An Information-Theoretic Account of Availability Effects in Language Production

Area of Science:

• Computational Linguistics
• Psycholinguistics
• Cognitive Science

Background:

• Language production involves complex cognitive processes.
• Existing models like Uniform Information Density explain some aspects of word choice.
• Availability effects in language suggest word selection is influenced by ease of retrieval.

Purpose of the Study:

• To present a novel computational-level model of language production.
• To integrate information theory and control theory for modeling word choice.
• To account for availability effects in language production through a unified framework.

Main Methods:

• Developed a computational model combining information and control theories.
• Formulated word selection as maximizing communicative value under capacity constraints.
• Applied the model to analyze English relative clause complementizer dropping.
• Utilized corpus data for English and experimental data for Mandarin.

Main Results:

• The model predicts a trade-off between production ease and communicative accuracy.
• The model successfully reproduces observed phenomena in English and Mandarin.
• Demonstrated the model's ability to explain availability effects in word selection.

Conclusions:

• The proposed model offers a viable alternative to existing theories like Uniform Information Density.
• The model provides a promising general framework for language production.
• The theory is closely aligned with emerging computational neuroscience approaches.