¹ Zellig S. Harris and Henry M. Hoenigswald, by their suggestions and questions, have greatly helped to improve the adequacy of this system. It was outlined in a paper read before the Linguistic Society in New York in December 1944. The following books and papers are cited by authors' names only:

Bernard Bloch and George L. Trager, Outline of Linguistic Analysis, 1942.

G. L. Trager and Bernard Bloch, The Syllabic Phonemes of English, Lang. 17.223–46 (1941).

Leonard Bloomfield, Language, 1933.

Charles F. Hockett, A System of Descriptive Phonology, Lang. 18.3–21 (1942).

George L. Trager, The Theory of Accentual Systems, LCP 131–45 (1941).

² Bernard Bloch, Phonemic Overlapping, American Speech 16.278–84 (1941).

³ Yuen Ren Chao, Non-Uniqueness of Phonemic Solutions of Phonetic Systems, Bull, of the Inst, of Hist, and Philology (Acad. Sinica) 4.363–98 (1934).

⁴ Hockett 9; Morris Swadesh, The Phonemic Interpretation of Long Consonants, Lang. 13.1–10 (1937). See especially the first page of Swadesh's article.

⁵ Trager 132, paragraph 4.

⁶ Hockett §5.31.

⁷ The fact that vowels and consonants, as well as pitches and stresses, are abstracted features rather than segments does not require us, with Hockett (8), to regard phonemes as classes of classes of features, or even as classes of features, instead of as classes of segments. One segment can be assigned to two phonemes at once, provided the two phonemes are of different dimensions. Suppose that there are just two pitches, P and Q, and just two vowel or consonant qualities, A and E. Then as primary classes there will be AP (segments with the quality A and the pitch P) AQ, EP, and EQ. The phoneme A = AP + AQ = A(P + Q), since P + Q is the total class of pitches; likewise the pitch phoneme P = AP + EP = P(A + E), since A + E is the total class of qualities.

⁸ Hockett 9, §6; Bloch and Trager 34; Trager and Bloch §2.

⁹ Hereinafter abbreviated SS.

¹⁰ Trager 131 (par. 5), 132 (par. 4); Bloch and Trager, implied 39 (par. 2), 41, 47–52.

¹¹ It is chiefly in this negative sense that juncture phonemes (Bloch and Trager 41; Hockett §6, §§7.74 ff.) are to be regarded as SS, though Bloch and Trager (34–6) regard the members of juncture phonemes as phonetically SS in some sense.

¹² Hockett §§5.31–2.

¹³ Cf. Trager 135 (par. 2).

¹⁴ Reported by Hockett §7.61.

¹⁵ The argument in favor of regarding pitch contours as morphemes is given below in Part 4.

¹⁶ 116; cf. 163.

¹⁷ Bloomfield 114 (par. 2).

¹⁸ Trager; Trager and Bloch; Bloch and Trager.

¹⁹ Mere number of phonemes, as proposed by Hockett §6.12, is not always an acceptable measure of simplicity, nor is simplicity always a last resort as a criterion. Often economy and pattern (distribution) favor different analyses.

²⁰ Allophones of pitch phonemes can, of course, differ from each other only in the one dimension or respect of pitch.

²¹ Cf. Bloomfield §11.2 on favorite sentence-types.

²² Distinctive, and having a kind of meaning which we will agree (cf. Bloomfield 114) is not merely cultural but linguistic.

²³ English Intonation, 1922.

²⁴ In the Grammar of Spoken English by himself and F. G. Blandford (2d ed. 1939), Palmer divides his former No. I into two: a high-falling and a low-falling (§§20–39). This is a change of details, not of method.

²⁵ Simultaneous Components in Phonology, Lang. 20.181 ff. (1944).