БАЛТУ КАЛБУ ВЕИКСМАЖОДЖИО ТЫРИНЕЙИМАИ ЛИЕТУВОС ТСР МОКСЛУ АКАДЕМИЯ

V. N. TOPOROV

A FEW REMARKS ON BALTIC VERBS IN -sta IN CONNECTION WITH THE ORIGIN OF THIS FORMANT (AN INDO-EUROPEAN PERSPECTIVE)

In recent years, the attention of many researchers has been directed toward Baltic present stems with the element -st(a). These formations have been analyzed both synchronically¹ and historically², resulting in a large body of material that has made it possible to draw substantial generalizations about the character of this type of present stem. It is now appropriate to return to the question of the origin of the formant -st(a) itself, especially since there is not only no consensus on this matter, but a whole range of competing views, often diametrically opposed.

The difficulty in determining the most reliable solution probably stems from the fact that, on the basis of Baltic data, it is difficult or even quite impossible to provide an internal reconstruction of that part of the verbal system in which the formant -st(a) took shape. When one turns for help to the Indo-European state, one often finds a picture that, paradoxically, is chronologically later than the Baltic state. Hence the numerous misunderstandings, distortions of the relative chronological perspective, and evident desire to prove not only the secondary nature but also the generally late (recent) origin of this type of present-stem formation. In this respect it is useful to distin-

J. Otrębski, Gramatyka języka litewskiego, II, Nauka o budowie wyrazów, Warszawa, 1965, pp. 320—326; A. Senn, Handbuch der litauischen Sprache, Heidelberg, 1966, pp. 266—268; Lietuvių kalbos gramatika, II, Morfologija, Vilnius, 1971, pp. 226—228; Mūsdienu latviešu literārās valodas gramatika, I, Fonētika un morfoloģija, Rīga, 1959, pp. 673—675, and others.¹

² См. J. Kazlauskas, Lietuvių kalbos istorinė gramatika, Vilnius, 1968, стр. 316—336; J. Endzelīns, Latviešu valodas gramatika, Rīga, 1951, стр. 751—763, — в том, что касается отдельных балтийских языков, и также: Chr. S. Stang, Das slavische und baltische Verbum, Oslo, 1942, стр. 131—138; его же, Vergleichende Grammatik der Baltischen Sprachen, Oslo—Bergen—Tromsö, 1966, стр. 338—349; J. Endzelīns, Baltu valodu skaņas un formas, Rīga, 1948, стр. 190—192; E. Fraenkel, Die baltischen Sprachen, Ihre Beziehungen zueinander und zu den indogermanischen Schwesteridiomen als Einführung in die baltische Sprachwissenschaft, Heidelberg, 1950, стр. 90 и след., — в том, что касается сравнительно-исторического анализа этих образований. Ср. также обращение к глаголам на -st(a) в работах по индоевропеистике — прежде всего: J. Kuryłowicz, The Inflectional Categories of the Indo-European, Heidelberg, 1964, стр. 50—51; В. В. Иванов, Общеиндоевропейская, праславянская и анатолийская языковые системы (сравнительно-типологические очерки), М., 1965, стр. 154 и след.

151

tinguish the older view of K. F. Johansson, who was inclined to see in the Baltic formant -sta a reflex of the old 3rd-person mediopassive sigmatic aorist suffix -s-tо³ from all the other views, whose essence is the assertion that this type⁴ arose secondarily and late (*-st- < *-sk- or *-st < *-t-t-, and the like).

Before turning to the question of the possible sources of Baltic -st(a) in present stems, it is useful to answer two questions: what is meant when the secondary and late origin of this type is discussed, and what place this type of stem occupies in the system of Baltic verbal derivation and inflection.

The secondary nature of the formant -st(a) in Baltic can be discussed only in the sense that it apparently consists of elements that can also occur separately, retaining one or another grammatical or derivational meaning—that is, in an unbound form. Compare verbs in -soti, such as murksōti, vēpsōti, snopsōti, etc., where -s-, by virtue of its nominal origin (cf. murksà and mùrksa, vēpsà, snòpsa, etc.) and intransitivity (denoting a state), is comparable, on the one hand, with -s- in ilgsta or sénsta (cf. the original nominal types ìlgas and sēnas, and the intransitive meaning of the corresponding verbs), and, on the other, with verbal formations containing the element -t- (cf. predicative types such as kárvē pašērta or pēdos pālikta, or verbs denoting an instantaneous, single action in -telēti: žvilgtelēti⁵, čiūptelēti, báubtelēti, etc.). The existence of pairs consisting of a root + telēti and a root + stelēti (i.e. -t-: -st-), given that in a considerable number of cases these formatives occur with the same root (cf. dùrtelēti and dùrstelēti, etc.), may likewise be regarded as an argument for segmenting -st- into constituent elements.

Finally, the secondary and late origin of the type in -st(a) is often discussed with reference to the expansion of this type among verbs that previously conveyed the same meaning by other means⁶. Such a conception of secondary origin is unconstructive, since it takes into account only this particular set of facts, and not all possible instances of use of the formant under consideration. In other words, this approach refutes only the fact of the presence—

³ См. К. F. Johansson, Eine analoge Neubildung der Verbalflexion im Altindischen und Baltisch-Slavischen, — KZ, Bd. 32, 1893, стр. 476; ср. также H. Hirt, Indogermanische Grammatik, Bd. IV, Heidelberg, 1928, стр. 236.

⁴ Обзор этих точек зрения см. в кн.: Chr. S. Stang, Das slavische und baltische Verbum, стр. 134 и след.; J. Kazlauskas, Op. cit., стр. 330 и след.

⁵ Compare žvilgtelēti in Daukša’s “Postil” (335, 44) and similar formations in dialects.

⁶ Cf. J. Kazlauskas, Op. cit., pp. 324–325: “That the formant -sta- in these verbs is later is shown not only by the double formation of their present-tense forms (infix and -sta-), but also by the fact that these verbs often have the formant -sta, in some dialects, whereas in others they are infix-only; cf. glęžta and glęžta; gęsta and gęnsa, gęsa;... glūžta and glūna, etc. There is no doubt that the infixed forms are older; the -sta formatives are discussed here later, especially when considering the cases. Cf. also Chr. S. Stang, Vergl. Gramm., p. 342: “The fact that it (scil. the suffix -sta-) occurs as a supplement to the nasal infix, i.e. that its distribution is governed predominantly by the presence or absence of the nasal infix, indicates that the productivity of this suffix does not go particularly far back. The fact that s after r and k is not represented by š points in the same direction: we are therefore dealing with an expansion of the suffix in the form -st-…”

152

…tion of -st(a) in a certain set of verbs, but the very fact that -st(a) is used is not refuted. Moreover, when it is said that -st(a) appears only when the nasal infix is lost or—especially importantly—in verbs whose root sound structure made a nasal infix impossible, the question is ignored of how the meaning of the verbal stem in question was expressed, given that it could not be expressed by an infix. Finally, when considering the question of the secondary and late origin of a particular formant in unique archaic languages such as the Baltic languages, it should be remembered that many processes which had already ceased to operate by the time of the earliest records of the Indo-European languages continued to operate in the Baltic languages and affected new strata of evidence⁷. Just as it would be unreasonable to consider the -i stems in Lithuanian secondary, since this means of formation remained productive for a long time and continued to affect a whole series of new nouns, so too we have no grounds for considering -st(a) secondary and to have originated only at a relatively late date. The difference in the interpretation of these two formatives should, I believe, be explained by the fact that the stem-forming -i is well known in many different Indo-European languages, whereas -st-(a) is regarded as an exclusively Baltic formation, from which the unwarranted conclusion is drawn: Baltic, therefore post-Indo-European. But more on this below. In any case, the view of its secondary status adopted here should in no way prevent the search for Indo-European correspondences to this formant. In doing so (and this too is very important), correspondences should be understood to include not only cases in which the formant is used isofunctionally (in relation to the Baltic evidence), but also cases in which this formant (or even its constituent parts) occurs with other meanings and even changes status (root extension, morphological or word-formative element, part of a compound, etc.). All this, of course, holds provided that one can construct an arbitrarily long chain of reconstructions that ultimately makes it possible to connect the formants being compared in different Indo-European languages.

