I
Indogermanische Forschungen 115. Band
II
III
Indogermanische Forschungen Zeitschrift für Indogermanistik und allgemeine Sprachwissenschaft Begründet von Karl Brugmann und Wilhelm Streitberg Herausgegeben von Wolfgang P. Schmid und Eckhard Eggers
115. Band 2010
De Gruyter
IV Alle für die Indogermanischen Forschungen bestimmten Aufsätze und kleineren Beiträge senden Sie bitte an Apl. Prof. Dr. Eckhard Eggers, Nikolausberger Weg 63, D-37073 Göttingen, bzw. als .pdf-File per E-Mail an:
[email protected] Rezensionsexemplare senden Sie bitte an Apl. Prof. Dr. Eckhard Eggers, Redaktion Indogermanische Forschungen, Arbeitsstelle Niedersächsisches Wörterbuch, Kreuzbergring 50, D-37075 Göttingen.
Mit der Annahme zur Veröffentlichung überträgt der Autor dem Verlag das ausschließliche Verlagsrecht und das Recht zur Herstellung von Sonderdrucken im Rahmen des Urheberrechts. Dazu gehört insbesondere auch die Befugnis zur Einspeicherung in Datenbanken, der Verbreitung auf elektronischem Wege (Online und/oder Offline) sowie das Recht zur weiteren Vervielfältigung zu gewerblichen Zwecken im Wege eines fotomechanischen oder eines anderen Verfahrens. Der Autor hat das Recht, nach Ablauf eines Jahres eine einfache Abdruckgenehmigung zu erteilen.
Ü Printed on acid-free paper which falls within the guidelines of the ANSI
to ensure permanence and durability.
© Copyright 2011 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this journal may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage or retrieval system, without permission in writing from the publisher. Copying in the USA! Authorization to copy items for internal or personal use, or for the internal or personal use by specific clients is granted by Walter de Gruyter, for libraries and other users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of US $ 3.00 per copy is paid to CCC, 222 Rosewood Drive, Danvers, MA 01923, USA. 0019-7262/11 © by Walter de Gruyter GmbH & Co. KG Berlin New York Printed in Germany ISBN (Print) 978-3-11-022280-7 ISBN (Online) 978-3-11-022281-4 ISBN (Print + Online) 978-3-11-022282-1 ISSN (Print) 0019-7262 ISSN (Online) 1613-0405 Indexed in Current Contents
V
Inhalt I. Aufsätze: A g b a y a n i , Brian, G o l s t o n , Chris. Second-position is first-position: Wackernagel’s Law and the role of clausal conjunction . . . . . . B r o s m a n , Paul W. The nt- participles and the verbal adjectives in *-to- . . Ve r b e k e , Saartje. Postpositions and agreement in Rajasthani languages: A comparative analysis of alignment features in Marwari and Harauti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . P a t r i , Sylvain. Qu’est-ce qu’un pronom en anatolien? . . . . . . . . . . . . S h i e l d s, Kenneth. Hittite sˇia-‘1’ and its implications for the etymology of Indo-European numerals . . . . . . . . . . . . . . . . . . . . . . . S c h ü r r , Diether. Lykische Genitive . . . . . . . . . . . . . . . . . . . . . . Wo o d h o u s e , Robert. Greek « Λ and further evidence for laryngeal as resonant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M a n c o , Alberto. Two unexplained terms in Iliad: and . M a d a r i a g a , Nerea. The development of Indo-European middle-passive verbs: a case study in Ancient Greek and Old Church Slavonic . . B l a s c o F e r r e r , Eduardo. Iberian *ortubeles´ and ordumeles, PalaeoSardinian Ortumele. New horizons in substrata research . . . . . P â r v u l e s c u , Adrian. Lat. servus . . . . . . . . . . . . . . . . . . . . . . . A n g h e l i n a , Catalin. A note on the phoneme schwa (a˘) in Romanian . . . V i l l a n u e v a S v e n s s o n , Miguel. Baltic sta-presents and the Indo-European desiderative . . . . . . . . . . . . . . . . . . . . . . . . . . . W i l l i , Andreas. Hera, Eros, Iuno Sororia . . . . . . . . . . . . . . . . . . .
1 22
35 65 110 118 127 135 149 179 190 198 204 234
II. Besprechungsaufsatz: K n ü p p e l , Michael. Fragen zu den von V. M. Illicˇ-Svitycˇ nachgelassenen Materialien . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
268
Druckfehlerkorrektur . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
275
Zur formalen Gestaltung von IF.-Manuskripten . . . . . . . . . . . . . . . .
277
VI
Inhalt
Second-position is first-position
1
Second-position is first-position: Wackernagel’s Law and the role of clausal conjunction* Abstract We show that the notion “second-position” (Delbrück 1878, Wackernagel 1892) involves a misunderstanding of coordinate structures and of the prosodic weight of sentence-initial words in early IE languages. Conjunctions begin and end life between their conjuncts; material from the right conjunct is moved only if postpositives (Dover 1960) require it. This puts “second-position clitics” either between clauses (conjunctions) or at the beginnings of clauses (pronouns and particles), leaving nothing in second-position.
1. Introduction Based on the comparative syntax of early Indo-European (IE) languages, Delbrück (1878) and Wackernagel (1892) proposed that PIE had a set of “second-position clitics” that followed the first stressed word of the sentence. Their thesis has been so influential that secondposition in a sentence is often referred to as Wackernagel’s position, even outside of IE. Within IE, their conjecture has attained the status of a law: “One of the few generally accepted syntactic statements about Indo-European is Wackernagel’s Law, that enclitics originally occupied the second position in the sentence” (Watkins 1964: 1036). Much contemporary work has sought to reevaluate the status of Wackernagel’s Law for early IE languages within current linguistic theory. This body of recent work offers a descriptive precision that was not available to the frameworks within which Delbrück and Wackernagel worked; and it has given rise to several analytic trends with respect to second-position phenomena in these languages (Garrett 1990, 1996, Luraghi 1990, 1998, 2001 for Hittite; Hale 1996 and Hock 1996 for Sanskrit; Janse 1992, Hock 1996, and Taylor 1990, 1996 * We would like to thank Craig Melchert for discussion on the Hittite data and the analysis presented here. We also thank Cheryl Chan, Sean Fulop, Andrew Garrett and E.F.K. Koerner for comments on earlier drafts of this paper. Any remaining errors are the sole responsibility of the authors.
2
Brian Agbayani and Chris Golston
for Greek; see also Anderson 1993 for more general discussion of the status of Wackernagel’s Law in current theory).1 We argue here that nothing regularly occurs in second position in any of these languages and offer an analysis under which these items lie outside of the clause (conjunctions) or in clausal first-position (pronouns and particles). In what follows we will use Dover’s (1960) term “postpositive” to refer to these items. In early IE languages, postpositive conjunctions (Greek ‘and’, Latin =ve ‘or’) are always the first in any string of such elements. We propose that such a conjunction lies external to the clause that forms its right conjunct, and that all subsequent postpositive elements are therefore clause-initial. These postpositive elements (pronouns and particles) always follow conjunctions when they are present. Again, conjunctions fall between the clauses they conjoin, so any elements that immediately follow them are clearly clause-initial rather than clause-second. We base our conclusions on the syntax of three early IE languages for which the ‘second position’ phenomenon is especially plain: Ancient Greek, Latin, and Hittite. We expect exactly similar results for Sanskrit, Avestan, and other early daughters of PIE, though we lack the language expertise to state this with any degree of confidence. The conjunctions we have in mind include normal (for lack of a better term) and postpositive conjunctions like the following: (1)
1
Conjunctions normal
postpositive
Greek
‘and’ $ ‘but’
= ‘and’
‘and’
Latin
et ‘and’ at ‘but’
=que ‘and’ =ve ‘or’
Hittite
nu ‘and’
=ya ‘and’ =ma ‘but’
We do not consider here analyses of the placement of second-position elements in modern languages, for which there is an enormous body of literature (see the seminal work of Klavans 1982, the collected articles in Halpern & Zwicky 1996, and important references such as the work of Franks & King 2000, among many others). It is an open question as to whether the analysis proposed here for early IE may extend to all modern cases.
Second-position is first-position
3
Some postpositives are phonological clitics (‘=’) and others are not. Thus Greek = ‘and’ is a clitic and loses its tone to a preceding word, while ‘and’ is not enclitic and retains its tone. Latin too has postpositives that are clitics (=que ‘and’) and postpositives that are not (enim ‘for’). Hittite is unique among these languages insofar as all postpositives are enclitic (=ya ‘and’, =ma ‘but’, etc.). Regardless of their clitic status, all postpositive conjunctions show up in the same part of the sentence as normal conjunctions do, or so we will try to show. Traditional analyses place postpositive conjunctions inside the righthand conjunct, on a par with other elements in ‘second-position’. We argue here that postpositive conjunctions fall between their conjuncts syntactically, as all conjunctions do, never within one of their conjuncts. This puts postpositive elements that follow conjunctions in first position.
2. Conjunction below the clause Our analysis of postpositive elements rests upon a proper understanding of conjunctions, so we begin here with simple cases of conjunction below the clause, specifically, conjoined noun phrases. As we will see, normal conjunctions (, et) come between their conjuncts and postpositive conjunctions appear to follow the first word of their second conjunct. Most early IE languages have full form conjunctions that come between their conjuncts in the familiar fashion. Using square brackets to indicate syntactic constituency we represent conjoined phrases as follows: (2)
[ ]NP λ [ ]NP sceptre and fillet ‘sceptre and fillet’ (Homer, Iliad 1.28)
(3)
[montem Iuram]NP et [flumen Rhodanum]NP Mount Jura and river Rhone ‘Mount Jura and the Rhone river’ (Caesar, Bello Gallico I.6)
Following much recent work in the syntax of coordination, we assume that conjunctions appear between the elements they conjoin in a configurational structure that groups the conjunction with the righthand conjunct (Munn 1993, Johannessen 1998, Zoerner 1999):
4 (4)
Brian Agbayani and Chris Golston
configurational structure for coordination
NP [montem Iuram]
& et
NP [flumen Rhodanum]
But our argument does not depend on having this exact structure for conjunction: we require only that a conjunction is distinct from the elements it conjoins. Even if the conjunction forms a constituent with the following conjunct (as it does in 4), it is clearly not part of either conjunct: the first word of the second conjunct above is not et but flumen. This corresponds with the semantics of the construction, where the coordinated terms are [montem Iuram] and [flumen Rhodanum], and et functions as a Boolean operator that takes the individual terms and yields their semantic coordination (creating a single category of the same type as the individual terms). This much should be noncontroversial. As we have seen, a number of early IE languages also have postpositive conjunctions. We assume that the syntactic and semantic structure for these is still [conjunct & conjunct], as in the following case from Latin: (5)
syntactic constituency
NP [dies]
& =que
NP [noctes]
The actual spoken form is of course quite different, with the second conjunct fronted to the left of =que: (6)
dies noctes=que _____ days nights=and ‘days and nights’
(We underline here the word that has moved and indicate the position from which it has moved with underline as well.) Noctes has clearly been moved from the position in which it is interpreted semantically, but the movement has probably not taken place in the syn-
Second-position is first-position
5
tax. For one thing, syntactic movement of a conjunct is banned across languages under the Coordinate Structure Constraint (Ross 1967). Second, syntactic movement from the complement of a head (=que) to a position within that head’s maximal projection is also banned across languages under constraints on Extreme Locality or Anti-locality (Grohmann 2001, Abels 2003, Kayne 2005). Finally, there is the problem of where noctes would move to if it did move to a syntactic position. Noctes is a phrase but the specifier position to the left of =que is already filled with dies, blocking movement of noctes into that position. The only other position available is the position of the head, already filled by =que; even if =que were to allow something else in its slot, it could only allow a head, not a phrase. For these reasons, a syntactic analysis of the movement is untenable. We attribute the movement to the phonology (Agbayani & Golston 2010): the second conjunct (or part of it, see below) moves to the beginning of its phonological phrase because phonological phrases in Greek must begin with real words rather than postpositives. We flesh out that proposal now for sub-clausal coordination and extend it to clausal coordination below. The term second-position is not generally used for sub-clausal coordination, but it has exactly the same properties as clausal coordination and therefore bears looking into. Following Selkirk (1986, 1995) we assume that each maximal projection in the syntax forms a phonological phrase () at its right edge, forcing the conjunction into the phonological phrase formed around the second conjunct. Using parentheses to indicate prosodic constituency the difference between syntactic and prosodic constituency looks like this: (7)
prosodic structure: right alignment of XP with
NP [dies]NP ⇓ (dies)
& que (=que
NP [noctes]NP ⇓ noctes)
The right edges of the NPs above give us the right edges of the phonological phrases (dies) and (=que noctes). So while =que is not part of either NP, it is part of the phonological phrase that contains the sec-
6
Brian Agbayani and Chris Golston
ond NP. Note that the semantics is read off of this representation, where the conjunction sits between its conjuncts. The syntax of (7) is fine, but the phonology has a rough patch: the phonological phrase that includes the second conjunct begins with what Dover (1960) calls a postpositive (=que), a word that cannot occur at the beginning of a phonological phrase. Given a constraint against phrase-initial postpositives, the syntactically bizarre but pronounceable order (noctes=que) is preferable to the syntactically faithful but unpronounceable order (=que noctes). In cases like this, where the second conjunct consists of a single word, the conjunction is phrase-final because the moved word (noctes) constitutes the entire final conjunct. This is shown for Greek below: (8)
( ) ( «= _____) scepter honors=and ‘scepter and honors’ (Aeschylus, Prometheus Vinctus 171) (κ ) (φ = _____) end head=and ‘end and head’ (Plato, Timaeus 69a)
Thus the left edge of the second phonological phrase begins with a conjunction (λ in (2) above), or with some word that follows ( « and φ in (8)) in case the conjunction is a postpositive. In either case, we assume that the conjunction stays in its base position. We stress this point to contrast our analysis of postpositive conjunctions with analyses where postpositive elements prosodically “flip” or “drop down” into the following phrase (Janse 1992, Halpern 1995, Anderson 1996, Garrett 1996, Hale 1996, Hock 1996, Taylor 1996, Embick & Noyer 2001). Such analyses take these postpositive elements to be prosodically light, stressless elements that need to be phonologically incorporated into a prosodically heavier “host”. In this sense they follow closely the original conception in Delbrück 1878 and Wackernagel 1892, who incorrectly stated that second position clitics had to be preceded by ein betontes Wort. But the apparent host for a postpositive is often prosodically as light or lighter than the postpositive itself and not all postpositives in these languages are phonologically enclitic, as we have seen. Thus the host may be the same prosodic weight as the postpositive, and both may bear pitch accent, as shown below (L = light syllable, H = heavy):
Second-position is first-position
(9)
7
L L (μ ξ _____ ») the and making ‘and the making’ (Aristotle, Poetics 1454a)
Or the host may be the same prosodic weight as the postpositive element, but lack pitch accent: (10)
H H (¹ _____ ) the for same ‘for the same ones’ (Aristotle, Poetics 1455a)
Or the host may be lighter than the postpositive element and lack pitch accent: (11)
L H (² σ _____ P "«) the so Polymarchus ‘for Polymarchus …’ (Plato, Republic 327a)
Given data like this, it is clear that the phonological weight of both the postpositive word ( ξ, , σ ) and its so-called host (μ, ¹, ²) are completely irrelevant for their linear ordering. Cases like (9–11) simply cannot be construed as a postpositive following ein betontes Wort. Let us now turn to the corresponding facts in Latin and Hittite. The postpositive conjunction pattern for Latin is illustrated below, where =que is postpositive and cannot occur at the left edge of a phonological phase, forcing the movement of some other word from its righthand conjunct. (12)
(Labiemun) (Terbonium=que _____) Labienus Trebonius=and ‘Labienus and Trebonius’ (Plautus, Mostellaria 1.22) (oppida) (vicos=que _____) towns villages=and ‘towns and villages’ (Caesar, Bello Gallico 1.28)
Again, the first word of the second conjunct is fronted so that the phonological phrase won’t begin with =que, a postpositive. Similarly in Hittite, where the conjunction =ya (=a after a consonant, with gemination) forces a word from its right-hand conjunct to move:
8 (13)
Brian Agbayani and Chris Golston
(UD.KAM-ti) (GE6.KAMti=ya _____) by.day by.night=and ‘by day and by night’ (KUB 33.98 ii 11) (nepis) (tekann=a _____) heaven earth=and ‘heaven and earth’ (KBo 6.29 ii 12–13)
When the second conjunct is longer than a single word, only some of the words move, as seen in the following example from Greek, where Ν and « move but %λ and & « remain in place: (14)
(Z ) (Ν _____ %λ) Zeus other and gods ‘Zeus and other gods’ (Homer, Iliad 6.476) ) (ρ« $ «) (« %’ _____ & «) (( )9 goats flaying hogs and roasting in yard ‘flaying goats and roasting hogs in the yard’ (Homer, Odyssey 2.300)
(15)
phonological movement from the right conjunct
(Z ) Zeus
(Ν te other=and
_____ %λ) gods
Note that since the conjunction has not moved, Ν is no longer the first word of the second conjunct. It has left NP-initial position and is now outside of the right conjunct altogether. Thus = is in situ and Ν is outside of the NP it is interpreted with. Latin data show the same thing: (16)
(cunctis oppidis) (castellis=que desertis _____) defeated towns fortresses=and deserted ‘the towns defeated and fortresses deserted’ (Caesar, Bello Gallico 2.29) (vir magni ingeni) (summa=que _____ prudentia) man great talent superior=and wisdom ‘a man of great talent and superior wisdom’ (Cicero, Legibus 3.45)
Second-position is first-position
9
A moment’s reflection on the semantics of coordination again requires that castellis and summa have moved from a position following the conjunction: (17)
[magni ingeni] =que [summa prudentia] prudentia) (magni ingeni) (summa =que _____
The same holds in Hittite, where postpositive conjunctions like =ya cannot be phrase-initial and require some word from the following conjunct to precede them, like patanna ‘of.feet’ or lú.mesˇis.gusˇ kin ‘golden grooms’ below: (18)
(ginuwas gad.hi.a) (patann=a _____ gisˇgìr.gub) for.knees veils of.feet=and stool ‘veils for the knees and a stool for the feet’ (StBoT 25.25 I 10) (ansˇu.kur.ra.mesˇ) (lú.mesˇis.gusˇ kin=ya _____ humandan) charioteers golden.grooms=and all ‘charioteers and all the golden-grooms’ (StBoT 24 ii 60–61)
Interestingly, Hittite has no non-postpositive counterpart to =ya for phrasal conjunction: all conjunctions below the clause are postpositive. The analysis that we have proposed moves words to an easily defined and independently motivated position, the beginning of a phonological phrase. The movement is driven by words that are postpositive and thus cannot occur phrase-initially. The fact that the moved element always comes from the following conjunct suggests that the conjunction is more closely connected to what follows than to what precedes, and this is what motivates the structure in (4).
3. Conjunction of clauses Our analysis of clausal conjunction parallels our analysis of subclausal conjunction. Our only assumptions are that conjunctions fall between their conjuncts and that postpositives cannot occur at the beginning of a phonological phrase. That said, there are three types of clausal conjunction to discuss here: cases where the conjunction is not postpostive, cases where it is postpositive, and cases where it is absent (asyndeton).
10
Brian Agbayani and Chris Golston
If the conjunction is not postpositive (Greek , Latin et, Hittite nu) it surfaces between the clausal conjuncts, as the syntax and semantics would lead us to expect (§ 3.1). However, if the conjunction is postpositive (Greek or =, Latin enim or =que, Hittite =ya or =ma), the first word of the second conjunct moves to the front of the phonological phrase that contains it (§ 3.2). Finally, if the conjunction is merely implied (asyndeton), nothing has to move because the phonological phrase that contains the rightmost conjunct already begins with a proper word; if there are other postpositives at the beginning of the clause, a word from that clause moves to keep the postpositives from occurring phrase-initially, just as it would if one of those postpositives were a conjunction (§ 3.3). 3.1 Normal conjunctions We begin with the simplest case, clauses conjoined by conjunctions that are not postpositive: (19)
[( % * π « ,«]CP λ [- M . ]CP there waited days three and came Menon ‘they waited there three days and Menon came’ (Xenophon, Anabasis 1.2.6)
Again, we take it as uncontroversial that falls between the clauses it conjoins and that noone will be tempted to claim that the verb - is in second-position just because it follows the conjunction orthographically. Simarly for regular conjunctions like et in Latin, which fall between their conjuncts as expected: (20)
[consulem interficerat]CP et [eius exercitum sub iugum consul had.killed and his army under yoke miserat]CP sent ‘he had killed the consul and sent his army under the yoke’ (Caesar, Bello Gallico 1.12.5)
11
Second-position is first-position
Hittite clauses conjoined by ubiquitous nu receive the same treatment: (21)
nu [=kán and =prt
Mursilin Mursilis
kuennir]CP killed
nu and
[eshar blood
ieir]CP shed
nu [Hantilis nahsariyatati]CP and Hantilis was.afraid ‘And they killed Mursilis and they shed blood and Hantilis was afraid’ (2 Bo TU 23 1 33–35) The proposed syntax for a case like this is as follows: (22)
CP [eshar ieir] they shed blood
& nu and
CP [Hantilis nahsariyatati] Hantilis was afraid
As long as the conjunction isn’t postpositive, nothing more transpires and everything is pronounced where it is interpreted. Note that while there is good reason to think that the conjunction and the following conjunct form a syntactic constituent of some kind (as well as a prosodic constituent), we do not take the conjunction to be part of the clause that follows. Hantilis is the first word in its clause, not the second. Now consider the first clause in (21), repeated below, with its syntactic bracketing: (23)
nu[=kán Mursilin kuennir]CP and =prt Mursilis they.killed ‘and they killed Musilis’ (2 Bo TU 23 1 33)
Orthographically (i.e, in cuneiform, not show here), the particle kán appears as a suffix on nu and is thus said to be second in the sentence. However, given the structure for coordination adopted here, nu (by virtue of its semantics and syntax) falls outside of the clause that =kán belongs to; this is another case where a postpositive (=kán) is clearly at the beginning of the clause. The only way to treat it as clause-second is to treat the conjunction as clause-initial, which is untenable.
12
Brian Agbayani and Chris Golston
(24)
[…]
nu and
[=kán Mursilin kuennir] =prt Mursilis they.killed
The same holds for all other sentential postpositive pronouns and particles in Hittite: they are clause initial and can be placed there by syntactic means. No notion of second-postion is required. Similarly for Greek, where postpositive pronouns like min are actually clause-initial, not in second position as traditionally claimed: (25)
[ φ. « * 01 ]CP and him addressing words winged spoke ‘and addressing him, he spoke winged words’ (Homer, Odyssey 15.259)
Again, if the conjunction lies outside of the clause, min is straightforwardly clause-initial: (26)
[…]
[ φ. « * 01 ]
Postpositives in Latin like enim ‘surely’ show the same thing once we realize that a conjunction is never the first word of its right-hand conjunct: (27)
at [enim nimis hic longo sermone utimur] but surely too.much here long speech we.use ‘But surely we are making our discussion too long here’ (Plautus, Trinummus 3. 3. 79)
(28)
[…]
at
[enim nimis hic longo sermone utimur]
If at sits between its conjuncts, as it must, enim sits at the beginning of its clause, not in second position, a notion that is no longer required. Returning to Hittite, this language has many more postpositives than the simple case above would suggest. (21) above shows the com-
Second-position is first-position
13
mon pattern, with the particle =kán immediately following the clausal conjunction nu. A short text illustrates how common this is and how clear it should be that Wackernagel clitics like =us ‘them’ and =as ‘he’ are clause-initial in the most mundane sense: (29)
nu [utnee arha tarranut] n[=us arunas and countries away strengthened and=them of.sea irhus ieit] boundaries made ‘And he weakened the countries and he made them boundaries of the sea,’ uru Halpa pait] nu n[=as [uruHalpan harnikta] and=he Aleppo went and Aleppo destroyed ‘and he went to Aleppo, and he destroyed Aleppo’ (2 Bo TU 23 1.27–28)
Each of these clauses is conjoined to the preceding clause with nu; the object =us ‘them’ (whose vowel forces elision of the vowel in nu) and the subject =as ‘he’ (whose vowel also elides the vowel in nu) are first in their respective clauses, clause-initial just like their cousins in Greek and Latin. These clause-initial postpositives are of course pronounced with the preceding conjunction, as the vowel elision clearly shows, but this doesn’t make them part of that conjunction any more than it makes the conjunction part of its right-hand conjunct. 3.2 Postpositive conjunctions Greek, Latin, and Hittite also have postpositive conjunctions, as we have seen, which provided much of the impetus for the traditional notion of second-position (except that Delbrück and Wackernagel didn’t know about Hittite, which had yet to be deciphered). Like their phrasal counterparts, postpositive clausal conjunctions like ‘and’ (often elided to ’ before vowels) cannot occur at the beginning of a phonological phrase and so require phonological movement of something else to phrase-initial position:
14 (30)
Brian Agbayani and Chris Golston
(* ’ _____ ( & ) (%0 _____ awoke and out sleep godly and
$ φ"’ 5 φ) him engulfed voice ‘he awoke from sleep and the godly voice engulfed him’ (Homer, Iliad 2.41) (γ« *φ’ #A «) (, ’ _____ so spoke Alcinous them and
(
pleased
%«) speech ‘so spoke Alcinous and his speech pleased them’ (Homer, Odyssey 13.16) 4 «) (γ« φ) (2 ’ _____ Ν ’ so spoke went and then Dream ‘so he spoke and Dream then left’ (Homer, Iliad 2.16) Again, we assume here that the conjunction always surfaces between its conjuncts, but cannot occur first in its phonological phrase. For this reason some word that follows (* , %0, , , 2) moves just past the postpositive to shield it from the left edge of the phonological phrase. As with the sub-clausal conjunctions discussed above, we merely assume that the base position for conjunctions is between conjuncts and that postpositives may not be phraseinitial. (31)
movement from the right conjunct
(γ« φ)
(2 ’
_____ Ν ’ 4 «)
Identical data occur in Latin, where the postpositive conjunction =que cannot occur phrase-initially and thus requires a word from what follows (duas and multos below) to move to phrase-initial position: (32)
(duas=que ibi _____ legiones) (conscribit) two=and there legions enrolls ‘and enrolls two legions there’ (Caesar, Bello Gallico 1.10.3)
Second-position is first-position
15
(arcebat longe Latio) (multos=que per ____ annos) kept far Latium many=and for years (errabant) wandered ‘she kept them far from Latium, and for many years they wandered’ (Virgil, Aeneid 1.31–2) Similarly for Hittite, with the conjunctions =ya ‘and’ and =ma ‘but’:2 (33)
(apass=a _____ ARAD) (DINGIR-LIM eesdu) he=and servant of.deity become ‘and may he be a servant of the deity’ (StBot 24 iv 79) (kedani=ma ANA _____ BULUG) (GIMan hasatarset this=but to malt as offspring NU.GÁL) not.exist ‘but as with this malt there is no offspring’ (KBo VI 34 II 31)
The conjunctions (=ya and =ma) logically belong between their conjuncts, as the syntax requires, the semantics makes clear, and the phonology fails to allow. A word following the postpositive (apass, kedani) must therefore move to the front of the phonological phrase to keep =ya and =ma from being phrase-initial. Other postpositive elements line up at the front of their clause following the conjunction, in what is clearly clause-initial position: (34)
2
(apiya=ya=at _____ QATAMMA=pat taparta) that=and =it same.way=very ruled ‘and at that (time) he ruled it in the very same way’ (KUB 14.4 i 11–12) (sessar=ma=wa=si akuwanna udandu) _____ beer=but=quot=him drinking they.bring ‘but ‘they will bring him beer for drinking’ he said’ (KUB 33 102 C II 26)
Hittite is a head-final language (OV with postpositions), so heads like eesdu ‘become’ and NU.GÁL ‘not exist’ are phrased phonologically with the XPs to their left (i.e., with their complements).
16
Brian Agbayani and Chris Golston
Again, =ya and =ma are in situ and lie outside of the clause, where sentential conjunctions belong. This puts the postpositives (=at, =wa, =si) in clause-initial position. Because =ya and =ma are themselves postpositive, some word to the right of =ya and =ma (apiya and sessar in these cases) are forced to move to the beginning of the phonological phrase. In these languages the postpositive conjunctions (Greek , =; Latin enim, =que; Hittite =ya, =ma) are always the first in any string of postpositives, a fact that we must clearly attribute to the syntax, where the conjunctive head is always external to the clause that contains the other postpositives. The postpositive elements that immediately follow the conjunction are syntactically initial in their clause and merely lean on the conjunction phonetically. There is no reason to think that this phonetic cliticization moves them out of their clause; it is just that they are parsed with the preceding material in speech. A parallel case might make the argument clearer. In an English sentence like Kate’s nice [keits nais], the auxiliary verb (‘is’) is phonetically cliticized to the subject, as shown by the voicing assimilation between it and the preceding [t]. Nonetheless, we do not count [s] as part of the grammatical subject, or claim that it is no longer part of the predicate, or say that it has moved out of the predicate to cliticize to the subject. Similarly for =at, =wa, and =si above: they are phonetically cliticized to the preceding word but belong syntactically with the clause that follows. Not one of these postpositives is second in its clause. The conjunctions (=ya, =ma) are in situ between their conjuncts and are not part of any clause. The rest of the postpositives (=at, =wa, =si) are clauseinitial, just phonetically enclitic on what precedes them. So far this is all completely parallel to conjunction below the clause. If the conjunction is a proper word, its conjuncts surface on either side of it. If the conjunction is a postpositive, a word from the following clause must be fronted past it. That word is usually a single word syntactically. But a phonological word (.) can also consist of one or more function words (preposition, article, etc.) and a following content word (noun, verb, adjective, adverb), as has been shown by Selkirk and others. The following Latin examples show how this works: (35)
(sub occasum).=que _____ solis before setting=and of.sun ‘and before the setting of the sun’ (Caesar, Bello Gallico 2.11)
Second-position is first-position
17
(ob eas).=que _____ res from those=and things ‘and from those things’(Caesar, Bello Gallico 2.35) (We assume that =que is subsequently incorporated into . but omit this for clarity). Note that such cases provide another strong piece of evidence that the movement here is not syntactic – sub occasum and ob eas do not form syntactic constituents here and so cannot have been moved syntactically: (36) =que [ppsub [np occasum [npsolis]]] and before setting of.sun =que [ppob and from
[np eas [npres]] those things
They do, however, form prosodic words, supporting the claim that the movement is phonological (Agbayani & Golston 2008, 2010). When function words are followed by words that are themselves XPs, they are incorporated into the phonological phrases those XPs form. We see this with (9 9 0), (( 7 * % ), and (λ 7 ’ 8) 7) in the examples that follow: (37)
0 ’) (_____ $
) ( (9 9 the tenth and to.assembly called μ ) (#A"1«) host Achilles ‘and on the tenth day Achilles called the host to assembly’ (Homer, Iliad 1.54) 7 * % ξ) (_____ $ ;) (( from the previous and consider ‘and consider this from the previous cases’ (Plato, Cratylus 389A) (λ 7 ’ 8) 7 ξ) (_____ 2 2 . ) also the near himself and barbarians (( ,) took.care.of ‘and he also took care of the barbarians near him’ (Xenophon, Anabasis i.i.5)
18
Brian Agbayani and Chris Golston
Note that these data argue against a morphological affixation account of conjunction placement, since the conjunction does not function as an affix that attaches to any kind of morphological or morphosyntactic word (Embick & Noyer 2001). As the final case above makes very clear, the apparent host (ob eas ‘from those’) is neither a morphological word nor a syntactic constituent. Thus, there is no straightforward morphological or syntactic source for postpositive conjunctions. The crucial observation here is that the postpositive conjunction is always preceded by something that is not itself postpositive, suggesting that what drives the surface position of the conjunction is neither morphological nor syntactic. 3.3 Asyndeton It often happens in these languages that things are conjoined asyndetically, without an overt conjunction. Consider the following from Greek, where the postpositive particle follows the adverb ‘then’: (38)
( _____ ¹ ) (% ) then prt him appeasing let.us.persuade ‘then let us persuade him by appeasing him’ (Homer, Iliad 1.100)
This clause is conjoined asyndetically to the preceding clause, i.e., without an overt conjunction like or . Since the elements
and are postpositives a proper word from what follows ( in this case) must be fronted. The same takes place in Latin, where the interrogative =ne is postpositive: (39)
(tantae=ne _____) (animis caelestibus) (irae) such=inter souls heavenly anger ‘is there such anger in the heavenly souls?’ (Virgil, Aeneid 1.11)
Since =ne is postpositive, a word from what follows (tantae) moves to the beginning of the phonological phrase. Hittite provides similar cases, including many with strings of postpositive elements like =mu=za=kan below: (40)
(istamassanzi=tta _____) listen=you ‘they listen to you’ (KUB 21.27 iv 31)
Second-position is first-position
19
(dingir-lum=mu=za=kán ____) (gasˇan-y) goddess=me=refl=prt lady-my (humandaza=pat daskisi) always=prt rescue2s.iter ‘goddess, my lady, you always rescue me’ (StBoT 24 i 50) All of the postpositive elements here are clause-initial. The single word that precedes them is not in situ but has been moved to the beginning of its phonological phrase to keep the postpositives postpositive. We can therefore do away with analyses that place these elements in “second-position in the sentence”. Note that from a syntactic perspective this is a desirable result, since second-position is hard to define in a configurational syntax (Keenan & Stabler 2001).
4. Conclusion This paper argues that the notion second-position is superfluous for early IE languages. We began with the notion that conjunctions are syntactically external to their conjuncts. For phrasal coordination this means that the conjunction is part of neither XP syntactically; for clausal coordination it means that the conjunction is part of neither clause. Once this is granted, it turns out that “second-position clitics” are uniformly found in clause-initial position, except of course for the postpositive conjunctions ( , =; enim, =que; =ya, =ma), which surface in situ between their conjuncts. Thus a proper understanding of conjunctions eviscerates the notion second-position because the relevant elements are demonstrably clause-initial. Second-position (in the sentence) is actually first-postion (in the clause). This allows us to dispense with the syntactically difficult notion “second-position” for the very languages it was designed for over a century ago.
References Agbayani, Brian, and Chris Golston. 2008. PF movement in Latin. Tenth Annual Conference in Diachronic Generative Syntax. Agbayani, Brian, and Chris Golston. 2010. Phonological movement in Classical Greek. Language 86.1: 133–167.
20
Brian Agbayani and Chris Golston
Anderson, Stephen. 1993. “Wackernagel’s Revenge: Clitics, morphology, and the syntax of second position”. Language 69.1: 68–98. Anderson, Stephen. 1996. “How to put your clitics in their place”. The Linguistic Review 13: 165–191. Bosˇkovic´, Zeljko. 2000. “Second position cliticization: Syntax and/or phonology?” Clitic Phenomena in European Languages ed. by Frits Beukema & Marcel den Dikken, 71–119. Amsterdam & Philadelphia: John Benjamins. Delbrück, Berthold. [& Ernst Windisch]. 1878. Syntaktische Forschungen. Vol. III. Halle/S/ Max Niemeyer. Embick, David & Rolf Noyer. 2001. “Movement operations after Syntax”. Linguistic Inquiry 32: 555–595. Franks, Steven & Tracy Holloway King. 2000. A Handbook of Slavic Clitics. Oxford: Oxford University Press. Garrett, Andrew. 1990. The Syntax of Anatolian Pronominal Clitics. Doctoral dissertation, Harvard University. Cambridge, Mass. Garrett, Andrew. 1996. “Wackernagel’s Law and unaccusativity in Hittite”. Halpern & Zwicky, eds. 1996. 85–133. Hale, Mark. 1996. “Deriving Wackernagel’s Law: Prosodic and syntactic factors determining clitic placement in the language of the Rigveda”. Halpern & Zwicky, eds. 1996. 165–197. Halpern, Aaron. 1995. On the Morphology and Placement of Clitics. Stanford: CSLI. Halpern, Aaron & Arnold Zwicky, eds. 1996. Approaching Second: Second position clitics and related phenomena. Stanford: CSLI. Hock, Hans Henrich. 1996. “Who’s on First? Toward a prosodic account of P2 clitics”. Halpern & Zwicky, eds. 1996. 199–270. Janse, Mark. 1992. “The prosodic basis of Wackernagel’s Law”. 15th International Congress of Linguists, Quebec, August 9–14, 1992. Johannessen, Janne Bondi. 1998. Coordination. Oxford: Oxford University Press. Keenan, Edward L. & Edward P. Stabler. 2001. Bare Grammar: A study of language invariants. Chicago: University of Chicago Press. Klavans, Judith L. 1982. Some Problems in a Theory of Clitics. Bloomington: Indiana University Linguistics Club. Luraghi, Silvia. 1990. Old Hittite Sentence Structure. London: Routledge. Luraghi, Silvia. 1998. “The grammaticalization of the left sentence boundary in Hittite”. The Limits of Grammaticalization ed. by Anna Giacalone Ramat & Paul J. Hopper, 189–210. Amsterdam and Philadelphia: John Benjamins. Luraghi, Silvia. 2001. “The development of local particles and adverbs in Anatolian as a grammaticalization process”. Diachronica 18:1. 31–58. Munn, Alan. 1993. Topics in the Syntax and Semantics of Coordinate Structures. Doctoral dissertation, University of Maryland. College Park, MD. Selkirk, Elizabeth. 1986. “On derived domains in sentence phonology”. Phonology Yearbook 3: 371–405. Selkirk, Elizabeth. 1995. “The prosodic structure of function words”. University of Massachusetts Occasional Papers 18: Papers in Optimality Theory ed. by Jill N. Beckman, L. Walsh Dickey, & S. Urbanczyk, 439–470. Amherst: University of Massachusetts.
Second-position is first-position
21
Taylor, Ann. 1990. Clitics and Configurationality in Ancient Greek. Doctoral dissertation, University of Pennsylvania. Philadelphia, PA. Taylor, Ann. 1996. “A prosodic account of clitic positioning in Ancient Greek”. Halpern & Zwicky, eds. 1996. 477–503. Wackernagel, Jacob. 1892. “Über ein Gesetz der indo-germanischen Wortstellung”. Indogermanische Forschungen 1. 333–436. Watkins, Calvert. 1964. “Preliminaries to the reconstruction of Indo-European sentence structure”. Proceedings of the 9th International Congress of Linguists ed. by H. G. Lunt, 1035–1045. The Hague: Mouton. Zoerner, Ed. 1999. “One coordinator for all”. Linguistic Analysis 29: 322–341.
Department of Linguistics California State University, Fresno 5245 N. Backer Ave. M/S PB92 F r e s n o , CA 93740 USA
[email protected] [email protected]
Brian Agbayani and C h r i s G o l s t o n
22
Paul W. Brosman, Jr.
The nt- participles and the verbal adjectives in *-toAbstract It is proposed that in Hittite the possession of passive meanings by the participles of verbs employed transitively was without exception and that it was the usage attested in Hittite which was inherited by both Anatolian and Indo-European. The change from passive to active which this view requires to have occurred in the meanings of the participles of transitive verbs in Indo-European is explained as having originated among verbs which could be used either transitively or intransitively. It is considered to have led in its turn to a compensatory change to almost uniformly passive of the originally variable meanings of the verbal adjectives associated with transitive verbs, which resulted in the incorporation of the verbal adjectives in the verbal system.
In both Indo-European and Hittite there occurred participles derived with the suffix -nt-. Although these forms are held to have been related, it is well known that they differed strikingly in function. Throughout Indo-European they were invariably active in meaning whether formed from intransitive or transitive verbs. In Hittite, on the other hand, although they could be formed from intransitive verbs with meanings which were necessarily active, when belonging to transitive verbs they were primarily, if not uniformly, passive in meaning. It is commonly held that two exceptions to the exclusively passive meaning of the Hittite participles of transitive verbs are known to have existed, the forms of eku- ‘drink’ and ed- ‘eat’, which could occur as either active or passive (Friedrich 1960: 144–5; Kronasser 1966: 257; Sihler 1995: 613–4). However, it seems likely that the apparent exceptions should be otherwise explained. In many languages there are verbs which can function as both transitive and intransitive. Among numerous examples showing that Hittite should be included among them one may cite arai-, ariya- ‘rise, raise’, arriya- ‘rouse, be awake’, asas-, ases- ‘seat, set, sit’, es-, as- ‘sit, inhabit, occupy’ and iya-, ie-, i- ‘do, make, act’ (Puhvel 1984–: 1.123, 138, 207; 2.291, 335). In a number of these languages, including contemporary English and German, two such verbs are those for ‘eat’ and ‘drink’. It thus is plausible that the same was true of Hittite. That the forms of eku- and ed- cited by Puhvel support the view that this was in
The nt-participles and the verbal adjectives in *-to-
23
fact the case is indicated by, among others, the following passages: kinun-za edmi ekumi ‘now I eat [and] drink’; parna-ssa paisi eszi euksi ‘you go to his house, eat [and] drink’; kuwat-za U¯L ez[atti] [kuwat] U¯L ekutti ammel isha-mi ‘why do you not eat, why do you not drink, my lord?’; nu U¯L ekutta ‘you did not drink’; SAL-za U¯L ezzazi ‘the woman does not eat’; namma-at-za adana esantari ‘then they sit down to eat’; nu-wa-za ezza ‘now eat!’ (Puhvel 1984–: 2.261–3; 316–7). It therefore seems significant that no object of the verbs is expressed in the examples cited by Puhvel (1984–: 2.264, 317) of the occurrence of the participles of eku- and ed- with active meanings. In view of this evidence it appears that in Hittite the participles of transitive verbs were consistently passive in meaning and that the occurrences of those of eku- and ed- with active meanings are to be explained as instances of the intransitive use of verbs which could be employed either transitively or intransitively. The nt- participle is attested also in Luwian, where it often, but not always, appears with the extension in -i- frequent in other Luwian nominal forms, and in Hieroglyphic Luwian. Since in each case it possessed a passive meaning when belonging to transitive verbs (Friedrich 1960: 193), it seems clear that the Hittite usage was inherited from Proto-Anatolian. That the contrasting Indo-European usage was also a common inheritance seems equally safe to say, for it or vestiges of it occurred in every traditional Indo-European dialect (Brugmann 1906: 454–61; Sihler 1995: 613–4). Presumably one should conclude that one usage or the other prevailed in Proto-Indo-Hittite and was inherited by both Indo-European and Anatolian, where it was retained unchanged on one side and fundamentally altered on the other. Such a conclusion seems clearly preferable to the only apparent alternative that the meanings of the participles of transitive verbs inherited from Proto-Indo-Hittite had somehow come to be active in some cases and passive in others and that they then evolved in different directions in Indo-European and Anatolian. Although a decision as to which of the two attested usages should be held to have originally been inherited cannot yet be made with a similar degree of confidence, it appears that the one found in Hittite should be regarded as distinctly the more likely choice. It is difficult to see how the absolute uniformity of meaning seen among the IndoEuropean participles could have begun to break down in the absence of confusion brought about by phonological change, a development
24
Paul W. Brosman, Jr.
which clearly did not take place. On the other hand, a system such as that of Hittite, in which transitive and intransitive verbs were treated differently, would have contained a potential source of instability which would have permitted the possibility of its alteration analogically, especially since the same verb can often function both transitively and intransitively. Even without the support of the apparent examples of eku- and ed- it is plausible that in such a system the participles of a number of verbs would have occurred with meanings which could be either active or passive in accord with the way in which the verbs were employed. It is among such verbs that a spread of active participles from intransitive to transitive verbs is most likely to have begun. Among them it presumably would have taken place first in a few verbs which for various reasons occurred often as intransitive and only occasionally as transitive, since the usual occurrence of their participles as active would have encouraged the apparent regularization of the rarer instances in which they were passive. Once some of these verbs had acquired in this way participles which were consistently active in meaning, the process could have been extended to others capable of occurring with participles of either type. A continuation of its extension would have led eventually to a point at which the system had been altered to one in which verbs employed as intransitive, exclusively or not, typically possessed active participles and passive participles were virtually restricted to verbs occurring exclusively as transitive. In other words, the basis for determining whether a participle was active or passive had become the verb to which it belonged rather than its use in a particular construction. After this stage had been reached, little explanation would be required for the replacement of the remaining passive participles by active forms in the interest of uniformity. Although the proposal here of the manner in which a system identical to that of Hittite could have been converted into the one found in Indo-European appears plausible, it cannot be considered acceptable without corroborating evidence. In this connection it appears potentially significant that the usage which evolved among the Indo-European verbal adjectives in *-to- resembled that found among the ntparticiples in Hittite, since each form could be derived with active meanings from intransitive verbs but when belonging to transitive verbs was virtually or entirely passive in meaning. Because Hittite possessed no forms corresponding to the verbal adjectives of Indo-
The nt-participles and the verbal adjectives in *-to-
25
European, it appears possible that in Indo-European the participles underwent the proposed change of meaning when the verbal adjectives arose there and began to acquire an approximation of their original function. It thus is pertinent to examine the history of the verbal adjectives to see if it could be related to developments among the participles in Indo-European. That the verbal adjectives in *-to- were an Indo-European innovation seems clear. Not long ago I discussed briefly their origin and that of the verbal adjectives in *-t- which presumably served as their immediate source in an article supporting the view that Anatolian separated from the parent linguistic stock significantly earlier than the languages related to it (Brosman 2002). At that time it was noted that verbal abstracts derived with the suffix -t- are attested in both Hittite and Indo-European but that Indo-European also possessed a variety of related forms which included, in addition to the verbal adjectives in *-t- and *-to-, verbal abstracts in *-to- (m.), *-to¯(i)-, *-tiand *-tu-, and adjectival abstracts in *-to- (n.) and *-ta¯-, none of which is found in Hittite. It was proposed that the latter forms provide eight distinct examples of common Indo-European innovations, which in combination with two other groups of innovations of similar size appear adequate to demonstrate the early separation of Anatolian. In this connection it was necessary to discuss the origin of each of the proposed innovations. Because of the number of such discussions required, the evidence concerning the adjectives in *-t- and *-to- was restricted to a minimum which seemed sufficient to indicate that they were indeed Indo-European innovations. However, further material pertinent to their origin and to subsequent developments related to the manner in which they had arisen could have been cited and will be included here in a fuller history of the two verbal adjectives, a subject which is regarded as of interest in its own right in addition to its possible relevance to questions concerning the nt- participles. Prior to the discovery that in Hittite verbal abstracts in -t- are attested but that the identically formed adjectives are not, Brugmann (1906: 422) had already suggested that among the t- forms the nouns were the earlier and that the adjectives had arisen from them through use of the nouns as the second element in adjectival compounds. One can apparently be confident that he was correct. One reason is that, as he pointed out, among the attested adjectives, in contrast to the nouns, compounds predominated by a wide margin. Another is that
26
Paul W. Brosman, Jr.
compound origin would permit an explanation for the remarkable variation in the meanings of the t- adjectives, which when derived from transitive verbs could be either active or passive (Brugmann 1906: 422), since the first element of the compound would provide a context making interpretation of the second in one way or the other appear appropriate. For example, ‘god-praise’ would clearly refer to praise for the gods and thus could be employed to mean ‘god-praising’, whereas ‘god-punishment’ would refer to punishment by the gods and could be used to mean ‘god-punished’. Since a single noun could often be used in more than one adjectival compound, the same form could sometimes be used either actively or passively in accord with the differing meanings of the elements with which it was combined, as is seen in Gk. $ --2« ‘shield-throwing, shieldthrower, deserter’ and -2« ‘thrown forward, prominent’. When simple adjectives were subsequently abstracted from the compounds, they would have been active or passive depending on their use in the compounds from which they had been taken. An adjective could therefore emerge with both meanings in cases in which the same form had been abstracted from two separate compounds in which it functioned differently. It is in this way that one can account for adjectives such as Gk. $ <« ‘unknown, not knowing’, which is attested beside similarly ambiguous examples among the adjectives in *-to-, such as synonymous Gk. Ν .«. It may be noted that because in the attested Indo-European languages the compounds which had served as the vehicles for the conversion of the abstracts into adjectives still greatly outnumbered the simple adjectives, the latter were presumably of relatively recent origin. That Hittite contains virtually no evidence of nominal composition of any sort (Sturtevant 1951: 67; Friedrich 1960: 42) provides further support for the view that they were an Indo-European innovation. Like the verbal adjectives in *-t-, those in *-to- could be either active or passive when formed from transitive verbs (Brugmann 1906: 422). The citation above of the ambiguous pair Ν .« and $ <« shows the degree to which the parallelism between the two types extended. Since the t- and to- forms functioned in precisely the same manner, it seems clear that the latter originated through thematicization of the former. In this case there can be no doubt that the adjectives in *-to- were an Indo-European innovation, for adjectives were not thematicized in Hittite (Brosman 1998: 75; 2002: 12).
The nt-participles and the verbal adjectives in *-to-
27
Another conceivable source of to- adjectives should be mentioned here, but only as a possible supplement, rather than alternative, to the explanation above, for it would not exclude thematicization of the adjectives in *-t-. Since abstract nouns in *-to- were included among the Indo-European forms identified earlier as proposed innovations, it might seem that some of them could have been converted into adjectives in the same manner as were their athematic counterparts. However, such a development appears unlikely to have taken place, for the occurrence of a large number of compound adjectives in *-t-, which presumably had served as the vehicles for the conversion of the athematic nouns, was not duplicated among the adjectives in *-to-. As the products of thematicization the verbal adjectives in *-towere at first confined to roots which had possessed derivatives in *-t-. However, they proliferated, spreading beyond the roots among which they had originated and becoming incorporated in the verbal system, where those belonging to transitive verbs came to be used primarily, if not exclusively, as passive. By the end of the period of common development they could apparently be formed from any primary verb. That their originally variable meanings became nearly uniform during their expansion may be readily explained, for they could not have proliferated as long as they continued to possess two such incompatible meanings as active and passive. In particular, their incorporation in the verbal system required that they be assigned a specific role there in order for them to function within the system. However, it is not so immediately clear why the meaning which survived was passive. Also remaining to be explained is why the to- forms began to become productive at a time when they still possessed their disparate original meanings. Although it is recognized that other forms are known to have proliferated for no apparent reason, in the present case such a development appears remarkable. For example, the German plural in -er is uniformly held to be the reflex of the derivational suffix of s- stem nouns, though for reasons unknown it spread from a handful of Old High German forms which included only one inherited s- stem (Lehmann 1973: 193). However, the productive German suffix was associated with the specific meaning of plurality when its expansion began. As has already been noted, that the original to- adjectives possessed two conflicting meanings should have been an obstacle to their expansion. It is also unclear whether their incorporation in the verbal system was the ultimate result of their extreme productivity or was its cause.
28
Paul W. Brosman, Jr.
It is at this point that developments among the verbal adjectives and the participles could plausibly have become related, for the change in the meaning of the participles which was suggested here earlier would provide answers for each of the questions concerning the history of the verbal adjectives. That the passive meaning was being eliminated among the participles would account for its survival and expansion among the verbal adjectives in order to compensate for the loss of passive non-finite forms. The assumption in this manner of the role of the participles by the verbal adjectives could have caused the early incorporation of the latter in the verbal system, to which the participles already belonged. In its turn the entry of the verbal adjectives into the verbal system for this reason would provide an obvious explanation for their proliferation.1 That the meanings of the participles and verbal adjectives would have evolved in opposite directions more or less concurrently appears improbable. However, it is not necessary to hold this to have been the case. Since the participles were inherited and the verbal adjectives were not, the developments proposed for the participles could have begun for the reasons of their own suggested here earlier and have progressed appreciably before the verbal adjectives originated. It thus is possible that the final stage, at which the participles of verbs consistently transitive were being altered from passive to active, had been reached by the time that the to- adjectives began to expand. Another significant difference between the two forms was that the inherited participles could be formed from any verb but the original verbal adjectives were restricted to roots which had possessed derivatives in *-t-.
1
That developments among the participles and verbal adjectives were related was proposed some time ago by Kronasser (1956: 210), who held the consequences of the relationship to have been the reverse of those suggested here. On the basis of his belief that the verbal adjectives had been lost in Hittite, Kronasser explained the Hittite passive participles by holding that in Hittite the participles had taken over the function of the verbal adjectives when the latter disappeared. However, he did not attempt to describe the process through which the change of meaning among the participles was accomplished. A sufficient reason for rejecting this proposal is that the verbal adjectives clearly were an Indo-European innovation. Kronasser himself appears later to have abandoned his explanation for other reasons, for there is no mention of it when he discusses the participles in his second volume on Hittite historical grammar (Kronasser 1966: 256).
The nt-participles and the verbal adjectives in *-to-
29
It is probable that the expansion of the verbal adjectives began when a few of the transitive verbs which did not possess verbal adjectives of any sort were provided with passive forms to compensate for their loss of passive participles. Since passive verbal adjectives already existed, they could have provided the model for the new forms, even though they were not the only type paired with transitive verbs. A continuation of this process in combination with completion of the elimination of the passive participles would eventually have caused every transitive verb which had originally had no to- form beside it to possess a passive verbal adjective. By the time that the stage described above had been reached, the passive forms would have become clearly predominant among the verbal adjectives occurring beside transitive verbs. After their dominance had become unmistakably established, it is probable that the original verbal adjectives possessing active meanings would have begun to be converted from active to passive when associated with transitive verbs. The initial conversions would have been facilitated by the previously noted existence of at least a few forms which had originally possessed both active and passive meanings, since the change of an individual form from active to passive would have been carried out gradually rather than accomplished abruptly among all its occurrences. In such a progressive conversion the essential first step, in which a form previously employed exclusively as active began to be used also as passive on occasion, would not have resulted in an innovation, since forms of that sort already existed. Thereafter, whatever the ultimate fate of the original forms with dual meanings, there would at all times have been partially completed conversions in progress to provide similar models for the first step in later conversions. Prior to the end of the period of unity the forms with active meanings had in this manner been altered in meaning to passive virtually without exception. The citation here of Gk. Ν. « shows that the transformation of the meanings of the verbal adjectives associated with transitive verbs from variable to passive fell slightly short of complete. However, a number of other forms, such as Lat. po¯tus, Skt. pitá- ‘having drunk’, Lat. pransus ‘having lunched’, Skt. bhaktá- ‘having eaten’ and Lat. circumspectus ‘circumspect, cautious’, which have been cited as apparent exceptions (Friedrich 1960: 145; Puhvel 1984–: 2.264, 317; Sihler 1995: 622), are probably to be explained for the most part in the same manner as were the active occurrences of the participles of Hitt.
30
Paul W. Brosman, Jr.
eku- and ed- rather than as relics of the active use of verbal adjectives belonging to transitive verbs. It should be noted that the previously proposed conversion from passive to active of the participles of verbs used exclusively as transitive could have been carried out in a fashion similar to that suggested above for the conversion of the verbal adjectives in the opposite direction, though the models for the initial conversions would have originated differently. In the case of the participles it is likely that the generalization of active meanings among those belonging to verbs used both intransitively and transitively had not been accomplished entirely without exception before the first exclusively passive participles of verbs uniformly transitive began to undergo a change in meaning. There thus should have remained a few participles occurring as both active and passive when belonging to a verb being employed transitively at the time that the conversion of those of the permanently transitive verbs started to take place. The few remaining forms of this sort could have provided models for the first step in the initial conversions of the passive participles in the same manner as did the few to- adjectives of dual meanings which had resulted from the thematicization of synonymous t- adjectives in the case of the conversion to passive of the consistently active verbal adjectives. At the same time as the meanings of the verbal adjectives were being altered from variable to nearly uniform, a similar development was taking place with respect to their form. Although it has been said that they originally contained unaccented zero grade of the root (e.g., Sihler 1995: 621, 623), it seems safe to say that this was not the case. As the products of thematicization, the original to- adjectives would have contained the same grade of the root as the t- forms from which they had stemmed. Since the t- adjectives were by no means associated with a particular grade of the root (Brugmann 1906: 422–4), there should originally have been much variation in this regard among the adjectives in *-to-. However, when the latter began to spread beyond the roots among which they had originated, the speakers were free to choose the form in which to cast them. With a high degree of consistency the ultimate result of this choice was zero grade of the root with the accent upon the suffix. In attempting to account for the generalization of the zero-grade forms it may be noted that, if the proposal concerning the spread of the passive meaning is correct, all of the verbal adjectives produced
The nt-participles and the verbal adjectives in *-to-
31
during their expansion were passive forms created to replace the passive participles which were being shifted to active in meaning. It thus is plausible that the form of the newly derived adjectives would have been influenced by that of the participles they were replacing. In that case there presumably would have been at least some degree of hesitation concerning the grade of the root of the verbal adjectives at first, since the participles were not uniform in that regard. However, it is probable that at the time that the expansion of the verbal adjectives began, which may be held to have been well before the end of the period of unity, athematic verbs remained distinctly in the majority. It therefore seems likely that the reason that zero grade of the root came to prevail among the verbal adjectives is that zero grade of the root occurred in the participles of the root verbs and zero grade of the stem in those of the other athematic types. Apparently the only other possibility is simply that zero-grade forms were somewhat more common among the original to- adjectives, though the evidence of the t- forms cited by Brugmann indicates that their margin of superiority would not have been great. In any event, the forms produced during the expansion of the verbal adjectives presumably possessed zero grade of the root in a steadily increasing majority of instances until the point was reached that every adjective derived thereafter occurred with zero grade. With the addition of many newly derived zero-grade forms, forms of that type became by far the most numerous among the verbal adjectives as a whole. After the zero-grade type had thus been established as predominant, its dominance was extended as most forms of other types were modified analogically to conform to it. However, not all forms originally possessing other grades were eliminated. Although there is no reason to question the accepted view that most attested deviations from zero grade resulted from various analogies within the separate dialects (Brugmann 1906: 397; Sihler 1995: 623), it is possible to point to forms which may confidently be identified as full- or lengthened-grade relics of the original to- adjectives. Among them are Gk. .« = Skt. jña¯tá- = Lat. no¯tus ‘known’ and Skt. pra¯tá= Lat. pletus ‘filled, full’, which are indicated by the correspondences among them to have been inherited and presumably confirmed as relics by Gk. $ <« and Lat. locu-ples ‘full of landed property, possessing large landed property’, which contain the adjectives in *-tfrom which they were formed. There are also isolated but apparently
32
Paul W. Brosman, Jr.
reliable examples such as Gk. .« ‘swimming’ beside <« ‘swimming, a fish’ and Lat. mansuetus, mansues ‘tame’. Since it seems safe to say that a number of t- adjectives were eliminated prior to their attestation by the occurrence beside them of synonymous adjectives in *-to-, it is reasonable to assume that there also exist original to- adjectives which cannot be identified in the same manner because the tforms from which they stemmed failed to survive. It thus is probable that most, if not all, to- adjectives attested as full- or lengthened-grade forms which through correspondences elsewhere can be established as inherited have retained their original grade of the root unmodified, whether they acquired it through thematicization of a t- form or as the product of an early stage in the expansion of the forms in *-to-. Unlike zero grade of the root, the other formal feature of the verbal adjectives, the accented suffix, presumably was usual, though not uniform, among the original to- forms. The reason that *-to- probably was accented at first is that the same was generally true of the other adjectives among the o- stems, where adjectives were commonly distinguished from substantives by being accented upon the suffix as opposed to the root (Brugmann 1906: 27–8). Since the o- stems as a class were inherited and the to- forms were not, it is likely that when they came into existence, the to- adjectives were usually made to conform to the prevailing pattern. In this connection it may be observed that *-to- was not a distinct suffix until it arose through resegmentation when the to- adjectives began to expand beyond the roots among which they had originated. Support for the view that the two features of the verbal adjectives arose separately is provided by the forms identified above as original to- forms, Gk. .«, Skt. jña¯tá-, Skt. pra¯tá- and Gk. .«, each of which was accented upon the suffix but did not contain zero grade. From what has been seen here, the history of the verbal adjectives apparently provides an appreciable measure of corroboration for the proposal concerning the Indo-European participles, since a number of questions raised by developments among the verbal adjectives, including those regarding their proliferation and incorporation in the verbal system, as well as the generalization among them of the passive meaning and of zero grade of the root, may each be answered on the basis of the change of meaning proposed to have occurred among the participles. In addition, it is possible to answer on the same basis the question which is fundamental to the others, that of why and how such
The nt-participles and the verbal adjectives in *-to-
33
an unlikely candidate for productivity as the verbal adjectives, which occurred with two conflicting meanings, would have begun to undergo expansion at all. Although the implications of the evidence concerning the verbal adjectives seem significant, the corroboration which they provide is indirect. However, it is possible to point also to an apparently credible relic of the proposed inherited passive participles in Skt. pr0´sˇant‘speckled, spotted’, one of the handful of nominal forms which are participial in form but not in function (Brugmann 1906: 459–61). That the meaning of pr0´sˇant- could have developed from that of an active participle appears improbable. However, since the verbal root which the form contains could occur as transitive ‘sprinkle, scatter, bespatter’ (Pokorny 1959: 823), its meaning of ‘speckled, spotted’ may readily be explained as stemming from that of the passive participle of a verb used transitively. Although it is apparently the only such example which can be cited, its isolation does not appear to present a problem in view of the small number of potential relics attested. Included among the latter are possible additional examples which, however, seem too uncertain to be considered reliable. Pokorny (1959: 390) suggests that the root of Skt. járant- ‘frail, old, old man’ = Gk. . ‘old man’ originally meant ‘rub, grind’ and cites Skt. járati ‘makes frail, causes to grow old’ (< *grinds, wears away’), which is suitable as the reflex of a transitive verb whose passive participle would have yielded járant-. However, the meaning of járant- and . could also have developed from that of the active participle of a verb meaning ‘become frail, grow old’, as has been assumed previously. Since verbs of this sort are attested containing the same root, though not in a form so precisely appropriate, one cannot be entirely certain how the original participle arose. In the case of Skt. br0 hánt ‘tall, high, large, strong’, the transitive verb Skt. br0 mháti ˙ ‘strengthen, enlarge, elevate’ is one of two pertinent verbs attested (Pokorny 1959: 140). However, because the other is Skt. barháyati ‘augment’, apparently the causative of intransitive ‘become large’, the origin of br0 hántalso appears uncertain. With the addition of the element of apparent direct support supplied by pr0´sˇant- to what was seen in connection with the verbal adjectives, it appears that the proposal concerning the Indo-European participles deserves to be considered probable. It should also be noted that, whatever the acceptability of this proposal and of others made
34
Paul W. Brosman, Jr.
here, it seems clear that the contrast in usage between the participles of Indo-European and Hittite provides an additional indication of the early separation of Anatolian. Since it was seen at the outset that one usage or the other presumably was inherited on both sides, one can apparently be confident that a common innovation took place in one instance, whether it should be identified as Indo-European, as was suggested here, or as Anatolian. The unlikely alternative that the inherited usage constituted some sort of middle ground between the two attested extremes and then evolved in different directions in Indo-European and Anatolian would not alter this conclusion, for in that case a common innovation would have occurred on each side.
References Brosman, Paul (1998): On the origin of the PIE o- stems. In: Folia Linguistica Historica 19.65–78. –, (2002): Evidence in support of Proto-Indo-Hittite. In: Folia Linguistica Historica 23.1–21. Brugmann, Karl (1906): Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. 2nd ed. vol. 2. pt. 1. Strassburg: Trübner. Friedrich, Johannes (1960): Hethitisches Elementarbuch. 2nd ed. Heidelberg: Winter. Kronasser, Heinz (1956): Vergleichende Laut- und Formenlehre des Hethitischen. Heidelberg: Winter. –, (1966): Etymologie der hethitischen Sprache. vol. 1. Wiesbaden: Harrassowitz. Lehmann, Winfred (1973): Historical linguistics: an introduction. 2nd ed. New York: Holt, Rinehart and Winston. Pokorny, Julius (1959): Indogermanisches etymologisches Wörterbuch. vol. 1. Bern: Francke. Puhvel, Jaan (1984–): Hittite etymological dictionary. Berlin and New York: Mouton de Gruyter. Sihler, Andrew (1995): New comparative grammar of Greek and Latin. Oxford: Oxford University Press. Sturtevant, Edgar (1951): A comparative grammar of the Hittite language. 2nd ed. vol. 1 New Haven: Yale University.
1550 2nd St. N e w O r l e a n s, Louisiana 70130 USA
[email protected]
P a u l W. B r o s m a n , J r .
Postpositions and agreement in Rajasthani languages
35
Postpositions and agreement in Rajasthani languages: A comparative analysis of alignment features in Marwari and Harauti Abstract Alignment differences in two Rajasthani languages, Marwari and Harauti, illustrate the distinct functioning of two mechanisms that determine alignment, case marking and agreement. In both languages, the agreement pattern is consistently ergative in sentences constructed with a perfect finite verb form. However, the argument marking pattern differs between the two. Arguments in Rajasthani are either unmarked or marked with a postposition. In the related language Hindi, postpositional markings have been well-identified as case marking. If this were also the situation in Rajasthani, the marking would be accusative. In Marwari, such alignment would form an exception to the prototypical definition of ergativity, which says that ergative agreement is not (or rarely) combined with accusative case marking. To solve this contradistinction, I argue that the argument marking postposition nai in Marwari and Harauti, contrary to Hindi, does not display all the features of the traditional definition of case marking; for instance, the postposition does not indicate a relationship between the head it marks and the verb, but rather serves other purposes. In Marwari, the occurrence of the postposition is subject to Differential Object Marking based on semantics, and in Harauti the postposition is even less of a case marker, as the same postposition may be used for A and O. This points to the conclusion that postpositional marking in Rajasthani is first and foremost a semantic marking that adds an extra dimension to the argument it marks and has no influence on the relationship between the argument and the verb. The latter relationship is expressed by the agreement pattern, which fulfills a distinguishing function.
1. Introduction1 Alignment refers to the way in which the intransitive subject (S) is aligned either with the transitive subject (A) or with the Direct Object (DO).2 S, A, DO and IO are here considered to be the core arguments 1
2
The following abbreviations are used in this article: cs: causative; cv : converb; f : feminine; impf: imperfect; m: masculine; nom: nominative; obl : oblique; pres : present; sg : singular. A more common designation for the patient argument is O. However, in the IA languages there is a functional difference between the grammatical relations of Direct Object (DO) and Indirect Object (IO), and in some IA languages (e.g., Kashmiri) there is agreement with IO. In this context the term DO is preferred.
36
Saartje Verbeke
of a sentence (cf., Van Valin and LaPolla 1997: 26–30). Two broad types of mechanisms are included under alignment: case marking and agreement. Both are considered to be mechanisms that treat (align) S as (with) A or as (with) DO (cf., Bickel and Nichols 2009: 305). If A is treated in the same way as S, the alignment of the language is accusative. Conversely, if DO is treated in the same way as S, the alignment of the language is ergative. These are not the only options, however. A, S and DO could all be treated differently or all in the same way, and different types of alignment may occur in one language (cf., Dixon 1994, Comrie 1978). The Indo-Aryan (IA) languages in particular illustrate how more than one kind of alignment can be used in a single language. This article focuses on the less-studied argument structure of two IA languages belonging to the Rajasthani3 language group, Marwari and Harauti. They are closely related to Hindi, but their argument structure is remarkably different. However, like Hindi (and most IA languages), these two languages display a split in alignment based on tense/aspect; the alignment of the present tenses differs from that of sentences constructed with a participle which indicates perfect aspect and/or past tense. Notably, in Hindi and Rajasthani, ergative features only occur when there is a form of this perfect participle in the verb construction; the participle may occur alone or combined with a present or other form of the verb ‘to be’. In combination with this auxiliary, the perfect participle functions as a participle indicating the perfect aspect. However, when the perfect participle is used as a finite verb (without an auxiliary), it is generally interpreted to indicate a simple past tense (cf., Davison 2002). Tense and aspect thus overlap in IA, and the perfect participle is necessary for the past tense as well as for an ergative construction. The past progressive, on the other hand, is not constructed with a perfect participle; sentences with a past progressive therefore do not show any ergative feature. The tense/aspect split is one characteristic of variable alignment in the IA languages (cf., Garrett 1990). Other splits in alignment are related to the different functions of agreement and case marking.
3
The name ‘Rajasthani’ was coined by Grierson (1906–1928) to describe the Indo-Aryan languages spoken in the area of Rajasthan, a state situated in the Western part of India.
Postpositions and agreement in Rajasthani languages
37
Agreement as a mechanism of alignment is also known as cross-referencing (e.g., in Dixon 1994) or head-marking. If the verb agrees with A or S, the agreement is considered to be accusative. If it agrees with the DO or S, it is considered to be ergative. Klimov (1973, Tournadre 1996: 17) speaks of “verbal ergativity” when the agreement indicates an ergative pattern. When the markings on the arguments themselves show that A is marked in the same way as DO, Klimov’s terminology refers to “nominal” ergativity. “Mixed ergativity” is the third option, where the verbal agreement as well as the case marking display ergative features. These three kinds of ergativity indicate again that splits in alignment may occur on every level, here in particular between case marking and agreement. Ergative agreement or verbal ergativity may seem to be a contradictory concept, because ergativity in the first place refers to the (ergative) case marking of A. However, in Rajasthani, the ergative type of agreement with the DO is a much more consistent way of indicating grammatical relationships than ergative case marking. Verbal ergativity in Rajasthani is easily noted, while the marking of the arguments themselves is more complicated and in general serves a different function than agreement. This process can be clearly seen in case marking. This article defines “case” as “the systemic differences in the form of all declined nominals, adjectives and participles, corresponding to differences in their functional and categorial status” (Dik 1978: 158). In a traditional definition, case marking refers to inflectional marking, i.e., a change in the form of the noun, adjective or participle (cf. Blake 1994: 1, Dik 1978: 158). However, the concept of case has been expanded to other forms of marking, including adpositional marking (cf. Blake 1994: 9) or marking by means of clitics (Butt 2006: 11). In the IA languages, one generally discerns two different inflectional case markings: a nominative or direct case and an oblique case. Historically, other inflectional case markings were present, but they have disappeared through merging. In many IA languages, postpositions have been added to the oblique form of the noun as if they were replacing the lost case inflections. There are two syntactic kinds of postpositions, those that are marking core arguments and those marking adjuncts. Although the same argument-marking postposition occurs in Hindi, Harauti and Marwari, its distribution differs in each language. In some instances, this distribution is more determined by semantic criteria than that the postposition is used as an index of a grammatical
38
Saartje Verbeke
relation. This observation leads one to ask whether postpositions behave in the same way as inflectional case marking, or if their occurrence is determined by different factors. I argue that postpositions in Marwari and Harauti are primarily semantic markers. The argument that postpositions are not case markers is partly based on Spencer (2005, see also Lidz 2006). Spencer argues from an LFG perspective that the postpositions in Hindi cannot project a phrase and thus lack a substantial feature belonging to a case marker. In Spencer’s view, the oblique case fulfills the case-marking function. However, Spencer’s argument is directed at postpositions in Hindi, while I contend that the postpositions in this language are more evolved towards a case marker than the postpositions in Rajasthani. This paper is constructed in the following way. Section 2 presents theoretical terms and background concepts that are used in the body of the article; these concepts are illustrated with examples from the well-established argument structure in Hindi. Section 3 provides examples of the argument structure in Marwari and outlines the different functions of case marking and agreement. Section 4 examines alignment in Harauti to illustrate similarities and differences in the use of postpositions in closely related languages. This comparison of Harauti and Marwari emphasizes the semantic role of the postpositions. Section 5 concludes the paper.
2. The function of case markers and agreement block: an illustration from Hindi The historical merging of inflectional cases and the rise of postpositions in Old and Middle IA continue to be evident in Modern IA, particularly in the system of “layers”. Masica (1991: 231) discerns three layers of different markers in IA. Layer I suffixes are the inflectional cases that remain after merging. In Hindi, the nominative and oblique cases form layer I. Additionally, the oblique case never occurs on its own in Hindi; it always takes a postpositional marking. Postpositions form layer II or layer III. Layer II includes simple postpositions that consist of a clitic and mark the core arguments; an example is the ergative postposition ne in Hindi. Layer III postpositions are combined postpositions that consist of a layer II postposition combined with a
Postpositions and agreement in Rajasthani languages
39
lexical element, as in the expression us=ke=nice, ‘this=layer II=under’ > ‘under this’. Postpositional marking in IA differs from inflectional marking in the sense that postpositions are invariable clitical forms that are added only to the head of the phrase and not to every part of it. For instance, in Hindi one would say is laflrki=ko, ‘this girl=to’, and not *is=ko laflrki=ko, ‘this=to girl=to’. While postpositional clitics (layer II) do not formally behave as cases in the strict sense of case inflections (layer I), it has been argued that functionally these clitics are on par with cases because they also indicate grammatical relations (cf., for Hindi, Mohanan 1994: 59, Butt and King 2004: 170). Layer III postpositions are adjunct markers and do not occur in combination with core arguments. In general, the case marking of core arguments, i.e., A, S, and O (DO and IO), has two functions: to characterize the argument it marks and to distinguish one core argument from the other in the sentence. Of these two functions, the latter one has been considered to be the most important (cf., Dik 1978: 161, Comrie 1989, Siewierska and Bakker 2009: 291), because a distinguishing function of case marking is the most economic one. In IA, only layer I represents formal case marking. The motivation for the layer I case inflections in Hindi appears to be fully syntactical: the nominative is the default case and the use of the oblique is determined by the postposition that is added to it. In contrast, the postpositional marking in Hindi is not completely syntactical, so there is a difference between the function of layer I and the function of layer II. DO, for instance, is optionally marked with a postposition. When DO is marked, it is not treated in the same way as S, so there is no ergative pattern of case marking. The conditions in which DO is marked are semantic: an animate and/or definite DO is marked with the postposition ko, as in ex. (1); other kinds of DOs are unmarked in the nominative case. (1)
ma˜ı=ne laflrki=ko dekh-a¯ h-ai I=ne girl.f.sg.=ko see-past.m.sg. be-pres.3sg. ‘I have seen the girl.’
ko is also used to mark the IO and the Experiencer; both core arguments are typically high in definiteness and/or animacy. Furthermore, ko can be used as an adjunct postposition of time or location, e.g., ra¯t=ko ‘at night’ and samu¯dr=ko ‘to the sea’. Here, also, the use of ko indicates a certain kind of definiteness, referring to a well-defined place
40
Saartje Verbeke
or time. Whereas the ergative postposition ne in Hindi functions only as the syntactic marker of A (cf., Masica 1991: 367 and ex. (1)), ko in Hindi is very much multifunctional; its occurrence is determined by definiteness and not by an argument-distinguishing purpose. In monotransitive sentences with a perfect finite verb form, A is clearly distinguished with the postposition ne, and the second argument can only be DO, marked or unmarked. Therefore, positing a distinguishing function of ko is superfluous. The characterizing function of ko defines it as a semantic marker of animacy and definiteness. It is much less obligatory as an indicator of a syntactic core argument than the postposition ne, which is only used for A in sentences with a perfect participle. To what degree, then, does ko still function as a case marker? Blake (1994: 1, followed by Siewierska and Bakker 2009: 291) points out that “case marks a relationship of a noun to a verb at the clause level, or of a noun to a preposition, postposition or another noun at the phrase level.” According to this definition, we can assume that ne is the case marker of A, as it always and only occurs when the verb is transitive and perfect. There is, therefore, a clear grammatical relationship between the verb and the postposition as an indicator of A. The multifunctionality of ko, however, poses a greater problem. If the occurrence of ko is in the first place determined by the animacy/definiteness of its head, how does it mark the relationship between a noun and a verb? In Hindi, ko influences the agreement pattern of the verb and is, therefore, related to the verb in the perfect form. Agreement in Hindi in sentences with a perfect verb form is normally ergative, i.e., there is agreement between the verb and the DO. However, the occurrence of ko ‘blocks’ the agreement between the DO and the verb. The sentence in ex. (1) illustrates this: the verb dekh-a¯ is in the default form (m.sg.), i.e., it does not agree with any argument in the sentence. Thus, while ko has a characterizing function and identifies the DO as animate and/or definite, it can also be considered a case marker, as it indicates the relationship between the DO and the verb. This case from Hindi is well known in the literature (cf. Mohanan 1994, Butt 2004, Comrie 1983, amongst others). The postpositions have been defined as case markers in which the distinguishing and characterizing function is alternately emphasized.4 Next, I will take a 4
The ne postposition in Hindi, for instance, has also been identified as a semantic marker (cf., Mohanan 1994: 55 ff.).
Postpositions and agreement in Rajasthani languages
41
look at the marking in another IA language, Marwari, where the situation is slightly different. I will also examine agreement in Marwari, because agreement and marking both function to distinguish and characterize the arguments.
3. Alignment in Marwari Marwari is the standard dialect in the state of Rajasthan. It is spoken throughout this state in western India, but mainly in and around the city of Jodhpur (5,622,600 speakers, according to Ethnologue 2009). Emigrants have brought the language outside Rajasthan to other parts of India and the world. Marwari displays a split in alignment determined by tense/aspect (cf., Section 1). On the one hand, sentences without a perfect finite verb form have accusative case marking and agreement in Marwari, i.e., they treat A and S as different from DO. S and A are in the unmarked or nominative case, and DO takes the postposition nai when it is animate and/or definite. This postposition also marks the indirect object (IO). On the other hand, constructions with a perfect finite verb form in Marwari display an alignment pattern that differs from that of non-perfect constructions. The focus of this investigation will be on constructions with a perfect finite verb form; these constructions are more challenging for alignment typology than the accusative pattern that is found in constructions with a present participle. Next, I will give a general overview of the alignment. First, in sentences with a perfect finite verb form, the case marking of the core arguments is as follows: nouns and pronouns take two forms, a formally unmarked or nominative form and an oblique form (layer I). The oblique form is not used as an independent form. It always takes a postposition in the form of a clitic (cf. Butt and King 2004: 170). Because it marks a core argument, this postposition belongs to layer II. Thus far, marking in Marwari is the same as in Hindi. However, Marwari uses the unmarked nominative case for S and A in both the present and past tenses. In contrast to Hindi, there is no ergative marker. In turn, the marking of the DO is again similar to that in Hindi, although a different postposition is used. There is Differential Object Marking (DOM): the DO is in the oblique case and is
42
Saartje Verbeke
marked with the postposition nai (or colloquial variants like na˜ı, ne or n˜e) if it is animate or definite. When the DO does not show a high level of definiteness or animacy, according to the cline of human > animate > inanimate (Silverstein 1976: 112–171, Aissen 2003: 437), the form of DO is the same as the nominative form used for A and S. Consider examples (2)–(3) below. Ex. (2) is an intransitive sentence with a nominative subject mha˜ı. The subject in ex. (3) is also nominative, and its form is the same as the nominative subject in ex. (2). In ex. (3), s´aran=na˜ fl ı is marked according to the criteria of DOM. (2)
mha˜ı bego hi va¯m=rai=kana˜ fl ı ja¯y pu¯g-yo I.nom. fast emph. he=gen.=close go.cv. arrive-past.m.sg. ‘I got to him in a hurry.’
(3)
mha˜ı s´aran=na˜ fl ı dekh-i I.nom. Sharan.f.sg.=na˜ı see-past.f.sg. ‘I saw Sharan.’
Next, consider the agreement pattern in Marwari. The transitive verb always shows agreement with DO, irrespective of whether the latter takes the accusative postposition or not. In example (2), for instance, DO is marked, but the verb dekh-i is feminine singular and is in agreement with DO. Agreement in Marwari is, therefore, clearly and consistently ergative.5 This is different from Hindi, where agreement is blocked because of the marking of DO with ko, as in ex. (1). In Marwari, therefore, we see two variations from the Hindi pattern that require an explanation: (i) the apparent accusative case marking pattern, in particular the absence of an ergative postposition combined with ergative agreement, which poses a problem for alignment theory, and (ii) the consistent agreement between the verb and the DO, whether it is marked or not. These two related problems will be treated in the next sections; the first problem will be examined in Section 3.1–3.3 and the second will be examined in Section 3.4.
5
Magier (1983: 250) reports that in some dialects of Marwari the participle form agrees with DO while the copula agrees with A. However, this is an exceptional pattern in Marwari that requires an extensive historical explanation, and I do not address these forms in this article.
Postpositions and agreement in Rajasthani languages
43
3.1. Accusative marking and ergative agreement? Theoretical preliminaries In Marwari, agreement and case marking apparently follow a different pattern, i.e., the agreement pattern displays a different kind of alignment than that suggested by the case marking pattern. While the agreement is ergative, the marking found in constructions with a perfect verb form is the same as that in the present tenses, i.e., there is no ergative marker, and there is DOM. This points to an accusative casemarking pattern (cf., Khokhlova 1995). It has been observed that it is very unlikely or even impossible for a language to have this kind of accusative case marking combined with ergative agreement (cf., Comrie 1978: 340, Anderson 1977: 330, Bossong 1984): We might go on to ask whether there are any languages that combine the nominative-accusative morphological system with the ergativeabsolute morphological system, having one for nominal case-marking and the other for verb-agreement. In fact, there are many languages which have an ergative-absolute system for nominal case-marking and a nominative-accusative system for verb-agreement (the inverse is rare or nonexistent). (Comrie 1978: 337–338). Dixon (1994: 94) calls this type of alignment split a ‘bound’ versus ‘free’ split, referring to the split in alignment between the ‘bound’ cross-references (agreement markers) on the verb and the ‘free’ case markings. It occurs, for instance, in the language Enga6, where the case marking is ergative and the verb agreement is accusative (cf., Bossong 1984, Van Valin and LaPolla 1997: 254). Agreement is bound to prefer an accusative pattern because agreement markers are supposedly derived from bound pronouns. It is generally accepted that pronominal marking is most often accusative (cf., Dixon 1994: 94). While the accusative agreement pattern appears to be primarily historically motivated because cross-references originate in bound pronouns, in Marwari, the agreement is definitely ergative and in the next section, I argue that this ergative agreement pattern is not
6
Enga is a language spoken in Papua New Guinea.
