Zur Vorgeschichte des Inessivs im Urostbaltischen

ACTA LINGUISTICA LITHUANICA LI (2004), 59 - 67 To the prehistory of the Inessives in the Proto-Eastern Baltic! ILJA SERZANT Universität zu Köln The article deals with the prehistory of the inessive in East Baltic. The iness. s.g. The morphology of ä-stems in *-äjen is shown to be a very old formation that must have arisen before the loss of laryngeals in Baltic, and must be the oldest among the inessive forms. The fact that the inessive postposition *-€n also is unclear within Baltic entitles us to look for an explanation reaching back to IE. An immediate parallel can be found in Old Indic, where the locative sg. ending of the a-stems -äyäm provides a phonetically exact match for Baltic *-aién. In Indic, the postposition -am is also restricted to the locative of äand i-stems (the latter have not survived in Baltic). The distribution of the postposition and its phonetic shape point to an IE postposition *-&m, an optional additional marker of the locative. 0. PRELIMINARY REMARKS The in the inessive form Sg. the @-stamp incorporated old local ending of Lithuanian and Latvian, which is to be used as *-ai (,, “stands for the acute), cf. lit. galvoj-¢ and lett. galva, comes in this form in no other idg. Speak out. In those idg. In languages where the distinction between long and short diphthongs has remained intact, we find a short diphthong *-ai, cf. e.g. Article B. dial. Boyot. rautn, wo -€ < *-ai < *-eh,-i (Rix 1976: 133). The reconstruction of an Urbaltic *-ai (e.g. Stang 1966: 199) therefore requires an explanation. 1. THE SINGULAR OF THE *a-, *-iiä-(¢)-ROD 1.1. The ending of the inessives is not old simply because it contains the unabbreviated *-ai- (as Zinkevičius 1982: 21). The local ending *-ail can ONLY be explained as a prevocal Sandhi variant of the Indo-European local morpheme: *-ah,-i-V > *-äj-V. It is striking that we also find it in the prevocal position, i.e. before the inessive particle or postposition *-én (for this approach see below). Otherwise, *-ä would not have arisen at all due to other syllabifation. Ą This is a more complete version of two lectures: „HeKOoTOpkIe 3aMeTKH O BOCTOYHOÖAJITHÄCKOM JIOKaTuBe“, XXXIII International Philological Conference, St. Petersburg, 5-6. March 2004, section “Baltic languages: historical grammar, history of literary language” and “DaZas piezimes par latviešu lokativu” 40. Conference “Artura Ozola dienas”, Riga, 19-20. March 2004. I would like to thank Prof. J. L. Garcia-Ramön (Cologne), Prof. D. Petit (Paris) and Prof. A. Holvoet (Vilnius) for their valuable critical comments and other help, although they did not agree with me on all points of this work. 60 | ILJA SERZANT Although it is theoretically possible that the prevowel Sandhivariant would have also arisen before other vowel-launching words, the fact that it occurred just before idg. *-En has been preserved, speaks for a very high age of the fixed attachment of the postposition idg. *-En to the old locative form. Probably this occurred before the shrinkage of the larynx, which is why the prevocal Sandhi variant could be preserved at all. 1.2. The expected preconsonant Sandhi variant idg. *-eh,-i > *-ai > urbalt. *-ai (?) is possibly still present in the forms of the adessives, cf. mergaip /mergaip < *-ai-p. Furthermore, the following speaks for a high age of the form of the inessive singular of the -ä stems: in the case of a prevowel Sandhi variant *-äj would have been in the 2nd person singular. Accent class according to de Saussure before the addition of the postposition a shift of the iktus to the ending to be expected, i.e. *rank-oj > **ranköj. This form would then be after the extension of the form by *-¢ to **ranköje or in analogy to the Mobilia, where the marginality principle? acted, became **rankoje. Such an accentuation is not found, however, and one must therefore assume that the extension by *-¢ precedes the effect of the Saussurean law, since in three-syllable words the Saussurean law did not work between the first and the middle syllable. 