A Corpus-Based Semantic Analysis of Semi-Technical Vocabulary in the field of Medicine

Abstract

The multiplicity of meaning has long been a central issue in lexical semantics, lexicography and corpus linguistics. In lexical semantics, multiple meanings contribute to lexical ambiguity—polysemy and homography, where a word form has multiple related and unrelated meanings. Semantic studies have yielded contradictory findings about methods of distinguishing polysemy and homography and approaches to the mental representation of polysemy. In lexicography, there has been little agreement on distinguishing and presenting different senses of the same polysemous word and its homographs, and the conventional format of numbered word senses in dictionary entries does not fully depict relationships between polysemous words, potentially confusing users. In corpus linguistics, polysemy and homography present a challenge to the task of word sense disambiguation (WSD) in corpus-derived wordlists. Continuous attempts have been made to achieve more precise and satisfactory corpus-derived outcomes when assigning appropriate senses to given words. This thesis revisits multi-meaning challenges in lexical semantics, lexicography and corpus linguistics. The focus is on semi-technical medical vocabulary, which has yet to be adequately addressed in dictionaries and corpus-derived wordlists. The study is divided into three journal articles. Paper 1 examines Hsu’s (2013) 595-word Medical Word List (MWL) and uncovers the lack of indication of polysemy and homography resulting from its corpus-based automatic analyses of semi-technical medical word forms regardless of their meanings. The examination of the MWL entailed a core meaning-based analysis which reconciles different lexical semantic methods (etymology and native speaker judgement) and approaches (monosemy and polysemy) to identify and distinguish polysemes and homographs. The examination of the MWL resulted in 302 words whose polysemes and homographs are anticipated to pose pedagogical difficulties. Paper 2 presents SemiMed, a lexicographic resource for semi-technical medical vocabulary, as an alternative to word form-based lists. SemiMed is based on a semantic analysis underpinned by lexical semantic theories (Lakoff’s (1987) radial categories and Tyler and Evans’s (2004) Principled Polysemy). A corpus-based analysis employs the WSD method of one-sense-per-collocation to validate the semantic analysis. SemiMed’s headword templates transfer corpus-based WSD results into semantic networks, thus visualising relations between polysemes and homographs in MWL words. This non-conventional format aims to minimize confusion associated with the traditional entry-structured layout. Paper 3 analyses SemiMed’s practicality and usefulness in a pilot study in which a 40-word e-version of SemiMed was introduced to 18 medical students with English as a foreign language to use while role playing medical scenarios. Student participants’ feedback on their experience in using SemiMed was gathered through focus groups. The students’ preference for SemiMed over conventional dictionary entries highlighted the benefits of SemiMed’s non-conventional format in facilitating the understanding of polysemy and homography in semi-technical medical vocabulary. The main implication of the study’s findings is that some of the challenges of learning and teaching words with multiple meanings may be resolved by an interdisciplinary approach, where theoretical and methodological frameworks from lexical semantics and corpus-based WSD inform lexicographic practices to better manage polysemy and homography.