The processing of relative clause attachment in Thai and the role of experience in sentence processing

Abstract

We adopt experience-based accounts’ techniques in investigating the processing of relative clauses that can modify either of two nouns in Thai (e.g., “the coach of the runner that is good at drawing”) with the aim to understand the nature of experience in sentence comprehension. Two issues are addressed. Firstly, we investigate whether experience with a construction with identical surface word order, namely nominal sentential complements, can affect the processing of relative clauses. Secondly, we investigate whether previously-reported effects of experience in sentence processing reflect general learning that can change participants’ preferences or whether it only reflects strategic learning that helps participants perform better in a specific situation.

We report a corpus count showing that local attachment (e.g., attaching the relative clause to “the runner”) is more frequent than non-local attachment (e.g., attaching the relative clause to “the coach”) but context can obscure such a local-attachment preference.

With contextual effects and segmentation kept under control, three reading experiments demonstrate that in comprehension, native Thai speakers prefer attaching relative clauses to the local noun. The results confirm that a previous report of a non-local attachment preference in Thai was likely to have been tainted by contextual and segmentation factors. As in previous studies, the results of this dissertation support the claim of experience-based accounts as they show that experience with the target construction can affect its later processing. However, we expand previous findings by showing that participants are sensitive to experience manipulation in experiments even when the distributions used in the experiment diverge minimally from their daily experience.

Crucially, we find that the effect of experience can be transferred to a different situation, indicating that participants can learn from their experience and generalize it. Moreover, the results of this dissertation pose a challenge to similarity proposal as in all three experiments, experience with nominal sentential complements do not affect processing of relative clause attachment. This suggests that superficial similarity in terms of word order is not a sufficient condition to cause a processing transfer.