Herman Kølln, „Oppositions of Voice in Greek, Slavic, and Baltic“

Herman Kölln, Oppositions of Voice in Greek, Slavic, and Baltic, Copenhagen, 1969, 65 p. In his work “Oppositions of Voice in Greek, Slavic, and Baltic,” the Danish linguist Herman Kölln examines questions concerning the semantic differentiation of the preterite formations in these languages. The author confines himself to several preterite types and analyzes in detail only those that he was able to establish as common to the Baltic, Slavic, and Greek languages. H. Kölln’s conclusions are so significant that it is worthwhile to annotate in greater detail those chapters of the book that deal principally with the Baltic languages. As is well known, cognate, single-root verbs in the Baltic languages generally have two preterites: one with the formant -ē-, the other with the formant -ā- (cf. keitė: ki-to). In the present-tense system, this opposition corresponds to an opposition between infixed or sta-stem verbs and je | o-stem verbs, and in the plane of content, to the opposition between transitivity and intransitivity. With this in mind, the problem posed and addressed by H. Kölln can be formulated as follows: could the opposition between the Baltic-language ē preterite and ā preterite have originated from the opposition between the sigmatic aorist and strong aorist in Slavic and Greek? Before proceeding with the annotation, it should be noted that the semantic opposition between effectiveness and ineffectiveness introduced in H. Kölln’s hypothesis plays a major role. It directly concerns the sphere of voice (diathesis), expressing the agent’s relation to the action and its object. H. Kölln calls an action ineffective if it is incidental with respect to the object, does not affect the object, and leaves it as it was once the action is complete. Conversely, if during an action the object is created or changes its state, H. Kölln considers verbs denoting such an action effective. It is not difficult to see that what is actually being discussed here is the opposition between middle and active voice, well known from other Indo-European languages (Greek, Indo-Iranian). This distinction between ineffective and effective verbs is well founded, since it allows their semantic relations to be understood more deeply and expressed more precisely. For example, Baltic infixed and sta-stem verbs are better termed ineffective: the term “intransitivity” is clearly too narrow to characterize their content, since they also include transitive verbs, cf. jūnita, rañda. From the perspective of the opposition between ineffectiveness and effectiveness, H. Kölln proceeds to examine the semantics of the individual types, first those of the ā preterite and then those of the ē preterite. As is well known, the ā preterite is found with all infixed and sta-stem verbs in the Baltic languages. Semantically, as already noted, they are intransitive, sometimes transitive, often inchoative, but always ineffective. Consequently, their ā preterite also has an ineffective meaning. The ā preterite is also used with verbs of the keṛta, kiṛto type. Semantically, these verbs seem to have a peripheral connection with infixed and sta-stem verbs. Some of them are intransitive, cf. sniēgti | sniēga, brēda, slēnka, leñda, telpa, sērga. Denoting slow movement, they differ from most infixed and sta-stem verbs, though they too sometimes have this character, cf. sniñga, brēda, dial. slīnksta, tilpsta. Other verbs of this type, though transitive, are nevertheless also ineffective, cf. lieka, perka, perša, renka, velka, since they do not denote a change in the object; the object remains as it was. kreñta, kērpa, kērta denote destructive actions. Only kemša, melža (preterite milžo) and treñkia. are clearly effective. 