44
Saartje Verbeke
strictly historically motivated. Dixon’s view on agreement is related to the opinion that case marking of A, S and O is considered in the first place as having an argument distinguishing function (cf., also Comrie 1989: 126): by using case marking, A and O are clearly distinguished in the mind of the speaker/listener. But in Marwari, it is the ergative agreement pattern, and not the postpositional marking, that distinguishes the arguments in the perfect tenses in Marwari. The marking, therefore, primarily has a semantic function that is related to the lack of an ergative marking equivalent to the ergative postposition in Hindi. In the next section, I elaborate this view on marking in Marwari. 3.2. ‘Bound’ versus ‘free’ split in Marwari: determining the marking of the subject 3.2.1. Pronouns Pronouns in Marwari exist in many forms and phonological variants due to differences in pronunciation and writing and to the influence of local variants and neighboring languages (cf., Table 1). However, the distinction between an unmarked nominative case and an oblique case with postposition also holds for the pronouns. Table 1: pronouns in Marwari nom. 1 sg. me, m˜e, m(h)ai, m(h)a˜ı, ma¯i hum, mhu, mhu, ˜ mhu¯ 2 ta˜ı tu¯m fl 3 (u)vo, o
obl. nom. me, mai pl. mhe mha¯ ta˜ı the, tame tha¯, tama¯ unfl (u)vai
obl. mhe the, tame va¯m fl
For the first person singular pronoun alone, we find me (m˜e), m(h)ai (mha˜ı), hum, mhu (mhu), ˜ etc. in the nominative case. For the second person singular there is tu¯m fl and ta˜ı (cf., Gusain 2004: 23–24, Drocco 2008: 98). The forms with the phoneme -u- come from the former nominatives of Old Rajasthani, whereas the forms with -e- or -ai- were formerly used only as obliques. In Old Rajasthani, the oblique was not only used as the case before a postposition, it was also
Postpositions and agreement in Rajasthani languages
45
the case of A, which is in fact the ergative case (cf., Khokhlova 1992). Thus, in this stage of the language, there was nominal ergativity. In contemporary Marwari, these forms of the pronouns occur not only as obliques with a postposition, but also as unmarked pronouns without a postposition in both the S and A position. The situation of the third person pronouns is different and includes a more distinctive difference between the nominative and the oblique. Consider the following examples (4), (5), (6) and (7). They illustrate that today, the hum, mhu (mhu) ˜ forms and the me (m˜e), mhai (mha˜ı) forms are both used for A and S, despite their different historical backgrounds. Although ex. (4) shows an intransitive construction that always takes a nominative S, the first person form mha˜ı is used. In ex. (5), the same pronoun (the variant ma¯i) is used as A of the verb ‘to say’. Ex. (6) has the pronoun mhu¯m fl as A. In the second part of the example, the second person pronoun tu¯m fl is A, while historically this was a nominative form restricted to intransitive sentences. Ex. (7) combines both forms mhu¯m fl and ma˜ı in exactly the same construction, thus proving that there is no apparent difference in function between the two forms of the pronoun in contemporary Marwari. (4)
mha˜ı bada fl ¯ ba¯bu¯=ri tebal=rai=sa¯ma˜ı au¯bho I.nom. big boss=gen. table=gen.=in front standing h-o be-impf.m.sg. ‘I was standing in front of the table of the big boss.’
(5)
la¯rla¯ dina¯m fl sagafll-a¯ patamg fl uda fl ¯ vai h-a¯, passed days all-nom.m.pl. kite.f.sg. fly.pres. be-impf.m.pl. ma¯i mamma=na˜ı ka-yo I.nom. ma=na˜ı say-past.m.sg. ‘In the past days, everyone was flying a kite – I said to mamma.’
(6)
mhu¯m fl I.nom.
pappu¯=na˜ı pu¯ch-yo tu¯m fl P.=na˜ı ask-past.m.sg. you.nom.
ma¯’flt sahib=na˜ı kyu¯m fl koni bula¯-yo? master sir=na˜ı why not invite-past.m.sg. ‘I asked Pappu – Why did you not invite master?’
46 (7)
Saartje Verbeke
mhu¯m fl mammi=na˜ı ka-yo, pã mammi I.nom. ma=na˜ı tell-past.m.sg. but ma.nom. isai ka¯m=m˜e phãsar-i h-i ke mammi=na˜ı such work=in sunk-f.sg. be-impf.f.sg. that ma=na˜ı ya¯d memory.m.sg.
koni rai-yo, ma˜ı pa¯pa¯=na˜ı not stay-past.m.sg. I.nom. pa=na˜ı
ka-yo pã pa¯pa¯ bhi me=ri ba¯t tell-past.m.sg. but pa.nom. also I.obl.=gen. word.f.sg. koni sun-i fl not hear- past.f.sg. ‘I told mamma, but mamma was sunk in so much work that mamma did not remember, I told papa – but papa also did not listen to my words.’ The oblique form of the third person singular unfl rarely occurs as A or S. However, if it is used, there is free variation between the use of the oblique and the nominative pronoun vo for A (cf., Magier 1983: 244). According to Khokhlova (1992), the strict opposition between the ergative and the nominative is only maintained in the m.pl. paradigm. The form va¯m fl is the oblique form of the third person plural pronoun, nominative vai, in combination with a postpostion. In some constructions, however, va¯m fl is also used for A, as in ex. (8) and (9). Furthermore, in accordance with Khokhlova’s research, I have found no examples where va¯m fl occurs as S. (8)
va¯m fl pu¯ch-yo ka¯m=m˜e koi dikkat o he.obl.?pl. ask-past.m.sg. work=in some problem then nim fl a¯-i? not come-past.f.sg. ‘He asked – Hasn’t any problem come up in the job?’
(9)
va¯m fl sava¯l gafllat bata¯-yo he.obl.?pl. question.nom.m.sg. wrong speak-past.m.sg. ‘He asked the question wrongly.’
The pronoun va¯m fl is nevertheless not the only pronominal form used for A. The following ex. (10) is widely attested; the original nominative form vai is used, although it is in the function of A.
Postpositions and agreement in Rajasthani languages
(10)
47
vai bada fl ¯ het=su¯˜ mha¯=ri a¯di fl he.nom.pl. great care=with I.obl.=gen. problem.f.sg siraka¯-i solve-past.f.sg. ‘He solved my problems with great care.’
If case is defined as a system of formal and functional oppositions (cf. Dik 1978, Masica 1991), va¯m fl is not the structurally ergative case of the third person pronoun, given that the variant vai is used for A as well. The use of vai shows that there is no systematic difference between the marking of the transitive and intransitive subject in Marwari, just as the other pronouns show. It is therefore maintained that the concept of an ergative case cannot be applied to contemporary Marwari. Historically, however, there was an ergative pronominal case; residues of this case can be found in the third person plural paradigm only, where the use of the oblique case of the pronoun is restricted to A and not S. An explanation for this fact may be found in the closeness of the third person pronoun and the nominal declension. The third person pronouns can also be used as demonstrative adjectives, and as such they display more similarities to nouns than to the first and second person pronouns. It is known from the literature that nouns, and by extension demonstratives, tend to be more marked than pronouns if they function as subject/agent (cf. Dixon, 1994: 86). Therefore, in Marwari, although both the historically ergative and nominative form of the first and second personal pronoun are still used, it is only in the third person plural that a functional difference is still made; in that instance, the oblique form is restricted to A. 3.2.2. Nouns Nouns generally possess a nominative and an oblique case form, but often there is no overt difference between the two cases. The oblique case is only used and recognized when it is followed by postpositions. However, in some paradigms we can again notice residual forms of an older ergative case, especially for the masculine plural paradigm; not coincidentally, these forms resemble the paradigm of the third person pronoun. It is only in the masculine paradigm ending in a vowel that the different cases take distinct inflections (cf., Table 2).
48
Saartje Verbeke
Table 2: nouns ending in a vowel m.sg. f.sg. m.pl. f.pl. Nom. -o/-au -i - a¯ -i Obl. -e/-ai -i - a¯m fl -i Nouns ending in a consonant remain unchanged in the nominative plural and oblique singular and take -a¯m fl for the oblique plural. The nominative is the case used for all subjects. Consider example (11); bha¯yala¯, a noun ending in a vowel, is in the nominative plural case. (11)
sagafll-a¯ bha¯yal-a¯ pu¯ch-yo a¯j all-nom.m.pl. friend-nom.m.pl. ask-past.m.sg. today mha¯=nai sagafll-a¯m=nai fl kiya¯m fl ya¯d kari-ya¯? I.obl.=nai all-obl.=nai why memory.m.pl. do-past.m.pl. ‘All the friends asked – Why do you remember us all today?’
As in the third person plural pronominal paradigm, the paradigm of masculine plural nouns is most susceptible to traces of the older division between a nominative and an oblique without postposition (= ergative). According to Khokhlova (1992: 87), “in m.sg. paradigm of nominal stems in -o/-au, there exist free variation of Nominative and Ergative markers for A while S is generally marked by Nominative”. She gives the following examples (12) and (13), in which the stem of the noun, however, does not end in -o/-au, but in a consonant. (Khokhlova 1992: 87–88, own glossing and modified transcription, SV): (12)
anek vidva¯n … likha-’r nahi bhej+sak-ya¯ many scholar.pl. write-cv not send+can-past.m.pl. ‘Many scholars could not send (their) articles.’ (Chandrasimh 1966:2) < after they had written [them], many scholars could not send [them].
(13)
bidva¯n-a¯m fl thor-au ma¯rg dikha¯-yau scholar-pl. some-m.sg. way.m.sg. show-past.m.sg. ‘The scholars showed some way.’ (Bhandari 1966:20)
For Khokhlova, the difference between the form vidva¯n as S and bidva¯na¯m fl as A is illustrative of the discrepancy between the nominative and ergative cases. However, in the feminine paradigm, nouns
Postpositions and agreement in Rajasthani languages
49
do not end in -i; the ending -a¯m fl is often used for S, as ex. (14) shows, while the expected form would be the invariant form a¯mkhya fl ¯. The ending in -a¯m fl should, therefore, be considered a mere variant of -a¯ for the nominative feminine plural, and it is not confined to marking A. In ex. (14), the plural a¯mkhya fl fl ‘eyes’, ending in -a¯m, fl is used ¯m as S. (14)
abkai va¯m=ri fl a¯mkhya fl fl mha¯=ra¯=pai ¯m now he.obl=gen. eye-nom.f.pl. I.obl=gen.=on jamyod-i fl h-i fixed-f.pl. be-impf.f.pl. ‘This time his eyes were fixed on me.’
In conclusion, the case marking of pronouns and nouns functioning as subject/agent is not ergative in Marwari because there is no systematic differentiation between the case marking of A and S. In my findings, there is only one restriction on this: the masculine plural third person pronoun va¯m fl and some forms of masculine plural nouns are rarely, if ever, used for S. The ergative case has disappeared everywhere, except in these instances. In the next section, the marking of the object is investigated. 3.3. The marking of the object Two semantic criteria determine the object marking in Marwari: definiteness and animacy (cf., Masica 1991: 365). The marking is differential, as in Hindi. Kittilä (2006) suggests a typology of object-, animacy- or role-based strategies for object marking. Object marking includes both the marking of the DO and the marking of the IO. In Kittilä’s typology, the marking of the DO in Marwari is animacybased, although animacy is expanded to definiteness in most IA languages. The DO is obligatorily nai-marked when it is animate and definite, e.g., the proper name S´aranfl in ex. (15). (15)
mha˜ı s´aran=na˜ fl ı dekh-i I.nom. Sharan.f.sg.=na˜ı. see-past.f.sg. ‘I saw Sharan.’
When the DO is inanimate but definite, it is optionally nai-marked. In ex. (16), the DO is construed with na˜ı, and the postposition excludes any possibility of interpreting this DO as indefinite.
50 (16)
Saartje Verbeke
mha˜ı kha¯mosi=ri simv=na˜ fl ı pa¯r kar+g-yo I.nom. silence=gen. limit.m.sg.=na˜ı. pass do+go-past.m.sg. h-o be-impf.m.sg. ‘I passed the limit of silence.’
The DO is never nai-marked when it is (i) inanimate and indefinite or (ii) animate and indefinite. On the other hand, the IO in Marwari is marked with nai on a more regular basis; this is to be expected because the IO is typically high in animacy and definiteness (cf., ex. (17)). (17)
ba¯bu¯ mha=na˜ı baiflth ja¯vnai=ro fl isa¯ro boss.nom. I.obl.= na˜ı sit.cv. go.inf.=gen. sign.nom.m.sg. kar-yo make-past.m.sg. ‘The boss made me a sign to sit down.’
The fairly consistent marking of the IO may indicate that nai is structurally assigned to this argument. However, languages with DOM based on animacy are open to what has come to be called EDOM, Extended Differential Object Marking (cf., Kittilä 2006: 14). In EDOM, the use of the marker of the DO has been “extended” to mark the function of the IO as well, using the same criterion of animacy. In this view, the extension of the DO marker to the IO does not imply that the use of the marker is being determined by something other than animacy. 3.4. Agreement in Marwari. Discussion Additional evidence of the semantic nature of nai is found in the pattern of agreement in Marwari. This pattern is different from Hindi, where the agreement is blocked because of the marking of DO (cf., Section 2). In Marwari, the agreement is not blocked (cf., ex. (15)). Thus, the occurrence of nai as a marker of a noun has no influence on the agreement of the verb; it does not function as an indicator of a relationship between the noun and the verb, like ko does in Hindi. In fact, nai only indicates the definiteness or animacy of the DO. It is not a case marker, but a semantic marker. Mistry (1998) proposes a similar but slightly different approach in his study of verb agreement and object marking in Gujarati. Gujarati shares a common ancestry with Ra-
Postpositions and agreement in Rajasthani languages
51
jasthani through the ancestral “Old Western Rajasthani” (cf., Tessitori 1914–1916), and Marwari displays minor differences from the agreement pattern in Gujarati. Mistry observes that the verb in Gujarati agrees with the marked DO; this is similar to Marwari and different from Hindi. In addition, Mistry remarks that this agreement pattern forms a problem for any theoretical approach which suggests that verb agreement is restricted to the nominative (unmarked) case. For Hindi, this approach is valuable, as there is only agreement with the unmarked case. However, Gujarati (and Marwari) are “marked” (in the sense of exceptional) languages in this respect because the verb agrees with the marked DO (cf. Mistry 1998: 429, citing Falk 1991: 227). To solve this contradistinction, Mistry argues that the postpositional marker of the DO ne is not a case marker, but a specificity marker that functions on semantic grounds. It is thus not able to restrict or block agreement (Mistry 1998: 433–437). Thus far, Mistry’s approach is essentially the same as the one supported here. However, he notices that some Gujarati verbs take objects obligatorily marked with ne (NE in his article), and they agree by default (i.e., they do not agree with the object). Examples of these verbs include mafll- ‘to meet’, vadhfl ‘to rebuke’, karadfl ‘to bite’, coflt- ‘to cling’, bheflt- ‘to embrace’, cum- ‘to kiss’, nam- ‘to bow’, paranfl ‘to marry’ and adakfl ‘to touch’ (Mistry 1998: 429). Mistry argues that in constructions with these particular verbs, NE functions as a case marker, i.e., an object marker; NE is homophonous with the earlier-mentioned ne postposition that serves as a specificity marker and does not restrict agreement. A number of objections can be raised against Mistry’s analysis. First, the verbs in question do not take a DO as Mistry assumes (and as their translations do in English). On the contrary, they are intransitive verbs that take a beneficient/recipient. 7 The obligatoriness of NE indicates this, as IOs and Experiencers (i.e. non-DOs) are always marked in Marwari and Gujarati (cf. Section 3.3). Moreover, the use of cognate verbs in Hindi points in the same direction. For instance, the Hindi verb milna¯ ‘to meet’ is intransitive, and in expressions such as ex. (18), a¯dmi takes the postposition ko; obviously, the verb does not agree with the marked object, but with the subject. 7
A more literal but grammatically incorrect English translation of the verbs would be ‘to join with’, ‘to rebuke to’, ‘to bite in’, ‘to cling to’, ‘to embrace with’, ‘to kiss with’, ‘to bow to’, ‘to marry with’ and ‘to touch to’, respectively.
52 (18)
Saartje Verbeke
vah a¯dmi=ko kal mil-a¯ he man=ko yesterday meet-past.m.sg. ‘The man met him yesterday.’ lit: ‘He joined with the man yesterday.’
The proposed homophony between the case marker NE and the specificity marker ne is in fact reducible to another instance of the multifunctionality of the object marker in the IA languages: it marks the DO under certain semantic conditions, the IO according to EDOM, and also Experiencers. The verb in Gujarati and Marwari always agrees with its DO, whether it is marked or not. However, agreement with an IO or an Experiencer, also marked with the same postposition, is out of the question.8 Another argument for the semantic nature of the postposition nai is the fact that agreement in Marwari consistently marks the DO. It is therefore the verb agreement that fulfills the argument-distinguishing and relational function that is attributed to ko in Hindi. Although ko is also a postposition that in essence indicates definiteness/animacy, it may have grammaticalized to a case marker under the influence of the strictly syntactic use of the ne postposition as a marker for A. In contrast, in Marwari, A does not have a postposition, which makes the notion of a case marking postposition in Marwari rather unsuitable. Change in another direction can be found in Kashmiri, an IA language without postpositions. In Kashmiri, DOM is completely abandoned in the past tenses. The agreement and marking are consistently ergative. Kashmiri shows mixed ergativity as it is traditionally defined: S and O are treated in exactly the same way. In the following ex. (19) from Kashmiri, the animate and definite DO ‘I’ is in the nominative case, and the verb agrees with it.
8
Experiencer verbs in Hindi and Marwari generally agree with their grammatical subject and not with the Experiencer, which is marked, respectively, with ko or nai. Agreement between the Experiencer verb and the Experiencer is possible outside the IA languages, as shown by Bickel (2003) in a study of the TibetoBurman language Bilhari. The IA languages included in Bickel’s study, Maithili and Nepali, do not display Experiencer agreement and follow the general IA pattern.
Postpositions and agreement in Rajasthani languages
(19)
53
tami on-u-s bi she.erg bring.past.-m.-1sg. I.nom. ‘She brought me’
In Kashmiri, there are no postpositions. The accusative case inflection, which is used in the present tenses to mark the DO, does not appear in the past tenses. Again, this seems to indicate that in Marwari, the postposition nai is not strictly a case marker. 3.5. Intermediate conclusion The preceding data were presented to illustrate and solve the two problems outlined in Section 3.1.: (i) the exceptional combination of ergative agreement and accusative case marking and (ii) the agreement between the verb and the marked DO. It has been shown that these issues are related to each other and are caused by the different functions of argument marking and agreement. From the above data, it is clear that the ‘bound’ versus ‘free’ split in Marwari is not caused by the pronominal nature of argument marking, as Dixon argues (cf., Section 3.1.). Rather, it is the different motivation of postpositional marking and agreement that lies at the basis of the split. Agreement in Marwari distinguishes the arguments, as the verb always agrees with the DO, whether it is marked or unmarked. This distinction happens so thoroughly that the postpositional marking is free to mark the arguments for semantic reasons. Therefore, an ergative marking of A is not present in Marwari, and the differential marking of DO is a semantic marker rather than a case marker. In conclusion, there is no split between ergative agreement and accusative marking in Marwari. It is perhaps more appropriate to speak of verbal ergativity combined with differential postpositional marking. In Harauti, this ‘split’ between marking and agreement is even more extreme, as illustrated in the next section.
54
Saartje Verbeke
4. Features of alignment in Harauti Harauti is a Rajasthani language that is related to Marwari but spoken on a slightly smaller scale in and around the twin cities KotaBundi (4,730,000 speakers according to the 2001 census). In Harauti, as in Marwari, sentences with a perfect finite verb form display a different form of alignment than present tense sentences. In tenses without a perfect verb form, the agreement is accusative. The case marking is as follows: A and S are in the nominative, and the markedness of the DO is again determined by animacy and definiteness. The DO postposition is in the same form as in Marwari, i.e., nai, often nasalized to nai. In the perfect form, Harauti has consistent ergative agreement, just as in Marwari. This shared agreement pattern points to the possibility that the marking in Harauti is similarly purely semantic and that the postposition na˜ı does not function as a structural case marker. There is an additional reason the postpositional marking does not qualify as a case marker in Harauti, though. In Marwari, the use of na˜ı is limited to DO and IO, but in Harauti, na˜ı is the case marker of DO and A. This does not lead to ambiguity, because there is a rule specifying that the postposition occurs only once in the sentence, preferably marking A (Allen 1960: 10). Whenever A is marked, DO remains in the nominative case. Moreover, DO receives no marking at all when it is indefinite or inanimate. IO, which takes the same postposition as DO in Marwari and Hindi, is marked in Harauti by other postpositions, depending on the verb. However, na˜ı does appear in roles that are associated with the postposition of IO in the other languages, i.e., as the marker of the Experiencer. 4.1. na˜ı marking A In most sentences, A is constructed with na˜ı with the finite verb in the perfect verb form. Consider the following examples (20) and (21), in which the same verb pa¯rfl - is used. Ex. (20) is a transitive sentence in the perfect tense, and the subject is constructed with na˜ı. Ex. (21) is very similar to (20), but this time the verb is in the imperfect tense with a subject in the third person. In ex. (21), the subject does not take the postposition na˜ı, and the verb agrees with the subject va¯; the verb does not agree with DO as in ex. (20).
Postpositions and agreement in Rajasthani languages
(20)
55
mha˜ı=na˜ı ha¯mk-a fl ¯ pa¯rfl -ya¯ ar pa¯chai I.obl.=na˜ı scream-nom.m.pl. scream-past.m.pl. and after bhi bha¯g-i also run-past.f.sg.
panfl but
mofltar bus.nom.f.sg.
nhã not
dhab-i fl stop-past.f.sg. ‘I screamed and I ran after it, but the bus did not stop.’ (21)
va¯ ha¯mk-a fl ¯ pa¯rfl -ti she.nom.f.sg. scream-nom.m.sg. scream-pres.f.sg. r-i ar mofltar=kai=pa¯chai bha¯g-ti prog-past.f.sg. and bus=gen.= after run-pres.f.sg. r-i prog-past.f.sg. ‘She was screaming and running after the bus.’
In ex. (22), A is used with the postposition na˜ı. Both elements of the present perfect construction, i.e., the past participle dekhi and the auxiliary chi, agree with the f.sg. DO jhalak, which is referred to by the relative pronoun jyo. (22)
dokari=ki fl a¯mkhya fl fl e ya¯cana¯=ki ¯ m=m˜ old woman=gen. eye.f.pl.=in begging=gen. asi jhalak ch-i jyo mhã=na˜ı this glimpse.f.sg. be-impf.f.sg. which I.obl =na˜ı im=su fl ¯m fl phali nhã dekh-i ch-i she.obl.f.sg.=from first not see.-past.f.sg. be-impf.f.sg. ‘In the eyes of the old woman there was a glimpse of begging which I had not seen from her before.’ 4.2. na˜ı marking Experiencers
Unlike the ne postposition in Hindi, the use of na˜ı is not limited to the marking of A. The following are examples in which the postposition na˜ı marks the Experiencer rather than A. Experiencer verbs taking a non-nominative subject are common in the IA languages (cf., Verma and Mohanan 1990). In Hindi and Marwari, these Experiencers are marked with the same postposition used for DO and IO. In Harauti, they are marked with the same postposition as
56
Saartje Verbeke
A. The identical marking of the Experiencer and A cannot possibly conflict or lead to ambiguity in any of the examples because the Experiencer and A never occur together. Ex. (23) is a sentence constructed with the verb lag ‘to attach’ and an Experiencer with a na˜ı postposition. (23)
dokari fl mhã=na˜ı asi la¯g-i old woman.nom.f.sg. I.obl.=na˜ı such attach-past.f.sg. jasya¯m fl sa¯mcya¯ni fl im=kai fl ta¯mi fl khaja¯no such just she.obl=gen. that treasure.m.sg. mala+g-yo ch-o im=kai fl find+go-past.m.sg. be-impf.m.sg. she.obl=gen. ta¯mi fl khaja¯no mala+g-yo ch-o that treasure.m.sg. find+go-past.m.sg. be-impf.m.sg. ‘The woman seemed to me as if that treasure of her had just been found.’
The typical function of IO with verbs like ‘to bring to’ or ‘to say to,’ however, is expressed with other postpositions. For instance, in ex. (24), su¯m fl ‘with’ is used, and in ex. (25), kai ‘for’ is used. (24)
im=su fl ¯m fl ja¯da¯ va¯ ka¯mi fl bhi nha he.obl= su¯˜ more she.nom.sg. something also not bol+sak-i say+can-past.f.sg. ‘She could not say more to him.’
(25)
pota¯-potya¯m=kai fl bei a¯mba¯ la¯-i grandchild.obl.m.pl.= kai those mangoes bring-past.f.sg. ch-i be-impf.f.sg. ‘I’d brought those mangoes for my grandchildren.’
Another example of a ditransitive construction is with the prototypical verb ‘to give’. Ex. (26) uses the expression ‘to give an order’, and na˜ı is used as the marker of the transitive subject. The IO takes a different morpheme, pai, a locative postposition with the meaning ‘on’.
Postpositions and agreement in Rajasthani languages
(26)
57
ek marad=na˜ı dalevar=pai fl hukam one man=na˜ı driver=pai order-nom.m.sg. cala¯+d-yo, dalevar fl sa¯b! go+give-past.m.sg. driver sir ‘A man ordered the driver: driver sab!’
In summary, na˜ı is used as a postposition of an Experiencer verb. With typical ditransitive verbs such as ‘to bring’, ‘to say’ or ‘to give’, the IO receives a different postposition. 4.3. na˜ı marking DO DO is the third possible argument for which the multifunctional postposition is used. Harauti, just like Marwari, has DOM, which likewise depends on the different criteria of animacy and definiteness. When the DO is human, it almost always takes na˜ı. When it is inanimate, it is only marked when there is some kind of definiteness associated with it (cf., Masica 1991: 365). In Harauti, another restriction applies to the optional marking of the DO: nai is only used once in a clause, so the DO is not marked with the postposition when A or an Experiencer is present in the sentence. In ex. (27), there is no overt subject; the main verb pu¯chajyo is an imperative. The inanimate but definite DO mofltar of ruk- ‘to stop’ takes na˜ı. Thus, na˜ı is used to indicate an argument in the sentence, but not to indicate the subject. Ex. (28) is a similar example with an animate DO. (27)
a¯va-ti mofltar=na˜ı ruk-a¯-’r pu¯cha+j-yo ke driving-f.sg. bus=na˜ı stop-cs-cv. ask+go-imp.2sg. that ek thelo to nhã mal-yo one bag then not be found-past.m.pl. ‘After stopping the driving bus, ask once if there hasn’t been found a bag.’
(28)
dekh-o im fl dokari=na˜ fl ı! see-imp.2sg. this old woman=na˜ı ‘Look at this old woman!’
Ex. (29) is in the present tense. A is the interrogatory pronoun ‘who’. DO is inanimate but definite and determined by a genitive phrase.
58 (29)
Saartje Verbeke
kha¯ba¯=ki cij=na˜ı kunfl choflr-ai eat=gen. thing=na˜ı who.nom.m.sg. loose-pres. ch-ai? be-pres.3sg. ‘Who loses a thing to eat?’ 4.4. Salience
Clearly, na˜ı is a multifunctional postposition. In Hindi, as in Harauti, the related form ne is used for A, while in Marwari na˜ı is used for DO and IO. In Harauti, however, na˜ı also occurs as a marker of DO and the Experiencer; it therefore appears that a high level of animacy and definiteness is associated with na˜ı and its phonological variants in every language in which they occur. However, the multifunctionality of the postposition is raised to extreme levels in Harauti. Harauti exemplifies a system in which only one form of postposition is used to mark the main arguments. With regard to the distinguishing function of case markers, as in Marwari, the consistent ergative agreement in Harauti fulfills the function of distinguishing the arguments, in particular the DO. The marking does not necessarily have a distinguishing function. Furthermore, the distribution of na˜ı in Harauti seems to indicate that it is not a case marker. It does not stand in functional opposition to other case markers like ko and ne do in Hindi. It does not block agreement as ko does in Hindi. It is a semantic marker as far as it indicates definiteness and animacy and in particular marks Experiencers, A’s and animate or definite DOs. However, it only appears once in a clause and can therefore often leave an animate DO unmarked. Furthermore, it can mark inanimate A’s as well. Animate and inanimate arguments are equally marked with na˜ı, as illustrated in the following example (30), where an inanimate subject is marked in the same way as all other transitive subjects: (30)
mofltar=nai pa¯m-pa fl ¯ m=ki fl a¯va¯j kar-i bus=na˜ı paw-paw=gen. sound.nom.f.sg. make-past.f.sg. ar ek jhafltako de’r flthaharag-i and one shake.m.sg. give.cv. stop-past.f.sg. ‘The bus made the sound of paw-paw and after giving a shake it stopped.’
Postpositions and agreement in Rajasthani languages
59
Animacy and definiteness are thus not enough to account for the distribution of na˜ı in Harauti. A possible criterion that unites the three uses of na˜ı, and which puts the na˜ı-marked arguments in opposition to the other unmarked arguments, may be found in discourse importance, i.e., the high level of salience the na˜ı-marked arguments possess in conversation. Salience may be defined as the level of “communicative importance” of the arguments in the sentence and the influence they exercise on the effect accomplished by the sentence (cf., Malchukov 2008: 216). In this view, the main semantic purpose of the na˜ı-marking is to characterize the argument as the main discursive argument of the sentence, and, as such, to highlight it vis-à-vis the other arguments. Definiteness and animacy are two criteria contributing to a high level of salience (cf., Croft 1988: 165). For example, an animate and definite argument tends to be more salient in discourse than inanimate and indefinite arguments (cf., Hopper and Thompson 1980). Animacy and definiteness are, however, not the only factors determining marking (cf. ex. (30), where an inanimate agent is marked). It is true that the DO, for instance, is only marked when it is definite and animate; however, it is primarily marked when it is the foregrounded argument, i.e., when it is the most important argument in the sentence. If the DO is only animate or definite but not foregrounded, the marking with na˜ı is not certain. This also explains why the IO of typical ditransitive verbs such as ‘to give’ or ‘to tell’ do not generally take na˜ı. The IO in Harauti is not the most salient argument in the sentence; for this reason, the IO is not marked by the postposition for discourse salience, but by another “adjunctive” kind of postposition. Still, in the rather exceptional situations where the IO has a great discursive importance, it may be marked with na˜ı (see, for example, ex. (31)). The animate IO para¯r ‘another’ takes na˜ı. Although this is unusual, it is an emphasized argument of the sentence: of all people, the buffalos are sold to an utter stranger. The salience of the IO is moreover stressed by its position at the beginning of the sentence: (31)
para¯r=na˜ı sya¯na fl ¯ =m˜e other=na˜ı cold season=in
mhu¯m fl bha˜ısya¯m fl bec’r I.nom. buffalo.nom.m.pl. sell.cv.
pa¯cho ja¯ r-yo ch-o first go prog- past.m.sg. be-impf.m.pl. ‘It was to a stranger that I first sold my buffalos in the cold season, and then I was leaving.’
60
Saartje Verbeke
One could be tempted to suggest that the marking of A with na˜ı is syntactically obligatory, since it seems as if all As of a verb in a perfect tense are marked. That this is not the case, however, is illustrated in ex. (32), where mha˜ı, while being A, does not take the na˜ı postposition: (32)
adha-khuli pha¯fltak=ko kudo ˜ da fl ¯ va¯ hath=su¯m fl half-open door=gen. part.nom.m.sg. left hand=with pakaflr’ r mha˜ı jivana fl ¯ hath= su¯m fl take.cv. I.nom. right hand=with
pofltali bhitar luggage.nom.f.sg. inside
pha¯mk fl d-i throw give-past.f.sg. ‘While taking a part of the half open door with my left hand, I threw my luggage inside with my right hand.’ The absence of na˜ı-marking indicates that in this particular sentence the emphasis is on the action, which happens in a hurry, and not on the subject, which is more or less obvious and already known from the previous context in the story. This is the opposite of the situation in ex. (32). In the present tenses, of course, A is never marked, as the alignment is accusative. The discrepancy between present and perfect constructions is a consequence of historical change that led to the split alignment situation in most IA languages. Agreement in the present tenses is always with the subject. While the marking of A differs for present and perfect constructions, the distribution of na˜ı as a marker of the Experiencer and the DO in the present tenses is the same for the sentences with a perfect verb form. This supports the assumption that the use of na˜ı as a marker of A in Harauti is a later development that compensates for the absence of agreement between the perfect verb form and A. Diachronic research needs to be undertaken to confirm this hypothesis. The case marking pattern in Harauti illustrates the variation in the functions of formal marking, while the function of agreement is to distinguish the argument most closely linked with the verb, in this instance, the DO. Again, as in Marwari, the marking of the arguments has no influence on agreement.
Postpositions and agreement in Rajasthani languages
61
5. Conclusion This paper has dealt with the alignment pattern of two closely-related languages of the Rajasthani branch of Indo-Aryan languages, Marwari and Harauti. Both languages have the same agreement pattern; there is verbal ergativity in every construction with a perfect finite verb form. However, the case marking patterns of the two languages differ considerably. Both languages make use of the same form of postposition; however, this postposition is used in radically different ways. In Marwari, nai is the marker of DO and IO. Its use is determined by the level of definiteness/animacy of these arguments. The postposition has a characterizing function and the distinguishing function of traditional case marking is absent. This is confirmed by the fact that nai as a marker of DO does not block agreement between the verb and DO, as happens in Hindi. The use of na˜ı in Harauti further illustrates that postpositions can function as something other than inflectional case markers. In Harauti, na˜ı is an argument marker; what argument it marks is determined by the salience of that argument. Salience is defined as the communicative importance of the argument in discourse, a criterion related to the level of definiteness and animacy of the argument. Marking on the arguments is thus not purely distinguishing or identifying in these languages; it adds meaning and is thus semantic. Agreement, on the other hand, is a far more consistent syntactic system for distinguishing the arguments linked with the verb.
References Aissen, Judith, 2003, Differential object marking. Iconicity versus economy. Natural language and linguistic theory 21: 435–483. Allen, William, 1960, Notes on the Rajasthani verb. Indian Linguistics 21: 1–13. Allen, William, 1964, Transitivity and possession. Language 40(3): 337–343. Anderson, S. R., 1977, On mechanisms by which languages become ergative. Mechanisms of syntactic change, Charles N. Li (ed.), 317–363. Austin & London: University of Texas Press. Bahl, K. C., 1989, Adhunik rajasthani ka samracanatmak vyakarana. Jodhpur: Rajasthani Sahitya Samsthan. Beames, J., 1872–1879, A comparative grammar of the Modern Aryan languages of India. London: Trubner & Co. Bhati, N. and S. Shekhavat, 1972, Rajasthani vyakarana: ek adhyayana. Jodhpur: Rajasthani Shodha Samsthana.
62
Saartje Verbeke
Bickel, Balthasar and Johanna Nichols, 2009, Case marking and alignment. The Oxford handbook of case, Andrej Malchukov & Andrew Spencer (eds.), 304–321. Oxford: Oxford University Press. Bickel, Balthasar, 2003, Referential density in discourse and syntactic typology. Language 79(4): 708–736. Blake, Barry J., 1994, Case. Cambridge: Cambridge University Press. Bossong, Georg, 1983–1984, Animacy and markedness in universal grammar. Glossologia 2–3: 7–20. Bossong, Georg, 1984, Ergativity in Basque. Linguistics 22: 341–392. Butt, Miriam, 2001, A reexamination of the accusative to ergative shift in IndoAryan. Time over matter. Diachronic perspectives on morphosyntax, Miriam Butt and Tracy H. King (eds.), 105–142. Stanford: CSLI Publications. Butt, Miriam and Tracy H. King, 2004, The status of case. Clause structure in South Asian languages, Veneeta Dayal and Anoop Mahajan (eds.), 153–198. Dordrecht: Kluwer Academic Publishers. Butt, Miriam, 2006, Theories of case. Cambridge: Cambridge University Press. Chatterji, Sunit K., 1949, Rajasthani bhasa. Udaipur: Vishva Vidyapita. Comrie, Bernard, 1973, The ergative, variations on a theme. Lingua 32: 239–253. Comrie, Bernard, 1978, Ergativity. Syntactic typology. Studies in the phenomenology of language, W. P. Lehmann (ed.), 329–393. Sussex: The harvester press. Comrie, Bernard, 1984, Reflections on verb agreement in Hindi and related languages. Linguistics 22: 857–864. Comrie, Bernard, 1989, Language universals and linguistic typology. Oxford: Basil Blackwell Publishing. Croft, William, 1988, Agreement vs. case marking in direct objects. Agreement in natural language: approaches, theories, descriptions, Michael Barlow and Charles A. Ferguson (eds.), 159–80. Stanford: Center for the study of language and information. Davison, Alice, 2002, Agreement features and projections of TENSE and ASPECT. The yearbook of South Asian languages and linguistics. Rajendra Singh (ed.). 11–27. New Delhi: Sage. Dik, Simon C., 1978, Functional grammar. Amsterdam: North-Holland Publishing Company. Dixon, Robert M. W., 1979, Ergativity. Language 55(1): 59–138. Dixon, Robert M. W., 1994, Ergativity. Cambridge: Cambridge University Press. Drocco, Andrea, 2008, L’ergatività in hindi. Studio diachronico del processo di diffusione della posposizione ne. Alessandria: Edizione dell’Orso. Falk, Yehuda, 1991, Case: Abstract and morphological. Linguistics 29: 197–230. Filiminova, Elena, 2005, The noun phrase hierarchy and relational marking: Problems and counterevidence. Linguistic typology 9: 77–113. Garrett, Andrew, 1990, The origin of NP split ergativity. Language 66: 261–296. Givón, T., 1979, On understanding grammar. Academic Press: New York. Grierson, George A. G., 1906–1928, Linguistic Survey of India. Delhi: Motilal Banarsidass. Gusain, Lakhan, 2001, Shekhawati. München: Lincom Europa. Gusain, Lakhan, 2003, Mewati. München: Lincom Europa.
Postpositions and agreement in Rajasthani languages
63
Gusain, Lakhan, 2004, Marwari. München: Lincom Europa. Hopper, Paul J. & Sandra A. Thompson, 1980, Transitivity in grammar and discourse. Language 56, 251–299. Kachru, Yamuna and R. Pandharipande, 1979, On ergativity in selected South Asian languages. South Asian languages analysis 1: 193–209. Kellogg, Samuel H., 1938, A grammar of the Hindi language. London: Kegan Paul, Trench, Trubner and Co. Khokhlova, Liudmilla V., 1992, Trends in the development of ergativity in New Indo-Aryan. Osmania Papers in Linguistics 18: 71–98. Khokhlova, Liudmilla V., 1995, The development of patient-oriented constructions in Late Western NIA languages. Osmania Papers in Linguistics 21: 15–54. Khokhlova, Liudmilla V., 1999, Stative participles in Western NIA languages. Vidyopaasanaa: Studies in honour of Harivallabh C. Bhayani, B. Modi (ed.), 91–112. Mumbai/Ahmedabad: Image Publications. Khokhlova, Liudmilla V., 2000, Typological evolution of Western NIA languages. Berliner Indologische Studien 13/14: 117–142. Khokhlova, Liudmilla V., 2001, Ergativity attrition in the history of Western IndoAryan languages. The yearbook of South Asian languages and linguistics, R. Singh (ed.), 159–184. New Delhi: Sage Publications. Kittilä, Seppo, 2006, Object-, animacy- and role-based strategies. A typology of object marking. Studies in language 30 (1): 1–32. Klaiman, Miriam H., 1987, Mechanisms of ergativity in South Asia. Lingua 71: 61–102. Klimov, G.A., 1973, Ocˇerk obsˇcˇej teorij ergativnosti. Moskva: izdat. Nauka. Kothari, D., 1991, Rajasthani bhasa aur uski boliyam. Udaipur: Sahitya Samsthan. Lazard, Gilbert, 2001, Le marquage différentiel de l’objet. Language typology and language universals. An international handbook, M. Haspelmath et al. (eds.), Vol.2, 873–885. Berlin/New York: de Gruyter. Lidz, Jeffrey, 2006, The grammar of accusative case in Kannada. Language 82 (1): 10–32. Magier, David, 1983, Components of ergativity in Marwari. Papers from the annual regional meeting of the Chicago Linguistic Society 19(1): 244–255. Magier, David, 1985, Case and transitivity in Marwari. Conference on participant roles: South Asia and adjacent areas. Chicago: Indiana University Linguistics Club. Malchukov, Andrej L., 2008, Animacy and asymmetries in differential case marking. Lingua 118: 203–221. Mistry, P.J., 1998, Objecthood and specificity in Gujarati. The life of language. Papers in linguistics in honor of William Bright. Hill, Jane H., Mistry P.J. Lyle Campbell (eds.), 425–442. Berlin: Mouton de Gruyter. Mohanan, Tara, 1994, Argument structure in Hindi. Stanford: CSLI Publications. Masica, Colin, 1991, The Indo-Aryan languages. Cambridge: Cambridge University Press. Rajpurohit, N., Dhoor mein pagliya, New Delhi: Sahitya Akademi. Sakariya, A. B. and B. Sakariya, 1984, Rajasthani-Hindi Shabd Kosh. Jaipur: Pamcashil Prakashan.
64
Saartje Verbeke
Saraswat, R. and Premji, 1984, Aaj ri Rajasthani kahaniyan. New Delhi: Sahitya Akademi. Sharma, R., 1992, Rajasthani Lokagathavan. New Delhi: Sahitya Akademi. Siewierska, Anna and Dik Bakker, 2009, Case and alternative strategies: word order and agreement marking. The Oxford handbook of case, Andrej. L. Malchukov and Andrew Spencer (eds.), 290–304. Oxford: Oxford University Press. Silverstein, Michael, 1976, Hierarchy of features and ergativity. Grammatical Categories in Australian Languages, Robert M. W. Dixon (ed.), 112–171. Canberra: Australian Institute of Aboriginal Studies. Spencer, Andrew, 2005, Case in Hindi. Proceedings of the LFG05 Conference. Miriam Butt and Tracy Holloway King (eds.). CSLI Publications (http://cslipublications.stanford.edu/) Stump, G. T., 1983, On the elimination of ergative patterns of case-marking and verbal agreement in modern Indic languages. Ohio State University Working Papers in Linguistics 27: 140–164. Tessitori, L.P., 1914–1916, Notes on the grammar of Old Western Rajasthani. Indian Antiquary 43–45. Tournadre, Nicholas, 1996, L’ergativité en Tibétain. Approche morphosyntaxique de la langue parlée. Louvain: Editions Peeters. Tsunoda, T., 1981, Split case-marking patterns in verb types and TAM. Linguistics 19: 389–438. Van Valin, Robert D. and LaPolla Randy J., 1997, Syntax. Structure, meaning and function. Cambridge: Cambridge University Press. Verma, Manindra and K.P. Mohanan, 1990, Experiencer subjects in South Asian languages. Chicago: The University of Chicago Press. Zakharyin, Boris, 1979, On the formation of ergativity in Indo-Aryan and Dardic. Osmania Papers in Linguistics 5: 50–71.
Research Foundation – Flanders (FWO)/Ghent University Blandijnberg 2 B-9000 G e n t Belgium
[email protected]
S a a r t j e Ve r b e k e
Qu’est-ce qu’un pronom en anatolien?
65
Qu’est-ce qu’un pronom en anatolien? Abstract In Ancient Anatolian languages, pronouns exhibit in their morphology some specific properties connected with number and gender reference. Number can account for the formal differenciation of personal pronouns and proforms while headnoun’s gender accounts for the use by dependant proforms of a specialized set of pronominal endings. As a consequence, demonstrative pronouns (or ‘3rd person pronouns’), when used in core grammatical cases, index a larger amount of information than the corefered inanimate noun.
1. Introduction En dépit d’un matériel morphologique abondant et de procédés diversifiés, les langues indo-européennes anciennes utilisent peu la morphologie pour différencier les classes de mots entre elles. Si les noms et verbes sont, le plus souvent, formellement reconnaissables comme tels indépendamment du contexte, les adjectifs, numéraux, adverbes, adpositions, connecteurs, conjonctions, etc., ne présentent pas de propriétés morphologiques génériquement attachables à une classe donnée. La position des pronoms est, de ce point de vue, souvent ambivalente: les pronoms personnels présentent généralement une morphologie particulière alors que les autres pronoms oscillent entre des organisations idiosyncrasiques (souvent distinctes de celles des pronoms personnels) et une morphologie analogue à celles des noms. Cette dernière situation mise à part, dans les langues indo-européennes anciennes, la flexion des pronoms tend, de façon générale, à se différencier de celle des autres mots déclinables par plusieurs propriétés: (i) (ii) (iii)
la déclinaison comporte un nombre généralement réduit de cas par rapport à celle des noms et adjectifs; les désinences peuvent être différentes, au moins partiellement, de celles qu’utilisent les noms et adjectifs fléchis aux mêmes cas; la flexion tend à être supplétive, avec un thème de nominatif (éventuellement, d’accusatif) distinct du thème utilisé aux autres cas.
66
Sylvain Patri
Selon les langues et, à l’intérieur de chaque langue, selon la nature des pronoms, ces propriétés sont plus ou moins saillantes1. La présente étude se donne pour objet de préciser les aspects qu’elles revêtent dans les langues anatoliennes en s’interrogeant sur le rôle qui leur est dévolu dans la syntaxe. La question à laquelle on tentera de répondre, dans un premier temps (§§ 2–7), sera donc celle des critères en fonction desquels on peut distinguer les mots dont la catégorisation est indépendante du contexte de ceux qui ne révèlent la classe à laquelle ils appartiennent que dans un contexte syntaxique donné. Ensuite, dans un second temps (§§ 7–11), on s’efforcera d’établir dans quelle mesure un rôle spécifique peut être reconnu aux critères en question dans l’organisation de la référence comme dans la syntaxe. L’étude porte sur les langues anatoliennes, mais on doit préciser d’emblée que les données de la comparaison au sein de l’anatolien sont décevantes: le hittite mis à part, peu de paradigmes complets sont parvenus jusqu’à nous, de sorte qu’il est difficile de discerner si les données éventuellement divergentes entre elles du lycien, du lydien, du palaïte et dans une moindre mesure du louvite reflètent des données de l’anatolien commun ou si elles ne font que prolonger autrement la situation reflétée en hittite. On adoptera ici cette dernière attitude par défaut non parce qu’elle est démontrablement vraie, mais parce que c’est celle qui demande le moins de conjectures invérifiables.
2. Pronoms interlocutifs 2.1. La flexion2 Suivant une tendance largement répandue dans les langues du monde, le nominatif des pronoms interlocutifs est, en vieux hittite, morphologiquement inanalysable par rapport au reste du paradigme3. 1 2
3
. Voir Brugmann 1904, Schmidt 1978, Beekes 1988, Bader 1990, Katz 1998. Pour la description des flexions pronominales hittites, on renvoie une fois pour toute aux dictionnaires (HW2; Puhvel, HED) et à la récente grammaire de Hoffner & Melchert 2008. Les études de Benveniste 1953 et de Laroche 1960, 174–185, restent importantes, mais ne sont plus à jour des données. J’emploie le terme de supplétisme pour désigner toute alternance entre des morphèmes de même signifiant dont les signifiés présentent des différences formelles irréductibles selon les règles de la morphonologie ou de l’allomorphie.
Qu’est-ce qu’un pronom en anatolien?
67
Les autres formes ont des bases radicales invariantes à tous les cas obliques: 1 sg. amm-, pl. anz-; 2 sg. tu-, pl. summ-. Le jeu des désinence est uniforme, à l’exception du génitif, où l’ancien cas en -enzan, quand il est attesté, est remplacé par -el dans les textes plus récents4. Le seul changement d’organisation observable au cours de l’évolution est qu’en conséquence de la tendance conduisant à la confusion du nominatif et de l’accusatif, les formes de nominatif adoptent, au pluriel, des radicaux indifférenciés par rapport au reste du paradigme5. (1)
pronoms de première personne en singulier (“je”) nominatif u¯k ) amm-uk acc., dat.-lock. amm-uk ) u¯k génitif amm-el ablatif-instr. amm-edaz
(2)
pronoms de deuxième personne en hittite singulier (“tu”) pluriel (“vous”) nominatif zik, zik summ-es ) summ-a¯s acc., dat.-loc tu-k summ-a¯s ) summ-es génitif tu-el summ-enzan f summ-el ablatif-instr. tu-edaz summ-edaz
hittite pluriel (“nous”) wes f anz-a¯s anz-a¯s anz-el anz-edaz
En louvite, où le cas génitif a été, de façon générale, pratiquement éliminé, ce qu’on peut discerner de la flexion pronominale (Payne 2004: 24 sq., Melchert 2003: 189 sq.), indique une complète homonymie des deux formes attestées (par hasard?) aux 1ère personne singulier et pluriel, ce qui, en l’état actuel de la documentation, conduit a considérer que les pronoms sg. hiér. (a)mu /amu/ et pl. cun. anza¯s, hiér. á-zu-za /anzunz/ seraient complètement sortis de la flexion. En lycien et en lydien, une forme unique est pareillement utilisée dans 4
5
Dans KBo 3.27: 15, passage mutilé de l’Édit de Hattusili Ier sur copie récente, une restauration proposée par Forrer 1926, 17, et admise par Kammenhuber 1962, 377, postule l’existence d’un forme de génitif 1 sg. *am]m-enzan qui n’est pas plausible, -enzan étant une désinence de pluriel (§ 2.3 et 5.2); voir, dans le même sens, avec d’autres arguments, CHD L-N 145a, P 93a, Hoffner & Melchert 2008, 135 n. 8. Dans les tableaux ci-dessous, ‘A f B’ signifie que A est éliminé par B, tandis que ‘A f B’ indique que B entre en concurrence avec A.
68
Sylvain Patri
des rôlés sémantiques du nominatif et du datif, lyc. amu, emu, e˜ mu, lyd. amu. La seule restitution sûre à laquelle conduit la comparaison anatolienne est donc la base oblique 1 sg. *amu-. A la deuxième personne, la flexion différencie trois cas avec un nominatif supplétif au singulier: (3)
pronoms de deuxième personne en louvite singulier (“tu”) pluriel “(vous”) nominatif cun. ti, hiér. ti /ti/ hiér. u-zu-za /unz-unz/ acc., dat.-loc tu-Ø cun. u¯nz-as, hiér. /unz-unz/ ablatif-instr. hiér. /tuw-ari/ hiér. u-za+ra/i /unz-ari/
La désinence abl. -ari reflète le “rhotacisme” caractéristique du louvite hiéroglyphique en face de la forme -ati bien attestée dans la flexion nominale (Oshiro 2000). Les données du lycien, du lydien et du palaïte font reconnaître certaines correspondances (Melchert 1983), mais trop fragmentées pour permettre la restitution d’un paradigme. 2.2. Désinences nominales et pronominales La seule désinence des flexions interlocutives qui peut être en toute sûreté rapportée à l’état anatolien commun comme représentative d’un cas donné est celle de l’accusatif-datif-locatif -a¯s (hitt. anz-a¯s, summ-a¯s, louv. cun. u¯nz-as), laquelle est identique à la désinence -a¯s du datif-locatif pluriel des noms. Au même cas, en louvite, l’apparition d’un -Ø n’est qu’une conséquence de l’élimination des finales répondant à hitt. -(u)k. Les désinences nominales, hitt. -es (hitt. summ-es) ou louv. -unz, ne se rencontrent que là où le supplétisme a disparu, ce qui suggère des emprunts analogiques à la flexion des noms. Seules les désinences de génitif sont spécifiquement pronominales. (4)
désinences des pronoms interlocutifs en anatolien hittite louvite nominatif inex. ou pl. -es inex. ou pl. -unz acc., dat.-loc sg. -(u)k : pl. -a¯s sg. -Ø: pl. -a¯s, -unz génitif -el, pl. -enzan – ablatif(-instr.) -edaz -ati ~ -ari
Les pronoms personnels forts de l’anatolien sont donc des formes utilisant des thèmes amalgamant la personne et le nombre combinés
Qu’est-ce qu’un pronom en anatolien?
69
avec des affixes flexionnels tantôt nominaux (-es, etc.), tantôt pronominaux (-el, -enzan)6. 2.3. Cas particulier du pronom de 2ème personne du pluriel La morphologie du pronom 2 pl. est originale à divers titres: (i)
(ii)
(iii)
il atteste, au génitif, une désinence -enzan (supplantée par -el dans les textes récents) qui, là où elle est attestée, amalgame une information relative au cas et au nombre (§ 5.3), singularité indiquant que “vous” – à la différence de “nous” – est marqué comme pluriel; sa flexion repose sur un radical invariant summ- (on attendrait un nominatif reposant sur quelque chose comme *yús), ce qui l’assimile formellement au régime général des noms et adjectifs; le radical summ- n’a pas d’équivalent dans les autres langues indo-européennes (essais d’étymologie: Melchert 1984: 27, Katz 1998: 136sq.), ce qui indique son caractère récent dans l’évolution.
Le pronom 2 pl. peut donc être considéré, du point de vue morphologique, comme le moins personnel des pronoms (ou le plus pronominal des indices de personne), singularité que confirme le fait que, parmi les déterminants possessifs, l’indice de co-variation avec le possesseur #s(a)m- est le seul à neutraliser la deuxième personne avec la troisième. Il semble raisonnable d’estimer que cette situation résulte d’un traitement en relation avec la moindre aptitude de “vous” à l’associativité par rapport à “nous” (détails dans Danièl’ 2005a: 120–123). 2.4. Dérivation adjectivale En hittite récent, le thème d’ablatif-instrumental des interlocutifs ammed-, tued-, anzed-, summed- sert de base à la création d’un paradigme nouveau fléchi selon le modèle des noms ou des délocutifs forts (§ 3.3). Le nouveau pronom qui en résulte semble limité à un emploi 6
La grille typologique de Danièl’ 2005b, est contestable en ce qu’elle postule que les désinences prises par les thèmes pronominaux sont soit nominales, soit pronominales, ce qui ne coïncide ni avec la situation de l’anatolien, ni avec celle de la plupart des autres langues indo-européennes anciennes.
70
Sylvain Patri
de possessif référé à une tête à laquelle est assigné (par hasard?) un cas oblique, locatif ou ablatif7: (5) a. KUB 36.90 Ro 21–22 (CTH 386, rédac. récente, Haas, 1970, 178) mDudhaliyan lukkattas#kan UD.KAM-ti au-matin##adv. jour-dat./loc. T.-acc. URUHakmis tued-as a¯ssiyant-as pªed-as 2poss-loc.pl. préféré-loc.pl. lieu-loc.pl. H. ˇ SˇUM URUNerik AS LúSANGA-UTTIM iskanzi N. pour objet prêtrise oindre-3pl.
“au matin, ils consacreront Tudhaliya à la prêtrise en tes lieux favoris, Hakmis (et) Nerik” b. KUB 8.48 ii 11–13, dupl. KUB 17.3 i 5–6 (CTH 341, Laroche 1969, 131) tued-aza memiyan-az UL#war#an#kan nég.#quot.#3sg.acc.#adv. 2poss.-abl. parole-abl. kuenne[r] tuer-3pl.prét. “ne l’ont-ils pas tué [le taureau] selon ton commandement?” (6) a. KUB 36.35 i 2–4 (CTH 342, restauré d’après i 10–15, Otten 1953, 126) [(nu#wa#ta#kkan ammed-az)] memiyan-az conn.#quot.#2sg.acc.#adv. 1poss.-abl. parole-abl. GAM-ta tamas[(mi sous écraser-1sg. “je vais t’écraser de mon autorité” b. KBo 3.4 iv 45–46 (CTH 61.I, Götze 1933, 126) nu#za ke KUR.KUR LÚ.KUR INA hostile cp. conn.#intens. dém.-nom./acc.pl. pays ˇ MU.10.KAM ammed-az SU-az tar(a)hhun 10 ans 1poss.-abl. main-abl. soumettre-1sg.prét. “en dix ans, j’ai conquis ces pays hostiles de ma main” 7
Haas, 1970: 181, Francia 1996, 212–213, Luraghi 1997, 13–14, 24, Hoffner & Melchert 2008, 137, 278 (§ 18.6).
Qu’est-ce qu’un pronom en anatolien?
71
L’émergence de ce nouveau pronom est contemporaine à celle du possessif délocutif apellaz (ablatif) fondé, pour sa part, sur le génitif fléchi apel (infra, § 3.3). Le fait que ces formes de possessif sont en relation avec l’ablatif-instrumental, tantôt comme base dérivationnelle, tantôt comme forme préférentiellement fléchie, suggère que leur processus repose sur une opération faisant de la provenance “de N” une cause d’appartenance.
3. Pronoms délocutifs 3.1. La référence délocutive La notion de “pronom de troisième personne” s’applique au fait que des référents autres que “je” et de “tu” peuvent être désignés par des pronoms. En hittite, les seuls pronoms exclusivement co-référés à un antécédent lexical sont les formes clitiques (§ 6). Toutefois, comme dans beaucoup de langues, les aptitudes référentielles des pronoms clitiques peuvent aussi être assumées par des pronoms forts, en l’occurrence par les deux démonstratifs ka¯- et apa¯-. Ces pronoms cumulent trois emplois8: () En tant que déictiques, ka¯- et apa¯- peuvent tenir le même rôle syntaxique que n’importe quel constituant nominal pour dénoter une ostension concomitante à l’acte d’énonciation: (5) a. KUB 26.12 ii 5–6 (CTH 255, rédaction récente, von Schuler 1957, 24) nu kıª memai { …} conn. dém.-nom.-acc.sg. dire-3sg. apa¯t memai { …} apa¯s#ma dém.-nom.sg #conj. dém.-nom.-acc.sg. dire-3sg. “et qu’il dit ceci (ka¯-) { …}, mais, qu’en fait, il dit cela (apa¯-) { …}” 8
Le mot ani- semble, en hittite, se comporter comme un démonstratif, mais il n’est attesté que comme forme figée en marge de l’usage (de son côté, annisemble être un emprunt livresque à l’accadien annisˇ ; cf. Kammenhuber, HW2 I: 81, Kloekhorst 2008, 174).
72
Sylvain Patri
(2) En tant que modifieurs d’un constituant nominal, ka¯- et apa¯peuvent prendre comme tête des noms dont le référent est a priori dépourvu de localisation spatio-temporelle: (6) a. KUB 14.1 i 13–15 (CTH 147, rédaction et tablette moyenne, Götze 1928, 4–5) linkiya [ka]ttan nu#t[ta] conn.#3sg.dat. serment-dat./loc. sous kªe udda¯r dais dém.-nom.-acc.pl. parole-nom.acc.pl. placer-3sg.prét. “il a placé ces paroles (ka¯-) envers toi sous serment” b. KUB 13.9 iii 19–20 (CTH 258, rédac. moyenne, tabl. récente, Westbrook & Woodard 1990, 644) nu apa¯t uttar { …} parkuwanzi conn. dém.-nom.-acc.sg. parole-nom.acc.sg. clarifier-3pl. (si quelqu’un remet en cause un jugement) “on met au clair cette parole (apa¯-)” Les deux pronoms peuvent aussi être co-référés à des noms désignant des entités situées par rapport au locuteur dans un rapport de distance a priori similaire comme, par exemple, “dieu”: (7) a. KBo 16.24 i 46–49 (CTH 251, rédac. moyenne, tabl. récente, Rizzi Mellini 1979, 522–523) [nu] apu¯n kªe NISˇ DINGIR.MESˇ conn. dém.-acc.sg. dém.-nom.pl. dieux. appandu saisir-3pl.opt. “que ces dieux (ka¯-) du serment se saisissent de celui-ci (apu¯n)” b. KUB 36.75 ii 10–11, dupl. KUB 30.11 rev. 3–4 (CTH 374, rédac. ancienne, tabl. moyenne, Lebrun 1980, 123, 129) nu#ssan DINGIR-LUM apa¯s dém.-nom.sg. conn.#adv. dieu takni ma¯n ne[(pisi)] ma¯n#as conj. ciel-loc. conj#3sg.nom. terre-loc. “que ce dieu (apa¯-) soit au ciel ou sur terre”
Qu’est-ce qu’un pronom en anatolien?
73
Le mécanisme en fonction duquel opère la sélection de ka¯- envers apa¯- est fondé sur l’identifiabilité pragmatique du référent: plus le référent du nom tend vers un emploi auxquels le contexte discursif attribue un caractère spécifique, déjà connu, immédiatement ou facilement accessible, c’est-à-dire “proche” dans une représentation mentale des rapports spatio-temporel (Lakoff 1974), plus ce nom sélectionnera ka¯-; à l’opposé, plus ce référent présente un caractère générique, peu ou pas connu, hors de portée, et donc “distant”, plus il sélectionnera apa¯-. La notion de “distance” en fonction de laquelle on rend compte, traditionnellement, des emplois respectifs de ka¯- et de apa¯- doit être entendue dans une acception purement métaphorique. Les sphères d’emplois respectives de apa¯- et de ka¯- sont déterminées par la perception que le locuteur se fait ou qu’il entend donner de l’identifiabilité d’une entité au moment où il désigne celle-ci9. () Enfin, en tant que substituts de constituant nominal, les deux pronoms peuvent réaliser l’anaphore; dans les exemples vieux-hittites suivants (8), ka¯- et apa¯- sont intégrés dans des séquences anaphoriques dont le premier chaînon est un pronom clitique: (8) a. KBo 3.60 ii 14–16 (CTH 17/1, rédac. ancienne, tabl. récente, Güterbock 1938, 106, CHD S 143b) mKaniu ¥ ze¯andan da¯s ¯s UZU SˇAH s#an mDUMU-dEN!.LIL-as pera¯n da¯is si#quot.#adv. ta#kku#wa#ssan kıª hazzizi ta#wa DINGIR-[LUM] dém.acc.sg. reconnaître-3sg. si#quot.#adv. “Kanius prit de la chair de porc cuite et la (#an) plaça devant mDUMU-dEN!.LIL: s’il la (ka ¯ -) reconnaît [comme étant du porc], alors il (est) un dieu!”
9
Goedegebuure 2003a, 340, exprime une conclusion proche en des termes différents: “the demonstratives could also be used metaphorically to indicate that something belonged to the cognitive domain of one of the speech participants”. Le recours à la notion de “domaine cognitif” semble toutefois discutable dans le sens où celui-ci préjuge de connaissances (ou d’événements) extralinguistiques construits dans l’esprit des locuteurs, alors que des exemples comme (6)-(7) montrent que la discrimination de ka¯- par rapport à apa¯- ne fait pas nécessairement appel à un univers de représentation partagé, encore moins à une opération mémorielle.
74
Sylvain Patri
b. KUB 43.26 i 4–6, dupl. KBo 21.25 i 46–48 + KUB 34.123 i 16–18 (CTH 631, rédac. ancienne, tabl. ancienne et moyennes, Archi, 1978, 24) [o o #m]a 2 GUD.MAH ¥ GE6-TIM karpanzi {…} [(t#us anda Éh)]alentiu pedanzi [h(ukanzi)] nu apu¯ss#a conn. dém.acc.pl.#coord. sacrifier-3pl. On choisit deux taureaux noirs {…} on les (#us) conduit au palais et on les (apa¯-) sacrifie” Le pronom apa¯-, typiquement co-référé à des entités “distantes”, est beaucoup plus utilisé dans l’anaphore que ka¯-, mais (8a) et d’autres témoignages réfutent la conception enseignée par Hoffner & Melchert (2008: § 5.6) selon laquelle la résomption anaphorique serait du ressort de apa¯- seulement10. Traditionnellement, les grammaires du hittite décrivent apa¯comme “pronom personnel” de troisième personne et ka¯- comme “pronom démonstratif”, mais sans exposer clairement le ou les critères censés motiver une telle différence de statut. Or du moment où les aptitudes référentielles et syntaxiques de ka¯- et de apa¯- sont, comme on vient de le voir, identiques, on ne voit pas ce qui justifierait de les opposer, l’un comme forme “personnelle” (et “démonstrative”?), l’autre comme forme “démonstrative” seulement11. Une telle catégorisation est non seulement arbitraire, elle conduit à des incohérences descriptives puisque apa¯-, censé être “personnel”, peut être démonstratif et résompteur anaphorique, tandis qu’un “démonstratif” comme ka¯- peut tenir le même rôle référentiel qu’un pronom clitique. Il paraît plus approprié de reconnaître que les démonstratifs apa¯- et ka¯- admettent l’un comme l’autre des emplois adnominaux comme 10
11
Melchert 2009, fait de cette conception un paramètre classificatoire dans une étude sur la déixis pronominale en partie fondée sur les interprétations contestables de Goedegebuure 2003 (infra, § 8.2). Cette partition paraît avoir été, sinon affirmée, du moins suggérée, par Sturtevant 1933, 201, au motif que apa¯- est “the commonest pronoun refering to an antecedent” (ce qui devient dans Sturtevant 1951, 110 “the more emphatic pronoun of the third person”). Quoi qu’il en soit, la distinction est fixée dans les descriptions grammaticales de Friedrich 1940 § 99 = 1960: §§ 98, 111, Held, Schmalstieg & Gertz 1988, 32, Luraghi 1997 § 2.3.2, Hoffner & Melchert 2008, 143sq.
Qu’est-ce qu’un pronom en anatolien?
75
pronominaux et que, dans cette dernière situation, l’un et l’autre peuvent assumer des emplois référentiels similaires à ceux des pronoms clitiques délocutifs12. En toute rigueur, les seuls pronoms “de troisième personne” du hittite sont les pronoms clitiques. La situation ne semble pas différente en louvite. 3.2. Restriction casuelle sur la référence pronominale Une des différences entre pronoms forts et pronoms clitiques est que les premiers peuvent être fléchis au génitif, mais pas les seconds. Or, les aptitudes référentielles des pronoms, qu’ils soient interlocutifs ou délocutifs, sont beaucoup plus étroitement spécialisées au génitif qu’à n’importe quel autre cas. Les délocutifs forts au génitifs n’ont qu’un seul emploi: l’expression d’un rapport possessif (9): (9) a. KBo 7.28: 17 (CTH 371, Friedrich 1957, 219) ki#ma tuel estu ceci#conj. 3sg.gén. être-3sg.opt. “que ceci soit tien” (litt. ‘de toi’) b. KBo 7.28: 24–25 (CTH 371, Friedrich 1957, 219) sumenzan es[tu … ki#ma ceci #conj. 2pl.gén. être-3sg.opt. “que ceci soit vôtre” (litt. ‘de vous’) c. KUB 8.81 + KBo 19.39 ii 8 (CTH 41, del Monte 1981, 216) apa¯s#wa ammel cela#quot. 1sg.gén. “cela est mien” (litt. ‘de moi’) La même caractérisation vaut lorsque le pronom est modifieur d’un nom:
12
Selon l’enquête de Bhat 2005, portant sur 184 langues, 54 % des langues utilisent des pronoms délocutifs distincts des pronoms démonstratifs; 28 % utilisent tous les démonstratifs comme équivalent de délocutifs; 10 % utilisent comme équivalent de délocutifs les démonstratifs spécialisés dans la déixis distale; 8 % utilisent comme équivalent de délocutifs les démonstratifs spécialisés dans la déixis proximale.
76 (10)
Sylvain Patri
KBo 3.1 i 18–19 (CTH 19, réd. ancienne, tabl. récente, Hoffmann 1984, 16) apªell#a SˇU-i URU.DIDLI.HI.A ¥ 3sg.gén.#conj. main-dat.-loc. cité GAL.GAL-TIM tittiyantes eser grandes établir-ptcp.nom.pl.. être-3pl.prét. “les grandes villes furent attribuées à sa main (litt. ‘à la main de lui’)” [= placées sous sa souveraineté]
En hittite moyen et récent, l’élimination des déterminants possessifs conduit, dans les constructions essives, à généraliser l’emploi des pronoms clitiques au datif comme expression du possesseur, comme dans ces variantes textuelles du texte ancien de la Prière au soleil où (11a) reflète l’état ancien et (11b) l’état nouveau: (11) a. KUB 31.127 i 24, tablette récente (CTH 372, Güterbock 1980, 46) nu hannesnas pedi dariyashas#tis conn. jugement-gén. endroit-loc. fatigue#poss2sg. NU.GÁL ne-pas-être b. KUB 24.3 i 48, tablette récente nu#tta hannesnas pedi tarriyashas NU.GÁL conn.#3sg.dat. “à l’endroit du jugement, tu n’as pas de fatigue” (“‘fatiguetienne’ f ‘fatigue à toi’ n’est pas”) La spécialisation des pronoms forts au génitif dans la possession apparaît définitivement cristallisée dans le fait que ceux-ci ne sont jamais utilisés pour exprimer la possession au datif. Les pronoms clitiques, n’ayant pas de génitif, sont toujours au datif(-accusatif), mais les pronoms forts, qui, eux, peuvent prendre les deux cas, sont toujours au génitif quand ils expriment un rapport de possession. Même le pronom interrogatif(-relatif) au génitif reflète cette spécialisation: (12)
KUB 6.45 iii 48, dupl. KUB 6.46 iv 17 (CTH 381, rédaction récente, Singer 1996, 22, 41, 66–67) kue¯l walliyatar nu#za conn.#intens. interr./rel.-gén. louange-nom.-acc. “de qui (suis-je) la louange?” [suite: ‘ne suis-je pas la louange du dieu de l’orage pihassassi-, mon seigneur?’]
Qu’est-ce qu’un pronom en anatolien?
77
De même qu’à la forme clitique #mu “à moi”, on ne peut pas opposer un génitif “de moi”, il n’y a pas de concurrence possible entre le génitif apa¯s ammel “ceci (est) de moi f ceci est mien” et le datif *apa¯s ammuk / uga “ceci (est) à moi”13. En hittite, la concurrence entre le génitif et le datif dans le marquage du possesseur est limitée aux noms et ne concerne pas leurs subsituts14. Il s’ensuit que les génitifs des pronoms forts ne sont pas des formes casuelles au même titre que les autres; par leur comportement, les génitifs se rapprochent de pronoms possessifs plus que des pronoms personnels15. L’analyse même des génitifs comme formes paradigmatiques plutôt que dérivationnelles pourrait facilement être mise en cause, au moins à partir du moyen hittite où les désinences pronominales de génitif ne varient plus selon le nombre (caractéristique, il est vrai, reproduite dans la flexion nominale où, à la même période, la désinence -as, anciennement de singulier, se généralise au détriment de l’ancien pluriel -an). L’émergence tardive d’un ablatif apellaz directement formé sur le génitif fléchi va pareillement dans le sens d’une extraction de apel de la flexion: (13)
KUB 14.4 iv 23 (CTH 70, rédac. récente, Hoffner 2006, 194) nu#kan DAM-YA apellaz BA.ÚSˇ conn.#adv. épouse-mienne poss.-abl. mourir.3sg. “mon épouse mourut à cause d’elle” (suite aux malédictions lançées par la reine)
Sur le processus de formation d’adjectifs possessifs en hittite tardif, voir § 2.4.
13
14
15
Je ne considère pas comme possessive, mais comme emphatiquement attributive, la fameuse phrase de la veuve du pharaon à Suppiluliuma nu#war#as ammuk LÚMUDI-YA INA KUR URUMizrima#war#as LUGAL-us “pour moi, il sera mon époux; pour l’Égypte, il sera roi” (KBo 5.6 iv 10–12, Güterbock 1956, 97). Cette contrainte n’est pas prise en compte dans les études classiques de Benveniste 1949, 1960; Watkins 1967, dont les conclusions, sont, pour cette raison et d’autres, discutables. Ce point a été justement discerné, quoique non démontré, par Pedersen 1938, 55.
78
Sylvain Patri
3.3. Les flexions Les flexions de ka¯- et de apa¯- sont identiques sauf au nominatif-accusatif singulier des inanimés, où apa¯-t montre une désinence {-d} (les noms et adjectifs ont -Ø ou -n), là où que ki-Ø présente une désinence zéro. En louvite, le nominatif inanimé de apa¯- n’est pas attesté, mais za¯-Ø (= hitt. ka¯-) présente pareillement un zéro16: (14)
paradigme du démonstratif apa¯- en hittite
nom. acc. génitif dat.-loc. instr. abl.
singulier pluriel animés inanimés animés inanimés apa¯s ape ) apu¯s apa¯t ape apu¯n f apa¯n apu¯s ) ape apel apenzan f apedas apedani apedas apet, apedanda apez (f apellaz)
(15)
paradigme du démonstratif apa¯- en louvite singulier pluriel animés inanimés animés inanimés nom. apa¯s ? ? ? acc. apa¯n apinza dat.-loc. apati apa¯sanza abl.-inst. apin
(16)
paradigme du démonstratif ka¯- en hittite singulier pluriel animés inanimés animés inanimés nom. ka¯s ke ) ku¯s ki ) ke ke ) ki acc. ku¯n f ka¯n ku¯s ) ke génitif kel kenzan f kedas dat.-loc. kedi f kedani kedas instr. ket, kidanda abl. kez
16
Comparer les nominatifs masc. véd. sá ~ sás, gr. ² ~ Ρ«; Wackernagel & Debrunner 1930, 541.
Qu’est-ce qu’un pronom en anatolien?
(15)
79
paradigme du démonstratif za- en louvite singulier pluriel animés inanimés animés inanimés nom. za¯s zinzi za¯ za¯ acc. zan zinza dat.-loc. zati za¯san[za abl.-istr. zin
Sur les formes hiéroglyphiques zin, apin, voir la récente analyse de Goedegebuure (2007). Le lycien semble avoir éliminé anat. *ka ´ ¯ - pour ne conserver que le démonstratif eb-: (18)
le démonstratif lycien singulier pluriel animés inanimés animés inanimés nominatif ebe eb˜ei eb˜e e/abeija accusatif eb˜e(nn˜e/i) eb˜eis, ebeijes génitif ? eb˜e/eh˜e datif-locatif ? ebette 3.4. Changements et alternances
Au cours de l’histoire du hittite, la flexion des délocutifs subit divers changements tant locaux qu’architecturaux. Certaines variations sont justifiables au titre de tendances généralisées en hittite (confusion du nominatif et de l’accusatif en hittite récent; remplacement de gén. -enzan par -el), d’autres sont propres aux démonstratifs: à l’accusatif singulier, l’apparition récente de apa¯n et de ka¯n semble être fondée sur une réinterprétation analogique du nominatif ap-a¯s (k-a¯s) f apa¯-s (ka¯-s) d’après les désinences nominales plutôt que sur un simple remplacement de -u¯n par -a¯n nominal car le nominatif pluriel keus, forme rare pour laquelle il ne semble pas exister de répondant *apeus, reproduit un mécanisme similaire (ke + -us). L’ablatif apellaz, également récent, est, de son côté, formé sur le génitif ap-el f apel-. La chronologie des autres cas de concurrence est moins nette: au datif-locatif, les variantes keti et kedani sont à peu de choses près également attestées en vieux hittite, mais, dès le moyen hittite, kedani supplante clairement keti:
80 (19)
Sylvain Patri
occurences du datif-locatif (textes authentiques)17 vieux hittite moyen récent ke-e-da-ni x 3 (KUB 33.59 iii 7–9) x 10 x 51 ke-e-ti x 2 (KUB 43.23 Ro 4, 17) 0 0
Le témoignage du louvite zati confirme que keti est la forme la plus ancienne. A l’instrumental, les variantes apet: ket, apedanda: kidanda sont trop mal attestées dans les textes authentiques pour se prêter à des dénombrements significatifs.
4. Pronom relatif-interrogatif A l’instar des délocutifs, le pronom relatif-interrogatif est soumis à la variation en genre: (20)
le pronom relatif-interrogatif kui- en hittite singulier pluriel animés inanimés animés inanimés nominatif kuis kues ) kuies, kueus kue ) kuit kuius ) kuis, kuie, accusatif kuin kuie kueus datifkuedani kuedas locatif génitif kuel abl.-instr. kuez(za)
Le paradigme de singulier reste stable au cours de l’histoire. Au pluriel, le renouvellement morphologique est, au contraire, rapide et diversifié: comme l’indique la comparaison indo-européenne, la forme héritée est nom. anim. kues (v. lat. ques); des formes comme kuies (emprunt au thème de singulier) ou, plus rarement, kueus (remplacement de -s par -us) semblent parallèles aux tendances analogiques que l’on constate en latin où ques est concurrencé par quia en latin impérial (Meillet & Vendryes 1948: 501). A l’accusatif, seules les formes kuius (anim.) et kue (inan.) sont susceptibles d’être secondai-
17
Je remercie H. Craig Melchert pour l’obligeance avec laquelle il m’a permis d’utiliser son corpus électronique de textes authentiques (tablettes contemporaines aux rédactions) à partir duquel ces calculs ont été effectués.
Qu’est-ce qu’un pronom en anatolien?
81
rement affixés par -ki/-ga et par -a, ce qui indique leur ancienneté relative (confirmée, au plan comparatif, dans le cas de kue, infra § 9.1). Les quelques formes attestées du palaïte (nominatif et accusatif sg.) sont identiques à celles du hittite. En louvite, où la flexion du pronom est également peu documentée, les données suggèrent une opposition thématique similaire à celle du hittite: une désinence de nom. anim. pl. -nzi (h *kwi-ns) identique à celle que prend le démonstratif za¯face à une forme de nom.-acc. inanimé kui-Ø (cf. KUB 35.133 ii 3: kui#pa#ku#ni#as kui a[-) dont le zéro résulte d’un processus phonologique d’élimination des plosives finales (comp. nom.-acc. louv. malli-Ø KUB 35.39 ii 12, face à hitt. milit-Ø “miel” KUB 35.5 ii 9). (21)
le relatif-interrogatif kui- en louvite singulier pluriel animés inanimés animés nominatif kuis kuinzi kui-Ø accusatif kuin abl.-instr. kuwatti
La flexion du relatif lycien ti- préserve l’opposition thématique entre cas nucléaires (nominatif et accusatif) et cas périphériques, mais modifie la distribution des désinences en neutralisant la distinction entre nominatif et accusatif. (22)
le relatif-interrogatif ti- en lycien singulier pluriel animés inanimés animés inanimés nominatif ti ti tijei, ti tija accusatif ti ? dat.-loc. tdi
Le thème du datif-locatif tdi (h *tedi h *kwadi) est confirmée par la forme dérivée d’indéfini tdike (Laroche 1960: 176).
5. Pronoms clitiques 5.1. Morphologie Considérons maintenant la série des clitiques, les seuls pronoms spécialisés, en tant que classe, dans la représentation d’un antécédent
82
Sylvain Patri
lexical. Le simple fait que les pronoms clitiques des langues anatoliennes se trouvent correspondre entre eux de façon immédiatement évidente (27)-(28)18, alors que tel n’est pas le cas des pronoms forts, indique, à l’image de la situation attestée dans beaucoup de langues, que ceux-ci sont plus anciens que ceux-là. Les deux formes interlocutives n’appartiennent pas à la flexion même si leurs emplois sont restreints aux rôles sémantiques correspondant aux cas accusatif et datif-locatif19: (27)
les pronoms clitiques interlocutifs anatolien hittite palaïte f #mu #mu sg. 1 *#mu f #tta/ddu 2 *#tu f #nas pl. 1 *#nos f #smas 2 *#smos
Seul le pronom délocutif forme un paradigme, lequel est réduit à trois cas: (28)
paradigmes du clitique délocutif en anatolien nominatif anim sg. pl. inan. sg. pl. accusatif anim sg. pl. inan. sg. pl. datif-loc. anim sg. pl. inan. sg. pl.
anatolien *#os *#oi *#od *#oi *#om *#ons *#od *#oi *#soi *#smos f f
f f f f f f f f f f – –
hittite #as #e f #at #at #e f #at #an #us f #as #at #e f #at #se #smas
palaïte #as #as #ata #ata #an #as #at #e #du #mmas
louvite lycien #as #ata
#an ##(n#) #ata #ata #du #ñne
##(n#) #ede #e(de) #i
Ainsi que le montre ce tableau, la forme délocutive tend également à s’extraire de la flexion, notamment au genre inanimé, réduit, en hittite récent et en louvite, à une seule forme invariante. 18 19
Kammenhuber 1969, 308, Meriggi 1980, 316sq., Garrett 1990a, 22sq. Meillet 1936, 291, 1937, 333, estimait qu’en indo-européen, les interlocutifs clitiques n’étaient pas soumis à la flexion; la tentative de Rasmussen 1987, visant à établir le contraire ne repose que sur des conjectures.
