We examined the
neural correlates of semantic ambiguity by measuring changes in MEG recordings during a
visual lexical decision task in which the properties of ambiguous words were manipulated. Words that are ambiguous between unrelated meanings (like bark, which can refer to a tree or to a dog) were accessed more slowly than words that have no unrelated meanings (such as cage). In addition, words that have many related
senses (e.g., belt, which can be an article of clothing or, closely related in
sense, a fan belt used in machinery) were accessed faster than words that have few related
senses (e.g.,
ant). The findings are inconsistent with accounts that posit that both kinds of ambiguity involve separate lexical entries, but instead offer both behavioral and
neurophysiological support for separate entry accounts only for homonymy, and a single-entry model of polysemy. The findings also provide
neural correlates for a behavioral study of lexical ambiguity that demonstrated that the frequently reported ambiguity advantage in
lexical decision tasks is not due to the presence of many unrelated meanings but to the presence of many related
senses.
