Lexical representation of phonological variation in spoken word recognition

There have been a number of mechanisms proposed to account for recognition of phonological variation in spoken language. Five of these mechanisms were considered here, including underspecification, inference, feature parsing, tolerance, and a frequency-based representational account. A corpus analysis and five experiments using the nasal flap (found in a production of gentle in American English) both in isolation and in biasing sentential context failed to fully support any of these accounts. The results support a strong phonological representation for the [nt] form and a gradient strength representation in the lexicon for the nasal flap that is influenced by production frequency. The results are discussed in terms of orthographic and phonological experience with word forms in the formation of lexical representations.