Segment realization in relation to a carrier word’s uniqueness point

Abstract

This paper determines whether the acoustic realization of English phones differs depending on their position relative to a carrier word’s uniqueness point, defined as the earliest point at which that word differs from all the words in the lexicon that share the same initial string. As words are confidently identified at their uniqueness point [Marslen–Wilson (1987)], phonetic material past that point may play a less crucial disambiguating role and show signs of reduction akin to those reported in Wright (1997), who found that vowels in monosyllabic words with relatively few perceptual neighbors are acoustically reduced. The acoustic duration of segments located at and past a carrier word’s uniqueness point is compared (e.g., -erry in blueberry and strawberry, respectively, because of blueberry’s competitors bluebird, blueblood, etc.). No consistent pattern of differences is found. This result suggests that the domain of Lindblom’s (1990) hypospeech and hyperspeech model is higher than the word. The implications of this result for DeJong’s (1995) characterization of stress in English as word-level hyperarticulation are discussed.