Understanding same subject-verb agreement differently: ERP evidence for flexibility in processing representations involved in French subject-verb agreement

In an ever-changing environment such as a situation with a variety of linguistic information, individuals have to adapt by selecting the most relevant and appropriate information. In event-related potential studies that manipulated the syntactic agreement between a subject and a verb, it was shown that morphosyntactic features (e.g., number or person feature) are used to compute syntactic dependencies. Furthermore, statistical language information seemed to play a role in the production of subject-verb agreement. We thus investigated flexibility in the processing of morphosyntactic features and co-occurrence frequency between a subject and its verbal inflection. Pronoun primes and verbal targets were presented auditorily and the flexibility of the representations in French subject-verb agreement was studied by manipulating the task to be performed on the target. In Experiment 1, the task was a lexical decision task to induce the use of co-occurrence frequency between a subject and its verbal inflection; in Experiment 2, the task was a grammatical categorization task to amplify the use of morphosyntactic features. Results showed that statistical information affected the processing of the verb earlier than the use of morphosyntactic features, whose violation produced the classic biphasic reaction with negativity followed by positivity. Our findings suggest that there is flexibility in the use of both statistical and abstract morphosyntactic feature representations, although the flexibility of the use of features depends more on task strategies.