ACTA LINGUISTICA LITHUANICA

LIETUVIŲ KALBOTYROS KLAUSIMAI XLII (2000)

Axel HOLVOET

Lietuvių kalbos institutas, Vilnius

LITHUANIAN DATIVUS CUM INFINITIVO AND LATVIAN INFINITIVE RELATIVE CLAUSES

This article discusses the development of certain infinitive constructions in Lithuanian and Latvian, with particular attention to the Lithuanian dativus cum infinitivo construction and its genetic counterparts in Latvian¹.

Lithuanian has two variants of the dativus cum infinitivo construction expressing the purpose of an object: in one, the noun in the dative functions as the object, e.g., I bought a horse to plough the fields; in the other, it functions as the subject, e.g., I poured milk for the puppies to lap up (examples after Vytautas Ambrazas 1995). These variants, however, derive from a single construction. It seems that the dativus cum infinitivo arose at a time when the infinitive was still a nominal form, most likely the dative of an action noun [nomen actionis],² and can therefore be described as “a double dative of purpose (goal) dependent on the sentence nucleus” (Ambrazas 1995: 79). Thus, only later, as the infinitive became a purely verbal form, could a secondary relation have developed between the infinitive and the noun, as between a predicate and its dependent object or subject. Now that the infinitive has lost its original nominal character, we can speak of the noun functioning as the object or subject with respect to the infinitive. The change in the nature of the infinitive may create conditions for further development of the dativus cum infinitivo construction and, ultimately, for its disappearance. This is precisely the process that occurred in the Slavic languages, where the dativus cum infinitivo is attested but is no longer used anywhere (Ambrazas 1995: 83 cites attestations in Old Church Slavonic and Old Czech, as well as the relevant literature).

Latvian has counterparts only to the second of the aforementioned variants: as a relic phenomenon, a dativus cum infinitivo is attested in which the dative functions as the subject of the infinitive (see Endzelīns 1951: 990):

lai paliek galotnīte putniņiem uzmesties

‘so that a little treetop remains for the birds to perch on’.

¹ The author is grateful to Prof. Vytautas Ambrazas, who kindly agreed to read and comment on the first draft of this article. The author alone is responsible for any inaccuracies and errors remaining in the article.

² Cf. Stang (1966: 447–448), Ambrazas (1995: 74–75).

105

It seems that in standard Latvian and most Latvian dialects, the first variant with an object dative has no equivalent³.

The sentence mentioned has also been attested with the relative pronoun kur before the infinitival construction (see Endzelīns 1951: 990):

lai paliek galotnīte, kur putnam uzmesties ‘so that a little top remains for a bird to land on | where a bird could land’.

In this way, an infinitival relative clause is formed—a syntactic type very characteristic of Latvian, but almost unknown in Lithuanian. More precisely, Lithuanian (like the Slavic languages) uses only one variant, in which the relative clause depends on the formally unexpressed subject of an existential sentence (e.g., there is | there is not anything to eat) or on the formally unexpressed object of a possessive construction with the verb have (e.g., I have | do not have anything to eat); for these constructions, see Ambrazas et al. 1994: 670. Equivalent forms of this variant are found already in Old Church Slavonic and modern Slavic languages, cf. OCS ne imatъ česo ěsti, l. (nie) mają co jeść, and so on. In Latvian, such relative clauses can also modify nouns, e.g., Es paņēmu grāmatas, ko lasīt atpūtas laikā ‘I took books to read (| so that I could read) during my free time’. These infinitival relative clauses differ from ordinary relative clauses (in which the predicate is expressed by a finite verb form) in that they express not an actual state associated with the object denoted by the noun being modified (e.g., Grāmatas, ko es lasu ‘The books that I am reading’), but only a possibility associated with that object, the purpose of that object (Grāmatas, ko lasīt ‘books for reading, books that can be read’).

