In this paper the productivity of reduction processes as a function of the frequent or infrequent nature of words or combinations is studied and the nature of the cognitive representation of reduced forms is considered. An experiment is presented where speakers of American English were asked to produce i) new strings of words which have a low probability of combining together, and ii) phonetically equivalent frequently occurring sequences. The experimental results show that reduction processes apply rather consistently to highly frequent combinations, but they hardly apply to sequences of words which are rare or unusual. The claims made in this paper on the basis of the evidence presented are: 1) Phonological rules can not provide a cognitive representation that accounts adequately for the automaticity of frequent sequences and the high percentage of reduced forms they present. 2) The difference in reduction processes in frequent and infrequent sequences can be explained in terms of two modes of speech perception and production. An automatic or template mode for frequently recurring words of phrases, where style-dependent variant pronunciations are stored as separate lexical entries, and an analytical or derivational mode for new, propositional speech, where rules might be used to produce variant pronunciations. This proposal is in line with psycholinguistic, neurological and learning models of speech behaviour. This proposal obviously reformulates the size and nature of the phonological inventory.
Cite as: Solé, M.J., Ohala, J.J. (1991) The phonological representation of reduced forms. Proc. ESCA Workshop on Phonetics and Phonology of Speaking Styles, paper 049
@inproceedings{sole91_ppst,
author={Maria Josep Solé and John J. Ohala},
title={{The phonological representation of reduced forms}},
year=1991,
booktitle={Proc. ESCA Workshop on Phonetics and Phonology of Speaking Styles},
pages={paper 049}
}