In this thesis, the source of gradience in acceptability judgments is discussed (Scholes 1966) and a set of experiments is performed which attempt to attribute gradience more concretely to either phonotactic knowledge or lexical knowledge. Two phonotactic acceptability judgment tasks are implemented to better understand whether reaction time can lessen the influence of lexical information on phonotactic acceptability judgments. Following results from Fox (1984) which show weaker influence from lexical information when less response time is allowed, I hypothesize that phonotactic information should be immediately accessible for participants, but that a lexical search takes more time to perform. In turn, an acceptability judgment task which allots less response time to participants should result in less influence from lexical information in their responses. By comparing the resulting participant judgments to gradient and categorical language models, I show that lexical access is still present at early reaction times, meaning reaction time was not useful in removing the influence of lexical information from phonotactic acceptability judgments in this set of experiments. This prompts a discussion of other possible models which can feasibly be used to understand these judgments and the source of their gradience.
Read