Since Latvian infinitival relative clauses have no equivalents in either Lithuanian or the Slavic languages, it may be assumed that such clauses are innovations in Latvian. Clearly, in terms of function they are very close to the dative with infinitive. Both constructions have an intermediate function between an attribute and an adverbial of purpose. Vytautas Ambrazas writes of the dative with infinitive that this construction only occasionally has a purely attributive function (e.g., in sentences such as That axe for chopping firewood has become completely blunt), whereas it is usually “associated with a verb and the noun or pronoun governed by it,” e.g., I brought poles for putting up a fence (Ambrazas 1995: 77), so its function is close to that of an adverbial of purpose. Latvian infinitival relative clauses, as relative (attributive) clauses, are formally linked to a noun, but can also be understood as adverbial clauses, cf.: paņēmu grāmatas, ko lasīt atpūtas laikā | paņēmu grāmatas, lai tās lasītu

According to the information provided by Lidija Leikuma, dativus cum infinitivo constructions with a dative functioning as an object also occur in Latgalian dialects. Until data on the nature and distribution of this construction in Latgalian dialects are published, it is difficult to assess their significance for interpreting the development of the dativus cum infinitivo construction in Latvian.

106

atpūtas laika. Latvian infinitival relative clauses belong to the type of clauses that F. R. Palmer (Palmer 1986: 178–179) calls “relative purpose clauses” (relative purpose clauses). In clauses of this type, which occupy an intermediate position between attributive (relative) clauses and adverbial clauses of purpose, verb forms characteristic of adverbial clauses of purpose are used: these are often infinitives, but conjunctives or periphrastic constructions with optative meaning also occur⁴.

There are also differences between the dative with infinitive and Latvian infinitival relative clauses: in the former, the dative can denote only the agent or object of the action, whereas the latter are not subject to such restrictions.

Since infinitival relative clauses dependent on the unexpressed subject of an existential construction (clauses of the there is something to eat type) are attested in Lithuanian and the Slavic languages, it is possible that such clauses spread further in Latvian and came to be used also to modify nouns. In this case, however, infinitival relative clauses must have taken the place of other, older constructions, and it seems quite likely that they came to be used precisely instead of infinitival constructions.

The subject of the infinitival action in Latvian infinitival relative clauses, if formally expressed, is always expressed in the dative, as in any other infinitival clause. Thus, the dative with infinitive is replaced by an infinitival relative clause without any additional morphosyntactic changes: it is sufficient to insert a relative pronoun or adverb before the infinitive (galotnīte putnam uzmesties → galotnīte kur putnam uzmesties). In cases where the dative functions as an object, the difference between the two constructions is more pronounced, since the object of the infinitive in an infinitival relative clause is expressed not by the dative but by the accusative:

Nopirku zirgu, ar ko laukus art.

Latvian infinitival relative clauses must be a fairly old phenomenon, since they probably underlie the structure of the modern debitive as well. As is known, the debitive derives from a complex sentence type with a relative clause, and the debitive prefix jā- (in some dialects ju-, derived from the original accusative or instrumental form *juo) is a trace of the relative pronoun. The construction lauks ir jāar must have developed from

⁴ In Latin, the conjunctive was used here, e.g., Scribebat... orationes, quas alii dicerent ‘He used to write speeches for others to deliver’ (Cicero, Brutus, see Palmer 1986: 178). In present-day Latvian, alongside the infinitive, the indicative with the particle lai is also used in relative purpose clauses; it is often considered a substitute for the third-person imperative, e.g., nolēma iecelt kasieri, kas lai to naudu uzglabātu (A. Deglavs) ‘they decided to appoint a cashier to keep the money safe’. The distribution of infinitives and constructions with lai in Latvian relative purpose clauses has not been investigated.

