Accounting for graded structure in adjective categories with valence-based opposition relationships

Abstract

In contrast to noun categories, little is known about the graded structure of adjective categories. In this study, we investigated whether adjective categories show a similar graded structure and what determines this structure. The results show that adjective categories like nouns exhibit a reliable graded structure. Similar to nouns, we investigated whether similarity is the main determinant of the graded structure. We derived a low-dimensional similarity representation for adjective categories and found that valence differences in adjectives constitute an important organising principle in this similarity space. Valence was not implicated in the categories' graded structure, however. A formal similarity-based model using exemplars accounted for the graded structure by effectively discarding the valence differences between adjectives in the similarity representation through dimensional weighting. Our results generalise similarity-based accounts of graded structure and highlight a closely knit relationship between adjectives and nouns on a representational level.

Keywords:

• concepts
• adjectives
• typicality
• valence
• GCM
• similarity

Acknowledgements

This work was supported by research grants awarded by the Research Foundation – Flanders (FWO) to the first three authors and by the interdisciplinary research project IDO/07/002 awarded to Dirk Speelman, Dirk Geeraerts and Gert Storms. Dan Navarro was supported by an Australian Research Fellowship (ARC grant DP0773794). We also wish to express our gratitude for the helpful comments by Amy Perfors, Brendan Weekes and two anonymous reviewers. An earlier version of this work was published as De Deyne, Voorspoels, Verheyen, Navarro, and Storms (2011). Comments may be sent to the author at [email protected].

Notes

1. Example from Calvin and Hobbes.

2. Antonymy has a more restricted meaning than opposition (e.g. Cruse, 2004) and we shall argue later on that a specific type of opposition based on valence is an important determinant of adjective representation.

3. All data are available as a downloadable file from http://ppw.kuleuven.be/concat/.

4. The range of these ratings was similar to the range in typicality values found for the 12 natural language noun categories, which varied between 4.64 for vehicles and 5.66 for musical instruments (De Deyne et al. 2008).

5. We also considered related models. One model that we considered was a central prototype model (e.g. Minda & Smith, 2010; Voorspoels et al., 2008). According to this model, the typicality of an adjective is derived by comparing it to an average representation of the category exemplars. The pattern of results of this model was very similar to the GCM results and only differed in terms of the absolute correlations that were achieved, which were slightly lower, on average 0.71 for the prototype model compared to 0.73 for the exemplar-based model.

6. We can think of at least two factors that might explain the lack of independence between arousal and valence. A first explanation is based on the differences between instructions in the rating task. For reasons of consistency with other rating procedures in Verheyen et al. (2012), seven-point rating scales were used, in contrast with studies that used a Manikin rating scale (e.g. Bradley & Lang, 1994). Second, our adjective categories cover a wide variety of properties instead of focusing on emotion words only.