Speech perception is of interest to linguists and psychologists alike. Psychologists seek for linguistic units to enable them to explain processes involved in speech; linguists try to establish what these units may be, whether distinctive features, phonemes, syllables, words or even larger units. Although the phoneme was for some time considered to be the most likely candidate, experimental evidence is increasingly pointing to some larger unit, particularly in view of the fact that no one-to-one acoustic correlation with the phoneme nor with distinctive features can be found (cf. Reddy, 1967: 336, Ladefoged, 1967: 146, Denes, 1963: 892). Furthermore, if the phoneme were to be the unit of perception, in any sort of processing involving matching a perceived pattern with one already stored, far too many operations would be involved because of the large size of vocabularies and large number of sentence types in a language; such processing would have to be too rapid to be feasible, bearing in mind the constraints of memory span. There is now more sympathy for the syllable or larger stretch as the unit of perception (e.g. Laver, 1970: 68, Maclay and Osgood, 1959, Ladefoged, 1959: 402), and there seems to be good evidence for the planning of speech to be in stretches longer than a word, e.g. Ladefoged's experiments with placing ‘dot’ at different parts of a sentence (Ladefoged, 1959).