¹ Grundriss² 2.3.§27(1916).

² Ibid. 48.

³ Indogermanische Grammatik 4.§127 (1928).

⁴ By ‘Indo-European proper’ we mean all that is Indo-European in the broadest sense, except Hittite and its closest congeners (‘Anatolic‘). While we concede that Tocharian and Armenian may have phonological archaisms not paralleled elsewhere in IE proper, their verb-systems are obviously of the type found in IE proper; cf. the appropriate sections of the Table, §4. For the Armenian cf. further Lang. 18.226-8 (1942).

⁵ For the purposes of this paper we mean by a ‘characterized stem’ one that contains at least one formant that can be isolated by comparison with an etymologically related control-stem. We do not consider all thematic stems as ipso facto characterized, and where there is reason to believe that a formant once confined to the present stem has been analogically extended to all the other stems of a verb-paradigm, we treat such a present-stem as simple, in the absence of other characterization. Hirt, Idg. Gr. 4.§97, is misleading when he says of the so-called ‘present-formante’; ‘im Germanischen sind sie bereits so gut wie verschwunden’; they have not disappeared, but have been analogically extended to the non-present stems.

⁶ Of course there are other types of suppletion found within what is normally a single stem-system, as New Lith. sg. 1 esù ‘am’ : sg. and pL. 3 yrà, but it is only the stem-system suppletives with which we are concerned.

⁷ All complete weak conjugations belong to the separate history of the continuant languages, even when their inflectional procedures involve generalization of IE structural types formerly of restricted incidence. For denominatives and new formations in general nothing further need be said, but when a verb formerly strong becomes weak, any characterization of its present stem is analogically extended to its other stems, and differences of stem-internal ablaut (unless they themselves have become inflectional procedures, as often in Germanic) are levelled out.

⁸ E.g. we do not concern ourselves with the ultimate relegation of earlier ‘imperfects’ to aorist (etc.) status by the emergence of neologistic imperfects, as Arm. ‘aor.’ sg. 3 eber : neo-ipf. berēr, OCS ‘aor.’ sg. 3 krade ‘stole’ : neo-ipf. kraděaše.

⁹ It is of course only in a very restricted sense that we can speak of ‘weak verbs’ in Sanskrit, but such re-formations as Classical Skt. fut. sg. 3 (replacing earlier satsyati) : pres. s↓ti ‘sits’, or the isolated fut. pl. 1 ‘shall obtain’ (Whitney, Skt. Gr. §936c), prepare us for the thoroughgoing recasting of Sanskrit strong verbs as weak in Middle Indic; cf. especially the recent papers of Franklin Edgerton on Buddhist Hybrid Sanskrit.

¹⁰ Old Latin shows traces of a third finite stem, a ‘Subjunctive Present’ (probably of aoristic origin; cf. Old Irish and Tocharian) : sg. 2 attigās, sg. 3 advenat, etc.

¹¹ It is probably only because of the paucity of attested P-Italic forms that more of the seven types of the Table cannot be here represented; e.g. many Oscan verbs with present stems of Conjugation I may have been weak, forming a neo-perfect with suffixal -tt-, but we are unable to cite corresponding presents to Osc. pf. sg. 3 prúfatted, etc. Cf. Buck, A Grammar of Oscan and Umbrian² §128 (1928).

¹² The vestigial status of the strong verb in Middle Welsh is shown by the fact that occurrences of ‘aorists’ other than s- aorists are largely confined to poetry and to the sg. 3; cf. Strachan, An Intrudiction to Early Welsh §133 (1908).

¹³ A parenthesis enclosing a type-number in the Table calls attention to the existence of a ‘special situation’ (§4); in this instance Goth, leihan, though ‘strong’ in the sense of Germanic grammar, could be considered ‘weak’ in our sense.

¹⁴ We assume that OCS sta, zna, bbra continue IE *stā-t, *ĝnō-t, *bhᕻrā-t (the first two from simple stems, the last characterized by suffixal -ā-; cf. certain Latin and Old Irish subjunctives), not s- aorists of type *stā-s-t, etc. Cf. van Wijk, Geschichte der altkirchenslavischen Spr. l.§58 (1931): ‘es ist sehr gut möglich, dass alle Formen der 2. 3. Person regelrecht auf indogermanische Formen ohne Tempusbildendes s zurückgehen.‘

¹⁵ The learner of Hittite, whatever his other difficulties, does not have to cope with ‘principal parts’.

¹⁶ The essential tenet of the Indo-Hittite hypothesis is that Hittite (or ‘Anatolic') became disengaged from the general Indo-European complex much earlier than the other dialects; this is denied by a number of scholars. A proposed canon has been the establishment, if possible, of common innovations within ‘Indo-European proper’ as against Hittite.

¹⁷ The Hittite mi- conjugation of course includes both thematic and athematic stems.

¹⁸ E.g. J. H. Moulton, Grammar of New Testament Greek 1³.108-9 (1908).

¹⁹ Sturtevant, HG §291.

