Reflexes Of Indo-European ‘Ē-Statives’ In Old Indic

23
REFLEXES OF INDO-EUROPEAN ‘ E-STATIVES’ IN OLD INDIC By ILYA YAKUBOVICH Moscow State University/Russian Academy of Sciences ABSTRACT The paper addresses the reconstruction of the Indo-European verbal derivational category that is conventionally called ‘ e-statives’ can be illustrated by Lat. sen- e-re ‘to be old’ or tac- e-re ‘to be silent’. Its empirical contribution consists in tracing the Old Indicreflexes of ‘ e-statives’, which have not been systematically investigated up to now. The analysis of the new data is conducive to the conclusion that that the Indo-European stative suffix *-eh 1 -i ̯ e/o- was originally typical of denominative verbs belonging to the Caland System. Its subsequent spread to primary verbs was implemented indifferent ways in individual ancient Indo-European languages, and therefore need not be projected into Proto-Indo-European. 1. INDO-EUROPEAN E-STATIVESA controversial problem of Indo-European linguistics is the form and function of a verbal derivational category conventionally called ‘ e-statives’. 1 There is a broad agreement among scholars that a number of Indo-European languages feature reflexes of a special suffix, which contained IE * e (resp. *eh 1 ), was normally attached to the zero grade of the verbal root or to the adjectival root/stem, and used for the derivation of verbal stems denoting uncontrollable states or achievements. But here similarities end and differences begin to pile up. The stative stems can pattern with adjectives (e.g. Lat. alb- e-re ‘to be white’ vs. albus ‘white’, sen- e-re ‘to be old’ vs. senex ‘old’), be ostensibly deverbative (either atelicizing, e.g. Lat. sed- e-re ‘to sit’ vs. sidere ‘to sit down’, or anticausative, e.g. cand- e-re ‘to shine, be bright’ vs. in-cendere ‘to burn’), or have no synchronic derivational base (e.g. Lat. tac- e-re ‘to be silent’, hab- e-re ‘to have’). In some languages the denominative ‘ e-statives’ primarily denote change of state. Such is the normal situation in Hittite, where we find mars-e-zzi, later mars-es-zi ‘becomes false’ vs. marsa- ‘false’ and tanatt-e-zzi, later tanatt-es-zi ‘becomes desolate’ vs. danatta-, danant- ‘empty, desolate’ (this class is identified in Watkins 1971). Accordingly, one should try to 1 The preliminary versions of this paper were presented at the University of Oxford (Spring 2011), University of Cambridge (Spring 2012), and Universitat Erlangen-Nurnberg (Summer 2012). I am grateful to Dr. Philomen Probert, Dr. James Clackson and Prof. Norbert Oettinger for the organization of the respective talks, and to their audiences for constructive feedback. Prof. Jay Jasanoff, Dr. Leonid Kulikov, Dr. Martin Kummel, Dr. John Lowe, Prof. Alexander Lubotsky, and Dr. Elizabeth Tucker have kindly agreed to provide feedback on the more advanced versions of this paper. I am also grateful to Stephen Durnford (Brighton) for improving the style of this paper and to the two anonymous reviewers for their helpful comments. Needless to say, I am the only person responsible for all the remaining shortcomings. A distant antecedent of this paper is Yakubovich (1999), written in the first year of my graduate studies at the University of California in Berkeley. Although I arrived there at certain results that are reiterated below, neither the overall methodology, nor many of the conclusions of that conference paper are satisfactory from my current perspective. The present work is meant to serve as a full replacement of Yakubovich (1999), extending at the same time the scope of its investigation. Transactions of the Philological Society Volume 00 (2013) 1–23 DOI: 10.1111/1467-968X.12039 © The author 2013. Transactions of the Philological Society © The Philological Society 2013. Published by John Wiley & Sons, 9600 Garsington Road, Oxford OX4 2DQ and 350 Main Street, Malden, MA 02148, USA.