107

of the earlier structure *ir lauks juo art, where *juo corresponds to the modern relative pronoun kuo. It is true that in the modern language the prefix ją- is attached not to the infinitive form (except for the form jabūt ‘it is necessary to be’, which is based precisely on the infinitive), but to the third-person present-tense form. Not all researchers agree with the idea proposed by Prellwitz and Endzelīns that the third-person form here took the place of the earlier infinitive. Mühlenbach (Mühlenbach 1907: 322) thought that the debitive prefix was originally attached to a third-person form used impersonally, and interpreted man (ir) jubrauc as ‘I have something to ride on’. It seems that Fraenkel, too, was inclined to regard the third-person form as the original basis of the debitive, since he compares the Latvian debitive with constructions found in Lithuanian dialects that have a third-person form in place of the infinitive, such as jam nebebus, kas ėda ‘he will have nothing left to eat, he will have nothing left to eat’ (Fraenkel 1928); this suggests that he saw a genetic relationship between them. However, it has not yet been established that the above-mentioned Lithuanian dialect constructions are old. Fraenkel also cites from the dialects of the Vilnius region the construction nėr kam ažtaria rūsičios žadėlios ‘there is no one who... could intercede for me with angry words’ (Fraenkel 1936); a similar construction is also given in the Grammar of Modern Lithuanian: Nėr kam turgun važiuoja (Ambrazas et al. 1994: 671). The use of the dative here is intelligible only if we agree that the third-person form took the place of the infinitive: there is no one to go to the market, there is no one to intercede. The third-person form probably took hold here because it was used in constructions in which the relative pronoun denoted the agent of the action, e.g., nėra kas dirba ‘there is no one who works’ (cf. Ambrazas et al. 1994: 671). Thus, the Lithuanian constructions mentioned above do not provide sufficient grounds for rejecting Prellwitz and Endzelīns’s proposal that the debitive prefix jā- was originally attached to the infinitive, although Endzelīns himself seems later to have abandoned his earlier view, probably under Fraenkel’s influence (see Endzelīns 1951: 885–886; the author does not repeat his earlier view here, but instead cites constructions from Lithuanian dialects given by Fraenkel).

The reasons why the infinitive in the Latvian debitive was replaced by the third-person form have not yet been explained. It is possible that the reasons for this replacement were similar to those that led to the inclusion of the third person in the aforementioned Lithuanian constructions, but certain arguments speak against this. As mentioned earlier, the Lithuanian constructions nėra kam turgun važiuoja, nėra kam užtaria must have arisen under the influence of constructions in which the relative pronoun denoted the agent of the action (e.g., nėra kas dirba), and such constructions could not have influenced the later development of the debitive, since no variant of the debitive is based on them. First, as the forms of the prefix show (*juo, *jā), the debitive arose from constructions with an accusative, instrumental, or genitive in the relative clause; we find no traces here of the nominative or dative (corresponding to similar forms in Lithuanian constructions nėra kam užtarti / nėra kas užtaria). Second, the emergence of the particularly distinctive Lithuanian dialect construction jam nebebus, kas ėda may have been facilitated by the still very much

108

living phenomenon in Lithuanian of the nominative object: the use of nominative kas here should probably be compared with constructions of the type kas reikia daryti, kas buvo daryti (see Ambrazas 1995: 91–93). Such a nominative object could have arisen in relative clauses with an infinitive (nėra kas ėsti = nėra ką / ko ėsti), into which the third-person form was subsequently introduced (nėra kas ėda ‘nėra ko ėsti’). In Latvian, such an influence of the nominative object seems unlikely. On the other hand, the claim that the original basis of the debitive was specifically a relative clause with an infinitive can be supported by several arguments: 1) the relic form jabūt⁵; 2) the fact that even today Latvian uses infinitival relative clauses, which may be directly related to the original ones preserved in relic form within the debitive, with only the stem of the relative pronoun *yo- having been replaced by the pronominal stem *kʷo-; 3) the fact that relative clauses of purpose, to which the relative clauses discussed here belong, are more characteristically used with verb forms other than the indicative⁶; 4) the fact that in some cases relative clauses of purpose with an infinitive are attested in both Lithuanian and Slavic.

If Prellwitz and Endzelīns’s view that an infinitival construction underlies the Latvian debitive is correct, it may be assumed that Latvian infinitival relative clauses are a fairly ancient phenomenon, dating back to the prehistoric period in the development of the language. This is indicated by the archaic use of the pronominal stem *yo- as a relative pronoun in place of the later *kʷo-⁷, preserved as a relic in the debitive. However, they probably did not yet exist in Proto-Baltic: there, the infinitive

⁵ More such relic forms might be found, but since the old Latvian texts contain many morphological errors, their interpretation will remain debatable. Unclear examples include, for instance, the forms used by Elger daudze man no jums iarunat und iatēsōt (Evangelia et epistolae 19r), especially since alongside them we also find mums nenek wairak iarunna (ibid. 12r, see Evangelien und Episteln ins Lettische übersetzt von Georg Elger... Bd. 1. Texte, hrsg. von Kārlis Draviņš, Lund 1961).

