• Aucun résultat trouvé

Relationships between all tasks

6. General discussion

6.3 Relationships between all tasks

In hypothesis D, ToM was linked with complex syntactic structures, and in particular the syntax of complement clauses. Correlations were calculated between different tasks to test our hypotheses concerning the link between syntax and ToM.

In hypothesis D.i we expected to find significant positive correlations between the verbal false-belief task and complement clause understanding. It has indeed been found in other studies (Astington and Jenkins, 1999; Shick, de Villiers, de Villiers, Hoffmeister, 2007) that complement clause understanding predicts performance on false-belief tasks in TD children and children with impairments such as deafness. This link has been shown with communication verbs and mental state verbs in children with ASD (Tager-Flusberg &

Joseph, 2005; Lind & Bowler, 2009). Poltrock (2011) also found significant positive correlations between TD 3-year-old children’s performance on false belief tasks and complement clause comprehension using the perception verb to see.

Table 16

Correlations between the different syntactic and ToM tasks. Significant correlations are marked by **

when p<.001, and * when p=.05.

Our results show no significant correlation between complement clause understanding with a perception verb and responses on the verbal false-belief tasks (r = .314, p=.30).

Furthermore, hypothesis D.ii predicted that the ToM verbal task would correlate with complement clause comprehension regardless of the semantics of the matrix verb. Given that our results show a lack of correlation, this should be accounted for. Perhaps only verbs allowing the expression of misrepresentation of reality, such as mental state verbs, somehow allow children with ASD to succeed in understanding verbal FB tasks. The correlations found by Poltrock could be suggesting that TD children develop ToM in a way that is linked to all sentential complements, and our results could suggest that ASD children follow a different developmental route for ToM understanding. Our hypothesis that it is the syntax of the complement clause that allows children with ASD to access a representation of a false belief, and not its semantics, is therefore not verified by our results. Another explanation for the lack of correlation could lie in the format of the task. The complement clause task

differed from the two other tasks, in that it was a truth-value judgment task using a puppet.

As mentioned in section 4.4.6, children with ASD may have a more pronounced difficulty in rejecting the puppet’s utterance, thereby affecting results and masking any link between ToM and complement clauses.

In hypothesis (D.iii), we expected no correlation between the verbal ToM task and children’s understanding of wh-questions and relative clauses. In accordance with our hypothesis, no significant correlation was found between results in all conditions of the wh-question task and results in the false-belief question of the ToM verbal task (r=.338, p=.20). However, relative clause understanding significantly correlates with the false-belief ToM verbal task (r=.674 , p<.001 ). Although this is contrary to our initial hypothesis, an explanation for these results may lie in an alternative hypothesis stating that ToM is linked to the ability to embed, such as in complement and relative clauses. Smith, Apperly and White (2003) had indeed found that relative clause comprehension is linked to ToM performance. Our results would support the hypothesis that embedding plays a role in understanding ToM verbal false-belief tasks. As no significant correlation was found between understanding of all wh-questions and the verbal false-belief task, perhaps embedding is the crucial component of the CP that allows for ToM to develop. However, as we found no correlations between complement clause comprehension and the verbal false-belief task, there is probably another component accounting for the relationship found in previous studies between ToM and complex syntax (complement clauses and relative clauses).

Hypothesis D.iv predicted no significant correlation between results on low-verbal content ToM tasks and complement clause understanding. In accordance with our hypothesis, no correlation was found. However, contrary to our prediction, the Intention condition of the picture sequencing task, although requiring little use of language, was found to correlate significantly with all the relative clause and wh-question tasks, except for the in-situ condition which was nearly significant (p=.07). This could point to a general developmental relation between the two tasks, i.e. that general cognition underlies results to both tasks. It could also be indicative of the presence of a specific component that is necessary to process both the sequencing task and the syntax tasks.

In accordance with our prediction, the two conditions in Level 1 of the syntactic complexity metric do not correlate with the ToM low-verbal FB task. However, contrary to our prediction, the results on the overall group show that the low-verbal ToM false-belief

condition correlates significantly with overall performance on wh-questions and overall relative clause comprehension. There are significant positive correlations between the three highest levels of the syntactic complexity metric and the non-verbal ToM false-belief condition. These findings point to a link between the understanding of complex syntactic structures requiring movement and ability to attribute a false-belief to an agent even in the absence of verbal directions.

Our final hypothesis, D.v, pertaining to the relationships between ToM tasks, was verified by our results. As predicted, the False-belief question of the verbal ToM task did not correlate with either the Intention condition of the picture sequencing task (r (14) = 0.271, p=0.31), or the False belief condition of the low-verbal ToM task (r (10) =.477, p=.12).

Furthermore, statistics show that the Intention condition of the picture sequencing task correlates significantly with the False belief condition of the low-verbal false-belief task (r(11)= .63, p=.02). This is in keeping with our hypothesis that both low-verbal ToM tasks would correlate with each other.

6.4 Methodological considerations and future perspectives