Hostname: page-component-89b8bd64d-sd5qd Total loading time: 0 Render date: 2026-05-05T14:48:30.261Z Has data issue: false hasContentIssue false
  • English
  • Français

On Defining the Phoneme

Published online by Cambridge University Press:  22 April 2026

W. Freeman Twaddell
Get access Rights & Permissions [Opens in a new window]

Extract

For many years, now, and particularly within the last decade, there has been increasing use of the term ‘phoneme’, ‘wenn man sich auch nicht immer über seine definition einig ist’. Parallel to this increased use of the term, there has been the re-emergence of a concept of some unit of spoken language, a unit which is not the same as a ‘sound of speech’. It has long been known to phoneticians and linguists in general that the sounds of speech, even within the narrowest restrictions of time and place, even within the usage of a single individual, present an almost infinite variety. Scarcely any two speakers of a given dialect pronounce the same word exactly alike, either as to their articulatory movements or as to the sound-waves which those movements set up in the atmosphere. And yet, within the communicative and expressive medium, those (articulatorily and acoustically) slightly different processes are still the same word; and the ‘sounds’ which comprise this same word are in some way the same sounds, within the frame of that communicative and expressive medium which is the language of the community. It is the recognition of this sameness, this effective unity, which has found expression in the term ‘phoneme’ as a unit of spoken language.

Information

Type
Research Article
Information
Language , Volume 11 , Issue 1-Part2: Language Monograph No. 16: On Defining the Phoneme , March 1935 , pp. 5 - 62
Copyright
Copyright © 1935 Linguistic Society of America

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable

References

page 5 note 1 In the preparation of this study I have drawn heavily upon several of my friends. The discussions with Mr. Martin Joos have been especially fruitful in numerous invaluable suggestions and criticisms. My colleagues Professor A. Senn, Professor E. I. Haugen, and Mr. David Sheldon have been most helpful. I am indebted to Professor Leonard Bloomfield for his patience and kindness in discussing the phoneme and our views on it personally and by correspondence.

page 5 note 2 Jespersen, Linguistica 214.

page 6 note 3 Journal de Psychologie 30.229.

page 6 note 4 Le Maître Phonétique (1933) 29.

page 6 note 5 Le Maître Phonétique (1934) 46; cf. also H. Pedersen in Litteris 5.154.

page 6 note 6 Language 2.153.

page 6 note 7 Language 8.220 ff.; 10.32 ff.; see also Modern Philology 25.211 ff.

page 7 note 8 Cf. L. Weisgerber, Vorwort zur Probe eines Wörterbuchs der sprachwissenschaftlichen Terminologie 5 ff.; Beiheft IF 51.

page 7 note 9 The definition to be found in The Cyclopedia of Medicine 6.149 f. (F. A. Davis Company, Philadelphia, 1933) appears to have little bearing on linguistic methodology: ‘A phoneme is an auditory hallucination consisting of spoken words’.

page 8 note 1 Norsk Tidsskrift for Sprogvidenskap 1.22 ff.

page 8 note 2 Cf. Trubetzkoy, in Actes du deuxième congrès international de linguistes 109: ‘Die Phonologie befasst sich ... mit den Phonemen, d.i. mit den sich in den Sprachlauten realisierenden, im Sprachbewusstsein lebenden Lautabsichten.‘

page 8 note 3 Donum Natalicium Schrijnen (1929) 36.

page 8 note 4 Travaux du Cercle Linguistique de Prague (hereafter referred to as TCLP) 1.39.

page 8 note 5 TCLP 1.10 f.

page 9 note 6 TCLP 4.53 ff.

page 9 note 7 Modern Philology 25.211 ff.; his Language 32 ff.; and reviews in Language 8.220 ff.; 10.32 ff.

page 9 note 8 The principles presented in this section are limited strictly to applications of scientific methodology, and accordingly do not involve a discussion with those to whom behaviorism is unacceptable as a psychological assumption or (irrelevantly) as a ‘philosophy’. Whatever our attitude toward mind, spirit, soul, etc. as realities, we must agree that the scientist proceeds as though there were no such things, as though all his information were acquired through processes of his physiological nervous system. In so far as he occupies himself with psychical, non-material forces, the scientist is not a scientist. The scientific method is quite simply the convention that mind does not exist: science adopts the nominalistic attitude toward the problem of the universals, in matters of procedure.

page 10 note 9 Das morphonologische System der russischen Sprache (TCLP 5.2) 35.