250 Su kêrta, kirto type verbs are related in their inflection to verbs with vowel gradation, such as dialectal dėla, dilė, dialectal svėla, svilė, dialectal gema, gimė, mėna, minė, gėna, ginė, vėja, vijo. The older ā preterite of these verbs is attested in old dictionaries (e.g., gimau in Milkė’s and Nesselmanno’s dictionaries) and in Latvian, cf. dilu, svilu, dzinu (ā). The ā preterite is also found in a group of verbs with an e/o present tense and an acute root vowel, cf. augo, bėgo, etc. In Lithuanian and Latvian, we find cases in which the preterite formant ā arose not because of semantic factors but was generalized in accordance with phonological features. In Lithuanian, the ā preterite is now found everywhere after j. It was apparently generalized also in verbs of the lipa, lipo type, where it replaced the older ē preterite still preserved in dialects (lipė): the ā formant, which is usually associated with the zero grade of the root, could readily have entered here because verbs of this type have precisely the reduced-grade root vocalism. Thus, the verbs of the types examined constitute a semantically defined group whose predominant meaning is ineffective. Several verbs denoting destructive actions are morpho(no)logically associated with them. Verbs of the types examined have, or have had, a preterite with the formant ā, whose meaning, accordingly, is also ineffective. H. Kölln established that in the preterite systems of Slavic and Greek, the strong (root and e/o) aorist has an ineffective meaning. H. Kölln makes a persuasive case, through etymological parallels, that the Baltic ā preterite is historically related to the strong aorist in Slavic and Greek. For example, Slavic nasal verbs generally have an e/o (thematic) aorist, cf. sędą: aor. sědo, “I sat down.” In terms of inflection, Lithuanian lieka, liko exactly corresponds to Greek pres. λείπω “I leave, throw,” aor. ἔλιπον “I left, threw.” Destructive verbs, which at first glance seem to contradict H. Kölln’s hypothesis, in fact support it, since in Greek, with their e/o aorist, they too pattern with ineffective verbs, as in the Baltic languages; cf. Lithuanian kirto, kirpo and Greek ἔταμον “I cut, chopped”; krimto and Greek ἔδακον “I bit, gnawed.” The counterparts of verbs of the auga, augo type in Slavic and Greek always have a thematic aorist, cf. bėgo, Sl. běgo, puolo, Sl. padъ, sėdo, Sl. sědъ, etc.; kando, Greek ἔδακον “I bit, gnawed.” The logical conclusion following from these correspondences, which H. Kölln also draws, is that the verbs of the types examined in the Baltic languages formerly also had a strong (root and thematic) aorist. The ā preterite was apparently once characteristic also of verbs of the kāsa, kāšė type, that is, verbs with an a present tense and a root vocalism. Their old preterites lako (inf. lakti), rako (inf. rakti), kaso, baro, kalo, malo, aro, kavo, šavo are found in old grammars, dialects, and Latvian. With regard to these verbs, it should be noted that, unlike in the ineffective verbs mentioned above, the preterite formant ā here is associated with an effective meaning. Verbs of the kāsa, kāšė type denote a continuous, rhythmically repeated action, that is, an action characteristic of the original o-vocalism verbs. In connection with the ā preterite, H. Kölln next surveys verbs with the so-called “second” stem in ā, cf. ieško (Univ. ieszku), ieškojau, ieškoti. Having a full grade in the root and denoting a state, verbs of this type contrast with the zero-grade inchoative infixed and sta-stem verbs, cf. ieško: dial su-ýška. Verbs with an 17* 251 The ā-stem is more widely represented in the Slavic languages, cf. iska, iskachǫ, iskati. A group of these verbs with a je/o present denotes an action of long duration, repeated rhythmically, cf. posati ‘to write’, lozati ‘to lick’, tesati ‘to hew’, stenati ‘to groan’, etc. H. Köllnas believes that this continuative meaning was precisely the reason why verbs with a je/o present and the infinitive in -ati were formed from the strong aorist of perfective (eventive-aspect) verbs, e.g., emljǫ, imati from jьetъ ‘to take’ (2nd–3rd sing. aor. jьetъ), dušǫ, dъchati from dъchnǫti ‘to breathe, to blow’. There are only five verbs with full-grade vocalism and a second ā-stem (miegōti, giedōti, raudōti, ieškōti, saugōti); evidently, unlike in the Slavic languages, this verb type remained undeveloped in the Baltic languages. Thus, Baltic verbs which, judging from their Slavic cognates, ought to have acquired a second ā-stem often have that ā only in the preterite, which is apparently an older state of affairs. Researchers had already noted that Baltic ā-preterites in certain cases correspond to Slavic verbs with a second