The second question, on whose resolution much depends in the problem of the origin of -st(a), concerns determining the place occupied by this formant in the system of Baltic verb types. In this connection, it will be necessary to recall several more or less familiar facts, giving some of them a more specific emphasis or placing them in a somewhat different context.

One achievement of Indo-European studies in recent decades should be considered the recognition of the fact that the appearance of this set of stem-forming formatives in the present with a particular root is not a matter of chance—

⁷ In particular, this naturally explains the numerous cases in which stems in -st- coexist with other stems in a number of dialects (cf. in the Upper Latvian dialects: l’ēpiri and lēpstu from l’īpt’, tÿuki and tÿukstu from mukt, klÿuri and klÿupstui from klupt’, etc.), or the presence in a number of dialects of stems in -st- in verbs whose present is formed by other means in other dialects or in the standard language (cf. Latvian/dialectal izsmōkstu, smokstu, apsūpst, etc. (Kalupē). See M. Rudzīte, Latviešu dialektoloģija, Rīga, 1964, p. 360; cf. also in part K. Draviņš, V. Rūķe, Verbalformen und undeklinierbare Redeteile der Mundart von Stenden, Lund, 1958, pp. 14–15.

153

…be accidental; that this phenomenon (the coexistence of stems) is governed by sufficiently strict (in principle) regularities of a systemic nature; and that, consequently, the presence of these formatives allows us to infer, with greater or lesser probability, the presence of other formatives systemically connected with them. The fact that a fairly considerable number of verbal roots have been identified in various Indo-European languages which, by means of a range of formatives (-n-, -s-, -sk-, -io-, etc.), form series of similar present stems should be interpreted as practical proof, first, of the reality of connections among different types of present stems and, second, of the possibility of dating these regularities to a sufficiently early period, when connections among different parts of the Indo-European linguistic area were still operative.

In this connection, it should be emphatically stressed that the relationship between present stems in -st(a) and present stems with a nasal infix in the Baltic languages⁸ is exceptional. Baltic present stems in -st(a)- and those with a nasal infix (-n-, -m-), which go back to very ancient formations, developed extensively in the history of the Baltic languages⁹. Although they acquired a number of new features (in particular, semantic ones), they nevertheless retain certain highly archaic traits that make it possible (at least in general terms) to reconstruct some relations of the most ancient type, for example, those belonging to the period before temporal oppositions had formed and become established. These two types of present formation are in complementary distribution, from which follow their shared features, allowing the two types to be united at a somewhat higher level (that of grammatical semantics), as well as their differences at the level of formal characteristics.

Among the features common to these two types, their intransitive-inchoative meaning should be noted first and foremost. At the same time, there are no grounds for considering these two means of forming present stems (-st(a)- and the nasal infix) the original or exclusive means of expressing this meaning, since outside the present-tense system it is conveyed by other means. In this connection, it is also appropriate to note another feature common to the two types of present-stem formation mentioned: both form the preterite with the element -ā, usually associated specifically with intransitivity (as opposed to—

⁸ The Lithuanian evidence is especially instructive, particularly that of dialects in which the nasal infix was not lost, unlike in Latvian, for example. As for the Old Prussian evidence, it is too scant to support independent conclusions (cf. wiršt, polinka, sindats / syndens, cf. poprestemmai / issprestun).

⁹ It should be noted that the exceptional productivity of the present type with a nasal infix in the Baltic languages forms a kind of pendant to the similarly widespread stems in -sta. As in the latter case, the nasal infix occurs with a considerable number of Lithuanian verbal roots, even though the corresponding roots in other Indo-European languages (including those, such as Ancient Indian, Hittite, and Tocharian, that have well preserved this type of present stem) show no infixal formation of present stems. The expansion of stems with a nasal infix in Lithuanian is all the more surprising given how rare it is in Slavic (sed̑a, leģa, b̑eḑe, cf. Gothic rēsti with a transitive meaning), while in Germanic it is represented by only one example (cf. Gothic standan-stop).

154

... vowel from -ē, which is associated with transitivity)¹⁰; exceptions are minimal and can be explained¹¹. The third feature worth mentioning here is the general (overall) set of phonemes that conclude the root to which -st(a) is added or into which the nasal infix is incorporated. Here, however, divergences also emerge, which should be attributed to phonetic phenomena. Compare, on the one hand, at least ambiguous cases such as gel-sta (but: gelto), nīrsta (but: nīrto), výsta (but: výto), výsta (but: výdo), žýsta (but: žýdo), skýsta (but: skýdo), and the like, and, on the other hand, cases such as džīuva, lýja, drýska, tīška, dīna, and the like, where a nasal infix is assumed¹².

For the purposes of reconstruction, however, more important are the differences between stems in -st(a) and those with a nasal infix. Among these differences, it is important to note the presence of accentual oppositions in the root of -st(a) stems; compare ė: ė̃ (drėksta, plėksta, but: šėlsta); ó: õ (nóksta, slópsta, sprógsta, but: blõgsta, lóbsta); ū: ũ (džīūsta, rīgsta, slūgsta, trāksta, but: dūksta, rūksta); ý: ỹ (dýgsta, klýsta, but: dýksta, klýpsta, krýpsta, nýksta, pýksta, výksta); il: il (ilgsta, tvilksta, but: silpsta, smilksta); ir: ir (tvirksta, but: mirksta, sirpsta, tirpsta); in: in (brinksta, stingsta, tiñsta, tvinsta, pažįsta: pažinti, but: dingsta, linksta, sprīngsta, tvinksta); im: im (kimsta, rimsta, but: gimsta, klim̄psta); ul: ul (smūlksta, but: stulbsta); al: al (ālksta, but: alpsta). In present stems formed by a nasal infix, accentual differences are absent, and the circumflex occurs throughout: aīka, teīka, krīnta, jūnka; taīra, leīra, līt́ra, kluītra; bāla, skāra, glēra, kȳla, bȳra, skȳla, svȳra, gūra; drȳska, mȳzga, īȳska, etc.

This relationship of freedom versus constraint of accent in the present, found in -st(a) stems and in stems with a nasal infix and distinguishing them from each other, takes a different form in other parts of the verbal system. Stems in -st(a) are characterized by the same accentuation for a given verb in the present, preterite, and infinitive (cf. ālksta, ālko, ālkti; alpsta, alpo, alpti; kimsta, kimo, kimti, etc.)¹³. Stems with a new infix, as if compensating for the lack of a choice of accentual patterns in the present, correspond to preterite and infinitive stems whose accentuation is generally different (in at least one of these stems) from that of the present. Compare present—preterite, but infinitive: raīnda, rādo, but rāsti; senka,

¹⁰ See Chr. S. Stang, Das slavische und baltische Verbum, p. 98, 189; P. Arumaa, Von der Eigenart des Ablauts und der Diathese im Baltischen, — ZfsIPh, vol. 26, 1957, pp. 118—149.

