Published online by Cambridge University Press: 22 April 2026
The major portion of this article is devoted to a criticism of Twaddell's monograph ‘On Defining the Phoneme.‘ The criticism is adverse throughout. Still, if the result of negating a negative is positive, it may be justifiable to presume that if this is a contribution at all, it is of a constructive character. Some negative assertions are made in said monograph concerning the results thus far achieved in the physical analysis of speech sounds. If we do not misconstrue those assertions, they are at variance with the literature on the subject. We wish to present the evidence.
1 W. Freeman Twaddell, On Defining the Phoneme, Publications of the Linguistic Society of America, Language Monographs, No. 16.
2 Ibid. 24.
3 Cf. C. E. Parmenter and S. N. Treviño, The Length of the Sounds of a Middle Westerner, American Speech 10.129: ‘The exact limits of the sounds were sometimes difficult to determine, particularly in the case of the voiced continuants. But by comparing the troublesome cases with other occurrences of the same sound, it was possible to arrive at fairly reliable approximations. In a few cases, such as the combination [ər], it was impossible to find a basis for division and the time of the two sounds was arbitrarily divided equally between them.‘
4 Ibid. 24.
5 Leonard Bloomfield, Language 80 (New York 1933).
6 L'analisi elettro-acustica del linguaggio (Milan 1934).
7 Op. cit. 29.
8 Ibid. 91
9 Ibid 68
10 Ibid. 91 (footnote).
11 Ibid. 66, 67.
12 American Speech 10.150. (The brackets are ours.)
13 ‘Tout oeil exercé lit dans les oscillogrammes (à part quelques rares exceptions) chaque consonne comme chaque voyelle, ce qui prouve que la courbe oscillographique a une physionomie caractéristique même pour chaque consonne.‘ A. Gemelli et G. Pastori, Quelques recherches sur la structure des consonnes 99 (Pisa 1934). (From the French translation of a portion of their work.)
14 Perhaps it was not long before or after Twaddell prepared his manuscript that the announcement of O. H. Schuck's recent invention, the sound prism, appeared in the Proceedings of Radio Engineers 22.1295–1310, Nov. 1934. It is a ‘new rapid acting heterodyne wave analyzer’, which enables the experimenter to make observations on the components of sound waves in one tenth of a second, and photograph their spectra at any given moment during the observation.
15 Quoted by Twaddell 23, 24.
16 A familiar instance is that of spelling a word in a telephone conversation. We have to say ‘p as in Peter’, ‘s as in Sam’, due to the fact that the telephone does not transmit the high frequency characteristic of these consonants. Twaddell says (24) that we understand speech transmitted by the telephone, that its low frequency-response does not affect the phonemes. But notice the artificial manner in which the telephone operators have to repeat the words ‘five’ and ‘nine’ to verify what they hear. Notice the difficulty of distinguishing consonants with high frequencies over the telephone when the context does not help much, as in the words ‘fifth’ and ‘sixth’.
17 One of these tests is described in detail by John C. Steinberg, Effects of Distortion upon the Recognition of Speech Sounds, The Journal of the Acoustical Society of America 1.121–137 (1929).
18 Ibid. 24.
19 The statement concerning the differentiae of the [ĭ] as given above is not made by any of the authors cited. It was inferred by us from their data. I do not cite Harvey Fletcher (Speech and Hearing, New York 1929) because his figures (58) are approximate averages of the results obtained by Stumpf, Miller, Paget, Crandall, and perhaps his own observations. With regard to Miller, it should be noted that he does not give two formants for all the vowels. Our statement, therefore, holds strictly only for [i] and the three other vowels for which the phonodeik showed two formants, but it is not in disagreement with the data on the vowels for which Miller gives only one formant. The reports referred to are: Irving B. Crandall, The Sounds of Speech 26, Bell Telephone Laboratories, Reprint B-162-1 (Nov. 1935); Dayton C. Miller, The Science of Musical Sounds (New York 1926); Sir Richard Paget, Human Speech (New York 1930); T. Chiba, Research into the Characteristics of the Five Japanese Vowels (Yokohama 1931); Gemelli and Pastori, op. cit.
20 Op. cit. 24.
21 Op. cit. 31.
22 Op. cit. 33, 34.
23 Three of the 92 known elements are on the borderline between metallic and non-metallic, while a common and important characteristic of the elements, namely, their capacity for chemical combination, is not exhibited by 6 peculiar gasses. Cf. Bernhard Bavink, the Natural Sciences 3, Eng. trans. of the 4th Germ. ed. (New York 1932).
24 Op. cit. 58.
25 Ibid. 58.
26 'However interesting these and other fictions, such as angels, devils, pixies, spirits, etc. may be for the logical theory of existential propositions, they are of minor importance for our present theme. At most they concern us in so far as such a judgment as “matter consists of atoms” or “the curved line consists of infinitesimals” is to be understood only as a fictive judgment in which no existence is predicated.' Then, referring to mathematical conceptions: 'All these concepts are contradictory fictions, mathematics being based upon an entirely imaginary foundation, indeed upon contradictions.' H. Vaihinger, The Philosophy of 'As if', Eng. trans. (New York 1925) 82 and 51, respectively.
27 Op. cit. 87.
28 A. D. Ritchie, Scientific Method 98 (London 1923).
29 Ibid. 49.
30 Ibid. 37 (footnote).