1.4 Given the fact that the inessive ending -oje seems to be old, a comment on Latvian is allowed at this point. One must assume that a similar allomorphy of the locative (*-ai before # or K, *-äj-V) existed also in Urlatin. The second variant *-ai must have resulted from the general abbreviation in Urlatin *-i, which is not proven, but can be reconstructed with certainty. But this *-i also agreed with the locative ending of the o-stems *-o-i > *-ie > *-jüberein, and since this ending was ambiguous, the a-stammed ending was transferred to the o-stems according to the following pattern: a-stamm o-stamm Primary ending *-1 *-1 Extended ending *-äji (< *-äjen) x x = *-äji. Only the coincidence of the primary localive endings of the ound ä@ stems made it possible to transfer the extended ending of the a stem to the o stems in Latvian’. Indirectly, therefore, Latvian also speaks on the one hand for the fact that in the to According to this principle, the mobile accent in the three-syllable word forms could only emphasize the first or the last syllable, never the middle one, cf. Stang (1966: 290 and 490 note 1); Zubaty (1896: 276). It resembled the Dat. Sgt. and Nom. You. of the a-tribe, the Lok. Sgt. of the ound ö tribes and the Nom. PI. of the o-tribe. Vanags (1994: 125) explains the equality of the local endings of the ound a-vowel by the phonetic coincidence of the old illative endings -an > -@ and *-änä > *-an > -an > -ä. However, this cannot apply to those dialects that have preserved the old lokative ending, e.g. the ending of Lok. Sgt. the ound a stem -ę < -ai < -ūji in Low Latvian. On the one hand, there was a double representation of the Indo-European ending of the singular locative of the a-tribe, and on the other that the extended Lithuanian singular locative of the o-tribe miske must be quite young (cf. 2.1.3). 2. THE OTHERS INESSIVENDUNGEN In the following (2.1-2.3) it is shown that the postposition urbalt. *-¢ was originally ONLY at home in the locative singular of the -4 stems (and *-iia-(-£) stems) and has spread relatively late by analogy to the plural as well as to the singular and plural of the other stems. 2.1. Singular of the *0 stems 2.1.1. The old idg. The ending of the singular locative is in fixed forms such as namie, ankstie (recorded in Zinkevičius 1966: 209 § 278), as well as in the adjective Sg. namiep(i) transmitted. The intonation is to be used as a circumflex due to the above-mentioned primary forms. This continues exactly the Indo-European locus ending *-o-i > *oi > ie. Stang (1966: 182-3) suggests that the development could be imagined as follows: *-oi (cf. ai. ašve < *ekuoi) > *-ai > *-& > -ie. The common ending of the inessives is -e, e.g. miske (‘in the forest’). This extension contains the original idg. Ending, which was additionally extended, like all other locatives, by the inessive postposition *-&° (cf. Stang loc. cit.). This inessive form (miske) is to be explained as follows: *misk& (< *-ai < *-o-i) + ¢ > *miškę“ > (Leskien’s law) miske (Stang loc. cit.) or > miski in those dialects where every *en > i. It follows from this that the expansion of the o stems with *-¢ took place relatively late. The emergence of a form misk-e at an earlier stage, when *-ai (< i.e. *-oi) was still diphthongic, would be inexplicable, since *-ai + ¢ would have become **-aje. The fact that in Lithuanian a large number of fixed locutives Sg. which has received *0 stems, in contrast to the -4 stems. 2.1.2. The addition of the postposition *-¢ must have taken place before or during the effect of the Saussure- 's law, since only by the merger of the ending *-€ and the postposition *¢ inrate (2nd accent class) could the acute result, which is a prerequisite for the shift, in the second and last syllable. Before the contraction, the iktus would not have been shifted, since it must not skip any intermediate syllables during the shift according to de Saussure. 2.1.3. For a relatively young age and an exclusively Lithuanian development of the The inessive form of the o stems also indirectly speaks the situation in Latvian (see 1.4). 3 Here and further <> stands for the intonation (acute), not for the iktus. 