⁶ V. Ambrazas also mentions constructions with present participles, which he rightly compares with dativus cum infinitivo and with the Latvian constructions from which the debitive arose: nebuvo kas bedirbęs, nėra kas valgą (see Ambrazas 1990: 115–116). In general, the use of participles in relative clauses may be regarded as an archaic phenomenon inherited from the proto-language (cf. Ambrazas 1989; 1990: 121–122), so a priori it seems not entirely plausible that participles entered the aforementioned constructions only recently, taking the place of the infinitive. One might rather expect participles to be replaced by other forms than to spread further. Nevertheless, the nominative object in the participial construction nėra kas valgo ‘there is nothing to eat’ is probably connected with the original use of the infinitive in place of the later participle, since it would be difficult to explain it otherwise than as a trace of nominativus cum infinitivo. In cases where the relative pronoun functions as the subject (nebuvo kas bedirbęs), the use of the participle appears archaic. When the third-person form was introduced in place of participles, a similar process may have taken place at the same time in infinitival relative clauses.

⁷ Although the relationship between the relative-pronoun stem *yelo- and the demonstrative-pronoun stem *i- in the Baltic languages is not entirely clear (see Brugmann 1911: 331, 347–348, Stang 1966: 233–234), there is no doubt that the relative use of the stem *ye/o- was inherited from the proto-language; consequently, the archaic character of the use of the pronominal stem preserved in the debitive is beyond doubt. A. Rosinas (1975: 169) denies the relative use of the pronoun jis, ji (which he does not distinguish from the relative *ye/o-) in Proto-Baltic, but he does not take the origin of the Latvian debitive into account.

109

was most likely still a form of nominal character. It is precisely the Lithuanian dativus cum infinitivo that attests to the infinitive’s original nominal character.

The aforementioned structural type *ir lauks juo art gradually acquired a modal function: in the present-day language, lauks ir jaur simply means ‘the field needs to be ploughed’. In some dialects, however, the original meaning of these constructions has partly survived; this is illustrated by the sentence variant mentioned earlier, with a debitive used instead of an infinitive (see Endzelīns 1901: 73 = 1971: 149–150):

lai stāvēja virsūnīte putniņiem jāuzmetas.

Labai įdomus yra Mühlenbacho (1907: 316–317) iš Vietalvos tarmės cituojami panašaus debityvo vartojimo pavyzdžiai. Šioje tarmėje įvyko dviejų debityvo variantų diferenciacija: vienas iš jų, su priešdėlio forma jā-, turi dabartinės latvių bendrinės kalbos debityvo modalinę funkciją:

man nau jābrauc ‘I must not go’.

The second variant, with the prefix form ju-, has retained the original function of an infinitival relative clause. In the examples cited by Mühlenbach, the prefix ju- functions as an instrumental:

man nau jābrauc ‘I have nothing to travel by’; brāl’am nau zirga jūecē ‘my brother has no horse with which to harrow’; ienas ūdeni jumazgājas ‘bring water for washing’; būtu mums bijs pūlvers jušain ‘if only we had had gunpowder to shoot’.

This function is also compatible with the prefix form, since the prefix ju-, shortened in proclitic position from *juo, can be interpreted either as an accusative or an instrumental (for the accusative form *juo, formally corresponding to the Lithuanian feminine form ją, see Stang 1966: 437, Endzelīns 1951: 518). Among the examples cited by Mühlenbach, we find none in which the prefix ju- can be understood as an accusative—perhaps because the object accusative was displaced by the genitive, which spread from negative constructions (for the generalization of the prefix form jā-, cf. Holvoet’s article submitted for publication). The variant with a relative pronoun functioning as the object became the basis for the debitive as a modal construction: this is also shown by the example from the Vietalva dialect man nau jābrauc ‘I must not go’. In this variant, the modal meaning eventually displaced the original meaning, which still survives in the construction found in old Latvian texts teems ne bij ja-ehd ‘they had nothing to eat’ (Glück’s Bible, Mark VIII.1, see Mühlenbach 1907: 319). The variant attested in the Vietalva dialect with the pronoun in the instrumental, which later became the prefix *juo- > ju-, retained its original meaning. It is precisely this variant that can be compared with the Lithuanian dativus cum infinitivo variant, in which the dative denotes

110

the object of the verb. Since the prefix *juo- > ju- here is an instrumental, not an object accusative, the object position in this adnominal debitive construction remains unoccupied and can be filled by a noun in the accusative:

panām kač pavadu zirgu jusapin ‘take at least a bridle with which to hobble the horse’ Lithuanian: take at least a bridle to hobble the horse with.

