Syntactic ambiguity resolution and revision processes in complement sentences
Temporarily ambiguous sentences are sometimes misanalysed and require revision during sentence processing. Previous studies have reported that non-syntactic information such as verb subcategorisation information does not always prevent misanalysis. However, there is contradictory evidence about whether non-syntactic information is immediately used to recover the globally correct analysis. Previous studies have also reported that initially assigned misinterpretations linger after disambiguation. Some recent studies have suggested that this lingering misinterpretation does not result from a failure to conduct syntactic revision. However, the current evidence for syntactic revision is scarce, limited to a syntactic structure and eye-movement while reading task, and crucially does not necessarily prove that syntactic revision is successfully conducted. The present study reports three self-paced reading experiments that investigate these issues, using temporarily ambiguous complement sentences. Experiment 1 showed that temporarily ambiguous complement sentences are misanalysed during sentence processing, which subsequently causes garden-path effects and lingering misinterpretation. Experiment 2 suggested that non-syntactic information is immediately used to recover the globally correct analysis. However, there was an indication that the incorrect analysis remains activated. Experiment 3 revealed that syntactic revision is conducted in complement sentences without regressive eye movements. The present study argues that the good-enough account can explain these results if this account assumes that a syntactic processing heuristic such as the Canonical Sentoid Strategy is used during the processing of temporarily ambiguous complement sentences.