Transcript of Reflexes Of Indo-European ‘Ē-Statives’ In Old Indic

REFLEXES OF INDO-EUROPEAN ‘ �E-STATIVES’ IN OLD INDIC

By ILYA YAKUBOVICH

Moscow State University/Russian Academy of Sciences

ABSTRACT

The paper addresses the reconstruction of the Indo-European verbal derivationalcategory that is conventionally called ‘�e-statives’ can be illustrated by Lat. sen-�e-re ‘to beold’ or tac-�e-re ‘to be silent’. Its empirical contribution consists in tracing the OldIndicreflexes of ‘�e-statives’, which have not been systematically investigated up to now.The analysis of the new data is conducive to the conclusion that that the Indo-Europeanstative suffix *-eh1-ie/o- was originally typical of denominative verbs belonging to theCaland System. Its subsequent spread to primary verbs was implemented indifferentways in individual ancient Indo-European languages, and therefore need not beprojected into Proto-Indo-European.

1. INDO-EUROPEAN ‘�E-STATIVES’

A controversial problem of Indo-European linguistics is the form and function of a verbalderivational category conventionally called ‘�e-statives’.1 There is a broad agreement amongscholars that a number of Indo-European languages feature reflexes of a special suffix, whichcontained IE *�e (resp. *eh1), was normally attached to the zero grade of the verbal root or tothe adjectival root/stem, and used for the derivation of verbal stems denoting uncontrollablestates or achievements. But here similarities end and differences begin to pile up. The stativestems can pattern with adjectives (e.g. Lat. alb-�e-re ‘to be white’ vs. albus ‘white’, sen-�e-re ‘tobe old’ vs. senex ‘old’), be ostensibly deverbative (either atelicizing, e.g. Lat. sed-�e-re ‘to sit’ vs.sidere ‘to sit down’, or anticausative, e.g. cand-�e-re ‘to shine, be bright’ vs. in-cendere ‘toburn’), or have no synchronic derivational base (e.g. Lat. tac-�e-re ‘to be silent’, hab-�e-re ‘tohave’). In some languages the denominative ‘�e-statives’ primarily denote change of state. Suchis the normal situation in Hittite, where we find mars-e-zzi, later mars-es-zi ‘becomes false’ vs.marsa- ‘false’ and tanatt-e-zzi, later tanatt-es-zi ‘becomes desolate’ vs. danatta-, danant-‘empty, desolate’ (this class is identified in Watkins 1971). Accordingly, one should try to

1 The preliminary versions of this paper were presented at the University of Oxford (Spring 2011), University ofCambridge (Spring 2012), and Universit€at Erlangen-N€urnberg (Summer 2012). I am grateful to Dr. PhilomenProbert, Dr. James Clackson and Prof. Norbert Oettinger for the organization of the respective talks, and to theiraudiences for constructive feedback. Prof. Jay Jasanoff, Dr. Leonid Kulikov, Dr. Martin K€ummel, Dr. John Lowe,Prof. Alexander Lubotsky, and Dr. Elizabeth Tucker have kindly agreed to provide feedback on the more advancedversions of this paper. I am also grateful to Stephen Durnford (Brighton) for improving the style of this paper and tothe two anonymous reviewers for their helpful comments. Needless to say, I am the only person responsible for all theremaining shortcomings. A distant antecedent of this paper is Yakubovich (1999), written in the first year of mygraduate studies at the University of California in Berkeley. Although I arrived there at certain results that arereiterated below, neither the overall methodology, nor many of the conclusions of that conference paper aresatisfactory from my current perspective. The present work is meant to serve as a full replacement of Yakubovich(1999), extending at the same time the scope of its investigation.

Transactions of the Philological Society Volume 00 (2013) 1–23 DOI: 10.1111/1467-968X.12039

© The author 2013. Transactions of the Philological Society © The Philological Society 2013. Published by John Wiley & Sons, 9600Garsington Road, Oxford OX4 2DQ and 350 Main Street, Malden, MA 02148, USA.

distinguish between statives proper (with or without cognate transformative verbs),deadjectival essives, and deadjectival fientives.

