Different models of word recognition make contrasting predictions about the effects of semantic ambiguity. These predictions are critically influenced by the nature of the representations that are involved. Models in which words are recognised as familiar distributed patterns of semantic activation predict that, for ambiguous words (e.g. bark), interference between their meanings should delay their recognition relative to unambiguous words. In contrast, models that represent words as abstract nodes may predict an advantage for ambiguous words that have multiple entries in the race for recognition.
Surprisingly, the effect of semantic ambiguity on the recognition on individual spoken words has not to our knowledge been investigated. However, reports in the visual domain of faster lexical decisions for high-ambiguity words are problematic for distributed semantic models. More recently, Rodd, Gaskell & Marslen-Wilson (1999) have shown that such findings reflect a confound between lexical ambiguity and word senses. They report a disadvantage for ambiguous words when this confound is removed.
Here, we establish that this distinction between word meanings and word senses is also important in spoken word recognition. Two auditory lexical decision experiments found slower responses for ambiguous words (e.g. bark) than for unambiguous words (e.g. frog). We argue that this ambiguity disadvantage is the result of semantic competition between the meanings of ambiguous words. Further, responses to words with many senses (e.g. slide) were faster than for those with few senses (e.g. shirt). This can be explained by assuming that words with many senses are associated with a large store of readily available semantic or contextual information. Alternatively, it may reflect a difference in the structure of the attractor basins that develop within distributed semantic representations. The attractor basins associated with words with multiple senses will be broad and shallow, and contain more than one stable state. It is possible that settling into the correct attractor may be quicker for such broad attractors, or that the multiple stable states within the attractor may lead to faster settling times.
These contrasting effects, of an ambiguity disadvantage and of a multiple sense advantage, support models of spoken and visual word recognition which employ distributed semantic representations, and they pose problems for recognition models that do not actively involve semantic representations in the process of lexical access and selection.
Rodd, J.M., Gaskell, M.G., and Marslen-Wilson, W.D. (1999). Semantic Competition and the Ambiguity Disadvantage. In M. Hahn & S. C. Stoness (Eds.), Proceedings of the Twenty First
Annual Conference of the Cognitive Science Society (pp. 608-613). Mahweh, New Jersey: Lawrence Erlbaum Associates.