Compensation for coarticulation in prosodically weak words

The goal of this study was to compare production differences in consonant-on-vowel coarticulation in accented and unaccented words with the extent to which listeners compensate differently for these coarticulatory effects. Native speakers of German produced nonsense words such as /pʊp/, /pʏp/, /tʊt/, and /tʏt/ that were either accented or unaccented. In a perception experiment, the same speakers made vowel quality judgements in /pʊp – pʏp/ and /tʊt – tʏt/ continua embedded in contexts in which they were prosodically strong or weak. Consonant-on-vowel coarticulation was found to be greater in the production of prosodically weak words. Listeners compensated for these coarticulatory effects in both prosodic conditions to approximately the same extent. These findings indicate that there was a mismatch between the production and perception of coarticulation in unaccented words. The results are discussed in terms of increased likelihood for sound change to occur in prosodically weak contexts.