Acta Linguistica Lithuanica LIX (2008), 57 – 80

On the Expression of Evidentiality in Baltic: the Lithuanian Particle nevà

For Marianne Giedra

Daniel Petit

École Normale Supérieure, Paris

Evidentiality is here understood as the linguistic marking of the speaker’s distance to the contents of his utterance. In Lithuanian, evidentiality can be expressed synthetically by means of the so-called relative mood, consisting in the use of participles instead of finite verb forms, or analytically by means of adverbs or particles. This paper deals with the origin of one of these particles, Lith. nevà ‘allegedly’. It is not attested before the 19th century and seems to consist of the negative nè and the presentative particle và. There are typological difficulties, however: evidentiality is never expressed by a bare negation; and presentative particles (like Lat. ecce, Russ. вот, French voici, voilà) are known to be non-negatable. This suggests that nè in nevà cannot have had its basic negative meaning. We could therefore assume it to have been used in its comparative function (‘as, like’), which has left numerous relics in Baltic and could go back to Indo-European (cf. OInd. ná ‘not’ and ‘as, like’).

I. INTRODUCTION¹

Recently, the concept of evidentiality has been given great importance in the literature on linguistic typology². By evidentiality

¹ Paper presented at the XIV Conference of the Indo-European Society (Salzburg, September 22, 2008). I would like to thank Björn Wiemer, Georges-Jean Pinault, and Axel Holvoet for discussion and comments.

² Während die englischsprachigen Sprachwissenschaftler (z.B. Aikhenvald und Dixon 2003, Aikhenvald 2004) jetzt weitgehend die Bezeichnung evidential, bzw. evidentiality verwenden, hat sich in der französischen Linguistik zumindest seit Lazard (1956: 149, Fußnote 2) und neuerdings Guentcheva (1996, 2007) die Bezeichnung médiatif, bzw. médiativité durchgesetzt – explizit mit dem Streben, der Spezifizität des Worts

57

Daniel Petit

one understands all linguistic means that express information about the speaker’s stance toward the content of an utterance. Evidential features show how the speaker came to know about an event, whether they experienced it directly or not, and, finally, whether or not they guarantee its truth. According to Guentcheva, evidentiality in the strict sense of the word can in principle be defined as non-appropriation of the informational content³. Evidential expression strategies can vary from one language to another. There are languages in which evidentiality can constitute a very complex category. For example, in the language of the Tuyuca (Brazil, Colombia), up to five sentences can be regularly formed, depending on the source of information and on the speaker’s stance toward the truth value of the utterance⁴:

(1a) Díiga apé-wi “He played football [I saw him play]”

(1b) Díiga apé-ti “He played football [I heard him play, but did not see him]”

(1c) Díiga apé-yi “He played football [I saw evidence showing that he played football, e.g. his footprints on the field]”

(1d) Díiga apé-yigi “He played football [I got the information from someone]”

(1f) Díiga apé-hỹi „Er hat Fußball gespielt [die Vermutung liegt nahe]‘

Evidentiality is more often expressed within a binary contrast, in which a distinction is made between external information from which the speaker dis-

évidence in French, as opposed to English (cf. Guentcheva, 1996: 13, Lazard, 2000: 209-210). Other terms have also been proposed, e.g. médiaphorique by Hagège (1995: 15). Terms such as testimonial, indirective, distancive, imperceptive, quotative, etc., are also occasionally encountered. Johanson (2000: 61) uses the term indirective.

³ Guentcheva (1996: 13): „non-prise en charge du contenu informationnel“. Vgl. auch Joseph (2003: 97): „ Evidentiality can be defined as the indication of the source of a speaker’s information, of the modality by which that information was gained, and/or of the speaker’s stance (i.e., attitude) towards the truth of the information.“

⁴ Klassisches Beispiel nach Barnes (1984: 255–271), s. auch Guentcheva (1996: 13). Ein anderes Beispiel ist Fasu, eine Papuasprache, mit 6 verschiedenen Möglichkeiten (vgl. Joseph, 2003: 98).

58

Zum Ausdruck der Evidentialität im Baltischen

is distinguished from internal information, which the speaker accepts as true. A well-known example is the Oblique Mood of the Finnic languages (Finnish, Estonian, Livonian), according to which participial forms used predicatively express an evidential meaning, i.e. the speaker’s distance, e.g. Estonian Ma lugevat ‘Apparently I read; they say that I read; it is said that I read’ (with the participial ending -vat⁵).

The number of studies in linguistic theory and linguistic typology on evidentiality and its marking has increased considerably in recent times⁶. It is striking, however, that studies oriented toward Indo-European linguistics have, with a few exceptions⁷, so far paid almost no attention to the question of evidentiality—perhaps simply because reconstructible Indo-European was considered a language in which evidentiality played no role, and evidential formations developed only in later stages of the individual Indo-European languages. As is well known, evidentiality as a morphosemantic category is found in only a few modern languages, particularly Albanian, Bulgarian, New Persian, New Armenian, and Lithuanian. It is probably reasonable to assume that the category of evidentiality was foreign to the Proto-Indo-European parent language. The claim that the Proto-Indo-European injunctive should be assigned an evidential basic meaning, as Joseph recently (2003) proposed, is, in my view, supported by too little evidence.

This does not mean, however, that the question of evidentiality is of no interest to Indo-Europeanists. For one of the main tasks of modern Indo-European linguistics is to investigate how inherited linguistic structures led to the state of the historical individual languages. The question of how evidentiality can arise is a topic that has so far been almost unexplored in Indo-European linguistics.

From a typological perspective, one must generally reckon with two important means of expressing evidentiality. Either evidentiality is synthe-

⁵ Cf. Hasselblatt (²2001: 135).

⁶ Z.B. Chafe und Nichols, Hrsg. (1986), Guentcheva, Hrsg. (1996, 2007), Johanson und Utas, Hrsg. (2000), Aikhenvald und Dixon, Hrsg. (2003), Aikhenvald (2004).

⁷ Vgl. Gren-Eklund (2000), Joseph (2003). Vgl. auch eine allgemeine Darstellung bei Ramat (1996).