¹¹ Cf.: “sta-stem verbs in the past tense always also have the stem-final -ā, with the exceptions of gimsta, gimė and miršta, mirė. However, the verb gimsta acquired the stem -sta recently, cf. gėma: gimė, while the ē-stem past-tense form mirė apparently arose when the ē-stem past tense ceased to be an indicator of transitivity.” See J. Kazlauskas, Op. cit., p. 319. It is interesting that miršta is also an exception with respect to the change of s to š after r; see Chr. S. Stang, Vergl. Gramm., pp. 342—343. An entirely different solution is also possible (-š- in miršta from -sk’-); see J. Endzelin, Über den slavisch-baltischen Reflex von idg. sk’, — ZfsIPh, vol. 16, 1939, p. 114; J. Kazlauskas, Op. cit., p. 330.

¹² Cf. also: bāla, būra, glīma, kīrža, bīzga, etc.

¹³ Exceptions are rare and explicable; cf. sēnsta, sēno, sērti; tēnsta, tēmo, tēnti. The same pattern is found in Latvian stems in -st-, cf. tirpstu, tirpu, tirpt; klistu, klidu, klīst; tūlstu, tūlzu, tūlzt; rīgstu, rīgu, rīgt, etc.

155

sėko, but sėkti; šlam̃pa, šlāpo, but šlāpti; stẽmba, stẽbo but stẽbti; sãla, sãlo, but sálti; gvẽra, gvẽro, but gvẽrti, etc.; present, but preterite—infinitive (with root -in-, -un-, -im-, -um-, -y-, -ū-): kiñta, but kito, kisti; buñka, but būko, būkti; kiñba, but kibo, kibti; grūmba, but grūbo, grūbti; dỹla, but dilo, dilti; žiūra, but žiūro, žiūrti; rỹzga, but rizgo, rizgti, etc.

Another difference, related in one way or another to those already noted, is that the range of root vowels in present stems in -st(a) is broader (virtually unrestricted) than in stems with a nasal infix, where -i- and -u- clearly predominate (drīñka, kiñta, kriñka, kriñta, miñga, miñta, driñba, kiñba, liñpa, stiñpa, šim̃pa; juñta, kruñta, pluñka, puñta, siūñta, buñda, buñka, guñda, šūñta, truñka, čiūñpa, duñba, grūmba, kluñpa, skuñba, etc.)¹⁴; -a- and -e- occur much more rarely and perhaps secondarily (añka, kañka, rañda, šlam̃pa, tam̃pa, bãla, sãla, skãra, šãla; geñda, señka, smeñga, teñka, gleñba, kneñba, leñpa, skrem̃ba, šeñpa, glẽra, gvẽra¹⁵.

These differences in the structure of the vocalic center and accentual paradigm between verbs forming the present in -st(a)- and verbs forming the present with a nasal infix are significant above all because they attest to the ancient isofunctionality of these types and to their formal opposition (during a certain period). Only within these two classes, taken together, are the possibilities characteristic of the whole realized (all kinds of root vocalism, all types of accentual patterns—both within the present and across the entire set of verbal stems). Taking these features into account, one may hypothesize which types are dominant, and in what respect (that is, which facilitate the greatest differentiation of possibilities)¹⁶. But here and now it is more important to emphasize that both the common elements and the differences between the -st(a) type and the nasal-infix type provide irrefutable evidence of the closest connection—both current and historical—between these two types of present stem, and of their extensive correlation and balance, which in many cases approaches complementary distribution and, above all, entirely rules out the possibility of assuming that this correspondence is accidental and very recent. The existence of different ways of forming present stems across dialects (cf. glẽta, an -st(a) stem: -st(a): gleñza, an infix stem; gẽsta: geñsa, kiūžta: kiuñza, gniūžta: gniuñža, dūžta:

¹⁴ Cf. Latvian tirpstü, gibstu, atzilstu, vīstu, atbilstu, klīstu, svīstu, nīstu, škīstu, likstu, mirstu, nikstu, slīkstu, trīkstu, digstu, ligstu, sirgstu, atspirgstu, stingstu, birstu, irštu, mirstu, silstu, svilstu, dilstu, pazīstu, grimstu, rimstu, dzimstu, bīstos, [dzīst; skurbstu, stulbstu, plūstu, trūkstu, tū̄kstu, kūstu, listu, skumstu, grīstu, klūstu, pūstu, žīstu, and the like

¹⁵ Cf. Latvian pampstu, slāpstu, skābstu, kalstu, alkstu, sprāgstu, vārģstu, salstu, etc.

¹⁶ In particular, the expansion of -st(a) stems in the present may be explained precisely by the fact that they offer the greatest possibilities for verbs with the most diverse kinds of root vocalism and accentuation to enter the present system (which, however, does not preclude the possibility of expressing other oppositions, for example, intransitivity versus transitivity depending on the root-vowel grade; cf. Lithuanian dỹgsta: diegia, Latvian digstu: diẽdzu, etc.). Verbs acquiring a nasal infix in the present, by contrast, prove most suitable for distinguishing verbal stems (in particular, tense stems) by means of accentuation.

156

duñža, priñsta: pruñsa, gniñsta: gniñsa, dūsta: duñsa, čyžta: čiñža, šąžta: šañša, sūsta: suñsa, dvýsta: dviñsa, gyžta: giñža, glyžta: gliñža, etc.¹⁷, strictly speaking, presupposes not only the diachronic adjacency of the infix -n- / -m- to the stem-forming type -st(a) (dūsti) du-n-ś(a): *dūs-sta, but also the synchronic co-occurrence of these formatives of the type *du-n-s- + sta > dūsta, inevitable for all stems with an infix that acquire the formative -st(a). Moreover, forms of the type dūsta или grýsta < *gris-sta < *grins-sta must undoubtedly, during the initial period of their formation, be perceived not as simple stems in -sta: dus-sta или gris-sta, but specifically as present stems with two terms — -n and -st(a), especially since ū and y̆ in the root of these verbs (as, indeed, in other analogous cases) are archiphonemes in which ū and un (um), y and in (im) are neutralized.

Given such a close and long-standing connection between stems with a nasal infix and stems in -st(a), not to mention evidence from other ancient Indo-European languages, the explanation of the origin of -st(a) recently proposed by J. Kurylowicz¹⁸ must be considered debatable, since it does not take into account the relationship between these two stem types. It would be highly peculiar if the Baltic pattern, which presupposes a connection between stems in -st(a) and stems with a nasal infix, merely happened to imitate the Indo-European state of affairs. J. Kurylowicz is undoubtedly right, however, in outlining general principles governing the emergence and spread of formatives, with due regard to the relations between primary and secondary semantic functions. Of particular importance is that this schema implies a fundamental mobility in the semantic values of this phonetic complex—from zero to derivational and inflectional meanings. In various Indo-European languages, including Baltic, the element -s- marks precisely different semantic and grammatical ranks and transitions from one rank to another. Incidentally, these transitions were facilitated by the archaic system of conjugational inflections, which has become increasingly clear as a result of recent research (cf. the inflectionless third person singular of the verb, the lack of distinction between second- and third-person inflections in the conjugation, the connection between the inflections of the second conjugation series (Hittite -ḫi) and the mediopassive and perfect, the identification of a broad stratum of injunctive formations, etc.).