5 The objection of Mažiulis (1970: 133.1) that a contraction must necessarily have resulted in circumflex intonation is not entirely justified. Firstly, this is a much later contraction than, for example, that of the Indo-European vowels (as, for example, in the Dat. Sg. of the o stems *-o-ei), and secondly, the addition of the *-En in analogy to the lok occurred here. Sgt. of the a-stem, consequently one cannot exclude the analogous transfer of the acute intonation to the contraction result (*-&-&n > *-&n analogous to the loc. of the ä-St. *-aj-én > -En). 62 | ILJA SERZANT Probably there was no analogous form at all, rather Latvian simply used the inesitive form of the -4 stems. 2.2. The inseminations of the i and u strains 2.2.1. The Indo-European ending of the u-tribe was *-Öu, cf. ai. Son, sl. cnınoy, somit sind lit. sänuje /sünuj, hochlett. tirguö (< *tirguöji) e.g. young, since the -jder postposition only comes from the Lok. Sgt. der -ūor -i-root is to be declared. The locative *sünü, which is e.g. contained in sünuje/sünuj, can also only be explained by the Leskien law of *-Öu-#’. The enlargement -je must have occurred sometime afterwards. 2.2.2. The situation is different with the i-trunks. Here the modern ending -yje (avyje) has been transformed to other stems and is therefore young, but in the Zemgaling dialect we find -ie < -€ (Zinkevičius 1966: 231), in some Latvian dialects -ie, -€ (Endzelin 1951: 412) as well as in the Catechism of Mažvydas nakteie, schirdie, naktie, manieie (Stang 1966: 209, 211), which go back to *-ėję or partly to *-iej&, and thus the old i-stemmed ending of the singular locative *-či, cf. ved. lamb, gr. Always. röAnı, enthalten’. The question of whether these forms are at home in the -i stems or whether they are borrowed from the -€ stems via the *_(i)jas stems is difficult to decide with certainty. One could consider the extension with *¢ due to the fluctuating preform *-ėjė vs. *-iėję (Stan 1966: 210f) so explain that the old locative form *-&i was extended with the extension -je abstracted from the locative singular of the 4-stems and initially developed into *-€ije’’. Then, however, the individual dialects have treated this form differently, namely, some have simplified it —*-£ije > *-ėje > lit. -ėje, > lett. -ė —the others, on the other hand, retained the diphthong and later changed it to -i: *-€ije > *-ieje > -ie. The Zemaitic ending -ie (< ė < *&je) may also have been adopted from the *(i)jo stems into the i stems, in the *ijo stems it may in turn come from the £ stems, which in Zemaitic and Latvian coincide partly (Endzelin 1922: 31). Such or similar development must be taken as a starting point, because otherwise one would face the difficulty of having to take the extension with *-ė abstracted: *-j¢ -in different areas relatively chronologically for two far apart times: either for the time of the presence of *éi (Zemait., let. dialects) or for the time of the presence of ie (let. dialects). However, one should consider that there is a complicated development in between: *či > *ei > *& > ie. Whichever of the two proposals one follows, Zinkevičius (1982: 27) explains this ending as having originated in analogy with the i-roots (-i), where it is also supposed to have originated from the consonant roots (Zinkevičius 1982: 20-21). However, the two forms can be explained verbally by using the Leskian law: -u (< *-ūo < *-öu), -i (< *-ie < *-€i). Especially since the intermediate stages -uo and -ie have been incorporated in the Latvian and Zemaites dialects. Certainly the old ending is contained in the infinitives on *ti, cf. lit. nordwestzem. -te, refl. - You. See debesesu in the Catechism of Mažvydas, 291, (about Kazlauskas 1958: 82) Lok. E.g. with a č-(< *ijä-) stem ending in an i-stem (originally s-stem) seen synchronously. “Only in this way does Kazlauskas reconstruct it (1968: 154). Anders Stang (1966: 210). ON THE PRECEDENTS OF THE INESSIVS | 63 one must assume that the extension with *-&/-j¢ was done much later than in the locomotive. Sgt. of the -a stem, by analogy to the latter. 2.3. The inessive plural The old plural ending is preserved in only a few dialects, cf. 1966) 1937 – Zbigniew Brzezinski, Polish politician (d. Here, the form of the o-, iand u-stems was first adapted to the Illative Pl., according to the following analogy to the 4-stems: Ill. Pl. -osna /Lok. Pl. -osu: Ill. Pl. -uosna (-ysna, -üsna) /Lok. PI. x, x = -uosu (-ysu, -üsu —-uose / -uosu according to the o stems), cf. Stang (1966: 186, 213, 218)". This has also occurred in Latvian, cf. darbuös. With the -4-trunks is the old Indo-European Lok. Pl.