¹ A set of postulates for phonemic analysis, Lg. 24.3-46 (1948), esp. 22-6, Postulates 25-31; and the definitions prefixed to an account of Japanese phonemes, Lg. 26.89-90 (1950).

² See Eli Fischer-Jørgensen, Remarques sur les principes de l'analyse phonémique, TCLC 5.213-34 (1949); John Lotz, Speech and language, Journal of the Acoustical Society of America 22.712-7 (1950); Kenneth L. Pike, More on grammatical prerequisites, Word 8.106-21 (1952).

³ For the definition of technical terms, see the two passages cited in fn. 1. Specific references: for allophone, Lg. 24.38 (§54.2); for dialect, 24.8 (§3.2); for environment, 24.22 (§25.3), 26.89; for phone, 26.89, and cf. 24.12 (§12.3), 24.35 (§52.3); for range, 24.30 (§42.3).

⁴ Thus, the range of [n] (the class of all voiced nasalized apical flaps or one-tap trills) in some sample of English can be very simply defined : members of the allophone occur after a vowel or semivowel and before a weak-stressed vowel (as in winter, painting, counter, carpenter, Mount Olympus). The range of [η‘] (the class of all front or prevelar voiced nasals), again in some sample of English, requires a general definition somewhat more complex: members of this allophone occur (1) final in a syllable after a front vowel (as in bingo, sang, making) or after the semivowel [j] (as in boing or a monosyllabic pronunciation of being, saying, trying) ; (2) after a front vowel or [j] before prevelar [k‘] in a strong-stressed syllable (as in think, winked, jinx, Schenck, Jenks, bank, Manx, Schumann-Heinck, oinkoink); and (3) after the vowel [e] before [þ] in a strong-stressed syllable (as in length). If the choice of defining features is correct, any environment in which a member of one of the allophones occurs will be found to exhibit one of the features referred to in the definition. But there will be countless environments, both among those that actually occur in the sample studied and among the vastly greater number that do not occur but can be invented, which exhibit one of these features and yet contain no member of the allophone in question.

⁵ Provided that each environment is interpreted as composed of allophones rather than phones. The statement in the text can be more pedantically paraphrased as follows: To any environment of P consisting of the phones a ₁b ₁c ₁ ... there corresponds an environment of Q consisting of the phones a ₂b ₂c ₂ ..., such that a ₁ and a ₂, b ₁ and b₂, c ₁ and c ₂, etc., belong respectively to the same allophones—and conversely.

⁶ These terms are adopted from Rulon S. Wells, Automatic alternation, Lg. 25.99-116 (1949), esp. 104. (Wells uses the terms communis and propria for parts of morphs.)

⁷ This postulate is intended to replace the seven postulates referred to in fn. 1. Other parts of my 1948 set of postulates are similarly in need of restatement and simplification.