¹⁷ See J. Kazlauskas, Op. cit., pp. 324–325. For Latvian examples of this type, see J. Endzelīns, Latviešu valodas gramatika, pp. 762–763.

¹⁸ See J. Kuryłowicz, Op. cit., pp. 50–51. The source of this formative is held to be participial forms in -tas from verbs whose roots end in -d, -t, -s, -š, -z (-sk, -šk). These forms allow for a double interpretation, for example, vìrstas = 1. virt-tas (primary form) or 2. virt-stas (secondary form). Cf. below (p. 51): “This possibility is exploited if the respective verb shows a double function, e. g. if, fundamentally transitive, it has in certain contexts an intransitive value or if it is a verb of state with a secondary durative function. Thus a participle like vìrstas having the primary passive meaning ‘which has been turned over’ and a secondary intransitive meaning ‘which has turned (over)’, can be split into vìrt-tas (primary semantic function) and vìrt-stas (secondary function). Purely virtual in the participle and in the infinitive in -ti, the split is actualized in the corresponding present tense: vìrt-tas: vìrt-st-as=vert’ù: virt-st-ù. According to § 36 we get the subtraction of -t- in virt-t-(as) entailing the restitution of the full grade vert’ since the zero-grade of the root is only a corollary of the addition of -tas, hence participle vert’-: virt-st-= present vert’-: virt-st-, representing a correct proportion…”, etc.

157

In view of the additional relationship -n-/-m-: -st- in Baltic present stems, a number of parallels and analogies suggest themselves. Bearing in mind that Baltic, Slavic, and Germanic languages adopted only one of the three types of Indo-European present with a nasal formative (1. *linekʷ-: *linkʷ-, 2. *k’l̥neu-: *k’l̥nu-, 3. *kʷrineH₂-: *kʷriṇ₂)—Baltic the first, Slavic the second, and Germanic the third—while retaining the general meaning of intransitivity, one must conclude that there is a certain legitimacy to comparing the Baltic infixal and Slavic suffixal types of present stems with the element -n-. In this connection, it is highly significant that originally Proto-Slavic stems in -ne- (*dvignoti, *dvigneši) formed exclusively the asigmatic aorist (*dviǵō), as, for example, is attested in Old Church Slavonic monuments. Thus, for a certain period, the following distribution of forms in the aorist can be reconstructed: present stems with a nasal (-n-) suffix → asigmatic aorist; present stems without a nasal suffix (though not all of them) → sigmatic aorist, i.e. *dvigneši — *dviǵō, but *znaješi — *znaxъ. If we take into account the stage at which specially marked stems began to emerge from the injunctive, the coexistence and balance of these same two types—with -n- and with -s-—can also be posited for Proto-Slavic. It is characteristic that Slavic languages, too, require us to take account of the expansion of the sigmatic type; cf. the late forms of sigmatic aorist formation from verbs with -n- in the present, dvigoхъ beside the older dviǵō. Moreover, as with Lithuanian forms, we must note the co-occurrence of the nasal element and -s- (-x-, -š-) within a single word form in sigmatic aorists from nasal-present stems, such as КОСНѦХЪ, ПО-МѦНѦХЪ, ОУСЬКНѦХЪ, and the like¹⁹ (i.e., in the ultimate analysis, *-šek-non-s-). These analogies, of course, make the supposition that -s- in Baltic present stems is connected with the Slavic aorist marker still more plausible, and lead to the hypothesis that -t- in -st- in Baltic forms is related to -t- in the aorist forms of the second and third persons singular, such as ritō, jetō, and perhaps also dadtō, if the latter is not seen as a reduplication. Since the view that this form is simple (*da-stō) has recently gained increasing acceptance, it is appropriate to compare it with Old Prussian dāst, Old Lithuanian dūosti, as well as a number of other formations; cf. Hittite daš (second–third person singular preterite of the ḫi-conjugation), Albanian dhashë (first person singular aorist, ‘I gave’), Venetic dona.s.to, and the like.²⁰ The examples cited above already show that -s- appears sometimes as a marker of present stems, sometimes as an aorist marker, and sometimes as an inflection (or part of one) marking person in verbal conjugation. This ‘mobility’ of -s- must certainly be taken into account when addressing the question of its origin.

Leaving further discussion of the Slavic aorist aside, it is appropriate only to recall that the distribution of its forms was determined by

¹⁹ См. Н. Ван-Вейк, История старославянского языка, М., 1957, стр. 342—345.

²⁰ See V. Rosenkranz, Historische Laut- und Formenlehre des Altbulgarischen, Heidelberg, 1955, p. 129; V. V. Ivanov, op. cit., p. 82; idem, Reflection of Two Series of Indo-European Verbal Forms in Proto-Slavic, — “Slavic Linguistics,” Moscow, 1968, p. 257; V. N. Toporov, On the Question of the Evolution of the Slavic and Baltic Verb, — VJaS, 5, 1961, p. 69.

158

not only by the type of stems, but also by the accentual paradigm of the verb. Cf. the accentual-morphological rule, which existed even before the appearance of written monuments of the Old Church Slavonic language and explained the distribution of aorist forms²¹: underived barytone verbs with consonant stems formed a simple aorist if the stem ended in a consonant, and a sigmatic-root aorist if the stem ended in a vowel or sonant. Underived oxytone verbs of the same type formed a sigmatic aorist of the type ВѢСЬ — ВЕДЕ if the stem ended in a consonant, and a sigmatic dental aorist if the stem ended in a vowel or sonant (see V. A. Dybo, op. cit., pp. 37–38). Hence: pirъ, but bi or idъ, but věsъ, and so on.

Whereas in Baltic and Slavic verbs the elements -n- and -s- compete with one another, appearing together in one and the same form only rarely (and such formations are either unstable, transitional in character, or imply that the former meanings of the co-occurring elements have been forgotten), some other Indo-European languages developed regular formations characterized by the combination of nasal and sigmatic elements.

First of all, the Tocharian A verbs of class VIII and the Tocharian B verbs of class X should be noted here; they are characterized by the suffixes -nās-, -nās- and nāšk-, -nāsk-, respectively. Cf. Tocharian A yār- ‘bathe’ — yārnāssi (Inf.), wäl — ‘die’ — wällāstār (< *wäl-nās-tär), etc.; Tocharian B yāp- ‘enter’ — yänmaskau (< *yäp-naskāu), päk- ‘intend’ — päknās-tär, käm- ‘come’ — kämmaššäm (mm < mn), täm- ‘be born’ — tänmastär (nm < mn), etc., not to mention a number of verbs whose roots end in -n: klin- ‘be obliged’ — klinassäm, aun- ‘begin’ (in the med.) — aunastär, pin- ‘blow’ — pinasle (Ger. I), rin- ‘leave’ — rinastär, sain- ‘defend oneself’ — sainaskentär, and so forth²². The segmentability of this Tocharian formant -nās- (-nās-), -nāšk- (-nāsk-) is ensured by the existence of a series of present stems characterized by only one of the two aforementioned formants, or in which these formants occur in other combinations. It is sufficient to recall such classes as Tocharian B VIII (-s-: -s-): er-: ersām, aip-: aiyesem, kāil-: kāilsām, kau-: kausām, and so on; IX (-sk-: -ss-): yām-: yamskau, we-: wskau, ai-: aiskau, and so on; XI (-s-…sk-: -s…ss-): auk-: auksāsšām, tu(twā-): twāssām, and so on (a mixture of classes VIII and IX), on the one hand, and classes VII (nasal infix): kätk-: kättankām, pik-: pinkām, and so on; VI (-nā-): alp-: alpanam, rap-: rapanam, and so on; XII (-ññ-): käsk-: käskannitär, klänts-: kläntsannī, and so on, on the other. In light of these relationships, the formation of class X in Tocharian B (and class VIII in Tocharian A) is explained comprehensively.