Differences in the morphological associations of ‘�e-statives’ are even more substantial. InLatin, the �e-like suffix can be reconstructed for the present stem (hab-e-�o ‘I have’, hab-�e-mus‘we have’), while in Attic Greek it is rather associated with the aorist stem (em�an-�e-n ‘Iwent mad’ vs. ma�ınomai ‘I am mad’). In Balto-Slavic one observes a clear differencebetween the grammatical behaviour of essives/fientives and statives proper. The deadjectivalessives/fientives show the suffix under discussion both in the present and infinitive stems,e.g. Old Church Slavic 1SG.PRES star-�e-jǫ, INF. star-�e-ti ‘to grow old’or Lithuanian 1SG.PRES

stor-ė-ju, INF. stor-ė-ti ‘to become fat’. By contrast, the statives proper display thealternation between the same type of infinitive stem and a present stem that apparentlydoes not contain reflexes of *�e, e.g. OCS. 1SG.PRES b�u�zdǫ, INF. b�ud-�e-ti ‘to be awake’, Lith.1SG.PRES bud�zi�u, INF. bud-ė-ti ‘to be awake’. The only incontestable Germanic reflexes of thestative suffix are attested in rare nominal derivatives, such as Gothic fah-e-þs ‘joy’.

This brief survey does not, of course, exhaust all the forms that are subsumed inscholarly literature under the label of ‘�e-statives’, but merely underscores the pitfalls ofreconstructing the function and combinatory properties of the stative suffix. A much moredetailed survey of the primary Indo-European data that are traditionally treated under thislabel can be found in Jasanoff (2002/2003: 127–129). I summarize his presentation inTable 1, adding to it Greek essive and fientive forms that were discussed in some detail inTucker (1990: 27–72). The forms in square brackets do not contain prototypical ‘�e-statives’but are assumed to contain cognates of the stative suffix in some scholarly accounts, as wewill see below.

The structural and functional diversity of the Indo-European ‘�e-statives’ represents achallenge for comparative linguists. If one assumes (as most people do) that all ‘�e-statives’ultimately derive from a single inherited category, it is reasonable to ask what thatcategory’s function was. A factor that complicates the problem is the existence of a different‘stative’ in Indo-European reconstruction. Oettinger (1976) cogently argued that recon-structed verb forms, which were now recognized as belonging to the middle voice of the ‘h2e-conjugation’ (e.g. 3SG. *kei-o(r) ‘lies’, *dhugh-o(r) ‘is available’) does not display the

Table 1. ‘�e-statives’ in individual Indo-European languages

Language Statives proper Deadjectival Essives/Fientives

Latin hab�emus ‘we have’ (pres.)tac�emus ‘we are silent’ (pres.)

rub�emus ‘we are red’ (pres.)sen�emus ‘we are old’ (pres.)

Hittite Not attested marsezzi ‘becomes false’ (pres.)tanattezzi ‘grows desolate’ (pres.)

Greek em�an�en ‘I went mad’ (aor.)e(w)�ag�en ‘I broke (intr.)’ (aor.)

thars�e�o, (e)th�ars�e-sa ‘to get courage’anth�e�o, �e

nth�esa ‘to be in bloom’

(pres./aor.)Old ChurchSlavic

[b�u�zdǫ], b�ud�eti ‘to be awake’[mĭnjǫ], mĭn�eti ‘to think’([pres.]/inf.)

star�ejǫ, star�eti ‘to grow old’bogat�ejǫ, bogat�eti ‘to become rich’(pres./inf.)

Lithuanian [bud�zi�u], bud _eti ‘to be awake’[mini�u], min _eti ‘to mention’([pres.]/inf.)

sen _e ju, sen _eti ‘to grow old’stor _eju, stor _eti ‘to become fat’(pres./inf.).

