Measuring Concept Concreteness from the Lexicographic Perspective

Measuring Concept Concreteness from the Lexicographic Perspective  Oi Yee Kwong Department of Chinese, Translation and Linguistics, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong Olivia.Kwong@cityu.edu.hk Abstract. The distinction between concrete and abstract concepts is psychologically valid but so far it can hardly be quantified in any objective way, which prevents it from being further studied in computational linguistics. This paper proposes a systematic way to measure concreteness from the surface structure of dictionary definitions. Comparing the scores from WordNet definitions with human ratings, the method apparently works better for concrete senses than abstract ones, and is not adequate for measuring concreteness on a finer scale beyond the simple dichotomous distinction. Future work thus includes exploring the possibility of deploying a wider range of surface clues from definitions for the purpose and investigating how the method works with definitions from different dictionaries. Keywords: Concept concreteness, Dictionary definitions, Lexicography, Word senses.  The work described in this paper was supported by grants from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. CityU 1508/06H) and the Department of Chinese, Translation and Linguistics of the City University of Hong Kong. Copyright 2011 by Oi Yee Kwong 1 Introduction Most people might find it more than normal that concrete concepts are easier to understand and learn than abstract concepts. Such common sense is supported by ample psychological evidence from lexical decision tasks and children’s spoken and reading vocabulary (e.g. Bleasdale, 1987; Kroll and Merves, 1986; Yore and Ollila, 1985). However, few studies have addressed how this apparently trivial observation might imply on the mental storage and organisation of words and their meanings, and whether modelling such distinction between concrete and abstract concepts in computational semantic lexicons might benefit natural language processing tasks like automatic word sense disambiguation. For example, how can concreteness be included in common ontological organisation of semantic lexicons? If concrete concepts are easier for people than abstract ones, should we expect concrete senses to be more easily disambiguated than abstract senses? If this is the case, should we take it into account when evaluating disambiguation performance, perhaps in addition to the similarity among senses (e.g. Resnik and Yarowsky, 1999; Palmer et al., 2006)? Should different information be employed for disambiguating concrete and abstract senses? One major obstacle to this kind of study is the subjectivity in deciding what is concrete and what is abstract, which is a matter of degree and confounded by many factors. In psychology studies, concreteness (or abstractness) has often been measured by averaging human ratings on a sample of words. Human raters give a score for each word on a scale (e.g. from 1 for highly abstract to 7 for highly concrete), often based on various factors like the familiarity and imageability of the concepts, and frequency of occurrence, but more heavily on their own 25th Pacific Asia Conference on Language, Information and Computation, pages 60–69 60 experience and probably private cognition. This way of measuring concreteness is …

Content truncated to 500 words