender. Because the acceptability of this word order depends on the previous discourse, this linguistic construction is evaluated at the interface of syntax and discourse. The Interface Hypothesis (Sorace & Serratrice, 2009; Sorace, 2011) argues that acquisition of such constructions present possibly insurmountable problems even in very advanced L2 speakers. I will discuss findings from Valenzuela (2005, 2006), Ivanov (2009), and Slabakova, Kempchinsky & Rothman (in revision), which demonstrate successful acquisition of Bulgarian and Spanish topicalization constructions. I will show that all these findings argue against a strong version of the Interface Hypothesis, and I will discuss why acquisition of this type is interesting to study.

The first language processing studies have shown that working memory (WM) plays an important role in many aspects of reading performance. This study investigates whether L2 learners' processing of Japanese relative clauses (RCs) is influenced by their WM capacity. Two processing-based hypotheses, Dependency Locality Theory (e.g., Gibson, 1998) and Structural Distance Hypothesis (e.g., O'Grady, 1987), are evaluated with respect to the performances of L2 learners' processing Japanese subject-gap and object-gap RC sentences in an online reading experiment. L2 learner participants were divided into two groups based on their WM capacity measured by a Reading Span Test (RST). L2 RST Low and High groups were not different in comprehension accuracy rates of object-gap RC sentences, but numerically different in subject-gap RC sentences. The online reading results support the DLT, but group differences were found in different regions of the RC sentences in the residual reading times. These findings suggest that one's WM capacity also affects RC processing in L2.

The current study explores how Japanese EFL learners use prosodic information in understanding auditory-presented English sentences. Two experiments were conducted to determine which information between prosodic and syntactic information Japanese EFL learners use in analyzing sentence structure, when the two suggest conflicting syntactic structures. The results showed that the prosodic boundary that signals sentence structures is important information for Japanese EFL learners in understanding English sentences. It was also revealed that learners with higher proficiency tended to rely more on syntactic information rather than prosodic information in analyzing the sentence structure.

This paper reports on how adult Japanese Learners acquire the universal and parameterised constraints which regulate the accessibility of Goals to Wh-Movement in English wh-questions. We present evidence that universal constraints on Goal Accessibility operate in early JLE grammars, and that learners initially transfer settings for parameterised constraints from L1 to L2, concluding that our overall findings are broadly consistent with the Full Transfer Full Access model of L2 acquisition developed in Schwarz and Sprouse (1994, 1996). We show that JLEs are able to reset some parameterised constraints (e.g. the P-Stranding Constraint) but not others (e.g. the Left Branch Condition), and argue that they are only able to re-set learnable parameterised constraints (i.e. those whose setting can be learned solely on the basis of positive evidence from input), not unlearnable parameterised constraints (i.e. those whose setting cannot be learned solely on the basis of positive input).