Old Irish Not attested? ruidi(d) ‘blushes’ (pres.)

Gothic [habaiþ] ‘has’ (pres.)Cf. fah-e-þs ‘joy’

[fastaiþ] ‘fasts’ (pres.)Cf. armaio ‘mercy’

2 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

reflexive semantics typical of the Indo-European middle but rather denotes states.2 Inreconstructable Proto-Indo-European there must have been some sort of originaldistribution between the ‘�e-statives’ and what I will call for the sake of convenience‘Oettinger statives’.3

2. EXISTING RECONSTRUCTIONS

In current Indo-European scholarship there are three main ways of approaching thereconstruction of ‘�e-statives’. The oldest one is the scenario outlined in Cowgill (1963) and stilldefended in Ringe (1996, 2006). It appeared before the discovery of ‘Oettinger statives’ anddoes not address the issue of functional difference between the two categories. Cowgillbelieved that the Indo-European statives originally belonged to the present stem and werecharacterized by the apophonic (‘ablauting’) suffix *-eh1-/-h1-. The main line of developmentpostulated for the statives is their thematization with the help of the suffix -ie/o-. Dependingon whether this secondary suffix was added to the full or the zero grade of the stative marker,one would get * -�eie/o- (as in Latin rubeo, rub�emus) or *-h1ie/o- (as in Germanic ‘Class III’weak verbs, e.g. Gothic habam ‘I have’, habaiþ ‘has’). After Pinault’s Law, which assumes theloss of ‘laryngeals’ in a position between consonant and *i already in PIE, gained popularityin Indo-European studies, this account has been modified. According to Ringe (2006: 132with references to the earlier literature), the suffix *-ǝie/o- reconstructed in habaiþ and similarGermanic forms as well as Tocharian Class III/IV presents is not a direct reflex of *-h1ie/o-but rather a result of proportional analogy involving the -to-participle: ptc. -to/prs. *-ie/o- =ptc. *-ǝto/X? X = *-ǝie/o-. The only language that could avoid the thematization of stativeswas Greek, but only because the Greek ‘�e-statives’ somehow shifted early enough from thepresent to the aorist system.4

Another approach is explicated in Harðarson (1998) and adopted in Rix (1998, 2001).According to this point of view, there is a contrast between the ‘Oettinger statives’, whichoriginally belonged to the present system, and the ‘�e-statives’ whose original locus was theaorist system and which were characterized by the suffix *-eh1- (with a zero-grade variant*-h1- postulated through internal reconstruction). Rix 2001, following Harðarson 1998: 327,reserves the term ‘statives’ for the first category, but defines forms in *-eh1- as ‘fientives’. Itis easy to see that the Greek situation emerges as the most archaic one under this analysis.Harðarson and Rix further assume that that primary ‘�e-aorists’, or ‘fientives’ serve as abase for the formation of secondary ‘essives’, which are reconstructed as present stemsprovided with the composite suffix *-h1-ie/o-. In Old Church Slavic, for example, theinfinitive b�ud�e-ti ‘to be awake’ is analysed as a derivative of the ‘fientive’ *bhudh-eh1-, whilethe present b�u�zdǫ ‘I am awake’ is taken as a reflex of the ‘essive’ *bhudh-h1ie/o-. Theformations that can be described as tertiary are ‘fientives’ with the suffix *-eh1-ie/o-

2 The concept of ‘h2e-conjugation’ is elaborated upon in Jasanoff (2003). The key assumption of his theory, whichis not universally accepted but widely followed, is the existence of an Indo-European contrast similar to that betweenmi- and ḫi-conjugations in Hittite. Although the Indo-European medio-passive is assumed to be originally alignedwith the ‘h2e-conjugation’, some mediopassive forms must have penetrated into the mi-conjugation before theseparation of Anatolian from Core Indo-European.

