Word meaning is both categorical and continuous.

Area of Science:

• Cognitive Science
• Psycholinguistics
• Computational Linguistics

Background:

• Traditional theories of word meaning fall into two camps: categorical (discrete senses) and continuous (trajectories in state space).
• Both categorical and continuous theories face empirical challenges in explaining how humans process word meaning.

Purpose of the Study:

• To introduce and test novel hybrid theories of word meaning that integrate discrete and continuous representations.
• To reconcile the dynamic, context-dependent nature of word meaning with evidence for category-like structures in lexical knowledge.

Main Methods:

• Developed two novel hybrid theories of word meaning.
• Conducted two behavioral experiments.
• Employed an analytical approach using neural language models to evaluate competing theories.

Main Results:

• Experimental results were best explained by a hybrid account combining discrete sense representations with a continuous meaning space.
• This hybrid model successfully accommodates both context-dependent meaning shifts and categorical lexical structures.
• Computational implementations of the hybrid account demonstrated significant predictive power.

Conclusions:

• Human word meaning is best characterized as both categorical and continuous.
• Hybrid models integrating discrete and gradient representations offer a more comprehensive explanation for lexical ambiguity.
• Future research should investigate the emergence of discrete sense representations and their role in cognitive processes.