²¹ This rule was formulated by V. A. Dybo; see his article: “The Accentuation of the Slavic Verb and the Forms of the Old Church Slavonic Aorist,” — “Brief Communications of the Institute of Slavic Studies,” issue 30, Moscow, 1961, pp. 33–38.

²² See W. Krause, Westtocharische Grammatik, Bd. 1, Das Verbum, Heidelberg, 1952, pp. 96 ff.; W. Krause, W. Thomas, Tocharisches Elementarbuch, Bd. 1, Grammatik, Heidelberg, 1960, p. 214; E. Sieg, E. Siegl i ng, W. Schulze, Tocharische Grammatik, Berlin, 1931, pp. 357 ff.

159

Among the Anatolian factors relevant here²³, Hittite verbs should be noted first of all, whose present stem is formed by combining a suffix with a nasal element and the suffix -šk- (-šš-), the former usually having a causative meaning and the latter an iterative one. Cf. arnušk- ‘bring repeatedly’ alongside ar- ‘remain’, ‘reach’ and arni- ‘bring’, ‘deliver’; also compare huš-nu-šk-, and so forth, or verbs in -nešk- (Luwian forms in -nešš-): cf. Hittite ti-ia-ni-eš-ni-mi beside Luwian ti-ia-ni-eš-šu-i; cf. Hittite tešḫannešk-, and so forth. Of particular interest are cases in which -šk- is attached to a present stem with a nasal infix; cf. ḫarninkisk- ‘destroy repeatedly’ alongside ḫarnink- ‘destroy’ and ḫark- ‘perish’. As with the Tocharian languages, in considering Hittite formations with a complex formant consisting of a nasal and -š(k)-, one should also bear in mind other types of present stems (-annai-, -nu-, -nin-, -eš-; cf. denominal verbs such as parkueš- ‘become clean’ beside parkui- ‘clean’, šalleš- ‘increase’ beside šalli- ‘big’, and so forth). A sequence -n-s- is also posited on the basis of the Luwian ↑e-n-s ‘set’, ‘dedicate’. Anatolian facts, and Hittite facts above all, are highly instructive, particularly in showing that the spread of the suffix -šk- is attested within the documented history of the Hittite language itself and is connected with the displacement of the suffix *-šša- (from *-so-). Therefore, the reconstruction of older present stems with a formant combining a nasal element with -š- (rather than -šk-) cannot be ruled out. However, the specialization of the suffix -šk- in Hittite (primarily for expressing iterativity) and its universalization (its ability to attach to any verbal stem) somewhat reduce the heuristic value of this formation for reconstructing the original range of cases in which a complex formant of this type was used. The same can, with due reservations, be said of Ancient Greek verbs of the type ἀλύσκάνω ‘avoid’, ‘escape’ (already in Homer and Hesiod), alongside ἀλύω ‘be in confusion’, ‘rush about’, ἀλύσσω (in Homer), or of older formations such as αὐξάνω ‘increase’, ‘grow’, and the like, beside αὔξω ‘make grow’, and so on, where the order of the elements is inverted compared with the examples considered thus far. The same inverted order is observed in Armenian verbs of the type harç-an-em ‘I ask’ (instead of the expected harçem), where the original sequence *-sk’- + *-n- is discernible (cf. Old Indic prcchámi, Latin poscō). In connection with the Armenian data, however, it is important to emphasize the following two features—the medial meaning of the element -n- in present stems (cf. melkel ‘weaken’ [active] — melkanal ‘become weak’ [middle]²⁴) and the formation of the aorist marker ç (cf. sireac from sirem ‘I love’, or gorceaç from gorce

²³ О них подробнее см. И. Фридрих, Краткая грамматика хеттского языка, М., 1952, стр. 82 и след.; G. Bechtel, Hittite Verbs in -sk-, Ann Arbor, 1936; В. В. Иванов, Общеиндоевропейская, праславянская и анатолийская языковые системы, стр. 139 и след., особенно стр. 147—148. О соотношении носового форманта и форманта *-sk- в индоевропейских языках см. W. Couvreur, Les dérivés verbaux en -ske/o- du hittite et du tokharien, — „Revue des études indoeuropéennes“, t. 1, fasc. 1, 1938; см. также C. Watkins, Indogermanische Grammatik, Bd. III: Formenlehre, Heidelberg, 1969, стр. 70 и след.

²⁴ See E. G. Tumanyan, Drevnearmyanskiy yazyk, Moscow, 1971, pp. 334–335; A. A. Abramyan, Uchebnik grabara, Yerevan, 1964, p. 116 et seq. (in Armenian).

160

‘I do’) from Proto-Indo-European -sk-²⁴а. The fact that -sk- functions as an aorist marker alongside -s- in the aorist of other Indo-European languages should probably be taken into account in connection with the relationship between Baltic -st- and -s- in a number of other related languages. Nevertheless, given the penetration of -sk- into the preterite system in Tocharian (and, as some following Karsten assume, in Slavic: -ахъ < *-ā-ske-), one may suppose that the opposition between a basic verb and a verb with the marker -sk- was known to the so-called “Pontic” languages (Anatolian, Tocharian, Armenian, Slavic)²⁴⁶.

In connection with the Baltic problem, however, examples featuring the sequence -n- + -s- are ultimately more important. Among them, the relationship between Homeric χνέω: ἔχυσσα deserves attention, as it permits the reconstruction *χu-vε-σ-μ²⁵, as do Vedic forms such as gr-niṣ-e (from gir-) or pu-niṣ-e (from pū-)²⁶, in which the marker -niṣ- is associated with the middle inflection; and Venetic forms such as zonasto, where -nas- is clearly segmented thanks to the parallel form zoto. Cf. mexo zona.s.to re.i.tiia.i. ner.ka lemeto.r.na “Nerka, daughter of Lemetor, dedicated me to Reitia” (PID 26. Este) or mexo zona.s.to sahnate.i re.i. tiia.i porah “Porah dedicated me to Reitia Sanatis” (PID 31. Este), but: vhremah.s.tna zoto re.i.tiia. i “Fremastna dedicated (me) to Reitia” (PID 19. Este); cf. Lydian ↑ens beside ↑es- ‘dedicate’, ‘place’ (as well as Lycian tas-: ta- or Hittite zikk- [t-sk-] from tehhi ‘place’, ‘set’)²⁷, not to mention a number of other cases²⁸.

After considering this material, it is appropriate to recall the undisputed conclusion that Hittite verbs in -s- (as well as certain other sigmatic forms) are connected with the -ḫi conjugation and, consequently, with the second series of Indo-European verbal forms²⁹. Hence the middle meanings of the counterparts to Hittite forms in other Indo-European languages, which explain the connection

²⁴а See R. Godel, Les aoristes arméniens en -ć-, — „Studia classica et orientalia Antonio Pagliaro oblata“, II, Roma, 1969, pp. 253—258.