59

Daniel Petit

tic in nature and is marked in the verbal form itself, usually by original perfect forms, as, for example, in Bulgarian (e.g. Иван спял ‘Ivan is apparently sleeping’⁸) or New Armenian (e.g. Gac‘er em ‘Apparently I have gone’⁹; Albanian, as is well known, has a specific mood, the so-called admirative, which can also have an evidential meaning (e.g. Punúakam ‘Apparently I have worked’¹⁰). Alternatively, evidentiality is expressed analytically by adding adverbial modifiers. This is the case in many modern languages, e.g. German (angeblich, vorgeblich), French (apparemment, prétendument), English (like), Russian (будто), Polish (rzekomo), Breton (hervez), New Armenian (eyer), etc. More complex means of expression are rare; one example is Irish is cosúil ‘apparently’ (< ‘there is an appearance’).

As is well known, the category of evidentiality plays an important role in the Baltic languages, particularly Lithuanian. Lithuanian can express the speaker’s distance from the content of a statement in two ways: either synthetically, through the so-called Relative Mood, or analytically, through adverbial forms. The Relative Mood, as is well known, involves the use of participial forms as predicates with an evidential meaning. A simple example:

(2a) Indicative, when the speaker considers the truth of the statement to be certain: Danutė dirba mokykloje ‘Danutė works at a school.’

(2b) Participle, when the speaker does not consider the truth of the statement certain, either because it seems doubtful to them or because they know the information not firsthand but from an external source. Danutė dirbanti mokykloje ‘Danutė apparently works at a school.’

⁸ Beispiel bei Guentcheva (1996: 54).

⁹ Beispiel bei Donabédian (1996: 90).

¹⁰ Beispiel bei Duchet und Pernaska (1996: 31–46).

60

Zum Ausdruck der Evidentialität im Baltischen

The literature on the Relativus mood in Lithuanian is extensive and difficult to survey. Here I refer only to the important monograph by Axel Holvoet, Mood and Modality in Baltic (2007), which devotes a chapter to the Relativus mood (pp. 81–127)¹¹. It is often assumed that the Relativus mood arose through contact with the Finnic languages of the Baltic Sea region. From a linguistic-typological perspective, however, it should be noted that evidentiality often derives from original perfect formations, which would also be a plausible source for the participial forms of Baltic, insofar as they play an important role in the development of periphrastic perfects. An internal interpretation is therefore entirely possible.

In addition to the Relativus mood, Lithuanian has a whole range of adverbs, or particles, that can carry evidential meaning. So far, they have been collected and examined only in outline, in a monograph by Roman Roszko (1993) and an important article by Björn Wiemer (2007). The question of their origin, however, has generally not been asked. The subject of the present study is the origin of just such an evidential particle, namely Lithuanian nevà ‘supposedly’.

2. THE PARTICLE nevà: PHILOLOGICAL EVIDENCE

Very little has been written so far about the origin of the Lithuanian particle nevà ‘supposedly’. It is treated only briefly by Fraenkel (LEW₁ 498); it is entirely absent from the recently published etymological dictionary of the Lithuanian language by Wojciech Smoczyński (2007). Even the philological evidence remains little studied. As far as I can tell, this particle is not attested in Old Lithuanian written sources. The first attestations date from the 18th century. The word nevà ‘supposedly’ appears for the first time in the handwritten dictionary of Jakob Brodowski (1713–1744). At the beginning of the 19th century, it is illustrated by the following example in the dictionary of Christian Gottlieb Mielcke (1800: 564):

(3) Christian Gottlieb Mielcke, Littauisch-deutsches und Deutsch-littauisches Wörter-Buch (1800):

Neva būk nekaltas esas

‘as if he were innocent [supposedly innocent]’

¹¹ Aus der älteren Literatur s. noch Ambrazas (1977). S. auch Wiemer (2006).

61

DANIEL PETIT

In the 19th century, the word slowly begins to appear in literature, interestingly at first only in informal usage or in less formal literary genres. The use of nevà in collections of folktales is characteristic, for example; these often draw on spoken dialect variants. For instance, in Lietuviškos Pasakos by Jonas Basanavičius (1898: I, 178)¹² we find:

(4) Jonas Basanavičius, Lietuviškos Pasakos (1898):

And the young man pretended to be drunk, lying down on the bed; he pulled on those trousers and pretended to sleep.

‘And the young man pretended to be drunk, lying in bed; he pulled on his trousers and pretended to be asleep.’

In the 20th century, the word is attested more frequently in literary works, especially in novels and novellas that seek to reproduce spoken language, for example in writers such as Mykolaitis-Putinas, Sruoga, and Tumas-Vaižgantas. For instance, in Juozas Tumas-Vaižgantas (1869–1933)¹³:

(5) Juozas Tumas-Vaižgantas (1869–1933):

Saviški, tušti Aleksio įrodymai labiau įtikindavo Jokūbą, negu labai išmintingi ir neva mokslingi privedžiojimai.

‘Aleksis’s peculiar, empty arguments convinced Jakob more than the very sensible and supposedly scientific rationalizations.’

Another literary genre in which the use of nevà seems to be preferred is private literature, such as personal letters or memoirs, characterized by a simple style close to spoken language. The particle nevà is found, for example, in the autobiography of Vilius Gaigalaitis (1870–1945), a Lithuanian from East Prussia¹⁴:

(6) Vilius Gaigalaitis (1870-1945):

¹² Similar usage in Lithuanian Folk Tales and Stories by Christoph Jurkschat (1898: 124). Cf. LKŽ_VIII 749.

¹³ Quoted from LKŽ_VIII 749.

¹⁴ S. Kaukiene (1997: 323).

62

Zum Ausdruck der Evidentialität im Baltischen

Buvome į blogą garsą įklampyti, neva neištikimi esanti? ‘Did we acquire a bad reputation, as if we were unfaithful?’