²⁰ Hirt, op.cit. 4.§97: ‘Anderseits wird das Präsens verschiedentlich durch besondere Präsens bildende Elemente verstärkt, deren Herkunft vorläufig noch ziemlich dunkel ist. ... Wenn also die Präsensverstärkungen, wie man annehmen darf, ursprünglich eine bestimmte Bedeutung gehabt haben, so ist diese doch frühzeitig verblasst und in den Einzelsprachen z.T. ganz verloren gegangen. Über die Entstehung der Präsensverstärkungen kann man nur Vermutungen aufstellen.’ We object in principle to the terms ‘Präsens bildende Elemente', ‘Präsensverstärkungen’, and the like; widespread as this usage is, it involves naming these morphemes from a later and merely apparent function, in complete disregard of their original one. While many of the languages of IE proper have an aspectual system the beginnings of which can fairly be traced from late IE times, correlations between the aspects involved and particular characterizations remain doubtful; where they appear to exist, they may well rest on secondary association. The chief reason for postulating an Indo-Hittite ‘aspectual’ conjugation is the existence of the characterizations themselves as specific morphemes; it is conservative, therefore, to distinguish between an IH ‘paleoaspectual’ system and a late IE ‘neo-aspectual’ one—without, of course, too categorical a denial of any thread of continuity connecting the two.

²¹ J. H. Moulton, l.c.; Brugmann, op.cit. §38, esp. 70-71.

²² Meillet, Introduction à l'étude comparative des langues indo-européennes⁷ 250 (1934): ‘Quoi qu'il en soit, l'opposition du présent et de l'aoriste est sans doute celle des particularités des verbes indo-européens qui a eu dans certaines langues pour le développement ultérieur de la flexion verbale les conséquences les plus importantes. Néanmoins, ce serait une erreur d'attribuer à l'opposition “présent : aoriste” en indo-européen l'importance qu'elle a en grec. Le hittite n'en offre pas trace.’ If by ‘indo-européen’ Meillet means our ‘IE proper’, then we cannot follow him in the latter part of this statement: structurally, the contrast between the ‘present’ and ‘aorist’ (or some other non-present category partly basing on earlier aorists) is quite as unmistakable in all the other languages of IE proper as in Greek; the absence of an aorist category in Hittite is significant, to be sure, but not in the sense in which Meillet would be appealing to it.

²³ Or ‘indicative’ (with primary ending) and ‘injunctive’ (with secondary ending); but it must be remembered that when the IE imperfect and aorist finally emerge, they are accounted ‘indicatives’, despite their secondary endings.

²⁴ Perhaps the most pointedly colorless way of stating a general truth was to use the colorless stem with a secondary ending, not in its preterital sense, but in its older timeless sense; this may survive in the ‘gnomic’ aorist of Greek and Slavic (otherwise explained by Brugmann, Grd.² 2.3.§670 (1916).

²⁵ Even if -s- formed inceptives in Indo-Hittite (Sturtevant, HG §315), a semantic parallel to its later functional fading in IE proper might be the use of ME gan as a mere auxiliary to form with the infinitive a periphrastic preterite. Cf. OED s.v. †gin, v.l: ‘In Middle English poetry the pa. t. gan was commonly used in a weakened sense as a mere auxiliary (— the modern did).‘

²⁶ The examples of Type 3 in the other languages involve a non-present stem which either bases on an IE reduplicated perfect or else is a neologism.

²⁷ Cf. Whitney, Skt. Gramm. §§856, 1046; Schwyzer, Griechische Grammatik 1.748.β.1 (1939).

²⁸ So, substantially, Sturtevant already in 1932: The s-aorist in Hittite, Lang. 8.119 ff.: ‘Although Primitive IE must have had an aorist of much the same morphological character that we see in Skt. and in Gk., the category is made up of such dissimilar elements that one is inclined to assume relatively recent origin for it. Consequently the lack of a morphologically composite but syntactically unified aorist in Hittite is not surprising, if Hittite is related to IE only by virtue of common descent from the earlier parent speech, which I have called Indo-Hittite.‘ In our opinion this constitutes an even more significant fulfillment of the canon referred to in fn. 16 above than his subsequent explanation of the demonstrative *so, *sā, Hod in IE (Lang. 15.11-9 [1939]) ; cf. however Goetze, JAOS 65.53 (1945).

²⁹ To be sure, the Hittite instances given are of variations between normal grade and o-grade, but Hittite orthography permits no sure inference as to whether variation between e.g. normal grade eu and zero grade u still existed in Hittite: Sturtevant, HG §102.

³⁰ The IE perfect, as a category, perhaps rests on a relatively late analogical imitation by many other verbs of a handful of inherited forms requiring the ‘alternative endings’ of §7— most of the models probably reduplicated; obviously, its emergence, unlike that of the aorist, involved wholesale creation of new stems. We are not concerned here with the still later morphological fusion of IE perfects and aorist s in Italic, Celtic, Germanic, etc.

³¹ Above all the Optative and the Subjunctive.