Qu’est-ce qu’un pronom en anatolien?
83
5.2. Aptitudes référentielles des pronoms clitiques L’existence, dans le paradigme délocutif, d’une forme de nominatif distingue l’anatolien des autres dialectes indo-européens où existent des pronoms similaires. Mais on doit aussitôt préciser qu’existe en anatolien une contrainte syntaxique généralisée interdisant la présence des pronoms clitiques au nominatif – donc délocutifs – dans les constructions avec objet à l’accusatif (Garrett 2003, d’après Watkins). Comme les interlocutifs, dépourvus de nominatif, ne peuvent pas assumer le rôle sémantique du sujet, et que l’entité susceptible de manipuler un patient est présumée animée, il s’ensuit que les pronoms clitiques s’avèrent, en tant que classe, inaptes à représenter le sujet transitif20. (29)
aptitudes morpho-syntaxiques des pronoms clitiques interlocutifs: délocutifs:
sujet intransitif sujet transitif accusatif datif-locatif + + + + +
Dans une langue où le verbe des constructions transitives est obligatoirement et exclusivement indexé au sujet, cette propriété a pour conséquences que le sujet d’une construction transitive peut être représenté par un pronom fort, mais jamais par un pronom clitique. En dépit de la divergence notable que représente l’existence de formes de nominatif #as / #at, la situation de l’anatolien ne diffère de celle des autre langues indo-européennes anciennes que sur un seul point: que les pronoms clitiques délocutifs ont accès au rôle de sujet dans les construction sans objet direct.
6. Analyse morphologique 6.1. Segmentation L’analyse morphologique de la flexion des proformes (pronoms délocutifs, pronom relatif) est problématique parce que la commutation par homologie avec les désinences des autres flexions ne coïncide pas avec la commutation par homologie avec les sections invariantes des bases. 20
Détails dans Patri 2007, 64sq.
84
Sylvain Patri
Le problème vient de la présence, à certains cas, d’éléments prédésinentiels ou post-radicaux -ed- et -an-. Quand -an- est présent, c’est uniquement à la suite de -ed-: dat.-loc. anim. aped-an-i, ked-an-i, instr. anim. aped-an-da, kid-an-da. La présence de -ed- est évidente au dat.-loc. anim. ap-ed-ani, k-ed-i, inan. ap-ed-as, k-ed-as (étendu par la suite au génitif); à l’instrumental ap-et, ainsi qu’à l’ablatif apez [apets], kez [kets]. Il semble également possible de postuler son existence aux génitifs apel, kel, car les séquences tl, dl sont, de façon générale, interdites dans la phonotactique du hittite21, ce qui, même en l’absence de témoignages positifs d’un changement *t/d f Ø /__ l, autorise à postuler un processus *-ed-l f -el. En fait, tous les cas, à l’exception du nominatif et de l’accusatif, s’avèrent, sûrement ou possiblement, concernés par l’extension radicale -ed(-an)-: (23)
cas obliques des proformes en hittite singulier pluriel génitif *-ed-l (f -el) *-ed-so¯m? (f -enzan) dat.-loc. *-ed-i, *-ed-an-i *-ed-as instr. *-ed-t, *-ed-an-da abl. *-ed-ti (f -ez)
On ne rencontre pas trace de -ed- et de -an- en louvite ou en palaïte, ce qui indique un développement particulier du hittite, en tout cas, certainement postérieur à l’anatolien commun22. Dans cette perspective, l’analyse de la désinence de génitif pluriel -enzan est indécise: pour l’anatolien commun, Oettinger (1994: 326) et Melchert (1994: 121, 2008: 370 n. 15) partent d’un un prototype *-oi-so¯m (infra, § 9.2) avec émergence d’une nasale en contexte [__s …N] suivie d’une épen-
21
22
Cf. Melchert 1994, 156–157, “the absence of dental plus */l/ is systematic”. Les flottements graphiques dont témoigne, par ailleurs, un terme vraissemblablement emprunté comme le titre et nom propre labarna: tabarna (Tischler 1988, Soysal 2005), s’expliquent probablement en relation à cette contrainte. Les tentatives de Pedersen 1938, 55–56, et de Benveniste 1962, 75sq., visant à expliquer les formes hittites en … ed-an- en les rapprochant de données indoiraniennes sont anachroniques (ce que l’état des connaissances à leur époque ne permettait pas de savoir). Petersen 1937, 314–315, conjecture des mouvements analogiques croisés. Hoffner & Melchert 2008, 144, se limitent à remarquer “unclear is the source of the elements -d- and -an- in the oblique cases”; Melchert 2008, 370, postule que -ed- reposait sur *-éd- “with regular lengthening in accented open syllable”.
Qu’est-ce qu’un pronom en anatolien?
85
thèse de [t] en contexte [n__s], *-ointso¯m f -enzan23. Mais dans la perspective d’une évolution strictement hittite, il paraît également possible de postuler le remplacement de *-oi- par *-ed- avec un prototype *-ed-so¯m suscitant pareillement l’émergence d’une nasale en contexte [__ts …N] (Oettinger 1994: 322), d’où *-entso¯m ou *-endzo¯m faisant reconnaître un thème unitaire pour tous les cas obliques. 6.2. Les désinences Une conséquence de (23) est de faire apparaître une organisation flexionnelle fondée sur des désinences casuelles distinctes de celles que suggère, à première vue, une commutation des paradigmes orientée par la gauche (ap-el, ap-edani, etc.). Aux cas qui ne présentent ni -ed-, ni -an-, les animés présentent des désinences identiques aux marques flexionnelles des noms de façon évidente au nominatif apa¯s, ka¯s, de façon moins clairement décidable à l’accusatif où, dans les désinences, ap-u¯n, k-u¯n, le vocalisme [u] est spécifiquement pronominal, mais pas la consonne [n]24. Au pluriel, la désinence pronominale -e neutralise les genres au nominatif. A l’inanimé, les délocutifs et le relatif-interrogatif utilisent une marque spécifiquement pronominale -t, laquelle alterne avec un zéro, occasionnellement caractéristique des noms. La datif-locatif sg. *-ed(-an)-i: pl. -as, est représenté par les désinences régulièrement affectées à ce cas dansla flexion des noms, mais dans des conditions qu’il convient de prendre en considération puisqu’une forme comme -edi ne peut représenter l’aboutissement d’un processus phonologique normal, la 23
24
La solution de Kloekhorst (2008, 192, 427), qui reconstruit *-énHsom, au motif que “-VnzV- can only reflect *-nHs- (whereas PAnat. *VnsV > Hitt. VssV)”, est plus problématique, un élement *-énH- n’ayant pas de base comparative, ni même d’explication en hittite. Les autres langues anatoliennes ont une désinence {-an}: louvite a-apa-an (KUB 35.103 iii 3) = palaïte a-pa-an (KUB 35.165 Ro 15, 20); louv. za-am#pa (KUB 35.103 iii 4). – Dans l’hypothèse où hitt. -u¯n serait le résultat d’un processus phonologique régulier, l’élévation et l’allongement concomitant de *[o] devant nasale se justifie, selon Kloekhorst 2008, 99, 192, 426 (contre Benveniste 1962, 71sq.), par le caractère accentué de la syllabe (*k´-óm f k-u¯n, mais *péd-om f ped-an). La solution de Kloekhorst n’est pas contrôlable empiriquement, mais elle est phonétiquement vraisemblable (sur les variations de hauteur des voyelles selon l’accent en contexte __N, voir les études typologiques de Foley 1975, Hajek 1997, 100sq.).
86
Sylvain Patri
palatalisation de la plosive coronale étant régulière devant __i (voir ci-après). Le génitif en -el est indifférent au nombre, alors que l’ancien génitif -(e)nzan, supplanté par -el, ne se rencontre, avec les pronoms délocutifs, que dans les paradigmes de pluriel et, parmis les interlocutifs, dans le paradigme de “vous”. En louvite, le génitif a été éliminé au profit d’un adjectif dérivé apassa/i- ; ce cas n’est pas documenté dans les autres langues anatoliennes. Les désinences -el et -enzan sont l’une et l’autre spécifiquement pronominales en anatolien, ce qui est également le cas, en indo-européen, de la désinence *-so¯m sur laquelle -enzan repose d’une façon ou d’une autre (§ 5.1). A l’instrumental, des variantes comme uddanit: uddanta “parole”, wetenit: witanta: “eau” rendent plausible la lecture d’une désinence {-t}, suscitant une syllabation tantôt CV = [-ta] (écrite -t/da), tantôt VC = [-it] (ainsi Neu 1979: 190, Hoffner & Melchert 208: 77). Dans cette perspective, ket ne doit pas nécessairement être vu comme le produit de l’évolution phonologique, mais, possiblement, comme le résultat d’une reconduction secondaire de la désinence sur le thème (*k-éd-t aurait normalement conduit à †kezta, selon le modèle illustré par 3sg. prét. *h1éd-t f e-ez-ta [etsta] “il a mangé”). A l’ablatif, l’élément -ed- se retrouve à l’ablatif-instrumental de tous les pronoms interlocutifs, où la divergence *-ed-ati f *-ed-ats f hitt. -edaz, mais *-ati f louv. -ati, rhot. -ari (tabl. 4), apporte une preuve supplémentaire du caractère spécifiquement hittite, c’està-dire récent, de la diffusion de *-ed-. Ici encore, kez peut tout autant s’expliquer en fonction d’un prototype *kéd-ti f kez selon le processus reflété par *h1éd-ti f e-ez-za-(a-)i “il mange” (où -i est secondairement restauré) que par reconstitution analogique secondaire où la désinence -az(a) = {-ts} issu de *-ti est simplement affixée à une base ked-. La seule chose sûre est que le datif-locatif, l’instrumental et l’ablatif des pronoms, sont, à des titres divers, des formes restructurées. Le paradigme de kui- n’est pas différent de celui des délocutifs en ce qui concerne la distribution des désinences, mais face au thème {kwed-} des cas obliques, on rencontre au pluriel du nominatif et de l’accusatif une diversité thématique irréductible à d’autres modèles.
87
Qu’est-ce qu’un pronom en anatolien?
6.3. Les thèmes La présence d’un élément -ed-(an-) entre les thèmes de base et les désinences des cas autres que le nominatif et l’accusatif (14) représente une organisation flexionnelle en apparence originale, mais non dépourvue de parallèles dans les langues indo-européennes. La flexion du démonstratif védique reflète une organisation comparable à celle du hittite avec un élément -sm- (fém. -sy-) inséré entre la base et les désinences des cas périphériques: (24)
flexion du démonstratif védique masc. neutre nominatif ta tad accusatif tam génitif tásya ablatif tá-sm-a¯d datif tá-sm-ai locatif tá-sm-in instrumental téna
fém. sa¯ ta¯m tá-sy-a¯s tá-sy-ai tá-sy-a¯m táya¯
Une hypothèse souvent (re)formulée est que -sm- / -sy- serait un morphème, autrement dit, que cet élément serait doué d’une signification, en l’espèce celle de *sem- “un, même” ou celle d’une “particule” ad hoc dont on imagine la valeur d’autant plus librement qu’elle n’est nulle part attestée25. Une approche de ce type est toutefois peu vraisemblable: outre que les flexions indo-européennes sont invariablement fusionnelle, on ne voit guère comment les pronoms, par définition dépourvus de référent lexical, viendraient à entrer en composition avec d’autres lexèmes, qui plus est, à certains cas seulement; dans la variété des processus conduisant à former des pronoms (Diessel 1999), nulle langue n’atteste quoi que ce soit de semblable. Les organisations morphologiques des flexions pronominales du védique et du hittite, sans être fréquentes, ne présentent, en réalité, aucun caractère exceptionnel. Considérons la flexion du pronom 2sg. en vieux slave qui, elle aussi, présente à tous les cas, sauf au nominatif, 25
Lane 1961, Cohen 1976, Beekes 1988, 78sq., Szemerényi 1996, 205–206, Carruba 2000, 2002, dont les conceptions sont évoquées avec réserves par Sihler 1995, 386–387, et par Katz 1998, 92–93, 156 n. 9.
88
Sylvain Patri
un élément constant -b- là où le hittite présente -ed-(an-) et le védique -sm- (18). Les données comparatives indiquent que -b- provient de ce qui était originellement un morphème de datif *-bh-: (25)
flexion du pronom personnel 2 sg. en vieux slave v. slave latin védique avestique nominatif ty tu¯ tvám tuuwm ¯ accusatif teb-e te tvám a˛m génitif teb-e tui táva tauua¯ O ˇ datif teb-e ti-bi tú-b ya(m) tai-biia¯ instrumental tob-ojo˛ tvá a¯ locatif teb-eˇ tvé
A un moment de l’évolution, le datif *te-bhei (lat. tibi, pruss. tebbei) f sl. *te-beˇ a été réinterprété *teb-eˇ en conséquence de quoi s’est formé un nouveau radical teb-, lequel a été analogiquement diffusé dans le reste de la flexion, sauf au nominatif. On a pris l’exemple du pronom “tu” parce que c’est celui qui a le plus de formes apparentées dans d’autres langues, mais on pourrait aussi bien citer le réfléchi sebe, sebeˇ, sobojo, sebeˇ, dont la flexion est similaire, à ceci près qu’elle n’a pas de nominatif. Un tel schéma d’évolution n’est lui-même qu’une variante d’un processus de nivellement paradigmatique des radicaux par absorption des désinences dont les exemples ne sont pas rares dans les langues dont l’histoire est documentée. En polonais, par exemple, le thème de présent du verbe “être” est passé de jes- à jest- en incorporant une ancienne désinence: (26)
flexion du verbe “être” en polonais langue ancienne état moderne sg. 1 jes´-m jest-em 2 jes´-Ø jest-es´ 3 jes-t jest-Ø pl. 1 jes-my jest-es´my 2 jes´-cie jest-es´cie 3 sa˛ sa˛
Dans les flexions verbales, la forme fléchie de 3ème personne semble être toujours celle qui sert de pivot à la restructuration analogique des radicaux (voir Hock 2003, pour d’autres exemples). Avec les pronoms, les choses sont plus complexes car il n’existe pas de cas oblique dont
Qu’est-ce qu’un pronom en anatolien?
89
on pourrait dire qu’il serait universellement plus ou moins marqué que les autres. En slave, l’élection du cas datif comme pivot (16) peut s’expliquer, entre autres hypothèses, en raison de son poids syllabique (la désinence en *-bh- est la seule du paradigme qui débute par une plosive), mais bien d’autres interprétations peuvent être imaginées. En hittite, la solution la plus directe, à défaut d’être contrôlable, serait de reconnaître les thèmes aped-, ked- comme reposant sur la réinterprétation de formes fléchies d’instrumental aped, ked (comp. les formes d’ablatif archaïque en latin med, ted, sed). Le fait que les seules formes interlocutives qui présentent -ed- sont des formes d’ablatif ammedaz, etc. (1)-(2) pourrait conforter cette hypothèse, mais celles-ci sont récentes et le témoignage du relatif kuez ne s’y conforme pas. On se limitera donc à tenir qu’à l’instar de -b- en slave, hitt. -ed-, -an-, ou véd. -sm- représentent, selon toute vraisemblance, d’anciennes désinences ou fragments de désinences amalgamées aux thèmes pronominaux, mais qu’en raison de la réinterprétation à laquelle ce matériel a donné lieu, il n’est plus possible d’en identifier le rôle précis par comparaison avec d’autres langues, en tout cas, pas de façon sûre. Un autre point commun à (14)–(16) est que, quelle que soit la désinence utilisée pour normaliser un thème pronominal, cette normalisation n’atteint jamais les cas nucléaires (jamais le nominatif, rarement l’accusatif), ce qui en définitive, revient à aligner l’organisation thématique des pronoms délocutifs sur celle des pronoms interlocutifs.
7. Catégorisation morphologique des pronoms 7.1. Paramètres Des données présentées ci-dessus, il ressort que l’identification d’une forme comme pronom repose sur la conjonction de trois critères morphologiques réguliers à toutes les époques pour toutes les catégories de pronoms, à l’exception des formes clitiques: (i) (ii)
présence de désinences -enzan et / ou -l au génitif; présence d’un radical en … ed(-an)- à au moins un des cas périphériques;
90 (iii)
Sylvain Patri
supplétion opposant le radical invariant des cas obliques aux autres cas (à l’exception de 2 pl. sum-: § 2.3)26;
auxquels ont peut ajouter deux autres propriétés caractéristiques des proformes seulement: (iv) (v)
présence d’une désinence (de pluriel) -e au nominatif animé et au nominatif-accusatif inanimé; présence d’une désinence -t au nominatif-accusatif singulier inanimé.
Certaines de ces propriétés peuvent isolément être reflétées dans certaines flexions: celle des numéraux *sia- “un” (gén. siel, dat.-loc. siedani) et “deux” (gén. 2-el)27; celle de quantifieurs comme tamai“autre” (gén. tamel, dat.-loc. tamedani, abl. tamedaz); dapi(ya)- “tout, entier” (abl. dapedaz, mais gén. dapiyas); la supplétion est régulière avec les noms inanimés “hétéroclitiques” du type nom.-acc. uttar-Ø: obl. uttan- “parole” et concerne aussi quelques formations isolées, noms (antuwahha- et antuhsa- “être humain”) ou verbes (tar- et te“dire” Oettinger 1979[2002]: 109sq.). Face à ces données, le point intéressant est que (i)-(iii) sont simultanément utilisées par les pronoms exclusivement et que seuls les mots dont le comportement syntaxique et référentiel est celui de pronoms cumulent ces propriétés. A l’exception de “vous” (§ 2.3), l’identification d’un pronom peut donc se conclure sur une base morphologique sans qu’il soit obligatoire de faire intervenir la syntaxe. Autrement dit, les pronoms, en anatolien, constituent une classe formelle. 7.2. Pronoms et autres formes co-variantes Une conséquence de cette observation est de révoquer certains usages traditionnels du terme de “pronom” pour identifier des mots ou classes de mots qui ne satisfont pas au cumul de propriétés caractéristique de cette classe:
26
27
On rejoint ici une observation déjà faite par Sihler 1995, 384 “suppletion can be used as a criterion for deciding what was and as not a pronoun in PIE”. Sur “1”, voir Goedegebuure 2006; sur la flexion des numéraux, voir Eichner 1992, 32 (“1”), 52 (“2”).
Qu’est-ce qu’un pronom en anatolien?
(a)
(b)
28
29
91
le déictique asi qu’on a proposé d’identifier comme un “pronom démonstratif” spécialisé dans la déixis distale (ainsi Goedegebuure 2003a/b, dont les conclusions sont adoptées par Hoffner & Melchert 2008: chap. 7), présente une flexion idiosyncrasique où on ne discerne ni radical unitaire aux cas obliques, ni désinence pronominale (génitif uniy-as, nom.-acc. inanimé in-i f en-i). A l’instar du numéral “un” et des quantifieurs, il ne se rapproche de la morphologie pronominale que par une seule propriété (dat.-loc. edi f ed-an-i). En vieux hittite, ce terme se limite à modifier des noms anaphoriques Ni …, asi Ni et n’est pas utilisable comme anaphorique. Son emploi répondant à la présomption que le N modifié par asi est le seul à satisfaire la propriété N au sein d’un domaine de référence, il est préférable de le catégoriser comme déterminant de définitude (Patri 2008), même si, comme c’est banalement le cas des déterminants (grec ancien, langues romanes), il dérive banalement d’un pronom (Patri 2008: 153–154)28; les possessifs clitiques souvent désignés comme “pronoms”29 utilisent une désinence pronominale -(e/i)t au nominatif-accusatif, mais des désinences nominales sg. -as: -pl. -an au génitif et ignorent la supplétion. En tant que formes co-variant avec le possédé (genre, cas et nombre amalgamés dans la désinence) et avec le possesseur (personne et nombre amalgamés dans le thème), les possessifs ne désignent pas un référent en particulier; ils ne peuvent pas non plus assumer un rôle de type adjectival (comp. catalan el meu cosí / amic “mon cousin / ami” [litt. ‘le mien cousin’], els seus cosins “ses cousins”, un meu llibre “un livre à moi”). Le comportement des possessifs en vieux hittite est exclusivement celui de modifieurs du nom, ce qui les assi-
Dans mon article de 2008, je n’ai pas tenu compte de certains passages grammaticalement fautifs où les emplois de asi ne se conforment pas à cette caractérisation: KUB 26.1 iii 37–40; KBo 17.17 (+?) KBo 30.30 Ro 6–7, dupl. KUB 43.53 i 16–18 (erreur de cas); KUB 48.119 Vo 11–13 (erreur de genre); le sens général de KBo 5.3 ii 32–38, est, pour moi, obscur. Ces exemples sont tous transmis par des textes ou copies tardives, ce qui tend à indiquer que l’emploi de asi n’était plus compris en hittite impérial. Ainsi, par exemple, Hoffner & Melchert 2008: chap. 6, mais Luraghi 1997, 22, parle d’“adjectifs”.
92
Sylvain Patri
mile à des déterminants possessifs (comme mon, ton, son en français, distincts des pronoms mien, tien, sien)30. Il paraît probable que si les déterminants possessifs comme le déterminant défini adoptent certaines caractéristiques pronominales, c’est en raison de leur proximité avec les démonstratifs avec lesquels ils forment une classe de “déterminants actualisateurs”. Le fait que le nominatif-accusatif “pronominal” -(e/i)t des possessifs co-varie avec le cas du nom possédé, et non avec celui du possesseur, est au demeurant significatif d’une captation analogique rigoureusement étrangère à la dénotation pronominale.
8. Intégration du nombre dans la morphologie 8.1. Supplétion et nombre Les pronoms anatoliens font de la catégorie du nombre un critère de structuration curieusement négligé dans la littérature spécialisée. On commencera par exposer les incidences des interactions entre le nombre et la morphologie pronominale avant d’en observer plus bas (§ 10.4) les conséquences dans l’organisation syntaxique de la référence. Les données résumées dans (30) semblent indiquer que l’orientation du supplétisme pronominal serait gouvernée par la situation d’énonciation: les pronoms interlocutifs opposent le radical de nominatif à tous les autres cas, alors que les pronoms référés à d’autres entités que “je” et “tu”, quel que soit leur type de spécialisation, ont en commun d’opposer les thèmes de nominatif et d’accusatif à tous les autres thèmes. Mais il est intéressant de constater que le rôle discriminant tenu par la situation d’énonciation est, de fait, équivalent à un paramètre d’accord. En effet, selon qu’un thème pronominal stipule une information relative au nombre, comme c’est le cas des interlocu30
Le recours au terme de “déterminant” est, par ailleurs, légitimé en observant que dans les constructions possessives, les noms dont le référent ressortit à la possession inaliénable sont obligatoirement hôtes d’un possessif clitique (Garrett 1998). Or la distinction entre défini et indéfini est normalement indifférente avec les noms relevant de la possession aliénable (mon livre = le livre à moi / un livre à moi), mais pas avec les entités relevant de la possession inaliénable (attas#mis “mon père” = le père à moi / *un père à moi).
Qu’est-ce qu’un pronom en anatolien?
93
tifs, ou qu’il est indifférent au nombre, ce qui est le cas des autres pronoms, on peut prédire si, à l’accusatif, ce thème sera assimilé soit au nominatif, soit aux autres cas. Une telle organisation est complètement différente de celle des noms et adjectifs qui différencient l’accusatif ou l’assimilent au nominatif selon le genre. En anatolien, le nombre est à la structuration flexionnelle des pronoms ce que le genre est aux noms. 8.2. Interlocutifs et nombre Un autre aspect du rôle tenu par le nombre dans l’organisation des pronoms concerne la délimitation entre les pronoms interlocutifs et les autres: les premiers utilisent des thèmes amalgamant la personne et le nombre tandis que les seconds utilisent des thèmes indifférents au nombre et au genre. En d’autres termes, selon qu’un radical pronominal dénote ou pas une information relative au nombre, on peut prédire si le pronom est interlocutif ou pas.
9. Pronoms et cas en vieux hittite 9.1. Supplétion et spécification casuelle En anatolien, l’organisation des pronoms interlocutifs comme délocutifs se caractérise par une opposition régulière entre les thèmes du nominatif et de l’accusatif (quand il n’est pas amalgamé à d’autres cas) et tous les autres. Avec les pronoms interlocutifs, ce supplétisme prend la forme d’un modèle où le nominatif présente une forme inanalysable par rapport aux autres cas; avec les autre pronoms, elle s’incarne dans un modèle où les formes fléchies utilisent un seul et même radical, sauf aux cas nucléaires. La situation de 2 pl. reste à part. (30)
la supplétion pronominale 1 sg. 1 pl. 2sg. nom.: u¯k wes zik nom.-acc.: autres cas: amm- anz- tu-
en hittite 2 pl. 3e dém. dém. rel.-int. sum#a- ka¯- apa¯- kuisum- #s- ked- aped- kued-
Cette situation est typologiquement banale, mais les données hittites font discerner que le pivot du supplétisme opère uniformément
94
Sylvain Patri
dans le sens d’une préservation ou d’un réaménagement du thème des cas périphériques en vue de le différencier du thème de nominatif, parfois d’accusatif. Tout se passe comme si, dans la référence pronominale, les paradigmes se trouvaient dans la nécessité, pour exister, d’introduire une sous-spécification formelle entre les cas nucléaires, les plus aptes à fonctionner en relation directe avec le verbe (sujet, objet), et les autres. 9.2. Spécificité casuelle des pronoms personnels Les proformes dans leur ensemble (démonstratifs, relatif-interrogatif) distinguent les cas datif(-locatif) et accusatif là où les pronoms personnels, qu’ils soient clitiques ou forts, les confondent: (31)
accusatif et datif dans les flexions pronominales en vieux hittite démonstratif relatif pronoms clitiques forts accus. datif
3sg. 2sg. 1sg. 2sg. 1sg. #an #tta / #du #mu tuk ammuk ku¯n, apu¯n kuin kedi, apedani kuedani #se
Les pronoms personnels opèrent un syncrétisme accusatif-datif là où les pronoms et les noms animés maintiennent la particularité de l’accusatif et où les pronoms et noms inanimés syncrétisent l’accusatif avec le nominatif. Il existe donc une corrélation dissociée entre l’animation et le marquage de l’accusatif. (32)
animation et marquage des cas grammaticaux en vieux hittite pronoms personnels noms animés noms inanimés N≠A
N≠A
N=A
A=D
A≠D
A≠D
La différenciation formelle du nominatif et de l’accusatif selon le caractère animé du référent est absolument régulière en anatolien (les participants du discours “je” et “tu” sont présumés animés). Or une telle propriété est typique des langues dont l’alignement prédominant est de type accusatif. En effet, Baerman, Brown & Corbett (2005: 47 sq.) ont récemment mis en évidence le fait – inattendu, a priori – que dans les langues à alignement ergatif, le syncrétisme équivalent (abso-
Qu’est-ce qu’un pronom en anatolien?
95
lutif-ergatif), quand il existe, n’est jamais corrélé à l’inanimation. Il s’ensuit que le marquage casuel peut, de façon générale, être corrélé par l’animation, par l’animation et l’inanimation, mais jamais par l’inanimation seule. Cette observation vient s’ajouter à la liste des données qui réfutent l’interprétation selon laquelle l’alignement prédominant des langues anatoliennes aurait été ergatif: dans une langue où seuls les noms ont un marquage ergatif, il est impossible que ces noms forment une classe fondée sur le genre inanimé31. Le syncrétisme accusatif-datif des pronoms personnels est en revanche sans équivalent dans les autres langues indo-européennes anciennes, même s’il peut se produire dans certains états de langue plus récents (arménien oriental, par exemple). 9.3. Spécificité casuelle des démonstratifs Les démonstratifs connaissent également une délimitation par rapport aux autres pronoms en l’espèce de l’architecure casuelle. Les démonstratifs distinguent un cas ablatif d’un cas instrumental là où les pronoms personnels et le pronom interrogatif-relatif utilisent l’ablatif pour l’instrumental. L’architecture des possessifs est similaire à celle des pronoms personnels, mais avec une expression opposée dans laquelle l’instrumental est, cette fois, utilisé pour l’ablatif. (33)
ablatif et instrumental dans les flexions hittites possessif relatif personnel démonstratif noms instrumental -e/it = abl. = abl. -e/it, -edanda -e/it, -anda ablatif = instr. -ez(z a) -edaz -ez -za(a)
Dans certaines locutions figées, il arrive qu’un démonstratif dépendant soit marqué à l’instrumental alors que le nom-tête est à l’ablatif: kit pantalaz “de ce moment”, apit pantalaz#pat “à partir de 31
Dans la version soutenue par Garrett à la suite de Laroche: “The Anatolian branch of Indo-European is characterized by a split-ergative case-marking system in which neuters [= inanimés] inflect ergatively and common-gender nouns [= animés] inflect accusatively” (Garrett 1990b, 261). Les arguments réfutant cette thèse sont développés dans Patri 2007, achevé avant que je prenne connaissance de Baerman, Brown & Corbett 2005. Les critiques que Melchert (s.p.) adresse à mon livre reposent, en grande partie, sur une confusion entre propriétés de comportement et propriétés de marquage des arguments nominaux.
96
Sylvain Patri
ce moment-là” (KUB 33.118 i/iv? 24, rédac. récente, Laroche 1969: 189), ce qui suggère que des formes comme kez, apez pourraient avoir été secondairement introduites dans la flexion des démonstratifs d’après celle des noms, c’est-à-dire par analogie (ce qui ne signifie pas obligatoirement que l’attribution d’une désinence spéciale au cas ablatif soit récente)32. Quoi qu’il en soit, le point à retenir est qu’il existe au moins un paramètre en fonction duquel l’organisation casuelle des démonstratifs peut être opposée à celle de tous les autres pronoms. 9.4. Neutralisation Les phénomènes de différenciation des pronoms entre eux selon le syncrétisme casuel que l’on vient d’observer sont limités au vieuxhittite. Dans les états de langue plus récents, la confusion du nominatif et de l’accusatif d’une part, la disparition du cas instrumental de l’autre, effacent les anciennes distinctions.
10. Anatolien et indo-européen 10.1. Désinences Les convergences entre l’anatolien et d’autres langues regardant les morphèmes flexionnels spécifiquement pronominaux se limitent, en définitive, à trois situations: (i)
(ii)
32 33
nominatif-accusatif singulier *-d (lat. i-d, véd. tá-t, got. i-ta, v. sl. *tod-je f tozˇde “le même”) f -t là où les noms et adjectifs de même genre utilisent -Ø ou -n, convergence limitée aux flexions de apa¯- (ka¯- utilise zéro) et de kui-33; nominatif animé et nominatif-accusatif pluriel inanimé *-oi (masc. véd. te, gr. dor. toí, lat. hi, got. pai) f -e, là où les noms
Jasanoff 1972, 126, Hoffner & Melchert 2008, 147. La désinence était voisée comme le montrent les témoignages du germanique et du slave; il n’est pas correct de postuler, avec Gamkrelidze & Ivanov 1984, 275–276 n. 1, une homologie de ce morphème avec la finale de nominatif-accusatif des noms inanimés du type de *yékw0r-t “foie” (véd. yákflr-t) qui était non voisée (gr. hàpatos).
Qu’est-ce qu’un pronom en anatolien?
(iii)
97
et adjectifs de même genre utilisent -es (anim.) ou -Ø / -a / -i (inan.), convergence limitée aux pronoms soumis à la variation en genre; génitif pluriel *-so¯m (véd. tàflsa¯m, lat. eo¯rum, v. sl. sixu˘) f -(en)zan, supplanté par -el dans les textes récents, là où les noms et adjectifs utilisent -an (supplanté par -as), convergence impliquant les démonstratifs, ainsi que l’interlocutif 2pl.
Ces trois propriétés sont appuyées sur les cas grammaticaux (nominatif, accusatif, génitif), jamais sur les cas locaux. Elles ne concernent que les proformes (pronoms délocutifs et relatif), compte non tenu du statut spécial de hitt. 2 pl. summenzan (supra, § 2.3). Les autres morphèmes flexionnels trouvant des correspondances dans d’autres langues indo-européennes (notamment le nominatif, l’accusatif et le datif-locatif) ne sont pas spécifiquement pronominaux. 10.2. Radicaux Les convergences des radicaux sont, elles, évidentes en ce qui concerne les bases de presque tous les pronoms personnels (2 pl. excepté), le relatif-interrogatif kui-, et le démonstratif ka¯-. Il est bien connu que ka¯- correspond en grec, latin, celtique, arménien, germanique, balte et albanais avec divers témoignages de pronom démonstratif plus ou moins figé (univerbé)34, mais on ne semble pas encore avoir signalé que les emplois syntaxiques du démonstratif *k´ comme une des formes apte à tenir le rôle de pronom de 3e personne étaient attestée en anatolien exactement comme dans un autre dialecte, en slave. En vieux slave, les emplois de si sont, à tous égards, similaires à ceux de hitt. ka¯-: il désigne un référent proche par opposition à onu˘ et tu˘ (Vaillant 1964: 140–141); il assume des emplois d’ostenseur (34), de modifieur du nom (35) comme d’anaphorique (36); il est le plus souvent préposé à la tête, mais admet aussi d’être postposé, comme dans (35); comparer les données exposées, supra, § 3.1, avec les exemples ci-dessous:
34
Tischler HEG I, 456, Phuvel, HED IV, 1997, 12, Kloekhorst 2008, 425sq.
98 (34)
Sylvain Patri
se mati moeˇ i dém.-nom.-acc. mère-nom.sg. poss.1sg. coord. bratrieˇ moeˇ frère-nom.pl. poss.1sg. π 0 λ ¹ $ (Mt. 12.49) “voici ma mère et mes frères”
(35)
na sodu˘ azu˘ vu˘ miru˘ prép. jugement-acc. 1sg. prép. monde.-acc. dém.-acc. si pridu˘ venir-1sg.aor >« (Ω >« μ @B (J. 9.39) “c’est pour un jugement que je suis venu en ce monde”
(36)
sı˘ pristopi ˛ ku˘ pilatu dém.-nom. se diriger-3sg.aor. vers P.-dat. « BΩ ) 7 P) . (L. 23.52) [survint Joseph d’Arimathie …] “il alla trouver Pilate”
Pour d’autres témoignages, voir SJSS IV: 373sq., Vecˇerka 1993: 56sq. Morphologiquement, le génitif pluriel *koiso ´ ¯ m f hitt. kenzan correspond exactement avec v. sl. sixu˘ où la flexion sur thème palatalisé est secondaire; voir, de même, lit. sˇiu˛˜, pruss. schieison (69/20)35. De telles correspondances peuvent être interprétées de deux façons: soit comme un héritage commun, ce qui est possible, mais difficilement démontrable en présence d’organisations autrement structurées dans la plupart des autres dialectes; soit comme une conséquence de la sélection de *k´ comme substitut de constituant nominal, mécanisme au terme duquel ce pronom se voit attribuer les rôles dévolus aux démonstratifs aptes à servir la “troisième personne”. Dans cette dernière hypothèse, le point mis en lumière ne serait pas tant la spécialisation de *k´ dans la déixis proximale (déjà attestée par les formes figées pour “aujourd’hui”: v. sl. d ˘ n ˘ -si, lit. sˇiañ-dien, v. sax. hi-udiga), que sa propension à servir de substitut nominal, propriété limitée à certains démonstratifs indo-européens seulement. 35
Détails dans Vaillant 1958, 383, Stang 1966, 233. Il est possible que le slavon russe seˇxu˘, attesté dans les Pandectes de 1296, reflète la banale confusion de i et de eˇ (Sobolevskij 1907, 189), mais on ne peut pas complètement exclure que cette forme reflète l’état le plus ancien.
Qu’est-ce qu’un pronom en anatolien?
99
10.3. Rétentions et innovations Jusque dans les années 1950, les études portant sur les pronoms de l’anatolien estimaient qu’on était face à des organisations foncièrement “autres” par rapport aux autres langues indo-européennes36. Depuis lors, l’accroissement du matériel empirique, la clarification de sa chronologie interne, la meilleure compréhension des textes, de façon générale, rend une telle conception caduque. L’architecture des flexions pronominales de l’anatolien ne diffère de celle de tous les autres dialectes pris ensemble que sur peu de points: (i)
(ii)
(iii)
existence d’une forme pronominale clitique pour le rôle sémantique de sujet (nominatif), forme au demeurant soumise à des contraintes syntaxiques telles (§ 6.2) que son emploi n’est possible que dans les constructions intransitives; en vieux hittite, neutralisation du genre au nominatif pluriel des démonstratifs (le hittite a étendu à l’inanimé la désinence *-oi reflétée comme masculin [h *animé] en grec, védique, gotique et latin); en vieux hittite, syncrétisme de l’accusatif et du datif des pronoms personnels (§ 9.1).
Les points (ii) et (iii) représentent nettement des innovations; le point (i) est moins clair: Garrett considère qu’il s’agit d’une innovation, mais, comme souvent dans les situations où l’anatolien manifeste une propriété distincte de la situation commune à tous les autres dialectes indo-européens, le point de vue opposé peut aussi être soutenu. Quoi qu’il en soit, autant il est clair que la morphologie de pronoms anatoliens reflète certaines innovations, autant celles-ci doivent d’autant moins être surestimées que le même phénomène peut, à des degrés divers, être observé dans les morphologies pronominales de tous les autres dialectes.
36
Ainsi Petersen 1937, 306 “the declension of demonstratives in Hittite […] shows only a slight resemblance to the pronominal declension of the other I[ndo]E[uropean] languages”; Benveniste 1953, 255 = 1962, 66 “la flexion pronominale [hittite] n’a […] à peu près rien de commun avec le système admis comme indo-européen”.
100
Sylvain Patri
Comme pour plusieurs autres dialectes, la situation de l’anatolien peut être grossièrement résumée en observant que la morphologie héritée des pronoms se localise principalement dans les radicaux pour les interlocutifs, dans les désinences pour les proformes. 10.4. La catégorie du nombre On a vu que le nombre constituait un paramètre apte à rendre compte la distinction morphologique entre délocutifs et interlocutifs (§ 7.3). Or presque tous les morphèmes spécifiquement pronominaux qui, en anatolien, sont clairement hérités de l’indo-européen ont en commun, d’amalgamer une information relative au nombre quel que soit leur statut (radicaux interlocutifs, désinences nom.-acc. sg. *-d et gén. pl. *-so¯m). Les morphèmes apparentés dans les autres langues indo-européennes ont également un statut spécifiquement pronominal et reflètent une information similaire. Le nombre s’avère donc être un paramètre organisateur de la morphologie des pronoms. Cette propriété est unique et ne se retouve dans la morphologie d’aucune autre classe de mots. Il s’ensuit qu’aux cas directs, les pronoms hittites sont invariablement marqués comme singulier ou comme pluriel, alors que ce n’est pas nécessairement le cas des noms du genre animé: (37) a. KUB 7.1 iii 27 (CTH 390, rédac. récente, Kronasser 1962, 158, 160) kıª#ma uttar nakki dém.-n.-acc.sg.#conj. parole-n.-acc.sg. important-n.-acc.sg. “toutefois, cette affaire (‘parole’) (est) importante” b. KBo 4.2 ii 22–23 (CTH 398, rédac. récente, Bawanypeck 2005, 30, 42) nu#wa kªe kallar conn.#quot. dém.-n.-acc.pl. mauvais- n.-acc.pl. uttar apiya paiddu parole n.-acc.pl. là aller-3sg.opt. “que ces paroles mauvaises s’en aillent là!” Par règle, le verbe n’indexe pas le sujet inanimé (van den Hout 2001, Patri 2007: 61sq.), si bien que, dans toute la construction (37b), la seule information relative au nombre du sujet est manifestée par le
Qu’est-ce qu’un pronom en anatolien?
101
pronom37. Dans les contextes où un constituant nominal du genre inanimé est sujet ou objet, ses propriétés sémantiques sont maximalement spécifiées par les pronoms dépendants, non par les noms-têtes eux-mêmes. Cette organisation de la référence syntaxique vaut pour les inanimés, non pour les noms animés, ce qui indique manifestement une motivation fondée sur l’échelle d’animation (où les pronoms occupent l’échelon le plus élevé)38. Or l’utilisation de marques spécifiquement pronominales est plus fréquente dans la flexion des démonstratifs inanimés que dans celle des animés correspondants, alors même que la non différenciation du nominatif et de l’accusatif reste typique des inanimés (noms et pronoms). L’hypothèse que l’on peut alors former est que, puisque les contextes dans lesquels les pronoms utilisent typiquement des désinences spécifiques sont les mêmes que ceux dans lesquels la morphologie des noms inanimés présente une information réduite, les pronoms indoeuropéens font usage d’une morphologie spécifiquement différente de celle des noms, pour stipuler une information – le nombre – que certaines classes de noms sont morphologiquement inaptes à manifester. Aux cas nucléaires, nominatif et / ou accusatif, normalement distincts dans les flexions de noms animés, mais confondus dans les flexions inanimées, les pronoms dépendants, qu’ils soient modifieurs ou anaphoriques, utilisent obligatoirement la même marque que leur nom-tête animé (38a), alors qu’avec les noms inanimés, au bas de l’échelle d’animation, la discordance des marques est absolument systématique (38b): (38) a. [animés] apa¯-s (…) haluga-s “ce message” KUB 7.57 i 2 b. [inanimés] apa¯-t eshar-Ø “ce sang” KUB 14.3 iv 52 apa¯-t ped-an “cet endroit” KUB 36.87 iii 9 Une situation similaire ou approchante est reflétée dans d’autres langues indo-européennes: quand la marque casuelle d’un pronom su-
37
38
A la différence de Luraghi 1997, 17, je ne considère pas comme établi que le nominatif-accusatif des “hétéroclitiques” du type de uddar- présente régulièrement au pluriel un allongement de la voyelle présidésinencielle (udda¯r-Ø); comp. Hoffner & Melchert 2008, 124sq. Dans le cas présent, cette conception est en discordance avec la graphie ut-tar de 37b. Pour des faits similaires ou typologiquement proches dans d’autres langues, voir Corbett 2000, 70sq.
102
Sylvain Patri
jet ou objet a la possibilité de coïncider avec celle du nom auquel il est co-référé aussi bien comme modifieur que comme anaphorique, c’est que le nom est animé; quand les marques sont dans l’impossibilité stricte de coïncider, c’est que le nom est inanimé (neutre), par exemple en latin: (39) a. [masculin] seru-us … i-s [féminin] fili-a … e-a b. [neutre] mar-e … i-d frig-us … i-d
“l’esclave …, il” “sa fille …, elle” “la mer …, elle” “le froid … il”
On peut résumer les données (38)–(39) en observant qu’aux cas nucléaires, si la marque casuelle d’un pronom dépendant est la même que celle de son nom-tête, ce nom est du genre animé; si les marques sont distinctes, le nom est du genre inanimé. Sous cette considération, la confrontation de (38) et de (39) livre une des meilleures bases démonstratives qui soient à l’appui de la restitution d’une système à deux genres seulement (animé et inanimé) en indo-européen commun.
11. Conclusion En hittite, les pronoms forment, avec les noms et les verbes, une classe de mots identifiables comme tels indépendamment du contexte syntaxique. Or, un énoncé ne peut transmettre un contenu propositionnel (une signification logiquement décomposable) que s’il contient au moins un mot appartenant à l’une de ces catégories. Pour qu’une construction syntaxique puisse être validée comme telle, elle doit donc être bâtie au moyen d’au moins un mot dont l’identification catégorielle est indépendante de la syntaxe. Ce point établi, quand on aborde le problème linguistique des pronoms, c’est le plus souvent sous l’angle de la référence et des interactions complexes entre syntaxe, sémantique et énonciation qui fondent ces relations, plus rarement en posant la question du rôle organisateur que tiennent les pronoms dans l’appareil formel d’une langue. Les deux aspects sont différents: dans le premier, on prend en considération la façon dont un pronom dénomme un référent, dans le second, on observe comment les pronoms agencent ces référents par rapport aux autres signes linguistiques et comment ces autres signes – particu-
Qu’est-ce qu’un pronom en anatolien?
103
lièrement les noms – interagissent avec les pronoms. Il va de soi qu’ainsi posé, le problème ne regarde que les proformes, mais ce serait justement commettre une erreur de perspective d’estimer qu’au seul motif que les pronoms personnels ont, par leur référence, un statut radicalement distinct de celui des autre pronoms, tous les pronoms dans leur ensemble doivent être abordés sous considération de leurs seules propriétés référentielles. Selon une conception sous-jacente à de nombreux travaux linguistiques, la quantité d’information grammaticale spécifiée par le pronom serait, par principe, inférieure à la quantité d’information grammaticale spécifiée par le nom39; ce point de vue est fondé dans beaucoup de langues, mais comme le montrent les données hittites, il ne présente pas de caractère absolu et peut être relativisé. Les pronoms peuvent, comme on l’a vu (37b), être dépositaires d’informations sémantiques que le nom auxquel il est co-référé est inapte à transmettre, qu’il s’agisse de sa morphologie ou de sa syntaxe, via l’indexation au verbe. Une telle situation n’est pas exceptionnelle40, mais il est beaucoup plus rare que, comme en hittite, elle soit (i) syntaxiquement régulière, (ii) déterminée par le genre ± animé du nom-tête et (iii) exprimée par l’emploi de marque spécifiques, distinctes de celles de noms. Quel que soit le genre du nom, le fait capital explicitement révélé par la morpho-syntaxe des pronoms est que l’existence d’un matériel morphologique différencié pour les noms et les pronoms peut être justifée syntaxiquement en relation avec le fait qu’aux cas normalement attribués aux sujet et objets prototypiques, les dépendants sont dépositaires d’une quantité d’information égale ou supérieure à celle de la tête, jamais inférieure41.
39 40
41
Voir, par exemple, Dixon 1994, 94–97. Comparer cet oiseau vole /st wazo vɔl/, ces oiseaux volent /sz wazo vɔl/, où il suffit de remplacer voler par dormir (3sg. dort, 3 pl. dorment) pour retrouver une organisation de la référence conforme au schéma dominant en français. En hittite, ce sont les dépendants qui, par défaut, manifestent l’ensemble des propriétés des têtes nominales, y compris si leur flexion n’est pas différenciée de celle des noms, comme c’est le cas avec les adjectifs; voir les données exposées dans Patri 2007, 52sq.
104
Sylvain Patri
Références I. Sources Archi, Alfonso. 1978. Note sulle feste ittite. I. Rivista degli studi orientali 52(1–2), 19–26. Bawanypeck, Daliah. 2005. Die Rituale der Auguren (Texte der Hethiter, 25). Heidelberg: Winter. Götze, Albrecht. 1928. Madduwattas (Mitteilungen der Vorderasiatisch-Ägyptischen Geselleschaft, 32/1. Hethitische Texte in Umschrift, 3). Leipzig: J. C. Hinrichs. Götze, Albrecht. 1933. Die Annalen des Mursilis (Mitteilungen der Vorderasiatisch-Ägyptischen Gesellschaft, 38. Hethitische Texte in Umschrift, 6). Leipzig: J.C. Hinrichs. Güterbock, Hans Gustav. 1938. Die historische Tradition und ihre literarische Gestaltung bei Babyloniern und Hethitern bis 1200. II, Hethiter. Zeitschrift für Assyriologie 44, 45–149. Güterbock, Hans Gustav. 1958. The Composition of the Hittite Prayer to the Sun. Journal of the American Oriental Society 78(4), 237–245. Haas, Volkert. 1970. Der Kult von Nerik. Ein Beitrag zur hethitischen Religionsgeschichte (Studia Pohl, 4). Roma: Pontificium Institutum Biblicum. Kronasser, Heinz. 1962. Das hethitische Rituale KBo IV 2. Die Sprache 8, 89–107. Laroche, Emmanuel. 1969. Textes mythologiques hittites en transcription. Paris: C. Klincksieck. Lebrun, René. 1980. Hymnes et prières hittites (Homo religiosus, 4). Louvain-laNeuve: Centre d’histoire des religions. Otten, Heinrich. 1953. Ein kanaanäischer Mythus aus Bogazköy. ˘ Mitteilungen des Instituts für Orientforschung 1, 125–150. Otten, Heinrich. 1960. Die Eidesleistung des Ashapala. Revue hittite et asianique 18 (fasc. 67), 121–127. Otten, Heinrich. 1973. Eine althethitische Erzählung um die Stadt Zalpa (Studien zu den Bogazköy-Texten, ˘ 17). Wiesbaden: O.Harrassowitz. Rizzi Mellini, Annamaria. 1979. Un istruzione etea di interesse storico, KBo XVI 24 + 25. Onofrio Carruba (éd.). Studia Mediterranea Piero Meriggi dicata (Studia mediterranea, 1). Pavia: Aurora, 509–553. Schuler, Einar von. 1957. Hethitische Dienstanweisungen für höhere Hof- und Staatsbeamte. Ein Beitrag zum antiken Recht Kleinasiens (Archiv für Orientforschung. Beiheft, 10). Graz: Im Selbstverlage der Herausgebers. Singer, Itamar. 1996. Muwatalli’s Prayer to the Assembly of Gods through the Storm-God of Lightning (CTH 381). Atlanta: Scholars Press. Westbrook, Raymond & Roger D. Woodard. 1990. The Edict of Tudhaliya IV. Journal of the American Oriental Society 110(4), 641–659. II. Études Bader, Françoise. 1990. Les pronoms dans les langues indo-européennes. Mémoires de la Société de Linguistique N.S. 1, 23–35.
Qu’est-ce qu’un pronom en anatolien?
105
Baerman, Matthew; Dunstan Brown & Greville G. Corbett. 2005. The SyntaxMorphology Interface. A Study of Syncretism (Cambridge Studies in Linguistics, 109). Cambridge: Cambridge University Press. Beekes, Robert S. P. 1988. The Origin of the Indo-European Pronominal Inflection. Mohammad Ali Jazayery & Werner Winter (éds.). Languages and Cultures. Studies in Honor of Edgar C. Polomé (Trends in Linguistics, 36). Berlin: Mouton de Gruyter, 73–87. Benveniste, Émile. 1949. Sur l’emploi des cas en hittite. Archiv Orientální 17(1), 44–45. Benveniste, Émile. 1953/1962. La flexion pronominale en hittite. Language 29, 225–262 = “avec plusieurs additions”, É. B., Hittite et indo-européen. Paris: Adrien-Maisonneuve, 1962: 66–77. Benveniste, Émile. 1960. “Être” et “avoir” dans leurs fonctions linguistiques. Bulletin de la Société de linguistique 54(1), 113–134. Bhat, D. N. S. 2005. 43. Third-Person Pronouns and Demonstratives. Martin Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie (éds.). The World Atlas of Language Structures. Oxford: Oxford University Press, 178–181. Brugmann, Karl. 1904. Die Demonstrativpronomina der indogermanischen Sprachen (Abhandlungen der Philologisch-Historischen Klasse der Königlich Sächsischen Gesellschaft der Wissenschaften, 22/6). Leipzig: Teubner. Carruba, Onofrio. 2000. Indo-European *sem / sm- in the Pronouns: ‘Singulative’ Plurals. Journal of Indo-European Studies 28, 341–357. Carruba, Onofrio. 2002. Plurali ‘singolativi’ nei pronomi personali indoeuropei. Atti del Sodalizio glottologico milanese 39–40, 195–206. Cohen, Gerald L. 1976. On the Origin of *-sm- in Indo-European Pronouns. Indogermanische Forschungen 81, 18–24. Corbett, Greville G. 2000. Number. Cambridge: Cambridge University Press. Corbett, Greville. 2005. Suppletion in Personal Pronouns: Theory versus Practice, and the Place of Reproducibility in Typology. Linguistic Typology 9(1), 1–23. Corblin, Francis. 1987. Indéfini, défini et démonstratif: constructions linguistiques de la référence (Langue & cultures, 17). Genève: Droz. Danièl’, Mixail Aleksandrovicˇ. 2005a. Tipologija associativnoj mnozˇestvennosti. Kandidatskaja dissertacija. Moskovskij gosudarstvennyj universitet, Filologicˇeskij fakultet. Danièl’, Mixail Aleksandrovicˇ. 2005b. Plurality in Independent Personal Pronouns. Bernard Comrie, Matthew S. Dryer, David Gil & Martin Haspelmath (éds.). World Atlas of Language Structures. Oxford: Oxford University Press, 474–477. Diessel, Holger. 1999. Demonstratives. Form, Function, and Grammaticalization (Typological Studies in Language, 42). Amsterdam: J. Benjamins. Dixon, Robert M. W. 1994. Ergativity (Cambridge Studies in Linguistics, 70). Cambridge: Cambridge University Press. Eichner, Heiner. 1992. Anatolian. Jadranka Gvozdanovic´ (éd.). Indo-European Numerals (Trends in Linguistics. Studies and Monographs, 57). Berlin / New York: Mouton de Gruyter. Foley, James. 1975. Some Rules Involving Nasals and their Implications. Charles A. Ferguson; Larry A. Hyman & John J. Ohala (éds.). Nasálfest: Papers from a
106
Sylvain Patri
Symposium on Nasals and Nasalization. Stanford: Stanford University, Department of Linguistics, 213–230. Francia, Rita. 1996. Il pronome possessivo enclitico in antico ittita: alcune riflessioni. Vicino Oriente 10, 209–259. Friedrich, Johannes; Annelies Kammenhuber & Inge Hoffmann (éds.). 1975 sq. Hethitisches Wörterbuch. Zweite, völlig neubearbeitete Auflage auf der Grundlage der edierten hethitischen Texte. Heidelberg: C. Winter Garrett, Andrew. 1990a. The Syntax of Anatolian Pronominal Clitics. Ph. D. Thesis, Harvard University. Garrett, Andrew. 1990b. The Origin of NP Split Ergativity. Language 66(2), 261–296. Garrett, Andrew. 1998. Remarks on the Old Hittite Split Genitive. Jay H. Jasanoff; H. Craig Melchert & Lisi Oliver (éds.). Mír curad. Studies in Honor of Calvert Watkins (Innsbrucker Beiträge zur Sprachwissenschaft, 92). Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck, 155–163. Goedegebuure, Petra. 2003a. Reference, Deixis and Focus in Hittite. The Demonstratives ka- “this”, apa- “that” and asi “yon”. Amesterdam: Academisch proefschrift ter verkrijging van de graad van doctor aan de Universiteit. Goedegebuure, Petra. 2003b. The Hittite 3rd Person/Distal Demonstrative asi (uni, eni, etc.). Die Sprache 43(1), 1–32. Goedegebuure, Petra. 2006. A New Proposal for the Reading of the Hittite Numeral ‘1’: sia-’. Theo Ph. J. van den Hout (éd.). The Life and Times of Hattusili III and Tuthaliya IV. Leiden: Nederlands Instituut voor het Nabije Oosten, 165–188. Goedegebuure, Petra. 2007. The Hieroglyphic-Luwian Demonstrative AblativeInstrumentals zin and apin. Studi Micenei ed Egeo-Anatolici 49, 319–334. Hajek, John. 1997. Universals of Sound Change in Nasalization (Publications of the Philological Society, 31). Oxford: Blackwell. Hawkins, John A. 1978. Definiteness and Indefiniteness. A Study in Reference and Grammaticality Prediction. London: Croom Helm. HED = Puhvel 1984sq. Held, Warren H. Jr.; William R. Schmalstieg & Janet E. Gertz. 1988. Beginning Hittite. Columbus: Slavica. Hock, Hans Henrich. 2003. Analogical Change. Brian D. Joseph & Richard D. Janda (éds.). The Handbook of Historical Linguistics. Oxford: Blackwell, 441–460. Hoffner, Harry A., Jr. 2006. The Hittite Degenitival Adjectives siela-, 2-ela und apella-. Theo Ph. J. van den Hout (éd.). The Life and Times of Hattusili III and Tuthaliya IV. Leiden: Nederlands Instituut voor het Nabije Oosten, 189–197. Hoffner, Harry A., Jr. & H. Craig Melchert. 2008. A Grammar of the Hittite Language. I, Reference Grammar. Winona Lake: Eisenbrauns. Hout, Theo Ph. J. van den. 2001. Neuter Plural Subjects and Nominal Predicates in Anatolian. Onofrio Carruba & Wolfgang Meid (éds.). Anatolisch und Indogermanisch (Innsbrucker Beiträge zur Sprachwissenschaft, 100). Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck, 167–192. HW2 = Friedrich & al. 1975sq.
Qu’est-ce qu’un pronom en anatolien?
107
Jasanoff, Jay H. 1972. The Hittite Ablative in -anz(a). Münchener Studien zur Sprachwissenschaft 31, 123–128. Kammenhuber, Annelies. 1962. Buchbesprechung: J. Friedrich, Hethitisches Wörterbuch. Orientalia 31, 371–380. Kammenhuber, Annelies. 1969 [1963, erschienen 1969]. Hethitisch, Palaisch, Luwisch und Hieroglyphenluwisch. Johannes Friedrich (éd.). Altkleinasiatische Sprachen. Leiden: E. J. Brill, 119–357. Katz, Joshua T. 1998. Topics in Indo-European Personal Pronouns. Ph. D. Thesis, Harvard University, Department of Linguistics. Kloekhorst, Alwin. 2008. Etymological Dictionary of the Hittite Inherited Lexicon (Leiden Indo-European Etymological Dictionary Series, 5). Leiden: Brill. Kurz, Josef & Zoe Hauptová (éds.). 1958–1997. Slovník jazyka starosloveˇnského. Praha: Nakl. Cˇeskoslovenské akademie veˇd. [= SJSS] Lakoff, Robin 1974. Remarks on This and That. Michael W. La Galy, Robert A. Fox & Anthony Bruck (éds). Papers from the 10th Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 345–356. Lane, George S. 1961. On the Formation of the Indo-European Demonstrative. Language 37(4), 469–475. Laroche, Emmanuel. 1960. Comparaison du louvite et du lycien (suite). Bulletin de la Société de linguistique 55(1), 155–185. Laroche, Emmanuel. 1970. Études de linguistique anatolienne. III. Revue hittite et asianique 28, 22–71. Luraghi, Silvia. 1997. Hittite (Languages of the World. Materials, 114). München: Lincom. Meillet, Antoine. 1936. Esquisse d’une grammaire comparée de l’arménien classique. Seconde édition, entièrement remaniée. Vienne: Impr. des PP. Mékhitharistes. Meillet, Antoine. 1937. Introduction à l’étude comparative des langues indo-européennes. Huitième édition corrigée [11903]. Paris: Hachette. Meillet, Antoine & Jules Vendryes. 1948. Traité de grammaire comparée des langues classiques. Deuxième édition revue et augmentée. Paris: Champion. Melchert, H. Craig. 1983. The Second Singular Personal Pronoun in Anatolian. Münchener Studien zur Sprachwissenschaft 42, 151–165. Melchert, H. Craig. 1984. Studies in Hittite Historical Phonology (Ergänzungshefte zur Zeitschrift für vergleichende Sprachforschung, 32). Göttingen: Vandenhoeck & Ruprecht. Melchert, H. Craig. 1994. Anatolian Historical Phonology (Leiden Studies in IndoEuropean, 3). Amsterdam: Rodopi. Melchert, H. Craig. 2003. Language. H. Craig Melchert (éd.). The Luwians (Handbuch der Orientalistik. Erste Abteilung, Nahe und der Mittlere Osten, 68). Leiden: Brill, 170–210. Melchert, H. Craig. 2008. Problems in Hittite Pronominal Inflection. Alexander Lubotsky; J. Schaeken & Jeroen Wiedenhof (éds.). Evidence and Counter-Evidence. Essays in Honour of Frederik Kortlandt. I, Balto-Slavic and Indo-European Linguistics (Studies in Slavic and General Linguistics, 32). Amsterdam: Rodopi, 367–375.
108
Sylvain Patri
Melchert, H. Craig. 2009. Deictic Pronouns in Anatolian. Kazuhiko Yoshida & Brent Vine (éds.). East and West. Papers in Indo-European Studies. Bremen: U. Hempen, 151–162. Melchert, H. Craig. s. p. The Problem of the Ergative Case in Hittite. Meriggi, Piero. 1980. Schizzo grammaticale dell’anatolico. Memorie dell’Accademia nazionale dei Lincei 377, Classe di scienze morali, storiche e filologiche, serie VIII, vol. 24(3), 244–411. Neu, Erich. 1979. Einige Überlegungen zu den hethitischen Kasusendungen. Erich Neu & Wolfgang Meid (éds.). Hethitisch und Indogermanisch (Innsbrucker Beiträge zur Sprachwissenschaft, 25). Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck, 177–196. Neu, Erich. 1997. Zu einigen Pronominalformen des Hethitischen. Dorothy Disterheft; Martin Huld & John A. C. Greppin (éds.). Studies in Honor of Jaan Puhvel (Journal of Indo-European Studies, Monograph Series, 20). Washington: Institute for the Study of Man, 133–169. Oettinger, Norbert. 1979. Die Stammbildung des hethitischen Verbums (Erlanger Beiträge zur Sprach- und Kunstwissenschaft, 64). Nürnberg: H. Carl = Nachdruck mit einer Kurzen Revision der Hethitischen Verbalklassen: Dresden: Verlag der TU Dresden, 2002 (Dresdner Beiträge zur Hethitologie, 7). Oettinger, Norbert. 1994. Etymologisch unerwarteter Nasal im Hethitischen. Jens Elmegård Rasmussen (éd.). In honorem Holger Pedersen. Kolloquium der Indogermanischen Gesellschaft vom 25. bis 28. März 1993 in Kopenhagen. Wiesbaden: L. Reichert, 307–330. Oshiro, Terumasa. 2000. Hieroglyphic Luwian tuwati and unzati. Y. L. Arbeitman (éd.). The Asia Minor Connexion. Studies on the Pre-Greek Languages in Memory of Charles Carter (Orbis Supplementa, 13). Leuven: Peeters, 189–193. Patri, Sylvain. 2007. L’alignement syntaxique dans les langues indo-européennes d’Anatolie (Studien zu den Bogazköy-Texten, ˘ 49). Wiesbaden: Harrassowitz. Patri, Sylvain. 2008. Le déterminant défini hittite asi : contraintes référentielles et syntaxiques. Indogermanische Forschungen 113, 149–175. Payne, Annick. 2004. Hieroglyphic Luwian (Elementa linguarum Orientis, 3). Wiesbaden: Harrassowitz. Pedersen, Holger. 1938. Hittitisch und die anderen indoeuropäischen Sprachen (Det Kongelige Danske Videnskabernes Selskab. Historisk-Filosofiske Meddelelser, 25/2). København: Levin & Munksgaard. Petersen, Walter. 1937. Hittite Demonstrative Pronouns. The American Journal of Philology 58(3), 306–319. Puhvel, Jaan. 1984sq. Hittite Etymological Dictionary. (Trends in Linguistics. Documentation, 1). Berlin: Mouton de Gruyter [= HED]. Rasmussen, Jens Elmegård. 1987. The Constituent Elements of the Indo-European Personal Pronouns. APILKU 6, 89–112 = J. E. R., Selected Papers on IndoEuropean Linguistics (Copenhagen Studies in Indo-European, 1). Copenhagen: Museum Tusculanum Press, 1999, 256–275. Rizza, Alfredo. 2007. I pronomi enclitici nei testi etei di traduzione dal hattico (Studia Mediterranea, 20). Pavia: Italian University Press. Schmidt, Gernot. 1978. Stammbildung und Flexion der indogermanischen Personalpronomina. Wiesbaden: Harrassowitz.
Qu’est-ce qu’un pronom en anatolien?
109
Sihler, Andrew L. 1993. The Anatolian and Indo-European First Person Plural. Bela Brogyanyi & Reiner Lipp (éds.). Comparative-Historical Linguistics. IndoEuropean and Finno-Ugric Papers in Honor of Oswald Szemerényi. III (Current Issues in Linguistic Theory, 97). Amsterdam: J. Benjamins, 171–185. Sihler, Andrew L. 1995. New Comparative Grammar of Greek and Latin. Oxford: Oxford University Press. SJSS: Kurz & Hauptová (éds) 1958–1997. Sobolevskij, Aleksej Ivanovicˇ. 1907. Lekcii po istorii russkogo jazyka. Izd. cˇetvertoe. Moskva: Universitetskaja tipografija. Sommer, Ferdinand & Adam Falkenstein. 1938. Die hethitisch-akkadische Bilingue des Hattusˇili I. (Labarna II.) (Abhandlungen der Bayerischen Akademie der Wissenschaften, Philosophisch-Historische Abteilung, N. F., 16). München: C. H. Beck. Soysal, Ögüz. 2005. On the Origin of the Royal Title tabarna / labarna. Anatolica 31, 189–209. Sturtevant, Edgar H. 1933. A Comparative Grammar of the Hittite Language. Philadelphia: Linguistic Society of America. Sturtevant, Edgar H. & E. Adelaide Hahn. 1951. A Comparative Grammar of the Hittite Language, I. Revised Edition. New Haven: Yale University Press. Szemerényi, Oswald. 1996. Introduction to Indo-European Linguistics. English Translation by D. M. Jones & I. Jones of 1990 German 4th edition. Oxford: Clarendon Press. Tischler, Johann. 1988. Labarna. Erich Neu & Christel Rüster (éds.). Documentum Asiae Minoris antiquae. Festschrift für Heinrich Otten zum 75. Geburtstag. Wiesbaden: O. Harrassowitz, 347–358. Vaillant, André. 1958. Grammaire comparée des langues slaves. II, Morphologie (Les langues du monde. Série grammaire, philologie, littérature, 11). Lyon / Paris: IAC. Vaillant, André. 1964. Manuel du vieux slave. I, Grammaire. Seconde édition revue et augmentée (Collection de manuels publiés par l’Institut d’études slaves, 6). Paris: Institut d’études slaves. Vecˇerka, Radoslav. 1993. Altkirchenslavische (altbulgarische) Syntax. II, Die Innere Satzstruktur (Monumenta linguæ slavicæ dialecti veteris, 34). Freiburg i. Br.: U.W. Weiher. Wackernagel, Jacob & Albert Debrunner. 1929–1930. Altindische Grammatik. III, Nominalflexion. Zahlwort. Pronomen. Göttingen: Vandenhoeck und Ruprecht. Watkins, Calvert. 1967. Remarks on the Genitive. To Honor Roman Jakobson (Janua linguarum. Series maior, 31). The Hague / Paris: Mouton, 2191–2198.
Université Lumière, Lyon-II & CNRS, UMR 5596 Faculté des Lettres 5, av. Pierre-Mendès-France F-69676 B r o n C e d e x – France
[email protected]
Sylvain Patri
110
Kenneth Shields, Jr.
Hittite sˇia-‘1’ and its implications for the etymology of Indo-European numerals Abstract This brief paper explores the implications of Goedegebuure’s (2006) recent identification of sˇia- as the Hittite numeral ‘1.’ Specifically, it is proposed here that sˇialends additional support to the notion that the lower numerals of Indo-European can be derived from deictic/demonstrative elements – in this case a contamination of an original deictic/demonstrative in *(e/o)s with another in *io (*yo). Moreover, the etymon of the Hittite form, in conjunction with the alternative Indo-European Proper root for ‘1,’ *sem-, a contamination of deictics in *(e/o)s and *(e/o)m, confirms the existence of an alternation in deictic extensions of *(e/o)s (*yo ~ *em) which parallels a similar alternation in deictic extensions involving the alternative Indo-European Proper root *oi- ‘1’ (*oi-no- ~ *oi-kwo-, etc.). *oi- itself is commonly traced to a deictic source (cf. Shields 1994).
The contribution of Hittite to the study of the etymology of IndoEuropean numerals is limited because Hittite scribes generally used logograms to represent these lexemes. However, in some cases syllabic renderings of numerals are attested directly, or insights into the pronunciation of Hittite numerals can be ascertained indirectly through their appearance in related word-forms. In an important recent article, Goedegebuure (2006) has proposed that the Hittite cardinal ‘1’ is sˇia-. By analyzing contexts in which sˇia- appears, she demonstrates that – formally – it is neither a demonstrative nor an accented anaphoric pronoun in function, thereby implying logically that it is a numeral, and that – semantically – its use is consistent with the meaning ‘one.’ Hoffner & Melchert (2008: 154) endorse this finding, characterizing it as “promising” in comparison to previous attempts at identification which “have proven uncertain.” Of course, Indo-European Proper attests two roots for ‘1’ – *oi-, which is variously extended by *-no- (e.g., Latin *oi-no- [u¯nus]), *-kwo- (e.g., Vedic *oi-kwo- [éka-]), and *-wo- (e.g., Old Persian *oi-wo- [aiva-]), and *sem- (e.g., Mycen. e-me, Toch. A sas (masc.), sam (fem.), Lat. semel, Go. simle ‘once, formerly’) – the former of which is traditionally associated with “singleness or isolation” and the latter of which, “togetherness or unity” (Hoffner & Melchert 2008: 154, cf. Waanders 1992: 370). “It is
Hittite sˇia-‘1’
111
possible that in Hittite also there were two words,” but the scribal practice alluded to earlier makes it impossible, at present, to verify this fact (Hoffner & Melchert 2008: 154).1 The purpose of this brief paper is to explore the potential implications of Goedegebuure’s hypothesis (2006) for the etymological origin of the numeral ‘1’ in early Indo-European itself. I shall argue that it provides significant confirmation for the view expressed in Shields (1994), where I maintain that the lower numerals can be traced to deictic/demonstrative elements. In my opinion, the system of IndoEuropean numerals developed gradually into the historical period, with the original system consisting of only a few digits originating as deictics/demonstratives and a so-called “limit of counting,” i.e., one, two, three, many. I developed this latter notion in a series of articles beginning in 1984 (see, e.g., Shields 1984, 1991, 1994, 2000); the same view was more recently – and apparently independently – formulated by Carol Justus beginning in 1988 (see, e.g., Justus 1988, 1999, 2004). The idea that, following common typological developments, the lower numerals of many languages have a deictic/demonstrative source is not new; indeed, both Conant (1896: 75) and Menninger (1969: 17) subscribe to this position. In Shields (1994: 179) I point out that Brugmann (1911: 311) reconstructs, as one of “die n-Demonstrativa,” a form in *oino-, which itself represents a contamination of the demonstratives in *oi- and the deictic/demonstrative element *no- and that extensions of *oi- by *kwo- and *wo- also “are homophonous with deictic particles which can be independently reconstructed for Indo-European” (cf. Hirt 1927: 11–13, Brugmann 1911: 349–350). I conclude that these deictics which were attached to *oi- may have originally expressed differing degrees of remoteness/proximity to the speaker, i.e., one-here, one-there, one-yonder, but that “as the proto-language evolved, the importance of specifying the deixis of the numeral ‘one’ lessened until various dialects merely generalized (and/or specialized) the original deictic variants” (1994: 182). I likewise propose there that *sem- derives from the demonstrative stem *se/o- (cf., e.g., Skt. sá-s, Gk. hó-s, Go. sa), which has been extended by the deictic *e/om, reconstructed for Indo-European, for example, by Hirt (1927: 13). 1
On the basis of paradigmatic variation exhibited by the Hittite logogram ‘1,’ Goedegebuure (2006: 185) speculates that Hittite had two roots for this numeral. However, she admits that other interpretations of the data are possible.
112
Kenneth Shields, Jr.
The deictic/demonstrative origin of the numeral roots *oi- and *sem- is, of course, controversial. Szemerényi (1996: 221–222), for instance, explicitly rejects deictic/demonstrative sources for any cardinal numeral, and Fortson (2004: 131) similarly describes the IndoEuropean word-forms for ‘1’ as independent lexical entities. On the other hand, Sihler (1995: 404–406) sees in *oi- “an ablaut grade of the pronoun root *i-,” although he posits no such deictic/demonstrative source for *sem-. This same position is endorsed by Buck (1949: 937) and Mallory & Adams (1997: 398–399). Among the few proponents of a deictic/demonstrative source for *sem- are Brugmann (1916/17: 160–161) and Hahn (1942). In support of deictic/demonstrative etyma underlying the IndoEuropean forms for ‘1,’ I have attempted to demonstrate in a series of articles that other lower numerals of Indo-European can easily be derived from such elements, which “frequently show a secondary affixation of … [various non-singular markers like *-(e/o)s, *-i, *u, *-e, *-(e/o)r, and *-t/d] as a means of hypercharacterizing their inherent non-singularity” (Shields 1985: 189). (See Shields 1985 and 2004: 572 for details.) For example, in Shields (1991: 270), I note that “if Schmidt (1978: 172) is correct in reconstructing *de-/*di- (cf. Hitt. da-, Lat. dis-) as variant forms of the Indo-European root for ‘2,’ cf. Shields 1984, then ‘2’ corresponds formally to the demonstrative stem in di(cf. ‘Iran. balt. di-: z. B. Akk. Sg. av. di-m preuss. di-n, Neutr. Av. di- t ˜ ‘es’ …’ [Brugmann 1904: 402]) and the Greek deictic particle -de,”2 while in Shields (2000), I posit the demonstrative/deictic *te/o- plus the non-singular affix -r as the basis of *ter- ‘3’ (cf. Hitt. ter-, Skt. ordi2
In Shields (2004b: 24), I argue that *de/o- was the primary form of the IndoEuropean numeral ‘2’: “Although *duwo- (e.g., Gk. dúo, Lat. duo, Ved. duvá, OCS dчva) is its most common reconstruction, a variant in *dw- (e.g., Ved. dvá-, Go. twai, Armen. er-ku) is also generally ascribed to the proto-language (cf. Szemerényi 1996: 222). However, as Sihler (1995: 407) argues, since ‘all other numerals have e-grade in the simplex forms, so the truly original form for ‘two’ must have been something like **dewo.’ I, too, believe that the latter form should be ascribed to the proto-language and that it should be segmented as *de-wo, with the truly original form for ‘2’ – *de – directly attested, for example, in the numeral *de-km 0 t ‘10’ (see Shields 1984 for details, cf. also Szemerényi 1960: 69 and Markey 1984: 284: ‘two hands (full)’). An o-grade variant of *de may be found in the Hittite numeral da- ‘2’ (e.g., da-ma-a(-i) ‘(an)other’), cf. Sihler (1995: 408).” The lexeme *di, as both numeral and deictic, probably represents a contamination of the deitic particles/demonstratives *de and *i.
Hittite sˇia-‘1’
113
nal *tr0 - [zero grade, cf. Benveniste 1962: 87]). I have also argued that the Indo-European Proper root for ‘4’ (*kwetwor-: Skt. catváras, Gk. téttares, Lat. quattor, Go. fidwo¯r) contains the root *kwe-, which “is probably related to the Indo-European interrogative-indefinite stem in *kwe-, an element that “bears close semantic and inflectional affinities to demonstratives (cf. Brugmann 1904: 379 …)” and that “probably represents an etymological demonstrative which has lost much of its deictic force (cf. Lane 1961: 469–470)” (Shields 1991: 269–270). To this root *kwe- has been added two hypercharacterizing non-singular markers, specifically *-u and *-or. Similarly, I have derived the Hittite numeral mey-u- ‘4’ from an Indo-European demonstrative stem in *me/o- (cf. Vedic áma- ‘this one’ < *e-mo-, i.e., a contamination of the demonstratives *e/o and *mo-, cf. Misra 1968: 81), to which the non-singular suffixes *i and *u were attached (Shields 2004a). Moreover, it is my opinion that even the Indo-European numeral for ‘6’ (cf. OIr. se, Lat. sex, Go. saihs, Gk. (w)éks) lends itself to analysis as an original demonstrative in *s(w)e, augmented with the non-singular elements *-k and *-(e/o)s. This original demonstrative also developed eventually into a reflexive pronoun (see Shields 1996 for details). However, my primary point in this brief paper is that a Hittite numeral in sˇia- ‘1’ provides very substantial support for my analyses of the etymology of *oi-no, etc. and *sem-. I am now convinced that the Indo-European deictic/demonstrative in *se/o- actually reflects an original deictic/demonstrative in *(e/o)s which has been contaminated with another particle in *e/o (cf. Hirt 1927: 10–11) or has been thematicized (Shields 1992: 29). Strong evidence for a deictic/demonstrative in *(e/o)s comes once again from Hittite in the form of “a pronoun/adjective asˇi,” which “is not an anaphoric pronoun,” as has been assumed (cf. Laroche 1979: 148 and Puhvel 1984), “but the 3rd person demonstrative ‘yon’ instead, accompanying the 1st person demonstrative ka¯- ‘this, near me’ and the 2nd person demonstrative apa¯- ‘that, near you’” (Goedegebuure (2002/03: 1). Goedegebuure carefully documents her assertion with textual data and by showing that the form meets all the typological criteria for demonstratives. Although Goedegebuure (2002/03) does not consider the etymology of asˇi, I argue in Shields (2007: 132–133) that “it derives from a late contamination of the deictic/demonstrative *(e/o)s (specifically, its o-grade) and a comparable element *i (cf. Hirt 1927: 11). Such a con-
114
Kenneth Shields, Jr.
tamination of deictics/demonstratives is a common development because of the need to reinforce the deixis of these forms.” Moreover, “Hoffner (2002/03: 81–83) proposes that the Hittite adverbial form a¯sˇma contains an original deictic element which ‘was not proximal (close to or on the ‘deictic center’), but distal’” and that it is to be etymologically connected to asˇi. In my opinion, “the long-initial of a¯sˇ-ma could have resulted from the contamination of the deictic particle *(e/o)s with the non-proximal deictic *e/o (cf. Hirt 1927: 10–11), i.e., *e/o + (e/o)s > a¯sˇ” (Shields 2007: 132–133).3 Like Hoffner (2002/03: 83), I see -ma as an affixed particle in -a/-ma, widely attested in Hittite. I should point out that Patri (2008) has recently questioned Goedegebuure’s conclusion that asˇi is indeed a demonstrative and presents evidence that it “est simplement un déterminant défini” (165), i.e., a definiteness marker. However, since Patri (2008: 169) acknowledges the well-known typological generalization that the etymological source of definiteness markers is deictic particles (cf. also Hazelkorn 1983: 110), the implications of this critique for the conclusions reached in Shields (2007) are minimal. In short, Indo-European possessed a deictic particle in *(e/o)s which underlies the demonstrative *se/o- of Indo-European Proper and the demonstrative (Goedegebuure) or definiteness marker (Patri) asˇi of Hittite. Now on the basis of this view that Indo-European possessed a deictic/demonstrative in *(e/o)s, it is reasonable to assert that Indo-European Proper *sem- shows *(e/o)s plus the deictic/demonstrative element *(e/o)m. Likewise, Hittite sˇia- can represent a contamination of *(e/o)s and the deictic/demonstrative form *io- (*yo-),4 which is to be connected etymologically to the so-called indefinite-interrogative pronoun *yo-.5 As Szemerényi (1996: 210) points out, “The origin of [indefinite-interrogative] *yo-s from a demonstrative, namely the 3
4 5
Hoffner (2002/03: 82) says of the difference in the length of the initial vowels of asˇi and a¯sˇma: “Since in Hittite texts from Bogazköy ˘ word-initial writings like a-a-, e-e-, i-i-, and u-u- are confined to Hurrian, Luwian, and Hattian words,” asˇi may have had a long vowel not indicated as such by the script, or, less likely, “an alternate account could suppose compensatory vowel lengthening accompanying contraction” for a¯sˇma. I feel that the contamination of deictic elements provides a less ad hoc explanation for the initial vocalic segment. On the plausibility of reading sˇia- as /sya-/, see Hoffner (2006: 191). In some dialects (Indic, Greek, Phrygian, and Slavic), this form also assumed a relative pronoun function (Szemerényi 1996: 210).
Hittite sˇia-‘1’
115
anaphoric *i-, is certain.” Thus, the Hittite numeral sˇia-, like *oi- and sem-, can easily be derived from deictic/demonstrative sources; and it appears to confirm the existence of an alternation in deictic extensions of *(e/o)s (*s-yo- ~ *s-em-) which is much like the alternation in deictic extensions of *oi-.6 Controversy surrounding the etymologies of Indo-European numerals will most certainly continue. However, new data do sometimes become available, and when they do, they most certainly can prove useful in assessing the explanatory power of current theoretical statements.
References Benveniste, É. 1962. Hittite et indo-européen: Études comparatives. Paris: Maisonneuve. Brugmann, Karl. 1904. Kurze vergleichende Grammatik der indogermanischen Sprachen. Strassburg: Trübner. –. 1911. Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. Vol. 2.2 Strassburg: Trübner. –. 1916/17. “Lat. aemulus, aequos, imita¯ri, ima¯go, griech. a¤ipsa, aipús, got. ibns.” Indogermanische Forschungen 37. 155–163. Buck, Carl. D. 1949. A Dictionary of Selected Synonyms in the Principal IndoEuropean Languages. Chicago: The University of Chicago Press. Conant, Levi. 1896. The Number Concept: Its Origin and Development. New York: Macmillan. Fortson, Benjamin. 2004. Indo-European Language and Culture: An Introduction. Malden, MA: Blackwell. Goedegebuure, Petra. 2002/03. “The Hittite 3rd Person/Distal Demonstrative asˇi (uni, eni, etc.).” Die Sprache 43. 1–32. –. 2006. “A New Proposal for the Reading of the Hittite Numeral ‘1’: sˇia-.” In: The Life and Times of Hattusˇili III and Tuthaliya IV: Proceedings of a Symposium Held in Honour of J. de Roos, 12–13 December 2003, Leiden, 165–188. Ed. T.P.J. van den Hout & C.H. van Zoest. Leiden: Nederlands Instituut voor het Nabije Oosten.