Characteristic is the fact that until the 1930s the particle nevà was not recommended by purists, or was even rejected as a barbarism. In 1911, the famous Lithuanian grammarian Jonas Jablonskis wrote in a letter that the word nevà in the meaning ‘like’ (= lýg) was incorrect¹⁵. In B. Sereiskis’s Lithuanian-Russian dictionary (1933: 511), the particle nevà is rendered by Russian будто, как-будто, будто-бы ‘allegedly’ and illustrated with an example:

(7) B. Sereiskis, Lietuviškai-rusiškas žodynas (1933): Jie sako, neva aš sutrukdžiau visą dalyką. ‘They say that I allegedly disrupted the whole thing.’

Sereiskis explicitly emphasizes, however, that nevà with the relative mood in the meaning ‘that... allegedly...’ is less correct than the standard construction kàd (+ relative mood). A few years later (1935), the grammarian Juozas Balčikonis criticized the writer Vincas Mykolaitis-Putinas in a review for using nevà in a sentence in his novel Altorių šešėly¹⁶. According to Balčikonis, nevà is a Slavism; he repeated the same opinion several times (cf. RR₁ 148 and 331). Only more recent dictionaries record the word as neutral. The comprehensive dictionary by Niedermann, Senn, Salys, and Brender (1951: II 180) contains numerous attestations, e.g.:

(8) Niedermann, Senn, Salys, and Brender (WISⅡ 180): žinios neva apie einančias derybas ‘news about allegedly ongoing negotiations’ (9) Niedermann, Senn, Salys, and Brender (WISⅡ 180): vaistai, kurie neva palengvina kentėjimą ‘remedies that supposedly alleviate suffering’

In the standard dictionary of the Lithuanian language (Dabartinės lietuvių kalbos žodynas, DŽ ²2000: 423), nevà is explained by tarsi, lýg ‘so to speak, like’

¹⁵ Vgl. Piročkinas (1986: 164). S. auch Jablonskio laiškai, hrsg. von A. Piročkinas, Vilnius (1985: 28, 29).

¹⁶ Mykolaitis-Putinas’s sentence reads: jausdamasis neva kaltas ‘feeling as if guilty’ (Altorių šešėly, I 233, 6).

63

DANIEL PETIT

and illustrated by two examples, of which I would like to cite only one here:

(10) Dictionary of Contemporary Lithuanian, DŽ ⁴2000 (cf. LKŽ_VIII 749):

Atrodo, neva serga.

‘It seems as though he were ill.’

In der heutigen litauischen Sprache ist der Gebrauch von nevà üblich. Roszko (1993: 53) und Wiemer (2007) führen mehrere Beispiele insbesondere aus der Presse an. Als charakteristisches Beispiel kann man folgendes vorbringen:

(11) Roszko (1993: 54):

The other astronauts who landed on the Moon laugh at stories in which they supposedly met aliens from other worlds.

‘Other astronauts, having landed on the Moon, laugh at the stories in which they supposedly were to have met inhabitants of other worlds.’¹⁷

It is no surprise, after all, that the particle nevà is fairly well attested in Lithuanian dialects, since dialect recordings, most of which were made from the 1950s onward, provide a true-to-life representation of spoken language. In the so-called “Kapsas” or “Zanavykas” dialects, which belong to the western Highland Lithuanian area (near the river Novà), the particle nèva (nèva) is frequently used. Since the relative mood has almost entirely disappeared in this dialect, nèva (nèva) is usually combined with indicative forms. Its meaning is often evidential, as the following example shows:

(12) Zanavykas dialect of Griškabūdis (ZŠŽ_II, 270):

Apparently, the woman is really looking at the cow grazing.

‘Supposedly the soldiers are asking for something to drink, but in fact they are looking at the girls.’

The contrast between nèva ‘supposedly’ and iš tikro ‘in fact’ is noteworthy

¹⁷ Cf. Roszko’s Polish translation: Inni astronauci, którzy wylądowali na Księżycu, śmieją się z opowieści, w których niby się spotkali z przybyszami z innych światów.

64

Zum Ausdruck der Evidentialität im Baltischen

However, it is striking that næva (nėva) in the same dialect sometimes has a more general meaning (‘so to speak, like’) without an evidential nuance, e.g.:

(13) Zanavykian dialect of Griškabudis (ZŠŽ_II 270):

Jiz dė:ši[m] mǟ:tu: nėva gā:vo:.

‘He reached approximately [< so to speak, like] ten years.’

In the same dialect, a variant nėvago·s (nėvagos) is occasionally also found¹⁸, e.g.:

(14) Zanavykian dialect of Griškabudis (ZŠŽ_II 270):

Nėvago·z bān̄do· stūba· pasistatīt.

‘Apparently, he is trying to build a house.’

Further examples (see LKŽ_VIII 749) show a clear predominance of dialectal sources.

There is therefore reason to assume that the particle nevà ‘allegedly’ is a popular word that was integrated only late into the written language, especially into writings oriented toward language standardization or purism. This may also explain why, as far as I know, the word is never attested in Old Lithuanian literature: this literature, which is almost exclusively religious in nature, belongs to an elevated style. It is striking that nevà was rejected by a certain purist tradition as incorrect or even as a Slavicism. Anyone concerned with the origin of the particle nevà must take this peculiarity into account.

3. ORIGIN OF THE PARTICLE nevà

Before attempting to answer the question of the origin of the particle nevà, I would like to point out a peculiarity of its use. Nevà can have two functions in a sentence: it can function as an adverb (15a = English allegedly, Polish rzekomo) or as a conjunction (15b = English that...allegedly, Polish jakoby). Here are two examples from Roszko’s monograph (1993: 53):

¹⁸ For further attestations, see Senkus (KD 130) and LKŽ_VIII 749,

65

DANIEL PETTIT

(15a) Roszko (1993: 53): Statistics from that period are presented, which are allegedly meant to show how enthusiastically Lithuania voted. ‘A statistic from that time is presented, which is allegedly supposed to show how enthusiastically Lithuania voted.’ ¹⁹

(15b) Roszko (1993: 53): Rumors began to spread that the M. Mažvydas National Library would allegedly not be operating. ‘Rumors began to circulate that the M. Mažvydas National Library would not be operating.’ ²⁰

In my view, this ambivalence can be explained by the reanalysis of originally paratactic constructions. Example 10 (atrodo, neva serga ‘it seems as if he were ill’) indeed leaves both interpretations open: either nevà is an adverb in an asyndetic construction (‘it seems, allegedly he is ill’) or a subordinate-clause conjunction (‘it seems that he is ill’). It was precisely in contexts like these that the use of nevà as a conjunction could arise. Originally, nevà was probably only an adverb.

