Poster #72 - French Children can use Explicit and Implicit Prosody to Guide Their Interpretations of Ambiguous Relative Clauses.

• In Event: Poster Session 09

Abstract

Several studies show that implicit (silent) prosodic representations can influence the parsing and comprehension of written sentences in adults (e.g., Fodor, 2002; Frazier & Gibson, 2015; Pynte & Colonna, 2000). For instance, when reading sentences with relative clauses (RCs) such as “He knows the daughter of the man who waits”, readers tend to impose an implicit prosodic break after the first noun (N1, e.g., “[He knows the daughter(N1)][of the man(N2) who waits]”, which leads them to interpret the second noun (N2) as the subject of the RC (i.e., “the man is waiting”). However, when reading a longer sentence containing the same words/ambiguity, they tend to impose a prosodic boundary after N2 (e.g., “[He knows the daughter(N1) of the man(N2)][who waits in the kitchen]”), which leads them to interpret N1 as the subject of the RC (i.e., “the daughter is waiting”). However, no one has investigated when children start relying on prosodic information to constrain their interpretations of relative clauses in the reading and spoken domain. Crucially, recent studies suggest that when processing complex syntactic structures with optional prosodic boundaries, children have trouble using prosodic information to constrain their parsing (e.g., Kolberg et. al., 2021; Snedeker & Trueswell, 2001).
To investigate these issues, we tested French-speaking children’s interpretation of RCs in two modalities: spoken sentences (with 7-to-9-year-olds (N=28) and 4-to-6-year-olds (N=27)) and written sentences (with 8-to-9-year-olds, N=20). With spoken sentences, children heard long-RC sentences containing a prosodic boundary either after N1 or after N2, as in the examples above. In the reading task, children read aloud either long- or short-RC sentences. Afterwards, participants heard or read a question about the subject of the relative clause (e.g., “who waits?”) while seeing two images side-by-side: one representing the N1 interpretation (e.g., a girl) and another representing the N2 interpretation (e.g., a man). They were then asked to point towards the right answer. With spoken sentences, the results showed that both age groups can successfully use prosodic information to guide their interpretations of RCs (see Fig.1). Although they showed an overall preference for choosing N1 as the referent of the long-RC, they chose N1 more often when listening to long-RC sentences with a prosodic boundary after N2 than after N1 (β= 0.84, SE=0.32, z=2.58, p<.01). The preliminary results of the reading task (see Fig.2) shows the same tendency, with children choosing N1 more often when reading long-RCs than short-RCs. However, the effect of condition is not significant yet (β= 0.51, SE=0.36, z=1.43, p=.15).
These results show for the first time that from age 4, children can use prosodic information in speech to constrain their interpretation of ambiguous relative clauses. Furthermore, by age 8, they seem to be able to generate representations of sentence intonation (i.e., prosodic phrasing) when reading sentences, which in turn affect their interpretations. This suggests that both explicit and implicit prosody can influence online sentence parsing in children, and should therefore be included in models of children’s spoken and written sentence processing and in methods of reading instruction to children.

Authors

• Letícia Schiavon Kolberg, Université Paris Cité
  Presenting Author

• Alex de Carvalho, Université de Paris