Speech errors as evidence for the reality of phonological units

Abstract

Speech error data collected by the author and from published literature are examined in relation to the issue of what implications they have for a theory of phonology. It is argued that separate (though related) theories are not required to account for anomalous as opposed to normal speech data. Rather, both types of data are regarded as congruent within the realm of linguistic activity and analysis; any noted difference between them is not regarded as a matter of theory, but as a matter of deviation from expected sound—meaning correspondences. This deviation, however, must be allowed as a part of the principle of linguistic creativity. Related to this, the COMPETENCE—PERFORMANCE distinction is criticised. Accountings proposed for speech errors and their implications for existent transformationalist and structuralist theories of phonology are examined and criticised. Explanations are proposed having the highly desirable general characteristic that they are all linked to a fundamental aspect of linguistic organization, viz. linearity, involving the broad, simple, linear order of traditional structural concepts such as Consonant and Vowel, for whose reduction speech errors provide clear behavioural evidence by correlation, and the order of traditional ‘paradigmatic’ units, involving the concept of neutralization. It is concluded that, contrary to the generally held position, there is no evidence for the ‘syllable’ or the ‘segment’ as independent linguistic units as these are assumed in transformationalist and structuralist phonologies, nor for the binary feature similarly and relatedly conceived.