As for the origin of nevà, some facts must first be taken into consideration.

First, the stress. In modern Lithuanian, stress on the final syllable is regular (nevà), but a dialectal variant with initial stress (nèva) is also known. This variation is very striking. Such accentual variation in a grammatically functional, indeclinable word is difficult to understand. Moreover, the variant nèva contradicts a sound law characteristic of Lithuanian, according to which the vowel e (as well as a) is lengthened in an open medial syllable and receives the circumflex accent: one would expect *nèva (as, for example, in Lithuanian kèvalas ‘eggshell’, klèvas ‘maple’, drẽvę accusative ‘hollow in a tree trunk’, etc.), not nèva with a grave accent. As is well known, in Lithuanian the grave accent marks stressed short vowels. Evidently, the expected lengthening did not occur here.

Another problem is the fact that a sequence *-ev- before a back-

¹⁹ Roszko: Statistics from that period are cited, which were supposedly meant to show how enthusiastically Lithuania was to vote.

²⁰ Roszko: Rumors began to spread that the M. Mažvydas National Library was supposedly not going to operate.

66

Zum Ausdruck der Evidentialität im Baltischen

vowel regularly becomes *-av- in Lithuanian; *-ev- is preserved only before a front vowel. Accordingly, we find, for example, sāvas ‘his, own’ (< PIE *seu̯-o-, Greek ἐός) alongside devyni ‘nine’ (< *dev-in- with secondary lengthening to *dev-yn- after aštuoni ‘eight’ < PIE *h₁neṷ-n̥, Latin nouem). The preservation of the sequence *-ev- before the back vowel *-a in nevà calls for an explanation²¹.

All these problems can be resolved if we assume that nevà is composed of two originally independent words, *nè + *và. Only this hypothesis explains why the vowel è was not lengthened within the word: it was originally not within the word, but at the end of an independent word (*nè), and the lengthening in question did not occur in this position. This assumption also explains why the sequence *ev did not become *av: the univerbation of *nè and *và must be later than the sound change *ev > *av before a back vowel. Finally, the accentual variation (nevà / nèva) is entirely understandable on the assumption that there are two originally independent words here; it reflects either the stress of the first element or that of the second. Nevà should therefore probably be understood as the late univerbation of two independent words, *nè + *và.

In view of these facts, it is reasonable to assume that the first element of nevà continues the Indo-European negation *nè. It is noteworthy that some of the peculiarities of the word nevà described above have parallels in negation: negation, too, shows no lengthening of the vowel e before (e.g. nèkalba ‘he does not speak’); the sound change *-ev- > *-av- is also alien to it (e.g. neválgo ‘he does not eat’); finally, the accent is similarly mobile (e.g. nèkalba / neválgo). That negation was originally independent in this structure is by no means surprising; regular univerbation (e.g. Lithuanian nesāko ‘he does not say’) is secondary²². The question now is what role a negation could have played in the emergence of an evidential particle. Here typological difficulties begin.

Lithuanian has other particles that have an evidential meaning

²¹ A variant, navà, is very rare.

²² In the prehistory of the Baltic languages, negation was still an autonomous word. This is attested by traces of tmesis in Old Prussian (Old Prussian neggi “still, and not” < *ne-gi + verb, III 31₄, 33₃, 45₉, 109₄) as well as in Lithuanian (e.g. nesidúoda “he does not give himself” < *ne-si-dúoda).

67

DANIEL PETIT

having²³. Semantically, these particles can be classified into several subgroups. They may be verbal forms, especially from the semantic field of speaking, such as sāko (< ‘one says’, from sakyti ‘to say’), tarši (< ‘you will say’, from tārti ‘to say’), tārtum or tarýtum (< ‘you would say’, also from tārti ‘to say’). They may also be based on the verb of existence būti ‘to be’, in the relative mood (esą, active participle, neuter) or in the imperative (būk, with the variant būktai, perhaps an imitation of Russian будто). Evidentiality can finally be expressed by comparative particles, for example by lýg ‘like’ (with the variant lýg-tai). Apart from nevà, the most widely used particle is now esą. The difference between the two particles seems to be that nevà expresses greater distance on the part of the speaker than esą. This question requires more thorough investigation. The same categories can also be found in other Indo-European languages in which evidentiality functions as a marked category. Evidential particles are often based on original verbal forms meaning ‘speak, say’, such as German angeblich (< angeben), French prétendument (< prétendre), Polish rzeko-mo (< rzec ‘to speak’), Hittite -ua(r)- (< *-uorh₁-, cf. Greek εἴρω ‘to speak’)²⁴. Just as often, evidential particles have developed from original comparative particles, such as English like (< ‘as’), Russian якобы (< jako ‘as’ + conditional by). Sometimes evidential particles derive from verbs of existence (cf. Russian будто < imperative singular, second person *bǫdi + -to) or from verbs denoting seeming or appearing (cf. French apparemment, Irish is cosúil ‘there is an appearance’). In Breton, hervez ‘allegedly’ is originally a preposition ‘according to, in accordance with’, indicating the indirect source of the information²⁵. Comparison with non-Indo-European languages confirms the same tendencies in the development of evidential particles.

It is striking, however, that no evidential particle is based on a negative formation. And this is probably no coincidence. Evidentiality expresses the various degrees of distance that may separate the speaker from what is said. It does not necessarily, or not in the first

²³ Zu den evidentialen Partikeln des Litauischen s. grundsätzlich Wiemer (2007).

²⁴ Zum Hethitischen s. Kloekhorst (2008: 958).

²⁵ Cf. Hemon (1975: 94): ‘according to’.

68

Zum Ausdruck der Evidentialität im Baltischen

In line, it presupposes a negation of reality. From a typological perspective, it is clear that evidentiality is nowhere understood as a negation of immediate experience.

