¹ The forms used here as illustrations are taken from the standard handbooks, particularly T. Maretié, Gramatika hrvatskoga ili srpskoga književnog jezika² (Zagreb, 1931), and dictionaries, especially Pravopis srpskohrvatskoga književnog jezika (Zagreb and Novi Sad, 1960); but they have been checked (except as noted below) against the speech of my wife, a native speaker of the Belgrade-based koine—one of the two standard varieties of Serbo-Croatian. Of these, the so-called ekavski variety is spoken in the eastern part of the Serbo-Croatian speech area, the ijekavski is spoken in the western part of the area. Some forms, in isolated and individual cases, have been checked also with other informants: members of my wife's family in Belgrade and other standard speakers encountered at various times and places. The ijekavski forms given here have not been so checked; but as I have had extensive contact with ijekavski speakers in Yugoslavia, I have reason to believe the handbook forms to be correct.

² The phonemic analysis underlying this study is that of Carleton T. Hodge, originally published in Lg. 22.112–20 (1946), amended and amplified in his ‘Serbo-Croatian stress and pitch’, General linguistics 3.43–54 (1958). The segmental phonemes are these: vowels /i e a o u , stops /p t k b d g/, fricatives /f s š v z ž h/ (the last [x]), affricates /c č ć , nasals /m n/, lateral /l/, trill /r/, high-front glide /j/.

The symbols I use in my transcription are those of the traditional Latin orthography (latinica), except that I write (for orthographic đ and dz̆). I analyse [1‘] and [n‘] as phoneme sequences, /lj/ and /nj/.

I depart from latinica in three respects. With Hodge, I write long vowels with doubled vowel symbols. Again with Hodge, I indicate the traditional falling stress with a single acute, the traditional rising stress (which extends over two syllables) with two acutes on the two vowels affected. Thus,

Finally, once more with Hodge (cf. his article in General linguistics, fn. 1), I write the traditional ‘vocalic r' as a sequence . The vowel occurs not only before r, but also in the Serbian names of the consonant letters , in acronyms for orthographic BSK ‘Belgrade Sport Club’), and in some recent loans before l and n .

³ Rarely too the stem may be zero. This occurs in prefixed verbs with the root where in the infinitive the is eliminated through the operation of automatic morphophonemic changes before the formant -ći. Thus, proóćí ‘to pass through’ is proó (prefix) + θ (root) + ći (infinitive formant),

⁴ Perfective verbs usually lack the present gerund and the verbal noun, imperfective verbs often lack the past gerund, and intransitive verbs lack the past participle. Intransitive verbs are those that do not occur with a direct (i.e. accusative) object; perfective vs. imperfective in Serbo-Croatian must be defined on the basis of syntactic criteria.

⁵ Cf. Pavle Ivić, ‘The study of standard Macedonian’, Word 9.355 (1953).

⁶ All automatic changes and most of the optional ones obtain throughout the language, not only in verb forms.

⁷ In terms of Serbo-Croatian phonemics, lengthening means addition of an identical phoneme; shortening means reduction of a sequence of two identical phonemes to one phoneme.

⁸ Items or alternations described and not exemplified in the same paragraph are illustrated in the sample paradigms below.

⁹ All changes described as affecting a given verb are, unless the contrary is stated, understood to apply to all other verbs with the same stem but with the addition of prefixes.

¹⁰ As in my study on the Polish verb, General linguistics 4.77–142 (1960), the grouping of the ending morphs into morphemes is performed on the basis of syntactic environment. Allomorphs are assigned to particular morphemes according to the possibility of their occurrence in the same construction with the nominative of the various personal pronouns. On this basis we may justify inclusion of identical morphs in two different morphemes. For a discussion of a similar problem see Z. S. Harris, Methods in structural linguistics 215, appendix to §13.43 (Chicago, 1950).

¹¹ As this form occurs primarily in writing, it is given in its written shape. So far as it is spoken at all, v would be replaced by f when š follows, except in extreme spelling pronunciations.

¹² However, some e-verbs with base-final vowel have alternant past part, forms with thematic e. Thus, bíti ‘to beat’ has both bíit and bíjén (as well as bívén with base suffix -v-).

¹³ Stems with bases ending in a vowel show some optional variation in the past participle and the verbal noun. Normally they have no thematic vowel, but there are optional variants with thematic e or, in some cases, with base suffix -v- plus thematic e. Thus the three variants cited for bíti in footnote 12. Because of this option, there may be a difference in stem formation (and consequently in formant) between past participle and verbal noun. Thus, from ‘to rub’, we have ‘rubbed’ but treénjé ‘friction’.

¹⁴ Most i/a-verbs have bases ending in palatal consonants (š, ž, č, j); i/ě-verbs have bases ending in nonpalatal consonants. This near complementation is however disturbed by i/a-verbs with nonpalatal C, as záspáti ‘to fall asleep’.

¹⁵ In colloquial usage final ao of the l-form is replaced by o.

¹⁶ The radical may be composed of a single root morpheme (which will frequently recur as the root of other verbs or of words of other grammatical categories, e.g. vód-i-ti ‘to lead’, vóod-θ ‘platoon’, pre-vód-í-ti ‘to translate’, preé-vód-θ ‘translation’) or of a root morpheme plus a root suffix or suffixes (which will also usually recur as the base of words of other categories, e.g. voć-aár-í-ti ‘to work as a fruitgrower’, vóć-áar-θ ‘fruitgrower’, vóć-e ‘fruit’). The suffixes -iir- and -is-, though they occur mainly in verb stems, are counted as root suffixes rather than base suffixes, since unlike the latter they do not enter into the system of verb aspect derivation, i.e. they do not serve to form perfective or imperfective derivatives from root verbs, as do -v- and -n/nu-. The suffixes -iir- and -is- occur mainly in recent foreign borrowings, such as organiziíráti ‘to organize’, manévrísati ‘to maneuver’, -is- has an allomorph -iz- which occurs before the root suffix -iir- and the presuffixal vowel and base-suffix -o/u + v-.

¹⁷ Archaic forms (known to persons versed in literary writing, but not occurring in colloquial usage) are cited with a superscript ^x.

¹⁸ This fact forces us either to posit as a morpheme in contrast with or to adopt some dubious fiction such as a morphologically determined juncture or zero consonant in na + .

¹⁹ Serbo-Croatian has two equally acceptable variants of the standard language, distinguished principally by divergent reflexes of the Common Slavic phoneme ě. In the ekavski variant, Common Slavic e and ě coalesce in the reflex e (or ee); in the ijekavski Common Slavic ě has a variety of reflexes (short je, e, i; long ije, jee, ijee).

²⁰ The i/ě-verbs zréti ‘to ripen’ and vréti ‘to boil’ have variant masc. sg. l-forms: regularly zrío, vrío; irregularly, but identical with the ekavski forms, zréo, vréo. Many speakers, both ekavski and ijekavski, have irregular (and for the most part nonstandard) variants of these verbs, which are treated as e-verbs with various root alternations.