²⁴⁶ See B. Rosenkranz, Zur Entstehungsgeschichte der indogermanischen Verbalflexion, — „Arbeitspapier“, N 16. Institut für Sprachwissenschaft — Universität, Köln, 1971, pp. 5—7. Rosenkranz immediately notes the presence in the languages mentioned of formations specifically of -l- origin, introduced into the verbal system, and the absence of an Indo-European-type perfect.

²⁵ Cf. similarly: dvigneši — dvigoxъ — dvignoxъ (cf. КОСНЖЖХЪ).

²⁶ См. об этих формах Т. Я. Елизаренкова, Аорист в Ригведе, М., 1960, стр. 29. Следует также напомнить, что в древнеиндийском, помимо трех презентных классов с носовым элементом, существовали многочисленные глагольные образования на -s-: сигматический аорист разных видов (в частности, на -sa- и даже на -ṣiṣ-; последний тип — результат неправильного анализа существовавших сигматических форм и аналогических выравниваний), будущее время, прекатив, дезидератив и даже презентные основы, см. F. B. J. Kuiper, Zur Geschichte der indoiranischen s- Präsensia, — „Acta Orientalia“, 12, 1934.

²⁷ Cf. also Phrygian ε-δαες ‘placed’.

²⁸ Cf., for example, Avestan činas ‘he will assign’ (from the assumed *ki-na-s-t?).

²⁹ See В. В. Иванов, Указ. соч., pp. 173—174; cf. also his Отражение двух серий..., p. 257 et al.

11. 902

161

-s- with the mediopassive in Tocharian, the middle voice in Vedic³⁰ and—and in this case this conclusion is fundamental—with intransitive-inchoative verb formations in -st(a) in the Baltic languages. It is plausible that the Lithuanian preterite of non-transitive verbs in -o (from *-ā)³¹ is likewise correlatable specifically with the second series of Indo-European conjugation.

In our view, Baltic -st- as a formative of present stems can, in light of all that has been said above, be compared with the inflection -st- found in a number of Indo-European languages. First and foremost, this concerns the second- and third-person preterite endings in the -ḫi conjugation (these endings being absent from the preterite of verbs in -mi), cf. dāišta, from dāi- ‘to set’, ‘to put’ (first-person singular present -teḫḫi), alongside the second-person plural imperative dāišten; temišta, from tema- ‘to speak’ (first-person singular present temaḫḫi), alongside the second-person plural imperative temišten; eš(š)ešta, from ešša- ‘to do’ (first-person singular present eššaḫḫi); aušta, from au(š)- ‘to see’, ‘to look’ (first-person singular present -uḫḫi), alongside the second-person plural imperative aušten and the second-person plural present aušteni, ušteni (and autteni)³², directly comparable with Lithuanian aušta (: aušo, aušti) and Latvian àust (: àussa, àust)³³. These same endings occur in a number of paradigms of a mixed nature, combining features of the -mi and -ḫi conjugations; cf. dalešta, from dala-, daliia- ‘to leave’ (first-person singular present dalaḫḫi and daliiam(i)), alongside the second-person plural present dalešteni, and so forth. Even more interestingly, alongside the ending -šta in the second and third person singular of the preterite, there are also more widespread endings that seem to be component parts of that ending, namely -š and -ta, cf.: dāš (second- and third-person singular preterite of dā- ‘to take’), dāiš (third-person singular preterite of dāi- ‘to set’, ‘to put’, alongside dāišta and even dāitta in the second person), paiš (third-person singular preterite of pāi- ‘to give’, with pešta in the same person and pāišta and even pāitta in the second person), nāiš (third-person singular preterite of nāi- ‘to direct’, ‘to send’, with nāešta in the same person), zāiš (from zāi- ‘to cross’), ḫalzāiš (from ḫalzāi- ‘to call’), tarnaš (second- and third-person singular preterite of tarna- ‘to leave’,

³⁰ This concerns not only forms such as puniše or griṇiše, but also the recently noted fact that medial forms of the aorist in -s- (aorist 4) decidedly predominate in the indicative. A scholar of the Vedic aorist is inclined to connect this tendency with the fact that vowel-initial endings predominate among secondary middle endings. It is characteristic that, among forms in which -s- stands before a consonant, most are likewise middle forms. Hence the plausible conclusion that this element -s- came to be associated with the middle paradigm in general. See T. Ya. Elizarenkova, op. cit., p. 78.

³¹ In the Latvian literary language, as in most dialects, the preterite in -ā became generalized, although a preterite in -ē is also attested in Middle Latvian texts of the sixteenth and seventeenth centuries and in some modern dialects (the southwestern part of Courland, most Upper Latvian dialects). See J. Endzelīns, Latviešu valodas gramatika, pp. 863 ff.; M. Rudzīte, Op. cit, pp. 367 ff. The Old Prussian data (preterites in -ā and -ē) are too scant to permit a reliable interpretation.

³² For Indo-European parallels, see V. V. Ivanov, Obshcheindoevropeiskaia…, pp. 153 ff.

³³ Cf. also J. Endzelīns, — FBR 17, 1937, p. 164; idem, — ZfslPh, vol. 16, 1938, pp. 107 ff., and elsewhere.

162

‘to set in place’, ‘to release’, alongside tarništa in the third person—a doublet form); šarraš (third-person singular preterite of šarra- ‘to divide’, ‘to break apart’, alongside šarratta in the second person), uaštaš (second- and third-person singular preterite of uašta- ‘to sin’), temaš (third-person singular preterite, alongside temišta in the second and occasionally the third person), dališ (third-person singular preterite)—as regards the ending -š; and šakta (second- and third-person singular preterite of šak- ‘to know’), datta (second-person singular preterite, alongside dāš in the third, and sometimes second, person), dāitta (second-person singular preterite, alongside dāiš in the third person and sometimes dāišta in the second person), daliiat (third-person singular preterite, alongside the competing forms dališ and dalešta), hašta (third-person singular preterite of haš ‘to generate’), etc.—as regards the ending -ta. Interestingly, these -š and -ta endings may correspond to the inflections of the respective forms in the -mi conjugation; cf. hatraēš and hatraīt (hatraiš also occurs), from hatrai- ‘to write’; daškēš and daškit, from dašk- ‘to speak again’; harši (more often ḥarti) and ḥarzi, from har(k)- ‘to hold’, ‘to have’, and the like. Compare, however, also the generalized endings of the second and third person singular preterite in the -mi conjugation, -t (-ta): epta from ep- ‘to seize’; kuenta from kuen- ‘to strike’, ‘to kill’; puništa from puniš- ‘to ask’; uuatet from uuate- ‘to bring’; iijat from iia- ‘to do’, and so forth. Consequently, the second- and third-person singular preterite forms most clearly demonstrate how the -šta inflection arose in Hittite, how readily and distinctly its component parts could be isolated, and even the direction in which this inflection spread (cf. imperative forms)³⁴. All this is relevant to the history of the formation of Baltic -st(a).

