Published online by Cambridge University Press: 22 April 2026
1 Bloomfield instances the spirant [γ] in some types of German pronunciation, where it is a variant of the phoneme [g ], stating that in Dutch and in Modern Greek [γ] and [g] occur as separate phonemes. But he himself now writes to me that these examples will not serve, for while the sounds are distinct in the languages named, [g] is in Dutch only a sandhi-product of [k] before [b] or [d], and in Modern Greek is the product of [k] after a nasal. For Modern Greek it might be added that [g ] occurs initially in native words if a preceding nasal has disappeared and also in borrowings, such as [gaz] 'gas', written γκáζ. But he is right in separating [y] and [g] in these languages, whereas they cannot be separated in German.
2 For this statement, I refer to Hossfeld's Japanese Grammar, by H. J. Weintz (printed by Peter Reilly, Philadelphia, 1914).
3 Bloomfield writes me: 'German has only one vowel of the e-type, hence no [ε] appears in German transcriptions.' But to my ear, the vowels of Bett and Beet are quite distinct in quality as well as in quantity. On this, Bloomfield writes that he regards the sign of length in [be:t] = Beet as an adequate indication of the difference, especially since the e-vowel is not uniform in quality among native speakers of German.
4 Cf. also footnote 9. My experience with speakers of French is that é in non-final syllables is likely to become è (i.e., [ε]), rather than the reverse. Thus I recall a Parisian professor who pronounces répéter as though rè-pè-té (or [rεpεte]), and my French colleagues tell me that his peculiarities of speech are typical of Parisian French of to-day.
5 Of course it would be impossible to use [a], [e], [i], [o], [u], with their English values, in any system; English is too aberrant in its system of vowels and diphthongs, to be used as a basis for a notation, except with very great limitations.
6 Cf. also the remarks of Franklin Edgerton in JAOS 53.295–7, on this and related matters, in a review of this same volume.
7 The examples illustrating the sounds under discussion are not all well-chosen. French gai is given by the Petit Larousse of 1925 as ghè, i.e., with the vowel of dette (an older dictionary in my possession gives it as ghé); we might better substitute été in the table. Also, the use of English examples ending in a nasal makes the comparison difficult, since the nasal somewhat alters the preceding vowel-sound (but not enough to make of it a different phoneme!); one might better list met, mat, bit, but respectively, for the first four.
8 For that matter, is not a third e-symbol required for French? The vowel that is nasalized in French vin is quite distinct from that in dette; but Bloomfield (228) transcribes [≠εη], considering that the sign of nasalization is an adequate indication of automatic change of the sound. However, his phonemic writing seems to me to require [vin] for 'structural' reasons (see latter part of review), cf. masc. fin, fem. fine.
9 Why should the 'structural' transcription be limited to vowels? Both in German and in Russian final voiced stops become voiceless; thus writings with [d], e.g. in German [pfe:rd] 'horse' and Russian ['gorod] 'city', are required in view of other case-forms, where [d] remains before the vowel of the flexional syllable. Bloomfield, however, writes that he 'transcribed [t] because in these languages [d] and [t] are distinct phonemes. Perhaps this principle is out of place, and I should really have used [d], but I believe the point is different from that of [a] vs. [o] in the second syllable of the Russian word, or of [e] vs. [ε] in the German word.'
10 I mean, of course, in a book of the type under discussion, where examples must be drawn from many languages. Where only one language is under discussion, either the phonetic transcription or the phonemic is practicable; there should in all cases be an accurate description of the values of the symbols, and for the phonemes there should be also a detailed account of their automatic variations and of the conditions in which the variations take place. The giving of such descriptions for a number of languages, which would be necessary for a book like Bloomfield's, would add unduly to its bulk. On the other hand, for a volume on a single language, the phonemic notation has many advantages, to which I am neither blind nor deaf; and for some purposes it is a great advance over the phonetic notation.