6
Goedegebuure (2006: 185) also briefly (in a footnote) speculates that sˇia- may be etymologically related to the demonstrative *se/o- and ultimately to IndoEuropean Proper *sem-; however, she does not explain in detail the developmental processes which underlie arguments for such a potential relationship. Moreover, she points out that Hittite itself may attest reflexes of *sem- in such forms as “1) sˇani- ‘the same’, 2) kisˇsˇan ‘in the same way’ (and not ‘in this one way’), and 3) sˇauitisˇt- ‘in the same year’” (2006: 185).
116
Kenneth Shields, Jr.
Hahn, E. Adelaide. 1942. “The Indefinite-Relative-Interrogative Stem sem-, sm-, smo.” Language 18. 83–116. Hazelkorn, Leena Tuulikki. 1983. “The Role of Deixis in the Development of Finno-Ugric Grammatical Morphemes.” Ohio State Working Papers in Linguistics 27. 89–139. Hirt, Hermann. 1927. Indogermanische Grammatik. Vol. 3. Heidelberg: Winter. Hoffner, Harry. 2002/03. “Hittite a-asˇ-ma.” Die Sprache 43. 80–87. –. 2006. “The Hittite Degenitival Adjectives sˇiela-, 2-ela, and apella-.” In: The Life and Times of Hattusˇili III and Tuthaliya IV: Proceedings of a Symposium Held in Honour of J. de Roos, 12–13 December 2003, Leiden, 189–197. Ed. T.P.J. van den Hout & C.H. van Zoest. Leiden: Nederlands Instituut voor het Nabije Oosten. – & H. Craig Melchert. 2008. A Grammar of the Hittite Language. Part 1. Winona Lake, IN: Eisenbrauns. Justus, Carol. 1988. “Indo-European Numerals and Number Systems.” In: A Linguistic Happening in Memory of Ben Schwartz: Studies in Anatolian, Italic, and Other Indo-European Languages, 542–541. Ed. Y. Arbeitman. Louvain-LaNeuve: Peeters. –. 1999. “Indo-European Numerals since Szemerényi.” In: The Emergence of the Modern Sciences: Studies on the Transition from Historical-Comparative to Structural Linguistics in Honour of E.F.K. Koerner, vol. 2, 131–152. Ed. S. Embleton, J. Joseph & H. Niederehe. Amsterdam: Benjamins. –. 2004. “On Language and the Rise of a Base for Counting.” General Linguistics 42. 17–43. Lane, George. 1961. “On the Formation of the Indo-European Demonstrative.” Language 37. 469–475. Laroche, E. 1979. “Anaphore et deixis en anatolien.” In: Hethitisch und Indogermanisch, 147–152. Ed. E. Neu & W. Meid. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. Mallory, J.P. & D.Q. Adams, eds. 1997. Encyclopedia of Indo-European Culture. London: Fitzroy Dearborn. Markey, T.L. 1984. “The Grammaticalization and Institutionalization of IndoEuropean Hand.” Journal of Indo-European Studies 12. 261–292. Menninger, Karl. 1969. Number Words and Number Symbols: A Cultural History of Numbers. Trans. P. Broneer. Cambridge, MA: The M.I.T. Press. Misra, Satya. 1968. Comparative Grammar of Sanskrit, Greek, and Hittite. Calcutta: World Press. Patri, Sylvain. 2008. “Le determinant défini Hittite asi: contraintes référentielles et syntaxiques.” Indogermanische Forschungen 113. 149–175. Puhvel, Jaan. 1984. “Review of J. Friedrich & A. Kammenhuber, Hethitisches Wörterbuch, Band I: A, Lieferung 6/7.” Journal of the American Oriental Society 103. 670–672. Schmidt, Gernot. 1978. Stammbildung und Flexion der indogermanischen Personalpronomina. Wiesbaden: Harrassowitz. Shields, Kenneth. 1984. “IE *dekm 0 (t) ‘10’: A New Etymology.” Balkansko Ezikoznanie 27. 4. 75–80.
Hittite sˇia-‘1’
117
–. 1985. “Speculations about the Indo-European Cardinals, 5–10.” Diachronica 2. 189–200. –. 1991. “The Indo-European Numeral ‘4’: A New Etymology.” In: Studia Etymologica Indoeuropaea: Memoriae A.J. van Windekens (1915–1989) Dicata, 265–272. Ed. L. Isebaert. Leuven: Peeters. –. 1992. A History of Indo-European Verb Morphology. Amsterdam: Benjamins. –. 1994. “Comments about IE *oi- ‘1.’” Journal of Indo-European Studies 22. 177–186. –. 1996. “Speculations about the Etymology of the Indo-European Cardinal ‘6.’” Folia Linguistica Historica 17. 3–12. –. 2000. “Some Comments about the Hittite Numeral ‘3.’” In: The Asia Minor Connexion: Studies on the Pre-Greek Languages in Memory of Charles Carter, 213–219. Ed. Y. Arbeitman. Leuven: Peeters. –. 2004a. “Comments on the Etymology of the Hittite Numeral ‘4.’” In: Sˇarnikzel: Hethitologische Studien zum Gedenken am Emil Orgetorix Forrer, 571–576. Ed. D. Groddek & S. Rössle. Dresden: Technische Universität Dresden. –. 2004b. “The Emergence of the Dual Category in Indo-European: A ‘New Image’ and Typological Perspective.” Indogermanische Forschungen 109. 21–30. –. 2007. “Hittite asˇi, a¯sˇma and the Deictic Origin of Indo-European Sigmatic Verbal Formations.” Studia Etymologica Cracoviensia 12. 133–138. Sihler, Andrew. 1995. New Comparative Grammar of Greek and Latin. New York: Oxford University Press. Szemerényi, Oswald. 1960. Studies in the Indo-European System of Numerals. Heidelberg: Winter. –. 1996. An Introduction to Indo-European Linguistics. 4th ed. Oxford: Clarendon Press. Waanders, Frederik M. J. 1992. “Greek.” In: Indo-European Numerals, 369–388. Ed. J. Gvozdanovic´. Berlin: Mouton de Gruyter.
3155 Grande Oak Place L a n c a s t e r , Pennsylvania 17601 U.S.A.
[email protected]
K e n n e t h S h i e l d s, J r .
118
Diether Schürr
Lykische Genitive Abstract The six different forms of the genitive singular in Lycian (A) are probably all variants of the form -he, which itself is going back to ie. *-osyo like Carian -s´ and Hieroglyphic Luwian -si. But arñnaha may alternatively be a genitive with an enclitic pronoun -a, like the ‘accusativus genitivi‘ in -hñ. A special trait of toponyms in -i is the ending -ih(e) instead of -eh(e). Titles after personal names in the genitive show the same ending -h, yet this is not a true genitive, but the suffix -a/ehi with apocope.
I „Lycian (A) displays the greatest number of ways of expressing possession among the Anatolian languages“ (Melchert, Genitive: 4): Neben dem Genitiversatz durch Adjektive auf -a/ehe/i- gibt es im Singular1 Genitive auf -h, was die Normalform ist, und auf -he2, selten endungslose Formen und Genitive auf -h◊ und -◊ (die beiden letzten Formen berücksichtigt Melchert in seinem Aufsatz nicht). Beispiele: udali zuhrijah 3tideimi{h} TL 1 im westlykischen Tel˜ messos; uhak˜ee: murazahe: tideimi TL 2 ebenda; q3ñtb˜e ddfla[p]ssmma 4tideimi TL 5 ebenda3; ijamara: terssi leh◊: tideimi TL 149 im ostlyki-
1
2
3
Der Genitiv Plural mit den Ausgängen -ãi, -˜ei ( ? ) und -˜e wird hier nicht behandelt. Lykische Inschriften werden wie bei Melchert 2004 (= DLL) und Neumann 2007 (= GL) angeführt, deren Lesungen aber in den Originalpublikationen überprüft werden müssen, N 334 nach Tekoglu ˘ 2002–2003. Mit KPN wird Zgusta 1964 angeführt. Für den Text seines Buches danke ich I. Yakubovich (Chikago). Bemerkenswert ist, daß von zehn Belegen der Datierungsformel „unter der Herrschaft des NN“ neun den Genitiv auf -he haben: TL 43, 61, 64, 67, 77, 103, 132, N 310, N 314a, dazu vielleicht auf dem zu N 324 oder 325 gehörenden Fragment Bousquet 1992: 186 a in ]he: ñta[ Z.5, nur einer anscheinend auf -h (N 315, schlecht erhalten). Statt Kalinkas Lesung [qñ]tbeddi ˜ s<s>mm[a(h) ˜ (Autopsie im Sommer 2007). Nasalierung vor d wäre irregulär; zu vergleichen sind qñtbeh TL 51, qñtb˜e TL 44d, 47 und ddapssmma ˜ TL 11.
Lykische Genitive
119
2tideimi TL 69 im zentrallykischen Rhodiapolis; ipresida◊: armpa◊ ˜ 4 schen Kyaneai . Außerdem gibt es auf Münzen noch arñnaha M 240a neben arñnahe M 240b (usw.) für Xanthos in Westlykien, za abaha M 134a und b neben za abahe M 134c und zagabah M 134d und M 109a, zagah◊ M 136 (offenbar aus *zágabh) für einen Ort in Zentrallykien (eher nicht der Avs¸ar Tepesi, wie Kolb – Tietz 2001 annehmen). Diese Formen werden auch Genitivvarianten und nicht Adjektivformen sein (contra DLL: 5 und 87 und GL: 22, während GL: 428 und 429 „Variante“ angenommen ist). Ein Sonderfall ist ñturiga ã: ãi tideimi in der frühen Inschrift TL 77 (Çindam bei Isinda): Da wird die Abtrennung in ñturiga und ããi zu berichtigen sein (vgl. Schürr 2006a: 1579). Letzteres sieht wie ein Genitiv Plural aus. Bei dem merkwürdigen a[. .]uflimfl fl eu 2tideimi in N 315 (Seyret) wird man auf eine verläßlichere Lesung warten müssen. Nach dem Photo des Abklatsches (Abb.15) scheint -mlu[h? möglich.
II Melchert geht von zwei verschiedenen indogermanischen Genitivendungen aus: lykisch -he soll auf *-(o)so zurückgehen, -h eine Verkürzung davon sein, was freilich „a genuine problem“ sei, „since there is no regular apocope of unaccented short *-o in Lycian“ (Genitive: 6, Anm.9). Die endungslosen Formen sollen dagegen auf *-os zurückgehen. Neumann 1970: 61 führt folgende Belege an: TL 5 in Telmessos (s. o.); TL 61 in Sebeda5 sbikezij˜ei: mre isa: tideimi neben Genitiv auf -he in der Datierungsformel; TL 119 in Limyra hura: iuba 3[ti]deri: ñteriwa..: tid<e>imi; TL 127 in Limyra sfl fltafl maha (…): epñ u a tideimi; TL 145 in Limyra hla: ñterubila: 2[6; TL 117 in Limyra siderija: pflerfl m[.]ne: tideimi = C « P « D« nach dem Faksimile. Kalinka hat p[ar]m[enah], ˜ und nach Neumann 1985: 247 ist pa[r]mnah fl
4
5
6
Da auch der erste Name im Genitiv steht, sollte man eigentlich *tideimehi erwarten. Port Sevedo, dann Bayındır Limanı, heute Limanagzı, ˘ ursprünglich der Hafen von Isinda. Vgl. aber E 2« KPN § 314, so daß der Name theoretisch trotz des Doppelpunkts unvollständig sein könnte
120
Diether Schürr
zu lesen. Kurioserweise hat Tischler in GL: 257 die überholte Lesung pa[r]mne beigefügt. Dieser Beleg ist also zu streichen. Melchert führt auch noch fünf weitere Inschriften an: TL 3 in Telmessos te2winez˜ei: <s>ppñtazah: asawãzala 3 tideimi (so nach DLL: 92). Aber Fellows hat in seiner Abschrift der nicht erhaltenen Inschrift am Beginn der dritten Zeile ein z, Loew Freiraum. Es ist also sicher in asawãzala
zu berichtigen. TL 16 in Pinara wazala: eppleme ti[ (DLL: 93). Aber da mag in dem Freiraum h zu ergänzen sein; [eh] schreibt Kalinka. TL 87 in Myra ahba: e[h]bi: wazzije: kbatra (DLL: 107) „seine Enkelin, des Wazzi Tochter“. TL 135 in Limyra [.]uwata: trbb˜enimeh: tideri: seb˜ela: ehet˜eme 2t[ (DLL: 93). Kalinka hat am Zeilenende [eh analog trbb˜enimeh. TL 104a in Limyra soll lusñ[tr]e auch Genitiv sein (DLL: 97 nach Neumann, der GL: 188 ein Fragezeichen setzt), aber das kann ohne weiteres ein Nominativ sein. Stattdessen dürfte TL 100 in Limyra in der singulären Formel ebe
upa me-tibeija ein Genitiv vorliegen (GL: 359 erwogen; anders Melchert DLL: 105 und Genitive: 4, Anm.6, aber ein mit tibe ‚oder‘ gleichlautender Name ist wenig wahrscheinlich, Genitiversatz durch ein Suffix -ija- nicht belegbar). Es fehlt also -h fünf- oder sechsmal nach -a (drei- oder viermal am Zeilenende), nur einmal sicher nach -e (wazzije). In Westlykien und Zentrallykien gibt es nur vereinzelte Belege, allein in Limyra mehrere. Von den Inschriften haben TL 5 und 87 ausgesprochen späte Zeichenformen. Daß diese endungslosen Genitive auf *-os zurückgehen, wäre zwar theoretisch möglich, ist aber m. E. unwahrscheinlich: Warum sollte denn eine ererbte unmarkierte Kasusform bewahrt worden sein, obwohl eine markierte zur Verfügung stand? Eine innerlykische Erklärung der hie und da auftretenden Formen ist a priori wahrscheinlicher, zumal das lykische h auch in anderen Fällen zum Verstummen neigt, siehe miñta in Kadyanda (TL 31) für miñtaha, eptte in Limyra (TL 121) für epttehe und auch die griechische Schreibung M « für mahanepi[ in Korydalla (N 302). Und die Normalform -h ist auch deswegen merkwürdig, weil lykische Wörter in der Regel auf Vokal enden. Selbst ein Nasal am Wortende wird nach -a- oder -e- durch die Zeichen ã und e˜ ausgedrückt, die schließlich durch -u und (manchmal) -i abgelöst werden, nach -i- und wohl auch -u- von vornherein nicht. Daher liegt nahe, daß in der Schreibung -Vh das [h] nur sehr schwach artikuliert wurde und diese Schreibung zu den endungslosen Formen überleitet: -Vhe > -Vh > -V. So hatte das
Lykische Genitive
121
auch Neumann gesehen: „Dieser vierte Typ des Genitivs, für den wir hier sechs Beispiele gesammelt haben, gehört sicher genetisch zu Typ a), von dem er sich nur durch den sekundären Abfall des (schwach artikulierten) -h unterscheidet“ (1970: 62). Bei arñnaha und za abaha könnte daher eine hyperkorrekte Schreibung für [-a] vorliegen. Bei -h◊ dagegen dient das seltene Sonderzeichen7 offenbar der Verstärkung des schwachen -h (ähnlich wie im Schriftdeutschen ch statt h), signalisiert also eine hyperkorrekte Aussprache. Diese Genitivvariante bildet demnach das Gegenstück zu den endungslosen Formen und stützt die Annahme, daß diese innerlykischen Ursprungs sind. Bei Zagaba dürften ein westlykisches, von arñnaha inspiriertes za abaha einerseits und zagah◊ andererseits einen Herrschaftswechsel signalisieren: Vergleiche zu letzterem z˜emuh◊ M 142ab8 und M 145b von z˜emure M 125 = Limyra (DLL: 89) in Ostlykien. Das in TL 69 in Kyaneai belegte -◊ wird eine Vereinfachung sein, während der frühe Beleg tumine◊ auf der MiBrapataMünze Zahle 1988: Fig.3 (420–400 datiert)9 m. E. anders zu verstehen ist: Es entspricht dem Genitiv [t]uminehih des Ortsnamens auf der etwas älteren Cheriga-Münze Zahle 1988: Fig.2, so daß es auf *tuminehh zurückgehen wird. Und ich denke, daß das Sonderzeichen auch zur Bezeichnung des aus *-hh resultierenden Lautes erfunden wurde, ähnlich wie für Dental + h.
III Alle lykischen Genitivformen gehen also m. E. auf -he zurück (nicht -h, wie in Schürr 2001a: 117 angenommen): Die Vielfalt ist ein inner7
8
9
In Westlykien nur auf der in die 2. Hälfte des 5. Jhs. datierten Scherbe N 313c mit zabr ah◊ in Xanthos. Diese Genitivform scheint also früher belegt als die endungslose. In Zentrallykien ist -h◊ auch noch in TL 54 (Phellos) belegt. Von diesen Münzen, die von dem Dynasten Trbbenimi stammen, ist M 142a wie Münzen mit za abaha, za abahe, zagabah im Podalia-Hort gefunden worden. Trbbenimi prägte außer in Limyra und Rhodiapolis (wedr˜ei) auch in Zagaba: M 135b hat zag t wie M 144b, wo sicher Trbbenimi gemeint ist, z˜em t (beide aus dem Podalia-Hort). Daher wird auch M 136 mit zagah◊ von ihm stammen. Der niederkniende Stier auf dieser Münze kehrt auf einem um 400 datierten Stater wieder, dessen Legende Neumann GL: 229 linksläufig mutuse liest. Das ist in tum und ein Monogramm zu korrigieren.
122
Diether Schürr
lykisches Phänomen; im nächstverwandten Karischen gibt es nur -s´. Und daß -he auf *-(o)so zurückgeht, ist zumindest nicht zwingend: Es kommt auch *-osyo in Frage, zumal das karische -s´ mit einiger Sicherheit nur darauf zurückführbar ist: siehe Schürr 2001a: 116 f., von Melchert nun auch für möglich (Genitive: 8) bzw. plausibel (Further Thoughts: 2, Anm.4) gehalten. Für den gleichen Ursprung dieser Endungen spricht, daß das Karische wie das Lykische über einen scheinbar vom Genitiv (und sicher nicht vom Adjectivum genitivale) gebildete Akkusativ Singular verfügt: karisch -s´ñ wie lykisch -hñ10, z. B. in N 320 (Trilingue vom Letoon): eseimiju: qñturahahñ: tideimi (a, 10 f.) = C K
« D (b, 8 f.) = SYMYN BR KDWRS (c, 9 f.). Ein starkes Indiz für die Zurückführung auf *-osyo liefert die lykische Dichtersprache (Lykisch B): Da geht der Genitiv auf -se aus wie bei kupr[l]lese TL 44d, 11 und m. E. auch auf -s bei arppa us: e˜ ti: tmpew˜ ˜ eti TL 44c, 57 f. (contra Melchert DLL: 92; wohl wegen des folgenden e˜ apokopiert), entsprechend arppa uh TL 44a, 1 f. und 30. Dagegen hat der ‚Accusativus genitivi‘ -zñ, z. B. in TL 44c, 48: timlu mawate: waxsadi: wizttasppazñ:) „den timlu … te er mit waxsa, den des Hystaspes“. Da in der Dichtersprache s vor ñ erhalten bleibt, dürfte diese Schreibung den palatalen Charakter der Genitivendung reflektieren11, während er sonst verlorenging (Schürr 2001a: 119). Damit gibt es keinen sicheren Beleg für einen Genitiv auf *-(o)so in den anatolischen Sprachen.
IV Allerdings hängt das von der Beurteilung der im Luwischen belegten Genitivformen ab. In den späteren hieroglyphenluwischen Inschriften sind zwei verschiedene Genitivendungen belegt, -a-sa und -a-si. Die 10
11
In DLL: xi ordnet Melchert diese Formen „h-stems“ zu und erklärt den klaren Genitiv hlah in TL 56 (= F!) und 129 zum Nominativ. Analog bucht Neumann GL: 295 sogar den Gen. puwejehe in TL 44a, 24 unter puwejehi-, weil in TL 44a, 39 und 40 auch puwejehñ belegt ist. Außer bei Personennamen ist der ‚Accusativus genitivi‘ auch bei Ortsnamen belegt, wie der Genitiv selbst: tlahñ in TL 44a, 47 zu Tlos. Das -ñ kann ursprünglich ein enklitisches Pronomen gewesen sein. So entspricht in der Dichtersprache die Endung -z des Nom. und Akk. Pl. c. der hieroglyphen-luwischen -nzi (aber karisch -sˇ).
Lykische Genitive
123
zweite Endung dürfte auch auf *-osyo zurückgehen (siehe schon Szemerényi 1990, 195), während die erste für *-as oder *-asa oder auch *-assa stehen kann. Denn letzteres ist möglicherweise im Keilschriftluwischen belegt (Yakubovich 2009, 38ff.), siehe vor allem in KUB 35.54: imrassa *Tarhuntassanza (ii 14 und iii 7), „das des Wettergotts des Feldes“, imrassa *Tarhunti (ii 35), „dem Wettergott des Feldes“, allerdings auch mit einer Adjektivendung imrassan *Tarhunti (ii 37). Im Hethitischen entspricht dem gimras *Tarhunni (KBo 22.137 iii 6’), so daß die Annahme plausibel ist, daß auch imrassa ein Genitiv ist.
V Im ‚Westgedicht‘ auf den Dynasten Cheriga ist in TL 44d, 8 erigasa belegt, eingeschoben zwischen je zwei weitere Wortformen auf -a: qñnãtba: uzrñta und tu[.]adrala: palaraima (offenbar Nom./Acc. Pl. n.). Das sieht so aus, als wäre hier ausnahmsweise ein Adjectivum genitivale von einem Personennamen gebildet. Man könnte aber auch annehmen, daß sich hier ein * erigase lediglich lautlich seiner Umgebung angepaßt hätte, also keine echte Kongruenz vorliegt. Konsequenter wäre aber, diese Form als eine dem ‚Accusativus genitivi‘ analoge Bildung zu verstehen, also erigas+a anzunehmen. Dann könnten auch die oben angeführten Formen arñnaha und za abaha so verstanden werden, mit Bezug auf die Währungseinheit ada.
VI Der lykische Genitiv ist auf Personen- und Ortsnamen beschränkt. Bei letzteren gibt es die Besonderheit, daß Ortsnamen auf -i diesen Vokal beibehalten: araihe, telebehihe, [t]uminehih, adawãtihe und
ãkbihe bzw. ãkbih (GL: 113), mit Ausname von adawãtihe in TL 35, 19 nur auf Münzen. Demnach wird auch bihñ in TL 44b, 5 Accusativus genitivi eines Ortsnamens sein, und da ãkbi in b, 7 folgt, werden zwei Zeichen übersprungen worden sein.12 12
Die Lesung ijãnihe: s<e>-trmmilihe ˜ in TL 29, 17 (Schürr 2001b: 136 und danach DLL: 28 und 70) ist nach Tekoglu ˘ 2006 durch ijãnazi: se-trmmili ˜ hrppi zu ersetzen.
124
Diether Schürr
Im Griechischen entsprechen diese Ortsnamen vermutlich Araxa und sicher Telmessos, anscheinend Tymnisos13, sicher Kadyanda und Kandyba. Nun sollte man bei telebehi, tuminehi und auch arai (vgl. tei ‚des Vaters, väterlich‘ gegenüber dichtersprachlichem tedesi) annehmen, daß sie mit dem Suffix -a/ehe/i- gebildet sind, wie beispielsweise der Personenname qñturahi = K
«, von dem der Acc. gen. qñturahahñ lautet. Die Genitive dieser ONN scheinen also regelwidrig zu sein. Außerdem kommt der Ausgang -ihe auch bei terihe (TL 58, N 309; N 324 dreimal) und kumalihe[ (TL 26, 19) vor, die beide bis auf N 324, 18 (wo ]j˜e: vorausgeht) nur nach tike belegt sind, wie [. .]edehe in TL 94. Klar sind alle drei Wortformen nicht, weder semantisch noch grammatisch. Götter haben im Lykischen eigentlich keine Namen14, denn bei ihren Bezeichnungen gibt es nur den Genitiversatz wie bei Appellativen, siehe beispielsweise in N 322: [ ? ]˜emudijah: tideimi: qelehi: kumaza „des [ ? ]emudija Sohn, des Qele- Priester“15. Im Karischen tritt der Genitiv dagegen auch an Appellative, z. B. in ùnemords´- múdons´ (Lesung nach Schürr 2002: 168 f.): „Des Ùnemord (Sohn), des múdon“. Genitiversatz scheint es im Karischen überhaupt nicht zu geben; das lykisch -a/ehe/i- entsprechende Suffix -s(i) ist nur in Personennamen belegt. In Bezug auf die Verwendung dieses Suffixes ist das Karische also konservativer als alle anderen südanatolischen Sprachen. Für die Fälle, wo auch im Lykischen ein Appellativ, das auf einen Personennamen folgt, wie dieser im Genitiv zu stehen scheint, ist nun die Felsgrabinschrift N 334 in Tlos aufschlußreich: Da folgen auf das Patronym aj˜eta[..]d[.]h16 der Titel armanazah17 und tideimi, schließt 13 14
15
16
17
Siehe das in Limyra belegte Ethnikon T (Wörrle 1995: 399 f.). Das refrainartig in TL 44a, 42–55 verwendete und in b, 56 nochmals belegte hãtahe wird nicht der Genitiv des luwischen Theonyms Sandas sein (DLL: 22), sondern eine rühmende, adverbiale Phrase: Dativ Plural von hãtahe/i- zu der Personenbezeichnung hãta- in TL 84, etwa ‚Inhaber‘ (Schürr 2006b: 119 f.). Also wohl so etwas wie ‚herrlich‘, worauf a, 50 „wie Herakles“ folgt. Das von Neumann, der nur ein Photo zur Verfügung hatte, am Beginn vermutete p ist sehr fraglich. Da der Sohn den Namen ipresida < #I 2 0« hat, könnte es sich um einen analog gebildeten griechischen Namen handeln. Aber welchen? Dazu Schürr 2006a: 1585: wohl lydisch arma- entsprechend, vgl. auch karisch armon ‚Dolmetscher‘ und lykisch (B) ermede in TL 44c, 62, wohl etwa ‚verkün-
Lykische Genitive
125
sich aber auch an ikuweh tedi in Z.6 se-pr[ñn]ezijeh an: „des Inguwe Vater und Hausgenosse“. Da ist der Endvokal von prñnezijehi apokopiert, wie das ja auch bei anderen Wörtern gelegentlich geschieht, und so wird das auch bei armanazah der Fall sein: ein Scheingenetiv. Das dürfte auch für <s>ppñtazah: asawãzala3 tideimi in TL 3 (Telmes2tideimi: sos), kudalah: ñtlah tideimi in TL 43 (Xanthos) und armpa◊ ˜ 18 tubure◊ in TL 69 (Kyaneai) gelten. Das Sonderzeichen ◊ tritt also in dieser Inschrift an die Stelle von -h und -hi. Tubure◊ kann nicht „von (aus der Stadt) Tuburi“ (GL: 385) bedeuten, weil eine solche Bildung singulär wäre. Und eine solche Stadt ist auch nicht belegbar, nur (1.) eine Personenbezeichnung, die in den Cheriga-Gedichten im Nom. Pl. tubuiz (TL 44d, 31 f.) und im Akk. Pl. tuburiz (c, 53 und 57) erscheint, (2.) in dem mythischen Personennamen T2 « bei Steph. Byz. s. v. 6Y wiederkehrt und in dem Personennamen E 2 « (KPN § 355–30 in Tlos und –31 leicht verschrieben in Xanthos) mit dem Namen des Mondgotts kombiniert ist, (3.) eine Stadt, deren Name davon abgeleitet ist: tuburehi TL 44b, 15, in dem schon Arkwright 1915: 103 T2 « südlich von Kyaneai erkannt hatte, (4.) der Personenname tebursseli in Limyra (TL 103 und 104), der auf das von diesem Stadtnamen in der lykischen Dichtersprache gebildete Ethnikon zurückgehen wird (Yakubovich apud Valério 2007: 5), vergleiche zum ersten Vokal urubli(je)- TL 44b, 34, 42 f., 55 und c, 9, aber später erublija TL 26, 1 und den Personennamen C
0« (KPN § 1396) zum Theonym C
« in Istlada (siehe Schürr 2008: 16). Literatur Arkwright, William, 1915, Notes on the Lycian alphabet. JHS. 35, 100–106. Bousquet, Jean,1992, Les inscriptions gréco-lyciennes. In: Fouilles de Xanthos 9.1. Paris, 147–203.
18
dete‘. Dazu könnte auch der Personenname E
/«/ (KPN § 355–18, mit zahlreichen Weiterbildungen bzw. Komposita) gehören: *armãta-, gebildet wie hãta-. In DLL sind asawãzala, xñtlah und tubure◊ unter den „Personal Names“ gebucht. In TL 28 auf einer Statuenbasis in Tlos ist sicher N1-hñ] 4ladu uwitahñ
ahb[u N2-hñ] 5apuwazahi p[r]ñnezijeh[i zu ergänzen: im Akkusativ „des N1] Gattin, des Uwinda Enkel[in, des N2], des apuwaza, Hausgenoss[in“. Da ist also apuwaza kein Name (so DLL: 91), sondern Titel und erscheint daher nicht im ‚Accusativus genitivi‘.
126
Diether Schürr
Kolb, Frank – Tietz, Werner, 2001, Zagaba: Münzprägung und politische Geographie in Zentrallykien. In: Chiron 31, 347–416. Melchert, H. Craig, 2004, A Dictionary of the Lycian Language. Ann Arbor / New York. Melchert, H. Craig, Genitive Case and Possessive Adjective in Anatolian. Online im Internet: URL http://www.linguistics.ucla.edu/people/Melchert unter anatoliangenitive (Stand: 6. 6. 2009). Melchert, H. Craig, Further Thoughts on Carian Nominal Inflection. Online im Internet: URL http://www.linguistics.ucla.edu/people/Melchert unter cariannoun (Stand: 6. 6. 2009). Neumann, Günter, 1970, Beiträge zum Lykischen IV. In: Die Sprache 16, 54–62. Neumann, Günter, 1985, Beiträge zum Lykischen VII. In: Die Sprache 31, 243–248. Neumann, Günter, 2007, Glossar des Lykischen. Überarbeitet und zum Druck gebracht von Johann Tischler (Dresdner Beitr. z. Hethitologie 21). Wiesbaden. Schürr, Diether, 2001a, Karische und lykische Sibilanten. In: IF. 106, 94–121. Schürr, Diether, 2001b, Bemerkungen zu Lesung und Verständnis einiger lykischer Inschriften. In: Kadmos 40, 127–154. Schürr, Diether, 2002, Karische Parallelen zu zwei Arzawa-Namen. In: Kadmos 41, 163–177. Schürr, Diether, 2006a, Elf lydische Etymologien. In: Festschr. Roberto Gusmani. Alessandria, 1569–1587. Schürr, Diether, 2006b, Karische Berg- und Ortsnamen mit lykischem Anschluß. In: HS. 119, 117–132. Schürr, Diether, 2008, Zur Rolle der lykischen Mindis. In: Kadmos 47,147–170. Szemerényi, Oswald 1990: Einführung in die vergleichende Sprachwissenschaft. 4., durchgesehene Auflage. Darmstadt. Tekoglu, ˘ Recai, 2002–2003, Three New Lycian Inscriptions from Tlos and Asartas¸. In: Die Sprache 43, 104–114. Tekoglu, ˘ Recai, 2006, TL 29: una nuova proposta di lettura. In: Festschr. Roberto Gusmani. Alessandria, 1703–1710. Valério, Miguel, 2007, ‚Diktaian Master‘: A Minoan Predecessor of Diktaian Zeus in Linear A? In: Kadmos 46, 3–14. Wörrle, Michael, 1995, Epigraphische Forschungen zur Geschichte Lykiens V. Die griechischen Inschriften der Nekropolen von Limyra. In: Chiron 25, 387–417. Zahle, Jan, 1988, Den lykiske by Tuminehi. In: Nordisk Numismatisk Unions Medlemsblad Nr.5, 98–104. Yakubovich, Ilya 2009: Sociolinguistics of the Luvian Language (Brill’s studies in Indo-European languages & linguistics; v. 2). Leiden – Boston Zgusta, Ladislav, 1964, Kleinasiatische Personennamen. Prag.
Mühlstraße 7 D-63584 G r ü n d a u [email protected]
Diether Schürr
Greek « Λ and further evidence for laryngeal as resonant
127
Greek Λ« and further evidence for laryngeal as resonant Abstract A long discarded etymology connecting the Greek word with Balto-Slavic and Germanic cognates is rehabilitated on the basis of one fresh semantic consideration together with the common lack of demonstrable e-grade forms, which is here explained by reconstructing h3 in the root onset. The resonant-like inhibiting effect of the laryngeal on assibilation of the assumed root-initial prevelar (~ palatovelar) in o- and zero- grade forms enables a wider range of Balto-Slavic cognates to be considered, thus providing a motivated answer to the ongoing question of whether the groups usually assigned to, e.g., OE géap ‘wide, spacious’ and OE géap ‘bent’ belong together, as well as drawing other suggested etymologies of « Λ (or their analogues) into the mix. A possible etymology for the previously essentially isolated group of Lith. ziaubti ˜ ziaubiù ‘eat with mouth full/open’ also emerges.
Greek « Λ ‘light, easy, nimble’ is without etymology according to J.N. O’Sullivan (LfgrE s.v.) and Beekes (2010 s.v.).1 The first judgement (published c. 1990) may or may not, the second certainly does, imply a tacit rejection of Van Windekens’ (1986 s.v.) ingenious proposal (viz. *k2os-2 ‘hair’ + *u(e)bh- ‘weave’),3 but both are most certainly in tune with the treatments of Chantraine (1968–1980 [1999] s.v.) and Frisk (1960–1972 s.v.), the latter also echoing Boisacq’s (1950 s.v.) unargued rejection of a number of suggestions. These suggestions may be reviewed briefly as follows.
1
2
3
Whether it is fair to cite here Liberman’s views, recently summarized by Urban (2008: 191 f.), on the generally excessive and arbitrary parsimony of anglophone etymological lexicographers, I leave to others to decide. In my bitectal notation for PIE, k1, g1, g1h represent prevelars which develop into palatovelars and pure velars in satem languages and plain velars in centum languages, while k2, g2, g2h are backvelars with positionally determined labialization developing into labiovelars and plain velars in centum PIE and plain velars in satem PIE (Woodhouse 1998; 2005); absence of subscripts indicates ‘precise nature uncertain’. For tritectal sources k1 etc. will simply indicate palatovelars, k2 etc. labiovelars; k(1) etc. indicates uncertainty in the source (pointing however to bitectal k1 etc.). I feel the proposed semantic development ‘made of hair’ > ‘light’ is not entirely satisfactory: a single hair would have been much lighter than a woven artifact – a single pigtail can after all be comparatively heavy.
128
Robert Woodhouse
Böhtlingk/Roth (1875: 259) connect « Λ with Skt. s´ubh- ‘dahinfahren’, a connection tentatively repeated by Uhlenbeck (1898–1899: 317a s.v. çóbhate) and thence Mansion (1904: 92). If the root is inherited it must have had s mobile; and although Mayrhofer (1956–1980, 3: 358; evidently unchanged in 1992–2001, 2: 657 f.) rejects the semantic analysis that ascribes the meaning ‘dahinfahren’ to the Skt. word, nevertheless the possibility of connecting « Λ with PIE *skeubh- ‘push, shove’ allegedly found only in Germanic and BaltoSlavic (Kümmel LIV2: 560) still remains to be considered. This is because the connection between ‘push’, ‘pull’ and ‘move’ – the latter being the basis of the chief alternative etymology for « Λ – is surely unimpeachable. The alternative etymology just referred to seems to have been first published by Bezzenberger (1878: 352), connecting « Λ with ChSl. gчnoti ˛ ‘movere’, gybati ‘movere, periclitari’, gybnoti ˛ ‘perire’, gybeˇlц ‘interitus’, gubiti ‘perdere’, gubitelц ‘perniciosus, tyrannus’, OPruss. gub-ans/-ons ‘gegangen’, per-gub-ans/-ons ‘kommend’ etc. as part of an argument against Johannes Schmidt’s proposal that u-diphthongs in Greek should be derived wholesale from nasal diphthongs. It is noteworthy that Bezzenberger (ibid. n. 1) specifically separated the above group from OCS preˇ-gybati ‘bend, bow’ etc. Fick (1883) appears to be the first to propose the fuller form of the same etymology by equally specifically ignoring Bezzenberger’s separation of the ‘bend’ group and connecting «, Λ « Λ and 1. with both ChSl. “gчne˛ti” (i.e. gчnoti), ˛ gybati ‘movere’ and gybчkч ‘flexibilis’, Lett. gubt ‘sich bücken’. Later Fick (1890: 414) came to agree with Bezzenberger’s separation (associating in the process 1; « ‘hangend’ with «) Λ and was followed in this, apparently, by Wharton (1890: 73) and Prellwitz (1905: 240). More recent scholarship, however, agrees with Fick’s earlier view of the common origin of all the above Balto-Slavic material, while totally ignoring the associated etymology of «, Λ no doubt in the wake of Berneker’s (1924 s.v. gчbezˇц) unargued turnaround (cf. Berneker 1896: 293) in dismissing it: see, e.g.,Vasmer (1953–1958 s.v. gubít’), Pokorny (1959: 450), Fraenkel (1962–1965 s.v. gaubti), ˜ Sˇanskij (1972 s.v. gnut’), Kümmel (LIV2: 188), Smoczyn´ski (2007 s.v. gaubti). ˜ A third possibility is due to Leo Meyer (1901: 285), who connects « Λ with NHG, MHG hüpfen, MHG hüpfen, hupfen, Eng. hop, evidently a Kluge’s law form so the precise quality of the root final
Greek « Λ and further evidence for laryngeal as resonant
129
consonant is immaterial to the comparison – a fact that once again facilitates a connection with PIE *(s)keubh-. In other words this third suggestion can be regarded as equal to the first. Osthoff (1910: 17 f.) found the semantic relations implicit in all of the above etymologies of « Λ unimpeachable and refused to attempt to choose between them. One could of course beat the semantic drum once again, but perhaps just a little harder, by pointing out that the gloss ‘easy-going’ supplied by Liddell/Scott/Jones (s.v.) in order to facilitate translation of the superlative of « Λ in Xenophon’s Agesilaus (11:12) rounds out the notion of ‘flexibility’ already suggested by the gloss ‘agility’ for 0« by the same authorities, thus pointing us in the direction of Fick’s original, fuller set of Balto-Slavic cognates. But equally, or perhaps, more compelling is the possibility of arguing for a point of root structure that seems to be shared by « Λ and the same set of Balto-Slavic cognates yet appears not to have been brought into the argument hitherto, viz. an apparent lack of e-grade forms. This is readily seen in the case of «, Λ Beekes (2010 s.v.), Chantraine (1999 s.v.) and Frisk (1960–1972 s.v.) all taking issue with the apparent o-grade plus barytonesis in an adjective. It is less self-evident in the case of the Balto-Slavic forms since there is a widespread tendency to replace naturally developed forms reflecting *eu by their o-grade equivalents (Stang 1966: 73 f.; Shevelov 1964: 272–275); but I agree with Kortlandt (1979: 57) following Pedersen (1935) that a rationale for the replacement needs to be found in each case. This might include such things as Endzelin’s assimilation rule, which was actually foreshadowed by Pedersen (1935: 151) and is supported and expanded by Stang (1966: 74). In the case of Lith. gaubti ˜ gaubiù ‘bend outwards; cover’, Smoczyn´ski (2007 s.v.) finds such a rationale by pointing out that the o-grade seems to have spread to the whole paradigm from the present gaubiù which was originally proper to the no longer existing iterative, inf. *gauby´ti. Plausible as this explanation may be, the fact remains that there appears to be no evidence whatsoever anywhere of a manifestly e-grade form either in Balto-Slavic or among the Germanic cognates adduced by Pokorny (1959: 450). This contrasts with the position of the superficially similar group of ON gaupn ‘die hohle Hand; both hands held together’, OHG goufana ‘handful, empty hand, palm’ (English glosses due to Orel 2003 s.v. *gaupno), Old Danish gøben, gjøben ‘handful’ (de Vries 1977 s.v.
130
Robert Woodhouse
gaupn), Lith. ziùpsnis ‘a pinch (e.g. of salt); a handful; an armful’, OE geopan ‘swallow’, geap ‘wide, spacious’ etc., which is carefully separated from the ‘bend/move’ set by Pokorny (1959: 449 f.) following Persson (1912: 835 n. 2) (though not by some other scholars, see Woodhouse 2008: 23 n. 14), since OE geopan directly attests the e-grade, perhaps represented in Lith. by ziaubti ˜ ziaubiù ‘eat with mouth full or wide open’ (i.e. ‘eat uncouthly, with little regard for niceties of etiquette’?) which seems a semantically fitting partner to OE geopan ‘swallow’ etc. but for which neither Fraenkel (1962–1965 s.v. ziauberoti) ˜ nor Smoczyn´ski (2007 s.v.) can suggest any extra-BaltoSlavic etymology, even though it is from a form such as this that Lith. ziùpsnis (perhaps once *‘a mouthful’?) must have got its palatalized initial. If this is accepted then an explanation that covers the apparent lack of e-grade forms in both Gk. « Λ and Lith. gaubti ˜ gaubiù etc. can be the presence of h3 immediately preceding the root vowel. Thus, given also the lack of Balto-Slavic evidence of palatality in the initial consonant, the root of these words can be provisionally reconstructed as PIE *g2hh3eubh-, the only problem with which is the Balto-Slavic non-acute. This non-acute, however, can be due to loss of the labialized laryngeal in Balto-Slavic posterior to the change *h3e > *h3o by dissimilation against either the initial aspera or the combined labiality of the coda *-ubh- of the root or a combination of both factors, since the three PIE laryngeals were probably still distinguishable from each other, i.e. h3 was still labialized, as late as the time of Winter’s law near the end of the Balto-Slavic period (Kortlandt 2002: 6). If the laryngeal can have been lost by dissimilation dialectally at this relatively late stage, then presumably it could also have been lost selectively (perhaps in o-grade and/or zero grade forms) in PIE, leading to the presence of two distinct roots of ultimately shared origin. Acceptance of such a proposal would enable us to posit a connection between ON gaupn, Lith. ziùpsnis, OE geopan etc. and the Balto-Slavic/ Germanic ‘bend/move’ group with the proviso that the initial consonant would then have been a prevelar, the assibilative palatalization of which was inhibited in o-grade and zero grade forms in the ‘bend/ move’ group by the presence of the laryngeal. On this basis the original root would have been *g1hh3eubh- ‘bend ( ? )’, giving rise directly to Gk. « Λ and the ‘bend/move’ group *ghHoubh- > *ghoubh-, representing both e- and o-grades, the original Balto-Slavic e-grade
Greek « Λ and further evidence for laryngeal as resonant
131
*g¤h(h3)e/oubh- having either disappeared or merged with the parallel root with lost laryngeal *g1heubh- ‘hollow/swallow’, the latter being then of course quite distinct from Gk. "« etc. (pace Pokorny 1959: 449), which must have had, if anything, h2. The above formulation of the inhibition of assibilation of the root initial prevelar by the laryngeal in o- and zero grade forms is based primarily on Kortlandt’s (1978) analysis of the many cases in which the same process occurs conditioned likewise by immediately following back vowel but where the trigger is a resonant rather than a laryngeal. It is, however, also to be added to other instances in which a laryngeal has been observed to have the same effect, such as PSl. *xoi-r/d- ‘grey’ with *k1h1 (Lubotsky 1989: 56 f.) and Russ. xápat’, soxá and (analogical) plesˇ’ with *k1h2 (Kortlandt’s “Postscript”, 2002: 20), in all of which the combination with voiceless laryngeal also results in aspiration > spirantization of the stop. Further examples are Russ. dial. kotorá ‘quarrel, strife’ (< *k1h3ot- beside Ved. s´átru- ‘enemy’ < *k1h3ét-, with normal lack of voicing by h3 in word initial clusters, Woodhouse in press § 5.5 [ii]) and Slavic gosц ˛ ‘goose’ in which Kortlandt (1985a: 119) attributes the lack of palatality of the initial to the presence of syllabic nasal in the gen. *g1hh2nsos, whereas it can with equal or better facility be attributed to the inhibiting effect of the immediately following h2 followed by non-palatal syllabic nasal. Finally, the claim of PIE *(s)k(1)(h3)eubh- to belong here as well can be briefly considered, it being remembered that when the s is present the tectal loses its palatal feature unless *i follows (Kortlandt 1979: 58 f.). The occurrence of the e-grade among the BSl. forms is somewhat uncertain, hence Russ. skubu ‘pluck, pull by the hair’ may well have original -k- owing to relatively late retention of the laryngeal (pace Kümmel LIV2 s.v. *skeubh- n. 4) while Pokorny’s (1959: 955) OCS “sˇcˇuti” (sc. sˇtuti?) ‘hetzen’ would, like the Germanic e-grade forms, reflect early loss of h3. Given the several possible factors involved in the two proposed dissimilations of h3 as well as my hypothesis of a preglottalized origin for the asperae (Woodhouse 1995; 2009: 110) and the evidence for a glottalic reflex of all three laryngeals in Balto-Slavic (Kortlandt 1985b: 186) it is possible that either or both of the dissimilations could have fallen either in the earlier period of preglottalized asperae or in the later period of aspirated asperae. The principal watershed between the two processes is the change *h3e > *h3o. The later Balto-Slavic dissimi-
132
Robert Woodhouse
lation can occur at any time posterior to this change or to the lapse of productivity of e: o ablaut, and prior to the onset of Winter’s law.
References Beekes, Robert Stephen Paul, 2010, Etymological dictionary of Greek. Leiden/ Boston: Brill. Berneker, Erich, 1896, Die preussische Sprache: Texte, Grammatik, etymologisches Wörterbuch. Strassburg: K.J. Trübner. –, 1924, Slavisches etymologisches Wörterbuch, 1. Heidelberg: Carl Winter. Bezzenberger, Adalbert, 1878, Homerische Etymologien, Beiträge zur Kunde der indogermanischen Sprachen, BB 4: 313–359. Böhtlingk, O and Roth, R., 1875, Sanskrit-Wörterbuch, 7. St. Petersburg [1966, Osnabrück/Wiesbaden]. Boisacq, Émile, 1950, Dictionnaire étymologique de la langue grecque étudiée dans ses rapports avec les autres langues indo-européennes, 4th ed. with Index by Helmut Rix. Heidelberg: Carl Winter Chantraine, Pierre, 1968–1980, Dictionnaire étymologique de la langue grecque: histoire des mots, [1999, 2. ed. avec un Supplément sous la direction de Alain Blanc et al.] Paris: Klincksieck. de Vries, Jan, 1977, Altnordisches etymologisches Wörterbuch. Leiden: E.J. Brill. Fick, August, 1883, Miscellen, 28, BB 7: 94 –, 1890, Wortschatz der Grundsprache, der arischen und der westeuropäischen Spracheinheit (= Vergleichendes Wörterbuch der indogermanischen Sprachen, 4th ed., ed. A. Bezzenberger, A. Fick and Wh. Stokes, 1). Göttingen: Vandenhoeck & Ruprecht. Fraenkel, Ernst, 1962–1965, Litauisches etymologisches Wörterbuch. Heidelberg / Göttingen: Carl Winter / Vandenhoeck & Ruprecht. Frisk, Hjalmar, 1960–1972, Griechisches etymologisches Wörterbuch, 3 vols. Heidelberg: Carl Winter. Kortlandt, Frederik, 1975, Slavic accentuation: a study in relative chronology. Lisse, Netherlands: Peter de Ridder. –, 1978, I.-E. palatovelars before resonants in Balto-Slavic, in: Recent developments in historical phonology, ed. Jacek Fisiak, (Trends in linguistics, Studies and monographs 4). The Hague/Paris/New York: Mouton. Pp. 237–243. –, 1979, Three problems of Balto-Slavic phonology, Zbornik za filologiju i lingvistiku 22/2: 57–63. –, 1985a, Long vowels in Balto-Slavic, Baltistica 21/2: 112–124. –, 1985b, Proto-Indo-European glottalic stops: the comparative evidence, FLH 6/2: 183–201. –, 2002, From Proto-Indo-European to Slavic, www.kortlandt.nl. Kümmel, see LIV2. Kurschat, Alexander, Litauisch-deutsches Wörterbuch, 4 vols. Göttingen: Vandenhoeck & Ruprecht.
Greek « Λ and further evidence for laryngeal as resonant
133
LfgrE = 1955–, Lexikon des frühgriechischen Epos, in Zusammenarbeit mit dem Thesaurus Linguae Graecae und mit Unterstützung der UNESCO und der Joachim-Jungius-Gesellschaft, Hamburg vorbereitet und herausgegeben von Bruno Snell. Göttingen: Vandenhoeck & Ruprecht. Liddell, Henry George and Scott, Robert, 1968, A Greek-English lexicon, revised and augmented throughout by Sir Henry Stuart Jones with the assistance of Roderick McKenzie and the co-operation of many scholars. Oxford: Clarendon Press. LIV2 = 2001, Lexikon der indogermanischen Verben, ed. Helmut Rix assisted by Martin Kümmel, Thomas Zehnder, Reiner Lipp, Brigitte Schirmer, 2nd rev. ed. by Martin Kümmel and Helmut Rix. Wiesbaden: Dr. Ludwig Reichert. Lubotsky, Alexander, 1988, The system of nominal accentuation in Sanskrit and Proto-Indo-European, (Memoires of the Kern Institute 4). Leiden / New York / Copenhagen / Cologne: E.J. Brill. –, 1989, Against a Proto-Indo-European phoneme *a, in: The new sound of IndoEuropean: essays in phonological reconstruction, ed. Theo Vennemann, (Trends in linguistics, Studies and monographs 41). Berlin/New York: Mouton de Gruyter. Pp. 53–66. Mansion, Joseph, 1904, Les gutturales grecques. Paris: Gand. Mayrhofer, Manfred, 1956–1980, Kurzgefaßtes etymologisches Wörterbuch des Altindischen. A concise etymological Sanskrit dictionary, 4 vols. Heidelberg: Carl Winter. –, 1992–2001, Etymologisches Wörterbuch des Altindoarischen, 3 vols. Heidelberg: Carl Winter. Meyer, Leo, 1901, Handbuch der griechischen Etymologie, 2. Leipzig: S. Hirzel. Orel, Vladimir, 2003, A handbook of Germanic etymology. Leiden: E.J. Brill. Osthoff, Hermann, 1910, Die Sippe leicht, Morphologische Untersuchungen auf dem Gebiete der indogermanischen Sprachen 6: 1–69. O’Sullivan, see LfgrE. Pedersen, Holger, 1935, Lit. iau, Studi Baltici 4: 150–154. Persson, Per, 1912, Beiträge zur indogermanischen Wortforschung. Uppsala/Leipzig: A.-B. Akademiska Bokhandeln/Otto Harrassowitz. Pokorny, Julius, 1959, Indogermanisches etymologisches Wörterbuch, 1. Bern/Munich: Francke. Prellwitz, Walther, 1905, Etymologisches Wörterbuch der griechischen Sprache, 2. verbesserte Aufl. Göttingen: Vandenhoeck & Ruprecht. ˇ eskij slovar’ russkogo jazyka, 1/ 4 (G). MosSˇanskij, N.M. (ed.), 1972, Etimologic fi cow: Moscow University Press. Smoczyn´ski, Wojciech, 2007, Słownik etymologiczny je˛zyka litewskiego, 2 vols. Vilnius: Vilnius University. Uhlenbeck, Christianus Cornelius, 1898–1899, Kurzgefasstes etymologisches Wörterbuch der altindischen Sprache. Amsterdam [1973, Osnabrück: O. Zeller]. Urban, Mateusz, 2008, Secretary bird, or how an etymological dictionary should be written, SEC 13: 191–199. Van Windekens, A.J., 1986, Dictionnaire étymologique complémentaire de la
134
Robert Woodhouse
langue grecque: nouvelles comtributions à l’interpretation historique et comparée du vocabulaire. Leuven: Peeters. Vasmer, Max, 1953–1958, Russisches etymologisches Wörterbuch, 3 vols. Heidelberg: Carl Winter. Wharton, Edward Ross, 1890, An etymological lexicon of Classical Greek. London [1974, Chicago: Ares]. Woodhouse, Robert, 1995, Proto-Indo-European injective asperes [sc. asperae], IF 100: 92–100. –, 1998, On PIE. tectals, IF 103: 40–60. –, 2005, Assibilative palatalization of tectals in Phrygian and the adequacy of bitectal frameworks for Proto-Indo-European, IF 110: 205–234. –, 2008, Some more etymologies not requiring PIE *b, in: Problemy zahal’noho, hermans’koho ta slov’jans’koho movoznavstva/Problems of general, Germanic and Slavic linguistics: do 70-ricˇcˇja profesora V.V. Levyc’koho/papers for the 70-th anniversary of Professor V. Levickij, ed. Gabriel Altmann, Iryna Zadorozhna and Yuliya Matskulyak. Cˇernivci, Ukraine: Knyhy-XXI. Pp. 18–26. –, 2009, The Sindhi implosives: archaism or innovation?, IF 114: 79–117. –, in press, Lubotsky’s and Beekes’ laws, PIE *(H)r-,*(H)i(V)-, *a and some other laryngeal matters, SEC 16.
School of Languages and Comparative Cultural Studies The University of Queensland B r i s b a n e QLD 4072, Australia [email protected]
R o b e r t Wo o d h o u s e
Two unexplained terms in Iliad: and
135
Two unexplained terms in Iliad: and
Abstract Homeric near-hapaxes and are unexplained terms that translators have interpreted as epithets for the eagle and transliterated or solved as chromonyms. This situation is paralleled by an absolute confusion concerning the identification of the raptor. As far as is concerned, it seems possible to assume that the original formal value contained in the chromonymic indication is not actually revealed, as it actually happens in the literature. On the other hand, , which is to be considered from a perspective (i.e. Iliad 24/316) where the apparition of the eagle is tantamount to a public, daytime, oniric vision, is a highly unusual name for the eagle suggesting a reference to . As it takes on the name of fate itself, that of the eagle can be thus interpreted as a very well-marked instance of omen-name that can be called “zooaptonym”.
The unusual forms (acc.) and (acc.), both found in Iliad 24/3161 as epithets for the eagle, are generally translated as chromonyms. The protagonist raptor is thus described according to a double chromatic specification, at least if one wishes to follow the traditional translations, here limited to the Italian cases: e [Zeus] subito mandò un’aquila, l’uccello sovrano, il cacciatore nero che chiamano anche “scuro”2 mandò subito un’aquila, il più perfetto tra gli uccelli, il cacciatore dal colore scuro che viene detto anche il Nero.3 mandò subito l’aquila, la più rapida fra i volanti, la cacciatrice nera, che dicono fosca4
1
2 3 4
) # >μ - 0 7 / %0 0 # χ λ μ
. Chantraine 1979, 194 includes « into a “groupe particulier et assez obscur […] constitué par les adjectifs de couleur, désignant surtout une couleur sombre”. Paduano 2007. Ciani 19902. Calzecchi Onesti 1950.
136
Alberto Manco
e mandò subito l’aquila, il più propizio fra tutti gli uccelli, il morphnos predatore, che chiamano anche “il nero”5. That double specification (an uneconomical repetition of the raptor’s color), albeit supported by the translation tradition, might be reconsidered here. We can begin with an argument concerning . The base * - in is regularly linked through a set of relationships with the notion of circularity, something to which the literature has hardly ever offered much attention. Among others, Benveniste himself belonged to the chromonymic tradition followed by the Iliad’s translators. In his argument on the French name of the pig, ‘porc’6 (its base pork- falling within the same lexical field as * -), the Scholar focused, although very carefully, on the animal’s color at an early age: “on pourrait rapprocher *porko- du radical *perk- qui fournit gr. perkos “mouchete; épervier”, perknos “moucheté de noir”, prox, prokos “faon”, skr. prsni- “tacheté, moucheté“”7. Even supposing it is appropriate to “rapprocher” the base *perk- and its cognate *pork-, it is also important to explain contextually the apophonic distance between the two forms8.
5
Cerri 2000. But the question is ancient and widespread, as evidenced by the few examples here related to old English translations: Jove sent his eagle, surest of all signs, The black-plum’d bird voracious, Morphnos nam’d, And Percnos (Cowper 1814) Forthwith he sent an eagle, surest of omen among winged birds, the dusky eagle, even the hunter, that men call also the black eagle (Murray 1957, p. 587)
6
7 8
And immediately sent the hunter eagle, the most sure augur of Birds, which they call Morphnos and Percnos (Anonymous 1833). Formally, Benveniste’s “porc” is found in English only as the culinary name for meat from the domestic pig (pork). As far as the linguist is concerned, however, English ‘pork’ and French ‘porc’ fall into the same class of things. In other words, the points at issue here are the general Indo-European conditions under which the designation of pig as pork, porc, porco etc. has arisen. Benveniste 1949, 74–91. For an investigation on the apophonic question, cf. Silvestri 2006.
Two unexplained terms in Iliad: and
137
In Ancient Greek, unlike other Indo-European languages, the base * - wasn’t normally employed to designate the young pig but, in the same way as * -, every object sharing the notion of circularity, even tridimensionally. Concerning * -, see «, “net”, and 0«, “ring, band”9; * -, on the contrary, would usually designate fish (fated to become perches10 par excellence to Philyllius, Hippocrates, Aristophanes, and Ephyppus11, among others) as well as
9
10
11
Homer, Iliad, 6/320 and, identical, 8/495: … % ξ μ« >" κ "0, λ ξ " 1« % 0«. Calzecchi Onesti cit.: “in cima splendeva la punta / in bronzo dell’asta, e intorno un cerchio d’oro correva”; Ciani cit., p. 120: “splendeva la punta di bronzo, splendeva l’anello d’oro che intorno correva”; Paduano cit.: “in cima alla [lancia] splendeva / la punta di bronzo, e intorno correva una fascia dorata”. Murray cit.: “blazed the spear-point of bronze, around which ran a ring of gold”. Extremely popular marine and freshwater fish (Genus Perca, Family Percidae, Order Perciformes). Despite the different local names, the group numbers, among others, the European Perch (Perca fluviatilis), the Comber (Serranus cabrilla), common in the Mediterranean and known in some Italian regions as “perchia”, and the “Clownfish” (Amphiprion percula). The translation of 0 as “dark fish” is prevailing in the literature and only in some dictionaries: freshwater fish with a dark back (“pesce d’acqua dolce, a dorso scuro”) in Rocci 1961, but only “a river-fish” in Liddell-Scott 1949. Chantraine 1979 translates the corradical « into “staked out” even when fish, not only snakes, are concerned: “le sens originel [emphasis added] semble être “avec de taches noires”, dit de raisins ou d’olives qui mûrissent (Poll. 1, 61; 5, 67), d’un serpent (Arist., Nic.), de poissons (Marc. Sid.), cf. encore Hp. VC 19; nom d’un aigle noir ou tacheté de noir, le même que le « (Il., 24,316)”. However, these hypotheses are somewhat puzzling. Although perches owe their name to a chromatic feature, they are neither “spotted” nor “dark”. Dark fish are countless, but many perches are not dark at all. Rather, they can display rings all around their bodies. Therefore, the perch is referred to as “the ringed one”. Philyllius Comic., fragment 13 line 3: 0, «, 1«, 0, , « Ephippus Comic., fragment 12 line 2: L 0«, , , 0« Hippocrates et Corpus Hippocraticum Med., De diaeta section 48 line 2: < «, , «, 0, % Aristophanes Comic., Fragmenta, fragment 18 line 1: LY< , τ
«, λ 7« . #A Q ³ λ 0 " .
138
Alberto Manco
raptors (hence ). In addition to the latter forms, more common but still infrequent, I should like point out a further form , gen. «, generally associated with the name of the fallow deer12. As a result, while the base * - appears to be restricted to an inanimate, static dimension, * - concerns an animate, dynamic one13. Yet there is further evidence beyond the chromatic dimension, more in favor of a reference to shape. Aside from Greek, the base *perk- normally refers to the feminine universe, for example to designate those characteristics that might distinguish it from the male one. This is a most ancient tradition, frequently used in a vulgar sense, as confirmed by some particular comic texts with a residual, obsolete Neap. “perchia”14, and the still existing Southern Italian “purchia”. The form “porca” (its base being *pork-), which indicates the mound normally produced between two drills after ploughing, is used in turn to designate something according to its shape: Qua aratrum vomere lacunam striam fecit, sulcus vocatur. Quod est inter duos sulcos elata terra dicitur porca15. Several other examples would be available, and the etymological indexes would confirm such tendency. Hence, if we 12
13
14
15
Cf. * – in overt association with those « mentioned by Herodotus (T7 ξ >"%1. (λ 1, »« « λ . «. Storie, V, 16, 21) – as well as on the form in Hesychius’s gloss α . Such forms, albeit unique, may encourage one to rethink what is traditionally considered an isolated zero-grade base, i.e. * -. Silvestri 2006 reminds that the apophonic grade can be employed to represent either staticity or eventivity. Galiani-Mazzarella Farao 1789, s.v.: “Pesce di mare di figura sconcia dal greco 0 [sic!], perca. [… H]a un’ampia bocca, e facilmente si fa coll’esca prendere da’ pescatori, adropasi a dinotar chi troppo parla, e non sa tener secreti, e si sa prendere in parole. […] Si trasferisce a dinotar donna vile, e disonesta, non che d’infelice fisionomia, ed è termine ingiuriosissimo” (“Saltwater, indecent-looking fish from Greek 0 [sic!], perch. [It] has a wide mouth, and is easily baitcaught by fishermen; used to denote he or she who speaks much, and cannot keep a secret, and rambles on. […] It can be used to denote vile, dishonest women, as well as poor-looking ones, and is an extremely insulting term”). Varro, R.R. 1, XXIX (cf. Benfey 1842, 362: “vgl. lat. porcae=sulci (Fest. p.218 M.)”). The literature has applied a distinction between perk¹- “Rippe; Rippengegend, Brust”, perk¹-, prek¹- “gesprenkelt, bunt” and perk¹- “aufreißen, aufwuhlen, aufkratzen” (Walde-Pokorny 1927, s.v.). This is the reason why the relationship between “abyss, crevice” and “dark”, between “black” and “pig”, between perch ( 0) and fallow deer ( ) remains unexplained.
Two unexplained terms in Iliad: and
139
wished to follow the systemic data, we should ultimately admit that while the reference to color is secondary, the primary sense of the base refers to shape. In this respect, we can notice two glosses by Hesichius. The first, “ 0« “ 1«“”, confirms the application of the base to represent circular entities. The second, “ « “1 «. ξ«
ξ μ 4μ 1 $ “”, confirms an indirect reference to birds and fish, again in close connection with the idea of circularity. The reference to color is absent, while the reference to outline is prevailing. Interestingly, this second case displays a marked correlation with 1 «, a net (with regard to fish) or cage (when birds are being referred to) now considered according to their typical round shape. A K1 «16 is the circular object with which different sorts of fish and other marine animals can be caught (octopi, squid, crustaceous). Contextually, « should be therefore interpreted as the circular tool by means of which fish is caught in general. It appears that the shift to a color indication through the bases *perk- and *pork- might have originally derived from a reference to shape. In view of that, one may put forward that perches (despite the incalculable number of speckled, streaked or flecked fish) used to be thus defined because of the rings around their bodies, regardless of their actual color (rings didn’t necessarily have to be “dark”). In the same way, raptors were referred to as percni, despite the vast number of speckled, streaked or flecked birds. Incidentally, compare also a few recent, coexisting designations of raptors in addition to those relative to the ancient world ( «17, but also vulture, a name in which the 16
17
Liddell-Scott cit., s.v.: 1 «: “convex [ecc.]”; cf., ibid., “gibbus” and, possibly, 1%« “pudenda muliebria”(Eup. 233; Ar. Ach. 782). – Cf. also Greek ", « “pudenda muliebria”, Lat. porcus “pudenda muliebria”; Southern It. purchia and purchiacca “pudenda muliebria”; It., dialect. sorca (>sulcus?) “pudenda muliebria”; Neap. vocca e’ perchia “bocca di perca = bocca larga” (perch mouth = large mouth), metaph. Cf. Iliad, 22/139: U « 4 ( « 0 7 . More hesitations and disagreement in translations. Note that the bird being referred to in this passage of the Iliad is about to pounce on a dove (cf. therefore pigeon hawk). – As for «, cf. Lat. circus. Calzecchi Onesti cit.: “come il nibbio sui monti, che è tra gli uccelli il più rapido”; Paduano cit.: “come sui monti il falco, che è l’uccello più rapido”; Ciani cit., 437: “come quando sui monti un falco – il più veloce fra gli uccelli – […]”.
140
Alberto Manco
idea of “turning” was ultimately established as the salient element for the way these birds move in the air). Such designations are in turn based on elements including some reference to the circularity of their flight (therefore, a formal value, albeit dynamic!), allowing for percno to be the bird “that flies in a circle”, or “which describes circles”. The reference to color realizes the slow conversion of the indication of a formal element (a more or less accomplished roundness). This is possible considering the perception of a pattern impressed on the animal’s body, or of an action typical of the animal (e.g. its way of flying), which eventually allows us to adjust the interpretation of and therefore elude that awkward, double definition – made with some twisting not always to be agreed with – of eagles and other raptors as “streaked” or “dark birds”, as found in passage 24/316 of Iliad. As a matter of fact, it seems more likely for a raptor to be defined according to its own, unique peculiarities (the flight in a circle appearing a major one), rather than to qualities shared by every other kind of bird. In conclusion, it seems possible to assume that if the original formal value in the chromonymic indication is not actually revealed – as it actually happens in the literature18 – it is even less plausible that such value might refer to a distinction between spots, streaks, rings and circles. This would ultimately answer for the fact that the corradicality between fallow deer and perches, nets and raptors has remained unexplained. We can now focus on the other near-hapax, . In Iliad 24/316 the deceptively explicit form (acc.) is interpreted by translators as an epithet for the eagle and transliterated or solved as a chromonym. It appears that the translation seeing a chromatic specification in survives beyond the fact that, in the general consideration, the term remains unexplained, as I should like to summarize here. What is more, this situation is paralleled by an absolute confusion concerning the identification of the raptor.
18
Murray cit.: “as a falcon in the mountains, swiftest of winged things”. Which leads to a similar conclusion: “PIE *pórk¹os “pig” (as one who creates a furrow-like track while rooting in the earth)” (Mallory and Adams 1997, s.v. *pórk¹os). For pórk¹os, however, one may expect a different derivative pattern.
Two unexplained terms in Iliad: and
141
A rapid survey of translations reveals a few exceptions in which the authors confine themselves to merely transliterating into morphnos. However, this should not encourage any identification of the Homeric specimen with Morphnus guianensis, the Guiana Crested Eagle, a raptor to be found only in some very specific areas of South America. Morphnus guianensis, in fact, is a breed belonging to a special Morphus genus. Tradition proves highly controversial: Cuvier 1836, 201, for example, who also attests it as “Guiana Hawk”, includes it amongst the hawks rather than vultures, and warns against any identification of the unknown Greek mòrphno[s] with Morphnus guaianensis: “Morphnus: nom grek d’un’espèce indéterminée d’oiseaux de proie”. In other words, Cerri’s cautious and suitable solution does not seem to unravel the puzzle of the raptor’s identity. Over the years, much curiosity – which should be regarded with suspicion – has arisen around the problem uncovering obsessively copied, distant antecedents with the recurring mistake/ambiguity nearly regularly turned into evidence based on ancient words. Inevitable, for example, is what is restated in AA.VV. 1818, s.v. Morphnos: “Le vrai morphnos des auteurs grecs est le petit Aigle”. Risso 1826, 28 uses morphnus to refer to both the goshawk and the pigeon hawk: “Morphnus: Aigle autour; Morphnus Palombarius: Aigle vulgaire”. And yet, on no account can these raptors, known all over Europe, be associated with Homeric [«]. In order to designate the pigeon hawk, in fact, Homer seems to employ the term «. Boraston 1911, 216, on the other hand, having judged the two epithets in Iliad 24/316 to be synonyms, connects “>μ« « or «, The Morphnos or Perknos Eagle” with the “Golden Eagle, Aquila Chrysaëtus” and, accordingly, with >μ« «, “The Black or Dark Eagle”. There have been several reconstructive attempts focusing on the identity of the bird, which prove much more openly inspired on the analysis of ancient texts. According to Mooney, it is central to consider Lycophrone’s account of as a noun19: “No maiden, but the eagle golden-sired”20. Russo 195021, instead, with reference to the Hesiodean occurrence 1., writes: 19 20 21
μ " 4 « %«. Mooney 1921, p. 90. Russo 1950, 105.
142
Alberto Manco
“ 1.: L’Et. M. v. 1« provides the equivalent « $ . Hesychios explains «, nowhere else found, as $μ« %«, a sort of bright-red eagle (.). M «, as attribute for eagle, is only present in [Iliad] 316; in Aristotle historia animal. 618b, 25 and in Lycophrone 838 designates, undoubtedly as a noun, an’”eagle” or “vulture” (as descrive in Suda)”. In focusing on size rather than color, Boraston writes: “Evidently this eagle was more than usually large, for Homer has in no other instance made special mention of the size of an eagle” (238). Further on, in dismissing once more the traditional association between and dark color, he adds: “Hesiod (Sc. Here. 134) writes of arrows being winged with feathers of the « 1« which I read, not as a colour epithet, but as the ‘flashing’ «, in allusion to the swift flight of arrows” (238). The clue – and this may be effortlessly ascertained – has not been taken into account by Homer’s translators, who have rather persisted in translating « into “black”, “dark”, “sullen”, and so forth. A few other remarks might be added here, along with those made by the three authors just mentioned. Firstly, with respect to Mooney’s translation, one wonders why Lycophrone would have used nominally a form bearing such an obvious chromatic value as « by further specifying it through an adjective denoting a bright red color. With regard to Aristotle, one may add that, in quoting Homeric , he commits a couple of infractions. The first concerns stress ( «). On this matter, Chantraine draws attention to the fact that the form is a variant found in some manuscripts of the Iliad but also suggests that the adjective « might have been nominalized into «22. This would entail that Aristotle resumes an adjectival function of the term no longer in use in Homer’s times. The second infraction concerns the fact that a raptor named “ ” is judged by 22
Chantraine 1968, s.v.
Two unexplained terms in Iliad: and
143
Aristotle the second (not first) among raptors as to size and strength, in groundless disagreement with the sacredness and primacy of the Homeric raptor. An additional observation is that the contextual terms within Aristotle’s discourse – « and 0 « – are entirely original, again as if to mark the necessity to facilitate a not so obvious individuation of some particular bird: 6E ξ « $ (λ χ « ,, 1 « % λ V< 9 0α >, ξ 2« λ Ν0 λ «, (, ξ « λ «α X λ 6O 0 « 0 ( 9 P ( ) .. As a result, the raptor in Aristotle’s text is an eagle qualified in some particular way: a means that, while showing the difficulty of framing in the best possible way the animal under consideration, escapes any precise identification: … «. As regards Hesychios, it seems appropriate to add that he seems distant from the chromatic value:: “ · ρ « $ [λ %«”23. As a matter of fact, Hesychios develops an equation between the terms and ρ «, both apt to cover values such as shape, idea, appearance (“ · > , ρ «”24). Not least, he comments on λQ through 1 “to nod, promise, approve”. The relationship between and , however, has not met with success among scholars. Among those scholars who have rejected any explicit link between and , Brugmann suggests the association “*morkÜsnos: aisl. miorkue ˛ “Finsternis“”25; Hirt himself puts forward a few solutions including – depending on the languages in which the outcome is to be found – values such as “sombre”, “obscurité”, “bigarré” or “scintiller”, “cligner les yeux”, “mort”. While still excluding any possible cognation with , he mentions other hypotheses allowing for I.E. *mer-gÜ, *mer-qÜ and *mer-gÜh, all affected by a value extension in base *mer- “briller” (which, incidentally, is associated in Greek with the outcome .).
23 24 25
Hesychii Alexandrini 1965, s.v. Hesychii Alexandrini 1965, s.v. Cit. in Hirt 1901, 226.
144
Alberto Manco
According to Chantraine, falls into “un groupe particulier et assez obscur par les adjectifs de couleur, désignant surtout un coleur sombre”26; therefore his conclusion on the “étymologie incertaine” in is relatively unsurprising. Windekens has suggested, or rather confirmed, the meaning “sombre (déja chez Homère)” and assumed a blending between the extremely rare 5 « and * "« “sombre”: “autrement dit le - initial de « proviendrait de * "« qui a exactement le même sens”27. On the other hand, Belardi28 has pointed out that the ancients would already find the term obscure (“già per gli antichi di significato oscuro”) and that it denotes an eagle identified by Aristotle as “duck hunter” (“cacciatrice di anitre”, which is inescapably disappointing as to the role played by the raptor in Iliad 24/316). Also, he has shown that Pisani had reconstructed as mr0 gwo-gwhno, the first part corresponding to O.Aisl. mr0 ga- “wild beast, game, gazelle, bird”, and the second being an agentive name of root *gwen- (“la prima parte corrisponde a ai. mr0 ga- “bestia selvatica, selvaggina, gazzella uccello” e la seconda è un nome d’agente della radice *gwen-”), but won’t agree with such problematic reconstruction, for it would require an intermediate * stage. At any rate, Belardi has not provided any alternative hypothesis. The connection with has been supported by Solmsen who, instead, as Boisacq illustrates, “revendique pour « le sens de “bigarré” et l’apparente a , dont le sens prim. aurait été “exterieur bigarré“”29. Hitherto, two possible conclusions can be drawn. One is that the hypotheses here considered and share the same chromatic value. This position allows the individuation of more or less adaptable antecedents. None of such explanations, however, seems utterly convincing nor, perhaps, does it wish to. Solmsen’s solution, for example, albeit realistic, is still to be motivated. How could a chromatic meaning such as “dark” or similar be possibly associated with ? Before establishing what bird actually corresponds to Homeric , 26 27 28 29
Chantraine 1979, 194. Cf. Liddel Scott, s.v.: “epith. of an eagle, dub. sens”. Windekens 1986 s.v. «. Belardi 1950. Boisacq 1916 s.v., 646.
Two unexplained terms in Iliad: and
145
one should rather discern why he actually names it . This does not automatically mean to reject the proximity of with , but rather to reconsider its value and suggest its possible semantic relationships. That might be associated with is supported by the solution found in Iliad 24/316 for the contextual μ , a form in which the morphological proximity suggested by the element - seems to lay beyond mere coincidence. Besides, there is a second level of proximity between and , both associated with the idea of partition-separation (in the case of those plumages with color bands clearly distinct from the others, for instance, but also in that of a “separation” from life) which, moreover, supports the apophonic association mor- / mer-: cf. « “part, portion”, but also “fate, destiny” on the one hand, and « “part, portion”, on the other. See also Hesychios “ · «”. An acc. is recurrent in Iliad with the general sense of “fate” (frequently with a negative nuance), while definitely circumscribed seems the base -, restricted to « and usually translated as “mortals”. The second conclusion is that, unlike the case with semantic “certainties” (a color), the literature provides a great number of accounts referring to some sort of formal dimension, with fluctuations ranging from meanings referable to the exceptional size of the raptor (Boraston), to those of a definitely smaller bird (Aristotle). These observations need be integrated with a few others. The mysterious meaning of cannot be deciphered without taking into due consideration the word’s broad morphological solidarity with (just another coincidence, or maybe more than that?). Note, however, as many points of difference: one, markedly referable to apophony (the -- in vs. -- in ), the second to a plainly accentual nature (oxytonesis vs barytonesis)30. In particular, it is of the utmost importance to consider that Iliad 24/316 shows a shift to a very high diaphasic register never explicitly noticed. The passage does not describe a human fact, but a natural event. As Hamann puts it in recalling the primitive ages evoked by Vico, every manifestation of nature was one word (“ogni manifestazione della natura era una
30
I am grateful to D. Silvestri for quite a few reflections on these last two aspects.
146
Alberto Manco
parola”31): Iliad 24/316 uses words to represent what has been announced by signs. Nonetheless, the passage is not a place of “agreed signs” – as Vico would have termed it – one of the common, earthy language serving to communicate life’s everyday needs, but a totally analogical, self-evident, phenomenologically marked place that the philosopher would have labeled “divine language”, and which in the literature recurs as “language of gods”. From a perspective in which the apparition of the eagle is tantamount to a public, daytime, oniric vision, the name of the eagle, , suggests a reference to . As it takes on the name of fate itself, that of the eagle can be interpreted as a very well-marked instance of omen-name. Accordingly, would be to as what offers an omen is to what flies in circles (which, in turn, is a divine sign itself). In consideration of this, it seems no less than peculiar that Iliad 24/316 might have chosen to use the solemn register to repeat, at least twice, judgments on color not really significant (black … dark32; dark … Black33; black … sullen34), rather than use descriptions of a highly meaningful behavior: that offers omen … the circling one. Seemingly, this solution refers to a directly “Homeric” textual context eluding reconstructive excesses of any sort, as if – to Homer – had already been the relic of the most inaccessible IndoEuropean remoteness. Instead, its proximity with might enable us to focus on its directly visible base ( -) as well as on the affixation (- ), rather than on other hypotheses marked by many more asterisks than the one just suggested.
References AA.VV. 1818: Nouveau Dictionnaire d’histoire naturelle, vol. min-moz, Deterville, Paris. Anonymous 1833 = Homerus [1833]. Aristote [1969]: Aristote, Histoire des animaux, tome iii, livres viii–x, trasl. by Pierre Louis, Les Belles Lettres, Paris. 31 32 33 34
Hamann 1967 cit. in Wohlfart 1981, 62. Paduano 2007. Ciani 19902. Calzecchi Onesti 1950.
Two unexplained terms in Iliad: and
147
Bailly 1950: Bailly Charles, Dictionnaire grec-français, Hachette, Paris. Belardi 1950: Belardi Walter, “Etimologie greche”, Doxa III, 1950. Benfey 1842: Benfey Theodor, Griechisches wurzellexikon, Zweiter Band, Reimer, Berlin. Benveniste 1949: Benveniste Emile, “Noms d’animaux en indo-européen. I. – Le nom du porc”, BSL 45. Boisacq 1916: Boisacq Emile, Dictionnaire étymologique de la langue grecque, Winter, Heidelberg. Calzecchi Onesti 1950 = Omero [1950] Cerri 2000 = Omero 2000 Chantraine 1968: Chantraine Pierre, Dictionnaire etymologique de la langue grecque: histoire des mots, Klincksieck, Paris. Chantraine 1979: Chantraine Pierre, La formation des noms en grec ancien, Klincksieck, Paris. Ciaceri 1901: Emanuele Ciaceri (ed. and translation), La Alessandra di Licofrone, Giannotta, Catania. Ciani 19902 = Omero [19902] Ciani 2005 = Omero [2005] Cowper 1814 = Homer [1814] Cuvier 1836: Cuvier Georges, Le règne animal distribué d’après son organisation, iii ed., tomo i, Hauman e. C., Bruxelles. Galiani-Mazzarella Farao 1789: Galiani Ferdinando, Mazzarella Farao Francesco, Vocabolario delle parole del dialetto napoletano, che più si scostano dal dialetto toscano, Tomo I, Porcelli, Napoli. Hamann 1967: Hamann Johann Georg, “Des Ritters von Rosencreuz letzte Willensmeynung über den göttlichen und menschlichen Ursprung der Sprache”, in Schriften zur Sprache, Suhrkamp Verlag, Frankfurt Hanson 1999: Hippocrates, On head wounds, in Maury Hanson (ed.), Akademic Verlag, Berlin. Hesychii Alexandrini 1965: Hesychii Alexandrini, Lexicon, post Ioannem Albertum recensuit Mauricius Schmidt, Hakkert, Amsterdam. Hett 1955: Aristotele, Minor works. On marvellous things heard, coll. LOEB, transl. by W. S. Hett, Heinemann (London) & Harvard University Press (Cambridge, Massachusetts). Hirt 1901: Hirt H., “Kleine grammatische Beiträge”, Indogermanische Forschungen 1901. Homer [1814]: Iliad, by William Cowper, Boston, Buckingham. Homerus [1833]: Homerus, The Iliad, by anonymous ed., London, Whittaker. Homer [1957]: The Iliad, by A. T. Murray, London, William Heinemann; Cambridge, Massachusetts, Harvard University Press. Liddell-Scott 1940: Liddell Henry George, Scott Robert, A Greek-English Lexicon, vol. ii, ix ed., Clarendon Press, Oxford. Louis 1969 = Aristote [1969] Mallory-Adams 1997: Mallory James P. and Adams Douglas Q., Encyclopedia of Indo-European Culture, Fitzroy Dearborn Publishers, London and Chicago.