3 An alternative name for this category, occurring in American literature, in particular in the works of Jasanoff, is‘stative-intransitive root presents’. It reflects the fact that the ‘�e-statives’ include occasional transitive forms (e.g. Lat.hab�ere ‘to have’), but no such forms are attested among the ‘Oettinger statives’. More rarely one encounters stative-intransative root aorists with the same set of endings, which can also be subsumed under the label of the ‘Oettingerstatives’.

4 The views of Cowgill and Ringe are partially misrepresented in Jasanoff (2002/2003: 129), where it is claimed that‘[a]ccording to the ‘Cowgill school,’ PIE *-eh1- with the zero grade *-h1- was an aorist suffix’. This erroneous claimobfuscates the crucial difference between the scenarios of Cowgill-Ringe and Harðarson-Rix.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 3

displaying the levelling of ablaut grade to that of the ‘essives’. These formations are typicalof deadjectival verbs, and therefore are not assumed to belong to the core of the Indo-European verbal system.

A third scenario of the development of ‘�e-statives’ was proposed in Jasanoff (1978) andunderwent further elaboration in Jasanoff (2002/2003). Jasanoff claims that the original locusof statives were denominative essives in *-eh1-ie/o- > *�e-ie/o-, which did not participate inapophonic alternations, because they were derived from the nominal instrumental suffix *-eh1.Once the ‘�e-statives’ were extended to primary verbs, they came into competition with theinherited ‘Oettinger statives’ (which Jasanoff prefers to call ‘stative-intransitive rootpresents’). The outcome of this competition varied by language family. In Latin the‘�e-statives’ completely won the field, while in Balto-Slavic and Germanic the ‘Oettingerstatives’ were supposedly retained in primary present stems. Peculiar to Jasanoff’s theory is hisclaim that the Baltic i- presents (e.g. bud�zi�u ‘I am awake’, b�udi ‘he is awake’), Slavic �ı-presents(e.g. b�u�zdǫ ‘I am awake’, b�udit�u ‘he is awake’), and Germanic ai/a-presents in (e.g. Goth. haba‘I have’, fastaiþ ‘fasts’) all represent the remodelled ‘Oettinger statives’. Nonetheless, thetraces of encroachment by ‘�e-statives’ upon primary verbs are found in the NorthwesternIndo-European dialectal area, for example in isolated nouns in Germanic, or Balto-Slavicinfinitive stems. The interaction between essives/fientives and statives proper had mostunexpected results in Greek, where the suffix -eh1- penetrated in the aorist system of primaryverbs (em�an-�e-n ‘I went mad’) via the presumed mediation of non-finite forms, while theoriginal denominatives retained the ‘stative’ suffix in the inflected forms of the sigmatic aorist(�enth-�e-sa ‘I went into bloom’).

Terminological differences complicate comparison between the three theories outlinedabove. Nonetheless, after one turns to the primary data, the relative advantages anddisadvantages of the three theories begin to emerge. Cowgill’s hypothesis has, in my opinion,the least to recommend itself in the light of our present knowledge. None of the three Indo-European families attested in the second millennium BC, namely Anatolian, Indo-Iranian, andGreek, displays regular ‘�e-statives’ functioning as presents derived from primary verbs. Thefirst two of these three families display the match between ‘Oettinger statives’ in the samefunction, while Greek does not contradict this reconstruction. For example, Luwian ziyari‘lies’ is morphologically comparable with Vedic �s�aye ‘lies’ on the assumption of IE *kei-o(r),while Hittite kittari and Greek ke~ıtai ‘lies’ bear witness to its expected restructuring to *kei-to(r) under pressure from the more productive class (cf. Rix 2001: 320).5 To this one can addthat the mechanism responsible for the shift of ‘�e-statives’ from the present to the aoristsystem within Greek remains without clarification within the paradigm of Cowgill-Ringe.

