¹ The basic assumption of this paper thus differs from the conclusions reached by Bernard Bloch, A set of postulates for phonemic analysis, Lg. 24.7, §1.7 (1948): ‘The totality of the possible utterances of one speaker at one time in using a language to interact with one other speaker is an idiolect. ... The phrase “with one other speaker” is intended to exclude the possibility that an idolect might embrace more than one style of speaking: it is at least unlikely that a given speaker will use two or more different styles in addressing a single person.‘ We suggest, however, that a speaker—even in the middle of a sentence—may suddenly lower his voice to a whisper, or burst into tearful speech, or shift to a caressing quality, or the like. Socially pertinent differences of style cannot safely be ignored; they must be handled in some way in our phonemic assumptions and procedures.

² The authors of this article approach phonemic study from widely different types of experience. Fries has concentrated on historical studies, Pike on reducing languages to writing. This article springs from discussions between them. Fries contributed especially to the discussion of historical change and to the principles outlined in Part 1; Pike was largely responsible for implementing these principles with the procedures and assumptions given in later parts of this paper.

An abstract of this paper was presented to the summer meeting of the Linguistic Society of America, 31 July 1948. Upon that occasion we were pleased to hear from Professor J. R. Firth of the School of Oriental and African Studies, University of London, that he had been making abstractions on various levels which might be related to these. See now fn. 30.

³ Taken from Kenneth L. Pike and Eunice V. Pike, Immediate constituents of Mazateco syllables, IJAL 13.79 (1947), with supplementary data from the second-named.

⁴ Brackets enclose phonetic symbols; virgules (slant lines) enclose phonemic symbols; quotation marks enclose letters of a suggested orthography.

⁵ We do not imply that a native's uncritical remarks are to be accepted as valid analyses. Rather we claim that a linguist can objectively observe the behavior of the native and draw certain structural implication from his observations.

⁶ This was the method of Pike in Phonemics: A technique for reducing languages to writing 142–3 (University of Michigan Publications: Linguistics, Vol. 3; Ann Arbor, 1947). It is precisely the premises given in the first part of the present paper which allow the gathering together of various items presented in the book, and the further procedural advance.

⁷ Even the most widely accepted procedures of the current type do not yet result in complete uniformity in the analysis, say, of English. Without considerable discussion it is highly improbable that we can get as much agreement on the new procedures as there is on the old ones. It appears to us, however, that the attempt to set up tentative procedures will be justified if it promotes more intensive investigation of pertinent problems.

⁸ The underlying premises and methods for handling them are to be found in Pike's Phonemics.

⁹ Some of them have been presented by Pike in his Phonemics.

¹⁰ Edward Sapir used this type of evidence in his article La réalité psychologique des phonèmes, Psychologie du langage 247–65 (= Journal de psychologie, Vol. 30 [1933]), summarized by W. F. Twaddell, On defining the phoneme 10–4 (Language Monograph No. 16; Baltimore, 1935).

¹¹ If, however, the student is told to write 'ə' whenever he hears the /ə/ unstressed, but to write 'ʌ' whenever he hears it stressed, he learns to apply the rule easily (though without necessarily hearing the quality difference) simply by listening to the stresses.

¹² For an elaboration of this evidence, see Pike, On the phonemic status of English diphthongs, Lg. 23.151–9 (1947).

¹³ In a dialect which distinguishes between matter and madder.

¹⁴ We assume that traditional spelling cannot be the whole source of the pressure which causes speakers of English to isolate the middle consonant of matter as [t^h].

¹⁵ For an experiment in the problem of transfer from Spanish to English nasals, see A. H. Marckwardt, Phonemic structure and aural perception, American Speech 21.106–11 (1946); for a similar experiment with voiceless sibilants and affricates, see Marckwardt, An experiment in aural perception, The English Journal 33.212–4 (1944). Marckwardt finds that in recording a series of dictated words there is less confusion when a phoneme appears in a strange position than when submembers of a phoneme of one language appear as separate phonemes in another language.

