Pronoun processing in anglophone late L2 learners of French: Behavioral and ERP evidence

• We examined cortical response in L2 acquisition of French clitic pronouns.
• All groups (Natives, High- and Low-intermediates) responded to in situ clitic placement with a P600.
• L2 groups demonstrated distinct capacities according to level.
• Both groups of learners showed a stronger ERP response to in situ pronoun placement than behavioral data would suggest.
• Our ERP data suggest grammaticalization of L2 pronouns prior to conscious use.

This study investigated offline, behavioral, and online cortical responses to French clitic and strong pronouns by two groups of anglophone learners (L2) and native French controls. English in situ object pronouns behave morphosyntactically like full lexical noun phrases (I saw Bill/I saw him), whereas in French they are cliticized to the verb (J'ai vu Guillaume/Je l'ai vu/*J'ai vu le/lui). We used ERPs, known to be sensitive to word order and morphological violations in natives, to investigate neural responses to word order violations of in situ pronouns. Results indicate native sensitivity to ungrammatical in situ pronoun placement in offline (grammaticality judgment (GJ)), as well as online end-of-sentence and ERP responses, notably a robust P600. L2 groups demonstrated distinct capacities according to level. Both high- and low-intermediates showed online cortical sensitivity to ungrammatical in situ prounouns, producing a P600 in response to such violations (though with distributions that sometimes differed from native controls). However, on behavioral measures the L2 groups differed, both from natives and from each other. Native controls correctly rejected more in situ clitic and strong pronouns than both the L2 groups, and the high-intermediate group correctly rejected more than the low-intermediate group. For natives, the similarity of online/offline responses confirmed a grammaticalized representation and processing of pronouns. Both groups of learners showed a stronger ERP response to in situ pronoun placement than behavioral data would suggest. This is especially true for low-intermediate learners who, despite high rates of acceptance of strong pronouns/clitics in situ, showed a cortical response in the P600 window though without the canonical topography of the native response. While both learner groups are apparently using (at least some) L1 English settings for pronoun placement, their cortical response clearly demonstrates unconscious increasing sensitivity to French parameters.