Cross-serial dependencies in Dutch: testing the influence of NP type on processing load

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Cognitive Science

Background:

• Sentence processing complexity is modeled by theories like Lewis (1996), Gibson (1998, 2000), and Gordon et al. (2001, 2002).
• Cross-linguistic validation of these models is crucial for understanding universal cognitive mechanisms.

Purpose of the Study:

• To test the cross-linguistic generalizability of three sentence processing complexity models.
• To investigate the on-line processing of cross-serial dependencies in Dutch.

Main Methods:

• Manipulated the number of subject-verb relations in Dutch clauses.
• Varied noun phrase types: indexical pronouns, proper names, and full definite NPs.
• Analyzed processing difficulty at verbs during on-line sentence comprehension.

Main Results:

• Processing difficulty at verbs was not influenced by proper names versus pronouns.
• Full NPs, compared to pronouns, increased difficulty in 3-verb constructions, supporting Gibson's model but contradicting Lewis's.
• Findings align with similarity-based inference models, where NP similarity, not discourse status or syntactic function, drives retrieval difficulty.

Conclusions:

• Similarity-based inference models provide a better explanation for the observed processing difficulties in Dutch cross-serial dependencies.
• The study highlights the importance of considering noun phrase similarity in sentence processing models.
• Divergence from Gibson's and Lewis's models suggests limitations in their predictive power for cross-linguistic data.