-ending only in | few dialects remained unchanged, cf. 1934) 1928 – Rudolf Steiner, German philosopher (d. -i is here analogous to the shortened Lok. "R.I.P.R.I.- P.R.I.P.R.I- .P.R." In most dialects and in high language, the terminating -u in the ends of all stems has been replaced by the -e of the inessive singular, cf. rankose, in the others by the -a of the illustrative singular (Stan 1966: 186). First the plural form of the -# stems was probably renewed: rankosu > rarikose (replacement, not addition: **rankosve), then the other stems as well. The formations with the extension *-ė are therefore young in the plural forms of the inessives. 3. THE POSTPOSITION URBALT. *-én 3.1. The following considerations lead to the reconstruction of the inessive postposition of the Proto-East Baltic as *-En (< *ė): a) the form Iness. Sgt. vaik-i in those dialects, where *-en > i and *-an > u, as well as the extended forms of the adjective, cf. dangujejis speak for a nasal in the pronoun (cf. Stang 1966: 182); b) the intonation must be in Iness. Sgt. of the *0 strains due to the effect of the Saussure law as acute. This is also confirmed by the fact that in all inesitive forms the Leskien's law (which reads as *-¢ > -e) must have worked. Compare here the forms of the accusative and instrumental singular of the -a stem: Akk.Sg.rarikq (< *-afi < *-am < *-eh,m) versus Instr. Sgt. ranka (< *-an < *-än). Should one therefore assume that the accentuated intonation only in the singular of the -o stems arose only as a result of contraction, one is faced with the difficulty that the other non-contracted —locative forms do not come from **-¢ (< **en), but from -e (< *¢ < *ėn). Since the intonation can be set as acute, one can assume *-En for Urbaltic (Stan 1966: 128). " Zinkevičius (1966: 238), on the other hand, thought that the locomotive. Pl. on -uose relies on the accusative plural and the postposition *-én. Stang (loc. cit.) has criticized this opinion with striking arguments. One could add a formal reason to Stangs’ explanation: there are no forms of the type **ratüose/**ratuose (2), as they should have been according to Zinkevi¢ius’ explanation, similar to the Illativ Plural (s.d.). Later Zinkevičius (1982: 28) joins Stangs opinion. 64 | ILJA SERZANT 3.2. Furthermore, it must be taken into account that the postpositions in Lithuanian, but also in Old Latvian, were subject to a special weakening in comparison to the corresponding (prepositions or) prefixes, cf. lit. dievop(i) (not from -pie), lett. majup ‘at home’ (< *pi, not *-pie, which would have become > **-pi) against pri£ /(pie), lett. pie, slav. npn, lit. rasikon (a) versus lit. nuo, lett. no, slave. and. Leskien's law does not explain this weakening, since all the aforementioned words are circumflexed. This weakening must have been very old, since it reflects the Indo-European decline. Lit.lett. nuo would be abbreviated to nu in both languages, not to -na. The form *na can be explained as a Baltic full step to the *nö (< (?) *noH, *nö) which is understood as a stretching step. This group is joined by idg. *som ‘together’, which in prefixed form in Lithuanian, Latvian and Slavic has the full degree for nouns, cf. lit. samdas, lett. sūods (< *sands) and otherwise in the Latvian prefix sa-(< *sam-), sl. cxa% against the prepositional form of Lithuanian su (Lett. ar —not related) and Slavic. ¢w. In analogy to this one would expect a weakened postposition to the well documented prefix ie- (< *en-) in Latvian, namely *-in (<*-n), which is also represented in Lithuanian as the preposition į!?. But we have *¢ < *-En. This fact speaks against the etymological connection of *-¢ and idg. *en, lett. ieund secondly, that the formation of the inessives Sg. of the @ stems should be kept outside of this sound system, and that is because, as in 1.1 and 1.3, had already been inherited from the basic language in this form. 4. THE EXTERBALTIC COMPARISON 4.1. The intralingual examination of