¹ The dialect here studied is a somewhat generalized northeastern variety of standard American English. For the methodological groundwork of this paper see Zellig S. Harris, Morpheme alternants in linguistic analysis, Lang. 18.169–80 (1942); Rulon S. Wells, Immediate constituents, Lang. 23.81–117 (1947); and C. F. Voegelin, A problem in morpheme alternants and their distribution, Lang. 23.245–54 (1947). Compare now also Charles F. Hockett, Problems of morphemic analysis, Lang. 23.321–43 (1947). Though my views in general agree with Hockett's, it will be observed that his treatment of certain problems of English inflection (especially in §24 of his paper) differs markedly from the one here proposed.

I have profited from discussions with R. S. Wells, W. F. Twaddell, and Martin Joos.

² Henry Sweet, A new English grammar logical and historical 1.391–428 = §§1283–493 (Oxford, 1892); Harold E. Palmer, A grammar of spoken English on a strictly phonetic basis 88–122 = §§199–270 (Cambridge, 1930); George O. Curme, A grammar of the English language, II. Parts of speech and accidence 241 ff., esp. 269–96 = §60, 304–19 = §63 (Boston etc., 1935); Charles Carpenter Fries, American English grammar 59–71 (New York and London, 1940); Otto Jespersen, A modern English grammar on historical principles, VI. Morphology 28–83 = Chaps. 4–5 (Copenhagen, 1942); Charles F. Hockett, English verb inflection, SIL, Vol. 1, No. 2 (1942). For an entertaining travesty of American English verb inflection see H. L. Mencken, The American language⁴ 427–47 (New York, 1936).

³ Wells, Lang. 23.93–8.

⁴ Leonard Bloomfield, Language 161 and 138 (New York, 1933).

⁵ The phonemic transcriptions in this paper necessarily reflect my own speech, except that a few distinctions not commonly made in other dialects of English have been disregarded. On the transcription of vowels and diphthongs see George L. Trager and Bernard Bloch, The syllabic phonemes of English, Lang. 17.223–46 (1941), and cf. Lang. 19.189 fn. 15 (1943). The following stressed syllabics occur in the dialect here transcribed: /i/ in pit, /e/ in pet, /æ/ in pat, /a/ in pot, /Λ/ in cut, /o/ in coffin, /u/ in put, /ə/ in just (adverb); /iy/ in beat, /ey/ in bait, /ay/ in bite, /oy/ in boil, /uy/ in ruin (monosyllabic); /aw/ in bout, /ow/ in boat, /uw/ in boot; /ih/ in theater, /eh/ in yeah, /æh/ in mad, /ah/ in calm, /oh/ in law, /əh/ in er (hesitation form); /ir/ in here, /er/ in there, /ar/ in part, /or/ in port, /ur/ in sure, /ər/ in curt; /ihr/ in beer, /ehr/ in bare, /ahr/ in bar, /ohr/ in bore, /uhr/ in boor, /əhr/ in burr. Pitch will not be marked in this paper: stress will be marked only in transcriptions of whole utterances. A space between words has no phonetic or phonemic significance.

Needless to say, the treatment of verb inflection offered here does not depend on the system of transcription. The cited forms could be written just as well according to any other system, so long as it recognized the existing phonemic distinctions in American English.

⁶ Read: minus /ey/ plus /u/. Cf. Zellig S. Harris, Lang. 21.121 and fn. 3 (1945).

⁷ We shall pay no attention in this paper to derivational suffixes (like the -er in waiter, the -ful in wakeful, or the -th in growth), except for a brief mention in §4.5. On the difference between inflection and derivation see Bloch and Trager, Outline of linguistic analysis 54–5 = §4.3 (Baltimore, 1942).

⁸ The verb be has inflected forms not included among these four categories. Since this verb presents a special and vexing problem, we shall postpone all reference to it until §5. Inflected verb forms used only with a subject in the 2d person singular (archaic forms like waitest, preterit waitedst, and Quaker forms like [thee] waits) are not regarded in this paper as part of present-day standard American English.

⁹ Hyphens in the phonemic transcription throughout this paper indicate morphological boundaries, not phonemic junctures.

¹⁰ On morpheme alternants and morpheme units see Harris, Lang. 18.170–3; and cf. Hockett, Lang. 23.341–2.

¹¹ A syllabic is a vowel alone or a vowel plus /y, w, h, r, hr/; ct. fn. 5.