Tocharian facts concerning the imperfect, preterite, conjunctive, and preterite optative bear many similarities to the Hittite facts. Let us compare just a few examples: Tocharian A — wolkašt (second-person singular preterite of wätk- ‘to command’, alongside third-person wotäk); kakläšt (the same form, as with the following example, from käl- ‘to carry’, ‘to endure’, alongside kakäl, third person); Tocharian B — yamassit (second-person singular imperfect

³⁴ It is for Hittitologists to determine the original state of affairs and the stages in the spread of the inflections -š, -ta, and -šta in the second and third person singular of the preterite. It is difficult to say anything definite on this point for now, since the surviving texts reflect two opposing patterns: a single form in the third person (dāš, dāiš), alongside competing forms in the second person (dāš and datta, dāišta and dāitta), and a single form in the second person (tarnaš, temišta), alongside competing forms in the third person (tarnaš and tarništa, temaš and temišta). The natural assumption that there was once a distribution of the type: second-person singular preterite -š, third-person singular preterite -ta (as apparently in Hieroglyphic Hittite: -sa/-s and -ta; Luwian: -s (?) and -ta; perhaps in Palaic: -si and -til-i, and Lycian, where, however, second-person singular preterite forms are unattested), from which the ending -šta, common to the second and third person, subsequently arose, naturally requires strictly Hittitological evidence. It seems possible to speak with greater confidence of the gradual spread of the more representative inflection -šta (partly because it combines distinct person markers). Incidentally, something similar was observed in the history of the inflections of simple past (sigmatic) tenses in the Slavic languages, especially during the period when this tense system was breaking down (cf. orientation toward the most characteristic form). On second- and third-person singular preterite endings, see I. Friedrich, op. cit., pp. 106 ff.; H. Kronasser, Vergleichende Laut- und Formenlehre des Hethitischen, Heidelberg, 1956, pp. 190–193; C. Watkins, Indo-European Origins of the Celtic Verb, I, The Sigmatic Aorist, Dublin, 1962, pp. 90 ff., 97 ff.; cf. also his Indogermanische Grammatik, vol. III, and elsewhere. J. Kuryłowicz also treats Hittite -š in the second (and third) person singular preterite as the result of a late, purely Hittite development (the same applies to -š in Old Persian akunaus ‘he did’, and the like); see J. Kuryłowicz, Op. cit., pp. 156 ff.

11*

163

from yām- ‘to do’, alongside yamaṣṣi (3rd person); śälasta (2nd person singular preterite of käl- ‘to bring’, alongside śala- (3rd person)); yamit (2nd person singular optative of yām-, alongside yami (3rd person))³⁵, etc.

Inflections of this type may in one way or another have been reflected in other languages as well (cf. Venetic zonasto or vhaxṣδo, Latin -īstī, etc.). Further enumeration of them in this connection is unnecessary. By contrast, the following facts of two kinds must be considered exceptionally important and relevant to the possible history of the Baltic verbal formant -st(a).

First, we are dealing with the expansion of -s- in the forms of the Old Indic precative (aorist optative), attested in written records³⁶. Compare the corresponding fragments of precative forms in Vedic (active):

1st person singular bhūyām 1st person dual bhūyāva 1st person plural bhūyāma 2nd person singular bhūyās 2nd person dual bhūyātam 2nd person plural bhūyāta 3rd person singular bhūyās 3rd person dual bhūyātām 3rd person plural bhūyur

and in the later Classical Sanskrit (active):

1st person singular bhūyāsam 1st person dual bhūyāsva 1st person plural bhūyāsma 2nd person singular bhūyās 2nd person dual bhūyāstam 2nd person plural bhūyāsta 3rd person singular bhūyāt 3rd person dual bhūyāstām 3rd person plural bhūyāsur

As comparison of these paradigms shows, the sigmatic forms of the 2nd and 3rd person singular served as the starting point for the spread of this -s- as a component of the precative marker into the 1st person singular, and then into forms of other numbers as well. It is entirely plausible that something analogous may have occurred in the development of this formant in the Baltic languages³⁷.

Second, for the hypothesized process of reanalysis associated with a change in the original (or, more precisely, earlier) locus of -st- in Baltic, it is highly relevant to take account of forms in which -st- can no longer, strictly speaking, be regarded merely as an inflection, but has not yet acquired the status of a marker of a verbal class within one and the same tense (for example, the present), and retains an intermediate position. Compare Ancient Greek λέκτο, for which an older form *λέχ-στο is very plausibly reconstructed alongside the secondary λέξατο; Vedic a-tam-sta (=Avestan mə-sta), etc.³⁸

³⁵ Cf. the corresponding Tocharian A forms: *kälñāsiṣt – *kälñāsi, and the like.

³⁶ See T. Burrow, The Sanskrit Precative, — „Asiatica. Festschrift Friedrich Weller“, Leipzig, 1954, pp. 35—42.

³⁷ Cf., for example, the following hypothetical scheme: I: 1st person singular -ō→ II: 1st person singular -ō → III: 1st person singular -ō→ IV: 1st person singular -stu 2nd person singular -s 2nd person singular 2nd person singular -st(i) 2nd person singular -sti 3rd person singular -(t) 3rd person singular -st (-s, -t) 3rd person singular -st(a) 3rd person singular -sta If this scheme is (in very general terms) plausible, the key question becomes that of the redistribution that led to the creation of a present-tense marker from material belonging to the inflection.

³⁸ Cf. H. Kronasser, Op. cit., p. 193.

164

Assuming that it was precisely the 2nd and 3rd person singular forms in which reanalysis could have begun, resulting in the transformation of -st- into an indicator of the present stem³⁹, it is important to recall the numerous cases in which the endings of the 2nd and 3rd person singular are indistinguishable in various Indo-European languages, as discussed to some extent above and repeatedly in the literature, particularly in recent years⁴⁰. At the same time, one must also bear in mind cases in which the ending differs in the 2nd and 3rd persons from the more usual pattern (cf. -s in the 3rd person, Vedic dhās and the like⁴¹, or -t in the 2nd person, Hittite datta, etc.)^41а. Of course, these conditions encouraged both the very beginning of the reanalysis process and the fact that these particular forms induced the spread of -st- throughout the conjugational paradigm in the Baltic languages. Characteristically, in the Slavic languages too, related types—in particular, the sigmatic aorist and the imperfect—do not distinguish the 2nd- and 3rd-person endings and may, at least in some cases, never have distinguished them (on this, see the author's other works)⁴². In general, it must be acknowledged that the numerous reconstructions of 2nd- and 3rd-person endings with -s- added to the original root-final consonant, of the type... (t)-s-s- (cf. Hittite daiš < *dejk’-s-s) or (t)s-t, once of heuristic and methodological value, increasingly obstruct the identification of an older and more specific pattern, later obscured by analogical tendencies. Paradoxical as it may seem, in discussing the origin of -st-, researchers should not bind themselves to rigid constraints. The answer may lie in other parts of the word (for example, in the ending rather than in the stem-forming element) and in other parts of the paradigm (for example, in the aorist rather than in the present). A characteristic example is the conjugation of the athematic verb dati in the present and aorist in Old Church Slavonic, and dúoti in Lithuanian⁴³.

³⁹ Strictly speaking, of the stem that later characterized the present. It might be more accurate to speak of injunctive stems, which in the Baltic languages entered the present system, in others the (sigmatic) aorist system, and in still others both the present and the aorist systems.

⁴⁰ See T. Burrow, An Archaic Verbal Termination in Early Indo-Iranian, — “Indo-Iranian Journal”, vol. 1, 1957, pp. 61—76; A. S. C. Ross and R. A. Crossland, Supposed Use of the 2nd Sg. for the 3rd Sg. in “Tocharian A”, Anglo-Saxon, Norse and Hittite, — “Archivum Linguisticum”, vol. 6, 1954, pp. 112—121; see also — KZ, vol. 69, 1951, pp. 150—164.

