Collocational Predictors of Polysemous Word Disambiguation in Business and General English

A Corpus-Based Study of “charge,” “interest,” and “order.”

Abstract

Polysemous vocabulary presents a persistent difficulty for learners of Business English. The lexemes “charge,” “interest,” and “order” each carry several unrelated senses, and learners who transfer the everyday meaning most familiar to them often miss the sense intended in business texts. This paper therefore asks whether the surrounding words make the intended sense predictable. A total of 24,369 keyword-in-context lines from COCA were divided into a Business sub-corpus of business and contractual texts and a General sub-corpus of non-specialist language. Each instance was assigned to a sense by an automated rule-based procedure built on a closed dictionary inventory, with cues weighted by position. Pearson chi-square tests with Bonferroni correction showed that the collocate immediately to the right of the node (R1) carries considerably more disambiguating information than the one immediately to the left (L1), with Cramér’s V reaching .66 at R1 against .31 at L1, and all twelve tests significant at p < .001. Mutual information (MI) identifies high-information cues such as “against,” “accrues,” and “desist,” and register narrowing emerges as the central mechanism by which the specialized financial and legal senses of “charge” and “interest” become conventionalized. The analysis yields explicit position-specific rules, for example R1 “rate” signalling the financial sense of “interest,” for use in ESP teaching and in Business English lexicography.