¹⁶ See the reference in fn. 12.

^16a In a recent paper, Juncture in modern standard German, Lg. 23.212–26 (1947), W. G. Moulton writes: 'The places where /+/ [open juncture] occurs usually coincide with syntactic and morphological boundaries. The only exceptions are a few words (all of foreign origin) in which open juncture and onset of strong stress precede a voiceless stop or a vowel: /+pa+^╹pi:r+ “paper”, ... /+ru:+^╹i:nen+/ “ruins”, etc.' (224). Since these loans are the specific data (cf. Moulton's note on 225) which caused his rejection of grammatical borders as criteria in his phonemic analysis, Pike suggested (Grammatical prerequisites to phonemic analysis, Word 3.171–2 [1947]) that 'one should hesitate to allow a small residue of words of foreign origin to prevent a general formulation, at least until vigorous attempts have been made to follow the analysis which otherwise would more easily represent the total grammatical-phonological structure of the language; possibly a descriptive expedient might be found which would preserve the easier formulation without doing violence to the facts.' We now raise the question whether the postulation of a coexistent system might be such an expedient.

¹⁷ When the present article was undertaken, neither Pike nor Eunice V. Pike had observed this fact. The latter now reports it to us. No count is as yet available to show the proportion of monolinguals or bilinguals using each of the two pronunciations. A detailed investigation, with close attention to the assumptions here listed, would probably yield further interesting observations.

¹⁸ Information from W. C. Townsend of the Summer Institute of Linguistics, as reported to Pike.

¹⁹ Observations of pattern impact on the phonetics of bilinguals, Sapir memorial volume (Language, culture and personality) 49–65 (Menasha, Wis., 1941). Many other illustrations of problems discussed in the present paper are to be found in Swadesh's article; but he does not attempt to set up a procedure to handle them.

²⁰ W. L. Wonderly, Phonemic acculturation in Zoque, IJAL 12.92–5 (1946), reprinted in Pike's Phonemics 202–6.

²¹ An attempt to deal with this problem in English will be included in Fries' History of the structure of English from 1100 to 1900.

²² Data from Eunice V. Pike, Problems in Zapotec tone analysis, IJAL 14.161–70 (1948); dialect of Villa Alta, Oaxaca, Mexico.

²³ Length is phonemic in the Zapoteco of Juchitan. Data gathered by Pike in collaboration with Velma Pickett and Marjorie MacMillan.

²⁴ For further discussion of these types see Pike's Phonemics 65–6, 124–5, 142–3.

²⁵ It is difficult, for example, to find whispering among the Mixtecos; Pike, after a considerable period of residence among them, cannot recall a single instance. For privacy, they speak in low voices. Pike found it difficult to teach Mixtecos to whisper when on a few occasions he wished to compare whispered with voiced speech. Similarly, in the dialect of San Miguel el Grande, Pike can recall hearing only three persons sing, without urging, even a single Mixteco song; and what they sang was heavily mixed with Spanish loan words. Yet certain other Mixteco dialects have numerous native songs.

²⁶ Pike, The intonation of American English 34–5 (University of Michigan Publications: Linguistics, Vol. 1; Ann Arbor: 1945).

²⁷ See G. M. Cowan, Mazateco whistle speech, Lg. 24.280–6 (1948).

²⁸ The investigator must also determine, by further analysis, whether these additional characteristics are in turn subdivided into an intricate system of contrasts (like the intonations of American English). So far, no such contrastive system has been observed for whisper. In song, such a system is obviously constituted by the tones of the musical scale, even though the contrasts are artificial (not ‘natural to the universe‘) and different in different cultures.

²⁹ In the Mazateco article cited in fn. 3, an asterisk was used for this purpose. A dagger seems better, to avoid confusion with the traditional use of the asterisk to mark unattested forms.