Words do not just label concepts: activating superordinate categories through labels, lists, and definitions

ABSTRACT

We investigate the interface between concepts and word meanings by asking English speakers to list members of superordinate categories under one of three conditions: (1) when cued by a label (e.g. animals), (2) an exemplar list (e.g. dog, cat, mouse), or (3) a definition (e.g. “living creatures that roam the Earth”). We find that categories activated by labels lead to participants listing more category-typical responses, as quantified through typicality ratings, similarity in word embedding space, and accuracy in guessing category labels. This effect is stronger for some categories than others (e.g. stronger for appetizers than animals). These results support the view that a word is not merely a label for a concept, but rather a unique way of accessing and organizing conceptual space.

KEYWORDS:

• concepts
• categories
• lexical semantics
• distributional semantics

Acknowledgments

This research was supported by NSF PAC 2020969 awarded to G. Lupyan. Thank you to Kevin Mui for JavaScript help. Thank you to Cognitive Science Society reviewers and to members of the Lupyan Lab for their feedback on this work. Thank you to all study participants.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Although the convention of writing requires us to describe these categories using language, we do not assume that people require language to represent any of these categories.

2 Participants were instructed to not consult dictionary definitions.

3 The three definitions for vegetables all received relatively low mean ratings (less than 70). We therefore constructed a definition for vegetables that we judged to be more appropriate.

4 An alternative solution is to use embeddings from contextual models such as BERT and its variants. However, this necessitates additional assumptions about what context to use when generating the embedding of the target word(s).

5 Adding Participant by Response Number random slopes prevented the models from converging.

6 The individual in the Exemplar condition who received truck, doll, blocks as their cue produced the list books, action figures, animals as their responses 4−6. With respect to toys, these responses had typicality ratings of 2.2, 7.6, and 2.8, respectively.

Additional information

Funding

This work was supported by NSF PAC awarded to G. Lupyan, Division of Behavioral and Cognitive Sciences [grant number: 2020969].