⁴¹ Об этих формах ср. J. N. Narten, Die sigmatischen Aoriste im Veda, Heidelberg, 1964, стр. 18 и след.; C. Watkins, Indogermanische Grammatik, Bd. III, стр. 53 и след.

^41а Recently, Rosenkranz pointed out that Hittite preterite endings correspond to Elamite personal endings (cf. 1st person singular -h, 2nd person -t, 3rd person -š, etc.). It is possible that much could be clarified in this connection by turning to other “Nostratic” languages, whose conjugational systems show clear parallels with the endings in question.

⁴² The situation seems to be somewhat different with examples of indistinguishable verbal endings in Prussian.

⁴³ For the conjugation of dúot in Latvian, see J. Endzelins, Latviešu valodas gramatika, pp. 723—725; M. Rudzīte, Op. cit, p. 359, and others; cf. Prussian dāst, etc.

165

Present tense Singular 1st person datь dasi 3rd person dastъ Dual 1st person (davē) 2nd person dosta 3rd person daste, dosta Plural 1st person datь 2nd person daste 3rd person dadētъ Aorist daxъ dastъ, da dastъ, da daxovē dasta daste, dasta daxomъ daste daše Present tense duomi duosi⁴⁴ duosti, duost duova duosta duosti, duost duome duoste duosti, duost

In this comparison, two features are particularly noteworthy (if we take the aim of this article as our point of departure): first, the homonymy, so to speak, of -st- (cf. -stъ as a 3rd-person singular ending, dastъ — da, and -sta in the 2nd-person dual as the result of recoding… d-ta, cf. ideta or sukata, as -sta, and the like); second, the cross-temporal occurrence of this -st- (cf. dasta, 2nd and 3rd person of the present and aorist), which, alongside many other facts, provides a basis for reconstructing a state preceding the separation of the tense system from the continuum of injunctive forms—a reconstruction that seems especially plausible for the Baltic verb. Finally, it is important to note that in the paradigms cited, -st- is concentrated precisely in the 2nd- and 3rd-person forms.

Finally, in discussing the involvement of -st- in the system of present-stem markers in Baltic, we must point out the difference in the conditions under which -st- appears where it functions as an ending (Hittite or Tocharian) and where it is a stem-forming element (the Baltic languages). In the first case, -st- is always preceded by a vowel (cf. tarništa, pahhašta, dāišta; šälästa, wotkast; dastъ, and the like); in the second, it is always preceded by a consonant (cf. alpsta, dȳksta, and the like). This difference in conditions must have decisively affected the subsequent fate of -st- in Baltic: it could not remain an ending, since in that case the final root consonant and -st- would form word-final three-consonant clusters: … p-st, … k-st, … t-st, etc. Structures of the type … a-st, and the like, were virtually absent (apart from isolated examples in athematic conjugation, where, moreover, the connection between -s- and -d- was clearly felt; cf. kliest, 3rd person singular of kliedeti; nusėst from nusėdėti, giest from giedoti, and the like).

Of course, the question of -st- as a marker of present stems is inseparable from that of the derived conjugation of verbs in -styti (barštyti: berti, dūrstyti: durti, dāigstyti: dėgti, and the like)⁴⁵. Typologically similar is the use of -s- to form the present or aorist, on the one hand, and derived conjugation (the desiderative), on the other, in Old Indian. Characteristically, in derived conjugation -st- may also be added to a vowel at the end of the root (klōbyti: klōti, krāustyti: krāuti, kāustyti: kāuti, etc.), setting this group of forms apart from present stems containing the same element⁴⁶ and thus providing additional grounds for determining the original-

⁴⁴ Cf. in Daukša’s “Postilė”: duosi, duosi, duost—forms of particular interest in connection with the Slavic parallels.

⁴⁵ См. P. Skardžius, Op. cit., стр. 537—539; J. Otrębski, Op. cit., т. 2, стр. 388—390 и т. д.

⁴⁶ Cf. also the usual transitivity (or reflexivity) of verbs in -styti.

166

initial conditions for the differentiation of these two types⁴⁷. It should be assumed that verbs in -soti, expressing the idea of duration⁴⁸, also share a number of features both with the types already mentioned and with other formations containing the formative -s-, which are not considered here. Taking all these formations into account, and bearing in mind the processes, not uncommon in the history of the Baltic verb, whereby the grammatical and semantic status of one and the same sound element, or a complex of such elements, changes⁴⁹, we may hope for a gradual resolution of the entire range of problems connected with the origin of verbal forms containing the element -s-, in particular -st-. It is difficult to predict the details of the overall picture, but everything currently known about the relative chronology of the evolution of the Baltic verb seems to favor the assumption that the formation of a series of s-forms of the verb, and above all stems containing the element -st-, took place under substantially different conditions from comparable phenomena in most other Indo-European languages, namely in the presence of a typologically earlier situation. It is difficult to say what this situation was, that is, what set of grammatical categories and individual grammes defined it. Moreover, at present this uncertainty constitutes the main obstacle to interpreting the meanings once associated with -st-. In comparing Baltic verbs in -st(a) with the third-person singular middle form of the sigmatic aorist, K. F. Johansson was right only insofar as these two formatives are comparable in principle and ultimately. Nevertheless, it by no means follows from this conclusion that the emergence of Baltic -st(a) presupposes the existence in the system of an aorist (sigmatic) or a middle voice of the Ancient Greek or Old Indian type (or of any other specific forms that might explain this -t-: participles in -to-, verbal inflections in -tha, the stem of the demonstrative pronoun *-to-, stem types in -to such as Latin plecto: explicāre, Ancient Greek πέπτο: πέλω, Hittite mant-: Ancient Greek μένω, and the like).

⁴⁷ The formal identity of such genetically related words as Lithuanian māsto ^{‘he thinks’}, ‘he reflects’ (from *man-stā; cf. māstyti: manȳti: mīnti; cf. P. Skardžius, Op. cit., p. 538), Vedic a-mam-sta (cf. a-mam-sata) ‘he thought’, and Avestan māsta ‘he thought’ (*man-sta) is highly significant, even though -sta in the Indo-Iranian examples cannot be related semantically to Lithuanian -sta.

⁴⁸ Cf. iterative verbs in -sa in hieroglyphic Hittite: pipas(a)- ‘to give’ from pi(a)- ‘to give’, aias(a)- ‘to deal with something’ from aia- ‘to do’ (cf. -s(a)- in Luwian), etc., in the absence—unlike Hittite—of the suffix -sk-. In this connection, the Khotan Saka verbs with the suffix -s- (traced back to Common Iranian -sa-, thought to derive from *-ske-/*-sko*) are of considerable interest; they have an intransitive-inchoative meaning (like the Baltic verbs in -sta), cf. stās- ‘remain’; sta- ‘stand’, pars- ‘be released’; parrij- ‘save’, etc.; a similar type is also attested in Shughni. Among other facts from Khotan Saka, cf. -sta stems of the type sūh-: śūsta ‘equip’, pyū-: pyūšta ‘hear’, and the like, where s is associated with the corresponding consonant at the end of the root (cf. also bai-: basta ‘bind’, from *bad-, as well as past participles in -sta). Cf. L. G. Herzenberg, Khotan Saka Language, Moscow, 1965.

⁴⁹ Of course, the same can be said of a number of other languages. On the emergence of -s in the third-person singular inflection and in the aorist, see: C. Watkins, Indo-European Origins of Celtic Verb, p. 90 ff.; cf. also his Indogermanische Grammatik, vol. III, p. 53 ff.