Processing relative clauses in supportive contexts.

Area of Science:

• Psycholinguistics
• Cognitive Science
• Computational Linguistics

Background:

• Processing complex sentence structures, like relative clauses (RCs), is crucial for understanding language comprehension.
• Subject- and object-extracted relative clauses (SRCs and ORCs) pose different processing challenges.
• Working memory models, such as decay-based accounts, attempt to explain these processing differences.

Purpose of the Study:

• To investigate the processing of SRCs and ORCs in different contextual support conditions.
• To test predictions of decay-based versus interference-based working memory accounts of dependency formation.
• To examine how contextual priming affects the processing of relative clauses.

Main Methods:

• Two self-paced reading experiments in English presenting SRCs and ORCs in both supportive and null contexts.
• A sentence completion study to assess the natural production rates and contextual expectations of SRCs and ORCs.
• Analysis of reading times and completion patterns to infer processing load and contextual influences.

Main Results:

• Object-extracted relative clauses (ORCs) were read significantly slower than subject-extracted relative clauses (SRCs) across both experiments.
• The processing advantage for SRCs over ORCs (extraction effect) was larger and persisted longer in supportive contexts compared to null contexts.
• Sentence completions revealed similar rates of SRC vs. ORC production, but RCs in supportive contexts were more lexically and semantically constrained.

Conclusions:

• The findings support interference-based or certain decay-based working memory models over accounts solely based on the number of intervening referents.
• Contextual support enhances the processing advantage for SRCs, likely due to lexico-syntactic priming and increased expectations for SRCs.
• Processing difficulties in relative clause extraction are modulated by both structural complexity and contextual predictability.