Revisiting Semantic Ambiguity in English Words: Nonarbitrary Polysemy-Form Mappings Influence Lexical Processing

Abstract

Most English word forms convey multiple meanings, that is, they are semantically ambiguous. A relatively small proportion of these ambiguous forms are homonyms that convey distinct meanings (e.g., bank may refer to a financial institution or the land bordering a river), while the majority are polysemes that convey interrelated senses (mouth may refer to the mouth of a person or a river). Empirical investigations have demonstrated an advantage for polysemous word forms across various lexical processing tasks, suggesting differences in the way they are organized in semantic memory. However, polysemous forms also tend to be more frequent, comprise fewer phonemes and syllables, and occur in more dense neighborhoods involving more similar sounding words. The nature and extent of these systematic polysemy-form mappings and their influence on processing have yet to be fully investigated. The present study reports an analysis of a corpus of English monomorphemic words (N = 4,466), confirming that phonological features predict a significant proportion of variance (16.8%) in the number of senses conveyed by a word. A series of experiments using relative weight analyses of megastudy data sets of word recognition and production tasks demonstrates that these systematic polysemy-form mappings have a relatively important influence on lexical processing compared to other lexical and semantic variables. These findings suggest that polysemous word forms might be structured systematically to minimize cognitive costs and maintain a compact or kernel lexicon. Implications for current accounts of lexical ambiguity based solely on semantic similarity are discussed.