Convergent probabilistic cues do not trigger syntactic adaptation: Evidence from self-paced reading

Area of Science:

• Psycholinguistics
• Cognitive Psychology
• Computational Linguistics

Background:

• Previous research suggested readers adapt to infrequent syntactic structures (reduced relative clauses) after exposure.
• Replication studies failed to find penalties for preferred structures, questioning rapid adaptation.
• The distinction between syntactic satiation and true expectation adaptation remains unclear.

Purpose of the Study:

• To investigate if contextual cues facilitate adaptation to reduced relative clauses.
• To determine if readers overcome a priori biases in syntactic processing.
• To differentiate syntactic satiation from expectation adaptation.

Main Methods:

• Three self-paced reading experiments were conducted.
• Participants were exposed to unbalanced statistical input (garden-path sentences).
• Co-occurring cues (comprehension questions, semantic cues, font color) were manipulated.

Main Results:

• Readers did not overcome preexisting expectation biases, even with convergent cues.
• No penalty was observed for the a priori preferred main-verb resolution.
• Syntactic repair times showed speedups after exposure to dispreferred structures.

Conclusions:

• Rapid adaptation to novel linguistic statistics is limited.
• Contextual cues do not appear sufficient to override strong prior biases.
• True expectation adaptation requires demonstrating a penalty for the initially preferred structure.