Cross-Linguistic Evidence for the Role of Phrasal Prosody in Constraining Parsing in Young Children

• In Event: Beyond Categories: Infant Syntax Above the Word Level

Abstract

Phrasal prosody has been proposed as a crucial cue for syntactic acquisition, since it correlates with syntactic structure, and infants are sensitive to it from their first months (e.g., Mehler et al., 1988). However, studies on young children's ability to use prosody to constrain parsing have reported mixed results, with some providing positive evidence (e.g., in French: Dautriche et al., (2014), de Carvalho et al., (2017); and in English: de Carvalho et al. (2016); Snedeker and Yuan (2008)), while others failed to observe this ability in children up to 6-years-old (e.g., Snedeker & Trueswell (2001) in English; Choi & Mazuka (2003) in Korean). We claim that these discrepancies are most likely due to the kind of syntactic structures tested rather than to a cross-linguistic difference. To further investigate this issue, we tested 3-4-year-old preschoolers (in French and Brazilian Portuguese (BP)) and 28-month-old toddlers (in French) with a type of syntactic structure ambiguity not yet studied.

We used French and Brazilian Portuguese sentences such as "Le tigre mange le dinosaure aussi"/"O tigre tá comendo o dinossauro também" ("The tiger is eating the dinosaur too"), which can have two interpretations, depending on the position of the prosodic boundary: when produced with a major prosodic boundary following the verb ("[The tiger is eating!] [The dinosaur too!]"), both the tiger and the dinosaur are agents of the action "eat", but the verb is ellipsed in the second phrase via a syntactic process called stripping; when produced with a minor prosodic boundary after the subject ("[The tiger] [is eating the dinosaur too!]"), the tiger is the agent of the action, and the dinosaur is the patient. Using a Preferential Looking paradigm, we randomly assigned children to one of two conditions: stripping vs. transitive prosody. While hearing the sentence, children watched two videos side-by-side: one depicting the simple transitive interpretation of the sentences (e.g., the tiger eating the dinosaur and a duck), and another depicting the stripping interpretation (e.g., the tiger and the dinosaur eating a duck, Figure 1).

The results showed that 3-4-year-olds who heard stripping sentences looked longer towards the two-agent videos than children who heard the transitive sentences (p<.001 for French children and p=.004 for Brazilian children). A cluster-based permutation analysis (Fig 2) also revealed two time-windows with a significant difference between conditions (for French: p=.04 and p<.001; for BP: p=.05 and p=.03). For 28-month-olds, no significant difference was observed, although there was a tendency in the expected direction (p=.074).

Our results provide cross-linguistic evidence for young children's ability to rely on phrasal prosody in constraining parsing, since both French- and BP-learning 3-4-year-olds correctly distinguished stripping from transitive sentences through their differences in prosodic phrasing. In addition, the 28-month-olds' results suggest that this ability might depend on the type of structure being tested, since children of the same age succeeded with simpler and more frequently used structures tested in previous studies (although, as will be discussed, methodological differences might have also played a role in children's performance).

Authors

• Letícia Schiavon Kolberg, Université Paris Cité
  Presenting Author

• Alex de Carvalho, Université de Paris
  Non-Presenting Author

• Anne Christophe, École Normale Supérieure
  Non-Presenting Author