Representation and processing of semantic ambiguity

One of the established findings in the psycholinguistic literature is that semantic ambiguity (e.g., “dog/tree bark”) slows word comprehension in neutral/ minimal context, though it is not entirely clear why this happens. Under the “semantic competition” account, this ambiguity disadvantage effect is due to competition between multiple semantic representations in the race for activation. Under the alternative “decision-making” account, it is due to decision-making difficulties in response selection. This thesis tests the two accounts by investigating in detail the ambiguity disadvantage in semantic relatedness decisions. Chapters 2-4 concentrate on homonyms, words with multiple unrelated meanings. The findings show that the ambiguity disadvantage effect arises only when the different meanings of homonyms are of comparable frequency (e.g., “football/electric fan”), and are therefore initially activated in parallel. Critically, homonymy has this effect during semantic activation of the ambiguous word, not during response selection. This finding, in particular, refutes any idea that the ambiguity disadvantage is due to decision making in response selection. Chapters 5 and 6 concentrate on polysemes, words with multiple related senses. The findings show that the ambiguity disadvantage effect arises for polysemes with irregular sense extension (e.g., “restaurant/website menu”), but not for polysemes with regular (e.g., “fluffy/marinated rabbit”) or figurative sense extension (e.g., “wooden/authoritative chair”). The latter two escape competition because they have only one semantic representation for the dominant sense, with rules of sense extension to derive the alternative sense on-line. Taken together, this thesis establishes that the ambiguity disadvantage is due to semantic competition but is restricted to some forms of ambiguity only. This is because ambiguous words differ in how their meanings are represented and processed, as delineated in this work.