Panašų pavyzdį (iš dainos teksto) cituoja Endzelynas (1901: 73 =1971: 149; 1951: 973) iš kitų tarmių (čia apibendrinta priešdėlio forma jā-):

devu savu kamelinu ruožu dārzu jāecē Lithuanian: I gave my mare to harrow the rose garden; lai stāv [nauda]... guovi jānopeŗk Lithuanian: let there be money to buy a cow.

Similar debitives with the prefix jā- in an instrumental sense are also attested in Old Latvian texts, such as Elger’s “Gospels and Epistles”:

vnd kad tems ne by iamākla, abbems pedewwe ‘non habentibus illis, unde redderent, donavit utrisque’ (64v = Lk 7, 42).

It seems quite likely that constructions with a relative pronoun in the instrumental were initially one of the principal variants of infinitival relative clauses; this also explains their grammaticalization, that is, the emergence of a particularly agglutinative grammatical form through the fusion of the pronoun and the verb. This type is probably genetically equivalent to the Lithuanian dativus cum infinitivo. Both constructions developed from Proto-language constructions with an infinitive expressing purpose, which in Latvian were altered by inserting the relative pronoun in the instrumental. The Lithuanian construction with a bare purpose infinitive (without a subject or object) corresponds in the Vietalva dialect to a debitive with the prefix ju-: the brother has no horse to harrow with = *brāl’am nav zirga juo ecēt > brāl’am nau zirga juecē. Here, the Proto-language infinitival construction was altered by inserting a relative pronoun; as in constructions with a noun functioning as the subject of the infinitive, no further changes were needed. The Lithuanian infinitive with a dative functioning as its object corresponds to the Vietalva variant of the ju- debitive with an object in the accusative: take a bridle to hobble the horse with = *paņem pavadu zirgu juo sapīt > paņem pavadu zirgu jusapin. Here, it was necessary not only to insert a pronoun, but also to change the case, since the noun became the object of the infinitive in the relative clause. It is clear that the absence in Latvian of a dativus cum infinitivo construction with a dative object is connected precisely with this process; what remains unclear is at which stage of the process the object accusative was introduced. There are two possibilities: 1) the object accusative was introduced into infinitival relative clauses and only

111

Later, infinitival relative clauses ousted the old dativus cum infinitivo (the connection between the emergence of infinitival relative clauses and the disappearance of the dativus cum infinitivo would in this case be indirect); or 2) the relative pronoun was incorporated directly into the dativus cum infinitivo construction, and only later was the dative replaced by the accusative. The position of the inserted pronoun provides strong evidence for this latter assumption: in constructions with an object, the pronoun could probably become attached to the verb only if the original word order was *panem pavadu zirgu juo sapīt, and not *panem pavadu juo zirgu sapīt. If the word order attested in the examples mentioned above is not accidental and the object (zirgu, ruožu dārzu, guovi) always stood before the pronoun-and-infinitive complex, then the later object of the infinitive must originally have been something other than an object; rather, it was dependent on the same noun with which the infinitive, before the pronoun was inserted, was also directly associated (as a dative of purpose or destination). In that case, it could have acquired its present case form (the accusative) only later, with the emergence of the infinitival relative clause, while its original case form must have been different: most likely, it was the dative. This means that the construction panem pavadu zirgu jusapin may have developed directly from the dativus cum infinitivo construction.

Considering the evidence from Lithuanian and Slavic, it is a priori plausible that both variants of the dativus cum infinitivo already existed in Proto-Baltic and were later replaced by other constructions in Latvian. It seems that Latvian data itself also allow us to trace the footprints of this construction in the development of Latvian infinitival constructions.

Received 1999 05 25

LITERATŪRA

A m b r a z a s V. 1989: Baltic languages and the reconstruction of the Indo-European complex sentence. – General Linguistics 29, 52–59.

A m b r a z a s V. 1990: Sravnitel’nyj sintaksis pričastij baltijskich jazykov / Vergleichende Syntax der baltischen Sprachen, Vilnius: Mokslas.

