Brown (2001) demonstrated that speakers adjust their production of coarticulation to accommodate potential lexical confusability (the possibility of one word being misheard as another), and that listeners' lexical recognition is facilitated by such adjustments. Speakers produced confusable, ``hard'' words (those with low-usage frequencies and many frequent, phonologically similar neighbors) with significantly more coarticulation than less confusable, ``easy'' ones (with high frequencies and few, low-frequency neighbors). The current study investigates whether these adjustments are general, applying to any potentially confusable words, or whether the adjustments are sublexical, sensitive to the particular neighbors a given word has. A frequency-balanced set of ``easy'' and ``hard'' words containing a vowel+nasal consonant sequence has been found such that half are confusable by the nasal and half are not. Six speakers read these words twice in carrier sentences. The degree of nasal coarticulation during the vowel is measured acoustically. From this it will be seen whether the coarticulatory cues to a segment are more enhanced (i.e., there is more coarticulation) in words confusable by that segment than in words with no neighbors differing by that segment. Preliminary results suggest that coarticulatory adjustments in response to lexical confusability are made at the word level rather than sublexically.
Acoustical Society of America Journal
- Pub Date:
- October 2003