The approach of Harðarson and Rix not only keeps apart the ‘�e-statives’ from ‘Oettingerstatives’ but also has the advantage of simplifying the reconstruction of Greek data, where the�e-suffix is synchronically associated with the aorist system of primary verbs. Since the Slavicaorist is derived from the infinitive stem, one could also attempt to argue that the Slavic‘�e-statives’ were at some point primarily associated with the aorist. But Latin presents theopposite picture. There the element �e extended its scope in the imperfect (forms like mitt-�e-bat‘he was throwing’ and veni-�e-bat ‘he was coming’ are built on the model of rud-�e-bat ‘he wasred’), but did not penetrate into the perfect system, which took over the function of theoriginal aorist. The stative derivatives deriving atelic verbs from their telic counterparts, likeLat. sed�ere ‘to sit’, likewise create a problem for Harðarson and Rix’s analysis. In Slavic, too,the element �e plays an important role in the formation of the secondary imperfect, and

5 Due to the scarce attestation of Mycenaean Greek verbal morphology, one has to extrapolate the data attested inthe first millennium BC. It is, of course, theoretically possible that a true ‘Oettinger stative’ *ke~ıoi ‘lies’ existed in thesecond millennium Greek.

4 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

this situation can be probably projected into Balto-Slavic (cf. Lith. ved-ė ‘he led’ vs. OCS.ved-�e-a�se ‘he was leading’ and see further Stang 1966: 387–388).6 Therefore one can say at thevery least that the aspectual distribution of the �e-suffix across Indo-European offerscontradictory evidence. This was, in fact, aptly noted in Garc�ıa-Trabazo (2009), whogenerally followed the Harðarson/Rix framework but asserted that the original meaning of*-(e)h1- neutralized the opposition between ‘essives’ and ‘fientives’. Yet once one makes thiscrucial step, there remains little reason, in my opinion, to follow the formal reconstructionadvanced by Harðarson and Rix.7

Not all the aspects of Jasanoff’s scenario are equally compelling, and not all of them areamenable to empirical testing to an equal extent. Thus, the instrumental origin of the stativesuffix *-eh1-ie/o- is not provable by the comparative method and belongs strictly to thedomain of internal reconstruction. The claim that presents of ‘stative’ verbs in Germanic,Baltic and Slavic developed directly from what is called here ‘Oettinger statives’ is based onrather complicated sequences of analogical changes, the intermediary stages of which are notcorroborated through independent evidence.8 Although Jasanoff is able to adduce a numberof lexical etymologies in support of his scenario, the late attestation of Germanic, Slavic, andBaltic data leaves enough time for ‘Oettinger statives’ to be replaced with completely differentformal models, which continued to be in a competition with ‘�e-statives’.9 Finally, one mayquestion the pervasive presence of the *-ie/o- extension, as opposed to the simple presents in*-eh1- in some ancestor languages (as did Jasanoff in his 1978 book).

But in comparison with the previous two theories, Jasanoff’s account provides a far betteroverall fit to the data summarized in Table 1. It is easy to see that there are no languageswhere the scope of ‘�e-statives’ is restricted to statives proper, but Hittite and perhaps Celticprovide instances of them being aligned to deadjectival essives/fientives. Since Anatolian

6 The comparison between the Baltic preterit and Slavic imperfect advocated by Stang is not universally accepted.L€uhr (1999) represents a recent attempt to explain the Slavic imperfect an original periphrastic construction involvingthe instrumental, e.g. *vo�na a- ‘was full of smell’ > vo�naa- ‘was smelling (intr.)’. This explanation curiously resemblesJasanoff’s account of the Indo-European ‘�e-statives’. I fail to see how it accounts for the similarity between secondary�e in OCS. ved-�e-a�se vs. Lith. ved-ė and similar forms, but in Slavic, at least, the spread of suffixal -�e- to the imperfect ofthematic stem is presumed to be due to the influence of the ‘�e-statives’ (cf. L€uhr 1999: 179). A specific problem thatconcerns Baltic is the synchronic alignment of the ‘�e-preterit’ with transitive stems. I believe that this correlation isepiphenomenal and reflects the secondary productivity of the etymological Balto-Slavic ‘�a-preterit’ with intransitivestems in Baltic, as a result of which the etymological ‘�e-preterits’ were ousted from this domain or acquired thesecondary suffixes of the ‘�a-preterit’. For the archaic reflexes of the ‘�a-imperfect’ in Baltic with transitive verbs, seeStang (1966: 384–385).