148
Alberto Manco
Mooney 1921: George W. Mooney (ed. and transl.), The Alexandra of Lycophron, Bell and sons, London. Murray 1957 = Homer [1957] Omero [1950]: Omero, Iliade, by Rosa Calzecchi Onesti, Einaudi, Torino. Omero [19902]: Omero, Iliade, by Maria Grazia Ciani, Marsilio, Venezia. Omero 2000: Omero, Iliade, by Giovanni Cerri, Milano: Rizzoli, 20002. Omero [2005]: Omero, Iliade, by Maria Grazia Ciani, 4th ed., Marsilio, Venezia. Paduano 2007 = Omero [2007] Omero, Iliade, by Guido Paduano, Mondadori, Milano. Hett 1936: Page T.E., et alii (eds.), On marvellous things heard, in Aristotle minor works, Loeb Cambridge. Risso 1826: Risso Antoine, Histoire naturelle, t. iii, Levrault, Paris. Rocci 1961: Rocci Lorenzo, Vocabolario greco-italiano, xiv ed., Lapi, Città di Castello. Russo 1950: Esiodo, Scutum, by Carlo Ferdinando Russo, La nuova Italia, Firenze. Silvestri 2006: Silvestri Domenico, “Apofonie indeuropee e altre apofonie”, in Studi linguistici in onore di Roberto Gusmani, volume III, R. Bombi, G. Cifoletti, F. Fusco, L. Innocente, V. Orioles (eds.), Edizioni Dell’Orso, Alessandria. Vegetti 1976: Opere di Ippocrate, in Mario Vegetti (ed.), UTET, Torino. Walde-Pokorny 1927: Walde Alois, Pokorny Julius, Vergleichendes Wörterbuch der Indogermanischen Sprachen, De Gruyter, Berlin-Leipzig. Windekens 1986: Windekens, Albert Joris van, Dictionnaire étymologique complémentaire de la langue grecque, Peeters, Leuven. Wohlfart 1981: Wohlfart Günter, “Vico e il carattere poetico del linguaggio”, Bollettino del Centro di Studi Vichiani 11.
Università di Napoli “L’Orientale” Piazza san Domenico Maggiore, 12 I-80134 N a p o l i Italien [email protected]
Alberto Manco
The development of Indo-European middle-passive verbs
149
The development of Indo-European middle-passive verbs: a case study in Ancient Greek and Old Church Slavonic* Abstract This work is a syntactic analysis of the change that affected the selectional properties of the Indo-European middle-passive verbs that take two arguments (namely, of the psych verbs denoting separation). Here, I will account for the differences in case marking of the objects of such “middle-passive” verbs, paying special attention to the contrast between the Ancient Greek pattern and the Old Church Slavonic pattern. I will show that the Greek and the Slavic middle-passive structures differ in a series of properties, and argue that they represent different points in the evolution from a PIE. intransitive / stative pattern to independently developed patterns: Greek developed into an active-middle pattern (similar to Latin deponents), while Slavic developed into a passive, namely, a reflexive passive pattern.
1. The puzzle Ancient Greek (Gk.) middle verbs denoting ‘separation’ and related notions, such as ‘fear’, ‘shame’ and ‘refusal’ (e.g. φ2,% ‘to be afraid’), like Indo-European (IE.) deponent verbs in general, seem not to undergo the so-called Burzio’s Generalization (Burzio 1986), according to which verbs that lack an external argument (here, a subject) do not license accusative case on the internal argument (here, an object). Consider, for example, the sentence in (1), where a Gk. middle verb of fear takes an accusative object:1 * I would like to thank Maia Duguine, Iván Igartua, and Aritz Irurtzun for their comments on this paper. My gratitude to the audience of the X DIGS conference for their valuable comments on my presentation. This work is part of the FFI2008–03816/FILO and the FFI2008–04786/FILO research projects, which are funded by the Spanish Ministry of Education and Science, and has been financed by Joseba Lakarra’s research group on historical linguistics and the HM-2009-1-1 and HM-2008-1-10 research projects (both funded by the Department of Education, Universities and Research of the Government of the Basque Country). 1 Most Greek examples have been extracted from the Perseus and the Titus internet pages (Perseus Digital Library: www.perseus.tufts.edu; Thesaurus IndoGermanischer Text und Sprachmaterialien: http://titus.uni-frankfurt.de), and
150
(1)
Nerea Madariaga
(Gk.: Thucydides, 2: 87) T« ’κ D 7 φ2, κ , )"λ
*". who part. because her of-you fears [the future]AC not truth has ‘Those of you who fear the future (battle) because of the last one are not right.’
On the other hand, Old Church Slavonic (OCS.) and other Slavic languages (i.e. Old Russian) conserved the PIE. ablative case (incidentally, we must note that its uses were conflated with the ones of the genitive case in Greek and Slavic) on the objects of the verbs of ‘separation’ and related notions (e.g. bojati se˛ ‘to be afraid’), observing the generalization that verbs that lack a subject do not license accusative case on the object, but mark it with another case (e.g. genitive case), as shown in (2): (2)
Ne uboite se˛ ubivajos ˛ ˇ cˇixч teˇlo. (OCS.: Mt. 10: 28)2 not fear refl. AC killing PART.GN body ‘Do not be afraid of those who kill your body.’
Finally, both languages have innovated with respect to PIE. in that they made occasional use of the mechanism of replacing the PIE. ablative case with a P(repositional) P(hrase) headed by an overtly (morphologically) realized preposition (Allan 2003, Luraghi 2003), as in (3): (Gk.: Sophocles, Trachiniae 671) (3) a. E , > , ( Ρ φ29 . explain if explainable from what GN fear ‘Explain, if you can, what you are afraid of.’ b. Nч ne boi se˛ otч mokч. ˛ (OCS.: Codex Suprasliensis 8b: 7) but not fear refl. AC from tortures GN ‘But do not be afraid of suffering.’ In this work, I will account for the contrast between the case marking patterns of middle-passive sentences in Gk. and OCS. (with more
2
some of the Old Church Slavonic ones, from the dictionary by Cejtlin, Vecˇerka, Blagova et alii (1999 [1994]). The OCS. New Testament references for which the source is not explicitly specified are all from the Codex Marianus.
The development of Indo-European middle-passive verbs
151
general reference to IE. and PIE., too). I will show that this contrast is not superficial, but has deep roots in the structure of IE. middlepassives itself and their development during the history of the different IE. groups of languages. This work is organised as follows: in section 2, I will overview the types and general characteristics of IE. middle-passive constructions, based on previous accounts about these constructions in PIE. and early IE. In section 3, I will introduce the main hypothesis of this work, and in sections 4 and 5, I will provide different arguments in favour of my proposal. In section 4, I will present evidence in favour of the structures triggered by the verbs under analysis, according to my proposal; and in section 5, I will show how these structures account for the differentiated object case marking of these verbs. Finally, in section 6, a conclusion is offered.
2. Observations about IE. middle-passive patterns in the literature In this section, I will provide the classifications of the middle-passive patterns in IE. that have been proposed in the literature, according to different properties: semantic and morphological, as well as from the point of view of their historical development.
2.1. Semantic middle-passive patterns and their classifications The semantic properties of middle-passives in PIE. and early IE. have been usually given in the literature in the form of hierarchies of constructions with respect to some semantic or pragmatic property. Here, I will review some of these proposals, namely, the ones that are relevant for the account I am developing here. Croft, Shyldkrot & Kemmer (1987) consider the semantics of middle constructions close to reflexives and related to transitive verbs from the point of view of the causal chain of events: the subject of middles is both the initiator and the endpoint of the action (vs. passive constructions, where the subject is the endpoint, but not the initiator of the action). Middle voice, therefore, does not consist in valence decreasing (as in passives), but in valence increasing (cf. also Gerritsen 1992, among others). Consider example (4):
152 (4)
Nerea Madariaga
Ivan breetsja. (Russian) IvanNOM shaves3SG-PASS ‘Ivan shaves himself / Ivan gets himself shaved (by someone else).’
Sentence (4) has two readings: a middle reflexive reading (‘Ivan shaves himself’) and a causative reading, where someone else shaves Ivan. In both cases, the middle form adds a second argument to the one we have in overt syntax (Ivan), namely, an implicit agent (be it Ivan himself or someone else). We will come back later to this notion of implicit arguments.3 Allan (2003) provides a semantic classification of the Ancient Greek middle construction, in a hierarchy according to the property “subject affectedness”, which, as we will see later, correlates with the morphological markers of the Gk. middle. Thus, his hierarchy goes from verbs with “a subject that undergoes a change of state” to the ones where “the subject does not undergo a change of state”: (i) (ii) (iii) (iv) (v) (vi)
Passive: ‘is made’. Spontaneous / mental processes: ‘I get burned’ / φ2 ‘I fear’. Body / collective motion: ‘I turn’ / $Q ‘to gather’. Perception / mental activity: 1 ‘I taste’ / Q ‘I consider’. Reciprocal: $. Q ‘I fight’. Direct / indirect reflexives: 1 ‘I get shaved’ / ‘I make for myself’.
The classification given by Savcˇenko (2003 [1974]) specifies the following meanings of the middle-passive forms in early IE. languages (I will concentrate on the early Gk. examples): (a) (b)
3
Intransitive meaning: he defines it as the most ancient one, as we will see later on. Stative meaning: also very frequent in Homer and early IE. languages; ex. , ‘I am lying’.
For implicit arguments in passive / middles in English, see also Bhatt & Pancheva (2005).
The development of Indo-European middle-passive verbs
(c)
(d) (e)
153
Middle in the classical sense (indirect reflexive), that is, the active meaning of the verb adding ‘for oneself’, as in Classical Gk. ‘I do something for myself’. Reflexive meaning: as in 1 ‘covers oneself’. Passive meaning: probably a late development (cf. below).
Finally, to complete the picture, I will briefly mention Vaillant’s (1965 [1924], 2002 [1948]) classifications of middle-passive meanings in OCS. (examples with the middle-passive reflexive se˛):4 (1)
(2) (3)
4
Intransitive forms (he calls them “reflexive forms”, because of the pronoun se˛), which can be divided into two groups: (i) the ones that are only “reflexive”, e.g. blistati se˛ ‘to shine’. The middle-passive verbs of mental state and perception that are conserved in OCS. are included here, e.g. radovati se˛ ‘to be happy’, diviti se˛ ‘to be surprised’, bojati se˛ ‘to be afraid’; and (ii) the ones that can be either reflexive or active. The pairs of verbs in this last group can have different meanings, depending on the presence of the reflexive se˛ or not, e.g. drчzˇati (+ accusative case) ‘to hold (an object)’ vs. drчzˇati se˛ (+ genitive case) ‘to observe (laws or principles)’. Reciprocal: sчbirati se˛ ‘to gather’. Passive: dastч se˛ vamч ‘it will be given to you’.
Another way to form passive constructions in Slavic is by combining a passive participle with a copular verb (OCS.: byla sчpasenaPAST-PASS.PART. ‘(she) was saved’). I will disregard here these constructions because their middle-passive status is not clear. It could be argued that these constructions are copular sentences, and not what is usually understood under “passive” voice for the following reasons: (i) they do not show the whole range of meanings that the se˛ form has (they form only passives, not medial or reflexive-reciprocals); (ii) as a consequence, they do not combine with middle-passive verbs of mental state, the ones denoting fear, refusal, and taking complements (our object of study; i.e. verbs like OCS. bojati se˛ ‘to be afraid’ do not form passive participles); (iii) they show very distinct properties with respect to the se˛ forms: e.g. the Ukrainian -no/to- participial constructions (passive constructions with an accusative object) are exclusively found with past participles, but not with se˛).
154
Nerea Madariaga
2.2. Middle-passive patterns from the point of view of their morphology and history The previous literature about the IE. middle-passives proposes different possible paths of development of these constructions in the history of the IE. languages. Jasanoff (1978) argues that statives and middles were closely related in IE., as opposed to the active paradigm. He shows that the perfect (and stative) desinences and the middle ones are etymologically related in many early IE. languages, such as Hittite, Gk. or Indo-Iranian, and have similar distribution in them.5 For example, the Gk. deponent present verbs ‘to become’ and ‘to get mad’, with middle desinences, correspond to the active perfect forms and 0 , respectively.6 Savcˇenko (2003 [1974]) also identified intransitive and stative meanings as the most ancient of the IE. middle forms. In fact, Homer middle forms do not have the classical indirect reflexive sense found later (“for oneself”). Thus, according to Savcˇenko (2003 [1974]), the Homeric verb \ means ‘to stand’ and not ‘to put for oneself’, (" is ‘to hold, to keep oneself’ and not ‘to have for oneself’, etc. The stative meanings were very frequent in Homer (purely statives: - ‘to be sitting’, , ‘to be lying’, or mental / sensitive states: ‘to want’, ΝQ ‘to be afraid’, ‘to think’) and other early IE. languages, e.g. Hittite (ar- ‘to stand’, iya- ‘to go’), Sanskrit (a¯ste ‘is sitting’, ihate ‘wants’, bha¯yate ‘to be afraid’), etc. In addition, some active stative verbs have middle desinences, which is a proof of their semantic relation; for example, the stative verb nahsariya- ‘to be afraid’ in Hittite (cf. Gk. φ2 ‘to be afraid’) had middle desinences. Finally, some middle verbs of process (non-eventive) in Homer could be transitive and have a direct object (e.g. $ ‘to pray’, $Q ‘to be afraid’); the same observation holds for Sanskrit (arthate ‘wants’, lokate ‘sees’).
5
6
Rix (1988) – cf. also Clackson (2007: 149) – proposes two initial paradigms: stative vs. active. Clackson (2007: 149) points out that there have been reconstructed different series of desinences for PIE. middle voice, ranging from two (middle / stative and active) up to four (active, middle, stative and perfect). According to Jasanoff (1978), in early IE., the old perfect forms lost their connection with the middles and merged with other past forms, e.g. Hittite *akha originally meant ‘I have died’, but later became ‘I am dead’.
The development of Indo-European middle-passive verbs
155
The passive meaning of these forms is found in the Indo-Iranian, Italic and Celtic languages, as well as in Greek and Gothic. However, Savcˇenko (2003 [1974]) also argues that the passive meaning of these forms is probably a late development of the stative meaning because: (i) it is semantically related to the corresponding stative, i.e. was always non-eventive (cf. Latin deponent verbs); and (ii) in the early texts, the passive constructions could not have an overt agent. In early Gk., Latin and Gothic texts, the passive constructions very rarely took an agent marked with an oblique case; later on, this agent appears more and more frequently. In conclusion, Savcˇenko (2003 [1974]) proposes the following path of evolution for IE. middle forms: stative constructions (nominative subject) > stative passive (nominative subject, without an agent) > passive with agents. In any case, in later IE. languages, the original IE. middle-passive desinences got lost and new middle-passive constructions arose by adding a reflexive pronoun to active verbs. This was the case of Romance and Slavic, where the 3rd person accusative (some times, also the dative) form of the reflexive pronoun (Latin se, OCS. se˛) was reanalyzed as the marker of the constructions analyzed here (cf. Haspelmath 1987 and Kemmer 1995 for Romance: reflexive > middle > passive).7 In this paper, I will concentrate on those middle-passive verbs that were able to take more than one argument. In Allan’s (2003) classification, these are mental processes (φ2 ‘I fear’), and perception and mental activities (1 ‘I taste’ / Q ‘I consider’). In Vaillant’s (1965 [1924]) classification, the crucial verbs are intransitives expressing mental processes and perception / mental activities (moliti se˛ ‘to pray’, drчzˇati se˛ ‘to observe (laws or principles)’, bojati se˛ ‘to be afraid’, stydeˇti se˛ ‘to be ashamed’, etc). We can characterize semantically all these verbs as processes or activities denoting perception. 7
Looking at the etymology of the verb bojati se˛ ‘to be afraid’, for example, we can track down its ancestor to an active reconstructed form *baie- (infinitive *bitei) > *bojati ‘to scare’ (Vasmer 2003 [1950–58]). The morphological formation of bojati se˛ is quite transparent: it should have been formed by adding the short accusative reflexive pronoun form se˛ to an active verb (the reconstructed form *bojati), although the corresponding active form is not attested in Slavic. In other Indo-European languages, however, active forms etymologically related to bojati se˛ are common, for example, the active form in Avestan bayente ‘to scare, to frighten’ (Vasmer 2003 [1950–58]).
156
Nerea Madariaga
From the point of view of their object case marking, I will concentrate on the perception / mental verbs denoting “separation” and related notions (e.g. “fear”, “shame”), that is, the ones that correspond to a PIE. verb with an ablative-case marked object; recall that the uses of the ablative case conflated with those of the genitive case in Greek and Slavic (hereby, I will call this case ablative-genitive case). The examples in (5a, 6a, 7a, 8a, 9a) illustrate some of these verbs in OCS.; compare their ablative-genitive objects with the accusative ones in the parallel New Testament Greek examples in (5b, 6b, 7b, 8b, 9b): (5) a. Ne strasˇo˛ se˛ protivnaago. (Codex Suprasliensis 490, 9) not fear ACT refl. enemy GN (St. John Chrysostom In sanctum pascha, 3) b. O) (λ φ . μ [] . not even fear ACT [the enemy]AC ‘I am not afraid of the enemies.’ (6) a. Posramleˇjotч ˛ se˛ s[y]na moego. are-ashamed ACT refl. [son mine]GN b. #E μ Dμ . Respect MID [the son my]AC ‘They will respect my son.’
(Mk. 12: 6)
(7) a. Postyditч se˛ mene i moixч slovesч. is-ashamed ACT refl. meGN and [my words]GN b. #E" %9 λ ]« ( ]« «. is-ashamed MID meAC and [the my words]AC ‘(The one who) is ashamed of me and my words.’
(Lk. 9: 26)
(8) a. Vцsego brasˇцna vчzgnos ˛ ˇ a se˛. [all food]GN loathed ACT refl. b. P» 2 7 (2 1. [all food]AC loathed MID ‘(He) loathed all kind of food.’ (9) a. Ne otцvrчgo˛ se˛ tebe. not will-deny ACT refl. you GN b. O) $ . not not you AC will-deny MID ‘I will not refuse you.’
(Psalm 106: 18)
(Mt. 26: 35)
The development of Indo-European middle-passive verbs
157
The examples in (5b, 6b, 7b, 8b, 9b) have been taken not from Ancient Greek, but from New Testament Greek, in order to show the exact contraposition between the two patterns (OCS. and Greek). But of course, Ancient Greek, right from Homer’s times, had already replaced ablative-genitive objects by direct accusative objects, as shown in (10): (Homer Odyssey 16, 446) (10) a. O) ξ % % Ν .. not something him death AC fear MID bid ‘I bid him not be afraid of death.’ (Xenophon Anabasis 7, 3, 26) b. Kλ $". 7 ) κ 9 0« μ . and retreating not not fear MID [the enemy]AC ‘And retreating, you shall not fear the enemy.’
2.3. Syntactic properties of middle-passives in Gk. vs. OCS. In this section, I will check some well-known syntactic properties of the middle-passive verbs which can take objects in Ancient Greek (Gk.) and Old Church Slavonic (OCS.). Some of the differences between old and new IE. middle-passives have been previously detected in the literature (see e.g. Comrie 1998, Jasanoff 1978), as mentioned in section 2.1: (i) As we said before, the Gk. morphological middle pattern corresponds to a PIE. series of middle desinences, shared with the Indo-Iranian group (Jasanoff 1978, Savcˇenko 2003 [1974]), directly inherited from one of the possible – reconstructed – PIE. middle systems. OCS., however, has innovated with respect to the IE. ancestor in that it has lost the middle conjugation and created a new pattern of passive voice by adding the reflexive accusative pronoun se˛ to an active verb. This is the evolutional path followed by other new IE. creations, such as Romance middle-passives. By the moment, I will call these two patterns the “old” middle-passive (Gk.) and the “new” middle-passive (OCS.). (ii) The old middle morphology could denote the old PIE. semantic values, i.e. stative and indirect reflexive meanings (an active verbal action in benefit of oneself), while the new pattern did not have those meanings: on the one hand, stative verbs are not se˛-verbs in OCS., as
158
Nerea Madariaga
shown in (11b) vs. the Gk. stative middle verb in (11a). On the other hand, in OCS., indirect reflexives are not middles, but construed with an active verb and a dative complement or a complex P(repositional) P(hrase), as shown in (12b) – cf. with the middle form in Gk. with an indirect reflexive meaning (12a): (Gk.: Odyssey 24, 75) (11) a.#E ) 7 , 1’ 5, φ ’ #A" . in it part. are-lying MID white bones illustrious Achilles ‘There are lying your white bones, glorious Achilles.’ b. I vц lozˇixч svoixч lje˛gotч. ˛ (OCS.: Sinai Psalm 103, 22) and in beds theirs are-lying ACT ‘And they are lying in their beds.’ (Gk.: Thucydides 4: 9) (12) a. O) @ Ρ ( ". ) . ( ) . %. not part. was arms AC in country uninhabited to-procure MID ‘It was impossible to procure arms for themselves in an uninhabited country.’ (OCS.: Lk 12: 21) b. Tako sчbiraje˛i sebeˇ a ne vч b[og]ч bogateˇje˛. thus collect ACT himself DT and not towards god being-rich ‘This is how is the one who makes money for himself and is not rich in God.’ (iii) In this paper, I have introduced an additional difference between the two patterns: recall from examples (1–2), repeated below for convenience, the different subcategorization of the Greek and Slavic middle-passive verbs of mental / perception processes with respect to their possible objects: Gk. middles could take an accusative object (see example 1 above), while their OCS. counterparts could not take an accusative object, but only a genitive one (example 2 above). Thus, the Gk. pattern does not seem to observe the so-called Burzio’s (1986) Generalization, while the OCS. one does observe it. This follows from the fact that the objects of passive verbs should not be marked with accusative case according to this generalization: “All and only the verbs that can assign a %-role (i.e. a semantic role) to the subject can assign accusative case to an object.” (Burzio 1986: 178). In the “strong” version of Burzio’s Generalization a “subject” is taken to be an argument with a semantic role “agent”. However, if we
The development of Indo-European middle-passive verbs
159
widen the notion of “subject” to other kinds of subjects, the Gk. pattern above can be accounted for without violating Burzio’s Generalization, as we will see later.
3. The hypothesis In this paper, I propose that the difference in the object marking of Greek vs. Slavic middle-passive verbs lies in their different nature and structures. I will argue that Burzio’s (1986) Generalization contradicts neither the Gk. nor the OCS. pattern because of the following reasons: the OCS. construction in (2) contains a pronoun, which is the one that receives accusative case and therefore blocks accusative case assigning to any other possible argument in the sentence. This pronoun does not have gender or number features, so that it cannot agree with the verbal tense and cannot be a nominative subject. The DP that surfaces as nominative is originally created in the object position, but it does have gender / number features and can therefore agree with the verbal tense and receive nominative case, i.e. surfaces as the subject, yielding a passive construction (the object becomes a nominative subject). As for the reflexive pronoun in (2), I will argue that, even if it has no gender / number features, being a pronominal element, it needs in Slavic to be marked with some morphological case. Its case materializes as the other case available in a regular ditransitive sentence: accusative case. This is represented in structure (13): (13)
T(ense)P NP NOM [gender/number] ····
····
····
····
···
(OCS.: Reflexive construction) T’
T [gender/number]
V(erb)P
Move
se˛ ACC [gender/number] ·· ·· ··
·· ·· ··
·· ·· ··
·· ·· ··
·
V
V’ NP
160
Nerea Madariaga
In this way, I simply support the association of gender / number features and case marking in the case of nominative arguments (i.e. a nominative subject generally agrees in gender / number with the tense of the sentence), but dissociate gender / number features and case marking in the case of accusative objects. This seems so at least in most IE. languages, where only the nominative argument “agrees” with the verb, but accusative case is no matter surfaced on the other argument (cf. Alboiu 2007 for a similar proposal in Romanian). (13) will be argued to be the OCS. pattern, where middle-passive verbs are new formations created by adding the reflexive accusative pronoun se˛ to the active voice (Schmalstieg 1983, Vasmer 2003 [1950–58]). I will call this pattern “reflexive construction”. As for the Gk. pattern, I will argue that we are not dealing here with a real passive construction, but with the old IE. middle-stative pattern, which could be either transitive or intransitive (cf. section 2.1). I will concentrate here on the transitive / active pattern (because it is the one that accept objects), which I will denominate “active middle”. Active middles are typical of Gk. and reconstructed also for PIE. (Jasanoff 1978, Meillet 1965 [1924], Meier-Brügger 2003). The verbs under analysis in this paper give rise to the regular transitive structure represented in (14). The presence of a regular nominative subject allows accusative marking on the object, as predicted by Burzio’s Generalization. T(ense)P NP NOM [gender/number] ····
····
····
····
···
(Gk.: active middle) T
T [gender/number] V
V(erb)P
··················································
(14)
NP ACC[gender/number]
Finally, I will argue that the development of the original PIE. constructions under analysis (perception / mental non-eventive verbs denoting “separation” and related notions) into Greek and Slavic was as follows: mental / perception processes were stative constructions in IE; the NP arguments that could be associated to them surfaced with the case required by the semantics of the relation between the verb and the NP, in our structures (verbs of “separation”), ablative-geni-
The development of Indo-European middle-passive verbs
161
tive case (cf. Meillet 1964 [1906], Savcˇenko 2003 [1974]). Then, in Greek, many of the NPs marked with such “adverbial” cases became simple objects (marked with accusative case), and the structure became transitive, as in (14). In Slavic, the middle desinences disappeared, and the new reflexive-passives arose; the change from PIE. “adverbial” ablative-genitive case to accusative case on the internal arguments (which took place in Slavic in active verbs, cf. Borkovskij 1978) was not possible in reflexive-passive constructions, because of their structure, which was really passive (the internal argument – the object – surfaced as nominative, and accusative surfaced on the passivizing accusative reflexive pronoun, in absence of another possible morphological case suitable for it). Therefore, the objects of the reflexive verbs of “separation” and related notions were forced to preserve ablative-genitive case on their objects. I will arrange the argumentation to prove this hypothesis around two main axes: in section 4, I will argue for a differentiated structure of the middle-passive verbs with objects in OCS. vs. Gk. Then, in section 5, I will account for the differences in object case marking of these verbs, taking into account the corresponding structures and their evolution in IE.
4. The structure of the IE. middle-passive constructions 4.1. The position of the nominative NP in (13) and (14) Formal studies on modern languages have investigated the syntactic properties of the traditionally considered “passive” constructions. For instance, Raposo & Uriagereka (1996) show that the so-called “indefinite se” constructions in Romance (a new formation similar to the one of Slavic) are, in fact, active structures. Previous work on this topic had already established that another Romance se, “generic se”, triggers regular transitive structures (Cinque 1988).8
8
In this respect, Cinque (1988) and Raposo & Uriagereka (1996) find a contrast between French and Portuguese se-constructions. (i) a. Omtem compraram-se demasiadas salsichas. (Portuguese) yesterday bought-se too-many sausages ‘Yesterday, too many sausages were bought.’
162
Nerea Madariaga
In the OCS. constructions under analysis, the NP that surfaces as nominative in (13) seems an object, rather than the subject. As for the nominative NP in the Gk. construction in (14), it rather seems a subject than an object. Here are the arguments in favour of this. 4.1.1. Semantic relations According to Croft (1993), mental perceptions and activities involve two participants: a stimulus and an experiencer, which usually involve two causal relations: the experiencer can direct his attention to the stimulus, and the stimulus can cause a mental affectedness in the mind of the experiencer. In the first case, the experiencer is more “agentive”; in the second case, the stimulus is more “agentive”. I will then assume that the experiencer can be the external argument (subject or agent) of a verb denoting mental / perception processes and activities (in which case the stimulus is the internal argument or object), or the other way round: the stimulus can be the external argument, while the experiencer should be the object.9 As I proposed in the structures in (13, OCS.) vs. (14, Gk.), the participants in the event differ in the two languages depending on their position in the structure. The crucial test to show that these structures are correct is the contrast between internal vs. external arguments. If (13) is correct, i.e. the NP that will occupy the subject position is the internal argument, the resulting structure is a real passive and therefore, we expect to be able to add an additional agent to such a construction, as in common passives. On the contrary, in (14), if the NP
9
b. La poubelle se vide au sous-sol. (French) the trash empties in underground-floor ‘One empties the trash in the underground floor.’ These authors argue that the Portuguese NP demasiadas salsichas and the French la poubelle have different properties; various syntactic tests for subjecthood, such as control, topicalization, and semantic properties, suggest that the French NP behaves like a subject, while the tests show that the Portuguese NP has direct object properties. A similar semantic effect was described in other accounts of psych verbs, such as Dowty (1991). This author claims that psych verbs can vary with regard to its semantic role properties. For instance, the arguments in John fears dogs and in Dogs frighten John are ambivalent with respect to their semantic roles; in each of them, John or the dogs are the agent, the experiencer, or the patient.
The development of Indo-European middle-passive verbs
163
that occupies the subject position is the external argument, while the other NP is the internal argument, the addition of an agent (a second external argument) is not expected. This prediction is borne out: in OCS., besides the original IE. ablative-genitive marking for the stimulus in the verbs of fear, we can have a stimulus that comes in the form of an agent, typically formed in OCS. with the preposition otч ‘from’ + genitive NP, as shown in (15). (15)
Nч ne boi se˛ otч mokч. (OCS.: Codex Suprasliensis 8b: 7) but not fear refl. AC from tortures GN ‘But do not be afraid of suffering.’
However, in Gk., agents (typically headed by D ‘under’, cf. Coulter 2005) are not compatible with the verbs under analysis (at least, until the New Testament Greek period); the prepositions that could marginally replace accusative case on the stimulus participant were various in Ancient Greek (( + genitive, « + genitive or accusative, >« + accusative), but never D (Allan 2003: 74). Examples of this are provided in (16a-b), taken from Allan (2003): (Sophocles, Trachiniae 671) (16) a. E , > , ( Ρ φ29 . explain if explainable from what GN fear ‘Explain, if you can, what you are afraid of.’ b. #A’> φ29 μ« … (Sophocles, Trachiniae 1211) but if fear towards that AC ‘But if you are afraid of that …’ ‘Feeling fear towards her husband and children.’ (Sophocles, Oedipus Tyrannus 980) c. C] ’>« 0 μ« κ φ2 φ1 . you towards the mother not fear marriage AC ‘Do not be afraid of marrying your mother.’ To appreciate the parallelism with (15) and (16), I give examples of regular agents headed by otч in a purely passive sentence in OCS. (17a) and by D in Gk. (17b): (17) a. Krчsti se˛ otч ioana vч iordaneˇ. (OCS.: Mk. 1: 9) baptized refl. from Johann in Jordan ‘He was baptized by Johann in the Jordan river.’
164
Nerea Madariaga
(Gk.: Aeschines, Speech on the Embassy 2, 172) b. T7 "7 Dμ 7 2 2 . .. . the walls under the barbarians destroyed PASS.PART. ‘Having being destroyed the walls by the barbarians …’ 4.1.2. Correferentiality of subjects in main and subordinate clauses Gk. verbs denoting fear could be followed by an argumental subordinate clause. This could not be an infinitive clause, but was always introduced by the conjunction κ and a subjunctive form, as in (18), being the subject of the main clause and that of the subordinate clause correferent (18a) or not (18b). (Xenophon Cyropaedia 3, 1, 22) (18) a. 2, κ (…) *" %9 0. is-afraid that the last-punishment will-suffer 3SG. ‘He is afraid that he will suffer the severest punishment.’ (Xenophon Anabasis 1, 10, 9) b. 5E ¹ 6E0 « κ μ« μ « λ (…) )]« ;
feared the Greeks that advance 3PL towards the flank and them cut-down 3PL ‘The Greeks were afraid that (the enemy) would advance against their flank and cut them down.’ The Gk. verbs of fear (middle or not) cannot take infinitive subordinate clauses for independent reasons. This restriction is derived from the fact that such constructions are direct inheritors of the PIE. system where two paratactic clauses were connected by a particle – here by , initially a prohibitive particle (Smyth 1956 [1920]: 500ff).10 However, other middle verbs of perception (e.g. the verbs of refusal) can take infinitive subordinate clauses. Compare (18) to (19), where the usual way how other perception verbs introduce argumental infinitive clauses is illustrated. Here, I provide an example 10
The origin of these constructions as independent sentences is the reason why (18a) does not mean ‘He is afraid that he will not suffer the punishment’, despite the fact that the conjunction ‘may not’ is negative in Gk.; (18a) rather means ‘He is afraid that he will suffer the punishment’ (this effect is usually explained as the parataxis of two original sentences, e.g. ‘He is afraid’ + ‘May he not suffer the punishment!’).
The development of Indo-European middle-passive verbs
165
of another verb of perception with an infinitive clause, where the subject of the main clause and that of the subordinate clause are correferent (19):11 (19)
(Sophocles Philoctetes 118) M%Ω ) _ $ 0 μ » . knowing part not part refuse the AC to-accomplish INF ‘When I am informed, I will not refuse to act.’
Notice that the infinitive clause is in a case-marked position, being introduced by the accusative neuter form of the article ‘the’, as is characteristic of Greek (cf. the accusative subject of the non-finite clause in example (i) in fn. 11). OCS., on the other hand, also can add an argumental subordinate clause to a verb of fear. In case the subject of the main clause correfers with that of the subordinate clause, an infinitival clause is used, as in (20a). If the subjects of the two clauses are not correferent, the conjunction da and a finite verb are used (with an overt subject), as in (20b): (20) a. I boeˇaxo se˛ vчprositi i. (Mk. 9: 32) and fear refl to-interrogate INF him ‘And they were afraid of interrogating him.’ b. Bljudeˇte se˛ da ne kto preˇlцstitц vasч. (Mk. 13: 5) guard refl part. not someone NOM deceive 3SG. you ‘Take care in order for someone not to deceive you.’ The difference in the argumental subordinate clauses of control with the verbs of fear in OCS. vs. in Gk. follows from the structures proposed in (13) and (14): in Gk. the internal argument in middle-ac-
11
Non-correferentiality between the subject of a psych middle verb and that of its subordinate clause is not found in Gk. because it is pragmatically difficult to find a situation where the experiencer of shame or refusal is not the subject of the subordinated event (i.e. one can “refuse to do something” or “be ashamed of doing something”, but not “refuse someone to do something” or “be ashamed someone else of doing something”). Instead, non-correferentiality between the subject of a verb of declaration and that of its subordinate clause is illustrated in (i): (i) O¹ π « ) φ» ρ Ν0 ² . (Xenophon Anabasis 4, 1, 21) the guides not say to-be INF [another road] AC ‘The guides say that there is no other way.’
166
Nerea Madariaga
tive constructions is the object. This is a case position in the Gk. constructions represented in (14), so any NP located there surfaces as an accusative object. Because of that, we can have an accusative infinitive clause in that position, as shown in (19), or a finite subordinate clause, as shown in (18). In OCS., however, being the constructions represented in (13) really passive, the internal argument position is not a case position (the internal argument raises and ends up as a nominative subject), so that it remains free to be filled with an infinitive clause which in Slavic, as is usual cross-linguistically (with some exceptions like in Gk.), cannot be in a case position. The contrast between the OCS. and Gk. structures triggered by the verbs under analysis is the reason why the OCS. verbs of fear need always to use a finite subordinate clause in cases of non-correferentiality between the subjects (20b), contrasting with Gk. (19). 4.1.3. Passive participles If the Gk. structure in (14) is correct and it is active, we expect to be able to derive from them passive verbal forms (passive participles, for instance), with the “middle” semantic value they usually have. On the contrary, if the OCS. counterparts in (13) are already passive, no additional passivization is expected. This prediction is borne out:12 in Gk., passive participles can be formed from the middle verbs under analysis, e.g. $ ‘to refuse’ > $ « ‘refusing’, and examples in (21), while in OCS. such formations are impossible.13 (21) a. M0 κ λ ’> «. (Gk. Homer Odyssey 3, 96) not something me be-ashamed PASS.PART. ‘Not respecting me (lit. feeling ashamed because of me).’
12
13
The data on word formation have been taken from the Perseus internet page, from Liddell & Scott (1883 [1968]) and Cejtlin, Vecˇerka & Blagova (1999 [1994]). Notice that I provide examples referred to “middle-passive only” verbs, that is, to verbs that lack active counterparts, in order to avoid the possibility to find passive participles that are in fact formed from the correspondent active form: Gk. > ‘to be ashamed’ and $ ‘to refuse’; OCS. bojati se˛ ‘to be afraid’. The same observation applies to section 4.1.4.
The development of Indo-European middle-passive verbs
167
b. #A 1 * . (Gk. Plutarch Parallel lives 2, 58a) deny PASS.PART praises ‘Negative (lit. denying) praises.’ In OCS., no passive participles can be formed from passive-reflexive verbs, e.g. (u)bojati se˛ ‘to be afraid’ does not form a passive participle like *(u)bojanч (past) or *(u)boimч (present). Only active forms of the participle + the reflexive se˛ are possible: (22)
Dalч esi boje˛sˇtiimч se˛ tebe znamenцe. (Psalm 59: 6) give aux. be-afraid ACT.PART.PRES. refl you GN standard ‘You gave the banner to those who were afraid of you.’ 4.1.4. Passivizing adjectival formations
The final argument I will present in this section involves facts dealing with word formation (inspired in Keyser & Roeper 1984). In order to distinguish external arguments (undergoers) from internal arguments, I will check “passivizing” word formation in the Gk. and OCS. verbs under analysis. If the Gk. structure (14) is correct, and the stimulus NPs of the psych verbs of separation in Gk. are internal arguments, then we expect to be able to form passivizing adjectives with the help of the suffixes -« or - « (comparable to English -able), as we do with . ‘to burn’ > 1 « ‘combustible’ (“burnable”). This prediction is borne out, and we can form the Gk. adjectives in (23): (23) a. A> ‘to be ashamed’ > > « / > « ‘respectable’ b. #A ‘to deny, to refuse’ > $ « ‘deniable’ In OCS., on the contrary, such formations are not possible, which is expected from the fact that the internal argument NP in these constructions is not the stimulus (but the experiencer), and that the construction cannot be passivized further, as it is already passive. In OCS., active verbs form passivized adjectives (equivalent to English -able) with the suffix -цnч or -tчjц, as in pecˇaliti ‘to sadden’ > pecˇalцnч ‘sad, pitiful’, and piti ‘to drink’ > pitii ‘drinking (water)’. But reflexive-passive verbs of separation do not: e.g. (u)bojati se˛ ‘to be afraid’ > *(u)boeˇnч / *(u)bojatii.
168
Nerea Madariaga
4.2. The position of se˛ in (13) Until now, I have given arguments for the position of the arguments in structures (13) and (14). Now let us analyze the se˛ element in the OCS. structure (13). I will argue that this element has case features, but not gender / number features, and that it is in a high position in the structure. 4.2.1. Se˛ has case features but not gender / number features First of all, let us show that se˛ has case features, but not gender / number features. The form se˛ is the accusative form of the reflexive enclitic (atonic) pronoun; it has overt non-default morphological case, therefore. In addition, it does not display gender / number features, as it remains invariant without regard to the person and number encoded in the verbal form, as shown in (24); (24a) illustrates 2nd person plural verbal agreement and (24b), 1st person singular. The reflexive pronoun se˛ remains invariable: (24) a. Azч esmц ne boite se˛. (Mt, 14: 27) I am not are-afraid 2PL refl. ‘It is me, do not be afraid.’ b. Uzˇe ne strasˇo˛ se˛ protivnaago. (Codex Suprasliensis 490: 9) already not fear 1SG. refl. enemy GN ‘I am nor afraid of the enemy any more.’ 4.2.2. Se˛ is higher than regular reflexive pronouns The se˛ in (13) is an enclitic pronoun, homophonous with the regular reflexive pronoun of 3rd person that can be placed in any argumental position. However, these two homophonous se˛ seem to occupy different positions in the structure; the two se˛ do not compete for the same position in the structure as they can co-occur (25). (25)
(Mt. 12: 25) Vчseˇkч gradч li domч razdeˇlч se˛ na se˛ ne stanetч. all city or house divided refl. against refl AC not stand ‘All city or house divided against itself will not stand.’
The development of Indo-European middle-passive verbs
169
Observe the position of the two different reflexive elements in the previous example. The enclitic se˛ that forms passive-reflexive constructions is higher, following the main verb, while the other reflexive pronoun is placed in the case position that would occupy if it were a regular NP, instead of a pronoun; namely, the second se˛ is the complement position of the preposition na, which requires accusative case. 4.2.3. Se˛ can be a second-position element Besides the immediate position following the verb, another possible position for this se˛ in OCS. was second position in the sentence, as shown in (26). (26)
Asˇte se˛ bi ne rodilч cˇ[eloveˇ]kч tч. if refl was not born man that ‘If that man had not been born.’
(Mt. 26: 24)
This is, according to Vaillant (1965 [1924]), an old pattern and the one that will prevail in later (Old and New) Bulgarian (Pancheva 2005).14 This observation suggests that this element is really placed high in the structure, like Romance se (cf. Raposo & Uriagereka 1996), confirming structure (13).
4.2.4. Coordination Another piece of evidence for a high position of the se˛ in OCS. comes from coordination. Vaillant (2002 [1948]) observes that it was possible in OCS. to coordinate two passive-reflexive verbs and have a single se˛ for both verbs, as shown in (27): (27) a. Vчveseliti zˇe se˛ i vчzdradovati. cheer part. refl. and be-happy ‘To cheer up and be happy.’ 14
(Lk. 15: 32)
Pancheva (2005) analyses the variation in the position of the clitics in OCS. and later Bulgarian. In general, it must be said that in some Slavic languages (e.g. Russian), the post-verbal position of the reflexive prevailed (the reflexive became the verbal suffix -sja/-s’, as in bojat’sja ‘to be afraid’). In other Slavic languages (e.g. Czech), the second-position pattern has prevailed (the reflexive form became the enclitic se always in second position in the sentence).
170
Nerea Madariaga
b. Da ne sчmosˇtaetч se˛ srчdцce vasˇe ni ustrasˇaetч.(Jn. 14: 27) let not embarrass refl heart our not fear ‘Let your heart not become ashamed or afraid.’ All the verbs in examples (27a-b) require the presence of the reflexive se˛ in order to be correctly interpreted as reflexive-passives, but the required element appears only once in each example, i.e. following the first coordinated verb. This suggests structure (28) for example (27a), which follows from the basic structure proposed in (13): (28)
VP se˛ AC
(OCS.: Coordinated reflexive-passive verbs = 27a) V’
V1
V2
vчveseliti vчzdradovati
4.3. Verbal morphology in Gk. To conclude this section, I will mention a morphological fact, which shows that Gk. middle constructions, as the one represented in (14) are really active structures and as such, they differed from other intransitive middles. Allan (2003) detects a correlation between the meaning and the verbal desinences related to the middle-passive Gk. verbs. Recall Allan’s (2003) semantic classification of middle constructions in Gk. given in section 2.1: passives > spontaneous / mental processes > motion > mental activities > reciprocal > reflexives. In Classical Greek, these constructions were divided into two groups according to their aorist desinences: sigmatic vs. aorist in -%0-. The distribution in Homer’s time was the following: passives and spontaneous processes were exclusively marked with -%0- aorist; on the contrary, mental activities, reciprocals and reflexives had exclusively sigmatic aorists.15 15
The distribution in Classical Greek was slightly different: passives, spontaneous processes and motions were marked with -%0- aorist, while mental activities, reciprocals and reflexives had sigmatic aorists, i.e. the -%0- aorist extended sporadically even to mental activities and reciprocals in Classical Greek, but never to reflexives.
The development of Indo-European middle-passive verbs
(29)
171
Homer’s Greek > -`H- AORIST SIGMATIC AORIST Reflexives, Mental activities, Reciprocals > Processes, Motion, Passive
This distribution shows that middle-passive constructions with two participants (reflexives, reciprocals and activities), at least originally, displayed a sigmatic morphological pattern in the aorist, while constructions with only one participant (processes, motion and passives) displayed a -%0- aorist pattern. In fact, the sigmatic aorist is often called “middle”, while the -%0- aorist is called “passive”. But this distribution is not exact: psych middle-active verbs of “separation” are not passive but rather transitives, as we are showing in this paper, but could form -%0- aorists in certain cases. More specifically, these verbs formed a sigmatic aorist when followed by a direct object (overt or elided), but could form a -%0- aorist in the absence of a direct object. This is shown in examples (30) vs. (31): the verbs > ‘to be ashamed’ and $ ‘to refuse’ must display a sigmatic aorist when they have an object, overt (30a) and elided (30b), but they can also surface in the -%0- aorist form in purely intransitive uses, without a direct object, as in the first part of (31): (Homer Odyssey 21, 28) (30) a. O) ξ %7 4 9 ’) ξ Q . not gods’ wrath AC is-ashamed not table AC “He had not regard for the wrath of the gods or for the table” (Herodotus Histories 3, 1, 2) b. LO ξ 5A « 9 1 7 P . $"% « (…) ) ρ" Κ Κ $ %. the prt. Amasis the power of-the Persians intimidated no could not to-give not to-refuse ‘Amasis, intimidated by the power of Persia, could neither give nor refuse (his daughter).’ (Homer Iliad 7, 92) (31) (P «) c % κ $ %, , ’D "%. (all) were-ashamed part. to-deny feared part. meet ‘They all were ashamed to deny him, and afraid to meet him.’
172
Nerea Madariaga
5. Case marking of objects of middle-passive verbs Once established that the structures under analysis are different constructions (active in Gk. / reflexive-passive in OCS.), let us account for the differential case marking of their objects: accusative in Gk. vs. ablative-genitive case in OCS. Recall examples (5a) and (10a), repeated below for convenience: (5a)
Ne strasˇo˛ se˛ protivnaago. not fear ACT refl. enemy GN
(Codex Suprasliensis 490, 9)
(Homer Odyssey 16, 446) (10a) O) ξ % % Ν .. not something him death AC fear MID bid ‘I bid him not be afraid of death.’ The verbs under analysis in this paper, verbs of “separation” or “departure” and related notions (e.g. fear, refusal, shame) were characterized in PIE. by having an ablative-genitive complement associated, as reconstructed by Lehmann (1974), Savcˇenko (2003 [1974]), Schmalstieg (1983). Some IE. languages (Latin, Sanskrit) conserved this ablative case as such, as shown in (32); in other languages (Greek, Baltic and Slavic), the ablative uses merged with the ones of the genitive case, giving rise to an ablative-genitive case, as shown in (33). The examples are from Savcˇenko (2003 [1974]: 347), and illustrate the ablative-genitive case of departure in a physical sense: (Latin: Caesar Commentaries on the Gallic War) (32) Suis finibus eos prohibent. [own borders]ABL them prevent ‘They restrain them from their border.’ (33) a. #A , ξ … 7 1 . (Gk.: Homer Iliad 179) Argons part. shipsGN were-prevented ‘And the Argons were expelled from the ships.’ b. Se otxozˇju sveˇta sego. (Old Russian: Laurentian Chronicle) prt. leave [world this]GN ‘Now I am leaving this world.’ The notion of separation could be also understood in a metaphoric sense; thus, the psych verbs denoting fear, refusal and shame also fall
The development of Indo-European middle-passive verbs
173
under this classification (‘to be afraid’, ‘to avoid’, ‘to distrust’, etc) and therefore, took ablative-genitive case-marked objects in Sanskrit, Avestan, Baltic, and Slavic (Meillet 1965 [1924], Borkovskij 1978, Schmalstieg 1983, Luraghi 2003). The following examples are from Sanskrit (34a) and Baltic (34b), also from Savcˇenko (2003 [1974]): (Sanskrit, Rig-veda) (34) a. Vís´vam bibha¯ya bhúvanam maha¯vadha¯t. whole fears world of-the-mighty-weapon ABL ‘The whole world is afraid of that with the mighty weapon.’ b. Vaika dazˇna bìjo tamsõs. (Lithuanian) children often fear darkness GN ‘Children are often afraid of darkness.’ In some IE. languages, like Greek and Latin, the semantic notion of “departure” was lost in the psych verbs (the ones with the metaphoric sense of separation), and the original ablative-genitive objects were reanalyzed as regular direct objects (and marked with accusative case). As said before, this is the pattern we find in the Gk. verbs of fear, shame and refusal from the Homeric times. The change from the PIE. ablative-genitive pattern into an accusative pattern in the Gk. verbs of fear was made according to a process that took place at different stages in various IE. languages, and which consisted in replacing lexical bare cases (“adverbial” cases) with structural ones.16 Sporadically, instead of a structural accusative case, an overt PP could be inserted instead (Borkovskij 1978, Luraghi 2003); recall examples in (3), repeated below for convenience: (Gk.: Sophocles, Trachiniae 671) (3) a. E , > , ( Ρ φ29 . explain if explainable from what GN fear ‘Explain, if you can, what you are afraid of.’ 16
For the purpose of this paper, I will differentiate between two general types of grammatical case: (i) lexical (adverbial) case, which is licensed on a NP by an element in the structure (usually a verb or a preposition), which determines the semantic role of that NP, too; (ii) structural (configurational) case, which surfaces on a NP not because of another element that determines its semantic role, but because the NP occupies a certain position in a certain configuration. The cases usually assumed to be structure-dependent are nominative, accusative, and sometimes genitive case.
174
Nerea Madariaga
b. Nч ne boi se˛ otч mokч. ˛ (OCS.: Codex Suprasliensis 8b: 7) but not fear refl. AC from tortures GN ‘But do not be afraid of suffering.’ As we cannot track this change in documented Gk. (the earliest appearances of the psych verbs of separation are already associated to an accusative object), let us see what the change looked like in a welldocumented IE. language, in Old Russian. This change process is analyzed in detail in Madariaga (2009) for the OR. verbs denoting “taking care”, which changed from a genitive lexical pattern to a structural accusative one. (11th century OR.: Anthology of 1076, 15) (35) a. Pacˇe zˇe bljudi slovesч jego. moreover part. guard words GN his ‘And over all, observe his words.’ (17th c.: Life of the Archpriest Avvakum, 246) ˇ ˇ b. Cistotoju devstvo sobljula. with-cleanliness virginity AC guard ‘She took care of her virginity by remaining chaste.’ The change that took place between (35a) and (35b), described in Madariaga (2009), was the following: the Old Russian system of grammatical case was transformed as a result of previous independent changes; basically, the old system of IE. bare lexical cases (initially, “adverbial” postpositional morphemes, according to Meillet 1964 [1906], and Savcˇenko 2003 [1974]), which denoted the specific semantic relation between verbs and NPs in a sentence was lost (as in most late IE. languages, cf. Savcˇenko 2003 [1974]). The bare lexical “adverbial” cases of early IE. were reinterpreted in late Old Russian either as structural (usually surfaced as accusative case) or inserted in a PP headed by an overt P (Borkovskij 1978, Lomtev 1954). Consider example (36), taken from Cˇernyx (1952: 269–70); which illustrates the regular replacement of bare genitive complements in verbs denoting “separation” with PPs headed by the overt P ot / iz ‘from’. (Note that example 36b is the equivalent of 36a, but taken from a later copy of the 1st Novgorod Chronicle): (13–14th century OR: 1st Novgorod Chronicle, 36) (36) a. Otstupi voleju Kyeva. left by-will Kiev GN
The development of Indo-European middle-passive verbs
175
(15th c.: 1st Novgorod Chronicle Com., 112b) b. Otstupi voleju is Kyeva. left by-will from Kiev GN ‘(Mstislav) moved away from Kiev of his own free will.’ In the case of the Old Russian verbs of ‘taking care’ in (35), the semantic relation (“separation”) existing between the verb and the related NP disappeared and the verbs started to take regular accusative objects. Something similar could happen with the psych verbs of fear in Greek analyzed in this paper. The crucial point is that, in Gk., this development from bare ablative-genitive into accusative case marking of the verbs of psychological ‘separation’ was possible thanks to the special structure of the active-middle verbs represented in (14), which did not violate Burzio’s (1986) Generalization, as it was active. On the contrary, in Slavic, the change from genitive to accusative in the reflexive-passive verbs of ‘separation’, represented in (13), did not take place because the NP denoting the one that experiences fear or shame is the internal argument, and only later it raises to become the nominative subject of the passive sentence. The reflexive pronoun, which is provided with case features, is marked with the following case available, accusative case, so that further accusative case valuing is blocked, and the stimulus of the fear or shame must continue surfacing as an adjunct with the lexical case that characterized it in PIE. (ablative-genitive), or as a PP (cf. ex 3b). Only much later, when the passive marker (the OCS. reflexive pronoun se˛) wore out phonetically and became a verbal suffix was accusative case valuing allowed on the stimulus. This happened as late as in Present-day Colloquial Russian, where the originally accusative reflexive pronoun form became the verbal passive morpheme -sja/ -s’ (< se˛). As a result, the new suffix could be reinterpreted as a middledeponent morpheme, similar to the middle desinences of the Gk. active constructions analyzed in this paper. Thus, besides a normative (conservative) paradigm where the NP associated to a -sja/ -s’ verb of fear is still the genitive inherited from IE. (37a), we have a colloquial pattern, where an accusative direct object is preferred (37b): (37) a. Ja bojus’ teti. I fear aunt GN ‘I am afraid of my aunt.’
(Normative Present-day Russian)
176
Nerea Madariaga
b. Ja bojus’ tetju. I fear aunt AC ‘I am afraid of my aunt.’
(Colloquial Present-day Russian)
Instead of the OCS. structure (13), the Colloquial Present-day Russian in (37b) can be represented as something similar to the Gk. structure (14) proposed here; as an active construction, with no accusative pronoun, it observes Burzio’s Generalization and licenses accusative case on the stimulus NP (Madariaga 2009).
6. Conclusion In this paper, I have shown that the different case marking on the objects of the Indo-European middle-passive verbs with two arguments (the psych verbs denoting separation) corresponds to two different structural and morphological patterns. The case studies presented in this paper illustrate the two patterns with the help of Ancient Greek and Old Church Slavonic, which represent different stages in the evolution of these verbs from the initial PIE. intransitive / stative pattern. Applying various mopho-syntactic tests (the position of the NP stimulus / experiencers, their semantic relations, correferentiality of subjects in main and subordinate clauses, passivizing participal and adjectival formations, the properties of se˛, and verbal morphology), I have shown that the Gk. psych verbs under study displayed an active-middle pattern (similar to Latin deponents), while their counterparts in OCS. displayed a reflexive passive pattern (similar to most modern Romance languages).
References Alboiu, Gabriela. 2007. Null Expletives and Case: the view from Romance. In: Pascual Masullo (ed). Romance Languages: Structure, interfaces, and microparametric variation. John Benjamins. Allan, R. 2003. The Middle Voice in Ancient Greek: a study of polysemy. Amsterdam: Gieben. Bhatt, Rajesh & Roumyana Pancheva. 2005. Implicit Arguments. The Blackwell Companion to Syntax v. II: Blackwell, 554–584. Borkovskij, V. I. 1978. Istoricˇeskaja grammatika russkogo jazyka: sintaksis, Nauka, Moskva.
The development of Indo-European middle-passive verbs
177
Burzio, L. 1986. Italian Syntax: A government-binding approach, Reidel, Dordrecht. Cejtlin, R. M., R. Vecˇerka, E. Blagova et alii. 1999 [1994]. Staroslavjanskij slovar’ (po rukopisjam X–XI vekov). Moskva: Russkij Jazyk. Cˇernyx, P. Ja. 1952. Istoricˇeskaja grammatika russkogo jazyka. Moskva: Ucˇpedgiz. Cinque, Guglielmo. 1988. On Si Constructions and the Theory of Arb. Linguistic Inquiry 19: 521–582. Clackson, James. 2007. Indo-European Linguistics: an introduction. Cambridge University Press. Comrie, Bernard. 1998. The Indo-European Linguistic Family: Genetic and Typological Perspectives. In: P. Ramat & A. Giacalone, eds. The Indo-European languages. Routledge. Coulter, George. 2005. Expressions of Agency in Ancient Greek. Cambridge Un. Press. Croft, William. 1993. Case Marking and the Semantics of Mental Verbs. In Pustejovsky (ed.) Semantics and the Lexicon. Kluwer: 55–72. Croft, William, Hava Bat-Zeev Shyldkrot & Suzanne Kemmer. 1987. Diachronic Semantic Processes in Middle Voice. In Anna Giacolone Ramat, Onofrio Carruba & Giuliano Bernini (eds.) Papers from the VIIth International Conference on Historical Linguistics. John Benjamins: 179–92. Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67: 547–619. Gerritsen, Nelleke. 1992. The Invariant Meaning of the Russian Reflexive Affix -sja. Acta Linguistica Hafniensis 25: 29–37. Jasanoff, J. 1978. Stative and Middle in Indo-European. Innsbrucker Beiträge zur Sprachwissenschaft. Haspelmath, Martin. 1987. Transitivity alternations of the anticausative type. Köln: Institut für Sprachwissenschaft. Keyser, Samuel Jay & Thomas Roeper. 1984. On the Middle and Ergative Constructions in English. Linguistic Inquiry 15/3: 381–416. Lehmann, Winfred P. 1974. Proto-Indo-European Syntax. Heidelberg: Winter. Liddell, Henry G. & Robert Scott. 1883 [1968]. A Greek-English Lexicon. Oxford: Clarendon Press. Lomtev, T. P. 1954. Iz istorii sintaksisa russkogo jazyka, Moskva. Luraghi, S. 2003. On the Meaning of Prepositions and Cases. John Benjamins. Madariaga, Nerea. 2009. The Triggering of Structural Case-marked Objects in Old and Present-day Russian. In: S. Birzer, M. Finkelstein & I. Mendoza Proceedings of the Second International Perspectives on Slavistics Conference (Die Welt der Slaven, series Sammelbände-Sborniki 36). Otto Sagner, Munich, 87–99. Meier-Brügger, Michael. 2003. Indo-European Linguistics. Walter de Gruyter. Berlin – New York. Meillet, Antoine. 1964 [1906]. Introduction à l’étude comparée des langues indoeuropéennes, University of Alabama Press. Meillet, Antoine. 1965 [1924]. Le Slave Commun, Paris: Librairie José Corti. [Translation to Russian 2001 Obsˇcˇeslavjanskij jazyk, URSS: Moskva]. Pancheva, Roumyana. 2005. The Rise and Fall of Second-Position Clitics. Natural Language and Linguistic Theory 23 / 1: 103–167.
178
Nerea Madariaga
Raposo, Eduardo & Juan Uriagereka. 1996. Indefinite se. Natural Language and Linguistic Theory 14/4: 749–810. Rix, Helmut. 1988. The Proto-Indo-European Middle: Content, Forms and Origin. Münchener Studien zur Sprachwissenschaft 49: 101–119. Savcˇenko, A. N. 2003 [1974]. Sravnitel’naja grammatika indoevropejskix jazykov. URSS: Moskva. Schmalstieg, W. R. 1983. An Etymology of the word ‘to fear’ in Indic, Baltic and Slavic. Lithuanus 29/3: 60–65. Smyth, Herbert Weir. 1956 [1920]. Greek Grammar. Harvard University Press. In Perseus web-site: www.perseus.tufts.edu/cgi-bin/ptext?doc=Perseus%3 Atext%3A1999.04.0007 Vaillant, André. 1965 [1924] Le Slave Commun. Paris: Librairie Honoré Champion. Vaillant, André. 2002 [1948]. Manuel du vieux slave. Paris. [Russian translation: 2002 Rukovodstvo po staroslovjanskomu jazyku. Moskva: URSS]. ˇ eskij slovar’ russkogo jazyka [Russian Vasmer, M. 2003 [1950–58]. Etimologic fi translation of Russisches Etymologisches Wörterbuch]. Moskva: Astrel’.
University of the Basque Country Facultad de Letras Paseo de la Universidad 5 V i t o r i a - G a s t e i z 01006 Spanien [email protected]
Nerea Madariaga
Iberian *ortubeles´ and ordumeles
179
Iberian *ortubeles´ and ordumeles, Palaeo-Sardinian Ortumele. New horizons in substrata research. Abstract Research on the primitive substratum of Sardinia, the Palaeo-Sardinian, had long come to an impasse, before structural and typological devices superseded the old historical-comparative method. The typological approach has shed light on the agglutinative organization of toponymic compounds, and at the same time highlighted striking ressemblances with Palaeo-Basque morphological and lexical material and with Iberian records. In this paper we contend that a careful examination of structural and semantic properties of some Palaeo-Sardinian microtoponyms related to the colour of mountain’s streams and grounds allow us to decipher one Iberian compound, and set up a new incisive line of cross-linguistic research between Old Iberia and Neolithic Sardinia.
1. Iberian International research on Iberian achieved a remarkable progress as Gómez Moreno deciphered the mysterious syllabic structure of that Pre-Indoeuropean language1, but the dramatic breakthrough took place when Jürgen Untermann, Javier de Hoz and other scholars began to apply structural devices to the corpus of inscriptions, thus managing to infer linguistic typology, morphosyntactic patterns and nearly a hundred lexical roots2. Still, further advances in semantic interpretations remained blocked by the lack of cross-linguistic correlations with other Mediterranean Indoeuropean languages. Particularly distressing, moreover, was the assessment that Basque could not help to decipher the almost 170 Iberian inscriptions, and that the scanty cognates displayed by the two ancient languages of the Iberian Peninsula were due to centuries of enduring contact (Sprachbund)3. Among the most persuasive correlations between Iberian and Palaeo-
1 2
3
Cf. Gómez Moreno (1949). Cf. Siles (1985), Untermann (1990, 2004), Velaza Frías (1991, 1996), Silgo Gauche (1994), de Hoz (2001 with foregoing bibliography). Cf. Tovar (1997:47–145), Gorrochategui (2002).
180
Eduardo Blasco Ferrer
Basque one must mention the adjectival morpheme for the colour ‘black’, attested as beles´ or beles4, and also in Latin transcriptions as meles5, as bele(x) in Aquitanian old anthroponyms6, and as bel(e) and beltz in Basque lexical terms and toponymic designations7. In the bronze inscription of Ascoli (Italy), dated 89 a.C., this morpheme occurs several times within the primary name, typically an epithet, used for Iberian soldiers in the Roman army. We select out of this inscription the sequence ordumeles, which neatly reflects the compound ortu(n) ‘?’ + beles´ ‘dark’, and will attempt an acceptable interpretation by means of external comparison with some Palaeo-Sardinian toponyms, which we presume to be formally and semantically tightly bound.
2. Palaeo-Sardinian Placed in umbilico maris Mediterranei – as the 16th century’s littérateur Giovanni Francesco Fara pregnantly put it – the island of Sardinia has long concealed the secret of its primitive Indoeuropean language. Italian scholars of the first half of the last century8 launched a thorough exploration of possible analogies within the whole Mediterranean, from Iberia to Palestine, and with their historical-comparative method managed to isolate a reduced set of lexical roots, still assumed to have been part of a primeval Mediterranean substratum9. Also the Swiss scholar Johannes Hubschmid and the Bavarian linguist Max Leopold Wagner, the “Meister” of Sardinian linguistics,10 remained wedded to the principles of a method which assigned total priority to the “casual homonimy” of forms, but that failed to unearth the very 4 5 6 7 8
9
10
Cf. Untermann (1990:216–217; 1998.:76). Cf. Velaza Frías (1996:34). Cf. Cf. Gorrochategui (1984:158–159). Cf. Michelena (1985:128,222). From Francesco Ribezzo, Alfredo Trombetti, Vittorio Bertoldi, Giovanni Alessio, Giandomenico Serra, Benvenuto Terracini to Carlo Battisti and Giacomo Devoto. For all them see the critical balance of Silvestri (1979–82) and Blasco Ferrer (2002: Introduction). Pellegrini (1994) has listed them (*carra ‘stone’, *mara ‘mare’, *sala ‘running water’, *tala ‘mountain’s stream’, among others). Cf. Hubschmid (1953, 1963), Wagner (1951, 1960–64).
Iberian *ortubeles´ and ordumeles
181
functioning of structures11. The situation has rapidly changed since two of the most archaic areas of the island have been exhaustively investigated, particularly in their toponymic configuration12. In 1993 we showed that a modern structural analysis of microtoponyms, with a thorough distributional and frequential examination of their segmental morphs, offered a syntactic account of the subjacent organization of the Pre-Indoeuropean language of central and eastern Sardinia. Thus, observing the archaic lexeme orgosa and the toponym Orgosa (and Orgòsolo), ‘well, river’s mouth’, we easily recognize two compounding roots, each generating further derivative forms: orga and orge [γ] ‘well, wet ground’, Orga, Orge, Org-ai, Org-ei, Org-oi, Org-osa, Org-ose, Orgol-ai, Orgose-goro, Orgos-ol-ai, and Osa (and Bosa, with prosthetic consonant), Osu, Os-oe, Flumen-d-osa, Os-alla, Ós-ana, Os-ol-ai, Ósolo, all terms for ‘wells, rivers, rivers’ mouths, wet terrains’. Now, one can readily deduce that Pre-IE. orga has been “glossed” by the second root, *osa, this connected with *aus-a, o¯sa, a productive Peri-Indoeuropean (Giacomo Devoto) or – less persuasively in Sardinia – PalaeoIndoeuropean (Hans Krahe) root13. Our tautologic toponym shares its complex structure with a handful of similar compounds, scattered all over the European continent: Vall d’Aran, Chateau-dun, Linguaglossa, Mon-gibello. Basing our research on further structural and typological examination of the microtoponyms of the central and eastern counties of Sardinia we managed to gather in recent years a vast amount of evidence about the real organization and the origin of the Palaeo-Sardinian language. For its “agglutinative” typology we may adduce a short array of selected examples, all microtoponyms: (1)
(2)
11 12 13
ard-ai, ili-ai, iri-ai, istil-ai, gurri-ai, nur-ai, ol-ai, on-ai, org-ai, ort-ai, sun-ai, tal-ai, turr-ai, ur-ai; aran-ake, ili-ake, iri-ake, nurake, tal-ake, ur-ake, berr-age; ili-ana, tal-ana, ós-ana, óv-ana, ártz-ana; árd-ara; ol-eri, tal-eri; sun-ele, turr-ele. os-ol-ai, ort-ol-ai, ov-ol-ai, artz-ol-ai; bid-on-i, mand-on-i, olon-é, ós-on-o; tale-turre, tala-suni-ai, berri-tal-ai, berru-nur-ai; ort-aran-i; orgo-r-isti, bid-istil-i.
A substantial evaluation can be read in Craddock (1969). Cf. Blasco Ferrer (1988), Wolf (1998). For Devoto’s Peri-Indoeuropean/mediterranean *aus-a and Krahe’s PalaeoIndoeuropean *au-sa see the balanced summary of Silvestri (1985–86).
182 (3)
Eduardo Blasco Ferrer
arau-nele (< aran), bidu-nele (< bide), istiu-nele (< istil), turrunele (< turri); org-ose-goro (< orga), turri-koro, talae-kore; talerthe (< tala).
Subset (1) shows a recursive use of suffixes (here -ai, -ake/-age, -ana, -ele, -eri, but all stressed/unstressed vowels and almost all kinds of diphthongs are documented) bound to a restricted set of roots: ard-, artz-, berri, bid-, ili, iri, isti(l), gurr-, mand-, nur, ol-, on, org-, ort-, os-, ov-, sun-. tal-, turr-, ur. Subset (2) shows the pertinent capacity of an agglutinative language to make compounds by simply adding roots and suffixes (with derivative and flexional marks): os- + ol-, ort- + ol-, ov- + ol-, artz- + ol-; bid- + on, ol- + on, os- + on, mand- + on; berr- + tal-, berr- + nur; tal- + turr-, tal- + sun-. Subset (3) is the most relevant for us, because it illustrates the faculty of deriving a suffix from a root, an outstanding feature of agglutinative languages: -gor/kor (+ paragogic vowel) is an offshot of Basque gorri ‘red’, a formal development well attested in Iberia and Aquitaine14, -erthe reflects neatly the bsq. root ertze ‘wedge, edge’, which neatly matches the denotata it specifies (tala + erthe = ‘edge of a mountain’s river’), and nele, at last, which is the adjectival unit for ‘dark’, on which we shall next focus our attention. The typical organization of Palaeo-Sardinian coincides entirely with that of Basque. The major part of the extracted lexical roots have an exact correspondence in present Basque, and in some cases in reconstructed Basque and Iberian: ardi ‘ow’ (and cf. ardiule, artule ‘wool’ and sd. Ardule, Ardauli in Sardinia), artz(ai) ‘shepherd’ (and cf. Artzu, cognomen, and Ártzana, a mountain’s town), aran ‘valley’, berri ‘new’, Iberian ili and Basque iri ‘settlement’, istil ‘marsh’, gorr(i) ‘red’ (and gurri: Gurriaran), mando ‘equus asinus’, ola ‘primitive hut’, obi (> sd. ov-) ‘cave’, (i)turri ‘spring, well’, ur ‘water’. Some other roots are still opaque, and we can not dwell on their structures and possible correlations in this paper, but only stress the fact that they can not be dismissed out of hand as potential Basque or Iberian units (so *tala ‘mountain’s river’, *nur, well represented in nurake ‘megalithic monument’, for us a plural form of ‘stone’).
14
Cf. Coromines (1981, I:111): “Que -corr soit le Basque gorri ‘rouge’, cela va sans difficulté’; Orpustan (1991:128,160).
Iberian *ortubeles´ and ordumeles
183
From a morphophonemic point of view it seems important to point out that in Palaeo-Sardinian compounds the final vowel of the first unit is subject to many qualitative variations (raising of mid-vowels, dissimilations), as in berri + nur- > berru-. In derivatives (1) dropping of final vowel is the general rule (tala + -eri > taleri), but ending -i turns out to be more resistent (ili + -ai > iliai against tala + -ai > tal-ai). Before going on to interpret the Iberian compound *ortubeles´ with the support of Palaeosd. roots, let us remind briefly that recent archaeological, historical and even genetic contributions15 all point unequivocally to one or more migrations from the neolithic Iberia to the island of Sardinia. We have contended in a recent History of the Sardinian language16 that the primitive inhabitants of neolithic Sardinia, the Ilienses and the Bàlari (with bal-, that ties narrowly Sardinia with the Balears), represent the result of Iberian and Palaeo-Basque colonization in search of obsidian, a ‘dark vulcanic mineral’ which we find exclusively all along the eastern coast of the south of France and Catalonia.
3. Cognitive semantics and reconstruction of *ortubeles´ Along with structural and typological approaches to isolated languages of the Mediterranean (Etruscan, Iberian, Minoic)17, cognitive semantics has played a relevant role in recent topics for active debate about diachronic reconstruction18. A vast amount of toponymic material widespread over Europe and Asia has clearly revealed that ‘black’ and ‘red’ were (and are!) usual adjectival qualities used to encode a rich variation of cromatic nuances of ‘earth’ and ‘water’ qualities, and that colour’s denotations were commonly described drawing on metaphoric and metonymic transferring (‘blood’, ‘fire’ > ‘red colour’)19. We undertake now a thorough discussion of the sem-
15
16 17
18 19
Cf. Rowland (2001:13,31), Mastino (2005, chapter I and II), Francalacci (2003 with foregoing literature). Cf. Blasco Ferrer (2009); further detailed eaminations in Blasco Ferrer i.p. Cf. Agostiniani (1993, 2000) for Etruscan, Velaza Frías (1996) for Iberian, Facchetti (2002) for Minoic. Cf. Taylor (1995), Geeraerts (1997), Blank/Koch (1999). Cf. Bach (1953, 272–299), Rohlfs (1960), Tischler (1977, 155–161).
184
Eduardo Blasco Ferrer
antic values of *ortu based on this assumption. The following list exhaustively records all place-names of central and eastern Sardinia which are allegedly derived or compounded with the root *ortu: ort-ai, ort-ana, ort-aran-i, ort-ei, ort-eni, ort-iai, ort-il-o, ort-il-á, ortil-ai, ort-ol-ai, ort-ol-o, ort-or-ai, orto-koro, (baxu < opa¯cus ‘cleft rocky ground, gorge where water runs’) ort-orgo, ort-os-ai, ort-ui, ortu-ene, ortu-eri, ortu-eria, ortu-mele. Particularly relevant for our digressions are the four compounds: Ort-aran-i20, Ort-orgo, and above all Orto-koro and Ortu-mele, non official toponyms (lieux-dits) of Nuorese (Bono) and High Ogliastra (Urzulei). We begin the discussion by examining the recorded compounds with -koro and -nele, and trying to ascertain their semantic fields. 3.1. -koro As we already mentioned, [’kɔre,-o] is an allomorph of the PalaeoBasque root *gor- (bsq. gorri ‘red’), used in nominal syntagmas to denote waters (Iturrigorri ‘red spring’) or qualities of ground (Mendigorri ‘red mountain’, Gurriaran ‘red valley’). Indeed gorri is also documented in Sardinia, once in an exactly equivalent hybrid compound, Funtana gorru (lat. fontana, from fons, fontis ‘spring’), as derivative in another oronymic designation, Nodu gurr-ai (lat. no¯dum ‘peak of rocky mountain’), and in further more complex toponyms: gurri-thókinu, gorro-ispa, gorr-osp-ai. Much more productive seems to be the suffix -gor,kor + Vowel: orgose-goro, orgose-kore (with orgose ‘well, spring’), ili-kore (Iberian ili ‘settlement’), iste-kor-í (bsq. istil ‘marsh, mare’), turri-kore (bsq. iturri ‘spring’), nur-koro and nurakoro (palsd. nur ‘stone’), talae-kore (palsd. tala ‘stream, mountain’s river’), enale-kore (lat. vena, -alis ‘spring’). There is little we can say about Funtana bar-kori, which displays a first common root bar- (as in barí, bar-úmini and baru-nele) of unknown meaning. A quick examination of our records confirms that except for Nodu gurriai, ilikore and nurkoro, all the remaining compounds adhere to a deep meaning of ‘water’ (‘springs, wells, running water in mountain’s rivers and val20
The correct original spelling of this microtoponym (a valley!) was confirmed by Dr. F. Cabras of Urzulei; as a result of assimilation it has been quoted as *ortorani in the newest administrative records.
Iberian *ortubeles´ and ordumeles
185
leys’). If we now briefly go through the general Sardinian (i.e. romance) toponymic compounds which exhibit the term ‘red’(rubeum > rúviu, ruju, orrúbiu), we notice that except for Nuraghe rúviu, related to the reddish colour of the typical stones of the megalithic monuments, Monte (mons, montem ‘mountain’) ruju, and Perda (petram ‘stone, rock’) rúvia, the most frequent composition is with lexemes denoting ‘water’: Abba rúvia, ruja (aqua ‘spring, well’), Funtana rúvia, Ludu (lutum ‘mud’) ruju. Interesting enough is also the metaphoric use of ‘blood’ for ‘red’, which we have detected in derivatives with the Palaeo-Basque root *dol (do-dol > odol ‘blood’)21. Striking is the survival of the iterative reconstructed protobase in High Ogliastra, together with the subsequent result: Do-dol-iai and Dol-ai, for ‘red granitic, calcareous rocks and mountains’, besides the commoner Rivu (rivus, -um) dol-ia, Bruncu (‘summit, highest rocky point’) dol-au (and cf. Bruncu rúviu), the plural doli-ake, with reference to ‘springs’, and the compounds artz-ana-dolu ‘red shepherd’s settlement’ (bsq. artz- ‘shepherd’, sd. Arzu cognomen, artz-ai, ártz-ana) and isti-dul-é, iste-dol-í (bsq. istil). 3.2. -nele One astonishing link with Old Iberia is the maintenance of the root ib. beles´, meles, bsq. bele, bel-tz ‘dark’, applied to geomorphologic denotata all over Sardinia. Again we find two successive stages of the root: mele and nele. First is still maintained in a few toponyms, such as: mara-mele (anie. *mara ‘mare’), Macu-mele ‘Macomer’, a town of west Sardinia erected on basaltic, dark terrain, with semitic maqo¯m ‘settlement’ in its first segment, Arriu (rivus, -um) mele ‘river with slimy waters’, tavara-mele (sd. túvara ‘heather’), kili-melis and kerumele, keré-mule, towns and terrains, designations where we recognize the typical Iberian segments kelti- (kelti-beles´ is recorded 7 times; lt = [l]) and kertu- (rt = [r], as in ilti- = ili). This first root has evolved through dissimilation to the typical Palaeo-Sardinian root/suffix nele, and some doublettes confirm this evolution: keru-mele and keru-nele, bidu-mele and bidu-nele (bsq. bide ‘way’), mortu-mele and mortu-nele. The following compounds, scattered abundantly in the central and
21
Cf. Lakarra (1995, 2004).
186
Eduardo Blasco Ferrer
eastern regions of the island, turn out to be most instructive to guide us to the deep denotation: turri-nele and turru-nele (bsq. iturri ‘spring’), istiu-nele (bsq. istil), orro-nele (bsq. orri ‘leaves’), aran-nulu, arau-nele and g-arau-nele (bsq. aran ‘valley’)22, risu-nele (rivus,-um), Funtana thiku-nele (ficus)23, mortu(ru)-nele (mortuus,-um ‘dead, dried’, as in Abba or Mitza morta, italian Pozzo morto ‘stagnating water in wells’), gutturu-nele (guttur > sd. gútturu ‘gorge, deep ravine’). A quick survey of lexical compounds based on solely latin roots lends further support to our equivalences: Badde (vallis,-em), Sa vena (venam), Littu (*elictum, ‘forest of oak trees’, as salictum), Rivu (rivus,-um), Ficu (ficus,-um), Funtana (fontanam) + nieddu, niedda, lat. nigellum,-am ‘dark’. We can also observe the perfect correspondences with present Basque toponyms, such as Aran-beltza < aran + bel(tz) and Beltza-baratz < bel + baratz ‘farmed land’. We can now turn our attention to Iberian *ortubeles´, completing our inspection with Sardinian compounds of *ortu. 3.4. *ortubeles´, ordumeles In the most archaic region of Sardinia we come across two compounds with *ortu, which may help to disentangle the real meaning of the Basque root orto-koro and ortu-mele, once with ‘red’ and once with ‘black’. But we have also mentioned ort-orgo, with *org (sd. orga,-e) ‘spring, wet ground’ and ort-aran-i, with bsq. aran ‘valley’, both combinations pointing at precise geomorphologic features with creeks, depressions, and flowing water. The last compound is, we believe, the solution of this rebus: as org-osa, ort-aran conceals a tautologic designation, ‘valley’ in Iberian and ‘valley’ in Palaeo-Basque, both roots encroaching in Sardinia, as ili and iri and the offshots of beles´ and bele. 22
23
Prothesis of non-etymological consonants seems to be a very productive rule in toponyms, triggered of course by romance parallels (ex¯i re > sd. bessire, vulpis, -em > urpe and then gurpe): iriai/biriai, oronai/moronai, orune/burune, ortei/gurtei, ósana/gúsana, urue/gurue, and also aran and garan + nele. In the Barbagia f - deletes, so that f¯i cum > [’iʔu] (with Knacklaut or glottal stop instead of [k]). Nonetheless, prothesis of interdental [θ] = th is not unusual, as orgosa, urgusa and thurgusa shows. So, we can safely argue, that [θiʔu’nele] reflects f¯i cus + nele, a compound which indeed is recurrent in romance composition: Ficu niedda.
Iberian *ortubeles´ and ordumeles
187
4. Conclusions and desiderata In this short article we have tried to point out the emergence of an intedisciplinary examination of Iberian, Basque and Palaeo-Sardinian microtoponyms, supporting the comparative instances with typological and semantic tools. Recent anchaeological, historical and genetic evidence suggest that Sardinia received one or more migrations from Old Iberia during the Neolithic Age. Our discovery of the agglutinative typology of Sardinian toponymy, and the structural affinities we have gathered between toponymic data and Iberian and PalaeoBasque roots, allow us to contend the hypothesis that behind the mysterious compound *ortubeles´, ordumeles we may find the exact correspondence of Palaeo-Sardinian ortu-mele, with the root ortu sharing the common primary denotation of ‘valley’. A mainstream of research has been opened between Old Iberia and Sardinia, and further efforts must aim at collecting new material which permits a cohesive structural and semantic interpretation of common roots in Iberian and Palaeo-Sardinian. This long way has just begun. Bibliography Agostiniani, L., 1993. La considerazione tipologica nello studio dell’etrusco. Incontri Linguistici 16, 23–44. Agostiniani, L., 2000. La lingua. In M. Torelli (ed.), Gli Etruschi, 485–499. Milano, Bompiani. Bach, A., 1953. Die deutschen Ortsnamen, I. Heidelberg, Winter. Blank, A./Koch, P. (eds.), 1999. Historical Semantics and Cognition. Berlin/New York, Mouton/de Gruyter. Blasco Ferrer, E., 1988. Le parlate dell’Alta Ogliastra. Cagliari, Della Torre. Blasco Ferrer, E., 1993. Tracce indeuropee nella Sardegna nuragica? Indogermanische Forschungen 98, 177–185. Blasco Ferrer, E., 2002. Linguistica sarda. Storia, Metodi, Problemi. Cagliari, Condaghes. Blasco Ferrer, E., 2009. Storia della lingua sarda. Cagliari, Cuec. Blasco Ferrer, E., i.p. Paläosardisch und Paläobaskisch: bide berriak. In Th. Vennemann/R. Mailhammer (eds.), The Linguistic Roots of Europe. Kopenhagen, Munksgaard. Blasco Ferrer, E., i.p. Il sostrato paleosardo: fine d’un rebus. Zeitschrift für romanische Philologie 127/1. Coromines, J., 1981. Estudis de toponímia catalana, 2 voll. Barcelona, Barcino. Craddock, J., 1969. Latin Legacy versus Substratum Residue. Berkeley, University Press.