ā-stem, cf. kāvo, kāuti, Sl. kovati, lāko, lāk­ti, O. Rus. lokati, kāso, kāsti, Sl. česati. From this it is easy to conclude that the second ā-stem in Baltic and Slavic (*ieškā-ti) arose through the generalization of the indicative preterite suffix -ā- (praet. *ieškā-). Since, as noted, verbs with a second ā-stem in the Baltic and Slavic languages denote continuous, rhythmically repeated (continuative) actions, it is logical to suppose that the old ā-preterite had a continuative meaning. But most Baltic verbs with an ā-preterite, especially infixal and sta-stem verbs, denote not the continuation of an action or state, but inchoation. The question arises how to explain the fact that these verbs, which are inceptive-ineffective and, in earlier times, as H. Köllnas believes, had an e/o aorist, now regularly show an ā-preterite in the Baltic languages, which formerly had a continuative meaning. In response to this question, H. Köllnas writes: “We must recognize that the e/o aorist and the ā-preterite once coexisted, that these verbs had both an e/o aorist and an ā-preterite, and that, as long as the thematic aorist was still alive, there was a semantic contrast between these two preterite types. Only after the e/o aorist disappeared, most likely for phonological reasons, and the ā-preterite became the sole preterite of these verbs, did it become capable of assuming all the functions of the preterite. As for the specific character of the opposition between these two preterites, I believe it was an aspectual opposition at the preterite level, similar to that known from other languages, such as the opposition between the aorist and imperfect … The continuative meaning of the ā-preterite leads me to suppose that there was an aspectual opposition here; contrasted with the other preterite, in casu the e/o preterite, the ā-preterite probably played the role of the imperfect” (p. 41). Comparing the aorist–imperfect opposition in the Slavic languages with the corresponding Baltic data, we see that the preterites known as aorists in the Slavic languages disappeared in the Baltic languages, while what survived there are the old imperfects, closely related to the Slavic imperfects. Researchers have long observed that the Baltic preterites ieško and sėdėjo are more similar to the Slavic imperfects iskaaсhъ, sěděachъ than to the aorists iskachъ, sěděchъ. Likewise, the Baltic ē-preterite vėdė more closely resembles the Slavic imperfect veděachъ (aor. vesъ). In the Slavic languages, imperfects formed from verbs with a strong (root or e/o) aorist have the formant -ēachъ, whereas, according to H. Köllnas’s hypothesis that ā-imperfects in the Balto-Slavic dialects were formed from the same verbs as the strong aorists, we would rather expect imperfects with the formant -aachъ. Eliminating this discrepancy, H. Köllnas showed quite convincingly that, from a historical point of view, the Baltic ineffective- 252 inchoative verbs’ ā-preterite is related to Slavic imperfective (imperfective-aspect) verbs with the formant -ati. Thus, Sl. dušą, dėchati has a second stem dōcha-, which can also be compared, in terms of its root vocalism, with the preterite of Baltic infixal and sta-stem verbs dūso. It may be supposed that the Slavic second stem with -ā- continues the imperfective ā-preterite, or ā-imperfect, of Balto-Slavic inchoative verbs, reflected in the Baltic languages as dūso. The second stem with -ā- (dōcha-) became associated in the Slavic languages with the je/o present (cf. dušą), most likely only after these stative verbs had been reinterpreted as imperfective. H. Köllnas believes that the initial stages in the formation of the imperfect were shared by the Baltic and Slavic languages. The ā-preterite, which initially served only as the preterite of continuative verbs (ieškoti, iškati), gradually spread and acquired a more specific meaning, becoming an imperfective preterite, or imperfect (dūso, dōcha-), opposed to the strong aorist of inchoative infixal verbs (dūsta, duńsa, dōchna). After the aorist–imperfect opposition had become firmly established in inchoative verbs, continuative verbs themselves were also aspectually differentiated at the preterite level: their ā-preterite split into an aorist iškachō and, with the formant -ā- added once again, an imperfect