page 10 note 10 Fakultative Sprachlaute, in Donum Natalicium Schrijnen 38.

page 10 note 11 Donum Natalicium Schrijnen 35.

page 10 note 12 Language 1.37 ff.

page 11 note 13 Journal de Psychologie 30.247 ff.

page 12 note 14 We learn from the Proceedings of the American Academy of Arts and Sciences 65.2.299, that he was Tony Tillohash, who at that time was ‘just about to complete his course at the Carlisle Indian School in Pennsylvania’.

page 13 note 15 Proc. Amer. Acad. 65.1.70.

page 14 note 16 Journal de Psychologie 30.247.

page 14 note 17 In S. A. Rice, Methods in Social Science (1931) 305.

page 14 note 18 In his Language 58.

page 15 note 19 For a presentation of other grounds of objection to the definition of the phoneme in terms of mental reality, see (TCLP 4): D. Čyževśkyj (3 ff.); K. Bühler (22 ff.); and W. Doroszewski (61 ff.).

page 15 note 20 In his Language (1932) xl.

page 15 note 21 Lexique de la terminologie linguistique, Paris, 1933.

page 15 note 22 English Studies 15.82.

page 15 note 23 Xenia Pragensia (1929) 435.

page 15 note 24 Donum Natalicium Schrijnen 47.

page 16 note 25 TCLP 1.68.

page 16 note 26 Xenia Pragensia 432 ff.

page 16 note 27 TCLP 1.10.

page 16 note 28 TCLP 4.311.

page 17 note 1 The definition of this term will be discussed later.

page 17 note 2 The definitions of K. Bühler appear to belong here: ‘(Jene) lautlichen Besonderheiten, ... die von den einzelnen Sprachen zum Range von diakritischen Zeichen erhoben worden sind’ (TCLP 4.44). ‘Une invariante qui sert de marque diacritique en vue de distinguer les significations intellectuelles’ (TCLP 4.295).

page 17 note 3 W. Doroszewski's definition may belong here: ‘Mais je pense qu'il est plus simple de considérer comme “phonème” le son tel qu'il est réalisé en position isolée, en traitant de “variantes” les produits combinatoires’ (TCLP 4.73).

page 22 note 4 These complexities connected with the use of the terms ‘feature’ and ‘features’ represent for me the greatest difficulty in determining precisely the meaning of Bloomfield's definitions of the phoneme. That difficulty is due not only to the lack of any preliminary definition of the terms as applied to events, but also to the unannounced introduction of different usages for the term. At all events, it is hard to see how any definition of ‘phoneme’ or ‘feature’ could warrant the optimism with which Bloomfield writes (Modern Philology 25.217): ‘The logical demand that a science speak in quantitative terms is met by linguistics because it speaks in terms of phonemes’. The phoneme as a quantitative unit (even of feature) is a difficult enough concept. But these units are by no means quantitatively comparable; one [r]-phoneme is not equal to 3.76 [p]-phonemes; and unless quantitative units are quantitatively comparable, it appears unjustified to claim that we ‘speak in quantitative terms’.

page 22 note 5 In so far as the ‘constant shape in successive utterances’ is only ‘relative’, the recurrence of the phonemes is only relative. But if the phoneme is ‘a minimum same’; if ‘the distinctive features of the word (pin) are three indivisible units’, and ‘each of these units occurs also in other combinations’ (Language 79), then the recurrence must in some sense be absolute; and in that same sense the ‘shape in successive utterances’ must be absolutely constant. If that ‘shape’ is only relatively constant, then the recurrence is only relative, the sameness of phonemes in successive utterances is only relative; and the entire procedure is necessarily invalid.

page 24 note 6 Cf. Hdb. d. Phys. 8.559 ff., 603; H. Fletcher, Speech and Hearing 27.

page 25 note 7 In H. E. Palmer, The Principles of Romanization (Tokyo 1931) 44.

page 25 note 8 Further, quite apart from technical considerations, it is difficult to see what reason there should be for expecting to find any common acoustic characteristic of the [p]s and [t]s, respectively, in pit and tip, especially if the final stops are pronounced without explosion.

page 26 note 9 In Proceedings of the International Congress of Phonetic Sciences, reprint distributed with Le Maître Phonétique (1933).

page 26 note 10 It should be noted that Palmer explicitly describes the ‘phone’ as an abstraction.