188
Eduardo Blasco Ferrer
de Hoz, J., 2001. Hacia una tipología del Ibérico. In F. Villar/M.P. Fernández Álvarez (eds.), Religión, Lengua y Cultura prerromanas de Hispania, 335–363. Francalacci, P. et alii, 2003. Peopling of three Mediterranean islands (Corsica, Sardinia, and Sicily) inferred by Y-chromosome biallelic variability. American Journal of Physical Anthropology 121, 270–279. Geeraerts, D., 1997. Diachronic Prototype Semantics. Oxford, Clarendon Press. Gómez Moreno, M., 1949. Misceláneas. Historia, arte, arqueología. Madrid, Aguirre. Gorrochategui, J., 1984. Onomástica indígena de Aquitania. Bilbao, Universidad del País Vasco. Gorrochategui, J. (2002). La lengua vasca = vascuence = euskara. In M.H. Mira Mateus (ed.), As línguas da Península Ibérica, 51–81. Lisboa, Colibri. Hualde, J./Lakarra, J./Trask, R., eds., 1995. Towards a history of the Basque language. Amsterdam, Benjamins. Hubschmid, J. 1963. Paläosardische Orstnamen. In Carlo Battisti (ed.), VII Congresso Internazionale di Scienze Onomastiche, II, 148–160. Firenze, Istituto di Glottologia. Hubschmid, J., 1953. Sardische Studien. Das mediterrane Substrat des Sardischen, seine Beziehungen zum Berberischen und Baskischen sowie zum eurafrikanischen und hispano-kaukasischen Substrat der romanischen Sprachen. Bern, Francke. Lakarra, J., 1995. Reconstructing the Pre-Proto-Basque Root. In Hualde et al., 187–207. Lakarra, J., 2004. Etimología y reconstrucción en el campo vasco. In Emilio Ridruejo (ed.), Las otras lenguas de España, 41–117. Valladolid, Universidad de Valladolid. Mastino, A., 2005. Storia della Sardegna antica. Nuoro, Il Maestrale. Michelena, L., 1985. Fonética histórica vasca. San Sebastián, Diputación Foral. Orpustan, J.B., 1991. Toponymie Basque. Bordeaux, Presses Universitaires. Orpustan, J.B., 2006. Nouvelle toponymie Basque. Bordeaux, Presses Universitaires. Pellegrini, G.B., 1994. Toponomastica italiana. Milano, Hoepli. Rohlfs, G., 1960. Europäische Flußnamen und ihre historischen Probleme. Proceedings of the 6th. International Congress of Onomastic Sciences, I, 23–50. München, Beck. Rowland, R.J., 2001. The Periphery in the Center. Sardinia in the ancient and medieval worlds. Oxford, Archaeopress. Siles, J., 1985. Léxico de inscripciones ibéricas. Madrid, Ministerio de Cultura. Silgo Gauche, L., 1994. Léxico ibérico. Estudios de Lenguas y epigrafías antiguas 1,1–271. Silvestri, D., 1977–82. La teoria del sostrato. Metodi e miraggi. Napoli, Macchiaroli. Silvestri, D., 1985–86. Ancora a proposito di elementi non indoeuropei nelle lingue germaniche. Filologia Germanica 28–29, 589–604. Taylor, J., 1995. Linguistic Categorization. Prototypes in Linguistic Theory. Oxford, Clarendon Press.
Iberian *ortubeles´ and ordumeles
189
Tischler, J., 1977. Kleinasiatische Hydronymie. Semantische und morphologische Analyse der griechischen Gewässernamen. Wiesbaden, Reichert. Tovar, A., 1997. Estudios de tipología lingüística. Madrid, Istmo. Untermann, J., 1990. Monumenta Linguarum Hispanicarum, vol. 3. Die iberischen Inschriften aus Spanien. Wiesbaden, Reichert. Untermann, J., 1998. La onomástica ibérica. Iberia 1, 73–85. Untermann, J., 1999, “Alteuropäisch” in Hispanien, in E. Eggers/J. Becker/ J. Udolph (Hgg.), Florilegium Linguisticum. Festschrift für Wolfgang P. Schmid zum 70. Geburtstag, Frankfurt a/M, 509–519. Velaza Frías, J., 1991. Léxico de inscripciones ibéricas (1976–1989). Barcelona, Universidad. Velaza Frías, J., 1996. Epigrafía y lenguas ibéricas. Madrid, Arco Libros. Wagner, M.L., 1951. La lingua sarda. Bern, Francke. Wagner, M.L., 1960–64. Dizionario Etimologico Sardo, 3 voll. Heidelberg, Winter. Wolf, H.J., 1998. Toponomastica barbaricina. Nuoro, Insula.
Sardische Sprachwissenschaft Facoltà Scienze della Formazione Unversità Cagliari /s Mirrionis 1 I-09123 C a g l i a r i Italia [email protected]
E d u a r d o B l a s c o Fe r r e r
190
Adrian Pârvulescu
Lat. servus Abstract Lat. servus does not date back to Indo-European, but is rather a Latin formation, a postverbal of serva¯re “to retain”, with the original meaning *“retainer”, “someone retained (for service)”.
The derivation of Lat. servus “slave” from serva¯re “to watch, observe, keep, retain, protect, preserve, save” or directly from the latter’s Indo-European ancestor *ser- “to guard, protect” + suff. *- u oˆ has a long history, servus being interpreted etymologically in antiquity as “(prisoner) saved (from slaughter)”1 and in modern times as “guard”, “shepherd”, or the like.2 The poor semantic support of this 1
2
Some late Roman sources interpreted servus as “saved” from serva¯re “to save” because the slaves were believed to be originally prisoners of war whose lives had been spared: see Institutiones Iustiniani 1.3.3 servi ex eo appellati sunt, quod imperatores captivos vendere iubent ac per hoc servare non occidere solent. For more examples see Lambertini 1984: 2386–7 and Maltby 1991: 564. A similar opinion at Pott 1867: 1278 who cites Vossius: “servi primum e captivis facti sunt, et dicti a servando, quia servati sunt, cum iure possent occidi”. This view was followed by Tovar 1971: 557–62 who considered the formation of servus vs. serva¯re as analogic to opi-parus vs. para¯re, arvus vs. ara¯re, conspicuus vs. conspica¯re, the original sense being “(vor dem Tode) bewahrt”. Tovar does not use the notion of postverbal and his list, which he defines as “Formen der ursprünglichen indogermanischen Typs”, mixes postverbals with *-u o- formations. ˆˇ ek 1881: 299–300, WP.: 2.498–9, For older references on this etymology see Vanic WH.: 2.525–7, EM.: 620, Tovar 1971: 556–560. To these ones add: Leumann 1977: 276 “substantiviert servus (s. homo Plt.) nach av. spa pasu- sˇ(h)aurva“vieh-hütender Hund”, vgl. die Bedeutung von serva¯re “hüten” (scil. das Haus) Plt.”; Rix 1994: 54–87 (esp. 81–7) “vorurital. *sor-u o-s “der Acht hat auf, Hirt” > (Transhumanz, Wanderhirtentum) > urital. *seruˆos “Sklave” (cf. kelt. *seru om ˆ ˆ “Hirtentum” > (“über das Phänomen der Transhumanz”) “Wanderhirtentum” > “Vagabundentum” > kymr. herw “Vagabundentum, Plünderei”, mittelir. serbh “Plünderei”); Untermann 2000: 670 “Dazu [sc. *ser- “bewachen, behüten”] mit suffix -u o – idg. *ser-u o- “Wächter, Hirte” in av. Adj. haurvaˆ “behütend”, lat. servusˆ *“Hirte” > “Sklave”; De Vaan 2008: 559 “PIt. *serwo[m.] “shepherd”, *serwo- [n.]/serwa [f.] “observation” < PIE. *ser-u-o- “guardian”, *ser-u-o/h2- “protection”. Other explanations: related to Ir. serbh “pillage”, Welsh herw “vagrancy, robbery”, herwr “banished, outlaw” (Loth
Lat. servus
191
etymology was excellently pointed out by Benveniste: “It is impossible to consider servus as a derivative of the verb serva¯re and to imagine that it was the function of the servus “to guard”. The verb serva¯re has a clear Indo-European etymology: Avest. haurva “who watches”, Gk. horân (² » ) “to observe, consider”. But servus indicates the legal and social condition of a slave and not a specific function. Surely the servus was not obliged to serva¯re”.3 Benveniste was perfectly right and one must add that no linguistic analogy to the presupposed semantic development “guard, shepherd” > “slave” has been presented so far. The theory of servus as *“guard”, “shepherd”, or the like should be discarded, but that does not necessarily mean that we must also abandon any relationship between servus and serva¯re. On the contrary, the two words are related by a different semantic link which the following lines will try to explain. It has been shown above that serva¯re has among its meanings also those of “to retain, keep”. For example: Plaut. Capt. 441 serva tibi in perpetuom amicum me “keep me your friend forever” (tr. Nixon); Merc. 228–9 dicit capram, quam dederam servandam sibi, / suae uxoris dotem ambedisse oppido “he says that the female goat [sc. young woman] I had given him to keep for me had completely devoured his wife’s dowry” (tr. Nixon); Capt. 930, Curc. 138, etc.: servare fidem “to keep faith, be faithful”; Cist. 199 servate vostros socios, veteres et novos “hold fast your allies old and new” (tr. Nixon); Cato agr. 141.3 utique … pastores pecuaque salva servassis “in order to … keep in good health the shepherds and the sheep”; etc. This sense of serva¯re enables us to find the clue of its connection with servus through anal-
3
1905: 211, followed by Brugmann 1906: 382–3, Vendryes 1935: 124–30, LEIA.: S-90–91); related to IE. *seru o- “running”, together with Skt. sárati “to run, ˆ flow”, Latv. sirt “to swarm around”, Gk. < L “to rush” (Brugmann 1906: 382–3, but discarded by Brugmann Grdr.: 2.3. 270–1); from Etruscan, cf. PN. Servi, Serve (Benveniste 1932: 429–440, also Benveniste 1969: 292, followed by Guenther 1960: 40–45); from PIE. *soru- “booty (particularly men, cattle, and sheep)” as *”he of the booty” (EIEC.:77). Other hypotheses at WP.: 2.498–9. Benveniste 1969: 292. F. de Saussure had expressed the same opinion: “Voici un example des témérities d’ autrefois: étant donné servus et serva¯re, on les rapproche – on n’en a peut- être le droit; puis on donne au premier la signification de “gardien”, pour en conclure que l’esclave a été a l’origine le gardien de la maison! Or, on ne peut pas même affirmer que serva¯re ait eu d’abord le sens de “gardien”” (Saussure 1968: 308).
192
Adrian Pârvulescu
ogy with the English pair retain vs. retainer. The English verb is attested since 1437 (reteinen, reteine, reteigne) with the meaning “to keep attached to one’s person or engaged in one’s service”, also “to engage, hire”.4 From the same time comes the derivative retainer used for “a dependent or follower of some person of rank or position; one attached to a house, or owing it service”5, “hired” or “retained (for service)”. The passive meaning of this -er derivative is a rarity among the approx. 978 -er formations recorded in English.6 One may cite weaner “a calf or lamb weaned during the current year”7, with an earlier variant weanyer from the year 15248 and with synonyms from the same root but a different suffixation such as wean-ling (a. 1532), wean-el (a. 1488) and a compound wean-calf “a calf newly weaned” (a. 1618).9 In the same line of formation appears to be keeper “one who continues or remains at a place” (a. 1611), also “a fruit, or other product, that keeps (well or ill)” (a.1843), derived from intransitive senses of the verb keep,10 and warder “one who wards or guards, watchman” (a.1400), but also “a person in ward, a prisoner” (a. 1584: from ward “to guard” + suff. -er).11 Lat. servus is similarly a derivative, in our case a nomen postverbale of serva¯re “to keep, retain” and must have meant originally *“retained (for service), hired” the same way retainer denotes “someone retained (for service), hired”. The implied passive meaning of this postverbal is a rarity also in Latin. Usually, postverbals form adjectives with active 4 5
6 7 8 9 10 11
OED.: 13.769; MED.: 8.580. OED.: 13.770; MED., ibid. The intransitive or passive senses “to adhere, be attached, be a retainer to one” are attested only a century later (a.1548, OED.:13.769). For the passive meaning instructive is also retinue “the fact of being retained in the service of another” (a. 1390), also “a number or company of persons retained in the service of someone” (a. 1375, see OED.: 13.781), from OFr. retenue “action de retenir, de detenir; engagement; gage, salaire” (Godefroy 1892:142–3) originally past passive participle of OFr. retenir “prendre et garder auprès de soi à son service”, “ne pas laisser, faire prisonnier” (a. 1100), “engager, louer” (a. 1315–7), see Tobler-Lommatzsch 1971: 1120; FEW.: 10.334a; TLF.: 14.1015. Category not found in Thorndike 1941 nor in the Suffixes 1982: 220–1. OED.: 20.44 from the year 1881. OED.: ibid. OED.: ibid. OED.: 8.376 OED.: 19.902.
Lat. servus
193
meanings: pro¯digus “prodigal” < pro¯digere “to waste”, -loquus “talking” < loqui “to talk” (mali-loquus, vani-loquus, fati-loquus, etc.), condus “one who stores (provisions)” < condere “to store up”, mergus “diver, sea-bird” < mergere “to dive”, etc. But passive or intransitive senses also are found, albeit rarely: dividus (Accius, Gellius), dividuus (Plaut.) “divided, separated” < dividere “to divide”, concinnus (Plaut.) “well adjusted” < concinna¯re “to set right, adjust”,12 truncus “lopped, trimmed” < trunca¯re “to lop off”, privi-gnus “stepson”, lit. “born single” < privus “single” + gignere “to beget, bear, produce” (mali-gnus, “malicious”, lit. “born evil”, beni-gnus, etc.), pro¯miscus (Plaut.), pro¯miscuus (Livy) “mixed, indiscriminate” < pro¯miscere “to mix”.13 If we judge from its Italic and IE. cognates, serva¯re denoted etymologically “to watch, observe, guard”: compare Umbr. seritu, aseriatu (anseriato) “to observe” (an- “on, to”), ooserclome “watch(tower)” (*op-seri-tlo-m = Lat. observaculum),14 Av. haraiti “to give attention to, watch over, protect”, med. “keep, preserve, save”, harwtar “watcher, guard”, harw- ra n. “care, upkeep”, ha¯ra- “watching over, 12
13
14
Leumann 1977: 398, but a different explanation at 268 (shortened from an original -atus). On these cases see Tovar 1971: 560 and Bader 1962: 202. The forms dividuus and pro¯miscuus show the influence of the large number of adjectives built with the help of the productive – u o – suffix carrying an active as well as a passive meanˆ ing. On nomina postverbalia see in general Brugmann Grdr.: 2.1.18. It must be said that Latin grammars have been very sparing with this category of nounformation. Meillet-Vendryes 1963: 366 reserves only four lines to it, mentioning pugna, abundus, and incurvus. Buck 1933: 313 refers only to pugna. Here is what P. Kretschmer had to say in 1951 on this matter: “Die postverbalen Nominalableitungen des Lateinischen haben nicht wie die griechischen und germanischen eine systematische und vollständige Feststellung und Untersuchung erfahren. Dadurch werden zuweilen etymologische Fehldeutungen verschuldet. Auch die Abgrenzung der Gattung ist nicht immer leicht. So will Leumann, Lat. gr. 196 die Ableitungen aus Komposita nichtabgeleiteten Verben wie advena confluus retrogradis nicht als Rück-, sondern als Suffix-ableitungen auffassen und verzeichnet S. 203 scriba ohne Erklärung” (Kretschmer 1951). In the revised edition of his grammar, Leumann mentions Kretschmer’s reference to scriba and a large list of postverbals, such as advena, are named “etymologisch durchsichtige lat. Komposita mit verbalem Schlussglied auf -a” (Leumann 1977: 280). The booklet of F. Brender, Die rückläufige Ableitung im Lateinischen. Diss. Lausanne 1920, 83 p. (advisor M. Niedermann) considers only the undisputed cases. Untermann 2000: 669–670.
194
Adrian Pârvulescu
observing”, hi-sˇara- “watching over”,15 Zoroastrian Pahlavi, NPers. zen-ha¯r “protection” (zen “having in one’s charge”), Khot. hadaa fl “watcher, guard leader”,16 all from IE. *ser- “to take care of, protect, preserve”.17 Lat. serva¯re is based upon an *-u - extension of this root, ˆ *ser-u -, together with Osc. serevkid n. “inspection, protection” ˆ (*seru -ikio- < *ser-u o- “observing”),18 Av. haurvaiti “to watch over, ˆ ˆ protect”.19 A g- extension appears in Lith. sér-g-eti “to watch”, sárg-as “watcher, protector”, Latv. sar-g-a¯t “to protect, watch over”.20 It is obvious that the connotations “to keep, retain, preserve” of serva¯re are late developments and this is proven also by parallel cases: OE. cepan “to watch, keep an eye upon, observe”, also “seize, lay hold of” > ME. kepen (cepen, quep, kipen, etc.) “to care about, receive”, also “keep, retain, hold”21 > NE. keep “to preserve, maintain, retain”; Gmc. *wardôn (OHG. warten, OSax. wardo¯n, OE. weardian) “to watch, look out for, wait” > OFr. garder “to look, protect, watch” > NFr. “to take care of, retain, keep guard, maintain”.22 One can conclude that Lat. servus does not date back to Indo-European, but it is rather a Latin formation, to be more exact, a late postverbal of serva¯re “to retain” with the original meaning *“retainer, someone retained (for service), hired person”. If this interpretation is correct, then the origin of servus would not indicate “exodouly” as the main source of Roman slavery,23 but probably a process similar to the one which led in the Middle Ages to the institution of serfdom and which presumably assimilated the condition of “hireling” with that of “prisoner of war”. Indeed, F. de Martino showed that “con le guerre di supremazia in Italia del IV secolo si ebbe un mutamento nella concezione della schiavitu ed essa venne ri-
15 16 17 18 19
20
21 22 23
Bartholomae 1905: 1787–90, 1806, 1816. Bailey 1979: 447, Rastorgueva/Edel’man 2007: 367–9. IEW.: 910. Untermann 2000: 669. Also visˇ-haurva [hauruua] “guarding the village”, pasusˇ-haurva “guarding the flock”. Pedersen 1893: 299 (but with a questionable explanation), Fraenkel: 762–3, 776–7, Karulis 1992: 2.155. MED.: 5.464–76. Buck 1949: 752–3, FEW.: 17.510–25. WH.: 2.527 “die idg. Völkern haben nur die “Exodulie” gekannt”, followed by Benveniste 1969: 292.
Lat. servus
195
collegata alla prigionia di guerra”.24 But the presence of servus in the Lex XII tab. (ca. 450 B.C.), characterizing a low human condition, because the penalty for harming the body of a servus is half that of a liberus “free man”, which is reminiscent of similar cases mentioned in the laws of Moses (e.g., Ex. 21.12 vs. 20–21) or the Code of Hammurabi (e.g., no.196 vs.199), shows that this process of assimilation to the condition of prisoners of war must have been much older.
Bibliography Alföldy, G. 1985, The Social History of Rome. Tr. D. Braund and F. Pollock. London/Sydney. Bader, F. 1962, La formation des composés nominaux du latin. Paris. Bailey, H.W. 1979, Dictionary of Khotan Saka. Cambridge. Barrow, R.H. 1928, Slavery in the Roman Empire. London. Bartholomae, Chr. 1905, Altiranisches Wörterbuch. Strassburg. Bradley, K.R. 1987, On the Roman Slave Supply and Slave Breeding. In: Classical Slavery. Ed. M.I. Finley. London, 42–64. Bradley, K.R. 1994, Slavery and Society at Rome. Cambridge. Brugmann Grdr. = K.B., Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. Strassburg 1930 (=1897–1916). 2 vols. In 4 parts. Brugmann, K. 1906, Zu den Benennungen der Personen des dienenden Standes in den indogermanischen Sprachen. In: Indogermanische Forschungen, 19.377–91. Benveniste, E. 1932, Le nom de l’esclave à Rome. In: Revue des Études Latines, 10.124–30. Benveniste, E. 1969, Indo-European Language and Society. Tr. E. Palmer. London. Buck, C.D. 1933, Comparative Grammar of Greek and Latin. Chicago. Buck, C.D. 1949, A Dictionary of Selected Synonyms in the Principal Indo-European Languages. Chicago. Capogrossi, L. 1978, Il campo semantic della schiavitù nella cultura latina del terzo e del secondo secolo a.C. In: Studi storici 19.717–733. De Martino, F. 1979, Diritto e società nell’antica Roma. Rome. 2 vols. De Vaan, M. 2008, Etymological Dictionary of Latin and the Other Italic Languages. Leiden/Boston. DNP. = Der neue Pauly. Stuttgart/Weimar 2001, vol.11.
24
De Martino 1979: 1.161 (see the entire chapter, p.130–161); see also Capogrossi 1978: 724–5. For a brief survey of the prisoners of war captured by the Romans, see Bradley 1987: 42–7 and Bradley 1994: 32–46; a detailed survey at Welwel 2000: 1–160. On the cases of selling of self into slavery, see Barrow 1928: 11–12 and Veyne 1991: 247–280. On debtors as a source of slavery, see Alföldy 1985: 11–12.
196
Adrian Pârvulescu
EIEC. = Encyclopedia of Indo-European Culture. Eds. J.P.Mallory, D.Q.Adams. London/Chicago 1997. EM. = A. Ernout, A. Meillet, Dictionnaire étymologique de la langue latine. Paris 19854. FEW. = W. von Wartburg, Französisches etymologisches Wörterbuch. Basel. Vol.10. Fraenkel = E.F., Litauisches etymologisches Wörterbuch. Heidelberg 1963–5. 2 vols. Godefroy = F.G., Dictionnaire de l’ancienne langue française. Paris 1892, vol.7. Guenther, R. 1960, Etr. serve- lat. servus. Ein Deutungsversuch. In: Acta Antiqua Academiae Scientiarum Hungaricae 8. 45–50. IEW. = J.Pokorny, Indogermanisches etymologisches Wörterbuch. Tubingen/Basel 1994 (= 1959). Karulis, K. 1992, Latviesˇu etimologijas va¯rdnica. Riga, 2 vols. Kretschmer, P. 1951, Zu den lateinischen Postverbalien. In: Glotta 31. 152. Lambertini, R. 1984, L’etimologia di servus secondo I giuristi latini. In: Sodalitas. Scritti in onore di Antonio Guarino. Napoli, vol.5. LEIA. = J. Vendryes, Lexique étymologique de l’irlandais ancien. Lettres RS. Dublin/Paris 1974. Loth, J. 1905, Contribution a la lexicographie et l’etymologie celtiques. In: Mélanges H.D’Arbois de Jubainville. Eds. Collinet et al. Paris. Maltby, R. 1991, A Lexicon of Ancient Latin Etymology. Leeds. MED. = Middle English Dictionary. Ann Arbor 1956–1999, 13 vols. Meillet/Vendryes = A. Meillet, J. Vendryes, Traité de grammaire comparée des langues classiques. Paris 1963 (= 19482). OED. = Oxford English Dictionary. Second edition. Oxford 1989, 20 vols. Pedersen, H. 1893, Servus und serva¯re. In: Beiträge zur Kunde der indogermanischen Sprachen. 19.298–302. Pott, A.F. 1867, Etymologische Forschungen. Detmold, Bd.2, Abt.2. ˇ eskij slovar’ Rastorgueva/Edel’man = V.S. Rastorgueva, D.I. Edel’man, Etimologic fi iranskix jazykov. Moskow 2007, Vol.3. Rix, H. 1994, Die Termini der Unfreiheit in den Sprachen Alt-Italiens. Stuttgart. Saussure, F. de 1968, Cours de linguistique générale. Eds. Ch. Bally, A.Sèchehaye, A. Riedlinger. Paris. Suffixes 1982 = Suffixes and Other Word-Final Elements of English. Eds. L. Urdang, A. Humez, H.G. Zettler. Detroit. Thorndike, E.L. 1941, The Teaching of English Suffixes. New York. TLF. = Trésor de la langue française. Paris 1990, vol.14. Tobler-Lommatzsch 1971 = A. Tobler, E. Lommatzsch, Altfranzösisches Wörterbuch. Wiesbaden, vol. 8. Tovar, A. 1971, Lat. seruus. Ein Indogermanisches Wort. In: Sprache und Geschichte. Festschrift für Harri Meier. Ed. E. Coseriu, W.D. Stempel. München. Untermann, J. 2000, Wörterbuch des Oskisch-Umbrischen. Heidelberg. Vanicˇek, A. 1881, Etymologisches Wörterbuch der lateinischen Sprache. Leipzig. Vendryes, J. 1935, À propos de lat. servus. In: Bulletin de la Société de Linguistique de Paris, 36.124–30.
Lat. servus
197
Veyne, P. 1991, La société romaine. Editions du Seuil. Welwel, K.-W. 2000, Sub corona vendere. Quellenkritische Studien zu Kriegsgefangenschaft und Sklaverei in Rom bis zum Ende des Hannibalkrieges. Stuttgart. WH. = A.Walde, Lateinisches etymologisches Wörterbuch. 3., neubearbeitete Auflage von J.B. Hofmann. Heidelberg 1954, vol.2. WP. = A. Walde, Vergleichendes Wörterbuch der indogermanischen Sprachen. Hrsg. J. Pokorny. Berlin 1927–1932, 3vols.
983C Ponderosa Avenue S u n n y v a l e , CA 94086–8946 U.S.A. [email protected]
Adrian Pârvulescu
198
Catalin Anghelina
A note on the phoneme schwa (a˘) in Romanian Abstract The causes that led to the emergence of the phoneme a˘ = [w] in Romanian are obscure. While the general consensus is that this emergence is due to the appearance of the morphological contrast between the definite and indefinite forms of the feminine nouns ending in -a, this paper proposes that this phoneme emerged much earlier, when a˘ = [w] came to be analogically used in the paradigm of the perfect.
The present paper treats some aspects of Romanian verbal morphology, which have not been definitively solved yet. These aspects regard the morphological role of schwa in differentiating verbal categories and may shed some light on the emergence of this phoneme in Romanian. In Romanian, the present indicative of a verb belonging to the 1st conjugation class, e.g., a la˘udá ‘to praise’, runs as follows: eu laúd ‘I praise’, tu laúzi ‘you praises’, el laúda˘ ‘he praises’, noi la˘uda˘2m ‘we praise’, voi la˘udát¸i ‘you praise’, ei laúda˘ ‘they praise’.1 One can see that the 1st person plural has a non-etymological stressed [w] ´ as ending: Lat laudámus should have given Rom *la˘udám, not la˘uda˘2m.2 This a˘ is present in the majority of the Romanian dialects, the most notable exception being Istro-Romanian. The usual explanation given for this is an analogy between the present and imperfect: IMPF 3rd SG la˘udá is to PRES 3rd SG laúda˘ what IMPF 1st PL is to PRES 1st PL. Thus, the PRES 1st PL becomes la˘uda˘2m and differentiates itself from the imperfect la˘udám.3
1 2
3
The Latin paradigm runs: laudo, laudas, laudat, laudamus, laudatis, laudant. The first a˘ = [w] is the result of the synchronic rule which turns, as in English for example, any unaccented a into w. Dimitrescu et alii (1978:301) considers it as an independent innovation in all the dialects of Romanian; I would rather see it as a Common Romanian innovation that did not spread to the dialect that will eventually become Istro-Romanian. Along the same lines, Sala (1976:192) concludes that “il est très probable que c’est la morphologie qui est intervenue … à marquer la difference entre le présent, imparfait …”.
A note on the phoneme schwa (a˘) in Romanian
199
I do not consider this solution very compelling for two reasons: first, the accent is situated on a˘ in the 1st person plural, but not on a˘ in the 3rd person singular; second, it is not clear what the connection was between these persons so that the analogy could work. I propose here a new solution, which is also based on an analogical process between the present and imperfect. Let us compare the 1st person plural of the present and imperfect forms from the 1st and 2nd conjugation classes, focusing on what happened from Latin to Common Romanian. In this scheme, I will assume that the synchronic rule by which any unaccented a turns into a˘ was already present in Romanian at this stage: Latin: 1st CONJ 2nd CONJ
PRES laudamus PRES videmus
Common Romanian 1st CONJ PRES la˘udámu PRES vedému 2nd CONJ
IMPF laudabamus IMPF videbamus IMPF *[lwudw(b)ámu] IMPF *[vede(b)ámu]
In this situation, the alternation e/ea between the present and imperfect of the 2nd conjugation may have influenced the 1st conjugation class and its alternations. The imperfect of the 1st conjugation would have had initially the suffix -*a˘a- < -aba-, where a˘ is the result of the original unaccented a. The analogical process then can be described as follows: if ea alternated with e between the present and imperfect in the 2nd conjugation, then *a˘a could alternate only with a˘ in the 1st conjugation for the same tenses. This analogy assumes the existence of a diphthong *a˘a at this stage of Romanian, which later contracted to a. There is, however, a problem with this solution: if, indeed, a˘ in the 1st person plural is the result of this analogy, why did the 2nd person plural not participate in this? Why does Romanian not have la˘uda˘2t¸i? The reason for this is obscure, but it may simply have to do with the sporadic character of the analogical processes in general. As I said above, the case of Istro-Romanian is different since this process did not take place there.4 In addition, this dialect innovated in the imperfect paradigm, where the 4th conjugation formed the basis 4
See note 3.
200
Catalin Anghelina
for all the other conjugation types: IMPF audiiam ‘to hear’ (< Vulg. Lat. audibam) was the basis for scapaiam ‘to drop’ (Lat. excapere) from the 1st conjugation. Another crux of the Romanian verbal morphology is the form of the 3rd person singular perfect for verbs belonging to the 1st conju´ For gation class.5 These forms have as ending an accented a˘ = [w]. example, Rom. INF a la˘uda [alwwudá] < Lat. laudare ‘to praise’ has ´ < Lat. PERF laudáuit.6 The normal PERF 3rd SG la˘uda˘ = [lwwudw] evolution should have been [lwwudá], which would have been identical to the imperfect. Romanian, however, displays a different form. While most opinions share the view that this has been an analogical process in order to distinguish the perfect from the imperfect, there is little agreement on how the analogy worked.7 A possible answer can be found in the paradigm of the 4th conjugation. A verb like Lat. audire, for example, had the PERF 3rd SG audiuit. This perfect type was parallel to that of the 1st conjugation class, having the suffix u added to the verbal root. Thus, for the 3rd person singular, audíuit was parallel to laudáuit. In the transition from Latin to Common Romanian, these two forms became very early *audí and *laudá. On the other hand, the present tense of such verbs must have been *[aúdi] and [laúdw]. This, then, may have offered the premise for the following analogy: 4th CONJ 1st CONJ
PRES *aúdô8 PRES laúda˘
PERF *audâ9 PERF X
The solution is exactly what we find in Romanian: la˘uda˘ [lwwudw]. ´ The above facts show how old the phonemicization of [w] may be. Given the fact that ô > e (*audô > Rom. aude) precedes the emergence
5 6
7
8
9
This is the Romanian ‘perfectul simplu’. The [w] in [lwwuda] is due to the fact that, synchronically, Romanian syllables must have onsets. Densus¸ianu (1997: 221) correlates this ending with that of the 1st person plural of an old perfect la˘uda˘m. It is, again, not clear what the connections are between these personal forms; cf. Dimitrescu et alii (1978: 309); Rosetti (1978:154). The keystone of this demonstration is that the analogy took place before i in *aúdi became e (fle) (Rom. aúde). This form gave in Common Romanian, after fricativization, *audzí > Rom. auzí. Aromanian still has audzí.
A note on the phoneme schwa (a˘) in Romanian
201
of the diphthong ea,10 it means that the phoneme /w/ could appear even before it was used in the alternation between the definite and indefinite nominal forms, e.g, casa˘ – casa, which is parallel to parte – partea. The perfect forms analyzed above, then, could be at the origin of the phoneme a˘.11 In addition, if, indeed, the analogical process proposed above occurred, this could mean that the phoneme a˘ emerged when the Latin vocalic system in the front series (i ô efl e˛) was still in place. This may also lead to the conclusion that the emergence of this phoneme occurred at a stage of the language, which was closer to Latin than to what will subsequently become Common Romanian. Besides, although there is obviously a relation between Romanian, Bulgarian and Albanian regarding the postponing of the definite article (e.g. Rom. INDEF fata˘ – DEF fata ‘girl’ is parallel to Alb. vajzë – vajza), which also means that this areal Balkanic feature may be due to either one of these languages or the substratum, the phonemicization of schwa seems to have occurred in Common Romanian in an independent way.12 A final discussion regards the morphological role of schwa in marking the plural of some feminine nominal forms, which display the alternation a – a˘ between the singular and plural. One should note again that the original Latin tonic vowel on these nouns was a:13 INDEF SG parte ‘part’, falca˘ ‘cheek’, ‘jaw’ scara˘ ‘ladder’ – INDEF PL pa˘rt, i, fa˘lci, sca˘ri < Lat. partem, *falca (falx), scala, etc. In examples of this sort the presence of schwa (a˘) in the plural has no phonological explanation, despite some attempts to find such a solution14. This is clearly shown by the fact that masculine nouns with similar phonological form do not enter the system of alternations: MASC SG frate < Lat. frater ‘brother’ – MASC PL frat, i [fratsi], MASC SG tata ‘father’-MASC PL tat, i [tatsi]. Even an isolated word like mart, i [martsi] ‘Tuesday’ < Lat. Martis (‘Dies Martis’) shows that we are not dealing here with a phonological process. On the other hand, from a synchronic standpoint, Chitoran (2002:209) rightly views
10 11
12 13 14
Cf. Rom. neagra˘ < Lat. nigra. Sala (1976: 194) argues that the phoneme a˘ occurred with the contrastive pair casa˘ – casa. For this areal feature, see e. g. Tagliavini (1973: 496), Orel (2000: 246). It is attested in all dialects with the exception of Istro-Romanian. Cf. Sala (1976:193) for a historical synopsis.
202
Catalin Anghelina
this alternation as corresponding to the alternations ea-e and oa-o, which are also used to additionally mark the category of number for some feminine nouns, e.g. INDEF SG seara˘ – INDEF PL seri ‘evening’ < Lat. sera, INDEF SG floare – INDEF PL flori ‘flower’ < Lat flos, ris. Chitoran’s synchronic analysis can lead to a solution for the ‘diachronic’ problem. Thus, this unetymological a˘ can be explained through the existence of the systemic alternations ea-e and oa-o between the singular and plural of the feminine nouns. In other words, these already existing alternations created the premises for the morphophonemic equivalent alternation a-a˘. Therefore, the additional marking of the plural through a˘ is the result of the morphological pressure created by the system of morphophonemic alternations ea-e and oa-o. This solution had already been stated by Phillipide (1927:21), but it has been since forgotten until Brâncus¸ (2000:295–300) recently reaffirmed it, rightly considering that the alternation must belong to Common Romanian since it can be seen both in Romanian and Aromanian. We can see thus that schwa’s very important morphological role begins with Common Romanian. It is, however, very possible that its oldest use as a phoneme came with the perfect forms analyzed above.15
Bibliography Brâncus¸, Grigore (2000). Studii s¸i Cercetari Lingvistice 51:295–300. Chitoran, Ioana (2002). The Phonology of Romanian. Berlin/New York: Mouton de Gruyter. Densus¸ianu, Ovidiu (1997). Histoire de la langue roumaine. Bucures¸ti (original edition 1901, Paris). Dimitrescu, Florica, Barborica˘, Elena, Ca˘la˘ras¸u, Cristina, Cvasnîi, Maria, Marta, Mihai, Ruxa˘ndoiu, Liliana, Pamfil, Viorica, Theodorescu Mirela, Toma Elena (1978). Istoria Limbii Române. Bucures¸ti. Orel, Vladimir (2000). A Concise Historical Grammar of the Albanian Language. Brill: Leiden. Philippide, Alexandru (1927). Originea Românilor. Ias¸i. 15
I would not exclude the emergence of the phonemic schwa through the ‘peripheral’ lexical opposition ca˘ ‘that’ (conjunction, < Lat. quod) – ca ‘than’ (preposition, < Lat. quam). These ‘little’ words can be found both in Romanian and Aromanian and, therefore, belong to Common Romanian.
A note on the phoneme schwa (a˘) in Romanian
203
Rossetti, Alexandru (1978). Istoria limbii române. Bucures¸ti. Sala, Marius (1976). Contributions à la phonétique historique du roumain. Paris: Klincksieck. Tagliavini, Carlo (1973). Orígenes de las lenguas neolatinas. México (original edition 1949, in Italian).
3569 Chowning Ct C o l u m b u s, OH 43220 U.S.A. [email protected]
Catalin Anghelina
204
Miguel Villanueva Svensson
Baltic sta-presents and the Indo-European desiderative* Abstract Baltic sta-presents are a productive verbal formation for anticausative and inchoative verbs. They stay in complementary distribution with nasal presents, stapresents being originally restricted to TERT- and TER-roots (the later with regular métatonie rude). The only reasonable cognate of the sta-presents is Slavic *orste- “grow”, which suggests that we are dealing with a Balto-Slavic formation of still disputed origin. In this article it is proposed that Baltic sta-presents (BaltoSlavic *ste/o-presents) originated in the 3 sg. middle of an athematic desiderative (Lith. mìrsˇta “dies”, vi˜rsta “turns into, becomes” h *mér-h1s-toi “wishes to die”, *Üért-s-toi “wishes to turn”). Forms like *mér-h1s-toi, *Üért-s-toi would be the regular middle counterpart of the Indo-European Narten desiderative that gave rise to the Baltic future tense. In some verbs they became unmarked present stems, a development favored by the common northern Indo-European trend to develop a productive class of anticausative-inchoative verbs and by the fact that infixal nasal presents could not be freely derived from all root structures. The zero grade of the sta-presents was taken from the thematic aorists with which they originally correlated. The present stem suffix *-(h1)ste/o- must be the result of a complex analogical chain *-(h1)s-tói f *-(h1)s-tó+ti f *-(h1)stó-ti f *-(h1)sté-ti, which can be linked to the decay of the middle voice in Balto-Slavic.
1. Lithuanian and Latvian have a large class of anticausative and inchoative verbs characterized by a present stem suffix -sta, e.g. diñgsta (inf. diñgti, pret. diñgo “disappear”), témsta (témti, t˜emo “grow dark”), pa-mi˜rsˇta < *-mi˜rsˇ-sta (-mi˜rsˇti, -mi˜rsˇo “forget”). The origin of this suffix remains a crux. Before offering a new proposal, I will present an overview of sta-presents as found in Baltic, paying special attention to what may be relevant for a historical elucidation of the type.1
* This article was written within the framework of the Project BALTLINGVA: Research on Baltic Linguistic Heritage and its Dissemination through Information Technologies, sponsored by the Lithuanian State Science and Studies Foundation. 1 For fuller descriptive surveys of nasal and sta-presents I refer to the standard grammars. Rich collections of data can also be found in Johansson (1893:476 ff.), Arumaa (1957), Pakalnisˇkiene˙ (1993).
Baltic sta-presents and the Indo-European desiderative
205
2. The position of sta-presents in the (East) Baltic verbal system is entirely parallel to that of nasal presents like sˇviñta (sˇvìsti, sˇvìto “dawn”), limpa ˜ (lìpti, lìpo “stick to”). Both formations constitute a productive class of anticausative and inchoative verbs. The vast majority of them can be classified into one of two derivational patterns: 2.1. Anticausative nasal and sta-presents beside full grade transitive ia-presents. This is the most common type,2 including sta-presents like diñgti, diñgsta “disappear, vanish” (: deñgti, deñgia “cover”), liñkti, liñksta “bend (intr.)” (: leñkti, leñkia “bend (tr.)”), lãzˇti, lãzˇta “break (intr.)” (: láuzˇti, láuzˇia “break (tr.)”), v˜ykti, v˜yksta “happen, occur” (: ve˜ıkti, ve˜ıkia “do, act”), vi˜rsti, vi˜rsta “turn into, become” (: ve˜rsti, ve˜rcˇia “turn over”), and nasal presents like kìsti, kiñta “change (intr.)” (: ke˜ısti, ke˜ıcˇia “change (tr.)”), plìsti, pliñta “spread (intr.)” (: ple˙˜sti, pl˜ecˇia “widen”), rùkti, ruñka “wrinkle (intr.)” (: raukti, ˜ raukia ˜ “wrinkle (tr.)”), sklìsti, skliñda “spread (intr.)” (: skle˜ısti, skle˜ıdzˇia “spread (tr.)”), tìkti, tiñka “be fit for” (: te˜ıkti, te˜ıkia “give”). 2.2. Inchoative nasal and sta-presents, used almost exclusively with preverbs, beside stative or durative verbs of several morphological types (mostly, but not exclusively with a second stem *-e-, *-a¯-).3 sta-presents: pra-gy´sti, -gy´sta “begin to chant” (: giedóti, gíesti/gíeda “chant”), pa-mìlti, -mìlsta “fall in love” (: mylqti, my´li “love”), pravi˜rkti, -vi˜rksta “begin to cry” (: ve˜rkti, ve˜rkia “cry”), pra-zˇy´sti, -zˇy´sta “begin to bloom” (: zˇydqti, zˇy´di “bloom”), pa-zˇìnti, -zˇ˛ísta “get to know” (: zˇinóti, zˇìno “know”). Nasal presents: pra-bìlti, -b˜yla “begin to speak” (: bylóti, bylója “say, tell”), pa-bùsti, -buñda “wake up” (: budqti, bùdi “be awake”), uzˇ-mìgti, -miñga “fall asleep” (: miegóti, mi˜egti/mi˜ega “sleep”), sˇvìsti, sˇviñta “dawn” (: sˇvi˜esti, sˇvi˜ecˇia “shine”), nu-tìlti, -t˜yla “fall silent” (: tylqti, t˜yli “be silent”). 2.3. Beside these major types, we find a wealthy number of denominative sta-presents like brañgti, brañgsta “become expensive” (: brangùs “expensive”), sve˜ıkti, sve˜ıksta “get better, recover” (: sve˜ıkas “healthy”), ru¯˜ kti, ru¯˜ ksta “smoke” (: ru¯˜ kas “mist, fog”), sénti, sénsta “grow old” (: s˜enas “old”), and nasal presents like ap-àkti, -añka “become blind” (: akìs “eye”), at-pìgti, -piñga “fall in price” (: pigùs 2 3
Pakalnisˇkiene˙ (1993:17 ff.) lists 206 such pairs in Lithuanian, 68 in Latvian. Pakalnisˇkiene˙ (1993:137 ff.) lists 150 verbs in Lithuanian (118 beside verbs with a second stem, 32 beside ia-presents), only 17 in Latvian, most of them not in a clear opposition to a stative verb.
206
Miguel Villanueva Svensson
“cheap”), sˇlàpti, sˇlampa ˜ “become wet” (: sˇlãpias “wet”), sˇlùbti, ˇslumba ˇ ˜ “become lame” (: slùbas “lame”). 2.4. Finally, nasal and sta-presents are also found among primary verbs with the appropriate semantics. We have sta-presents like aus ˜ ˇ ti, aus ˜ ˇ ta “dawn”, brvsti, brvsta “ripen”, mi˜rti, mìrsˇta “die”, pa-mi˜rsˇti, -mi˜rsˇta “forget”, Latv. svîst, svîstu “sweat”, and nasal presents like lìpti, limpa ˜ “stick to”, su-pràsti, -prañta “understand”, mìsti, miñta “nourish oneself”, sèkti, señka “sink”, sˇìlti, sˇy˜ la (dial. sˇìlsta) “get warm”. In some cases nasal and sta-presents may have replaced earlier types of presents, as gìmsta beside OLith. g˜ema “is born” or sniñga beside OLith. and dial. sni˜egti, sni˜ega “snows”. In other the derivational basis may have been lost, but for many verbs there is no inner-Baltic evidence suggesting that the sta-present must be a recent coinage. 3. Nasal and sta-presents share the same paradigm: they are typically built to the zero grade of the root, the infinitive stem is attached directly to the root, and both have the a¯-preterit.4 The present stems -sta and -n- stay in a near to complementary distribution. Stang (1942:132 f., 1966:340 ff.) gives the following principles:
4
Exceptions with the e-preterit are very few. In the literary language only two are known: mi˜rti, mìrsˇta, mìre˙ “die” (OLatv. nomirre Elger), gìmti (dial. gimti), ˜ gìmsta, gìme˙ “be born”. The dialects furnish further examples like dil˜ti/dìlti, d˜yla/dìlsta/d˜ela, dìlo/dìle˙ “wear out”, svil˜ti/svìlti, sv˜yla/svìlsta/sv˜ela, svìlo/svìle˙ “scorch”, tàpti, tampa, ˜ tãpo/tãpe˙ “become”, skàsti, skañta, skãto/skãte˙ “spring, hop”. mìrsˇta is the only sta-present affected by the ruki-rule. This, however, only guaranties a certain antiquity within Baltic (inasmuch as they are not relatively recent creations, the effects of the ruki-rule may have been easily removed in less salient verbs). For this root we can safely reconstruct a Áe/o-present *mr-Áétor (Ved. mriyáte, Lat. morior), still preserved in Slavic (OCS. Zo., Mar. umцretч, Slvn. mrjèm, cf. Tedesco 1956:520 ff., Barton 1980:263 f.). The irregular e-preterit mìre˙ (< *mir-iÁa¯-) is thus probably dependent on an earlier Baltic present *mir-ia-, see below footnote 41. The sta-present of gìmti is a fairly recent replacement of old and dialectal g˜ema. The preservation of the original e-preterit must be attributed to the frequency of this verb, no doubt supported by the e-preterit of its antonym mi˜rti. A recent transfer into the class of nasal and sta-presents probably accounts for other unexpected e-preterits as well.
Baltic sta-presents and the Indo-European desiderative
207
nasal presents are built to roots with a short vowel followed by a stop, a liquid, j or v, e.g. tìkti, tiñka “fit, be fitted”, bùsti, buñda “wake up”, lìpti, limpa ˜ “stick to”, bálti, bã˛la (< *banla) “become white”, svìrti, sv˜yra (< *svinra) “bend (intr.)”, gy´ti, g˜yja (< *ginja) “recover, get better”, pãti, pu¯˜ va (< *punva) “rot”. sta-presents are built to roots with a long vowel, to TERT/S-roots, and to roots ending in a nasal, e.g. dy´gti, dy´gsta “spring, sprout”, tõlti, tõlsta “move away”, diñgti, diñgsta “disappear”, ta˜rpti, ta˜rpsta “thrive”, rìmti, rìmsta “calm down”, sénti, sénsta “grow old”. 3.1. This distribution corresponds to the literary language. In the dialects there is considerable variation between nasal and sta-presents, which is usually interpreted as a tendency to spread sta-presents over the original domain of nasal presents.5 The spread of -sta is virtually complete for roots ending in a sibilant, which descriptively have both a nasal infix and the suffix -sta, e.g. gèsti, ge˜˛sta (< *gens-sta) “go out, die out”, mìsˇti, m˜ysˇta (< *minsˇ-sta) “mix (intr.)”. Forms like geñsa, miñsa are only occasionally found in the dialects. Among other root structures the spread of -sta has taken place only in some dialects, eventually leading to doublets in the literary language. So, from roots ending in -v, dzˇiãsta, zˇãsta belong in the literary language (beside dzˇiu¯˜ va, zˇu¯˜ va, inf. dzˇiãti “dry (intr.)”, zˇãti “perish”), but pãsta, griãsta (liter. pãti, pu¯˜ va “rot”, griãti, griu¯˜ va/griu¯˜ na “fall down”) are found only in some dialects, while sta-presents are very rare among roots ending in -j (e.g. Latv. dzît, dzîstu vs. Lith. gy´ti, g˜yja “recover”). Dialectal sta-variants are also rare among short vowel roots ending in a stop, e.g. gl˜ebsta, kniùbsta (liter. glèbti, glemba ˜ “become flabby”, kniùbti, kniumba ˜ “kneel down”). In two cases the redistribution of nasal and sta-presents seems to have taken place entirely in prehistoric times: 3.2. “Normal” zero grade is the general rule for nasal and sta-presents, but there is a large number of sta-presents with lengthened zero grade y, u¯ built to roots ending in a stop or a sibilant, e.g. rãgti, rãgsta “turn sour, ferment (intr.)” (: ráugti, ráugia “ferment (tr.)”), dy´gti, dy´gsta “sprout” (: díegti, díegia “plant”).6 The type with lengthened zero grade is particularly prominent among diphthongal roots with 5 6
See Kazlauskas (1968:324 ff.) for a detailed scenario. See Arumaa (1957:132 ff.) for a discussion of sta-presents with lengthened zero grade and an extensive list of examples.
208
Miguel Villanueva Svensson
acute intonation (due to Winter’s Law or to a laryngeal), to which in all probability it was originally restricted, later spreading to roots with circumflex intonation. For many verbs variants with normal zero grade and a nasal present are also attested, sometimes involving specialized semantics, e.g. sky´sti, sky´sta “liquify” beside skìsti, skiñda “become flimsy” (: skíesti, skíedzˇia “dilute, separate”), trãkti, trãksta “be lacking, burst” beside trùkti, truñka “last, continue” (: tráukti, tráukia “pull, draw”). As argued by Kazlauskas (1968:326 f.) and Young (2008:207 ff.), these variants are best interpreted as levellings from an earlier paradigm *sky´sti, skiñda, sky´do, *trãkti, truñka, trãko. Such a reconstruction is supported by OPr. sindats (I), syndens (II) : sidons, sidans (III) “sitzend”, see below § 5.7 3.3. sta-presents to roots ending in a liquid are well represented in the dialects, e.g. ìrti, ìrsta “disintegrate, fall to pieces”, kìlti, kìlsta “rise” (liter. y˜ ra, k˜yla). They seem to be the rule in northern Zemaitian and are the normal type of present in Latvian as well (e.g. Zem. bìrsta, Latv. bi˜rst corresponding to liter. Lith. b˜yra, inf. bìrti “pour out (intr.)”). In Auksˇtaitian nasal presents to roots in -r, -l regularly have acute intonation in the infinitive, irrespective of the original aniflt or seflt character of the root (as in bìrti, b˜yra, from *bher-). This fact is easily understood if they originally formed sta-presents, but would be difficult to explain otherwise. The assumption that roots ending in a liquid originally formed sta-presents finds firm support in the necessarily old mi˜rti, mìrsˇta “die”. We can thus set up the following rules for Proto-Baltic: T(R)ET/Sand T(R)EUT/S- roots made nasal presents, TERT/S- and TER-roots made sta-presents. 4. A formal feature not shared by nasal and sta-presents is that of metatony.8 There are cogent reasons to assume that the sta-suffix regularly triggered métatonie rude in the present stem of roots ending in a resonant. The assumption of an original restriction of the metatony to the present stem depends crucially on the contrast between pres. mìrsˇta and inf. mi˜rti “die”. Otherwise metatony has been leveled out. For the most part the acute intonation has spread to the infinitive. 7
8
The circumflex intonation of the present stem is obscure to me, but there is hardly any other way to handle the evidence. The facts on metatony in sta-presents are collected in Bu¯ga (1924:255 ff.) and Derksen (1996:285 ff.).
Baltic sta-presents and the Indo-European desiderative
209
As noticed above, this is the general rule for roots ending in a resonant, e.g. Lith. bìrti, Latv. bi˜rt “pour out (intr.)” (pres. Lith. b˜yra, dial. bìrsta, Latv. bi˜rst) beside trans. Lith. be˜rti, b˜eria, Latv. bçrt, be¸ru “strew, scatter” (also Lith. bérti, Latv. b˜ert, with acute taken from the intransitive *bìrti), Lith. ìrti, Latv. i˜rt/iˆrt “disintegrate, fall to pieces” (pres. Lith. y˜ ra/ìrsta, Latv. i˜rstu/iˆrstu) beside causative Lith. ardy´ti, a˜rdo, Latv. årdît “pull down, dismantle”. In addition, we find a number of dialectal doublets with generalized acute or circumflex intonation, as in Latv. mi˜rt, mi˜rstu and mìrt, mìrstu vs. Lith. mi˜rti, mìrsˇta “die”, Lith. rìmti, rìmsta, Latv. rimt, ˜ rimstu ˜ vs. Latv. rìmt, rìmstu “calm down” (: remti, ˜ r˜emia “support”, also rémti with acute taken from the intransitive), or Latv. grimt, ˜ grimstu ˜ vs. grìmt, grìmstu, Lith. grimti, ˜ grimsta ˜ “sink (intr.)” (cf. Latv. grèmdêt “sink (tr.)”). Métatonie rude is also found among roots of a different structure, e.g. pra-vìrkti, -vìrksta “begin to weep” (: ve˜rkti, ve˜rkia “weep”), Latv. àiz-mi˜rst, -mi˜rst beside -mìrst, -mìrst “forget” (Lith. pa-mi˜rsˇti, -mi˜rsˇta), or pa-bãgti, -bãgsta “become afraid” (: baugùs 4 “fearful”). In spite of occasional claims to the contrary (e.g. Schmid 1962:13 f., Kuryłowicz 1965:180 f.), metatony cannot be seen as a recent phenomenon. Métatonie rude in roots ending in a resonant is systematic as far back as we can trace East Baltic. Among other root structures metatony is significantly less regular and usually in competition with the expected circumflex intonation (cf. Derksen 1996:292 f.). It thus seems safe to assume that the original pattern for roots ending in a resonant spread to roots of a different structure. 9 5. The picture that emerges from East Baltic is perfectly clear. In Old Prussian, on the other hand, both nasal and sta-presents are very sparsely represented: wirst “wird” = Lith. vi˜rsti, vi˜rsta, Latv. vìrst, vìrstu “turn into, become”.10
9
10
There are some occasional cases of métatonie douce, e.g. pra-, pa-gil˜bti, -gil˜bsta beside -gìlbti “recover” (: gélbe˙ti, gélbti “save”, pagálba “help”), juõsti, juõsta beside júosti, júosta “become black” (: júodas “black”), but these are definitely rare, cf. Derksen (1996:166 ff.). OPr. wirst < *wirsta shows irregular shortening in an auxiliary verb. 1 pl. wirstmai, 2 pl. wirstai have been rebuilt on the 3rd person wirst.
210
Miguel Villanueva Svensson
1 pl. po-prestemmai “wir fühlen” (inf. is-sprestun “verstehen”), with sta-present against the nasal present of Lith. su-pràsti, -prañta, Latv. sa-prast, -prùotu “understand”.11 polinka, polijnku (inf. pola¯ikt “bleiben”),12 certainly replacing an earlier (a)thematic present (cf. dial. Lith. liñka beside liter. li˜eka, OLith. liekmi). Pres. act. ptcp. sindats (I), syndens (II), pret. act. ptcp. sidons, sidans (III) “sitzend”, pointing to a paradigm pres. *sinda- : inf. *sistwei. The agreement with OCS. seˇsti, se˛do, ˛ aor. seˇdч “sit down” points to a Balto-Slavic paradigm pres. *sínde/o- or *sìnde/o- : inf., aor. *sed-. Such a paradigm was entirely isolated in East Baltic and the present was replaced in Lith. sqsti, sqda and Latv. sêst, sêzˇu/s¸e¤stu.13 It is curious than only wirst and polinka agree with the East Baltic distribution of nasal and sta-presents. The sta-present of OPr. -prestemmai vs. Lith. -prañta is unclear to me.14 sindats : sidons probably preserves a type *señda- or *siñda : *sàsti that has been systematically leveled out in East Baltic (see above § 3.2). Apart from that, Old Prussian does little more than certifying that sta-presents are at least Proto-Baltic in date. 6. Outside of Baltic the only reasonable cognate of sta-presents is Slavic *orst ¤ o, ˛ *orstetц' , inf. *orstı' : “grow” (OCS. rasti, rasto, ˛ Ru. rastí, 0 rastú, Ukr. rosty´, rostú, Pol. rós´c´, rosna˛c´, Cz. rusti, rostu, SCr. rásti, ráste¯m, etc.), which has been frequently taken as an isolated ste-present to the root *h3er- “rise, start up” of Ved. rnóti fl “sets in motion”, Gk. 4 “raise”, - “rise”, Lat. orior, -iri “rise”, etc. The functional agreement with the Baltic sta-presents makes this etymology 11
12
13
14
The divergence in root vocalism between OPr. -sprestun and Lith. -pràsti is unclear to me. Smoczyn´ski (2005:83156) argues that this and other cases of apparent e-vocalism in Old Prussian are secondary for inherited a-vocalism. The apparent contrast in root vocalism between pres. polinka and inf. pola¯ikt has no correlate in East Baltic. Smoczyn´ski (2005:276 f.) argues for a diphthongization *palíkt > po-la¯ikt. An athematic present sqdmi also occurs, but its late attestation (Ruhig +) and the absence of the expected 1 sg. *sqmi < *sedmi indicate that it is a recent neologism. Endzelin (1923:561) suggests 3rd person *sest(i) f *sest(a) as the source of the dial. Latv. -sa¯st, but the sta-present needs not depend on an earlier athematic present. Go. frapjan “understand” may suggest that OPr. po-prestemmai and Lith. suprañta are (independent?) replacements of an earlier Áe/o-present *prat-Áa, cf. Villanueva (2006:300 f.).
Baltic sta-presents and the Indo-European desiderative
211
especially attractive, but from a formal point of view it has to face the following problems: (1) -st- is not limited to the present stem, but is part of the root, (2) the circumflex intonation of *orstı' does not agree with the Baltic metatony in roots ending in a resonant, (3) the full grade (or o-grade) of *orstı' does not agree with the regular zero grade of Baltic. None of these problems, however, is prohibitive or even too strong. If *orste- was the only remnant of a class of Balto-Slavic ste/o-presents, a pre-Slavic paradigm *or-ti, *or-ste-, aor. *or-e/o- would certainly have called for some type of analogical repair: resegmentation of the present stem as *orst-e- and extension of °st- through all the paradigm was only to be expected. The circumflex intonation may be only apparent. That of the present is expected in an Accentual Paradigm c. The final ictus of the infinitive *orstı' is regular both from *Hor-tài as well as from a hypothetical secondary *HorHtài (with -H- taken from the present *HorHste/o- at an early date), as Hirt’s law would not have applied in either case. As for the vocalism of *orsti, I believe here lies an additional argument in favor of this etymology. It is strongly reminiscent of the unexpected vocalism of the Indo-Iranian middle root aorist *ar-ta < *h3ér-to (Ved. a¯rta, inj. arta, GAv. 2 sg. imper. uz-a¯rwsˇuua¯), with full grade probably confirmed by Hitt. ar-ta “stands”,15 and TB. subj. ertär “will evoke” (Gk. τ is ambiguous). The vocalism of the present *or-ste-, then, depends on that of the aorist *Hor-to (or thematized *Hor-e-t) to which it was originally formed (see below §10.3).16 15
16
Hitt. ar-ta fits better in *h3er- than in *h1er- (Hitt. a¯ri “arrives”, Ved. rccháti, Gk. * " “goes”, etc.) and implies loss of word-initial *h3- in Hittite, a matter that cannot be discussed in this article. The prehistory of ar-ta is probably more complex than suggested here, cf. Villanueva (forthcoming:§ 3, 13.2). Recent work on the ablaut of the Indo-European middle root aorist has made the presence of full or o-grade in *Hór-to less surprising than it might traditionally have been, cf. Jasanoff (2003:150 ff.). My own views have been presented in Villanueva (2006, forthcoming). It is nevertheless a fact that zero grade has been generalized with few exceptions. The full grade of Ved. arta is thus best seen as a particular idiosyncrasy of this verb. For our present purposes it is not necessary to take a strong position on the problems surrounding the “große Bewegungswurzel *er-” (cf. most recently Vine 2005:274 ff., with literature). Whatever the choice one makes, either setting up two roots *h1er- “move (forward)” and *h3er- “rise, start up” (e.g. LIV 238, 299 f.), or only one root *h1er- “start moving” that split into *h1er- and “*h1or-” (e.g. Jasanoff 2003:212), the agreement in root vocalism between Slavic *orsti and Ved. arta remains striking.
212
Miguel Villanueva Svensson
If this account is correct (otherwise *orstı' has to be left as a locus desperatus of Slavic etymology),17 sta-presents become a Balto-Slavic rather than an exclusively Baltic problem. 7. Reasonable comparanda of the sta-presents are otherwise unknown. It must be an innovation, but it is far from self-evident where the origin of this formation should be sought. The following explanations have been proposed: 1.
Schleicher (1856:246), Porzˇezinskij (1901:99 ff.), Vaillant (1966:172 f.): resegmented from te/o-presents to roots ending in a sibilant or a dental.
The principal alternative takes *orstı' as a te-present to a root found in Lat. ˛ “steep” (< *ardharduus “high”, OIr. ard “high” (< *ard(h)Ío-), OIc. ordugr uga-), more distantly Gk. 5 %« “straight”, Ved. u¯rdhvá-, Av. wrwa- “upright” ˙ SSJ XXXII 199). This etymology is beset (references in Vasmer REW II 494, E with too many problems to be acceptable: 1) A connection with *orstı' is possible only if one sets up a “western” root *h2erHdh- and an “eastern” root *ÍerHdh- (cf. Chantraine DELG 819 for initial digamma in Gk. 5 %«). The two families, however, can hardly be separated from each other. A general “western” dissimilatory loss of initial *Í- would be curious, but cannot in principle be excluded (cf. Av. wrwa- < *rHdÍá- < *ÍrHdhÍó-, Myc. o-tu-wo-we, o-to-wo-we, o-two-we #O %ζ<ζ0« < *ζ %ζ<ζ0«, for parallels). Although conceivable in a Ío-adjective, there is no motivation for loss of initial *Í- in Slavic *orstı'. 2) The root(s) certainly contained a laryngeal, as shown by Ved. u¯rdhvá- < *ÍrHdhÍó- and Lat. arduus < *arauÍo- or *araoÍo- with vocalized internal laryngeal, *h2erdhÍo- would have given Lat. †arbuus (Greek (ζ)5 %« < *Íor[H]dhÍó- is due to Saussure’s Law; Av. wrwa- < Ir. *wrdÍá- < IIr. *[Í]rHdhÍá- is regular, cf. Cantera 2001; OIr. ard remains problematic, see the references given by Neri 2005:220 f.72–73). There is no evidence for a laryngeal in *orstı'. 3) The evidence suggests that we are dealing with derivationally isolated adjectives already within the parent language. Primary verbal formations of this root (or roots) are otherwise unknown. 4) te/o-presents are hardly represented in Balto-Slavic at all. Of the cases studied by Vaillant (1961) only two stand serious scrutiny: OCS. plesti, plete- “plait” (< *plek-te/o-, Lat. plecto¯, -ere, OHG. flehtan vs. Gk. .), certainly inherited and thus no evidence for a productivity of te-presents in Slavic, and leteˇti, leti- “fly” (< *lek-te/o-?, Lith. le˙˜kti, l˜ekia “run, fly”), for which the assumption of a te-presents is not really superior to other solutions (e.g. cross of the roots *lek- and *peth2-, or simply assuming an otherwise unknown root *let-, see the ˙ SSJ XIV 147 f.). discussion in E 17
Baltic sta-presents and the Indo-European desiderative
2.
3.
4.
5.
6.
7.
8.
9.
18
19
213
De Saussure (apud Leumann 1942:1181), Verner (apud Pedersen 1943:184): sta-presents stem directly from the Indo-European ske/o-presents. This is the most widespread theory, cf. especially Leumann (1942:113 ff.), Gorbachov (forthcoming).18 Johansson (1893:507 ff.): resegmented from the 3 sg. middle of sigmatic aorists and root aorists ending in a dental: rìmsta “calms down” < *é-r(e)m-s-to, vi˜rsta “turns into, becomes” < *é-Ürt-to. Schulze (1904[1933]:107 ff.): sta-presents derive from the IndoEuropean desiderative. The Baltic metatony is directly compared to the secondary seflt-character of Ved. kar-i-flsyá-ti “will do”, Gk. . “I will stay” < *men-e-so¯. Persson (1912:340 ff.), Brugmann (1916:370 f.): sta-presents continue an Indo-European suffix *-ste/o- (*-s- + *-te/o-), a rarer variant of *-ske/o- (*-s- + *-ke/o-).19 Stang (1942:135 ff., 1966:344), Snyder (1981): sta-presents are related to a small group of Germanic Schallverba in -st- like OHG. brestan, brastôn “burst, crack”, Go. kriustan “crackle”. Kuryłowicz (1964:50 f., 1965): a past passive participle like vi˜rstas was interpreted as *vi˜rt-stas, to which the present vi˜rsta was analogically created. The suffix -sta- then spread to other root structures. Kazlauskas (1968:332 f.): the participle *virsta in predicative usage was reanalyzed as virt + sta and reinterpreted as 3rd person present. Toporov (1973:162 ff.): related to 2nd and 3rd sg. preterit endings like Hitt. memi˘sta “said”, OCS. dastч “gave”, 2 sg. TB. -sta, TA. -flst, the Sanskrit precative, and some other forms. In addition to Leumann and Gorbachov, *-ske/o- > Baltic -sta is regarded as phonologically regular by Pedersen (1943), Hamp (1973), Kabasˇinskaite˙-Klingenschmitt (2006:18449), Klingenschmitt (2008:41720). This theory has also been (tentatively) accepted by scholars who do not regard the putative sound change *sk > st as proved or overtly assume a different development of *sk in Baltic, e.g. Klingenschmitt (1982:687), Gamkrelidze-Ivanov (1984:1211), Rasmussen (1987:151), Jasanoff (1988:236), Bammesberger (1991:275), Szemerényi (1996:2737), Forssman (2001:90, 180). Van Wijk (1933) and Kortlandt (1979:59) propose dissimilation *-ska > -sta after roots ending in -k or -g. So also Endzelin (1923:580, 1948:190), Fraenkel (1950:280 ff.), more recently Levet (1975).
214
Miguel Villanueva Svensson
7.1. Comparative evidence alone suffices to reject some proposals. te/o-presents are very rare and nowhere found with any degree of productivity. The Germanic Schallverba are evidently onomatopoeic and thus a poor comparandum for the Baltic sta-presents. The notion that *-ske/o- is to be analyzed as an etymologically composite suffix *-s- + *-ke/o- can be discussed in its own terms, but this doesn’t allow for the reconstruction of parallel suffixes like *-ste/o-.20 Leaving aside other problems, Toporov’ account is built on very doubtful premises concerning the Indo-European system of verbal endings. None of these theories, in any case, actually explains the formal and functional properties of the Baltic sta-presents. We are thus left with the “ske/o-theory”, Johansson’s “sigmatic aorist theory”, Schulze’s “desiderative theory”, and Kuryłowicz’ and Kazlauskas’ derivation from participles in *-to-, which at least start from unproblematic Indo-European formations. 7.2. Although both Johansson and Schulze offer some interesting prospects (see below), their theories are clearly unacceptable in the way they were presented. Johansson’s theory accounts for the functional value of sta-presents, but the development of an aorist formation into a present stem suffix remains entirely unclear. Schulze’s theory accounts directly for the metatony, but the characteristic anticausative value of sta-presents is left unexplained. Most importantly, Schulze was unable to explain the -t- of -sta. 7.3. Kuryłowicz’ scenario has the advantage of providing an innerBaltic account for a formation that is attested with certainty only in Baltic (if Slavic *orsti is not accepted as evidence for Balto-Slavic antiquity), but both the crucial reanalysis of *virsta- as *virt-sta- and the analogy required to obtain a new present suffix -sta- are hard to believe. A similar criticism applies to Kazlauskas’ proposal. 7.4. The “ske/o-theory” is the only one that derives the sta-presents 20
In addition to Slavic *orsti and the Germanic Schallverba in -st-, Levet (1975) presents the following evidence in support of Indo-European *-ste/o-: a) OHG. pret. konsta “knew, was able”, onsta “granted” (Otfrid, for normal konda, onda), which are almost certainly inner-German innovations (cf. Braune-Eggers 1987:300), b) some Greek verbs is -Q. like 2Q. “lift up, carry”,
Q. “doze”, c) some verbal roots ending in °st-, e.g. Ved. céflsfltati “stir”, Gk. 2 ., aor. *2 “sprout”, TB. kärst-, TA. kärflst- “cut off”, mostly of obscure etymology or analysis. It is evident that these facts do not allow the reconstruction of a new Indo-European suffix.
Baltic sta-presents and the Indo-European desiderative
215
from a well-established Indo-European present stem formation. This is an important advantage over all other theories and no doubt the main reason for its relative popularity. Although relatively straightforward from a functional point,21 this theory has to face some almost prohibitive objections: 7.4.1. The most important problem concerns the putative sound change *-sk- > -st-, which is contradicted by strong evidence like Lith. iesˇkóti, íesˇko, OCS. iskati, isko˛ “search for” (: Ved. iccháti etc., LIV 260), or Lith. skíesti, skíedzˇia “separate, dilute” (*skeid-, LIV 547 f.). The development of Indo-European *sk in Balto-Slavic remains disputed and cannot be adequately discussed in this article. I refer to Villanueva (2009) for a full treatment, where I have tried to show that *sk regularly yielded *sk. 7.4.2. Word-equations between Baltic sta-presents and Indo-European ske/o-presents are almost non-existent.22 Lith. pa-zˇìnti, -zˇ˛ísta, Latv. pa-z˜ıt, -z˜ıstu “get to know” is the synchronically expected inchoative of the stative Lith. zˇinóti, Latv. zinât “know”. A direct equation with Gk. <., Lat. (g)no¯sco¯ etc. is thus very doubtful.23 The sta-present of Lith. gìmti/gimti, ˜ gìmsta, Latv. dzìmt/dzimt, ˜ dzìmstu/dzimstu ˜ “be born” is a fairly recent replacement of OLith. g˜ema, OLatv. dz¸emu and cannot, accordingly, be equated with Ved. gácchati, Gk. 2.. The comparison of Latv. svîst, svîstu “sweat” with YAv. xvisat “id.” (e.g. LIV 607) is worthless, as both Baltic -sta and Avestan ˜ -sa- are productive formations. The Baltic sta-present may well have replaced an earlier Áe/o-present *sÜid-Áé/ó- (: OHG. swizzen, Toch. B /sya-/, perhaps late Ved. svidyati). Lith. aus ˜ ˇ ti, aus ˜ ˇ ta, Latv. àust, àust “dawns” and Ved. uccháti, YAv. usantfl “shine” < *h2us-ske/o- remains the only potentially acceptable equation. The root vocalism of aus ˜ ˇ ta must have been taken from ausˇrà “dawn” or some other nominals, but it suffices to show that aus ˜ ˇ ti cannot be linearly derived from an IndoEuropean prototype.
21
22 23
ske/o-presents have become productive in functions broadly similar to those of the Baltic sta-presents in Latin (lu¯cescere “grow light” : lu¯cere “be light”, albescere “turn white” : albere “be white”, etc.), and Iranian (YAv. twrwsaiti “becomes afraid”, tafsat “became hot”, etc.). ˜ to the contrary, e.g. Leumann (1942:119 ff.). In spite of frequent claims Cf. Villanueva (2008) on the prehistory of Lith. zˇinóti, with further arguments for taking Lith. pa-zˇìnti, Latv. pa-z˜ıt as a specifically East Baltic coinage.
216
Miguel Villanueva Svensson
In a broader context, ske/o-presents are very sparsely represented in the northern Indo-European languages. In all known cases (including Baltic!) -sk- has been resegmented as part of the root and there is no predilection for anticausative or inchoative value. This holds true for clearly inherited cases like Lith. iesˇkóti, OCS. iskati “search for”, or the Germanic denominatives OHG. eisco¯n “ask”, forsco¯n “investigate”, wunscen “wish” (: Ved. iccháti, prccháti, váñchati), as well as for far less secure possible traces of ske/o-presents like OHG. waskan “wash”, dreskan “thresh”, Lith. brqksˇti, brqksˇta “dawn; get dark (Zem.)” ~ OPol. brzeszczy sie˛ “dawns” (< *breh1g³ -ske/o-, LIV 92), Lith. blisˇkqti, blìsˇka “shine” (beside bliskqti, blizˇgqti, blizgqti) ~ OCS. blцsˇtati, blцsˇto˛ se˛ “sparkle, shine”, or Lith. pysˇkqti, py´sˇka “crack, crackle” ~ OCS. piskati “pipe”, Ru. pisˇcˇát’ “squeak”, Cz. pískati, písˇteˇti “whistle”.24 These facts suggest that ske/o-presents ceased to be productive quite early in these languages, perhaps having enjoyed a mild productivity as an iterative or intensive formation. When pared with the lack of acceptable etymologies between Baltic sta-presents and IndoEuropean ske/o-presents, it is necessary to conclude that the “ske/ o-theory” does not square well with what we know about the development of ske/o-presents in the dialectal area to which Baltic belongs. 7.4.3. Finally, the metatony is not directly accounted for by the ske/ o-theory. Most authors assume that it was extended from seflt-roots like Lith. pa-zˇ˛í sta, Latv. pa-z˜ıstu “get to know” < *zˇânsta < *g¤ nh3-ske/o-.25 Such a possibility can perhaps not be excluded, but there is no particularly obvious reason why something like this should have happened.26
24
25
26
Cf. Stang (1942:136), Vaillant (1966:338 ff., 396 ff.) for the class of expressive verbs in Lith. -ske˙ti, -ska (-sˇke˙ti, -sˇka), OCS. -skati, -sˇte-, -sˇtati, -sˇti- (with voiced variants Lith. -zge˙ti, -zˇge˙ti, OCS. -zgati, -zˇdati), mostly denoting sounds or emission of light. E.g. Klingenschmitt (1982:18629), Bammesberger (1991:276), Derksen (1996:294). I am aware of two attempts to explain the metatony directly. Rasmussen (1987:150 ff., 1997:254) derives mìrsˇta from *mr-h1-ské-ti, a ske/o-inchoative made from a stative aorist *mr-éh1-t (Lith. mìre˙). Gorbachov (forthcoming) identifies the -s- of *-ske/o- with the desiderative suffix *-(h1)s- and rewrites the suffix as *-(h1)ske/o- (*-h1s- after resonants, *-s- after stops). Both proposals simply lack comparative support (only *gwm-ské/ó-, *h1r-ské/ó- etc. can be
Baltic sta-presents and the Indo-European desiderative
217
8. In spite of the numerous attempts to explain the origin of the stapresents, it is probably safe to say that they are generally regarded as an utterly obscure and still unexplained formation. All accounts presented so far are seriously flawed in one or another respect and none has been able to embrace all facts within a single coherent framework. In what follows I will present a new proposal, which partly builds on the earlier insights of Johansson and Schulze. As Stang (1966:343 f.) observes, Schulze’s “desiderative theory” explains the metatony directly (here lies its main appeal), and it is possible to explain the inchoative value out of an earlier desiderative, but it doesn’t explain neither the -t- (a rather gross difficulty, as Schulze himself recognized), nor why sta-presents are characteristically anticausative as well as inchoative. These are precisely the two facts I wish to explain by proposing a modified version of Schulze’s theory (or, rather, a combination of Johansson’s and Schulze’s theories). In brief, I propose to see the origin of sta-presents in the 3 sg. middle of an athematic desiderative. Thus, if inherited, mìrsˇta “dies”, vi˜rsta “turns into, becomes” would go back, in the last instance, to *mr-h1s-tói “wishes to die”, *Ürt-s-tói “wishes to turn”. In order to substantiate adequately this proposal we must do two things: 1) to show that *mr-h1s-tói, *Ürt-s-tói (probably *mér-h1s-toi, *Üért-s-toi, see below § 9.3) are a perfectly acceptable starting point (§ 9), 2) to offer a reasonable scenario of how middle desideratives like these could have evolved into a Balto-Slavic class of anticausative-inchoative ste/o-presents (§ 10–11). 9. The precise reconstruction of the Indo-European desiderative is still disputed. Here I can only address the facts that have a direct bearing on the origin of the sta-presents.27
27
reconstructed). Rasmussen adduces the Latin type senesco¯ (which certainly does not prove *-h1-ske/o-), and Gk. . (usually traced back to an extended root *mneh2-, cf. LIV 447). Gorbachov tentatively adduces Ved. váñchati “desires” (h *váchati < *Ín-h1ské/ó), but the root of váñchati was almost certainly seflt and cannot be identified with the *Íen- of Ved. vanóti, Gmc. *winnan “win” (cf. Mayrhofer EWAia II 501, with references). I am basically following Jasanoff (1988:232 ff., 2003:132 ff.) in the reconstruction of the Indo-European desiderative. The Baltic future and cognate formations have been recently studied in detail by Hill (2004), to which I refer for a fuller treatment of the data and the literature.
218
Miguel Villanueva Svensson
9.1. The reconstruction of the desiderative formant as *-(h1)s(*-h1s- after resonants, *-s- after stops) can now be regarded as wellestablished. The Indo-European pattern is directly reflected in Vedic (fut. kariflsyáti “will do” vs. vakflsyáti “will say”, desider. cíkirflsati “desires to make” vs. vívrtsati “desires to turn”), Greek (fut. . “I will stretch” vs. . “I will show”), and Old Irish (fut. célaid “will conceal” < *ki-kla¯-s- < *ki-kl-h1s- vs. gigis “will pray” < *gwhi-gwhed-s-). The reconstruction of the laryngeal has often been doubted, but it would be quite incredible if Indo-Iranian, Greek, and Celtic would have independently implemented a non-trivial innovation in exactly the same phonological environment.28 9.2. It seems clear that the parent language had more than one desiderative formation. I regard the following as practically certain: a) reduplicated *gwhí-gwhn-h1se/o-, continued in the Indo-Iranian desiderative (Ved. jígha¯mm ˘˙ sati “desires to slay”, with restored nasal, cíkitsati “desires to know”), and the Old Irish reduplicated future (génaid “will kill” < *gigna¯s-, -cicherr “will put”). b) thematic *-(h1)se/o-, continued in the Greek future ((1 “I will come”, . “I will stay”), and in possible relics like Lat. uiso¯ “to go and look, visit” (: uideo¯ “see”), quaeso¯ “seek, request” (: quaero¯ “seek”). c) *-(h1)sÁe/o-, continued in the Indo-Iranian future (Ved. yakflsyá“will sacrifice”, gamiflsyá- “will go”, GAv. vaxsˇiia- “will say”), in Gaulish (pissíumí “I will see”, marcosior “I will ride”), and in the Greek relic . “I will lie, I desire to lie” (< *keÁesÁe/o- < *kei-h1sÁé/ó-).29 The future participle Ved. yakflsyánt-, YAv. bu¯sˇiiantfl has an exact match in Balto-Slavic (Lith. -siant-, RuCS. bysˇos ˛ ˇ t-/bysˇe˛sˇt- “about to be” < *-sÁont-). d) athematic, continued in the Baltic future (3rd person Lith. duõs “will give”, bùs “will be”), and in Sabellian (Um. 3 sg. ferest “feret”, 3 pl. staheren “stabunt”, Os. 3 sg. deiuast “iurabit”, 3 pl. censazet “censebunt”).30 Possible remnants in other languages include isolated
28
29 30
Whether this allomorphy is to be explained as an inner-Indo-European sound law deleting *h1 between obstruents, as per Jasanoff (2003:134 f.), or in some other way is not of direct relevance for our present purposes. Cf. Hollifield (1981:173 ff.). See Hill (2004:115 ff.), with references, on the Sabellian future and the Italic imperfect subjunctive *-se-. The Old Latin type faxo¯ “I shall do”, faxim “I might
Baltic sta-presents and the Indo-European desiderative
219
items like Ved. cáflsflte “looks at”, YAv. casˇte “teaches” (< *kwék-sor *kwek-s-, LIV 383 ff.), Hitt. gane˘ss˘ -mi “recognize” (< *g¤ nàh3-s-/ *g¤ néh3-s-, cf. Jasanoff 1988, 2003:133), and virtual neo-roots like *peh2-s- (Hitt. pa??˘s- “protect”, OCS. pasti, paso˛ “pasture”, to *peh2in Ved. páti, YAv. pa¯iti “protects”).31 9.3. It is the later type that is of direct importance for the origin of the Baltic sta-presents, as *mr-h1s-tói, *Ürt-s-tói would be the original middle counterpart of the desiderative type that actually surfaces as the future in Baltic. A middle *Ürt-s-tói would in principle demand a “normal” ablaut pattern act. *Üért-s-ti/*Ürt-s-énti. This, however, cannot be regarded as certain, as Baltic itself may point to Narten ablaut in the athematic desiderative. The Lithuanian dialects confront us with two main variants of the future paradigm, one with regular i-inflexion except for the 3rd person (1 sg. dúosiu, 2 sg. dúosi, 3rd person duõs, 1 pl. dúosime, 2 pl. dúosite, 1 du. dúosiva, 2 du. dúosita), the other with athematic plural and dual forms (1 pl. dúosme, 2 pl. dúoste, 1 du. dúosva, 2 du. dúosta).32 The 1st and 2nd singular could in principle be derived from *-sÁe/o-, but, in spite of frequent claims to the contrary, there is no possible way to derive the 3rd person and the plural and dual forms from such a suffix, nor can Stang’s “semithematic” *-si- (1942:203 f., 1966:398) be seriously entertained today. Endzelin (1948:204), Schmalstieg (1958:123 ff.), and Jasanoff (1978:105 ff.) have proposed deriving the i-inflexion from a 3 pl. *-s-nti > *-sinti, with -i- extended as a union vowel to the plural and dual endings. We would then reconstruct a (pre-)Baltic, probably Balto-Slavic paradigm *do¯smi, *do¯si, *do¯st(i),
31
32
do” probably does not belong here. See Cirilo de Melo (2007) for a convincing derivation from the subjunctive and optative of the sigmatic aorist. Whether Old Irish unreduplicated s-futures like seiss (saidid “sits”) or -ré (reithid “runs”) continue an athematic desiderative must at present be regarded as insecure. Original thematic inflexion cannot be excluded (discussion in Hill 2004:148 ff., with references). See McCone (1991:165 ff.) for an attractive connection with the Indo-Iranian type Ved. dípsa-, GAv. dizˇa- “desire to harm” (< IIr. *dhibzha- < *dhi-bh-se/o-, dissimilated from *dhi-dhbh-se/o-). I am also skeptic towards derivation of the Vedic “Prospektiv” stuflsé “I will now praise” from an athematic desiderative (Rasmussen 1985). Cf. Zinkevicˇius (1966:359 ff.), for the variants of the future in Lithuanian dialects. There are no secure traces of athematic plural forms in Latvian, cf. Endzelin (1923:659, 665).