A second typological difficulty is connected with the interpretation of the second element -và. Assuming that it is an originally autonomous word, it is tempting to suppose that it is identical to the deictic particle và ‘there! here! look there! look!’ This particle, which is used quite frequently in spoken language, is characterized by its colloquial character. It is entirely absent from Old Lithuanian literature; its earliest attestations date from the end of the 19th century (see Büga, RR_I 328, RR_II 47). In terms of its function, it recalls forms classified in other languages under the cover term presentative particles, e.g. Russian вот, Latin ecce, French voici, voilà, etc. The function of such forms is to present, in a strongly deictic manner, a reality that is immediately before one’s eyes. In most cases, và can be translated by the French presentative voici or voilà. Examples:

(16a) Jan Otrębski, GJL_III 371: Và, ką aš radau. ‘Look what I’ve found!’ (French ‘Voici ce que j’ai trouvé!’)

(16b) Highland Lithuanian dialect of Pivašiūnai (see LKŽ_XVII 763): Dabar và laiko y[ra]. ‘Now there really is time’ (French ‘Voilà, il y a du temps.’)

(16c) Highland Lithuanian dialect of Pivašiūnai (see LKŽ_XVII 763): Aš và pinigų turiu. ‘Look, I have money!’ (French ‘Voilà, j’ai de l’argent.’)

A variant vã (with lengthening) is attested less frequently, e.g.:

(17) Highland Lithuanian dialect of Miežiškiai (see LKŽ_XVII 763): Vã keturi arkliai – pasigauk kątrą nori. ‘Look, four horses—take whichever one you want’ (French ‘Voici quatre chevaux – prends celui que tu veux!’)

Assuming that an etymological connection exists between và and nevà, it should be noted that both forms are characterized by the same colloquial nature. Both are attested late, initially only in colloquial texts or dialect recordings, and both belong

69

Daniel Pettit

evidently belongs to spoken language. I suspect that the particle và is responsible for this, and that nevà owes this very character to it.

The fact that nevà was regarded by some Lithuanian grammarians as a Slavism is probably connected with the mistaken view that và was borrowed from Russian meaning ‘look there! look!’. This assumption, however, is unfounded. Büga (RR_II 46, RR_III 826) rightly considers và a purely Lithuanian word (savas žodis). A borrowing from Russian would have yielded not và but vàt. This form, vàt, does indeed occur in some Lithuanian dialects²⁶, but it must not be identified with the particle và. At most, one could assume that và and vàt came to be perceived secondarily as free variants of the same particle, given their complete synonymy and phonetic similarity. There can be no question of original identity. Interestingly, alongside và / vàt there is no corresponding pair nevà / *nevàt²⁷.

Alongside và there are variants with further deictic or emphatic particles, e.g. vàge (LKŽ_XVII 814), vàgi (LKŽ_XVII 815), and so on. Following the model of niekadà / niekadõs ‘never, never at all’, a lengthened-grade form võs was created from và (see LKŽ_XIX 957)²⁸. Similarly, alongside *neva-gà (with the particle -gà), a variant nèvagos was formed, attested, among other places, in the dialects of Šakiai, Geist ar ai, and Žalioji (near Vilkaviškis). Other variants may perhaps be classified here, e.g. vaĩ (LKŽ_XVII 831), vaigi (LKŽ_XVII 847), vajè (LKŽ_XVII 980), vajãĩ (LKŽ_XVII 979), although it remains uncertain whether these are in fact the same particle, since the meaning is somewhat different. Vaĩ and vajè tend rather to express an exclamation of astonishment. Finally, mention should also be made of the variant vè (LKŽ_XVIII 460), which has exactly the same meaning as và (‘look there! look!’) and is attested mostly in Low Lithuanian dialects; the variant vègi (LKŽ_XVIII 502) is extremely rare. Given the instability of ablaut patterns in the morphology of particles, it is scarcely possible to determine how the original final vowel should be reconstructed (*và or *vè).

²⁶ See the attestations in LKŽ_XVIII 391.

²⁷ A form nèvat is indeed occasionally attested in Lithuanian dialects, but only as a loanword from Polish nawet ‘even’. Cf. ZŠŽ_II 270.

²⁸ This form is very rare (probably because it is homonymous with the widespread particle võs ‘scarcely, hardly’).

70

Zum Ausdruck der Evidentialität im Baltischen

The origin of the particle và is obscure. It was probably originally a deictic stem. Other Indo-European languages have deictic or anaphoric particles belonging to a stem *uo-, e.g. Russian его ‘there, look there!’ and Serbo-Croatian ево ‘there, look there!’ (from *e-uo). One might assume, for example, that và continues a neuter form of this stem in adverbial use, perhaps Indo-European *uo(d) ‘in this way, thus’ (or with locative meaning, ‘here’). Various pronominal forms based on *e/o-uo- should also be mentioned: Slavic *ovo- (Old Church Slavonic овъ... овъ ὁ μέν, ὁ δέ, Russian овый, Serbo-Croatian òvāj, Slovenian όv, Czech ov, Polish ów ‘this’, cf. also Old Polish owo ‘ecce’), Indo-Iranian *ava- (Avestan auua-, Old Persian ava- ‘this’, Sanskrit avóṣ gen. du. ‘of those two’²⁹. A connection with the Old Indian emphatic particle vává ‘precisely, certainly, just’ (with reduplication and double accent) is possible. Whether Sanskrit váí ‘truly, indeed’ (cf. Avestan uōi ‘truly’, Y 36, 3) also belongs here, or is even directly comparable with Lithuanian véi ‘look there!’, must be considered uncertain until the structure of the particle has been clarified³⁰. Incidentally, it should be noted that in Old Indian the particle váí can be negated; cf. ná váí ‘truly not, not at all, certainly not’ RV + (see also Younger Avestan na-ua ‘absolutely not’).

It therefore seems plausible to assume that nevà consists of two parts {negation nè + presentative và}. Here, however, one encounters a second typological difficulty. It is well known from linguistic typology that presentative particles cannot be negated in most languages. In French, for example, voici, voilà cannot be negated: alongside voici/voilà des fleurs, *voici/voilà pas des fleurs is impossible³¹. In Latin, alongside the

²⁹ Trautmann (BSW 20), Vasmer (III 116-117), Hoffmann and Forssman (1996: 167), Kent (1953: 85). Russian сом does not belong here, since it is to be interpreted as *v-oto (with *v-prothesis) and compared with Polish oto ‘the same’.