A m b r a z a s V. 1995: Lietuvių kalbos bendraties konstrukcijų raida. – Lietuvių kalbotyros klausi- mai 33, 74–109.

A m b r a z a s V. ir kt. 1994: Dabartinės lietuvių kalbos gramatika, Vilnius: Mokslo ir enciklopedijų leidykla.

B r u g m a n n K. 1911: Grundriss der vergleichenden Grammatik der indogermanischen Sprachen, Bd. II, 2. T., Strassburg: Karl J. Trübner.

E n d z e l ī n s [Endzelin] J. 1901/1971: Ursprung und Gebrauch des lettischen D जैbitivs. – Be- zenbergers Beiträge 26, 66–74. (Perspausdinta: idem, Darbu izlase 1, Rīga: Zinātne, 143–150.)

E n d z e l ī n s [Endzelin] J. 1905/1971: Zur Entstehung des lettischen D जैbitivs. – Bezenbergers Beiträge 29, 320–321. (Perspausdinta: idem, Darbu izlase 1, Rīga: Zinātne, 288–289.)

E n d z e l ī n s J. 1951: Latviešu valodas gramatika, Rīga: Latvijas Valsts izdevniecība.

112

Fraenkel E. 1928: Litauische Beiträge. – Indogermanische Forschungen 46, 44–57.

Fraenkel E. 1936: Der Stand der Erforschung des im Wilnagebiete gesprochenen Litauischen. – Balticoslava 2, 14–107.

Holvoet A. (atiduota spaudai) On the rise and grammaticalisation of the Latvian debitive. – Linguistica Baltica.

Mühlenbach [Mjulenbachъ] K. 1907: O debitivė. – Izvĕstija Otdĕlenija russkago jazyka i slovesnosti Imperatorskoj Akademii Nauk 12/3, 313–333.

Palmer F. R. 1986: Mood and modality, Cambridge: University Press.

Prellwitz W. 1904: Zur Entstehung des lettischen Debitivs. – Bezenbergers Beiträge 28, 319.

Rosinas A. 1975: Ar baltų *i-, *io- resp. *i, *iā- kamienai buvo reliatyviniai? – Baltistica 11 (2), 165–170.

Stang Chr. S. 1966: Vergleichende Grammatik der baltischen Sprachen, Oslo etc.: Universitet-sforlaget.

THE LITHUANIAN DATIVUS CUM INFINITIVO AND THE LATVIAN NON-FINITE RELATIVE CLAUSES

Summary

The dativus cum infinitivo, an archaic infinitive construction which probably arose from a double dative of purpose (the dative being originally the dative of a verbal noun), is well retained in Lithuanian but has hardly left any traces in Latvian. Of the two varieties attested in Lithuanian, with the dative as subject of the infinitive (įpyliau pieno šuniukams užlaki ‘I poured out some milk for the dogs to lap’) and with the dative as its object (nupirkau arklių laukams arti ‘I bought a horse to plough the fields with’) only the former is vestigially attested in Latvian. In the paper an attempt is made to bring this fact in connection with the development of non-finite relative clauses in Latvian. In Lithuanian (as in Slavonic) the occurrence of such clauses is restricted to but a few types, and in Latvian their scope was probably extended as relative pronouns were inserted in constructions with infinitives expressing purpose. In reconstructing this process, evidence may be drawn from some dialectal varieties of the debitive, which, in a fossilised shape, represents an original non-finite relative clause type with the relative pronominal stem *ye/o-. In the dialect of Vietalva an adnominal variety of the debitive with the prefix ju- (an original instrumental) is attested, in which an object may occur (paņam kač pavadu zirgu jusapin ‘take with you a lead to shackle the horse with’). This seems to correspond exactly to the Lithuanian dativus cum infinitivo with the dative as an object (pasimk pavadį arkliui supančioti), not only functionally, but also genetically. Though the original dative was replaced with an accusative as the original infinitival construction developed into an infinitival relative clause, a direct connection with the dativus cum infinitivo is suggested, among other things, by word order, as the incorporation of the relative pronoun into the verbal forms (*juo sapīt > jusapin) suggests an original word order *paņem pavadu zirgu juo sapīt, where the second noun, before being reinterpreted as an object in the relative clause, must have been in adnominal position, and must therefore have been in the dative rather than in the accusative.

113