7 The specific claim of Harðarson (1998) about the existence of a robust class of -ai-/-ja- verbs in Proto-Germanic,which is also found in Kortlandt (1990) and Ringe (2006), has become a subject of critical examination in a recentOxford M.Phil. thesis (Goering 2012). Goering convincingly demonstrates that the only productive suffix of Proto-Germanic statives is *-ai-/-a-.

8 Cf. Jasanoff (2002/2003: 153) for Baltic and Jasanoff (2002/2003: 156–157) for Germanic. In the case of Slavic,Jasanoff acknowledges that the Slavic stative present in *-�ı- of the type mĭnjǫ, mĭnitъ owes its suffix to a transfer(‘influence’) from the causative conjugation.

9 For example, the suffix *-ai-/-a- of Germanic Class III weak verbs can be derived by sound law from the (virtual)Indo-European *-oie/o-. Granted, the existence of such a suffix in Proto-Indo-European is highly doubtful, but itcould arise at any time in the prehistory of Germanic as a new marker of denominative verbs formed from thematicstems (*-o-ie/o-). A small class of transitive –ai-/-a- verbs in Gothic, described in Dishington (1976), might represent avestige of the earlier state of affairs, when Germanic *-ai-/-a- < *-oie/o- was not yet associated with the statives anddenominative essives/fientives. While there is no obvious functional reason why the new Germanic denominativesshould have become productive in intransitive meaning, dislodging the inherited ‘�e-statives’, such a grammatical-ization would be typologically trivial. As an example of the opposite but structurally similar development, one canpoint to the secondary *-oie/o- denominatives in Greek, which were grammaticalized as transitive factitives in -o�o(type neό�o ‘I make new’). The whole account of this footnote is admittedly speculative, but I fail to see how it isinferior to the radical scenario of Jasanoff (2002/2003), where the Germanic Class III weak verbs are derived directlyfrom the ‘Oettinger statives’. I owe the main details of this suggestion to the personal communication of Prof. AndreasWilli (Oxford). Cf. also an alternative scenario of Dishington (2010), which nonetheless also implies themorphological renewal of Germanic statives, as opposed to the transformation of an inherited stative morpheme.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 5

represents an outlier within the Indo-European family, one can regard the Hittite distributionas an archaism and the pattern of other languages as a common innovation between them.Furthermore, under Jasanoff’s account one is not required to ignore Pinault’s Law (as doHarðarson and Rix), nor to invent ad hoc proportional analogies aimed at obtaining thesequence *-ǝye/o- of secondary origin (as does Ringe). Those forms that do not look like‘�e-statives’, and therefore appear in square brackets in Table 1, are simply not reconstructedas such. This solution comes at a very low cost, because none of the rejected present stemsexcept for the Germanic fastaiþ type belonged to the original semantic domain of ‘�e-statives’,as defined by Jasanoff.

