LLM Estimates of Word Characteristics

Psychological and linguistic characteristics of words (e.g., concreteness) are used to norm experimental stimuli and discover empirical relationships in the lexicon. One potential application of Large Language Models (LLMs) is to estimate these characteristics directly. Indeed, recently published work has shown that LLMs can be used to produce estimates of word features in different languages that correlate with gold standard human judgments. This chapter provides an overview of the use of word features in Cognitive Science and Psycholinguistics research generally, then surveys recent research using LLMs to generate those features. We also discuss various methods (such as prompting and fine-tuning) for producing higher-quality estimates from LLMs, as well as the potential limitations of this approach.