page 27 note 11 Cf. Definition of a phoneme (Le Maître Phonétique [1929] 43 f.): 'Two sounds of a language are said to belong to the same phoneme when they are related in character and it so happens that one of them never occurs in exactly the same situation as the other in connected sentences of that language. ... It would indeed be possible to group the sounds of a language into phonemes without knowing the meanings of any words.' Query: without knowing that they are words and different words? Cf. Jones's discussion (Le Maître Phonétique [1933] 30) of Palmer's principle of 'multiple identity'.

page 28 note 12 Cf. Bloomfield, Language 124 ff.

page 28 note 13 Particularly since it could fit equally well into the ‘internal organization’ within either group.

page 28 note 14 English Studies 15.89. Cf. also Vachek's Daniel Jones and the Phoneme, in Charisteria Guilelmo Mathesio (Prague 1932) 25 ff.

page 29 note 15 Le Maître Phonétique (1933) 53; (1934) 54.

page 29 note 16 Page 45, footnote 2.

page 30 note 17 Language 10.117 ff.

page 34 note 1 Linguistica 180 f.

page 34 note 2 Modern Philology 25.216 f.

page 35 note 3 Upper limit of 1230 cycles.

page 36 note 4 Upper limit of 870 cycles.

page 36 note 5 Cf. Bloomfield, Language 21 ff.

page 37 note 1 By regarding ‘things’ as entities of more than molecular magnitude, and ‘events’ as changes in position of such things, we avoid discussions of the identical nature of such abstractions as atoms and quanta.

page 37 note 2 Cf. Bloomfield, Language 78, 144.

page 45 note 3 This list of certain articulatory components is very sketchy, of course, omitting as it does factors of muscular tension and the time-factors of duration and acceleration. The list suffices, though, for our present purposes.

page 49 note 4 Cours de Linguistique Générale (1922) 166.

page 50 note 5 Here, as elsewhere, I make traditional use of the IPA symbols in transcribing American English rather than Bloomfield's, not because I regard that traditional use as inherently better—for I consider Bloomfield's transcriptions of themselves simpler and more rational—but quite simply because phonetic transcriptions, for the reasons given elsewhere in this study, are arbitrary and conventional; they partake at least in part of the nature of orthographies, and accordingly an established tradition of transcription is preferable to any new one which is based on the same principles. Cf. R.-M. S. Heffner, Concerning Transcription, in Language 10.288 ff.

page 51 note 6 Speech and Hearing 60.

page 52 note 1 Modern Philology 25.217.

page 52 note 2 TCLP 1.39.

page 52 note 3 TCLP 4.77.

page 52 note 4 Linguistica 391.

page 53 note 5 (6–1) + (6–2) + (6–3) + (6–4) + (6–5). Thus, bite: beet, bit, bait, bet, bat; beet: bit, bait, bet, bat; bit: bait, bet, bat; bait: bet, bat; bet: bat. Some readers may prefer the formulation: If x is the maximum number of significant phonological differentiations within a given articulatory range in a language, then 2x = n(n – 1), where n is the maximum number of phonemes in that range, and (n – 1) is the number of consecutive phonological relations within that range.

page 55 note 6 Linguistica 390.

page 56 note 1 TCLP 1.41.

page 56 note 2 Linguistica 134 ff.; quotation from 143 f.

page 57 note 3 That is, not only are the classes beet: bit: bait: bet: bat and seek: sick: sake: sec: sack similarly differentiated, but there is considerable objective similarity between the vowels of ‘beet’ and ‘seek’, ‘bit’ and ‘sick’, etc.

page 58 note 1 TCLP 1.41.

page 58 note 2 For an excellent discussion of this point, see K. Bühler, Phonetik und Phonologie (TCLP 4.22 ff.); see also Jespersen, Linguistica 213.

page 59 note 3 Hereafter, the term ‘set of micro-phonemes’ means ‘set of micro-phonemes which are similarly ordered in sequentially similar classes’. Thus the vowels of ‘beet: bit’ correspond to a class of micro-phonemes; the vowels of ‘beet, feet’ and of ‘beet, bean’ correspond to sets of micro-phonemes. A set of micro-phonemes is accordingly a fraction of a macro-phoneme.

page 60 note 4 It is worth noting that this assumption of the meaningless phoneme, which alone validates the claim of random sampling and thereby justifies operation with ‘functional burdening’, has sometimes been explicitly denied by the very linguists who speak of ‘functional burdening’.

page 60 note 5 Cf. B. Trnka, Some Remarks on the Phonological Structure of English, in Xenia Pragensia 357 ff.

page 62 note 6 Cf. Modern Philology 25.225.