³⁰ In Lithuanian, véi may be analogically related to néi ‘not, as’ (< *nei with an uncertain acute accent).

³¹ It should be noted that spoken French can form a negation of voilà, but only in the form of a rhetorical question: voilà-t-il pas, cf. Voilà-t-il pas Monsieur qui ricane déjà? (Molière, Tartuffe, I 1), cf. also Voilà-t-il pas de quoi pousser des cris sinistres! (Victor Hugo, Ruy Blas III 5). Further attestations in Grevisse (¹²1986: 642, § 387). The presentative particle voilà, however, is not in itself negatable.

71

Daniel Petit

Presentative ecce ‘look there, look!’ has no negative counterpart *non ecce. The same applies to the Russian presentative вот, which has no counterpart *невот, as well as to Lithuanian štai, which is likewise isolated (no *neštai). Robert Forest (2000: 21–31) has shown that this impossibility of negating presentative particles constitutes almost a universal feature of the languages of the world. In the language of the Peul (Africa), for example, a sentence such as padao i ‘there is a field!’ (French ‘voilà un/le champ’) cannot be negated. A negative structure *ne-và based on the presentative và would be a typological incongruity in Lithuanian.

This situation is exemplary of the problems confronting contemporary Indo-European studies. What should one do when an etymological reconstruction conflicts with language typology? It is no longer conceivable today simply to accept typologically impossible forms and behave as if they did not exist. Typological contradictions must be considered just as compelling as the evidence from historical phonology and morphology. That means, in our specific case, that if the reconstruction of nevà as {negation nè + presentative và} proves typologically impossible, it must somehow be wrongly formulated and must be replaced by another reconstruction.

In my view, the problem lies in the negation nè. Negation is unexpected here for the two reasons given above. First, evidentiality is nowhere expressed by negating reality. Second, a presentative particle such as và is normally not negatable. It must therefore be assumed that nè has a function other than negation.

4. NEGATION AND COMPARISON IN LITHUANIAN

It is well known that the negative particle *nè can also have a comparative function in the Baltic languages. The question has already been the subject of several studies. Here I refer in particular to the important article by Georges-Jean Pinault on negation and comparison in Vedic (1985), which also cites much Baltic material (see especially pp. 131–132)³². In contemporary Lithuanian this comparative

³² S. auch Pinault (1989: 53–74). Viel Material auch bei Būga (RR, 470–471).

72

Zum Ausdruck der Evidentialität im Baltischen

function is almost exclusively restricted to comparative constructions, where extended forms of negation nẽ or negù (less often nekâip) introduce the complement. Examples from Old Lithuanian (18a) and modern Lithuanian (18b):

(18a) J. Bretkūnas, Postilė (1591: 37, 17–18)³³ Tassai daug geresnis ira nei wissu praraku Seno Sokano. ‘This one is much better than all the prophets of the Old Testament’

(18b) High Lithuanian dialect of Naujamiestis (LKŽ_VIII 621) Geriau duoti, negu prašyti. ‘It is better to give than to ask.’

In standard Lithuanian, only nẽ or negù (and nekâip) are still used in this function, but in old texts and dialects simple nè is sometimes also found after comparative forms, for example:

(19) Jakob Brodowski, Lexikon Germanico-Lithvanicum et Lithvanico-Germanicum (1713–1744)³⁴ Keturios akys daugiaus mato ne viena. ‘Four eyes see better than one.’

Apart from comparative constructions, where the negation nè functions as a comparative particle (cf. the parallel construction: comparative + kaîp ‘like’), a freer use of nè as a comparative particle (‘like’) can be attested only in old texts³⁵. For example, in the first Lithuanian dictionary (Sirvydas DTL III 84) one finds:

(20) Konstantynas Sirvydas, Dictionarium trium linguarum (³1643) s.v. jakoby: Ne zerketas žiba. ‘He shines like a mirror.’

In the eighteenth century, the same usage is still found in J. Brodowski (1713–1744):

³³ Beispiel und Analyse bei Leskien (1903: 111).

³⁴ See LKŽ_VIII 596.

³⁵ See LKŽ_VIII 597.

73

Daniel Petit

(21) Jakob Brodowski, Lexikon Germanico-Lithvanicum et Lithvanico-Germanicum (1713–1744):

Auga vaikai ne giroj medžiai.

‘Children grow like trees in a forest.’

Extended forms of negation can also exhibit this function, e.g.:

(22) High Lithuanian dialect of Suvalkų Naumiestis (Būga RR_III 826):

piktas nēi velnias

‘as evil as the devil’

This ambivalence of the negation nè (‘not / like’) is an archaism of Lithuanian. A parallel is found in Vedic, where the particle ná can, as is well known, function either as a negator³⁶ or as a comparative particle³⁷. In the Rig-Veda, the comparative use of ná is even attested more frequently than the negative use; in this meaning, ná competes contextually with iva (iva occurs after a vowel, ná after a consonant)³⁸. It is to be assumed that the comparative function arose secondarily from the negative function, namely through reinterpretation of so-called ‘negative comparisons’ (comparaisons négatives). Let us take RV VI 10.2 as an example, following Pinault (1989: 64):

(23) RV VI 10.2

Ghṛtám ná śúci matáyah pavante.

‘Pure as ghee, (my) thoughts clarify.’

Originally, the sentence was to be interpreted as a negative comparison, something like: ‘It is not the ghee that is pure; pure are the thoughts that clarify.’ Later it was reinterpreted as ‘Pure as ghee, (my) thoughts clarify.’ This figure of speech, the negative comparison, is also found in Lithuanian folk poetry—

³⁶ E.g. RV IV 30.19: ná tát te sunnám āṣṭave ‘this goodness of yours cannot be attained [in words]’. Cf. Pinault (1985: 115).

³⁷ E.g. RV VI, 10.2: Ghṛtám ná śúci matáyah pavante ‘pure as ghee, (my) thoughts clarify’. Cf. Pinault (1989: 64).

