About sharing and commitment: the retrieval of biased and balanced irregular polysemes

We examined how the degree of semantic similarity between an ambiguous word’s meanings (homonyms vs. irregular polysemes) and meaning frequency (biased vs. balanced meanings) interact during lexical access and disambiguation. In Experiment 1, which was a continuous priming experiment, and with an ITI of 50 ms, we observed exhaustive access of meanings for all ambiguous words. With an ITI of 200 ms, we found a dominance effect for biased homonyms. There was no priming for biased irregular polysemes. For balanced homonyms and polysemes, we observed strong and roughly equivalent priming for target words associated with either meaning. In Experiment 2, using sentence reading, all ambiguous words elicited longer reading times in the absence of biasing context, while only biased and balanced homonyms also led to longer reading times in subsequent subordinate-biased context. Taken together, our data support a shared features model of irregular polyseme representation and retrieval.