220
Miguel Villanueva Svensson
*do¯sime, *do¯site, *do¯sint(i), with participle *do¯sÁont-. After reshaping of the 2 sg. *do¯si as *do¯sei (vel sim.), the replacement of 1 sg. *do¯smi by *do¯sÁo¯ would have been almost unavoidable.33 The athematic plural and dual variants dúosme, dúoste are easily explained as backformed to the 3rd person duõs.34 As observed by Jasanoff (1988:233, 2003:133), this explanation entails a desiderative with “Narten” ablaut for the prehistory of Baltic.35 Positive evidence confirming an athematic desiderative with “Narten” ablaut is (predictably) slim. Jasanoff (1988:233 f., 2003:133, 192) sees a direct relic in Hitt. gane˘ss˘ -mi “recognize” < *g¤ nàh3-s-/ *g¤ néh3-s-, and an indirect one in *h2íis-s- “seek”, inferred from the
33
34
35
A 1 sg. eismi “I will go” is actually attested in some hand-written additions to Sengstock’s 1612 edition of Vilentas’ writings, cf. Matusevicˇiu¯te˙ (1938:101). Given its isolation, it is unclear whether it should be taken as a strong archaism, as an occasional analogy on the present eimì, or as a mistake. So e.g. Stang (1942:204, 1966:399). This explanation has often been doubted (e.g. Endzelin 1923:664, Schmid 1963:58, Hill 2004:100 f.), but I still find it unproblematic. Baltic verbal paradigms are formed by adding personal endings to an endingless 3rd person (e.g. pres. v˜eda f 1 pl. v˜eda-me, 2 pl. v˜eda-te, pret. v˜ede˙ f v˜ede˙-me, v˜ede˙-te, etc.). There was thus enough systematic pressure to replace 1 pl. e˜ısime, 2 pl. e˜ısite with e˜ısme, e˜ıste on 3rd person e˜ıs. As observed by Kortlandt (1982:6) the East Auksˇtatian variants with accented endings 1 pl. eismà, 2 pl. eistà are found in dialects that generalize accentual mobility in the noun and are thus probably secondary. The i-inflexion can also be explained from the 3rd plural if the Baltic future is derived from the injunctive of the sigmatic aorist (e.g. Rasmussen 1985:384; similarly Kortlandt 1982:7, but taking -i- from 1 sg. *eisim < *-s-m), a possibility I find a priori unlikely. Notice, in any case, that the 3rd person duõs < *do¯st can not be taken as an argument for original secondary endings, as apocope of final *-i is well-known in Balto-Slavic. In the future apocopated *-t (< *-ti) was extended from trisyllabic (kalbe˙˜s “will speak”) to disyllabic stems (duõs for †dúosti), cf. Stang (1966:410), Jasanoff (2003:6064). If, on the other hand, athematic 1 pl. dúosme, 2 pl. dúoste are taken as original, the 1 sg. dúosiu could still be explained as analogical to 2 sg. *do¯sei, but it would be difficult to motivate the choice of i-inflexion for the renewed plural and dual endings rather than the more common ia-inflexion. I find a direct connection of 1 sg. dúosiu, 2 sg. dúosi with the Indo-Iranian sÁa-future, as recently advocated by Hill (2004:99), highly unlikely. The future participle dúosiant-, bãsiant(: RuCS. bysˇos ˛ ˇ t-, YAv. bu¯sˇiiant-) fl uncontroversially goes back to *-sÁont-, but the presence of a suppletive sia-future participle in Balto-Slavic, curious as it might be, does not by itself allow for an etymologically suppletive paradigm in the finite forms, which would be definitely aberrant.
Baltic sta-presents and the Indo-European desiderative
221
unexpected vocalism of Lith. iesˇkóti, íesˇko, OCS. iskati, isko˛ “search for” (< *h2eiske/o-), OHG. eisco¯n, Arm. haycˇem “ask” (< *h2eiske/o-), cf. Alb. njoh, Arm. aor. caneay (< *g¤ neh3-ske/o-) for a similar cross with a regular zero grade ske/o-present (Ved. iccháti, YAv. isaiti). From this perspective, Ved. cáflsflte can be derived from *kwék-s-(t)o rather than from *kwek-s-(t)ó with secondary root accent.36 The Italic evidence is basically neutral. Building on Schrijver’s account of variants like Gk. ˘ “plant” vs. “growth” as depending on accent position (Schrijver 1991:512 ff.), Hill (2004:119, 153) adduces the short vowel of Os. fust, Um. fust, fust “erit”, 3 pl. furent “erunt” < *õ-s- as evidence for original final accent, implying a desiderative *déh3-s-ti/*dh3-s-énti. I am not certain that this is the only possible explanation of Sabellian *õ-s-. Finally, Rasmussen’s interpretation of Ved. stuflsé “I will now praise” as a desiderative *stu-s-h2ái (Rasmussen 1985) would also imply normal ablaut. The type stuflsé, however, remains unclear. If it is of desiderative origin, we would certainly expect †stu¯sfl é < *stu-h1s-h2ái (as later recognized by Rasmussen himself 1997:258 f.). 10. Balto-Slavic then inherited an athematic desiderative with Narten ablaut that sooner or later evolved into the Baltic (BaltoSlavic) future tense.37 We can thus confidently postulate middle desideratives like *mér-h1s-toi, *Üért-s-toi for an early stage of the dialectal area from which Balto-Slavic emerged (the zero grade of mìrsˇta, vi˜rsta will be discussed below § 11.1).38 36
37
38
Jasanoff further observes that some of the Old Irish unreduplicated s-futures are built to roots with clear Narten affinities (e.g. *sed- “sit down” in saidid, fut. seiss, cf. Ved. sa¯dád-yoni-, OIr. sáidid “implant”, Lat. sedes “seat”, etc.), but see above footnote 31 on this formation. Needless to say, at an early stage at least the (h1)sÁe/o-desiderative must also have been present, but it is impossible to determine what the distribution of both desiderative types may have been or when they were combined into a suppletive paradigm. I attribute the paradigm reconstructed above (§ 9.3) to common Balto-Slavic. In reconstructing *mér-h1s-toi, *Íért-s-toi I am assuming that the (pre-)BaltoSlavic system of middle endings was of a typologically “modern” type: *-i rather than *-r as the marker of primary endings, 2 sg. *-so rather than *-th2e, 3 sg. *-to, 3 pl. *-nto rather than 3 sg. *-o, 3 pl. *-ro as the productive (or exclusive?) 3rd person endings. Thus, Hitt. pa?˘s-a(ri) “protects” < *péh2s-or (< *péh2-h1s-or?) may indicate that the athematic desiderative originally selected 3 sg. middle *-o,
222
Miguel Villanueva Svensson
As the next step I assume that in some verbs the original middle desiderative came to function as the normal present stem. That desideratives can become unmarked primary verbs is of course unremarkable. What is perhaps not so trivial is that in (pre-)Balto-Slavic they became not independent verbs, but a new present stem, and that this happened only in the middle voice and only in verbs with anticausative-inchoative value. It is thus necessary to examine whether anything in the evolution of the verbal system could actually favor such a process. 10.1. The origins of the productive Baltic patterns Lith. ple˙˜sti, pl˜ecˇia “widen” : plìsti, pliñta “spread (intr.)”, budqti, bùdi “be awake” : pa-bùsti, -buñda “wake up” are certainly older than Proto-Baltic. Nasal presents are also productive with anticausative-inchoative value in Slavic and Germanic. They are well-known in East and North Germanic, both involving transitivity oppositions (Go. brikan “break (tr.)” : us-bruknan “break (intr.)”, ON svída “burn (tr.)” : svidna “be burnt”), and stative : inchoative oppositions (Go. bi-leiban “stay” : aflifnan “be left over”, ON sofa “sleep” : sofna “fall asleep”). Stative : inchoative pairs are well-attested in Slavic (OCS. seˇdeˇti, seˇdi- “sit” : seˇsti, se˛de- “sit down”, bчdeˇti, bчdi- “be awake” : vчz-bч(d)noti, ˛ vчzbч(d)ne- “wake up”). Pairs involving transitivity oppositions with thematic or je-presents are not typical of Slavic, but one can quote cases like OCS. ze˛ti, ze˛be- “tear, rend” : pro-ze˛bnoti, ˛ pro-ze˛bne- “germinate, cut through”, or ve˛zati, ve˛zˇe- “bind, tie” : u-ve˛znoti, ˛ u-ve˛zne“be caught”.
but a general replacement of *-o by *-to in (pre-)Balto-Slavic would be essentially unproblematic. To be sure, it is not even absolutely certain that the Indo-European athematic desiderative possessed a middle, as desiderative formations show a remarkable defectivity as far as verbal voice is concerned (cf. Hollifield 1981:181 f.). Thus, in Greek the future tends to be inflected in the middle, while in Vedic it is usually inflected in the active, irrespective of the diathesis of the verb (e.g. Gk. : ". “suffer”, : . “bite”, Ved. mariflsyáti : mriyáte “die”, patsyáti : pádyate “fall, move”, etc.). Whether such a defectivity was present in the athematic desiderative can hardly be determined, but Hitt. pa?˘s-a(ri) or Ved. cáfls-flte suggest that it was not. If it is nevertheless supposed to have been predominantly inflected in the active, new middles could have been easily created in the prehistory of Balto-Slavic by supplying oppositional middles to active desideratives and by replacing active with middle desideratives in typically or exclusively middle verbs.
Baltic sta-presents and the Indo-European desiderative
223
The constitution of a class of anticausative-inchoative nasal presents is of course a major isogloss uniting these languages. The type has been recently studied in detail by Gorbachov (2007), where it is shown that all three branches presuppose a common, “northern Indo-European” prototype *li-n-kw-é-ti, *li-m-p-é-ti etc. The original present morphology, characterized by a nasal infix and thematic inflexion, has been faithfully preserved in Baltic. In Germanic and Slavic the nasal infix has been replaced by a nasal suffix *-ne/o-, the original infix being preserved only in relics.39 The original independence of the present stem is also most clearly preserved in Baltic (Lith. pres. li-m-p-a ˜ : inf. lìp-ti “stick to” vs. OCS. pri-lц(p)-ne-tч : pri-lц(p)-no-ti ˛ “cleave to”, Go. af-lif-ni-p : af-lif-na-n “remain”), but it is Slavic that surveys the original aorist. Slavic nasal presents of the type se˛de-, vчzbч(d)ne-, pri-lц(p)ne- are regularly associated with thematic aorists (OCS. seˇde-, vчz-bчde-, pri-lцpe-), thus pointing to an earlier paradigm pres. *bhundhéti : aor. *bhudhét “wake up”. As frequently assumed (e.g. Stang 1942:190, 1966:340, 378 ff.), this must have been the paradigm inherited by Baltic, where the thematic aorist was replaced with the a¯-preterit. There is no reason not to assume that it was the paradigm inherited by Germanic as well, where the aorist has been regularly lost. 10.2. A class *bhundhéti : *bhudhét is thus practically certain for Balto-Slavic, and at least very probable for “northern” Indo-European. Even more important than its formal expression is the fact that there must have been an early trend to create an independent class of anticausative-inchoative verbs. These were typically opposed to transitive or stative verbs, but primary verbs with inherent anticausative-inchoative value would have naturally tended to acquire their morphology as well. Although the type *bhundhéti : *bhudhét must have become productive quite early, this was surely not the only morphology allowed in this class. The thematic aorist was the regular aorist formation of this class, irrespective of the present stem, but other types can also be reconstructed (perhaps to the exclusion of the sigmatic aorist). Lith. stóti, 39
Infixal presents are still preserved in Go. standan, pret. stop “stand” and OCS. seˇsti, se˛do˛ “sit down”, lesˇti, le˛go˛ “lie down”, byti, bod ˛ o˛ “will be”, ob-reˇsti, -re˛sˇto˛ “find”. More relics of infixal nasal presents are discussed in Gorbachov (2007:21 ff., 109 ff.). Cf. also Vaillant (1966:185 f.).
224
Miguel Villanueva Svensson
stója, stójo, OCS. stati, stano, ˛ staxч “stand up” (: stative Lith. stovqti, stóvi, OCS. stojati, stoji- “stand”) evidently point to an inherited root aorist *sta¯-t < *stéh2-t. OCS. pres. Zo., Mar. u-mцretч (Slvn. mrjèm, otherwise replaced by u-mцre-) : aor. u-mreˇ(tч) “die” stems directly from Indo-European *mr-Áé-tor : *mér-t (Ved. mriyáte, Lat. morior : OHitt. merzi, medialized Arm. mer¯aw, Ved. ámrta), with preserved active root aorist. At the stage we are considering, OCS. rasti, rasto˛ “grow” may still have presented a middle root aorist *Hor-to (Ved. a¯rta, Gk. τ ), rather than a thematized *Hor-e/o-. As for the present stem, Tedesco (1948) established an older layer of Slavic intransitive je-presents like OCS. po-gyble- “perish”, i-sчsˇe“dry (intr.)”, RuCS. u-sчple- “fall asleep”, soon replaced by the nepresents po-gybne-, i-sчxne-, u-sч(p)ne-. Just like ne-presents, je-presents of this type are regularly pared with thematic aorists (po-gybe-, i-sчxe-, u-sчpe-). The pattern pres. po-gyble- : aor. po-gybe- is directly inherited from the parent language, the thematic aorist being for an earlier (middle) root aorist. It is best represented in Indo-Iranian, where intransitive ya-presents remained productive (Ved. mányate : ámata “think”, búdhyate : ábodhi “awake”, t´rsfl yati : trflsa¯náfl RV, trflsat AV “become thirsty” etc.). Intransitive Áe/o-presents are also attested in Germanic (ON liggja “lie”, sitja “sit”, Go. paursjan “be thirsty”, frapjan “understand”, etc.),40 and can probably be assumed for the prehistory of Baltic as well.41 It is evident that at an early date Áe/o-presents with anticausativeinchoative value must have enjoyed a certain productivity in the northern Indo-European area. Unlike in Indo-Iranian, however, Áe/ o-presents were not specialized in this usage alone. They were thus unfit to serve as the characteristic present stem of the emerging class of anticausative-inchoative verbs. Nasal presents of the type *bhundhéti eventually came to acquire such a status. The conversion of the nasal infix into a ne/o-suffix in Slavic and Germanic allowed for the 40
41
I regard the stative value of verbs like *ligjan “lie” or *sitjan “sit” as secondary, derived through decomposition from earlier inchoatives. The Baltic evidence is only inferential. If the Baltic e-preterit goes back to *-iÁa¯, an a¯-preterit built to *(i)Áa-presents (cf. Villanueva 2005 for a defense of this view), unexpected intransitive e-preterits like Lith. mìre˙ (mi˜rti, mìrsˇta “die”), gìme˙ (gimti, ˜ gìmsta “be born”), gùle˙ (gul˜ti, gùla “lie down”), or púole˙ (pùlti, púola “fall”) can be explained as built to an earlier Áe/o-present (cf. Villanueva 2005:248 ff., building on Barton 1980:257 ff.).
Baltic sta-presents and the Indo-European desiderative
225
creation of nasal presents to roots like *terp- (OCS. u-trц(p)noti, ˛ Ru. terpnút’ “grow numb”) or *ters- (Go. ga-paursnan “dry up”), but as long as the nasal infix was still kept as an infix the possibility to form a nasal present must have been constrained by root-structure, as it is still the case in Lithuanian. Other formations like simple thematic presents may have occasionally been found among anticausative-inchoative verbs, but none of them was selected to serve as a companion of the nasal presents.42 10.3. There was thus a gap to be filled. This provided a sufficiently strong motivation for the middle desiderative to become the unmarked present stem of some verbs belonging to the emergent anticausative-inchoative class. A root like *mers- “forget”, for instance, formed a middle root aorist *mórs-e/*mrs-ré (Ved. mrflstha¯s, TB. subj. ma¯rsam, fl inf. marsatsi), which would have evolved into a thematic aorist *mrs-é/ó- (Lith. pa-mi˜rsˇo). Both the Vedic ya-present mrflsyate and the Tocharian “duhé-present” TB. märsetär (< *mrs-ór) could in principle be old, but “duhé-presents” were specialized with stative value in Germanic (Class III of weak verbs) and Balto-Slavic (ì-presents), where in addition they are regularly pared with a second stem *-e-. A possible present †mirsˇi- beside the aorist *mrs-é/ó- was thus unacceptable, whereas *mrs-Áé/ó-, if really Indo-European and not just Indo-Iranian, may well have been dismissed in favor of a more expressive desiderative *mers-s-toi “desires to forget” f “forgets”. Similar processes may be envisaged for other verbs. Thus, *terp-s-toi “desires to grow numb” (Lith. ti˜rpti, ti˜rpsta “melt, grow numb”) may have become the present of the aorist *trp-é/ó- (OCS. u-trцpe-), once the
42
The Indo-European origins of the formations involved in the northern IndoEuropean “anticausative-inchoative” system cannot be discussed in detail in this article. The thematic aorist has been plausibly derived from a 3 sg. middle *-e recharacterized with *-t (*dh3-é f *dh3-é-t > Ved. á-dat “took”), cf. Watkins (1969:100 ff.), Hollifield (1977:57 ff.), more recently Peters (2004:268 f.). In my view the zero grade of *dh3-é and other middle root aorists has replaced an earlier ablauting paradigm *dóh3-e/*dh3-ré, cf. Villanueva (2006 and forthcoming).The Indo-European pedigree of intransitive Áe/o-presents is perfectly clear. The productive present type *bhundhéti is notoriously problematic. The main proposals are referred in Meiser (1993:292 f.), to which Gorbachov’s “h2e-conjugation” nasal present type (3 sg.) *bhu-n-dh-é should now be added (Gorbachov 2007:212 ff.). I would favor a middle origin in combination with secondary association with thematic aorists.
226
Miguel Villanueva Svensson
old present *trp-ór (OCS. trчpeˇti, trчpi- “suffer, endure”, Lat. torpeo, -ere “be numb”) had become part of a new paradigm *tirp-i- : *tirp-e-. The desiderative *h1rem-h1s-toi “desires to calm down”, from *h1rem“calm down” (Lith. rìmti, rìmsta), may likewise have replaced an earlier type of present (cf. Ved. rámate, YAv. ra¯mo¯iwm, which may also continue a root aorist subjunctive). Similarly, *h3er-h1s-toi “desires to ˛ may have replaced some of the presrise (> grows)” (OCS. rasti, rasto) ents of this root (LIV 299f. lists *h3i-h3(é)r-, *h3r-n(é)u-, and *h3r-Áé-). There is no point in insisting on the correctness of these examples. The original core of verbs that transformed middle desideratives like *mér-h1s-toi, *Üért-s-toi into unmarked presents is simply unrecoverable. It is equally impossible to determine whether this was a northern Indo-European or just a Balto-Slavic innovation. No examples are known in Germanic, but here, as in Slavic (where *orste- is the only relic of a ste-present), the creation of a suffix *-ne/o- and the general restructuring of the verbal system may easily have led to their total elimination. In what follows I will assume that it was a specifically Balto-Slavic innovation. 11. It is thus perfectly reasonable, I believe, to postulate pre-BaltoSlavic paradigms like pres. *mérs-s-toi : aor. *mrs-é-t “forget”, *térps-toi : *trp-é-t “grow numb”, *h1rém-h1s-toi : *h1rm-é-t “grow calm”, *Hór-h1s-toi : *Hór-to “rise”. The next question is how the present stems *mrs-sté/ó-, *trp-sté/ó-, *h1rm-h1sté/ó-, *Hor-h1sté/ó- were created out of full-grade athematic middle desideratives. 11.1. The zero grade is essentially unproblematic. Once established in a paradigm with thematic aorists, and once they were felt to belong to the class of *bhu-n-dh-é-ti : *bhudh-é-t, *lip-Áé-toi (or *lip-Áé-ti) : *lip-é-t etc., presents like *térp-s-toi simply adopted the vocalism of the aorist *trp-é-t, which typically displayed zero grade of the root. The principle was also followed when, for whatever reason, the aorist had a different vocalism (*Hor-h1s-toi to *Hor-to or *Hor-e/oin Slavic *orste-, Lith. témti, témsta “grow dark” perhaps to an active aorist *tém-t, etc.). 11.2. Adoption of the zero grade (recte, of the vocalism of the aorist) may have taken place before or after *térp-s-toi / *trp-s-tói gave rise to a present stem *térp-ste/o- / *trp-sté/ó-. The resegmentation involved is the most serious difficulty that this account of the Baltic sta-presents has to face, and the solutions one may offer can only be regarded as tentative. Clearly the process must have been something
Baltic sta-presents and the Indo-European desiderative
227
like *trpstói f *trpstóti f *trpstéti. The 3 sg. middle was recharacterized with the 3 sg. active *-ti. The resulting *trp-s-to-ti was reanalyzed as *trp-sto-ti, with suffix -st- and “thematic” o-vowel. The rest of the paradigm was then rebuilt on the 3rd singular (1 sg. *trp-sto¯ or *trpsto-mi, 2 sg. *trp-sto-si, etc.). Finally, the aberrant paradigm with persistent union vowel *-o- was adapted to the regular thematic inflexion, giving *trp-ste/o-, *Hor-h1ste/o- from which Lith. ti˜rpsta, OCS. rastetч regularly derive. 11.3. It is of course impossible to prove that this is what actually happened, but it is at least possible to give parallels for the different steps involved. That verbal paradigms are frequently remade from the third singular has been well-known since the principle was explicitly formulated by Watkins (1962:90 ff.). The redundant addition of active to middle endings has clear parallels like Ved. impf. ás´ayat, ás´eran “lay”, áduhat, áduhran “gave milk”, Ved. 2 sg. -tha¯s (probably *-th2e + *-e-s), Lat. 2 sg. -ris, -rus (*-so + *-s), or Gk. 1 sg. - 0 (< *-ma¯ < *-m-h2 + *-m). In some cases the addition of redundant morphological material has led to the constitution of a new paradigm.43 Thus, the thematic aorist has often been assumed to be based on a 3 sg. middle *-e recharacterized with *-t.44 The paradigm of the Germanic 3rd class of weak verbs (Go. mun-a, -ais, -aip, -am, -aip, -and “intend”) has been explained as based on a middle 3 sg. *munai < *mn(n)-ói recharacterized with the active 3 sg. *-pi.45 Finally, “thematic” paradigms with persistent *-o-, found exclusively in the middle voice, are attested in Hittite (?arratta(ri) “persishes”, tarratta “be able, can”, Üe?atta “turns (intr.)”, par?attari “chases”, etc.), Tocharian (deponent presents of class III and IV, e.g. TB. triketär, TA. trikatär “goes astray”, TB. orotär, TA. aratär “ceases” < *-o-tor), and Gothic (passive 1/3 sg. bairada, 2 sg. bairaza, 1/2/3 pl. bairanda). At least the Hittite and Tocharian paradigms almost certainly arose through renewal of athematic middles with 3 sg. ending *-or as *-o-tor.46 43
44 45 46
To be sure, most of the parallels I give are somewhat controversial. This is not the place to argue at length why I consider the explanations adopted here for the thematic aorist or the Germanic 3rd class of weak verbs to be preferable to other proposals. See above footnote 42. Cf. Jasanoff (1978:74 f., 2002–03:156 f.). Cf. Watkins (1969:86f.), Yoshida (2007), for Hittite, and Jasanoff (2002–03:158f.), for Tocharian. Whether the persistent -a- of the Gothic passive is to be explained
228
Miguel Villanueva Svensson
11.4. These formations, however, do not provide an exact parallel for the sta-presents. They are all based on a 3 sg. middle *-e or *-o, obsolete endings that were prone to be recharacterized and reanalyzed in several ways (if they were not simply lost). *trp-s-tói, on the other hand, contained the 3 sg. middle *-to, a productive ending neither in need of morphological repair nor susceptible of reanalysis as part of the stem. Both its recharacterization as *trp-s-tó-ti and the resegmentation of *-st(o)- as a stem marker seem thus to lack a clear motivation – at least as long as the middle voice was a living category of the language. The last caueat, however, is important, as the whole process makes more sense if it is seen in connection with the decay of the middle voice in Balto-Slavic. The details of the loss of the middle voice in Balto-Slavic are of course unknown, but it is reasonable to suppose that it was a gradual process. Middle endings would be specialized in some particular usage (e.g. as a passive, as in Germanic, or with “internal”, self-benefactive values, as in Indo-Iranian), other functions of the Indo-European middle voice being taken over by new formations or syntactic constructions. Thus, the evolution of the ancient desiderative into a future tense was perhaps accompanied by generalized active inflexion. Middle desideratives like *térp-s-toi, *Hór-h1s-toi thus became free, so to speak, to be recategorized as a present stem. The anticausativeinchoative class to which they now belonged was sufficiently characterized by its distinctive present stem affixes and thematic aorist to have tended to be inflected in the active alone. The active inflexion of the thematic aorist was probably a consequence of its origin in middle root aorists with 3 sg. *-e f *-e-t. Similarly, in Vedic thematic aorists are almost exclusively inflected in the active, in spite of their frequent middle functional value. Whatever explanation one may favor for the “northern Indo-European” present type *bhu-n-dh-é-ti, it is very probable that it was typically or exclusively inflected in the active. Intransitive Áe/o-presents were probably both active and middle in IndoEuropean, but it is reasonable to suppose that middle Áe/o-presents tended to be inflected in the active quite early. A parallel is again provided by Vedic, where active ya-presents are typically pared with from a thematic 3 sg. middle *bher-oi, as per Watkins (1969:213) and Jasanoff (2003:49), is less clear to me. See Cowgill (1985) for a plausible explanation in terms of reduction of morphological complexity in moribund categories.
Baltic sta-presents and the Indo-European desiderative
229
thematic aorists against middle ya-presents pared with “passive” aorists (cf. Kulikov 2001:558 ff.). Under these conditions, it was almost unavoidable for a paradigm like *trp-s-(m)ái, *trp-s-sói, *trp-s-tói to be activized in one way or another. If this took place at a time when the middle voice was already on the verge of disappearing, *trp-s-tói f *trp-stó-ti may well have been an activizing strategy with the additional advantage of obtaining a distinctive suffix to serve as the companion of nasal presents in the anticausative-inchoative class of verbs.47 12. Summing up, I believe a revised version of Schulze’s “desiderative theory” can account for all facts associated with the Baltic stapresents. The sta-suffix is not an exclusively Baltic formation, but goes back to a Balto-Slavic class of ste/o-presents presumably in a distribution with nasal presents broadly similar to that reconstructable for Proto-Baltic. *orsti, *orste- “grow” is the only remnant of this class of presents in Slavic, the general elimination of ste-presents being a consequence of the nasal infix having become a suffix *-ne/o- that could now be made from every verb regardless of its root structure. The desiderative origin of sta-presents explains directly the métatonie rude in roots ending in a resonant. The root athematic desiderative behind sta-presents is not only a reasonable construct from an Indo-European point of view, but is actually demanded by an internal analysis of the Baltic future. The assumption that it was the middle of the desiderative that gave rise to the sta-presents accounts for their anticausative and inchoative value. The conversion of some middle desideratives into a new present stem can be motivated within the general
47
Other scenarios can of course not be excluded. The rest of the paradigm, for instance, could have been thematized from the 3 sg. *-(h1)s-toi, with or without being immediately activized, yielding 1 sg. *trp-ste-(m)ai or *trp-sto¯, 2 sg. *trpste-soi or *trp-ste-si, 3 sg. *trpstoi, the 3rd sg. being later reshaped as *trpsteti. Another possibility would be assuming that the loss of the middle voice took place independently in Baltic and Slavic and that the paradigm *tirp-s-(m)ai, *tirp-s-sai, *tirp-s-tai was activized after *-a- (< *-o-) had been generalized as the thematic vowel in Baltic. Resegmentation of a suffix *-sta- would then be even easier. I am not certain, however, how the crucial 3rd person *tirpsta could be derived from *tirpstai, as a possible middle secondary ending *tirp-s-ta can hardly be supposed to have played any role at this stage. In addition, Slavic *orste- would have to be taken as a parallel but independent innovation, which I find a priori unattractive.
230
Miguel Villanueva Svensson
trends of the evolution of the verbal system in the northern IndoEuropean languages. Finally, the crucial recharacterization of the 3rd sg. *-(h1)s-toi as *-(h1)s-to-ti and its subsequent reanalysis as *-(h1)sto-ti (f *-(h1)ste-ti) can be seen as a particular offshoot of the general loss of the middle voice in Balto-Slavic.
References Arumaa, Peeter (1957), “Von der Eigenart des Ablauts und der Diathese im Baltischen”. ZslPh. 26: 118–149. Bammesberger, Alfred (1991), “Zur Vorgeschichte von lit. kláusia und mìrsˇta”. HS. 104: 269–278. Barton, Charles R. (1980), “Notes on the Baltic Preterite”. IF. 85: 246–278. Braune, Wilhelm, Hans Eggers (198714), Althochdeutsche Grammatik. Tübingen: Niemeyer. Brugmann, Karl (1916), Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. II 3. 2. Bearbeitung. Strassburg: Trübner. Bu¯ga, Kazimieras (1924), “Die Metatonie im Litauischen und Lettischen”. KZ. 52: 250–302. Cantera, Alberto (2001), “Die Behandlung der idg. Lautfolge (C)RHC- im Iranischen”. MSS. 61: 7–27. Chantraine DELG = Pierre Chantraine, Dictionnaire étymologique de la langue grecque. Histoire des mots. I–IV. Paris: Klincksieck, 1968–1980. Cirilo de Melo, Wolfgang David (2007), “The sigmatic future and the genetic affiliation of Venetic: Latin faxo¯ “I shall make” and Venetic vha.g.s.to “he made””. TPhS. 105: 1–21. Cowgill, Warren (1985), “Loss of morphophonemic alternation in moribund categories, as exemplified in the Gothic verb”. In: U. Pieper, G. Stickel (eds.), Studia linguistica diachronica et synchronica. Werner Winter sexagenario anno MCMLXXXIII. Berlin: de Gruyter, 145–149. Derksen, Rick (1996), Metatony in Baltic. Amsterdam-Atlanta: Rodopi. Endzelin, Ja¯nis (1923), Lettische Grammatik. Heidelberg: Winter. –, (1948), Baltu valodu skan¸as un formas. Riga: Latvijas valsts izdevnieciba. ˙ SSJ = Oleg Nikolajevicˇ Trubacˇev (ed.), E ˙ timologicˇeskij slovar’ slavjanskix jazyE kov. Moscow: Nauka, 1974-. Forssman, Berthold (2001), Lettische Grammatik. Dettelbach: J. H. Röll. Fraenkel, Ernst (1950), “Zum baltischen und slavischen Verbum”. ZslPh. 20: 236–320. Gamkrelidze, Tamaz V., Vjacˇeslav Vsevolodovicˇ Ivanov (1984), Indojevropejskij jazyk i indojevropejcy. I. Tbilisi: Izdatel’stvo tbilisskogo universiteta. Gorbachov, Yaroslav (2007), Indo-European Origins of the Nasal Inchoative Class in Germanic, Baltic and Slavic. Ph.D. Dissertation, Harvard University. –, (forthcoming), “The origin of the Baltic inchoative in -sta- and a new Proto-Baltic sound law”.
Baltic sta-presents and the Indo-European desiderative
231
Hamp, Eric P. (1973), “North European “1000””. In: Cl. Corum, T. C. Smith-Stark, A. Weiser (eds.), Papers from the Ninth Regional Meeting of the Chicago Linguistic Society, April 13–15, 1973. Chicago: Chicago Linguistic Society, 172–178. Hill, Eugen (2004), “Die sigmatischen Modus-Bildungen der indogermanischen Sprachen. Erste Abhandlung: Das baltische Futur und seine Verwandten”. IJDL. 1: 69–171. Hollifield, Patrick Henry (1977), On the System of Conjugation in Indo-European. Ph.D. Dissertation, Harvard University. –, (1981), “Homeric . and the Greek Desideratives of the Type ”. IF. 86: 161–189. Jasanoff, Jay H. (1978), Stative and Middle in Indo-European. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. –, (1988), “PIE *gˆne- “recognize, know””. In: A. Bammesberger (ed.), Die Laryngaltheorie und die Rekonstruktion des indogermanischen Laut- und Formensystems. Heidelberg: Winter, 227–239. –, (2002–03), ““Stative” *-e- revisited”. Sprache 43: 127–170. –, (2003), Hittite and the Indo-European Verb. Oxford-New York: Oxford University Press. Johansson, Karl Ferdinand (1893), “Eine analoge neubildung der verbalflexion im aind. und balt.-slavischen”. KZ. 32: 434–513. Kabasˇinskaite˙, Birute˙, Gert Klingenschmitt (2006), “Isˇ lietuviu˛ kalbos zˇodzˇiu˛ istorijos: 3. Lie. kùrmis”. Baltistica 41: 169–186. Kazlauskas, Jonas (1968), Lietuviu˛ kalbos istorine˙ gramatika. Vilnius: Mintis. Klingenschmitt, Gert (1982), Das altarmenische Verbum. Wiesbaden: Reichert. –, (2008), “Lit. úosˇvis”. Baltistica 43: 405–430. Kortlandt, Frederik (1979), “Three Problems of Balto-Slavic Phonology”. ZbFl. 22/2: 57–63. –, (1982), “Innovations which betray archaisms”. Baltistica 18: 4–9. Kulikov, Leonid (2001), The Vedic -ya-presents. Dissertation, Leiden University. Kuryłowicz, Jerzy (1964), The inflectional categories of Indo-European. Heidelberg: Winter. –, (1965), “Les présents lituaniens en -sta-”. BPTJ. 23: 175–182. Leumann, Manu (1942), “Idg. sk im Altindischen und im Litauischen”. IF. 58: 1–26, 113–130. Levet, Jean-Pierre (1975), “Les présents en *-st- de l’indo-européen: les données tokhariennes”. BSL. 70: 91–114. LIV = Lexicon der indogermanischen Verben. Die Wurzeln und ihre Primärstammbildungen. Unter Leitung von Helmut Rix und der Mitarbeit vieler anderer bearbeitet von Martin Kümmel, Thomas Zehnder, Reiner Lipp, Brigitte Schirmer. Zweite, erweiterte und verbesserte Auflage bearbeitet von Martin Kümmel und Helmut Rix. Wiesbaden: Reichert, 2001. Matusevicˇiu¯te˙, Izabele˙ (1938), “Prierasˇai Berlyno valstybine˙s bibliotekos Sengstacko originale”. Archivum Philologicum 7: 88–101. Mayrhofer EWAia = Manfred Mayrhofer, Etymologisches Wörterbuch des Altindoarischen. I–III. Heidelberg: Winter, 1986–2001.
232
Miguel Villanueva Svensson
McCone, Kim (1991), The Indo-European Origins of the Old Irish Nasal Presents, Subjunctives and Futures. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. Meiser, Gerhard (1993), “Zur Funktion des Nasalpräsens im Urindogermanischen”. In: G. Meiser (ed.), Indogermanica et Italica. Festschrift für H. Rix. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck, 280–313. Neri, Sergio (2005), “Riflessioni sull’apofonia radicale di proto-germanico *namo¯n “nome””. HS. 118: 201–250. Pakalnisˇkiene˙, Dalia (1993), Lietuviu˛ kalbos intarpiniai ir sta kamieno veiksmazˇodzˇiai. Dissertation, Vilnius University. Pedersen, Holger (1943), “Et baltoslavisk problem”. In: R. Brøndal, V. Brøndal, Chr. Møller, H. Olsen (eds.), In memoriam Kr. Sanfeld. Udgivet paa 70-aarsdagen for hans fødsel. København, Gyldendalske Boghandel Nordisk Vorlag: 184–194. Persson, Per (1912), Beiträge zur indogermanischen Wortforschung. Uppsala: Akademiska bokhandeln. Peters, Martin (2004), “On Some Greek nt-Formations”. In: J. H. W. Penney (ed.), Indo-European Perspectives: Studies in Honour of Anna Morpurgo Davies. Oxford-New York: Oxford University Press, 266–276. Porzˇezinskij, Viktor-Jan Karlovicˇ (1901), K istorii form sprjazˇenija v baltijskix jazykax. Moskva: Universitetskaja tipografija. Rasmussen, Jens Elmegård (1985), “Der Prospektiv – eine verkannte indogermanische Verbalkategorie?”. In: B. Schlerath, V. Rittner (eds.), Grammatische Kategorien, Funktion und Geschichte. Akten der VII. Fachtagung der Indogermanischen Gesellschaft. Wiesbaden: Reichert, 384–399. –, (1987), “Miscellaneous Morphological Problems in Indo-European Languages IV”. APILKU. 6: 145–156. –, (1997), “Processes of grammaticalization in Indo-European verbal derivation”. In: A. Lubotsky (ed.), Sound Law and Analogy. Papers in honor of Robert S. P. Beekes on the occasion of his 60th birthday. Amsterdam-Atlanta: Rodopi, 249–262. Schleicher, August (1856), Litauische Grammatik. Prag: J. G. Calve’sche Verlagsbuchhandlung. Schmalstieg, William R. (1958), “The Vocalism of the Lithuanian Sigmatic Future”. SEEJ. 16: 120–129. Schmid, Wolfgang P. (1962), “Lit. kláusti “fragen” und das altind. Futurum”. IF. 67: 1–15. –, (1963), Studien zum baltischen und indogermanischen Verbum. Wiesbaden: Harrassowitz. Schrijver, Peter (1991), The Reflexes of the Proto-Indo-European Laryngeals in Latin. Amsterdam-Atlanta: Rodopi. Schulze, Wilhelm (1904), “Lit. kláusiu und das indogermanische Futurum”. [quoted from: Kleine Schriften. Göttingen: Vandenhoeck & Ruprecht, 1933: 101–109]. Smoczyn´ski, Wojciech (2005), Lexicon der Altpreussischen Verben. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. Snyder, William H. (1981), “Das st-Verbalsuffix im Baltischen und Germanischen”. Sprachwissenschaft 6: 275–281.
Baltic sta-presents and the Indo-European desiderative
233
Stang, Christian S. (1942), Das slavische und baltische Verbum. Oslo: Dybwad. –, (1966), Vergleichende Grammatik der baltischen Sprachen. Oslo-BergenTromsö: Universitetsforlaget. Szemerényi, Oswald (1996), Introduction to Indo-European Linguistics. OxfordNew York: Oxford University Press. Tedesco, Paul (1948), “Slavic ne-Presents from older je-Presents”. Language 24: 346–387. –, (1956), review of L. Sadnik, R. Aitzetmüller, Handwörterbuch zu den altkirchenslavischen Texten. Language 32: 518–526. Toporov, Vladimir Nikolajevicˇ (1973), “Neskol’ko zamecˇanij o baltijskix glagolax na -sta v svjazi s proisxozˇdeniem e˙togo formanta (indojevropejskaja perspektiva)”. LKK. 14: 151–167. Vaillant, André (1961), “Les présents slaves en -te-”. BSL. 56: 15–20. –, (1966), Grammaire comparée des langues slaves. Tome III. Le verbe. Paris: Klincksieck. Van Wijk, Nicolas (1933), “Der Ursprung der litaulettischen Präsensklasse mit Formans -sta-”. Archivum Philologicum 4: 57–63. Vasmer REW = Max Vasmer, Russisches etymologisches Wörterbuch. I–III. Heidelberg: Winter, 1953–1958. Villanueva Svensson, Miguel (2005), “The Baltic e-Preterit revisited”. Baltistica, VI Priedas: 239–252. –, (2006), “Traces of *o-Grade Middle Root Aorists in Baltic and Slavic”. HS. 119: 295–317. –, (2008), “Lithuanian zˇinóti “to know””. Baltistica 43: 175–199. –, (2009), “Indo-European *sk in Balto-Slavic”. Baltistica 44: 5–24. –, (forthcoming), “Indo-European Middle Root Aorists in Anatolian”. Sprache. Vine, Brent (2005), “Remarks on Rix’s Law in Greek”. JIES. 33: 247–290. Watkins, Calvert (1962), Indo-European Origins of the Celtic Verb, I. The Sigmatic Aorist. Dublin: Institute for Advanced Studies. –, (1969), Indogermanische Grammatik. III/1. Geschichte der indogermanischen Verbalflexion. Heidelberg: Winter. Yoshida, Kazuhiko (2007), “The Morphological History of Hittite Mediopassive Verbs”. In: A. J. Nussbaum (ed.), Verba Docenti. Studies in historical and IndoEuropean linguistics presented to Jay H. Jasanoff by students, colleagues, and friends. Ann Arbor: Beech Stave Press, 379–395. Young, Steven (2008), “Winter’s law and etymologies, with special reference to Lithuanian”. Baltistica 43: 201–218. Zinkevicˇius, Zigmas (1966), Lietuviu˛ dialektologija. Vilnius: Mintis.
Vilnius University Perku¯nkiemio 13–84, LT-12114 V i l n i u s Lithuania [email protected]
M i g u e l Vi l l a n u e v a S v e n s s o n
234
Andreas Willi
Hera, Eros, Iuno Sororia Abstract This article aims to provide the name of Hera, the Greek goddess of brides, weddings, and marriage, with an etymology that is satisfactory both from a historical and from a linguistic point of view. Earlier suggestions are rejected because they either disregard the central features of Hera’s character or fail to do justice to the Mycenaean evidence favouring an etymon *Serah2. The latter is here explained as a well-formed vflrddhi derivative of an Indo-European root noun *sor(H)-/*ser(H)‘woman, female’, which has long been posited for independent reasons and which may belong with the verbal root *ser(H)- ‘to attach’ suggesting an original notion of ‘female partner’. Since other derivatives of this root *ser(H)- may irregularly show no initial aspiration in Greek, a further connection with the name of the love god Eros is possible via an s-stem derivative *ser(H)-(o)s. Outside Greece, apart from a possible link to the Indian Goddess Sarasvati, the same formation seems to underly the epithet of the Roman divinity Iu¯no¯ Soro¯ria, which – given the goddess’s function as a protector of adolescent young women – can have nothing to do with the Latin word for ‘sister’, but must also refer to the young women’s newly acquired ability and wish to enter a marital union.
1. Hera in archaeology, history of religion, and literature The goddess Hera is one of the major divinities of the Greek pantheon, but in some respects she is also a very elusive one. In the popular imagination she is primarily known as the wife of Zeus, but she also has a distinctive personality of her own; after all, Zeus himself is defined in a Homeric formula as the ‘loud-thundering husband of Hera’ (( « « 6H 0«). Archaeologically, Hera is among the first Greek divinities to be honoured by monumental temple buildings, as at Samos, Argos, and also Olympia, where the Heraion is even the oldest great sanctuary of the local cult district.1 Leaving aside the colonial area (Kroton, Paestum), the most famous Heraia were all located on the Greek islands
1
Cf. Kerényi (1972: 108), who detects here a ‘Vorliebe der Göttin für das Haus’, Burkert (1985: 131–132), Graf (1998: 358); Hera’s sanctuaries are one of the foci in de la Genière (1997).
Hera, Eros, Iuno Sororia
235
and on the Peloponnese, not for instance in Athens or in the North (Thessaly). This observation is in line with Hera’s own statement in the Iliad (4.51–52) e ( λ ,« ξ ] φ > 0«, 5A « C 0 λ ) M 0 ‘indeed, three cities are by far the dearest to me: Argos and Sparta and Mycene with its wide streets’. A divinity who is thus connected with the cultural centres of the Mycenaean world must be a highly archaic one – as we should expect anyway, given her marriage to the king of the gods. Together with Zeus, Hera often appears in sculpture and mythical iconography, as at Olympia (cf. Paus. 5.17.1) or on various vase paintings where she features as a young bride accompanying her husband on a wedding chariot.2 A similar picture emerges when we ask what the historians of religion have to tell about her.3 Here one important source of evidence is constituted by Hera’s epithets, above all 6H ‘Hera who brings the «, who fulfils’, but also Q ‘she of the yoke’, ,« ‘girl’, % « ‘virgin’, 1 φ0 ‘bride’. As 6H , and jointly with Z]« «, the goddess receives the main sacrifice at wedding ceremonies in Ancient Greece. In Nauplia, Hera takes an annual bath in the source Kanathos, which restores her virginity. In Olympia, the College of the Sixteen Women organises a girls’ race in her honour, the winning prize being a cow. In Corinth, seven young women and seven young men are appointed to serve Hera Akraia for one year, thereby undergoing some rite de passage. In Knossos, according to Diodorus (5.72), a ¹ μ« « or ‘holy wedding’ takes place, and a similar ritual can be imagined for the festival of Hera at Samos where, according to Varro (ap. Lact. Inst. div. 1.17.8), simulacrum (sc. Iunonis) in habitu nubentis figuratum et sacra eius anniversaria nuptiarum ritu celebrantur ‘the image of Hera is represented in the dress of a bride and her annual ceremony is celebrated according to the wedding rites’. In this context, one is further reminded of Hsch. 0 824, s.v. π "α % α ¹ ξ 8 , ¹ ξ ¹ ‘Herochia: the “% ” (“Banquet of the Gods”), or a festival, or a ceremony’. The compound recorded in this gloss may be derived from the name 2 3
Cf. Ley (1998: 360–361), Kossatz-Deissmann (1988: esp. 682–690, 718). See (with references) e.g. Nilsson (1906: 40–64) and (1967: 427–433), Eitrem (1912), Rose (1958: 102–107), Kerényi (1972: esp. 76–142), Burkert (1985: 131–135), Pötscher (1987), Graf (1998: 358–359), Bremmer (2005: 6.3914–3916).
236
Andreas Willi
of the goddess followed by the root *wegh- ‘to convey (in a chariot)’ of Gk. 5" ‘to drive in a chariot’, Lat. veho¯, Skt. váhati, OCS. vezo˛ ‘to convey’ etc. As the bridegroom would typically drive home the bride in a chariot, *wegh- could take on the additional meaning ‘to marry’, which is clearly attested for Skt. vah-4 and which may help to account for the Greek derivative ( ? ) 5"1. ‘to cover, copulate’ (with a further semantic specialisation) as well as, perhaps, the epithet of Poseidon f(ζ)"«: quite possibly this is not the god ‘who makes the earth tremble’ but the god ‘who marries the earth’ (as suggested by his main name if this is based on an ancient vocative P E»« ‘husband of Earth’).5 However that may be, Hera also appears as φ 0 ‘wedded bride’ at the Daidala of Plataiai in Boeotia, where her image is washed in the Asopos river, then seated on a chariot and pulled up the slopes of Mount Kithairon: at the mountain top a cow is sacrificed to Hera and a bull to Zeus. According to the aetiology of this rite, the goddess had gone into hiding after a quarrel with her husband, but a local hero had advised the latter to dress up like a bride a wooden doll and put it on a wedding chariot; full of jealousy Hera had reappeared and pulled off the veil of the presumed bride – only to find herself facing a doll and thus being appeased. Finally, in Argos too the young women had an important role to play in the cult of Hera: decked with flowers they danced for the goddess, and Hera’s priestess may have acted as a bridesmaid ( φ1 ) as she was driven on a chariot from the town to the sanctuary, presiding over a procession of the armed young men. The preceding list could be extended, but the essential point is clear: from the historian of religion’s viewpoint Hera is even more obviously a goddess of women and marriage than from that of the archaeologist – 9 « φ ‘she keeps the keys of marriage’, as Aristophanes puts it in the women’s hymn of Thesmophoriazusae (976). Other literary sources, notably epic poetry, concur with this. As a patroness of brides and newly-wed women, Hera arranges
4 5
Cf. Moussy (1980: 342–344), with references. Thus Borgeaud (1944: 222), while Kretschmer (1914: 303) had already compared 5"1.; for the alternative hypothesis see e.g. Chantraine (1968–80: 1.219, s.v. ) and for the above derivation of P »« etc. Fick (1875: 307) and Kretschmer (1909: 27–28), followed most recently by Willi (2007: 182–183 n. 34) on *E».
Hera, Eros, Iuno Sororia
237
the marriage between Peleus and Thetis, the mother of Achilles (Il. 24.59–61), and when she sides with the Greeks against the Trojans, this is also because Paris has violated the legitimate marriage of Helen to Menelaus. Of course, the Homeric Hera is not always presented in a favourable light, but her unfaithful husband does not make her life particularly easy either; and yet, she does know how to use the art of marital seduction when it suits her. In Iliad XIV, in order to counter Zeus’ will and facilitate Poseidon’s intervention for the Greeks, Hera famously borrows Aphrodite’s belt of love and buys the help of Hypnos (Sleep) by promising him one of the Charites as a wife. After these preparations she goes to the top of Mount Ida, […] c ξ φ0 Z1«. ³« # c , —« * .« « φ « $ φ; , g Ρ 7 ( %0 φ0 ‘[…] and Zeus, the cloud-gatherer, saw her. As soon as he saw her, desire enwrapped his shrewd mind, just as when they had made love to each other for the first time.’ (Il. 14.293–295) The god does not lose time beating around the bush: “Hera,” he says, “come and sleep with me, ) < # h %»« * « ) ξ μ« % μ ( λ % "%λ« ( , ) # ²# 0 #I 0« $", […] ) #Ρ E 0 « $ 0«, ) # ² F0 « ( «, ) ξ )«, ³« * ]« \ « ¹ ,. ‘for never before has the love of a goddess or a woman so much overcome and enslaved the passions in my heart, not when I fell in love with the wife of Ixion, […] nor with Demeter, the lady with the beatiful locks, nor with famous Leto – nor with you yourself, as I love you now, and sweet desire seizes me.’” (Il. 14.315–328) Although this is a time of great tension between the divine spouses, the memory of the first night persists; and it is this partly carnal, partly
238
Andreas Willi
sentimental memory which becomes one of the defining features of Hera’s character even in Homeric epic where she is no longer the actual bride of Zeus. Hence, the literary perspective too encourages us to view Hera as a goddess of brides, weddings, and marriage in all its phases, a goddess who protects married women and the marital union both under its social and sexual aspect.
2. Etymologies of Hera I: Pre-Greek Mistress or Greek Goddess of the Year? In view of the unanimity with which Hera has thus been characterised by archaeologists, historians of religion, and literary scholars, it may come as something of a surprise how little this has helped historical linguists in their search for a satisfactory etymology of Hera’s name. At first sight this might look like a minor, and exclusively linguistic, issue. However, a potential failure to come up with an etymological explanation which ties in with the other disciplines’ data also has a more general impact. If the origin of Hera’s name did not match the functions she undeniably takes on in historical times, would we not then have to conclude that the original Hera was different in character and that Hera as we know her is only a ‘secondary’ Hera? Etymology, after all, has its main value in the elucidation of prehistory. In the major etymological dictionaries, uncertainty prevails. ‘Comme pour beaucoup de noms de divinités, pas d’étymologie établie’ notes P. Chantraine, while dismissing an old theory which had looked for a connection with Lat. serva¯re ‘to protect’; for there is no trace of an original *-w- (*-ζ-) after the - - of 6H ¯ in either Attic (where the theonym is not †6H 0, while e.g. * ζ0 yields 0) or Mycenaean (cf. section 3).6 Chantraine then goes on to assume ‘une origine préhellénique’. The obvious advantage of this position lies in the impossibility of ever proving it wrong, and one can even build upon it an entire house of cards: according to a hypothesis by U. von Wilamowitz, which has found prominent followers in M. Nilsson and C. J. Ruijgh, 6H ¯ should be translated as ‘mistress’ and be understood as the feminine counterpart of j .«, the latter allegedly being 6
Chantraine (1968–80: 1.415–416, s.v. 6H ¯ ); for the lack of *-ζ- cf. already Schröder (1956: 65–67) and Pötscher (1961: 304) and (1965).
Hera, Eros, Iuno Sororia
239
the Pre-Greek word for ‘master’.7 Unfortunately this theory has no linguistic foundation whatsoever – and given the structure of the two words involved it is even a priori unlikely: with its distinctive suffix j .« looks more like a derivative of 6H ¯ than vice versa. More generally, to dispose of our etymological problem case in such a cavalier way seems problematic when Hera is sufficiently integrated into the Indo-European part of Greek mythology to be thought of as the wife of Zeus (*dyeus; cf. Skt. Dyauhfl, Lat. Iu¯-piter). Even so, neither Chantraine nor H. Frisk, whose basic assessment of the matter is virtually identical to that of his French colleague, feel inclined to follow the strongest Indo-European hypothesis which was available at their time and which, since then, has found its way into as influential a textbook as W. Burkert’s Greek Religion.8 In 1956, F. R. Schröder had proposed to explain Hera’s name on the basis of a stem *yer- (> Gk. π -) which is reflected by Goth. jer, OHG. ja¯r ‘year’ (cf. Engl. year) and, with a corresponding long o-grade (*yo¯r-), by Av. ya¯rw n. ‘year’, Russian Church Slav. jara ‘spring’, Lat. adj. ho¯rnus ‘of this year’ (< *gho-yo¯r-ino-), Gk. — ‘season’. Admitting a further analysis of *yer-/*yo¯r- as a derivative in *-r(o)- from the verbal root *(h1)yeh2- ‘to go’ (cf. Skt. yáti ‘to go’, Lat. ia¯nus ‘passage’),9 the ‘year’ would here be conceived of as the eternal passage or return of time. For Schröder, therefore, the Greek Hera turns into a primeval earth goddess and ‘Goddess of the Year’ (‘Jahresgöttin’), a ritual trace of whom might be preserved in the annual renewal of Hera’s virginity at Nauplia (cf. section 1) and in the tripartite cult of Hera ,« ‘girl’, ‘wife’, and " ‘widow’ at Stymphalos in Arcadia: these three hypostases would correspond to the three seasons spring, summer, and autumn. However, in the absence of any further relevant evi7
8
9
Nilsson (1941: 326, 401) and Ruijgh (1967a: 89 n. 75), (1967b: 51), and (1995: 75), after Wilamowitz (1889: 1.296); cf. also West (1978: 370–373). Moreschini Quattordio (1979: 195–198) thinks of a ‘resa fonetica greca del teonimo sumerico Eresˇ.ki.gal’. On j .« next to 6H ¯ see also fn. 22. Schröder (1956: esp. 64–69), cited by Burkert (1985: 131), though without full endorsement; cf. also Frisk (1960–72: 642, s.v. 6H ) and, critically against Hera as an ‘Earth Goddess’, Pötscher (1961: 306, 312–313, 320–322) and (1987: 9–19, 137–141). Cf. also R. Nedoma apud Peters (2002: 364 n. 37) (*Hyeh2-r-), but Peters prefers the more straightforward postulate of a heteroclitic neuter *Hyoh1-r, *Hyeh1-nfor the ‘year’ word.
240
Andreas Willi
dence in the substantial body of information we possess about the goddess it is much easier to interpret the Stymphalian cult, like all the other cults of Hera, as the cult of a goddess of women (in the various phases of life).10 So, we cannot accept Schröder’s theory unless we are prepared to believe (a) that Hera’s function as a divinity has completely changed – the former state of affairs not leaving any trace except in her name – and (b) that earlier on the traditional earth goddess Demeter, who does oversee the annual renewal of the crops, had something like a systemic rival. Despite these difficulties Schröder’s etymology seemed too attractive formally to be dismissed without further ado. As a consequence two other scholars, A. J. Van Windekens and W. Pötscher, tried to make it suit better the ‘historical’ Hera.11 According to the former the theonym did not initially designate a ‘Goddess of the Year’, but a ‘yearling animal’ or more precisely a ‘one-year-old calf’ (cf. Lat. vitulus, Skt. vatsá- ‘calf’ < ‘yearling (cow)’ next to *wetos, Gk. (ζ)« ‘year’). In support of this idea Van Windekens points to Hera’s Homeric epithet 27« ‘cow-eyed’, but again other evidence for a prehistoric ‘cow goddess’ Hera is missing12 and the Homeric epithet itself is no strong indicator as is it also, though less commonly, applied to other women.13 Hence, W. Pötscher favours a more narrowly philological approach: he underlines that the Greek adjective ³ ,«, a derivative of — (*yo¯r-), sometimes expresses the right moment for marriage since a young woman who has reached the relevant age may be called ³ . In this way, — would refer to the ‘good season’, the ‘maturity’ of the young woman, and Hera would be the goddess in charge of this maturity. Here too, though, not everything is as straightforward as it should be. Firstly, if the semantic specialisation described by Pötscher can be illustrated for the lexeme — , the same is not true for the base word *yer- from which Hera’s name is supposedly derived. Secondly, the expression ³ ‘ready for 10 11
12
13
Cf. e.g. Rose (1958: 105), Pötscher (1961: 311), Kerényi (1972: 104–105). Van Windekens (1958) and (1960: 213–214), Pötscher (1961: 303–309); cf. Pötscher (1987: 2–3). As rightly observed by Schröder (1956: 61): ‘Durch Kultordnungen und Mythen ist gesichert […], dass die Kuh der Hera besonders heilig war. Das beweist an sich gewiss noch nicht, dass sie selbst einmal kuhgestaltig war’; but Schröder then calls doubts about this ‘Hyperkritik’ (cf. already Pestalozza 1939). Cf. Pötscher (1961: 309, 317).
Hera, Eros, Iuno Sororia
241
marriage’ proves little when the adjective ³ ,« (fem. ³ ) does not evoke the ‘time of marriage’ on its own, without ; according to Greek usage someone can also be (e.g.) ³ ,« % ‘ready for death’, and if there is a — 0 in Homer, there is also a — 0 / / 1%. , a ‘time for sleep, a meal, story-telling’ etc.
3. Mycenaean Hera Meanwhile, the most serious obstacle to the Schröder-Pötscher theory (and indeed to any other hypothesis which operates with an etymon such as *yera¯14) lies elsewhere.15 In perfect accordance with the Mycenaean pointers in the text of Homer (cf. section 1) we find Hera mentioned already in the Linear B documents, notably on the famous tablet PY Tn 316 which records a list of human and material offerings to various divinities. Lines 8–10 (verso) read as follows: v.8 v.9 v.10
i-je-to-qe, di-u-jo, do-ra-qe, pe-re, po-re-na-qe a-ke di-we aur *213vas 1 vir 1 e-ra aur *213vas 1 mul 1 ] vacat di-ri-mi-jo di-wo, i-je-we, aur *213vas 1 [
Although not all the details are clear, line 8 uncontroversially refers to the place of the ceremonial offering (dat.-loc. di-u-jo ‘at the *Eζ /sanctuary of Zeus’) followed by the usual introductory formula containing the words 7 φ ‘and brings gifts’ (the subject perhaps being pu-ro ‘[the community of] Pylos’ as indicated elsewhere on the tablet). Lines 9 and 10 then name three divinities, each of whom receives one golden vessel and one human servant (a male for the two gods, a female for the goddess). The first is Zeus (dat. di-we = *Eζ), the second Hera (dat. e-ra = 6H ¯ , NB. not †e-wa = †6H ζ¯ ), the third an otherwise unknown Drimios son of Zeus (dat. di-ri-mi-jo di-wo i-je-we = *E . Eζμ« ¹ζ). From a mythological perspective this sequence is interesting in two respects. Firstly it
14
15
See also Kazansky (1995: 158–163), with an equivalence Gk. LH -« ~ Slavic *Jaroslavu˘, and Adams (1987: 175–177) who postulates an action noun *yeh1-r ‘deed’ (cf. Hitt. iya- ‘to do, make’, Toch. A ya-) on which a derivative *yeh1-r-eh2 ‘Strength, Vitality’ would be based; both formally and semantically this is inferior to the Schröder-Pötscher hypothesis. Cf. already Peters (2002: 365), Janda (2005: 214).
242
Andreas Willi
suggests that Zeus and Hera were already a couple in Mycenaean times and secondly it invites speculations about the (later) identity of the obscure Drimios. Some scholars have tried to identify him with Apollo or Dionysus,16 but a different hypothesis seems at least as promising: since the name *E « most probably derives from the adjective 1« ‘sharp, piercing’ and since 1« is applied to the noun "0 ‘fight, combat’ in Homer (Il. 15.696) and Hesiod (Scut. 261), Pötscher plausibly considers an equation of Drimios with the war-god Ares who is, together with Eileithyia and Hebe, one of the few children of the supreme couple according to classical myth (cf. esp. Hes. Theog. 921–923).17 For our present purposes, however, an orthographic point is more relevant. As is well-known, the Linear B syllabary contained a series of signs representing a sequence of *y- + V: ja, je, jo and probably also ju if this is the value of sign no. *65. From graphic doublets such as jodo-so-si ‘(those) who will give’ ( ? ) (~ later Greek θ <) next to o-do-ke ‘what he gave’ (~ χ (*) .), or ja-ke-te-re next to a2-ke-te-re for an agent noun of unclear meaning, it may be inferred that ‘le passage de *y- à *h- est sensiblement contemporain de l’époque mycénienne, où coexistent une orthographe traditionelle jo- et une orthographe o- (à entendre: ho-), tenant compte de l’affaiblissement de yod’.18 The Mycenaean scribes appear to have exploited this phonetic development in the case of o- next to jo- so as to differentiate between nom. pl. masc. \ < *yoi (henceforth written jo-) and nom.-acc. sg. ntr. Ρ < *yod (written o-).19 In the case of a2-ke-te-re next to ja-ke-te-re, the new spelling still indicated the initial aspiration arising from *y-, thanks to the existence of the special sign a2- = /ha-/. Thus the traditional spelling seems to have been abandoned when the new spelling could help to disambiguate, or at the very least when it did not create more ambiguity than had existed previously. Given this background two points are worth stressing as we look at Hera’s name. Firstly, we never read *je-ra, while e-ra (or e-[, e-ra[) is
16
17 18 19
Dionysus: Gallavotti (1956: 228–229); Apollo: Stella (1958: 26) and (1965: 248–249), Palmer (1963: 264); cf. Gérard-Rousseau (1968: 65–66), Aura Jorro (1985–93: 1.178, s.v. di-ri-mi-jo). Cf. Pötscher (1987: 19–27). Lejeune (1972: 168); cf. Heubeck (1979: 241–243), Willi (2008). See Probert (2008).
Hera, Eros, Iuno Sororia
243
found several times.20 Secondly, to replace *je-ra by e-ra would have rendered the reading more, not less, ambiguous: after the change of *y- into /h-/ a spelling *je-ra would have been interpreted automatically as /hera¯/ (since the sign je- would now have designated /he-/ regularly), whereas the spelling e-ra could in theory be decoded with or without initial aspiration (/hera¯/ or /era¯/). In other words such an ambiguous spelling for the theonym is fully justifiable only if there had never been an initial *y- to start with, for in that case a ‘traditional’ spelling *je-ra would never have existed either. Even if this consideration may not amount to an absolute proof, it is therefore fair to say that the Mycenaean material warns against the Schröder-Pötscher hypothesis. Instead it encourages the search for an etymology without *y-. Such an alternative would still have to account for the initial aspiration of Hera’s name, so that only one possibility is left: to postulate *sera¯, whose initial *s- must have turned into /h-/ in PreMycenaean times.
4. Etymologies of Hera II: Raped Girl or Goddess of the Galaxy? In recent years two etymologies which opt for *sera¯ instead of *yera¯ have been proposed. The first, by M. Peters, starts from the verbal root *ser- ‘to catch, take’ of Gk. ¹ . h *sr-ye/o- (with modification of the stem, yielding a vocalic verb21) and posits a related root noun nom. *ser, gen. *sr-os with the meaning ‘(violent) taking’ > ‘rape’ > ‘booty’ (cf. Hitt. sˇa¯ru < *soru ‘booty’). This root noun would have served as the basis for an exocentric derivative *ser-o- ‘belonging/relating to the rape, of the rape’ whose feminine *ser-a¯ would have meant ‘she of the rape’: ‘ein Mädchen […], das “gewaltsam genommen/geraubt worden” oder “(gut) zu nehmen/rauben” ist – und ergo […] die ideale Braut für einen idg. (jungen) Krieger (j .«) verkörpert hat’.22 Formally this 20
21 22
Cf. Gérard-Rousseau (1968: 94–96), Moreschini Quattordio (1979: 176–180) and esp. Aura Jorro (1985–93: 1.201–202, 1.232–233, s.vv. e-[, e-ra, e-ra[), with bibliography: PY Tn 316.9, TH Of 28.2, PY Un 219.8 (e-[ra) ( ? ), PY Un 853.1 (e-ra[) ( ? ). Note also the frequent toponym e-ra (with corresponding ethnonym e-ra-jo) at Knossos, which Ruijgh (1967: 228) wants to identify with the theonym 6H . Cf. Vine (1998: 48–49), Janda (2005: 209). Peters (2002: 372); for the root-noun type of his *ser, *sr-os cf. Schindler (1972:
244
Andreas Willi
theory is unobjectionable (especially if the postulated root noun were, despite the divergent semantics, reflected in Homeric ((λ) @ φ
‘to gratify’ < ‘to pay tribute’ vel sim.), but it seems most uncertain whether in the eyes of a (Proto-)Greek a raped (booty) woman could have become one of the legitimate wives who are protected by Hera. Moreover, the derivation presupposes that Hera herself must have been imagined as a ‘raped girl’ at some point – for which there is no evidence at all in the iconographic, cultic, and literary sources reviewed above –, and it overlooks the fact that Hera is a goddess not only of brides, but of (married) women throughout their adult life. Even so, Peters’s theory may carry more conviction than an alternative sketched subsequently by M. Janda. The latter follows Peters for the crucial parts of the formal argument but operates with the more widely attested root *ser- ‘to attach, connect’ of Gk. c . (< *serye/o-), Lat. sero¯ (< *ser-e/o-), and OIr. sernaid ‘to put in order, arrange’ (homonym of sernaid ‘to stretch out’ < *sterh3-23). Such a connection is plausible enough – and we will in fact return to it later on (cf. section 7) –, but Janda’s further contextualisation is highly speculative. For him the root noun *ser is supposed to have meant not just ‘that which is serially attached’ (‘das in einer Reihe aneinander Befestigte’), but also ‘the tense and stretched out arms’ (‘die angespannten
23
36–38). The (traditional: cf. section 2) link between 6H ¯ and j .« need not detain us further, but if it is correct (despite Moreschini Quattordio 1979), Pötscher (1961: 329–335) may be right when he sees in the primeval ‘hero’ the ‘young man’ # (" (cf. Peters 2002: 357–362), a counterpart of the nubile young woman protected by Hera; pace Pötscher, however, this need not imply a mythical couple Heros ~ Hera predating the couple Zeus ~ Hera (nor is there any evidence for the original j .« having been their son, as Schröder 1956: 69–72 suggests). Because of Myc. ti-ri-se-ro-e (dat. ‘to the triple hero’?, PY Fr 1204, PY Tn 316.5) one hesitates today to explain the formation of j .« with the help of a suffix *-o¯u- (cf. .« ‘paternal uncle’, .« ‘maternal uncle’; Chantraine 1933: 124): see Ruijgh (1967: 89) and Risch (1974: 160–161). Still, if a new paradigm was based on the acc. sg. *-ou-m > *-o¯m (cf. Peters 2002: 362–363) or if the /-w-/ in a form such as */hero¯wei/ was interpreted as a glide and omitted in writing (cf. i-je-re-u, but also i-e-re-u = ¹ 1« ‘priest’), the theory could be revived: compare, perhaps, Myc. to-ro-o (PY An 519.1) if this is a genitive /Tro¯os/ next to a nominative T <« (cf. Aura Jorro 1985–93: 2.365, s.v. ]to-ro) and if the stem of T <« originally contained a *-w-, as suggested by the probable equation of Gk. T with Hitt. Taruisa, Luv. Tarwiza. Cf. Thurneysen (1946: 133).
Hera, Eros, Iuno Sororia
245
und ausgestreckten Arme’); ‘[w]ir können *Serah2 somit versuchsweise auf jene spezifische, phraseologisch und mythologisch bedeutsame Verwendungsweise beziehen, die […] sich im Namen des Hermes äussert: auf die fest angespannten, ausgestreckten Arme, mit denen der göttliche Seelengeleiter den Himmel trägt, in denen die gewölbt ausgebreiteten Arme der Milchstrasse ihren mythologischen Ausdruck gefunden haben’. Against the historical evidence Hera thus becomes an old goddess of the galaxy (‘Göttin der Milchstrasse’), when even the initial semantic change ‘to attach, connect’ > ‘to stretch out one’s arms’ is hardly substantiated.24
5. An Indo-European word for ‘woman’ The preceding critical remarks illustrate why the safe Pre-Greek hypothesis à la Frisk and Chantraine may still represent the most attractive of all the unattractive options hitherto available. On the following pages a new attempt will now be made to come up with an Indo-European solution to the riddle of Hera’s name. Undoubtedly this too will fail to convince every sceptic. At the very least, however, it may show how much room for variation, and hopefully improvement, there is within an Indo-Europeanist framework. Our starting point is the Proto-Indo-European numerals ‘3’ and ‘4’. In addition to masculine *treyes, *kwetwores (> Gk. ,«, «, Lat. tres, quattuor, Skt. tráyas catváras), Sanskrit and Old Irish attest separate feminine forms going back to something like *tis(o)res ‘3’, *kwetes(o)res ‘4’ (> Skt. tisrás, cátasras, OIr. teoir, cethéoir). These may be further analysed as containing the numeral stems *tri- ‘3’ (with dissimilation *tris(o)res f *tis(o)res) and *kwet- ‘4’, as in the masculines, followed by a suffixal element *-s(o)r-.25 Since such a suffix 24
25
Janda (2005: 213–224), whose ‘kumulative Evidenz’ is meagre; on p. 318 Heracles also becomes ‘he who has the glory of the galaxy’ (‘den Ruhm der Milchstrasse habend’). Regarding the ablaut grade Oettinger (1986: 126) observes that ‘der einzige Beleg des Nom. Pl. tisrás und einer der Belege von Nom. Pl. cátasras sind metrisch überzählig (RV. V 35,2) und vielleicht *tisa¯ras, *catasa¯ras zu lesen’; he therefore considers an original distribution with *-sor- for the nominative, *-srfor the accusative, but the need to reconstruct *-sor- instead of *-sr- for the Old Irish forms is questioned by Cowgill (1957).
246
Andreas Willi
*-s(o)r- was not known elsewhere to form feminine derivatives, K. Brugmann and, more explicitly, R. Meringer suggested already in 1876 and 1904 to recognise here an old Proto-Indo-European lexeme for ‘woman, female’: so *t(r)i-s(o)r-es, *kwete-s(o)r-es would originally have meant ‘3 women/females’, ‘4 women/females’. Obviously such a theory would have been gratuitous if it had been completely isolated, but Brugmann – in the footsteps of A. F. Pott – further connected the Indo-European word for ‘sister’, *sweso¯r (> Lat. soror, Skt. svasar-, etc.; cf. also the Greek glosses Hsch. 4048, s.v. * α %0 , $ ;« ‘daughter, cousin’, Hsch. 4049, s.v. * «α «, ,« ‘relatives, kinsfolk’, and Hsch. 7105, s.v. ) φ [read: ( φ?]α ‘women’, which may point to a semantic widening of a more restricted initial meaning). A. Meillet, who followed Brugmann and Meringer, thus translated *swe-sor- as ‘le membre femelle du groupe’, the element *swe- being identical to the stem of the reflexive pronoun *swe with the corresponding possessive adjective *s(e)wos (> Gk. Ρ«, Lat. suus).26 Moreover, by establishing a link with the root *h1euk- of Arm. owsanim ‘to learn’ Meillet traced back Lat. uxor ‘wife’ to yet another compound containing *-sor- ‘woman, female’: an *(h1)uk-sorwould have been a ‘personne féminine habituelle, compagne’. Alternatively, one might still consider a connection first outlined again by Pott which discovers in uxor the root *wegh- ‘to convey’ (> ‘to marry’; cf. Skt. vah- and section 1), so that the Latin ‘wife’ would have started off as the ‘woman who is conveyed/married’, or the ‘conveyance/marriage woman’ (*ugh(s)-sor-).27 The entire file was further extended by É. Benveniste’s derivation of Av. ha¯irisˇi ‘female’ from a suffixed *sor26
27
Meillet (1931: 8–9), building upon Pott (1833: 126–127) and (1836: 554), Brugmann (1876: 393–394) and Meringer (1904: 171–172); cf. subsequently e.g. Benveniste (1969: 214–215), Linke (1985: 362). General doubts about PIE. *sor- are expressed by Mayrhofer (1952) and Normier (1980). Moussy (1980: esp. 339–346), after (e.g.) Pott (1833: 9) and Ambrosini (1963) (revising the version in Ambrosini 1962: 25–31, where a link with *wek- ‘to wish’ is proposed, as already by Ascoli 1864); de Vaan (2008: 649, s.v. uxor) calls this etymology ‘impossible morphologically’ (cf. Normier 1980: 60), but if the first part consists of a zero-grade s-stem noun (as in e.g. Skt. medhá- ‘wisdom’ < *mns-dh- next to mánas < *menos ‘mind’; cf. Normier 1980: 65–67 with a different explanation of uxor, Meier-Brügger 1989) the objection may be unjustified. The much-cited connection of uxor with the word for ‘ox’ (Skt. uksflán- etc.; cf. Pisani 1951: 4 and Mayrhofer 1952: 35, variously modified by Mastrelli 1980 and Van Windekens 1984) seems implausible in any case (Normier 1980: 60–64).
Hera, Eros, Iuno Sororia
247
Hs-ih2 or *ser-Hs-ih2 (cf. section 9),28 and the addition of Ved. strâ- f. – the categorical term for ‘woman, female’ as the opposite of púma¯ms˙ ‘man, male’29 – remains a possibility too, at least as long as no better explanation of the word is available; it is true that the zero-grade form of *sor- enlarged by the usual feminine-deriving suffix *-ih2- should presumably yield (*sr-ih2- >) *srâ- rather than strâ-, but an analogical remodelling of the former after the feminine agent noun formations in -tri < *-trih2 (cf. masc. -tar-) seems feasible.30 In Greek the same PIE. *sor- ‘woman, female’ may help to elucidate one, perhaps even two, of the Homeric words for ‘wife’. Homeric 4 may show the same copulative prefix 5- < *sm- (with Aeolic vocalisation and psilosis) as the adjective 4 « ‘of the same father’ and thus continue a compound *sm-sr- ~ ‘female companion’ (with generalised weak stem from the oblique cases; originally e.g. nom. *sm-so¯r);31 and Homeric , which has been variously but not definitively explained, could similarly derive from a tatpurusfla com28
29 30
31
Benveniste (1935), who, somewhat unnecessarily, posits *so¯r-Hs-ih2 (in a modernised notation); cf. Oettinger (1986: 126–127). Normier’s (1980: 46) alternative (compound *ha¯r- + isˇ- ‘Sättigung-Labung’) is implausible. See Kazzazi (2001: 212–220); cf. Eichner (1974: 26). Cf. Sturtevant (1949: 344–345) (with the intriguing additional suggestion that a ‘Proto-IE word *sri “woman”’ surfaces in Gk. Ll¯ h *V, the name of the mother of Zeus and Hera) and Benveniste (1969: 215); contrast Mayrhofer (1952: 35–37) (cf. *ster- ‘furrow’), Eichner (1974: 26–39) (cf. ster-ih2 ‘infertile’, but see Hamp 1981: 38), and Normier (1980: 44–47) (*sh2-trih2 from *seh2- ‘to nourish’). 4 (cf. B. Mader in LfgrE. 3.480–481, s.v. 4 , Gates 1971: 17–18) does not designate any ‘junge Frau’, as Janda (2005: 128) states in order to postulate an archaic base word *swosr next to which an old collective noun *sweso¯r ‘group of young women within a household, sisters’ > ‘sister’ would have existed (cf. already Janda 1999); note also that Janda’s ablaut pattern would probably imply a root *swes- or the like, thus rendering impossible any analysis of *swesor- as a compound. According to Gates (1971: 69 n. 55) and B. Mader in LfgrE. 3.480, s.v. 4 , metrical considerations might favour an initial *w-, but see Chantraine (1958: 90) on unshortened diphthongs (and long vowels) in arsi before vowel (as at the verse-end of Il. 5.486 $ n [5 ]) and Chantraine (1958: 104) on short syllables counting as long in arsi (as in Il. 9.327 $
«, 5 . o φ . ; cf. e.g. Il. 15.715 beginning with $
7
. ). Ruijgh’s (1967: 386–387) argument in support of *w- is bold, as he himself admits, and the same is true for Normier’s (1980: 70–72) etymological proposal (*wns-(h1)rh2 ‘love loving’; cf. Skt. vánas- ‘loveliness, desire’, Lat. Venus and Gk. * ‘to love’, on which see further below, section 8).
248
Andreas Willi
pound of *dom- ‘house’ + *s(o)r-, i.e. *dm-sr-, the shape of the first element being identical to that of Homeric ‘floor’ < *dm-pedom.32 Finally, the postulate of a PIE. lexeme *sor- ‘woman, female’ was substantially boosted in 1936 when J. Lohmann first wondered about its occurrence in a thematised motion suffix *-s(o)r-o- of the Anat¥ ¯ - ‘master’ (cf. Lat. erus < olian languages,33 as exemplified by Hitt. isˇha *h1esh2ó-) vs. isˇha-s ¥ ˇ sˇara- ‘mistress’, has ¥ ˇ sˇu- ‘king’ vs. (*)has ¥ ˇ sˇu-sˇsˇara‘queen’, as well as the female priest’s title sˇuppi-sˇsˇara- (cf. sˇuppi‘pure’), or Luv. nana/i- ‘brother’ vs. adjectival nana-sˇr-iya/i- ‘of the sister (lit. “female brother”)’. However, additional Anatolian material, from Luvian, has also cast doubt on the classical reconstruction. Luvian further shows an adjective asˇrul(i)- ‘female’ and a noun asˇrahit¥ ‘femininity’ whose initial a- may point back to an anlauting *(h1)e-. Hence, O. Szemerényi has argued that one should posit *esor(i.e. *h1esor- in laryngealist terms) rather than *sor- for PIE., and consequently analyse the ‘sister’ word as *su-(h1)esor- > *swesor- or the numeral ‘4’ as *kwet-(h1)esor-es rather than *kwete-sor-es.34 On balance, though, this revision does not seem advisable as it creates more problems than it solves. To start with Anatolian itself, the question arises whether we can simply operate with an ablaut variant *h1sor- in order to explain *-sˇsˇar(a)- (NB. not *-asˇsˇar(a)-, given sˇuppisˇsˇara- etc.); and, even supposing that Luv. asˇrul(i)- and asˇrahit¥ do be32
33
34
On see Risch (1974: 214). The stem in -- (gen. --« etc.) would then have to be secondary, as suggested already by Ruijgh (1967: 385 n. 166) in his explanation (after Schulze 1887, rejected by Szemerényi 1977a: 395–396 and 1977b: 78–79) of as ‘celle qui a soin de la maison’; cf. gen. (ζ) --« ‘lord’ next to the (certainly ancient) divine epithet (ζ) -« of the Dioscuri. Lohmann (1936: 291–292), followed e.g. by Sturtevant (1949: 343–344) but cautiously criticised by Kronasser (1966: 109–112); cf. the history of the question in Szemerényi (1967: 206–209). Melchert (1994: 151) posits *-sr-o-, Rieken (1999: 262) *-h1sor-o- (with assimilation *-h1s- > -sˇsˇ-; but see below on the problem of the initial laryngeal). Sihler (1977: 43–45) detects the same motion suffix *-s(o)r- within the Germanic suffix conglomerate *-stri(o¯)n for feminine nomina agentis. Szemerényi (1977a: 389–391) and (1977b: 32–47), improving on Szemerényi (1967: 215–221); similarly Carruba (1991: 159) posits *Hser-. Pace Oettinger (1986: 124), a proto-form *eso¯r without initial laryngeal (in an otherwise laryngealist framework) is hardly acceptable.
