The processing of referential expressions in second language (L2) is notably difficult (e.g., Friederici et al., 1999; Sabourin & Stowe, 2008), yet possible in high-proficient L2 speakers as demonstrated in a number of ERP studies looking at clitic pronouns processing (Rossi et al., 2014; Rossi & Prystauka, 2019). Crucially, those studies did not track what strategies are at play during on-line processing, and only focused on comprehension. Here, using eye-tracking and a production task we test sensitivity in comprehension and production to gender and number agreement in L2 processing.
Eighteen L1 Spanish (eight females; mean age: 28.72 yrs.; SD = 5.26), and 14 English-L2 Spanish bilinguals (eight females; mean age: 20.86 yrs.; SD = 1.03) were tested. During the eye tracking study participants read sentences that 1) correctly matched the antecedent i.e., Maria pela la manzanafem.sing y lafem.sing come; 2) violated grammatical gender, 3) violated number, and 4) violated both gender and number. During the sentence production task, participants were prompted with a picture and a sentence containing a clitic depicting an action, and needed to complete the sentence using another pronoun.
Eye tracking data revealed that native Spanish speakers are sensitive to gender and number violations (gaze duration F(1,17)=6.73, p=0.0189; right-bounded duration F(1,17)=4.52, p=0.0483; regression path duration F(1,17)=4.61, p=0.0465), while Spanish L2 speakers were sensitive to the double violation (regression path duration F(1,13)=6.89, p=0.0210). Data from the production task revealed a significant difference in performance between the L1 and L2 groups, with L2 learners producing fewer correct clitics (58% against 95%), and producing substitutions of the clitic with the full noun phrase (9.5% of errors). Overall, these data suggest that even L2 speakers show some level of sensitivity during on-line processing of gender and number agreement in the comprehension and production of referential expressions.