iskaa-chō. But in the Slavic languages, as noted, the ā-preterite (dōcha-) became associated with the je/o present (dušą); as a result, it lost its character as an imperfect and became the independent preterite of (imperfective) verbs. The result of this change was not only the creation of a second ā-stem, but also that imperfective verbs of the dušą, dōchati type followed the model of continuative verbs and developed their preterite in -ā- into a fully expressed imperfect in -ā- (from *-ājā-). H. Köllnas begins his analysis of the ē-preterite types with verbs of the lēsa, lēsē type. Verbs of this type have an e/o present and an ē-preterite and belong to a special class of transitive verbs. Since they (except for sėkti) denote a change in or creation of an object, they may be presumed to derive from a class of effective verbs. The corresponding Slavic verbs are also effective and have a sigmatic aorist, cf. vėsō “I led,” nėsō “I carried,” grēsō “I ruled” (praes. grebą), basō “I stabbed” (praes. bodą). H. Köllnas believes that verbs of this type also had a sigmatic aorist in the Baltic languages, since the sigmatic aorist has an effective meaning in Greek and Slavic. Another group of verbs in which the ē-preterite functions comprises je/o-stem verbs. Most of them are transitive and have alongside them root-cognate infixal and sta-stem verbs with an intransitive meaning. At the same time, however, among the je/o-stem verbs we also find some that do not differ in voice (diathesis) from infixal and sta-stem verbs; in other words, like the latter, they are ineffective. In that case, they form a state–inchoation opposition with the latter, cf. jaučia: junta. H. Köllnas’s attempt to establish the distribution of je/o-stem verbs—transitive-effective versus stative-ineffective—is noteworthy: if the subject of infixal and sta-stem verbs is a person or at least a living being (animate subject), then the corresponding je/o verbs are stative (state), cf. junta: jaučia, pra-virsta: verčia. If their subject is inanimate or passive (e.g., a sleeping person), the corresponding je/o-stem verbs are transitive (effective), cf. plýšta: plėšia, lūžta: 253 breaks. This rule, although it quite clearly highlights the semantic difference between the two groups of je/o-stem verbs, does not seem very satisfactory, since it is not always easy to say whether the subject is animate or inanimate, and, moreover, in certain cases—as H. Kölln himself also observed—infixal and sta-stem verbs have both transitive and stative je/o-stem verbs alongside them, cf. švinta: šviečia, šveičia. H. Kölln thinks that Baltic transitive-effective je/o-stem verbs may formerly have had a sigmatic aorist, since the corresponding Slavic verbs do have one, cf. praes. bljudą: aor. bljušь “I watched” (cognate with Lithuanian baudžia: baudė). But matters are less clear in the case of stative (ineffective) je/o-stem verbs: their je/o present may indeed be old, since it is also attested in Slavic, cf. emlją “I take,” dušą “I will breathe my last”; but it is hardly possible to suppose that there was also a sigmatic aorist here, even one unmarked for meaning, since the corresponding Slavic verbs have a second ā stem here, cf. imati, dočati. The striking fact is that these transitive (effective) verbs in Baltic languages take a je/o present: in other languages they have an e/o present, which apparently can also be reconstructed for Proto-Indo-European; cf. punishes, but Slavic bljudą, Greek πεύθομαι, Sanskrit bódhati; turns, but Latin verto, Sanskrit vártate. e/o or je/o presents appearing with transitive verbs that have an ē preterite apparently depends on the structure of the root: only verbs whose root structure is e/o present retained the present; their root structure is TeT-, whereas verbs with T(e)RT- roots acquired a je/o present (if they retained the ē preterite); cf. carries, but punishes, turns. The condition for this differentiation of present stems may have been the different relationship of verbal roots to ablaut. Roots of the TeT- type do not participate in ablaut, whereas roots of the TeRT- type readily become zero-grade morphemes. Only in roots of the latter type could pairs distinguished by transitivity have arisen. H. Köllnas believes that oppositions