Published online by Cambridge University Press: 22 April 2026
This paper presents a theory of the child's acquisition of the phonology of his native language. The theory predicts learning, measured by correct pronunciation of phones, as a function of ease of perception. The general prediction is that the more discriminable phones are learned earlier and the less discriminable ones later. Finally, the theory is compared with some previous suggestions in this field.
1 These formulations have been much improved by vigorous criticism from participants in the class in psycholinguistics at the Linguistic Institute 1962, University of Washington, particularly James J. Asher, David McNeill, Agnes Niyekawa, Anita Marten, W. Ballantyne, J. Connors, C. Davis, G. Drachman, N. Pitaro, and C. Yakulic. I am also indebted to Frank A. Logan, Susan Erwin, Jarvis Bastian, and George A. Miller for their helpful criticisms of a version circulated in dittoed form. An earlier version of the paper was read before the Linguistic Society at its annual meeting in 1962, and presented as a lecture before the linguistic group at the Moscow State University in 1963, where it provoked a vigorous discussion. I wish to thank Serge von Oettingen, Y. S. Stepanov, and Urie Bronfenbrenner for suggestions regarding the translation into Russian of certain technical terms in linguistics and psychology.
2 Cf. O. Hobart Mowrer, Learning theory and the symbolic processes (New York, 1960); N. E. Miller and John Dollard, Social learning and imitation (New Haven, 1941).
3 It is not asserted that acoustic qualities are produced by characteristic articulations; there is considerable evidence that this is not the case. Even the present postulate may be overturned by experiment; nevertheless, it is probably accurate by and large, and reasonable for the moment, though it is recognized that further work may require a complex restatement of it.
4 Here and in what follows, I lean heavily upon the work of George A. Miller and Patricia E. Nicely, ‘An analysis of perceptual confusions among some English consonants’, JASA 27.338–52 (1955), reprinted in Sol Saporta (ed.) Psycholinguistics: A book of readings (New York, 1961).
5 It is possible, though not certain, that this order of discriminability is universal and not limited to English. The principal evidence for English is the experiment by Miller and Nicely. Friction is used here for their affricatton.
6 First set forth in Roman Jakobson, Kindersprache, Aphasie und allgemeine Lautgesetze (Uppsala, 1941).
7 Werner F. Leopold, ‘Patterning in children's language learning’, in Saporta (ed.), op. cit. See also Leopold's Bibliography of child language (Evanston, 1952), and his great pioneering work, Speech development of a bilingual child (Evanston, 1939–49; 4 vols.).
8 An incidental by-product of the present theory is a possible explanation of certain puzzling phenomena from the babbling and cooing periods. It has been reported (cf. Leopold, op.cit.) that certain sounds pronounceable as early as the second or third month of life later drop out of the child's repertoire. Since the amount of vocalization is not decreasing, this apparent loss of responses poses a difficult problem for students of the subject. It is possible that velar consonants and high-back vowels are incidental by-products of the innate mechanisms of sucking and crying during the first few months of life and that they are articulated quite well at a time when the perceptual apparatus of the child is insufficiently developed to permit the kind of auditory feedback necessary to ‘fix’ the response in the hierarchy. Thus the necessity to relearn these phones later, when the auditory, perceptual and cerebral development has proceeded to the point where it is possible. In any event, the production of [k] before [i] is perhaps so different a task from the cooing [k] beore [u], that there is little or no transfer of response availability.