the forms of the locative and inessive in Baltic shows that the inessive form singular of the -# stems (as well as the -£-St. < *-ija-St.) is by far the oldest among the inessive forms. Their age goes back to the period before the shrinkage of the larynx. From this form, the element *-¢ (or abstracted: -je) spread to all other endings of the locative, and thus a “marked locative” emerged in Proto-East Baltic, i.e. Inessiv. The question of the origin and etymology of this postposition is still unresolved. Stang (1966: 182) puts it in connection with the preposition *h en(i) (Let. ie-k3a, Gr. év, eis, Lat. in u.a.). The difficulty, however, lies in the fact, as Stang (loc. cit.) himself notes, that for our postposition length (*En) has to be reconstructed due to the acute (3.2). In this form, however, the preposition *en does not occur in Indo-European. 4.2. I would like here, on account of the original distribution (only locally speaking of the -ä-stems- °) and on account of the sound shape'* the Baltic postposition *én in direct 12 Girdenis (1980) notes that in the dialect of Kretinga (in Zivatas, XVIII c.) for the aukšt. Preposition į is used when the preposition y, which goes back to *en, is used. Besides, but less often, i = aukšt is also used. in vor. This allomorphy only testifies, I think, that in this dialect the original distribution of the full (*en) and null (*n) is no longer maintained. In Old Indian, the postposition -äm originally occurs with the i-roots (devyam) and the pronoun (tdsyam). The -i stems have not been preserved in Baltic and the pronoun flexion has been preserved in Baltic in the locative Sg. replaced by the nominale (Stan 1966: 244-245). Ur-Baltic *-m becomes *-n according to phonological law (Stan 1966: 114). 13 14 THE PREHISTORIC INESSIVES | 65 Connection with the ancient Indian postposition -am (cf. to ai. year (f) ‘Heer’, Lok. Sgt. sénay-am) bring, and that would be as a common preform idg. *&m to start. The fact that this postposition only appears in these two language families could be explained by the fact that it was only used optionally in addition to the postpositionless form of the locative in pre-singular languages, as one must assume for Urbaltic and even for Urlithuanian and Urletian. Probably it served to specify local conditions. 43. The following facts speak for the antiquity of the extension -ūm in Indian: a) the extended form is mandatory not only in classical Sanskrit, but also in Vedic in the locative singular of the -4 stems. As a rule, the later extended forms of other cases occur only in one of the two languages; b) the previous explanation (Wackernagel-Debrunner 1930: 121; Thumb-Hauschild 1959: 46-7) of the Indian element -am (in senayam), which assumed an etymological affiliation to the Avestic postposition (in the Indian Preund postposition) -4 /a, encounters considerable difficulties, and namely this hypothesis must assume a double extension first by -4 (which also functions as post and preposition or prefix in Avestic), then by an unclear -am. In case of such a first Indian contraction, the two-syllabic reading should also appear in the Vedic metric texts, since here chronologically much earlier, Indo-European contraction cases (e.g. genitive plural *-0-om > *-öm) are partly still two-syllabic. However, this is not the case. Oldenberg (1888: 185) mentions the lok in his list of all cases where the two-syllable reading occurs in Vedic. Sgt. not. While in Indian only the singular form of the locative of the -@ and -i stems contains the postposition *-ēm", the other forms of the locative are only occasionally indicated with -a, e.g. ved. Lok. Pl. apsv-a, in Iranian and above all in Old Persian, in a similar way as in Baltic, with the help of the latter postposition, an "inessive" -4 has been formed, which can be seen from the fact that the postposition -@ in Old Persian had become mandatory for most local endings. It seems, therefore, that in Avestic and Old Persian in the locative Sg. of the -4-trunks the postposition -4 has displaced the older, already inherited from Indo-European in this form postposition *-ēm > **-äm. 5. CONCLUSION The investigation of the proto-East Baltic locative Sg. 1) The preservation of the prevocal Sandhivariant of the locative morphem *-eh, i-V just in a prevocal position indicates that the form of the inessive Sg. the -a stem is ancient in Baltic. 