Word and construction probabilities explain the acceptability of certain long-distance dependency structures

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Cognitive Science

Background:

• The acceptability of long-distance dependencies in sentences is a complex linguistic phenomenon.
• Previous research proposed probability-based accounts for specific constructions, such as wh-questions with complement verbs.

Purpose of the Study:

• To evaluate simpler probability-based models against existing accounts for wh-questions with long-distance dependencies.
• To investigate the applicability of probability-based explanations to wh-questions involving predicate adjectives.

Main Methods:

• Replication of a previous experiment using Liu et al.'s (2022) dataset.
• Analysis of alternative probability-based models on the existing and new datasets.
• Application of the same analytical framework to wh-questions with predicate adjectives.

Main Results:

• An alternative probability-based account demonstrated a better fit to the original data.
• Replication of the experiment yielded consistent results, supporting the proposed alternative model.
• Similar findings were observed when analyzing wh-questions with predicate adjectives.

Conclusions:

• Sentence acceptability, particularly in constructions with long-distance dependencies, is influenced by the probability of constituent words and structures.
• Probability-based models offer a robust framework for understanding syntactic acceptability.
• The findings generalize across different types of complementation in wh-questions.