The Role of Context in Lexical Recognition

Abstract

UC Berkeley Phonology Lab Annual Report (2007) The role of context in lexical recognition Jacqueline Palmer Introduction The nature of the mental lexicon and its stored forms remains an open and much disputed question. While the traditional view has suggested single abstract phonological forms for each lexeme, there has been much recent evidence to the contrary that requires a model of the lexicon that allows much more information to be encoded on several lexical represen- tations per lexeme. There have been many recent studies on a variety of phonetic aspects that seem to indicate that lexical forms need to be encoded in a more detailed manner to be able to account for listeners’ great sensitivity to the individual variation that is dis- played in everyday speech (e.g., Bybee, 2001; Palmeri, Goldinger, & Pisoni, 1993). Subjects in many experiments have been able to attend to fine-tuned phonetic detail that distin- guishes one talker from the next but traditionally has not been included in the abstract lexical representation, such as individual variation in voice-onset-time (Allen & Miller, 2004). These recent studies have suggested that these phonetic details be encoded in the representation of the stored lexical forms as listeners’ recognition of these details aided in and sped up lexical recognition. One question that remains is the status of utterances of words that are phonetically reduced; i.e., that are missing segments that would otherwise be included in a more care- ful pronunciation of the same word. The studies mentioned above used pronunciations of words that indeed differed from speaker to speaker in a given property, but in both cases the stimuli consisted of words that contained all of the segments that would exist in a careful pronunciation. Thus, it remains unclear whether phonetically reduced forms may be simply another part of interspeaker variation that requires representation in the lexicon just as forms varying in the dialect or gender of the speaker do. This leads one to wonder which phonetically detailed productions listeners remember, if listeners and speakers do indeed retain phonetically detailed representations in the lexicon. While sim- ple accounts of lexical access would likely have heavily reduced conversational tokens be activated by a single abstract representation during recognition, it could be that these re- duced forms are retained in memory to be deployed during the recognition of reduced versions of words. This study seeks to shed light on the nature of the perception of phonetically reduced forms in conversational speech; specifically, the factors that lead to listeners’ difficulty in