¹² Strictly considered, every verbal base has at least two alternants differing in stress see Wells, Lang. 23 108–14 = §§66–78. Thus, wait has the following alternants: (1) /weyt/, without inherent stress, when the base is accompanied in an utterance by the emphatic stress morpheme /´/, e g. Wait a minute /wéyt ə mînit/; (2) /wêyt/, with ‘reduced-loud’ stress, when the base is not accompanied by the emphatic stress morpheme, e.g. Let's wait here /lèt s wêyt hír/; (3) /wèyt/, with ‘medial’ stress, again when the base is not accompanied by the emphatic stress morpheme, e.g. Let's not wait here /lèt s nât wèyt hír/. (The first of these alternants would appear also if the base were pronounced without any phonemic stress; but this would scarcely happen with a word like wait.) In our discussion, we shall disregard all stress differences among morpheme alternants.

¹³ The argument: Phonemic ally different forms that occur in the same environment, and are not in completely free variation with each other, are morphemically different (§2.3). In the two participles shown /šow-n/ and showed /šow-d/, the phonemically different elements /n/ and /d/ both follow the base /šow/. They are not in free variation, since the two forms have different connotations of elegance and hence are not interchangeable. Therefore, either /n/ and /d/ are different morphemes, or the environment in which they occur is after all not the same for both. Since we wish to identify /n/ and /d/ as alternants of the same morpheme (suffix 3), we assume that the /šow/ that precedes /n/ is morphemically different from the /šow/ that precedes /d/. In other words, we choose to set up two different but homonymous morphemes /šow/, and to refer to them—rather than to the suffix alternants—the stylistic or connotative difference between the inflected forms shown and showed.

¹⁴ The form n't is best regarded as a separate morpheme, not as an alternant of the full form not. The two forms contrast, at least stylistically and in their connotations, in such phrases as I cannot go: I can't go.

¹⁵ Any form in this list not otherwise identified is both the uninflected form of the verb and the homonymous 3d-singular form with zero suffix.

¹⁶ It goes without saying that historical considerations play no part in a structural description. The actual historical relation between sing and song is irrelevant here; all that is relevant is their morphological relation in the structure of present-day English.

¹⁷ The optative use of be is paralleled by other verbs in a few formulas: God have mercy, God forbid; Perish the thought. The concessive use of other verbs than be is limited to such archaic locutions as Try they never so hard.

¹⁸ The alternant /wΛz/ occurs only with a subject in the 1st or 3rd person singular; the alternant /wahr/ occurs with all other subjects.

¹⁹ Or /biy/ in British English, identical with the alternant that appears in the uninflected form.

²⁰ Since be has more inflected forms than any other verb, we might have begun our discussion with it instead of saving it for the end. In that case we should have said that all other verbs (except the auxiliaries) have the same inflectional categories as be, but that only be formally distinguishes the general present from the uninflected form or the unreal from the preterit. This is essentially what Hockett did in his rigorously systematic treatment of English verbs (op.cit. in fn. 2). The treatment here adopted seems preferable because it results in a simpler statement.

²¹ One verb not listed in this section requires a word of comment. Beware is used as an infinitive and as an imperative (I told him to beware; Beware of the dog); inflected forms (bewares, bewared, bewaring) are either not used at all in present-day English or at best extremely rare. We may tentatively assign beware to conjugation type A1 as a regular verb whose inflected forms happen to be not in use.

²² The following symbols are used to characterize the several types of alternation: V = syllabic (not merely vowel), C = non-syllabic consonant, 0 = zero (i.e. absence of a phoneme), V₁ and V₂ = different syllabics. The symbol ~ everywhere means ‘in alternation with’.

²³ We have already agreed to disregard the difference between such alternants as /hæ̂v, hǽv, hæv/—a difference in stress alone (fn. 12). But even if we discriminate such alternants, the situation will not be affected; for the completely unstressed alternant /hæv/ is common and perfectly natural in the cited context.

²⁴ /kəŋ/ before /k/ or /g/ (I can come; I can go), /kən/ elsewhere.

²⁵ Some speakers use also a preterit form /št/, as in / should think so /ày št θíŋk sôw/. For such dialects we must set up another verb shall, with morpheme alternants /šəl/ uninfected and before suffix 1, /š/ before suffix 2.

²⁶ For pedagogical purposes, this description of atonic verbs—and of tonic verbs too, for that matter—would certainly be over-meticulous and ineffective. To the student learning to speak English, busy at his primary task of memorizing model sentences, it would be unhelpful at best, if not actively confusing. But the intent of this paper is not pedagogical.