of voice (diathesis) in the preterite system—that is, the difference between the ē preterite and the ā preterite (cf. carried: turned), or the sigmatic aorist and the strong aorist—also motivated the differentiation of present stems. The sigmatic aorist, which had an effective-transitive meaning and was later replaced in Baltic languages by the ē preterite, as H. Köllnas argues, could in the present system be associated only with the je/o stem, since the e/o stem was used by infixed and sta-stem verbs and by some other intransitive verbs of the shines type. The transitive meaning became associated with the je/o present stem gradually. The je/o stem became a productive model for transitive verbs only later, when the old e/o presents were replaced by je/o presents. A prerequisite for this replacement of present stems and the formation of the model just mentioned was the existence of transitive verbs that, as can be inferred from their Slavic etymological cognates, had long had a je/o stem; cf. licks, Slavic lizą, draws, Slavic pišą, cuts, Slavic rězǫ, etc. Transitive verbs with an e/o stem, having a different shade of meaning, could for a time have coexisted with transitive verbs from the same root with a je/o stem, much like the doublets found in Old Slavic; cf. vezą and vežą, leką and lečą. Moreover, the sigmatic aorist does not always require an e/o present; it can also be used (e.g., in Greek) with je/o presents. Although ėsti, dioti, imti are not effective, their ē preterite may nevertheless be old, in H. Kölln’s view. Their Slavic equivalents have a root aorist, cf. jasrъ, dastъ, jetъ (praes. jamъ “I eat,” damb “I give,” imą “I take”). Here H. Kölln is compelled, in addition to his 254 stated regular correspondence between the Baltic ē preterite and the Slavic sigmatic aorist, to acknowledge a historical identity between the Baltic ē preterite and the Slavic root aorist as well. This assumption undermines the coherence of H. Kölln’s hypothesis, according to which Slavic strong aorists (root and thematic) correspond to Baltic ā preterites. One may question whether the ē preterites of these verbs are really old: the ā preterite found in dialects (cf. ēdo) fits better with the ineffective meaning of this verb and with the strong aorist in Slavic languages (cf. jastъ “you ate, he/she ate”). In further support of his hypothesis concerning the origin of the ē preterites, H. Kölln emphasizes that these last three verbs with an ē preterite are athematic. H. Kölln thinks that the ē preterite first arose precisely in athematic verbs. In this view, H. Kölln relies on the fact that the athematic Slavic verb jesmъ (Lithuanian esmi) has the imperfect běchъ, the only ē preterite in the Slavic languages. The verb “to be” will also have had an ē preterite in the Baltic languages, cf. Old Prussian ast, bēi, bē, be. The stative imperfect, or ē preterite, běchъ formally coincides with the aorist of stative verbs in the suffix -ēti, cf. aor. sěděchъ from inf. sěděti “to sit.” This circumstance points to the model on which ē imperfects (preterites) were formed. H. Kölln thinks the starting point was the opposition between inchoative verbs and stative verbs with a second ē stem, Slavic sěsti: sěděti, Baltic sēsti: sēdėti. Their aorists—the strong aorist sědъ “I sat down” and the ē aorist (later sigmatized) sěděchъ “I was sitting”—were probably initially preterites without aspectual status, opposed to one another like other verb forms; that is, like Belarusian sěsti: sěděti, they constituted an inchoative opposition. On the model of sědъ: sěděchъ, the verb “to be,” alongside the root aorist bychъ, 2–3 sing. bystъ, develops a stative preterite bě-chъ “I was.” Thanks to the stative present jesmъ, no special state verb in -ēti was formed here, and the opposition bychъ “factus sum”: běchъ “eram” was interpreted as an aorist: imperfect opposition. From this verb, the ē preterite spread first to other athematic verbs with a root aorist, cf. Lithuanian ėdmi, ėdė, Slavic jambъ, 2–3 sing. aor. jastъ, imperf. jaděachъ. On the basis of these verbs, the expansion of the ē imperfect (preterite) began: verbs that had an old sigmatic aorist also formed this imperfect, cf. Slavic vedą, aor. věsъ, imperf. vedě-achъ, Belarusian věda, vědě. Since