Preschool Children Incrementally Interpret Causal and Concessive Connectives

• In Event: Integrating Morphosyntactic and Discourse Cues to Process Sentences: Insights from Turkish, Cantonese, and English

Abstract

Adults incrementally use discourse connectives as cues to construct a better representation of the discourse and predict upcoming content (Köhne-Fuetterer et al., 2021). Nevertheless, less is known about how children integrate discourse connectives on the course of processing. Studies showed that children acquire concessive connectors late because they are cognitively more complex than causal connectives (Knoepke et al., 2017). Also, children do not use concessive connectives to reason about the meaning of an unknown word in the discourse until age 8 (Rabagliati et al., 2018). Instead, they apply causal relationship across utterances regardless of whether the connector encodes a causal or concessive relation. Despite their failure in the reasoning task, a follow-up study indicated that children could actually interpret the meaning of the concessive connector (i.e., but) when the context did not require them to infer the meaning of an unknown word. The reason for this pattern may be related to the nature of the tasks. Since these tasks rely on end-sentence interpretations, they may be confounded by processing cost and working memory span (Snedeker & Huang, 2015). The present study investigates children’s online processing of the connectors by using a visual world eye-tracking study, modelled after Knoepke et al. (2017) and Rabagliati et al. (2018).
We tested 4 and 5 year-olds’ (N=22) and adults’ (N=15) interpretation of a causal (so/bu yüzden) and a concessive (but/ama) connector in Turkish. Participants listened to spoken utterances composed of a context sentence with a critical adjective (e.g., sweet) and a target sentence with a connector while seeing two pictures on the screen (e.g., pepper and chocolate). We examined whether children shifted their gaze to the target picture after hearing the connector. We used demonstrative pronoun in the target region to keep the target ambiguous. This was followed by a comprehension question asking the child to choose the correct picture.
Figure 1 indicates that in concessive/ama and causal/bu-yuzden condition, there was an increase in the looks towards the target image in both age groups. Generalized Additive Mixed Model (GAMM) showed that adults and children showed a similar non-linear trend in concessive condition (F=1.768, p<.14). Although adults’ overall look at the target image was more than the competitor in both causal and concessive conditions (Figure 2), the looks to target increased after hearing the connector for both age groups. However, accuracy scores showed that children performed poorly compared to adults in both causal [t(22.349)=-2.7229, p<.02] and concessive [t(22.64)=-12.459, p<.001].
These findings indicate that children integrate concessive meaning incrementally but they may not be able to use it to reason about the discourse. Although children’s looks to target were significantly greater compared to the looks to competitor right after hearing the concessive connector, they later shifted back to the competitor picture compatible with the adjective in the context sentence. This might be because the lexico-semantic activation for this adjective might be greater than that of the discourse connective which requires both lexico-semantic and discourse integration, and children might be having difficulty suppressing this stronger meaning.

Authors

• Özge Günay
  Presenting Author

• Duygu Özge