2018 Friday Session A 1615

Friday, November 2, 2018 | Session A, East Balcony | 4:15pm

Topichood and the Comprehension of Relative Clauses in French
A. Bentea, S. Durrleman

Children’s difficulties with object relatives (ORs) as compared to subject relatives (SRs) have been explained in terms of intervention effects, arising in ORs when the object and the subject share a (lexical) NP feature. Such effects can be modulated by featural mismatches in argumental or Phi features between the two DPs [1-3], but these have a selective impact cross- linguistically depending on whether a feature is syntactically ‘active’ in a given language, i.e. functioning as an attractor for movement by belonging to the feature set of the clausal inflectional head [3]. The question that arises is whether a mismatch in a feature that is neither argumental, nor quantificational in nature [4], such as +Topic, impacts OR comprehension. The presence of a +Topic feature on the moved argument distinguishing it from the intervener has been argued to improve children’s comprehension of verbal passives in English [5], and topichood has been shown to reduce the difficulty associated with the processing of ORs in adults [6].

In a series of three studies with 70 typically-developing French speaking children (age range 5;0 to 6;11), we explored whether the presence of a discourse-related feature like +Topic on the moved argument can modulate intervention effects associated with OR comprehension in children. The studies involved a character-selection task (Figure 1) and used the same visual stimuli and test sentences. The only difference consisted in the way the test sentences were introduced (examples 1-3). While no lead-in was provided in Study 1, Studies 2 and 3 used a short lead-in with the purpose of establishing the head of the relative as the discourse topic and thus triggering the presence of a distinctive +Topic feature on the moved argument. Crucially, the lead-in in Study 3 also associated the character established as the discourse topic to the role of Patient and hence gave further indication as to what argument role should be associated with the head noun in the case of ORs.

The data of all three studies (Figure 2) was fitted to a GLMM, with Sentence Type (SR vs OR) and Topic (NoTopic vs Topic vs TopicPatient) as fixed predictors. The results reveal a significant effect of Structure Type in all studies (object relatives yield lower accuracy scores than subject relatives, p<.001). The +Topic feature on the moved element does not significantly affect response performance (p>.05). Our findings show that the presence of a discourse-related feature like +Topic on the head noun in ORs does not modulate comprehension, not even in instances when the context associated the head of the relative to the role of Patient. This follows from the featural intervention account, as children seem to be sensitive to the presence of an NP feature on the moved object and the intervener even in cases when the two elements are distinguished by discourse-related features such as +Topic or quantificational features such as +WH [1].