³⁸ Zum Vedischen vgl. Pinault (1995–1996, 1997, 2004).

74

Zum Ausdruck der Evidentialität im Baltischen

widespread. Here is a simple example from an old collection of folksongs:

(24) August Leskien and Karl Brugmann, Lithuanian Folksongs and Tales (1882: 16):

Neí vėjes púte, “Not the wind blew, neí šódai užė, not the gardens rustled, tik aužulai lingáva only the oaks swayed back and forth”.

That is to say:

‘The oaks swayed back and forth, as the wind blew and the gardens rustled.’

It seems likely that this tendency to reinterpret negative comparisons was already Indo-European. This explains why negation still functions as a comparative particle in comparative structures in many languages; cf., for example, English than (<þat + negation ne), American English he is taller nor me, and so on.

The question now arises whether nevà could be based on precisely this comparative use of the negative particle. This approach would resolve the typological difficulties that we have discussed at length in this article. It has often been emphasized that evidential particles frequently derive from comparative particles, as, for example, Lithuanian lýg (< ‘like’). From a typological perspective, there is nothing to prevent the assumption that the particle nè in nevà actually has a comparative function underlying it. This was already Fraenkel’s view (LEW_I 498), who translated nè in nevà as ‘as it were, like’. In terms of formation, one may think of the parallel kaĩp và, which is used in some Lithuanian dialects with the meaning ‘such as’ (= in the standard language kaĩp antai). In contrast to và, which describes the reality immediately before one’s eyes (‘just as I see the thing hic et nunc’), the comparative particle nè in nevà introduces a semantic nuance of indeterminacy and imprecision (‘approximately so, only comparatively so, more or less so, but not quite so’), which could readily make it suitable for expressing evidentiality. For linguistic typologists, the origin of nevà is therefore interesting insofar as it shows how an evidential particle developed as the result of a process of comparison based on a presentative particle, which can naturally shed some light on the phenomenon of evidentiality as a whole.

75

Daniel Petit

REFERENCES

Aikhenvald, A. u. R. M. W. Dixon, Hrsg., 2003: Studies in Evidentiality. Amsterdam, Philadelphia: John Benjamins.

Aikhenvald, A. 2004: Evidentiality. Oxford: Oxford University Press.

Ambrazas, V. 1977: Nietiesioginės nuosakos (modus relativus) paplitimas ir kilmės problema. Lietuvių kalbotyros klausimai 17, 7–54.

Balčikonis, J. 1935: Kelios pastabos dėl V. Mykolaičio-Putino romano „Altorių šešėly“ kalbos. Gimtoji kalba 3 = RR₁ 144–146.

Balčikonis, J. 1978–1982: Rinktiniai raštai I–II. Vilnius: Mokslas.

Barnes, J. 1984: Evidentials in the Tuyuca verb. International Journal of American Linguistics 50, 255–271.

Basanavičius, J. 1898: Lietuviszkos pasakos. I Band. Chicago.

Bretkūnas, J. 1591: Postilla, Tatai esti Trumpas ir Prastas Ischguldi̇mas Euangeliu, Königsberg. In: O. Aleknavičienė, Hrsg., Jono Bretkūno Postilė. Vilnius: Lietuvių kalbos institutas, 2005.

Brodowski, J. 1713–1744: Lexicon Germanico-Lithvanicum et Lithvanico-Germanicum. Manuskript.

Būga, K. 1958–1961: Rinktiniai raštai, hrsg. von Z. Zinkevičius (RR). Band I, 1958; II, 1959; III, 1961. Vilnius: Valstybinė politinės ir mokslinės literatūros leidykla.

Chafe, W. u. J. Nichols, Hrsg., 1986: Evidentiality: The Linguistic Coding of Epistemology. Norwood, New Jersey: Ablex Publishing Corporation. (Advances in Discourse Processes, 20.)

Dictionary of Contemporary Lithuanian (DŽ). 4th edition. Vilnius: Mokslas, 2000.

Dendale, P. u. J. Nuyts, 1994: Bibliographie sélective de l’évidentialité. Langue française 102, 121–125.

Donabédian, A., 1996: Pour une interprétation des différentes valeurs du médiatif en arménien occidental. In: Guentcheva, Hrsg., 87–108.

76

Zum Ausdruck der Evidentialität im Baltischen

Duchet, J.-L. u. R. Pernaska, 1996: L’admiratif albanais: recherche d’un invariant sémantique. In: Guentcheva, Hrsg., 1996, 31–46.

Forest, R., 2000: Ce qui ne se nie pas dans les langues. Mémoires de la Société de linguistique de Paris, N.S., La négation: une ou multiple?, 4, 21–31.

Fraenkel, E., 1962–1965: Litauisches etymologisches Wörterbuch (LEW), I. Band, A-privekiuoti; II. Band, prივykėti-žvolgai. Göttingen: Vandenhoeck und Ruprecht.

Gren-Eklund, G., 2000: Evidentiality and Typology: Grammatical Functions of Particles in Burmese and the Early Stages of Indo-European Languages. In: Johanson u. Utas, Hrsg., 367–381.

Grevisse, M., ¹²1986: Le bon usage. Bruxelles: Duculot.

Guentchéva, Z., Hrsg., 1996: L’énonciation médiatisée. Louvain, Paris: Peeters. (Bibliothèque de l’Information Grammaticale.)

Guentchéva, Z., Hrsg., 2007: L’énonciation médiatisée II. Louvain, Paris: Peeters. (Bibliothèque de l’Information Grammaticale.)

Hagège, C. 1995: Le rôle des médiaphoriques dans la langue et le discours. Bulletin de la Société de Linguistique de Paris 90, 1–19.

Hasselblatt, C. ²2001: Grammatisches Wörterbuch des Estnischen. 2., völlig überarbeitete Auflage. Wiesbaden: Harrassowitz.

Hémon, R. 1975: A Historical Morphology and Syntax of Breton. Dublin: The Dublin Institute for Advanced Studies.

Hoffmann, K. u. B. Forssman, 1996, ²2004: Avestische Laut- und Flexionslehre. 2. Auflage. Innsbruck. (Innsbrucker Beiträge zur Sprachwissenschaft, 115.)