these latter verbs are transitive, the suffix -ē-, used to form an imperfect alongside the sigmatic aorist, loses its association with the intransitive (ineffective) meaning characteristic of stative verbs with a second ē stem. The suffix -ē- was apparently initially an indicator of non-present forms of stative verbs in Baltic and Slavic. In this connection, H. Kölln notes that stativity consists of three components: ineffectivity, continuativity, and consecutivity. The continuative component was what caused the suffix -ē- to become an imperfect marker. As in the case of continuative verbs with a second ā stem, stative verbs with a second ē stem were themselves differentiated as part of the preterite system: alongside the aorist sěděchъ, an imperfect sěděa-chъ was formed by adding the suffix -ā-. Such is H. Kölln’s hypothesis concerning the formation and development of the Baltic and Slavic ā preterite and ē preterite (imperfect) forms. From what has been said, it is clear that H. Kölln answers in the negative the question posed at the outset: in his view, the ā preterite resp. 255 ē preterite cannot directly continue the strong resp. sigmatic aorist of the Baltic and Slavic protodialects. ā and ē preterites (imperfects), on the one hand, and the strong and sigmatic aorists, on the other, were parallel preterite formations that existed side by side and were formed from one and the same verbs. ā and ē preterites (imperfects) gradually expanded their sphere of use, taking the place of the aorists and assuming their functions. Like every hypothesis, H. Kōlln’s also has stronger and weaker points. Methodologically, H. Kōlln’s efforts to seek a historical connection between the preterite formatives ē- and ā and the corresponding verb stems (the so-called “second” stems) are well-founded. H. Kōlln’s observation of the semantic equivalence between the Baltic ā preterite and the Slavic and Greek strong aorist, and between the Baltic ē preterite and the Slavic and Greek sigmatic aorist, opens up broad prospects for research on the Baltic verb. Researchers are faced with the question of how to interpret this possible correspondence diachronically, since H. Kōlln’s proposed solution—that the Baltic dialects had a sigmatic aorist which they later lost—is problematic. Given the conservatism of the Baltic languages, the complete disappearance of the sigmatic aorist would be difficult to understand and, moreover, it would scarcely be possible to motivate and explain this loss satisfactorily. It is true that the Baltic languages showed marked tendencies toward the unification and simplification of verbal paradigms. But it is also true that individual forms from older paradigms can now exist in the Baltic languages as independent lexical units. H. Kōlln established that the sigmatic aorist in the Slavic languages has an effective meaning, whereas in the Baltic languages the element -s-, as a derivational formative, functions with stative verbs in -ėti and -oti, cf.: dubšėti (praes. dùbši alongside dùbti, dùmba) and vėpsóti (alongside vẽmpa, vėpti). Verbs with the suffix -sā- often show lengthened vocalism; thus, here the formative -s- is associated with the lengthened grade of root vocalism, as often occurs in the Slavic sigmatic aorist, cf. Belarusian *vęp-s- and Slavic aor. vẽšo ‘I led’ from *vẽd-s-. According to the prevailing view, which H. Kōlln himself also holds, the sigmatic aorist is an innovation of the Slavic languages, and consequently its effective meaning would likewise have arisen later. In that case, when seeking correspondences in the Baltic languages, it might be appropriate to consider the semantic-derivational types of other verbs as well. For example, in the case of the root bēsti, Slavic bosti, the Baltic languages have the vocalism ō in stative-inchoative verbs, cf. bodėti (praes. bódi), bóstis (praes. -džiasi) and bósti (praes. bōsta), while in the Slavic languages this vocalism appears in the sigmatic aorist, cf. Slavic basъ ‘I prodded’ from *bōd-s-. The sigmatic element in the suffix -sį- functions here in the present system just as the suffix -sk- (-sk-) did: in the Baltic languages it was originally a present formative, whereas in Armenian it functions in the preterite system. S. Karaliūnas