Hera, Eros, Iuno Sororia
249
long here, how much do their initial vowels really tell us when the ‘[l]ack of /sR-/ is probably significant’ in this language,35 i.e. when Luvian phonology appears to have forbidden forms such as *sˇrul(i)- or *sˇrahit-? ¥ Turning to Greek, to derive 4 from *(h1)osr0 involving an o-grade, as Szemerényi would like, plays havoc with usual animate ablaut patterns, when Homeric 4 is clearly not a neuter (cf. Il. 9.327). As for the alleged *su-(h1)esor- ‘sister’, this is excluded by Vedic metre where such a proto-form should be reflected by a regular trisyllabic scansion *su(v)asar- of attested svasar- – a scansion which is not found once in the Rig Veda.36 Moreover, if Szemerényi’s argument may work for the numeral ‘4’, it fails with ‘3’ since PIE. *t(r)isor-es can hardly derive from either *tri-h1esor-es or *tri-h1sor-es.37 All things considered, the traditional reconstruction *sor- of Brugmann, Meillet, Benveniste etc. therefore remains preferable for this archaic PIE. lexeme for ‘woman, female’. And as we shall see (section 7), a welcome confirmation of this comes from the fact that *sor- need not remain an isolated item in the Indo-European vocabulary.
35 36
37
Melchert (1994: 267). Contrast e.g. svastí- ‘well-being’ (< *(h1)su-h1estí- or *(h1)su-nstí-: cf. Mayrhofer 1986–2001: 2.796–797, s.v. svastí-), which almost always appears as su(v)astí- (cf. Grassmann 1873: 1634–1635, s.vv. svás˙r, svastí). The same argument invalidates the divergent view of Pisani (1951: 7–8) and Normier (1980: 48–59) who analyse the word as *su-h1esh2o¯r ‘having the blood of the own family’ vel sim. (followed in different ways by Linke 1985: 354–356, with the unlikely idea that the words for ‘woman’ and ‘blood’ share the same origin, and Pârvulescu 1989: 67–72, who unacceptably regards the initial *h1e- of *h1esh2(o)r- as one of ‘various affixes’ added to ‘a base *ser, *sor or *sr’). Note further Hamp (1988) according to whom the ‘sister’ word must have been analysable as *swe-sor- in Proto-Albanian to trigger the formaiton of a new word for ‘brother’, vëlla¯ < *swe-loudh-a¯ ‘own people’. The second (unaccented) *-e- of *kwet-e-sor-es (next to masc. *kwet-wor-es), in turn, could have arisen from a (Pre-)PIE. epenthetic *-w- (vel sim.; cf. the origins of the thematic vowel?) to prevent a cluster *-ts- whose simplification might have obscured the morphological basis *kwet-.
250
Andreas Willi
6. Hera, Goddess of Women Given its formal appearance there can be no doubt that *sor‘woman, female’ is a root noun. Thanks to the systematisation by J. Schindler we know that one type of Proto-Indo-European root nouns had a full o-grade in the strong stem (nom./acc. sg.) alternating with a full e-grade in the weak stem (e.g. gen./dat. sg.). Thus, the nominative **dom-s > *do¯m of the word for ‘house’ (PIE. *dom-) survives in Arm. town (and, less directly, in Homeric 7 ‘house’), while the genitive *dem-s is reflected by the first member of the compound Gk.
0« < *dems-pot- ‘(house-)master’ (cf. Av. dw¯ ng fl paiti-).38 Hence, for PIE. *sor- we shall expect nom. *so¯r (< **sor-s), as in nom. *sweso¯r ‘sister’, but gen. *ser-s39 with regular weak stem *ser-. Looking at the semantics of these root nouns of the type *dom-/*dem-, Schindler finds among them (a) ‘substantifs féminins à valeur résultative ou passive’ (cf. *dom- ‘that which is built’) and (b) ‘noms d’agent (substantifs et adjectifs), souvent avec une nuance itérative, comme *bhor“celui qui emporte”’ (cf. Gk. φ< , Lat. fu¯r ‘thief’).40 Just like any other noun a root noun could serve as the basis for a vflrddhi derivative with the meaning ‘relating/belonging to X’. The Proto-Indo-European ancestry of this derivational process, which can be described as ‘-o-Ableitung mit zusätzlicher -e-Infigierung’41 and which becomes particularly productive in Indo-Iranian and Germanic, is firmly established by items such as *deiw-ó-s ‘divine’ (> Skt. devá- ‘god, divine’, Lith. di˜evas, Lat. deus, divus etc.) from the weakstem basis *diw- ‘Sky(-God)’ (cf. Gk. gen. E(ζ)-«, Skt. Div-áhfl vs. nom. Z1«, Dyauhfl < *dyeu-s) or *swekur-ó-s ‘(belonging to the father-in-law >) brother-in-law’ (> OHG. swa¯ger, Skt. s´va¯s´urá-) from the basis *swékuros ‘husband’s father, father-in-law’ (> Skt. s´vás´ura-, Gk. 8 «, Lat. socer, OHG. swehur).
38
39
40 41
Cf. Meier-Brügger (2002: 217–218), after Schindler (1972: 32) and R. Lipp in LIV. 115, s.v. 1. *demh2-. Obviously, *ser-s might have been remodelled into *sr-es or *sr-os at some point (cf. Schindler 1972: 32), but this is of little importance here. Schindler (1972: 36). Meier-Brügger (2002: 288); for the type see esp. Darms (1978), after the classic treatment by Schulze (1907).
Hera, Eros, Iuno Sororia
251
Starting from the weak stem (as in *diw- f *deiwó-), for our lexeme *sor-/*ser- ‘woman, female’ we thus obtain an adjectival vflrddhi derivative *ser-ó-s ‘relating/belonging to women, of women’. The feminine form of this adjective would be *ser-áh2, i.e. precisely the form needed to account for the proto-form *Sera¯ of Hera’s name.42 Even the accentuation need not worry us unduly. In the paradigm of theonyms the vocative has a pivotal role because of its use in invocations, so that for example the Latin nominative Iu¯piter is in reality an original vocative (*Dyeu pater) replacing the ‘true’ nominative Dies piter. At the same time, the Indo-European vocative typically retracted the accent, as in Skt. dúhitar ‘daughter’ vs. nom. duhitá, Gk. ‘master’ vs. nom. 0«, Ν φ ‘brother’ vs. nom. $ φ«. Hence the vocative *Sàra will have caused the retraction of the accent also in the nominative *Será (just as it did, incidentally, in a word like θ0 , which is accentuated after the vocative θ1 43). According to this new etymology, then, Hera is quite literally the goddess ‘of women’. This is perfectly in line both with her function as a deity protecting adolescent girls when they are about to become ‘adult women’ by getting married and with her continued interest in the well-being of women within the institution of marriage. Moreover, it may help to elucidate one of the oldest Homeric formulae: the verse-end (27«) 6H 0 whose antiquity is suggested by the lack of elision in .44 In such an archaic phrase the name of the goddess may well retain some of its original force as a descriptive adjective. Homeric 6H 0, or Proto-Greek *potnih2 serah2, would thus translate word by word as ‘mistress of women’.
42
43 44
A link between this PIE. word for ‘woman, female’ and Hera’s name is first adumbrated in Carruba (1991: 159), but how 6H ¯ could derive from a protoform *Hsar-a¯ remains obscure: surely the initial 6H- cannot be contracted from *(h)–? Cf. Schwyzer (1939: 381). See Ruijgh (1995: 75–76); Schmitt (1967: 63 n. 396) regards 6H 0 as modelled upon 0 , but in view of its peculiar semantics the priority of the latter phrase is unlikely.
252
Andreas Willi
7. Women, love, and attachment To speak of ‘women’ tout court is, however, a little imprecise. As we have seen (section 1), Hera is primarily the goddess of women as nubile beings and/or in wedlock. With this specification in mind, let us now take another look at PIE. *sor-/*ser- ‘woman, female’. Just as the PIE. root noun *dom-/*dem- ‘house’ must be related to the verbal root *demh2- ‘to build’ (cf. Gk. « ‘form [< construction]’ ¯ « ‘newly-built’ < *-dmh2-to-), *sor-/*ser< *demh2-s, Dor. - may be related to some verbal root. As mentioned before (section 4), there is more than one PIE. verbal root *ser- to choose from. The Lexikon der indogermanischen Verben lists (1) *ser- ‘to protect, guard, oversee’ (cf. Myc. o-pi o-ro-me-no ‘he who oversees’), (2) *ser‘to catch’ (cf. Gk. ¹ .: section 4), and (3) *ser- ‘to attach, connect’ (cf. Gk. c ., Lat. sero¯: section 4).45 In his entry for the last of these three, J. Pokorny follows an old suggestion by Meringer when he notes that one might consider a further link with the Germanic verbs ON. serd¯a ‘to fornicate’, OHG. sertan ‘to make love’ as well as, perhaps, with the Old Irish noun serc ‘love’.46 It is easy to see how a verb based on a root which designates a connection, attachment, or union could eventually have come to refer to a sexual act. Similarly, the PIE. *sorcould well be originally either (a) ‘the female who is attached/ coupled’ or (b) ‘the female who attaches herself’ to a male, i.e. the man’s ‘partner’ both socially and physically or emotionally; and since she becomes a potential partner only once she has reached sexual maturity, the adult age, she is specifically the nubile/adult woman. To be sure, the latter restriction of the initial semantics would no longer have been felt in the case of e.g. *t(r)i-sor-es or *swe-sor-, where *-sorjust indicates the female being in general; but we may equally remember in this context a series of derivatives of Homeric 4 < *sm-sr in which the notion of emotional attachment, ‘intimacy’, is paramount. Both 4 « and 5 Q. refer to the intimate conversation between a man and a woman and 5 1« designates an amor-
45 46
LIV. 534–535, s.vv. 1./2./3. *ser-. Pokorny (1959: 1.911), after Meringer (1904: 171): ‘Es empfiehlt sich daher, ein stammabstufendes *sór-es, *sr-bhís usw. und zwar mit der Bedeutung “Weiber” anzunehmen, das zur selben Wurzel *ser- gehört wie lat. series, sero, c .’.
Hera, Eros, Iuno Sororia
253
ous encounter.47 Not just by coincidence is Aphrodite’s belt of love (cf. section 1) therefore described with the words * %# * ξ φ0«, ( # \ «, ( # 5 1« ‘in which there is love, and longing, and intimacy’ (Il. 14.216). If we may thus conclude that the root *ser- ‘to attach, connect’ could refer not only to the act of physically attaching something, but also to an act of metaphorical attachment, be it social or sentimental, it follows that this act itself may have been expressed in Proto-Indo-European times with the help of a derived action noun. One of the oldest procedures to form such nomina actionis or nomina rei actae is the creation of neuter s-stem derivatives such as *genh1-os ‘generation’ (> Lat. genus, Gk. «, Skt. jánas-) from *genh1- ‘to generate, bring into being’ or *demh2-s ‘construction’ (> Gk. «) from *demh2- ‘to build’ (with an old apophonic variant found especially with roots in *-h2-, yielding Gk. -«, Skt. -ifls-).48 From *ser- we would thus expect a noun *ser-os n. ‘attachment, connection, union’. In Indo-Iranian, such a *seros n. should result in *sáras-, i.e. in a homonym of sáras- ‘pond’ (< *selos n.; cf. Gk. o« ‘pond, swamp’49). The latter sáras- unquestionably appears in the name of the Vedic river and river goddess Sáras-vati (< *Selos-wnt-ih2), which corresponds to the Old Persian toponym (= /harauvati-/, Av. haraxvaiti-). By and large, the Vedic Sarasvati is in fact endowed with the qualities one might expect of a river goddess.50 Yet, in comparison both with other river divinities (e.g. Sindhu, Ganga ˙ ¯ ) and with other Vedic goddesses more generally, Sarasvati has a special status as her personification is much stronger. For example, she has a son called Divoda¯sa (RV. 6.61.1), she is a friend of the Maruts (RV. 7.96.2), and she helps the As´vin (RV. 10.131.5); one of the Vedic hymns calls her quite simply ‘best’ not only of rivers but of mothers and goddesses (RV. 2.41.16). Very cautiously we might therefore wonder whether Sarasvati is not perhaps more than a mere river goddess in origin – all the more since she also has a male counter-
47
48 49
50
See Chantraine (1968–80: 2.771, s.v. 4 ), B. Mader in LfgrE. 3.481–482, s.vv. 5 Q., 5 1«, 4 «. Cf. Stüber (2002: 217–248). See Stüber (2000) and (2002: 147–149); differently Janda (2005: 217): ‘Sáras- […] geht auf *séros- zurück und bezeichnete ursprünglich die zusammenhängende “Ausdehnung” einer Wasserfläche’. Cf. Hillebrandt (1891–1902: 2.335–340), Macdonell (1897: 86–88).
254
Andreas Willi
part, Sárasvat. Together with him, and again rather exceptionally for a Vedic goddess, Sarasvati is invoked in two entire hymns (RV. 7.95, 7.96). In one of these, a verse relating to Sarasvat is particularly suggestive (RV. 7.96.4): janiyánto nú ágravahfl putriyántahfl sudánavahfl Sárasvantam fl hava¯mahe ‘longing for a wife and longing for sons, we, the unmarried men, invoke Sarasvat with rich offerings’ Apparently there is some connection between Sarasvat and marriage, just as there is a connection between Sarasvati ambitama¯ (‘best of mothers’) and the fact of having children. Is this due to nothing but the association of any river (or river goddess) with fertility? Or do we grasp here the trace of an older Sarasvati who is not ‘endowed with ponds’ (*selos) but ‘endowed with attachments/unions’ (*seros vel sim.: cf. section 8)? We cannot know, and the whole idea may be too speculative to be defended forcefully; but even if it were, it should not prevent us from looking for *seros or the like elsewhere. Let us therefore return to Greece once again.
8. Eros, personified ‘Love’ As pointed out above (section 4), the root *ser- ‘attach, connect’ is represented in Greek by the y-present c . < *ser-ye/o-. The absence of an initial aspiration in this form is remarkable. According to the etymological dictionaries it may be ascribed to the analogical influence of a synonymous compound - . in which the aspiration would have been lost word-internally.51 From parallel cases such as Ϊ. next to -. ‘to attach’ it is evident that such an explanation is entirely ad hoc. Moreover, borrowed words such as Lat. synhodus ‘indiquent encore, pour l’époque hellénistique, la prononciation 1 h «’,52 so that it cannot even be said that * h . should have lost the internal aspiration automatically. The problem is 51
52
Frisk (1960–72: 1.469, s.v. 1. c .), Chantraine (1968–80: 1.325, s.v. c .), who both note that a trace of an initial aspiration survives in EM. 304.30; more cautiously Schwyzer (1939: 304) observes that ‘[i]n einzelnen Fällen sind psilotische Formen über ihr eigentliches Gebiet hinaus gedrungen’. Lejeune (1972: 279).
Hera, Eros, Iuno Sororia
255
therefore a very real one and it may be appropriate to consider an – albeit equally tentative – alternative explanation. The shape of the Old Irish nasal present sernaid ‘to put in order, arrange’, which is cognate with Gk. c . and Lat. sero¯ (section 4), is best accounted for if the underlying root was *serh2- rather than *ser-.53 For all the other forms which we have previously traced back to a root *ser-, *serh2- would also be a viable possibility. If a root noun *dom-/ *dem- ‘house’ appears next to a seflt root *demh2- ‘to build’, the same may be the case for *sor-/*ser- ‘woman’ next to *serh2-: in a nominative **domh2-s or **sorh2-s, the laryngeal would have fallen away according to Saussure’s Law, so as to yield *dom-s > *do¯m and *sor-s > *so ¯ r;54 from here, the root variant without final laryngeal could easily spread throughout the paradigm. Similarly, the aniflt root *ser- of Lat. sero¯ may be extracted from thematic forms such as 1sg. pres. *serh2-o¯ > ser-o¯. In Greek, too, the root-final laryngeal would have disappeared regularly in a y-present *serh2-ye/o- > *ser-ye/o- (cf. e.g. *terh1-ye/o- ‘to pierce’ > *ter-ye/o- > . next to Lat. tero¯, but Gk. ‘drill’ < *terh1-tro-55). At the same time, the second laryngeal is the laryngeal which triggers an aspiration of the preceding consonant in Indo-Iranian, as in Skt. pflrthú- ‘wide’ < *prth2ú-.56 Greek 1« does not show the same effect, but this may be due to analogical levelling; for the 2sg. perf. ending *-th2e yields both Skt. -tha and Gk. -θa (e.g. véttha ~ ρ% ‘you know’). If, therefore, the aspiratory effect of *-h2- seems to have operated, at least occasionally, in Proto-Greek as well, it might be postulated that, before the loss of *-h2-, the preceding *-r- in *serh2-ye/o- was also aspirated, resulting in a voiceless *-rh-. After the change of *s- into /h-/, *serh2-ye/o- > *serh-ye/o- > *herh-ye/o- would then have lost the initial aspiration due to Grassmann’s Law, the ultimate outcome being unaspirated c ..57 53
54 55 56 57
Cf. Rasmussen (1989: 198), who also points to Latv. s˜ert ‘Getreide zum Trocknen zurechtlegen’, Lith. pasártis ‘Hängevorrichtung, Kleiderrechen’ and suggests ‘dass o nach c . […] umgebildet worden ist’. See R. Lipp in LIV. 115, s.v. 1. *demh2-. Cf. Beekes (1969: 234), Peters (1980: 81 n. 38). Mayrhofer (1986: 139). By contrast, the initial aspiration would have survived in derivatives like Homeric o ‘ear-rings’ because the voiceless *-rh- would have been resonorized before the following voiced consonant (*herh-m- > *her-m-).
256
Andreas Willi
Whether this is correct or not, what must be retained in any case is the observation that (a) a reconstruction of the underlying root as *serh2- instead of simple *ser- is possible, even likely, and (b) the initial aspiration which is expected in Greek is unquestionably absent in some lexemes belonging to this root. From this it follows that we should postulate, instead of the neuter s-stem noun *ser-os hypothesized in section 7, a proto-form *serh2-os or, given *demh2-s etc., *serh2-s.58 By regular sound law the latter would develop into historical *o « or, taking into account the occasional psilotic nature of Greek derivatives of *ser-, ** «. As such ** « is not attested, but there are a number of derivatives, like (
« ‘lovely’ (< *( - «), ( - « ‘lovely’, ( -« ‘lover’. Obviously these belong to the wider lexical family of * and ( . ‘to love’, two verbs which often carry sexual connotations in Homer (as exemplified, for instance, by Il. 14.328 quoted in section 1). Despite its importance and frequency, this whole group of Greek ‘love’ words has so far been lacking an established etymology and obvious relatives in other languages.59 On the basis of the preceding remarks we are now able to explain * , ( . with the help of the same root *serh2- ‘to attach, connect’ as found in Greek c ., no matter if the athematic (and thus 58
59
It should be stressed that the hypothesis concerning Sáras-vati in section 7 does not depend on the existence of a neuter *serh2os > *sáras- instead of *serh2-s (> *sarifls-; cf. below on Av. ha¯irisˇ-i), for the base word could be an animate s-stem *sarás- (cf. Skt. jarás- f. < *gerh2-e/os- next to Gk. « < *gerh2-s; Stüber 2002: 83–84); on such an animate formation derived from *serh2- see below. Moreover, one may cite parallels such as távas- n. ‘strength’, tavás‘strong’ but taviflsá- ‘strong’, or támas- n. ‘darkness’ but támisra¯- f. ‘darkness’ (Debrunner 1954: 233–234). Cf. Chantraine (1968–80: 1.363–364, s.v. * ), Frisk (1960–72: 1.547, s.v. * ). There has, however, been a recent suggestion by Weiss (1998: 35–47) who seeks to reconstruct a root *h1erh2- ‘to divide’ (hence middle ‘to divide for oneself’ > ‘to enjoy’ > ‘to love’) which would be reflected also in Lat. o¯ra ‘seacoast’, OIr. or ‘border’, Hitt. arha ¥ ¯ s ‘border’, as well as Gk. * « ‘shared picnic’ (against Brugmann 1902/3: 153–155, who posits initial ζ-) and * « ‘strife’ (< ‘division’?). Although elegant, Weiss’s proposal may account less well for the prominent early sexual nuances of * etc. (cf. H. W. Nordheider in LfgrE. 2.669 and 2.714–715, s.vv. * and * «) than for * « and * «. Incidentally, the latter of these (which is explained in yet another way by Haudry 1993: 179–181) could also be linked with *serh2- ‘to connect’ via an original i-stem derivative *serh2-i(d)- referring to an ‘encounter’ of enemies (cf. the Homeric meaning ‘fight, battle-strife’).
Hera, Eros, Iuno Sororia
257
presumably old60) * is a retrograde formation based upon ** « or an ancient root present *serh2-mai (with Narten ablaut and hence full-grade root in the middle). Semantically it will have meant ‘to attach oneself/be attached (to someone/something)’ in either case, alongside the noun ** « which referred to the emotional ‘attachment’ itself. The neuter ** « was at some point replaced by an animate counterpart, * .« ‘love’, which is also the name of personified ‘Love’ in Greek. The relationship between ** «, ( ., and * .« is exactly the same as that between *« ‘laughter’ < *gelh2-s n. (presupposed by derivatives such as -« ‘laughter’ and $--« ‘not laughing’), . ‘to laugh’, and .« ‘laughter’. Both .« and * .« show a t-stem inflection (gen. .«, * .«) in classical Greek, but t-stem forms are not yet found in Homer and Hesiod and must be as secondary as the thematized variant * « (attested already in Homer).61 Thus, despite its masculine gender and different accentuation62 * .« ultimately belongs to the same class of animate s-stem nouns as (e.g.) Gk. > <« f. ‘shame’ or (Ionic) <« f. ‘dawn’ ~ Skt. uflsás- f. (< *h2eus-o¯s). Whether these originated from ancient collectives (with nom. sg. *-o¯s < **-os-h2) or from ‘real’ animates (with nom. sg. *-o¯s < **-os-s)63 – or both – makes little difference for our purposes: * .« ‘love’ could equally well be conceived of as a collective group of ‘acts of love’ or as a directly personified, animate ‘Love’.
60
61
62
63
Cf. Weiss (1998: 37): ‘The one example of thematic inflection in Homer, ( % with diectasis from ( % at Iliad 16.208, is a metrically determined Streckform’. See Risch (1974: 88), Weiss (1998: 36–37), and Stüber (2002: 90). According to Meissner (2006: 158), ‘for - and ( - we do not even have to assume the existence of independent neuter nouns in -«’ as ‘these may simply contain the old ablauting stems belonging to .« and * .«’, but given the undeniable existence of equations such as Gk. (ζ)« ~ Skt. kravíhfl < *krewh2-s n. (next to Lat. cruor m. < animate *krewh2o¯s) it definitely cannot be excluded either; for the etymological argument developed here, the matter is of little consequence anyway. The latter must be due to different columnarisations in an originally amphikinetic inflection: cf. Stüber (2002: 207–208). Cf. on this issue Stüber (2002: 22–26) and Meissner (2006: 153–159).
258
Andreas Willi
9. A relative in Rome: Iu¯no¯ Soro¯ria Greek 5E .« would not have remained without an etymology for so long if there had been obvious cognate s-stem nouns in other IndoEuropean languages; but as long as one was looking for *h1erh2-s, *h1erh2-os- or the like none could be discovered. That things may be different now has already been tentatively suggested in the above discussion of Skt. Sáras-vati (section 7, and note fn. 58 on saras-, not †sarifls-). More confidently, we may now equate Gk. ** « < *serh2-s with the basis of the Avestan derivative ha¯irisˇ-i ‘female’ < *serh2s-ih2 highlighted by Benveniste (section 5) and literally to be understood as ‘she of the attachment/union’ (= ‘partner’).64 At first sight, unfortunately, Western Indo-European appears to have nothing to add: for even if Old Irish serc ‘love’ belonged to our word family (cf. section 7), it would clearly not represent an s-stem.65 But a closer look does reveal some interesting material in Rome. In Latin the Indo-European animate s-stem nouns are continued by the type in nom. -or, gen. -o¯ris, whose -r- is due to rhotacism (cf. e.g. honor, hono¯ris ‘honour’, next to an older nom. hono¯s and a derivative hones-tus ‘honourable’ in which the stem-final *-s- is still recognisable). Occasionally a neuter s-stem may coexist with an animate one, as with decus ‘distinction, glory’ (*dek-os) and decor ‘good looks, beauty’ (*dek-o¯s) from the root of decet ‘it is becoming’.66 For our root *serh2-, we should thus expect either *serus n. or *seror m., neither of which is found. However, a number of Latin s-stems nouns show an unexpected vocalism -o- instead of -e- in the root syllable. In color
64
65
66
The long -a¯- of ha¯irisˇ-i need not point to *-o- since lengthening in the first syllable of a plurisyllabic word is possible: cf. Hoffmann and Forssman (1996: 56–57). Alternatively, one may think of a vflrddhi derivative from (*serh2s >) IIr. *saris-, i.e. fem. *sa¯ris-i next to a hypothetical masc. *sa¯risa- ‘he of the attachment/union’. OIr. serc, which acts as the action noun of caraid ‘to love’, is conventionally connected with Gk. . ‘to love’, but *sterga¯ should not yield *sterka¯ > serc, Mod. Ir. searc. Janda (2000: 121–122) therefore proposes a link with the root *serk- of Lat. sarcire ‘to repair’, positing an original meaning ‘to connect’; but one might also think of a derivative in *-ka¯ (cf. e.g. Gk. θ0 ‘box’ from *dheh1‘to put’, Lat. mica ‘grain, particle’ from *meiH- ‘to diminish’) from our root *ser(h2)- once this had become synchronically aniflt in formations like ser-naid. Cf. Skt. *das´as- in das´asyáti ‘to honour’; Stüber (2002: 77).
Hera, Eros, Iuno Sororia
259
‘colour’ < *kel-o¯s (cf. cela¯re ‘to hide’, OIr. -ceil ‘to hide’), this may be due to the following velar -l-, but for honor < *ghen-o¯s and glom-us < *glem-os ‘ball-shaped mass’ vowel assimilation to the vowel-colour of the following syllable must be assumed (cf. similarly homo¯ ‘man’ < *hemo ¯ , as in nemo¯ < *ne-hemo¯).67 Both if such a (somewhat irregular68) assimilation could take place also over -r- and if cases like color and honor acted as analogical models, we might therefore consider the possibility of finding *soror (< *soro¯s) rather than *seror (< *sero¯s).69 In a post-rhotacistic period the former would of course have been a homonym of soror ‘sister’ (< *swe-so¯r; cf. section 5). The semantic oddity of some apparent derivatives of soror has long been noted. One is found in a fragment of Plautus, which is quoted by the lexicographer Festus (p. 380 L., restituted after Paul. Fest. p. 381 L., with Plaut. fr. 84): sororiare mammae dicuntur puellarum, cum primum tumescunt, ut fraterculare puerorum. Plautus in Frivolaria […]: <“Tunc> papillae pri<mulum fraterculabant; illud> volui dicere, so soro¯ria¯re is said of the breasts of girls when they swell for the first time, just like fra¯tercula¯re of those of boys. Plautus says in the Frivolaria […]: “Then, for the very fist time, the nipples ‘brothered’ – oh, no, I wanted to say they ‘sistered’.” In order to understand this entry, we must first stress that fra¯tercula¯re is merely a Plautine pun, whereas soro¯ria¯re (or soro¯rire) is attested with the same meaning as in Festus also in Pliny the Elder (NH. 31.66).70 Thus, Plautus clearly saw a connection between an existing verb soro¯ria¯re and the noun soror ‘sister’; for otherwise he would not have created fra¯tercula¯re. What is less clear is whether this is not just a folk-etymology. First of all, the semantics of soro¯ria¯re raises doubts since the verb has nothing to do with a ‘sister’. Moreover, soro¯ria¯re is patently derived from a Latin adjective soro¯rius, which does some67
68 69
70
Cf. Leumann (1977: 101), who also lists color (but see Leumann 1977: 47), Meiser (1998: 83). Note e.g. tenor, tremor beside honor, homo¯. A further possibility is a folk-etymological replacement of *sero¯r- by soro¯r- in *sero¯rios f soro¯rius etc. (on which see below). Cf. Gaertner (2003).
260
Andreas Willi
times mean ‘belonging to the sister, of the sister’, but which also appears in the peculiar term tigillum soro¯rium (cf. CIL. 6.2295). The tigillum soro¯rium near the compitum Acili in Rome was a wooden beam (tigillum) forming a doorway across a road, with an altar for Ia¯nus Cu¯ria¯tius to one side and an altar for Iu¯no¯ Soro¯ria to the other (Schol. Bobb. Milon. p. 277 Orelli, Dion. Hal. AR 3.22.7). The Roman tradition, which we find in Livy (1.26.13) and again in Festus (p. 380 L.; cf. Paul. Fest. p. 399 L.), preserves the following aetiological account. At a time when Rome was at war with Alba Longa over the supremacy in the alliance between the two cities, their respective kings agreed that the dispute should be finally resolved by a duel between the three Curiatii brothers representing Alba and the three Horatii brothers representing Rome. During the fight five of the six men are killed, but one of the Roman Horatii survives. He triumphally returns to his city where he is met by a crowd including his sister Horatia, who is also the fiancée of one of the Curiatii. On seeing the cloak of her beloved over her brother’s shoulder she calls out in desperation. Infuriated by this, Horatius seizes her and kills her with his sword. Everybody is shocked and Horatius is duly taken into court; but because of his recent service to the country he is acquitted. In order for Horatia’s killing not to remain uncleansed, however, the siblings’ father has to perform a sacrifice in the name of the state, and he erects the tigillum soro¯rium underneath which his son passes as under a yoke. Again this story tries to establish an etymological connection between the adjective soro¯rius and the word for ‘sister’, but again the result seems a little forced. As H. J. Rose has pointed out in an ingenious clarification of the whole complex, the essential point is the link of the tigillum with Iu¯no¯ Soro¯ria. More obliquely this goddess’s adjectival epithet is also at stake in a ceremony taking place each year on the occasion of the Roman Ma¯tra¯lia, an old festival for married women centred around Ma¯ter Ma¯tu¯ta, the ‘Early/Mature ( ? ) Mother’. According to our sources the Ma¯tra¯lia rites included a prayer said by the women not for their own children, but for those of their sisters (Ov. Fast. 6.553–562, Plut. Mor. 267e). This peculiarity, Rose argued, must have arisen from a misunderstanding. The relevant prayer formula will have been something like pueris soro¯riis, but this did not originally mean ‘for the sister’s children’; instead, it meant ‘for the children at the age of puberty’, cum primum tumescunt mammae puellarum, as Festus says in his entry on soro¯ria¯re. Hence, the passage underneath
Hera, Eros, Iuno Sororia
261
the tigillum soro¯rium is nowadays interpreted as originating from a rite de passage in the strict sense of the term, which took place when the young Romans became soro¯rii, i.e. when they reached puberty and turned into adults.71 So, to follow Rose in admitting (1) that soro¯rius must have meant ‘relating to puberty, adolescent’, having nothing to do with the word for ‘sister’, (2) that soro¯ria¯re referred to the physical changes happening to girls during puberty, and (3) that Iu¯no¯ Soro¯ria should therefore be understood as the goddess Juno as the patroness of adolescent young women (whereas Ia¯nus Curia¯tius will have taken on the same role for the young men as they became members in a cu¯ria72) means to account well for the usage of all these words in Roman culture and religion, but it does not yet explain their origin.73 Thanks to the above establishment of an early Latin *soro¯s as a formal correspondent of Gk. * .«, we may now fill the gap. That the Roman Iu¯no¯ is the equivalent of the Greek 6H ¯ both in her capacity as the wife of the supreme god Iuppiter ~ Z1« and as a divinity taking care of adult women is well known: every Roman man has a genius, every woman a iu¯no¯.74 It will not therefore surprise us if we also find Iu¯no¯ Soro¯ria protecting Roman young women like Horatia, adolescents who are about to become adults by getting married, when Hera does the same for the nubile young women of Greece who are ready for the * .« of marriage. In this role Juno is Soro¯ria because she accompanies the awakening of *soro¯s: the ability and wish to enter into an intimate marital union. 71
72 73
74
Rose (1925: 407–410) on the tigillum and (1934: 156–157) on the Ma¯tra¯lia; cf. Latte (1960: 97 and 133). Cf. Ogilvie (1965: 117). Rose (1934: 157) proposes a connection between soro¯rius and the root *swel(H)- ‘to swell’ (LIV. 609–610, s.v.? *sÍelH-), but no other Latin example of an assimilation *-l-r- > -r-r- is given and exactly the inverse is normally the case (cf. Leumann 1977: 231 on -r-r- and -l-l- > -l-r- or -r-l-). The speculations of Gaertner (2003: 247–248 n. 9) about a further link between *swel(H)- and *sweso¯r ‘sister’ are implausible, but he is right in rejecting as ‘naive’ the explanation of soro¯ria¯re as ‘être comme des soeurs’ in Mignot (1969: 281); as for Leumann (1977: 547), the comparison with Gk. π2. ‘to be in the bloom of youth’ regards only the formation in -ia¯re instead of -a¯re, but it is unnecessary when one recognises in soro¯ria¯re a derivative not of soror, but of soro¯rius ‘relating to puberty, adolescent’ (cf. Mignot 1969: 309–310). Cf. Latte (1960: 104), Graf (1999: 72–73 and 75).
262
Andreas Willi
References Adams, Douglas Q. (1987). ‘6H .« and 6H ¯ : Of men and heroes in Greek and Indo-European’. Glotta 65: 171–178. Ambrosini, Riccardo (1962). ‘Etimologia di lat. uxor’. Studi e Saggi Linguistici 2: 23–32. Ambrosini, Riccardo (1963). ‘Ancora a proposito di lat. uxor’. Studi e Saggi Linguistici 3: 86. Ascoli, Graziadio (1864). ‘uxor (vaça, vacca)’. Zeitschrift für Vergleichende Sprachforschung 13: 157–160. Aura Jorro, Francisco (1985–93). Diccionario micénico (2 vols.). Madrid: Consejo Superior de Investigaciones Científicas. Beekes, Robert S. P. (1969). The Development of the Proto-Indo-European Laryngeals in Greek. The Hague and Paris: Mouton. Benveniste, Émile (1935). ‘Un nom indo-européen de la “femme”’. Bulletin de la Société de Linguistique 35: 104–106. Benveniste, Émile (1969). Le vocabulaire des institutions indo-européennes. 1: Économie, parenté, société. Paris: Éditions de Minuit. Borgeaud, Willy (1944). ‘Homer. f"«’. Zeitschrift für Vergleichende Sprachforschung 68: 221–222. Bremmer, Jan (2005). Art. ‘Hera’, in Lindsay Jones (ed.), Encyclopedia of Religion (15 vols., 2nd edn.). Detroit and London: Macmillan. Brugman(n), Karl (1876). ‘Zur Geschichte der stammabstufenden Declinationen. I: Die Nomina auf -ar- und -tar-’. Curtius’ Studien zur griechischen und lateinischen Grammatik 9: 361–406. Brugmann, Karl (1902/3). ‘Wortgeschichtliche Miszellen’. Indogermanische Forschungen 13: 144–163. Burkert, Walter (1985). Greek Religion (trad. John Raffan). Cambridge, Mass.: Harvard University Press. Carruba, Onofrio (1991). ‘Searching for woman in Anatolian and Indo-European’, in Perspectives on Indo-European Language, Culture and Religion: Studies in Honor of Edgar C. Polomé, I. McLean, Va.: Institute for the Study of Man, 155–181. Chantraine, Pierre (1933). La formation des noms en grec ancien. Paris: Klincksieck. Chantraine, Pierre (1958). Grammaire homérique. I: Phonétique et morphologie (3rd edn.). Paris: Klincksieck. Chantraine, Pierre (1968–80). Dictionnaire étymologique de la langue grecque: Histoire des mots (2 vols.). Paris: Klincksieck. Cowgill, Warren (1957). ‘Old Irish teoir and cetheoir’. Language 33: 341–345. Darms, Georges (1978). Schwäger und Schwager, Hahn und Huhn: Die VrddhiAbleitung im Germanischen. München: Kitzinger. de la Genière, Juliette (ed.) (1997). Héra: Images, espaces, cultes. Naples: Centre Jean Bérard. de Vaan, Michiel (2008). Etymological Dictionary of Latin and the Other Italic Languages. Leiden: Brill.
Hera, Eros, Iuno Sororia
263
Debrunner, Albert (1954). Altindische Grammatik. II.2: Die Nominalsuffixe. Göttingen: Vandenhoeck & Ruprecht. Eichner, Heiner (1974). ‘Zu Etymologie und Flexion von vedisch strâ und púma¯n’. Sprache 20: 26–42. Eitrem, Samson (1912). Art. ‘Hera’. RE 8/1: 369–403. Fick, August (1875). ‘Beispiele zur Veranschaulichung des Verhältnisses zwischen den Voll- und Kosenamen im Griechischen’, in Georg Curtius (ed.), Studien zur griechischen und lateinischen Grammatik, VIII. Leipzig: Hirzel, 303–313. Frisk, Hjalmar (1960–72). Griechisches Etymologisches Wörterbuch (3 vols.). Heidelberg: Winter. Gaertner, Jan Felix (2003). ‘Plautus’ Frivolaria und die Wortgeschichte von sororiare und frat(e)r(cul)are’. Philologus 147: 245–253. Gallavotti, Carlo (1956). ‘La triade lesbia in un testo miceneo’. Rivista di Filologia e di Istruzione Classica n.s. 34: 225–236. Gates, H. Phelps (1971). The Kinship Terminology of Homeric Greek. Baltimore: Waverly Press. Graf, Fritz (1998). Art. ‘Hera. I. Kult und Mythos’. Der Neue Pauly 5: 357–360. Graf, Fritz (1999). Art. ‘Iuno. I. Kult und Mythos’. Der Neue Pauly 6: 72–76. Grassmann, Hermann (1873). Wörterbuch zum Rig-Veda. Leipzig: Brockhaus. Hamp, Eric P. (1981). ‘Remarks on *ster-’. Münchener Studien zur Sprachwissenschaft 40: 35–38. Hamp, Eric P. (1988). ‘*sor- “woman” and “Indo-Hittite”’. Journal of Indo-European Studies 16: 121–122. Haudry, Jean (1993). ‘Altindisch arí-, griechisch * «, ( -, $ - und der Gott Aryaman’, in Gerhard Meiser (ed.), Indogermanica et Italica: Festschrift für Helmut Rix zum 65. Geburtstag. Innsbruck: Institut für Sprachwissenschaft, 169–189. Heubeck, Alfred (1979). ‘Remarks on the sign-doublets ro2, ra2, ta2’, in Ernst Risch and Hugo Mühlestein (eds.), Colloquium Mycenaeum: Actes du sixième Colloque international sur les textes mycéniens et égéens. Genève: Droz, 239–257. Hillebrandt, Alfred (1891–1902). Vedische Mythologie (3 vols.). Breslau: Koebner. Hoffmann, Karl, and Forssman, Bernhard (1996). Avestische Laut- und Flexionslehre. Innsbruck: Institut für Sprachwissenschaft. Janda, Michael (1999). ‘Zur Herkunft von homerisch 4 ’, in Sigrid DegerJalkotzy, Stefan Hiller, and Oswald Panagl (eds.), Floreant Studia Mycenaea, II. Wien: Österreichische Akademie der Wissenschaften, 315–324. Janda, Michael (2000). Eleusis: Das indogermanische Erbe der Mysterien. Innsbruck: Institut für Sprachwissenschaft. Janda, Michael (2005). Elysion: Entstehung und Entwicklung der griechischen Religion. Innsbruck: Institut für Sprachwissenschaft. Kazansky, Nikolai N. (1995). ‘Indo-European onomastics as a historical source’. Journal of Indo-European Studies 23: 157–177. Kazzazi, Kerstin (2001). “Mann” und “Frau” im Rgveda: Mit einem Exkurs über Wörter für “Frau” im Atharvaveda. Innsbruck: Institut für Sprachwissenschaft. Kerényi, Karl (1972). Zeus und Hera. Urbild des Vaters, des Gatten und der Frau. Leiden: Brill.
264
Andreas Willi
Kossatz-Deissmann, Anneliese (1988). Art. ‘Hera’. Lexicon Iconographicum Mythologiae Classicae IV/1: 659–719. Kretschmer, Paul (1909). ‘Zur Geschichte der griechischen Dialekte’. Glotta 1: 9–59. Kretschmer, Paul (1914). ‘Literaturbericht für das Jahr 1911: Griechisch’. Glotta 5: 259–313. Kronasser, Heinz (1966). Etymologie der hethitischen Sprache. 1/I. Zur Schreibung und Lautung des Hethitischen, II. Wortbildung des Hethitischen. Wiesbaden: Harrassowitz. Latte, Kurt (1960). Römische Religionsgeschichte. München: Beck. Lejeune, Michel (1972). Phonétique historique du mycénien et du grec ancien. Paris: Klincksieck. Leumann, Manu (1977). Lateinische Laut- und Formenlehre. München: Beck. Ley, Anne (1998). Art. ‘Hera. II. Ikonographie’. Der Neue Pauly 5: 360–361. LfgrE.: Lexikon des frühgriechischen Epos. Göttingen: Vandenhoeck & Ruprecht. Linke, Uli (1985). ‘Blood as metaphor in Proto-Indo-European’. Journal of IndoEuropean Studies 13: 333–376. LIV.: Rix, Helmut (ed.) (2001). Lexikon der indogermanischen Verben: Die Wurzeln und ihre Primärstammbildungen. Wiesbaden: Reichert. Lohmann, Johannes (1936). Review of Edgar H. Sturtevant, A Comparative Grammar of the Hittite Language (Philadelphia: Linguistic Society of America). Indogermanische Forschungen 54: 285–294. Macdonell, Arthur Anthony (1897). Vedic Mythology. Strassburg: Trübner. Mastrelli, Carlo Alberto (1980). ‘Un falso arcaismo: lat.-fal. uxor. pel. usur’. Archivio Glottologico Italiano 65: 14–18. Mayrhofer, Manfred (1952). ‘Gibt es ein indogermanisches *sor- “Frau”?’, in Wilhelm Brandenstein (ed.), Studien zur indogermanischen Grundsprache. Wien: Gerold & Co., 32–39. Mayrhofer, Manfred (1986). Indogermanische Grammatik. I/2: Lautlehre [Segmentale Phonologie des Indogermanischen]. Heidelberg: Winter. Mayrhofer, Manfred (1986–2001). Etymologisches Wörterbuch des Altindoarischen (3 vols.). Heidelberg: Winter. Meier-Brügger, Michael (1989). ‘Verbaute schwundstufige -s-Neutra in der griechischen Wortbildung’. Historische Sprachforschung 102: 58–61. Meier-Brügger, Michael (2002). Indogermanische Sprachwissenschaft (8th edn.). Berlin: de Gruyter. Meillet, Antoine (1931). ‘Essai de chronologie des langues indo-européennes. La théorie du féminin’. Bulletin de la Société de Linguistique 32: 1–28. Meiser, Gerhard (1998). Historische Laut- und Formenlehre der lateinischen Sprache. Darmstadt: Wissenschaftliche Buchgesellschaft. Meissner, Torsten (2006). S-Stem Nouns and Adjectives in Greek and Proto-IndoEuropean. Oxford: Oxford University Press. Melchert, H. Craig (1994). Anatolian Historical Phonology. Amsterdam: Rodopi. Meringer, Rudolf (1904). ‘Wörter und Sachen’. Indogermanische Forschungen 16: 101–196. Mignot, Xavier (1969). Les verbes dénominatifs latins. Paris: Klincksieck.
Hera, Eros, Iuno Sororia
265
Moreschini Quattordio, Adriana (1979). ‘6H ed j .«: un tentativo di esegesi etimologica’. Studi e Saggi Linguistici 19: 167–198. Moussy, Claude (1980). ‘Une étymologie de lat. uxor’. Bulletin de la Société de Linguistique 75: 325–346. Nilsson, Martin P. (1906). Griechische Feste von religiöser Bedeutung mit Ausschluss der attischen. Leipzig: Teubner. Nilsson, Martin P. (1967). Geschichte der griechischen Religion. I: Die Religion Griechenlands bis auf die griechische Weltherrschaft (3rd edn.). München: Beck. Normier, Rudolf (1980). ‘Nochmals zu *sor-’. Indogermanische Forschungen 85: 43–80. Oettinger, Norbert (1986). ‘Avestisch ha¯irisˇi- “Frau” syn- und diachron’. Indogermanische Forschungen 91: 116–128. Ogilvie, Robert M. (1965). A Commentary on Livy: Books 1–5. Oxford: Oxford University Press. Palmer, Leonard R. (1963). The Interpretation of Mycenaean Greek Texts. Oxford: Oxford University Press. Pârvulescu, Adrian (1989). ‘Blood and IE. kinship terminology’. Indogermanische Forschungen 94: 67–88. Pestalozza, Uberto (1939). ‘B7« 6H 0’. Athenaeum 17: 105–137. Peters, Martin (1980). Untersuchungen zur Vertretung der indogermanischen Laryngale im Griechischen. Wien: Österreichische Akademie der Wissenschaften. Peters, Martin (2002). ‘Aus der Vergangenheit von Heroen und Ehegöttinnen’, in Matthias Fritz and Susanne Zeilfelder (eds.), Novalis Indogermanica: Festschrift für Günter Neumann zum 80. Geburtstag. Graz: Leykam, 357–380. Pisani, Vittore (1951). ‘Uxor. Ricerche di morfologia indo-europea’, in Miscellanea Giovanni Galbiati, III. Milano: Hoepli, 1–38. Pokorny, Julius (1959). Indogermanisches etymologisches Wörterbuch. Tübingen and Basel: Francke. Pötscher, Walter (1961). ‘Hera und Heros’. Rheinisches Museum 104: 302–355. Pötscher, Walter (1965). ‘Der Name der Göttin Hera’. Rheinisches Museum 108: 317–320. Pötscher, Walter (1987). Hera. Eine Strukturanalyse im Vergleich mit Athena. Darmstadt: Wissenschaftliche Buchgesellschaft. Pott, August Friedrich (1833). Etymologische Forschungen auf dem Gebiete der Indo-Germanischen Sprachen, mit besonderem Bezug auf die Lautumwandlung. Lemgo: Meyersche Hofbuchhandlung. Pott, August Friedrich (1836). Etymologische Forschungen auf dem Gebiete der indo-germanischen Sprachen. II: Grammatischer Lautwechsel und Wortbildung. Lemgo: Meyersche Hofbuchhandlung. Probert, Philomen (2008). ‘Mycenaean o- is accusative; jo- is nominative’. Glotta 84: 126–168. Rasmussen, Jens Elmegård (1989). Studien zur Morphophonemik der indogermanischen Grundsprache. Innsbruck: Institut für Sprachwissenschaft. Rieken, Elisabeth (1999). Untersuchungen zur nominalen Stammbildung des Hethitischen. Wiesbaden: Harrassowitz.
266
Andreas Willi
Risch, Ernst (1974). Wortbildung der homerischen Sprache (2nd edn.). Berlin and New York: de Gruyter. Rose, Herbert J. (1925). ‘De religionibus antiquis quaestiunculae tres’. Mnemosyne n.s. 53: 406–414. Rose, Herbert J. (1934). ‘Two Roman rites’. Classical Quarterly 28: 156–158. Rose, Herbert J. (1958). A Handbook of Greek Mythology including its Extension to Rome (6th edn.). London: Methuen & Co. Ruijgh, Cornelis J. (1967a). Études sur la grammaire et le vocabulaire du grec mycénien. Amsterdam: Hakkert. Ruijgh, Cornelis J. (1967b). ‘A propos de myc. po-ti-ni-ja-we-jo’. Studi Micenei ed Egeo-Anatolici 4: 40–52. Ruijgh, Cornelis J. (1995). ‘D’Homère aux origines proto-mycéniennes de la tradition épique. Analyse dialectologique du langage homérique, avec un excursus sur la création de l’alphabet grec’, in Jan Paul Crielaard (ed.), Homeric Questions. Amsterdam: Gieben, 1–96. Schindler, Jochem (1972). ‘L’apophonie des noms-racines indo-européens’. Bulletin de la Société de Linguistique 67/1: 31–38. Schmitt, Rüdiger (1967). Dichtung und Dichtersprache in indogermanischer Zeit. Wiesbaden: Harrassowitz. Schröder, Franz Rolf (1956). ‘Hera’. Gymnasium 63: 57–78. Schulze, Wilhelm (1887). ‘ - [without title]’. Zeitschrift für Vergleichende Sprachforschung 28: 281. Schulze, Wilhelm (1907). ‘Ahd. suagur’. Zeitschrift für Vergleichende Sprachforschung 40: 400–418. Schwyzer, Eduard (1939). Griechische Grammatik. I: Allgemeiner Teil, Lautlehre, Wortbildung, Flexion. München: Beck. Sihler, Andrew L. (1977). ‘The PIE origins of the Germanic feminine nomina agentis in *-stri(o¯)n’. Sprache 23: 36–48. Stella, Luigia Achillea (1958). ‘La religione greca nei testi micenei’. Numen 5: 18–57. Stella, Luigia Achillea (1965). La civiltà micenea nei documenti contemporanei. Roma: Edizioni dell’Ateneo. Stüber, Karin (2000). ‘Zur Bedeutung und Etymologie von altind. sáras, iranisch *harah- und griechisch o«’. Historische Sprachforschung 113: 132–142. Stüber, Karin (2002). Die primären s-Stämme des Indogermanischen. Wiesbaden: Reichert. Sturtevant, Edgar H. (1949). ‘An Indo-European word for “woman”’. Language 25: 343–345. Szemerényi, Oswald (1967). ‘The alleged Indo-European *sor- “woman”’. Kratylos 11: 206–221. Szemerényi, Oswald (1977a). ‘Das griechische Verwandtschaftsnamensystem vor dem Hintergrund des indogermanischen Systems’. Hermes 105: 385–405. Szemerényi, Oswald (1977b). ‘Studies in the kinship terminology of the IndoEuropean languages with special reference to Indian, Iranian, Greek, and Latin’, in Acta Iranica 16: Varia 1977. Leiden, Teheran, and Liège: Brill and Bibliothèque Pahlavi.
Hera, Eros, Iuno Sororia
267
Thurneysen, Rudolf (1946). A Grammar of Old Irish (2nd edn., trans. D. A. Binchy and Osborn Bergin). Dublin: Institute of Advanced Studies. Van Windekens, Albert J. (1958). ‘6H “(die) junge Kuh, (die) Färse”’. Glotta 36: 309–311. Van Windekens, Albert J. (1960). ‘Spuren einer mittelmeerischen Stierkultur im griechischen Wortschatz’. Sprache 6: 211–219. Van Windekens, Albert J. (1984). ‘Encore lat. uxor et i.-e. *ukson-’. Zeitschrift für Vergleichende Sprachforschung 97: 96–98. Vine, Brent (1998). Aeolic 4 and Deverbative *-etó- in Greek and IndoEuropean. Innsbruck: Institut für Sprachwissenschaft. Weiss, Michael (1998). ‘Erotica: On the prehistory of Greek desire’. Harvard Studies in Classical Philology 98: 31–61. West, Martin L. (1978). Hesiod: Works and Days. Oxford: Oxford University Press. Wilamowitz-Moellendorff, Ulrich von (1889). Euripides: Herakles (2 vols.). Berlin: Weidmann. Willi, Andreas (2007). ‘Demeter, Gê, and the Indo-European word(s) for “earth”’. Historische Sprachforschung 120: 169–194. Willi, Andreas (2008). ‘Genitive problems: Mycenaean -Ca-o, -Co-jo, -Co vs. later Greek -¯ , -, -’. Glotta 84: 239–272.
University of Oxford Worcester College GB-O x f o r d OX1 2HB [email protected]
A n d r e a s Wi l l i
268
Michael Knüppel
I I . B E S P R E C H U N G S AU F S AT Z
Fragen zu den von V. M. Illicˇ-Svitycˇ nachgelassenen Materialien Es mag Gelehrte geben, die in der Zeit eines nur vergleichsweise kurzen Wirkens ein sehr umfangreiches Opus hervorbringen und publizieren, aber auch solche, die in ebenso kurzer Zeit ein beeindruckendes Werk schaffen, dessen Herausgabe ihnen verwehrt bleibt und durch andere besorgt werden muß. Beides zugleich kommt eher seltener vor und einem jeden von uns fallen hier sofort Ausnahmeerscheinungen wie M. A. Castrén (1813–1852),1 der gemessen an seinen wenigen Lebens- und Schaffensjahren ein gewaltiges Œuvre zu Papier gebracht und einerseits publiziert, andererseits aber auch – wenngleich zu einem größeren Teil – nachgelassen hat,2 ein. Ganz ähnlich scheint es sich im Falle des „Begründers“ der Nostratik, wie sie uns in Gestalt der „Moskauer Nostratischen Schule“ begegnet, Vladislav Markovicˇ Illicˇ-Svitycˇ (1934–1966), zu verhalten. Dieser hat in den wenigen Jahren seines wissenschaftlichen Wirkens die Publikation von mehr als ein Dutzend „kleineren“ Beiträgen (Illicˇ-Svitycˇ 1963a, 1964a, 1964b, 1964c, 1965a, 1965b, 1966a u. 1966b) und einer Monographie (1963b) sowie das Erscheinen einer umfangreicheren Arbeit (dem „Makedonsko-russkij slovar’“ – Tolovski/ Illicˇ-Svitycˇ 1963), an der er mitbeteiligt war, erleben dürfen. Ein Teil seiner Schriften erschien unmittelbar nach seinem Verscheiden und scheint noch zu seinen Lebzeiten eingereicht worden zu sein – zu nennen sind hier die Publikationen aus den Jahren 1967–1971 (Illicˇ-Svitycˇ 1967a, 1967b, 1968a, 1968b, 1968c, 1971a – also noch vor dem Erscheinen des ersten Bandes seines „Opyt sravnenija nostraticˇeskich jazykov“). Hinzu tritt weiterhin der zweibändige „Karpatskij dialektologicˇeskij atlas“, an dessen Entstehung Illicˇ-Svitycˇ ebenfalls mitgewirkt hat (Bernsˇtejn/ Illicˇ-Svitycˇ 1967a u. 1967b).
1
2
Zu Leben und Werk M. A. Castréns vgl. hier nur stellvertretend für die kaum überschaubare Literatur Aalto 1971: 83–88 u. Korhonen 1986: 50–66. Die Herausgabe des größten Teils seines wissenschaftlichen Erbes besorgte dann in den Jahren nach seinem Tode der Sibirist Franz Anton Schiefner.
Fragen zu den von V. M. Illicˇ-Svitycˇ nachgelassenen Materialien
269
In den dann folgenden Jahren erschien schließlich auch noch das erwähnte dreibändige „Vergleichende Wörterbuch der nostratischen Sprachen“ („Opyt sravnenija nostraticˇeskich jazykov“, 1971, 1976 und 1984). Womit im Falle des Werkes von Illicˇ-Svitycˇ die Sache ganz ähnlich gelagert zu sein scheint, wie dies bei Castrén der Fall war – mit dem Unterschied, daß Illicˇ-Svitycˇ mit nur 32 Jahren verstarb, während der große finnische Gelehrte (- der zudem auch etwas früher seine Forschertätigkeit begann3 -) im 39. Lebensjahr verschied. Die Herausgabe von Aufzeichnungen Illicˇ-Svitycˇs hatte damit allerdings noch lange keinen Abschluß gefunden. Bis heute tauchen immer wieder Materialien auf, die aus seiner Feder zu stammen scheinen – und selbst aus den unscheinbarsten Angaben, die sich in seinen Publikationen finden, werden neue Beiträge produziert – so etwa aus einer Wortliste, die sich in einer (übrigens auch posthum erschienenen) Vorarbeit Illicˇ-Svitycˇs (1967a) zu seinem „Opyt sravnenija nostraticˇeskich jazykov“ findet (Parkinson 1990). Gelegentlich wurden diese Schriften aus seinem Nachlaß auf diversen Symposien zur Proto-Sprachen-Forschung oder speziell zur Nostratik einem interessierten Publikum bekannt gemacht. Es findet kaum eine nostratische Tagung oder Konferenz statt, auf der nicht die besonderen Verdienste des Begründers dieser Forschungsrichtung gewürdigt werden und bisweilen werden bei solchen Anlässen dann auch noch Übersetzungen von Auszügen aus seinen Werken oder nachgelassene Materialien vorgestellt.4 Dies auch noch nahezu fünf Jahrzehnte nach seinem Verscheiden. Es hat bisweilen geradezu den Anschein als sollen den Veranstaltungen hierbei Segnungen oder höhere Weihen durch den Ordensgründer erteilt werden. Nun lassen aber genau diese quasi-religiösen Züge im Verbund mit einer weit verbreiteten Stagnation im Bereich der nostratischen Forschungen, wie sie etwa in der Nicht-Berücksichtigung der methodischen Fortschritte in der Indogermanistik der vergangenen Jahrzehnte zum Ausdruck kommt (die Methoden der wenigsten Nostratiker sind von jenen der 1960er Jahre – also der Zeit, in der Illicˇ-Svitycˇ wirkte – verschieden),
3
4
Castrén hatte sein Studium mit 23 Jahren abgeschlossen (und zu dieser Zeit bereits eine Arbeit zur finnischen Mythologie geschrieben) und unternahm in jenem Jahr seine erste Forschungsreise (nach Lappland und Karelien). So etwa die Übersetzung des Vorworts zu seinem „Opyt sravnenija nostraticˇeskich jazykov“ (Illicˇ-Svitycˇ 1989a) oder die dreier Einträge aus selbigem (IllicˇSvitycˇ 1989b).
270
Michael Knüppel
sowie einer Reihe von Begleitumständen (vgl. hierzu unten) – auch ohne den „Nachlaßverwaltern“ die Anfertigung von Materialien, die dann dem Begründer der Nostratik zugeschrieben werden, unterstellen zu wollen – die beklemmende Frage aufkommen, inwieweit IllicˇSvitycˇ denn der Urheber all dieser Materialien ist oder sein kann. Jenseits des bereits angesprochenen Umstandes, daß dieser in nur wenigen Jahren (1961/2–1966) ein mehr als 400 Seiten umfassendes Opus, das in den Jahren 1963–1971 herauskam, produziert hat, das „Makedonisch-russische Wörterbuch“ einen Umfang von 576 Seiten aufweist und der zweibändige „Atlas der Karpatendialekte“ einen solchen von 489 Seiten (incl. Karten), war er der Urheber des insgesamt 695 Seiten umfassenden „Vergleichenden Wörterbuchs der nostratischen Sprachen“.5 Vor diesem Hintergrund ist es natürlich schon erstaunlich, daß das Werk eines so produktiven Gelehrten – das inzwischen eine so ungewöhnliche „Rezeptionsgeschichte“ aufzuweisen hat6 – noch immer unpublizierte Materialien umfassen soll. Merkwürdiger als dies ist allerdings, daß trotz der fortgesetzten Publikation von Werken Illicˇ-Svitycˇs bis heute eine systematisch angelegte Übersicht über die von ihm nachgelassenen Schriften ebenso fehlt, wie verläßliche Angaben hinsichtlich des Umfanges oder der Beschaffenheit seines Nachlasses. Es finden sich allenfalls einige verstreute Hinweise auf diesen. So läßt V. V. Sˇevorosˇkin uns in der Einleitung zu einem der Tagungsbände des ersten internationalen, interdisziplinären Symposiums für Sprache und Vorgeschichte in Ann Arbor (8.–12. 11. 1988) wissen: „…, Dybo and Bulatova, who have been carrying out the publication of Illich-Svitych’s heritage (it is not yet fully published)“ (Shevoroshkin 1989: 15), in einem anderen verweist er im Zusammenhang mit einer angestrebten engl. Übersetzung auf die Überarbeitung des Vergleichenden Wörterbuchs: „…, whereas in Russia a work on rewriting of the dictionary, with inclusion of new material, elimination of obsolete data (…), etc., has been undertaken by Vladimir Orël and his colleagues“ (Shevoroshkin 1990: 12),7 in einer Anmerkung zur Übersetzung des Vorworts zu Illicˇ-Svi5 6 7
Bd. 1 (1971): XXXVI+369 pp., Bd. 2 (1976): 155 pp., Bd. 3 (1984): 135 pp. Zu den Übersetzungen von Auszügen und Bezugnahmen aus das Werk vgl. unten. Ergebnisse – zumindest aus der Feder V. E. Orëls – hat es, abgesehen von Stellungnahmen oder „Vorberichte“ keine gegeben und sind nun, nach dem Verscheiden des russ. Indogermanisten im Jahre 2007 auch nicht mehr zu erwarten.
Fragen zu den von V. M. Illicˇ-Svitycˇ nachgelassenen Materialien
271
tycˇs „Opyt sravnenija nostraticˇeskich jazykov“ findet sich die kryptisch anmutende Notiz: „This article is the introduction to Opyt sravneniia nostraticheskikh jazykov (semitokhamitskii, kartvel’skii, indoevropeiskii, ural’skii, dravidiiskii, altaiskii), 2 vols. and vol. 3, part 1 to date (Moscow: Nauka, 1971-), vol. 1, 3–5“ (Illich-Svitych 1989: 111; also noch weitere Teile zu erwarten?) und J. Parkinson läßt uns in seinem Beitrag erahnen, was da noch so auf uns zukommen könnte: „There are 607 Nostratic words in Illich-Svitych’s foundation article „Materials for a Comparative Dictionary of the Nostratic Languages“ (1967), but only 378 words have yet been published in the Nostratic Dictionary volumes (1971-), and only eight beginning with r to z (or ). The Nostratic words from both sources are here combined and alphabetically sequenced; unreconstructed words from „Materials“ are included, with only three important words from other sources. In all, there are 977 entries, which comprise about 690 independent words“ (Parkinson 1989: 122). Weitere Angaben finden sich im Vorwort der Herausgeber des dritten Bandes des „Vergleichenden Wörterbuchs der nostratischen Sprachen“.8 Die Aufstellung solcher Hinweise ließe sich natürlich nahezu beliebig fortsetzen. Zu einem eklatanten Mangel an Bestandsangaben zu den von IllicˇSvitycˇ nachgelassenen Materialien tritt hinzu, daß wir von allen Bestandteilen, die ein wissenschaftlicher Nachlaß gemeinhin aufweist und die dann bisweilen auch an die interessierte Öffentlichkeit gelangen – wie Briefwechsel oder andere Materialien, die die Person Illicˇ-Svitycˇs an Konturen gewinnen ließen, mit Glossen des Verblichenen versehene Arbeiten Anderer u.dgl. – bislang keine Kenntnis erlangt haben. Darüber hinaus ist auffällig, daß in den einschlägigen Publikationen auch keine hand- oder auch bloß mit Anmerkungen versehenen maschinenschriftlichen Aufzeichnungen (etwa als Faksimiles9) gegeben worden sind, wie überhaupt den Begründer der Nostratik eine Aura des NichtGreifbaren – ja des Rätselhaften – zu umgeben scheint.10 8 9
10
Illicˇ-Svitycˇ 1984: 3–11. Von einem Abdruck des Faksimiles von Illicˇ-Svitycˇs bekanntem „nostratischen Gedicht“ (erstmals in Illicˇ-Svitycˇ 1971b: unpaginiert [5]) in dem von V. V. Sˇevorosˇkin herausgegebenen Tagungsband „Nostratic, Dene-Caucasian, Austric and Amerind“ (Shevoroshkin 1992: 82) einmal abgesehen. Rühmliche Ausnahmen – zumindest im Hinblick auf die Behandlung der Person Illicˇ-Svitycˇs – bilden einige der biograph. Skizzen, die gelegentlich vorgelegt wurden. Hervorzuheben ist hier besonders die Darstellung bei Bulatova (1989).
272
Michael Knüppel
Auffällig ist auch, daß der dritte Band des „Vergleichenden Wörterbuchs der nostratischen Sprachen“ augenfällige Abweichungen von den beiden vorangegangenen Bänden sowohl im Hinblick auf die Ordnung der Lemmata als auch in sprachlicher Hinsicht aufweist. Dies ist gerade auch insofern problematisch, als das Wörterbuch posthum erschienenen ist, der dritte Band – vom erwähnten Vorwort abgesehen – keine Einleitung enthält und ohnehin nicht erkennbar ist, welche Teile auf Illicˇ-Svitycˇ zurückgehen und welche möglicherweise vom Herausgeber (V. A. Dybo) und/ oder anderen Vertretern des Umfeldes von Illicˇ-Svitycˇ „ergänzt“ wurden – was dann ja wiederum die Abweichungen der benutzten Sprache erklären dürfte (wenngleich der Herausgeber derselbe, wie im Falle der vorangegangenen Bände war). Selbst dem unvoreingenommensten Betrachter muß sich hier der vage Verdacht aufdrängen, daß mit dem Nachlaß des sowjetischen Linguisten „irgendetwas nicht stimmt“. Den „Nachlaßverwaltern“ ist daher dringend anzuraten, nicht nur den Jüngern Illicˇ-Svitycˇs, sondern auch seinen (und ihren Kritikern!) – schon in ihrem eigenen Interesse – endlich Klarheit über Umfang und Beschaffenheit dieses Nachlasses zu verschaffen, anstatt sich an diesem, wie an einem für jedermann zugänglichen Steinbruch, zu bedienen und zusammenhangslos von Zeit zu Zeit irgendwelche Materialien in Gestalt von Arbeiten, aus denen kaum mehr ersichtlich ist, welchen Anteil Illicˇ-Svitycˇ an diesen hatte, und welchen sie selbst daran haben, zu publizieren. Vielleicht könnte ja auch dies – wie allgemein die Wirkung der Forschungen Illicˇ-Svitycˇs (- ganz gleich wie man diesen gegenübersteht [der Vf. des vorliegenden kleinen Beitrages nimmt diesen gegenüber bekanntlich eine äußerst distanzierte Haltung ein, was nicht mit einer prinzipiellen Ablehnung nostratischer Forschungen verwechselt werden sollte] -) – einmal das Thema für eine Konferenz abgeben!
Literatur Aalto, Pentti, 1971. Oriental studies in Finland 1828–1918. Helsinki. Bernsˇtejn, S. B./ Illicˇ-Svitycˇ, Vladislav Markovicˇ, 1963. Karpatskij dialektologicˇeskij atlas. 2 Bde. Moskva 1967. Bulatova, Rimma V., 1989. Illicˇ-Svitycˇ. A biographical sketch. In: Reconstructing languages and cultures: abstracts and materials from the first international interdisciplinary symposium on language and prehistory, Ann Arbor, 8–12 November 1988. (BPX, 20). Bochum, 14–28.
Fragen zu den von V. M. Illicˇ-Svitycˇ nachgelassenen Materialien
273
Illicˇ-Svitycˇ, Vladislav Markovicˇ, 1963a. Altajskie dental’nye: t, d, . In: VJa 1963 (6), 37–56. –, Vladislav Markovicˇ, 1963b. Imennaja akcentuacija v baltijskom i slavjanskom: sud’ ba akcentuacionnych paradigm. Moskva [engl. Übers. 6 1979]. –, Vladislav Markovicˇ, 1964a. Drevnejsˇie indoevropejsko-semitskie jazikovye kontakty. In: Problemy indoevropejskogo jazykoznanija. Study po sravnitel’no istoricˇeskoj grammatike indoevropejskich jazykov. Red. V. N. Toporov. Moskva, 3–11. –, Vladislav Markovicˇ, 1964b. Genezis indoevropejskich rjadov guttural’nych v svete dannych vnesˇnego sravnenija. In: PSG, 22–26. –, Vladislav Markovicˇ, 1964c. Sledy iscˇeznuvsˇich baltijskich akcentuacionnych sistem. In: KSISl 41, 18–26. ˙ timologija. –, Vladislav Markovicˇ, 1965a. Altajskie guttural’nye: *kc, *k, *g. In: E Principy rekonstrukcii i metodika issledovanija. Red. O. N. Trubacˇev et al. 1964. Moskva, 338–343. ˙ timologija. Principy rekonstrukcii i –, Vladislav Markovicˇ, 1965b. Caucasica. In: E metodika issledovanija. Red. O. N. Trubacˇev et al. 1964. Moskva, 334–337. –, Vladislav Markovicˇ, 1966a. Iz istorii chadskogo konsonantizma: labijal’nye smycˇnye. In: Jazyki Afriki 1966, 9–34. –, Vladislav Markovicˇ, 1966b. K istolkovamju akcentuacionnych sootvetstvija v kel’lo-itahjskom i balto-slavskom kralkie. In: Soobsˇcˇenija instituta slavjanovedenija 35, 63–72. –, Vladislav Markovicˇ, 1967a. Materiały po sravnitel’nomu słovarju nostraticˇeskich jazykov (indojevropejskij ałtajskij, ural’skij, dravidskij, kartvel’skij, semi˙ timologija 1965, Moskva, 321–373. tochamitskij). In: E –, Vladislav Markovicˇ, 1967b. Rekonstrukcija ural’skogo vokalizma v svete dannych vnesˇnego sravnenija. Tezisy doklada. ,Voprosy finno-ugorskogo jazykoznanija’, Vysˇ. IV. Izˇevsk, 95–100. –, Vladislav Markovicˇ, 1968a. Sootvetstvija smycˇnych v nostraticˇeskich jazykach. ˙ timologija 1966 [Problemy lingvogeografii i mecˇ’jazykovych kontaktov]. In: E Moskva, 304–355. –, Vladislav Markovicˇ, 1968b. Korrekturnye primecˇanija k stat’e V. M. Illicˇ-Svi˙ timologija 1966 [Problemy lingvogeografii i mezˇ’jazykovych kontaktycˇa. In: E tov], 401–404. –, Vladislav Markovicˇ, 1968c. Opyt sravnenija nostraticˇeskich jazykov (semitochamitskij, kartvel’skij, indoevropejskij, ural’skij, dravidijskij, altajskij). In: Slavjanskoe jazykoznanie. VI. mezˇdunarodnyj s’ezd slavistov. Doklady sovetskoj delegacii, 407–426. –, Vladislav Markovicˇ, 1971a. Licˇnyje mestoimenija mi ’ja’ i mä ’my’ v nostraticˇeskom. In: Issledovanija po slavjanskomu jazykoznaniju, 936–403. –, Vladislav Markovicˇ, 1971b. Opyt sravnenija nostraticˇeskich jazykov (semitochamitskij, kartvel’skij, indoevropeiskij, ural’skij, dravidijskij, altajskij). 1: V vedenie. Sravnitel’nyj slovar’ (b-k) fl po kartotekam avtora (otvectvennyj redaktor: V. A. Dybo. AN SSSR, Inst. Słavistiki i bałkanistiki). Moskva. –, Vladislav Markovicˇ, 1976. Opyt sravnenija nostraticˇeskich jazykov: (semitochamitskij, kartvel’skij indoevropejskij, ural’skij, dravidijskij, altajskij). [Otv. red. V. A. Dybo]. 2: Sravnitel’nyj slovar’: l-©´. Moskva.
274
Michael Knüppel
–, Vladislav Markovicˇ, 1979. Nominal accentuation in Baltic and Slavic (Imennaja akcentuacija v baltijskom i slavjanskom [1963]). V. M. Illich-Svitych. Transl. by Richard L. Leed and Ronald F. Feldstein. Cambridge Mass. –, Vladislav Markovicˇ, 1984. Opyt sravnenija nostraticˇeskich jazykov: (semitochamitskij, kartvel’skij, indoevropejskij, ural’skij, dravidijskij, altajskij). [Otv. red. V. A. Dybo]. 3: Sravnitel’nyj slovar’: (p-q). Moskva. –, Vladislav Markovicˇ, 1989a. The relationship of the nostratic family languages: a probabilistic evaluation of the similarities in question. In: Explorations in language macrofamilies. Materials from the first international interdisciplinary symposium on language and prehistory, Ann Arbor 1988. Bochum (BPX, 20), 111–113. –, Vladislav Markovicˇ, 1989b. Three entries from the Nostratic dictionary. In: Explorations in language macrofamilies. Materials from the first international interdisciplinary symposium on language and prehistory, Ann Arbor 1988. Bochum (BPX, 20), 122–127. Korhonen, Mikko, 1986. Finno-Ugrian Language Studies in Finland 1828–1918. Helsinki. Parkinson, Jim, 1990. A Nostratic word list: reconstructions by Illich-Svitych. In: Explorations in language macrofamilies. Materials from the first International Interdisciplinary Symposium on Language and Prehistory, Ann Arbor, 8–12 November, 1988. Bochum (BPX, 23), 128–162. Shevoroshkin, Vitaly, 1989. Introductory remarks. In: Explorations in language macrofamilies. Materials from the first International Interdisciplinary Symposium on Language and Prehistory, Ann Arbor, 8–12 November, 1988. Bochum (BPX, 23), 4–15. Shevoroshkin, Vitaly. 1990. Introduction. In: Proto-languages and proto-cultures. Materials from the first International Interdisciplinary Symposium on Language and Prehistory, Ann Arbor, 8–12 November, 1988. Bochum (BPX, 25), 8–12 Shevoroshkin, Vitaly (Hrsg.), 1992. Nostratic, Dene-Caucasian, Austric and Amerind. Materials from the first International Interdisciplinary Symposium on Language and Prehistory, Ann Arbor, 8–12 November, 1988. Bochum (BPX, 33). Tolovski, Dime/ Illicˇ-Svitycˇ, Vladislav Markovicˇ, 1963. Makedonsko-russkij slovar’: 30 000 slov. Sost. Dime Tolovski i V. M. Illicˇ-Svitycˇ. Pod red. Nikita I. Tolstogo. S pril. kratkogo grammaticˇeskogo spravocˇnika, sost. V. M. Illicˇ-Svitycˇem. Moskva.
Universität Göttingen Seminar für Turkologie und Zentralasienkunde Waldweg 26 37073 G ö t t i n g e n [email protected]
Michael Knüppel
Fragen zu den von V. M. Illicˇ-Svitycˇ nachgelassenen Materialien
275
Druckfehlerkorrektur aus IF. 114
S. 150, Beispiel (16) S. 156, Beispiel (28)
IF. 114 ( μ 7 $% .
Korrektur ( μ i 7i $% .
276
Indogermanische Forschungen, 113. Band, 339–340 © Walter de Gruyter 2008 ISSN 0019-7262
DOI 10.1515/IDGF.2008.017
277
Zur formalen Gestaltung von IF.-Manuskripten Die Schriftleitung der IF. bittet um die Berücksichtigung einiger Formalien bei der Gestaltung der Manuskripte: 1. Das Manuskript sollte auf beiden Seiten einen ausreichenden Rand (mindestens 3,5 cm) aufweisen. Es sollte ferner ein eineinhalbfacher (bei Schreibmaschinenmanuskripten besser: zweifacher) Zeilenabstand gewählt werden. 2. Auszeichnungen der Schriftarten (sofern sie nicht in der Textverarbeitung verfügbar sind) sollten folgendermaßen vorgenommen werden: doppelte Unterstreichung = fett; Wellenlinie = kursiv; unterbrochene einfache Unterstreichung = gesperrt. 3. Für die Verwendung der unter 2. genannten Schriftarten gelten folgende Richtlinien, die besonders beachtet werden sollten: Kursivdruck wird nur für Beispiele verwendet (z. B.: lat. fenestra ‚Fenster‘). Eine Ausnahme bilden griechisch geschriebene Beispiele, die stets nicht kursiv erscheinen. Andere Alphabete (z. B. kyrillisch, hebräisch usw.) müssen wissenschaftlich transliteriert werden und erscheinen wiederum kursiv. In der Bibliographie ist dagegen Kursivsetzung zu vermeiden. Titel von Aufsätzen, Büchern, Sammelbänden oder Zeitschriften erscheinen niemals kursiv. Ebenso ist es nicht zulässig, die Namen der Verfasser durch Kapitälchen hervorzuheben. Jegliche Sperrung ist ebenfalls nach Möglichkeit zu vermeiden. Fettdruck im Text ist nur dann zulässig, wenn die Konventionen der Einzeldisziplin dies erfordern: z. B. bei oskischen Beispielen. 4. Die Herausgeber der IF. folgen einer Tradition der Zeitschrift, indem sie auf die konsequente Setzung von Abkürzungspunkten bestehen. Dies gilt sowohl für die Abkürzung von Sprachen (z. B. mhd., ai., OE., OIr. usw.) als auch für abgekürzte Zeitschriftentitel (z. B. PBB., IF. usw.). Wir bitten, diese Konvention zu beachten – auch wenn die Setzung von Abkürzungspunkten v. a. im englischsprachigen Raum immer mehr zurückgeht. 5. Bibliographische Angaben: Es wird dringend empfohlen, eine gesonderte Bibliographie (die mindestens die im Aufsatz erwähnte Literatur enthalten sollte) am Ende des Aufsatzes anzufügen. Im Text wird dann in der Regel in der folgenden Kurzform zitiert: „Schon Boley (1993: 200 f.) weist hierauf hin.“ Nur in besonderen Fällen (z. B. bei sehr wenigen zitierten Titeln [max. ca. 5], die nur jeweils einmal auftauchen) können die bibliographischen Angaben statt dessen in den Fußnoten erscheinen. Bei der Gestaltung der Bibliographie ist v. a. zu beachten, daß keine Kursivauszeichnung für Aufsatz-, Buch-, Sammelband- oder Zeitschriftentitel verwendet wird und daß die Verfassernamen keinesfalls in Kapitälchen hervor-
278
Zur formalen Gestaltung von IF.-Manuskripten
gehoben sein dürfen. Auch sollten Abkürzungen jeder Art (also auch die Abkürzungen von Zeitschriften, wie z. B. B. IF., PBB., KZ., IBS. usw.) grundsätzlich einen Abkürzungspunkt aufweisen. Bei der Anordnung der Informationseinheiten werden dagegen den Autoren keine strengen Vorschriften gemacht, so daß hier nur einige mögliche Anordnungen beispielhaft vorgeschlagen werden: Boley, Jacqueline, 1993, The Hittite Particle -z/-za. (Innsbrucker Beiträge zur Sprachwissenschaft, 79). Innsbruck. Ebenso zulässig ist aber auch: Boley, Jacqueline. The Hittite Particle -z/-za. Innsbruck 1993. (Innsbrucker Beiträge zur Sprachwissenschaft, Bd. 79). Aufsätze könnten z. B. nach folgendem Muster zitiert werden: Fischer, O., 1991, The Rise of the Passive Infinitive in English. In: Kastovsky, D., 1991, Historical English Syntax. Proceedings of the Kellner-Festival, Held in 1988 at Schloß Liechtenstein. (Topics in English Linguistics, 2). Berlin, New York, 141–188. 6. Abbildungen: Beigefügte Abbildungen müssen eine so gute Qualität aufweisen, daß sie direkt für den Druck übernommen werden können. Insbesondere ist darauf zu achten, daß die verwendeten Schrifttypen denen der IF. in etwa entsprechen (v. a. sind Schreibmaschinenschriften zu vermeiden). Die Größe der Abbildungen sollte den Seitenspiegel der Zeitschrift berücksichtigen: es kann sinnvoll sein, eine Abbildung in größerem Format einzureichen – die Elemente müssen dann aber so gestaltet sein, daß bei einer Verkleinerung eine gute Erkennbarkeit gewährleistet ist. 7. Bitte geben Sie unter dem Beitrag Ihre vollständige Adresse an. An diese Adresse werden Ihnen u. a. die Korrekturfahnen übersandt, so daß eine Erreichbarkeit gewährleistet sein sollte. 8. Falls der Text Ihres Beitrages in computerlesbarer Form vorliegt, übersenden Sie bitte zusammen mit Ihrem Manuskript eine Diskette mit dem gespeicherten Text und vermerken Sie auf dem Diskettenaufkleber neben dem Titel des Beitrages und dem Verfassernamen auch, mit welchem Programm und unter welchem Betriebssystem der Text verfaßt wurde. Fügen Sie bitte, falls es Ihre Textverarbeitung zuläßt, zusätzlich auch eine Version im RTF-Format bei. 9. Alle für die Indogermanischen Forschungen bestimmten Aufsätze und kleineren Beiträge senden Sie bitte an Apl. Prof. Dr. Eckhard Eggers, Nikolausberger Weg 63, D-37073 Göttingen, bzw. als .pdf-File per E-Mail an: [email protected] Die Herausgeber
Wo l f g a n g P. S c h m i d Eckhard Eggers