2) The non-expanded ending must be for the Proto-East Baltic due to the adjective forms 5 Possibly contains according to Wackernagel-Debrunner (1930: 121) the form of the D. Abl. Instr. You. on -bhyam also this postposition. 66 | ILJA SERZANT be used as *-ai (1.2). Unfortunately, the type of intonation cannot be determined. One could, however, suspect that it was a circumflexed *-ai. In this case, this ending would continue the Indo-European locus ending *-eh,-i before a pause or consonant. 3) The linguistic- -historical comparison of all present-- day inessive forms has shown that the form of the inessive Sg. the -a stem is the oldest and that it formed the starting point for the spread of the abstracted postposition *-¢ or -je. 4) The points 1 and 3 open the possibility to change the form of the inessives Sg. on urbalt. *-ąjėn to be combined with the Old Indian form of the locative on -ūyam due to the phonological and distributive agreement. For this, it is necessary to assume that the distribution of the postpositions -4 and -ūm in ancient Indian, in contrast to the Iranian languages, represents the original (lectio difficilior), that the generalization of the postposition -a in Iranian for the marked locative is based on later analogy and that the unexpanded locative form was still alive in Indo-Iranian. BIBLIOGRAPHY ARUMAA, P. 1930: Lithuanian oral texts from the Vilnius region. In: University of Tartu University Press. Acta et Commentationes Universitatis Tartuensis, B Humaniora XXIII. Dorpat (Tartu). ENDZELIN, J. 1922: On the Baltic declination of the "ablautenden" (i)io-stems. Kuhns Zeitschr13-35. ift 50, ENDZELIN, J. 1951: Grammar of the Latvian language, Riga: Latvijas valsts izdevniecība. GiRDENIS, A. 1980: Del vieno prepositional forms in the 18th-century Northern Samogitian Kretingiskiy dialect. Baltistica 16 (2), 111-115. KAZLAUSKAS, J. 1968: Historical grammar of Lithuanian language, Vilnius: Mintis. Mažiuris, V. 1970: Relations between the Baltic and other Indo-European languages (Declination), Vilnius: Mintis. ME = Bolvenbachs-Endzelins: K. MULENBACHS, Latvian language verb dictionary. Redicted, supplemented, continued by J. ENDZELINS. 2nd unmodified edition. 6 Bde. Cikäga: Baltu philologist kopa, 1954. OLDENBERG, H. 1888: The Hymns of the Rig-Veda. Metric and Text Historical Prolegomena, Berlin: Wilhelm Hertz. OTREBSKI, J. 1934: East-Lithuanian Tver narrows. Part I: Grammar, Krakow: PAU. Rıx, H. 1976: Historical Grammar of Greek. Darmstadt: Wissenschaftliche Buchgesellschaft. Rosınas, A. 1999: On the origins and development of the inessive and adhesive forms. Baltistica 34 (2), 173-183. SERZANT, I. 2003: Intonations in the suffixes and endings of Latvian. Synchronicity and diachronicity. Baltu filologija 12 (1), 83-123. SERZANT, I. 2004 K Bonpocy 06 obpasoBaHuu aneccusa. Acta Linguistica Lithuanica 51, 49-57. STANG, Chr. S. 1966: Comparative Grammar of the Baltic Languages, Oslo etc. : University of California Press. THumB-HauscHiLD 1959: Handbook of Sanskrit. An Introduction to the Linguistic Study of Ancient Indian by Albert Thumb. Part 11: Theory of Forms. Third heavily revised edition by Richard Hauschild, Heidelberg: Winter. Tororov, V. V. 1968: Jlokamue e cnasanckux A3bikax, Mocksa: Hayka. Vıpusirıs, A. 1959: Some features of the Zietela dialect. Language Studies 2, 195-230. WACKERNAGEL-DEBRUNNER 1930: Ancient Indian grammar by Jacob Wackernagel. Volume 111: Nominal flexion-number pronouns by Albert Debrunner and Jacob Wackernagel, Göttingen: Vandenhoeck and Ruprecht. 67 Vanass, P. 1994: The development of case endings in the oldest Latvian language monuments. Linguistica Baltica 3, 121-130. ZINKEVICIUS, Z. 1966: Lithuanian dialectology, Vilnius: Mintis. ZINKEVICIUS, Z. 1982: Postpositional locutors in Lithuanian. Baltistica 18 (1), 21-38. ZuBATY, J. 1896: Baltic Miscellanea. Indo-European Studies 6, 269-307. Ilja A. SerZant Gauta 2004 11 04 University of Cologne Institute for Linguistics Comparative linguistics D-50923 Kėln [email protected]