Semantic Underspecification in Language Processing 论文

摘要

Abstract It is commonly assumed that when we encounter a word in a text, we automatically and immediately activate specific, detailed semantic information associated with that word and instantly integrate this information in the unfolding interpretation of the text. On‐line evidence of how we process polysemous words, that is, words with multiple semantically related interpretations or senses, suggests that instead of accessing a specific sense, language users initially activate a word's meaning that is semantically underspecified. Context can then help to make this meaning more specific, if there is a need for it. I will present an overview of the available evidence for this view, including new work that indicates that the type of task can influence how quickly we home‐in on a specific sense, address evidence that, at first sight, seems to contradict the underspecification view, and outline a number of issues that require further attention.