Holvoet, A., 2007: Mood and Modality in Baltic. Kraków: Wydawnictwo Uniwersytetu Jagiellońskiego.

Johanson, L. u. Utas, U., Hrsg., 2000: Evidentials. Turkic, Iranian and Neighboring Languages. Berlin: Mouton de Gruyter.

Jones-Bley, K., M. E. Huld, A. Della Volpe u. M. Robbins Dexter, Hrsg., 2003: Proceedings of the Fourteenth Annual UCLA Indo-European Conference, Los

77

Daniel Petit

Angeles, November 8-9, 2002. Washington DC, Institute for the Study of Man. (Journal of Indo-European Studies Monograph Series, 47.)

Joseph, B. D. 2003: Evidentiality in Proto-Indo-European? Building a Case. In: Jones-Bley, Huld, Della Volpe u. Robbins Dexter, Hrsg., 2003, 96–111.

Jurkschat, K. 1898: Litauische Märchen und Erzählungen. Heidelberg.

Kaukiene, A. 1997: Klaipėdos krašto vakarų aukštaičių tarmė. Klaipėda: Klaipėdos universiteto leidykla.

Kent, R. 1953: Old Persian, Grammar, Texts, Lexicon. New Haven, Connecticut: American Oriental Society.

Kloekhorst, A. 2008: Etymological Dictionary of the Hittite Inherited Lexicon. Leiden: Brill.

Lazard, G. 1956: Caractères distinctifs de la langue tadjik. Bulletin de la Société de Linguistique de Paris 52, 117–186.

Lazard, G. 2000: Le médiatif: considérations théoriques et application à l’iranien. In: Johanson u. Utas, Hrsg., 2000, 209–228.

Leskien, A. u. K. Brugmann, 1882: Litauische Volkslieder und Märchen. Straßburg: Trübner.

Leskien, A. 1903: Litauische Partikeln und Konjunktionen. Indogermanische Forschungen 14, 89–113.

Lietuvių kalbos žodynas (LKŽ), 1941–2001, Vilnius, Mintis (I-II: 1968–1969, 2. Auflage), Valstybinė enciklopedijų, žodynų ir mokslo literatūros leidykla (III–VI: 1956–1962), Mintis (VII–IX: 1966–1973), Mokslas (X–XX: 1976–2001), hrsg. von J. Balčikonis, später von K. Ulvydas und J. Kruopas.

Mielcke, Ch. G. 1800: Littauisch-deutsches und Deutsch-littauisches Wörter-Buch. Königsberg: Hartungsche Hofbuchdruckerei.

Niedermann, M., A. Senn, A. Salys u. F. Brender, 1932–1968: Wörterbuch der litauischen Schriftsprache (WLS). Heidelberg: Winter.

Otrębski, J. 1956–1965: Gramatyka języka litewskiego (GJL). 3 Bände. Warszawa: Państwowe Wydawnictwo Naukowe.

78

Zum Ausdruck der Evidentialität im Baltischen

Pinault, G.-J. 1985: Négation et comparaison en védique. Bulletin de la Société de linguistique de Paris 80, 103–144.

Pinault, G.-J. 1989: Les voies de la comparaison védique. In: P. Valentin, Hrsg., La Comparaison. Actes du colloque tenu les 23 et 24 septembre 1984 par le Département de linguistique à l’Université de Paris-Sorbonne. Paris: Presses Universitaires de Paris-Sorbonne (Linguistica Palatina, Colloquia, III), 53–74.

Pinault, G.-J. 1995–1996: Distribution des particules comparatives dans la Ṛk-Saṃhitā. Bulletin d’Études Indiennes 13–14, 307–367.

Pinault, G.-J. 1997: Distribution de la particule négative ná dans la Ṛk-Saṃhitā. Bulletin d’Études Indiennes 15, 213–246.

Pinault, G.-J. 2004: On the usages of the particle iva in the Ṛgvedic Hymns. In: A. Griffiths u. J. E. M. Houben, Hrsg., The Vedas: Texts, Language and Ritual. Proceedings of the Third International Vedic Workshop, Leiden, 2002. Groningen: Forsten, 285–306.

Piročkinas, A. 1986: Jono Jablonskio kalbos taisymai. Kaunas: Šviesa.

Ramat, P. 1996: „Allegedly John is ill again“: stratégies pour le médiatif. In: Guentcheva, Hrsg., 1996, 287–298.

Roszko, R. 1993: Wykładniki modalności imperceptywnej w języku polskim i litewskim. Warszawa: Polska Akademia Nauk, Instytut Slawistyki.

Senkus, J. 2006: Kalbotyros darbai (KD). Vilnius: Lietuvių kalbos institutas.

Sereiskis, B. 1933: Lietuviškai-rusiškas žodynas. Kaunas: Lapinas ir Volfas.

Sirvydas, K. 1642: Dictionarium trium linguarum (DTL³), Vilnius, 3. Auflage. In: J. Kruopas, Hrsg., Pirmasis lietuvių kalbos žodynas. Vilnius: Mokslas, 1979.

Smoczyński, W. 2007: Słownik etymologiczny języka litewskiego. Vilnius: Vilniaus universitetas.

Trautmann, R. 1923: Baltisch-Slavisches Wörterbuch (BSW). Göttingen: Vandenhoeck u. Ruprecht.

79

Daniel Petit

Vasmer [Fasmer], M. ³1996: Etymological Dictionary of the Russian Language. 4 volumes. St. Petersburg: Azbuka.

Wiemer, B. 2006: Evidencialumo kategorija lietuvių kalboje. Baltistica 41.1, 33–49.

Wiemer, B. 2007 (in press): Lexical markers of evidentiality in Lithuanian. Rivista di Linguistica 19.1.

Willett, Th. 1988: A cross-linguistic survey of the grammaticalization of evidentiality. Studies in Language 12.1–2, 51–97.

Zanavykų šnektos žodynas (ZŠŽ), 2 Bände. Vilnius, Mokslas, 2003–2004.

Daniel Petit Ecole Normale Supérieure, Centre d’études anciennes 45, rue d’Ulm. F-75005 Paris, France [email protected]

80