Speaker
Description
The purpose of this study is two-fold. First, we investigate whether academic words tend to show specialized meanings (e.g., resident as a physician being trained) in domain-specific corpora. Second, we offer suggestions for technical dictionaries based on our corpus investigation.
We focused on the medical domain, and developed a computational approach to automatically identify words which are more likely to show technical senses. Inspired by Yarowsky’s (1995) “one sense per collocation” principle, our approach automatically collected and compared words’ collocating words in medical vs. general-purpose corpora. We tested our approach with 1,140 nouns in the Academic Vocabulary List, with it suggesting 129 candidate words. Evaluated by experts, however, only 7 were judged to show medical senses. While identifying few technical senses, our approach collected numerous specialized usages; over 60% of the candidates and their collocates were found to form multi-word medical terms.
Furthermore, the candidates were checked against the Merriam-Webster’s Medical English Dictionary. Consistent with corpus findings, those which are listed as entries mostly have academic rather than medical meanings. Additionally, half of the medical terms identified here are not included in the dictionary. Accordingly, we suggest technical dictionaries include more collocations containing academic words, rather than the words only.