Jasanoff’s reconstruction is admittedly not conducive to a straightforward account of theGreek passive aorists (type em�an�en ‘I became mad’) as an isolated phenomenon, but actuallyemerges as a preferable alternative if one takes into consideration additional evidence fromGreek. With Ruijgh (2004), Greek passive aorists in -�e- should be considered together withdialectal Greek presents in -�e-, such as Aeolic thers�emi < thars�emi ‘I am bold’ or Arcadianapeith�emi < *apith�emi ‘I am disobedient’.10 I would like to propose that the Greek deadjectivalstatives had at some point the suffix -�e- (the mi-conjugation of such forms may or may not besecondary). Like all denominative verbs, they did not exhibit an opposition between thepresent and aorist stems, and in terms of Aktionsart, they could function as both essives andfientives (e.g. 1SG.IMPF *�ap�ıth�en ‘I was disobedient’ and ‘I became disobedient’). Later thepreterit in -�e- began to spread to primary stative verbs, where it was reinterpreted as an aoristbased on its telic, fientive reading and in view of the mismatch with the present stem (aor.em�an�en ‘I became mad’ vs. pres. ma�ınomai ‘I am mad’). By contrast, the original deadjectivalpreterits in -�e- could only be interpreted as imperfects because of their association with thematching present stems (impf. *�ap�ıth�en vs. pres. *apith�emi). Consequently, when aspectualoppositions were extended to denominative verbs, this resulted in the creation of the newaorist *�apith�esa ‘I became disobedient’, while the imperfect *�ap�ıth�en was confined to the non-telic, essive reading ‘I was disobedient’. The basic intuition behind this scenario is that it iseasier to explain how ‘�e-statives’ evolved into both present and aorist stems if one reconstructstheir original locus outside the system of primary verbs. For additional considerationspertaining to the formal development of ‘�e-statives’ in Greek, see section 7 with Table 6.

Whether or not the bulk of Indo-European scholars accepted the scenario of Harðarson/Rix, it probably remains better known to the students of Indo-European than Jasanoff’s viewson the origin of ‘�e-statives’. The reasons for this situation may be partly sociological andinclude such factors as the special status of the Lexicon der indogermanischen Verben (Rix2001) and the Indo-Europeanists’ overall better familiarity with Greek than Hittite. In anycase, it seems unlikely that the participants of the debate will modify their views unlessprompted to do so by the emergence of new data.

3. PREVIOUS RESEARCH ON INDIC STATIVES

The new empirical material that is conducive to further testing of competing hypotheses aboutthe origin of ‘�e-statives’ comes from the Indo-Iranian family, in particular from Old Indic.11

10 The dialectal Greek present stems in -�e- need be reconciled with the reconstruction of thematic presents in*-eh1- ie/o- > *-�eie/o-, which is supported by Balto-Slavic and also, as we shall see, by Indo-Iranian data. The Italo-Celtic and Hittite data appear to be ambiguous in this respect. For an attempt to derive the dialectal Greek -�e- stemsfrom the earlier stems in *-�eie/o- and *-eie/o- via analogical restructuring see Rau (2013). This issue would meritadditional investigation but is beyond the scope of this paper.

11 The contribution of the Iranian data to the problem under discussion is not significant, although some relevantforms are listed in section 6. The Avestan denominative forms collected and discussed in Tucker (2004) do not appearto include synchronic or etymological statives, which could be compared with the Vedic and Sanskrit forms treated inthis paper.

6 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

The Vedic and Sanskrit languages display, unlike Greek, a clear contrast between primaryand derived verbs. The latter group, united into the residual Class 10 of traditional Indicgrammar, has not been fully integrated into the aspectual system and its conjugation wasessentially limited to the derivatives of the present stem. Therefore one can hardly imagine ananalogy that would cause a denominative present suffix to spread to the primary aorist inVedic, or vice versa. Thus one can use the Old Indic data for discriminating between thetheories of Jasanoff and Harðarson/Rix without running into the same sorts of problemsthat one faces tackling the Greek data. On the formal side, Vedic (and Indo-Iranian ingeneral) clearly distinguishes between the reflexes of *-eh1ie/o- > *�aya-, *-eh1- >*-�a- and *-h1ie/o- > -ya-.12 Last but not least, Early Vedic poetry comes from the second millennium BC andthus, alongside Anatolian and Mycenaean Greek, belongs to the earliest known Indo-European texts. Unfortunately, Mycenaean Greek does not yield much information about theform and function of ‘�e-statives’, but a substantial agreement between Vedic and Hittitewould almost secure their Indo-European origin.

At present, only the formal reflexes of ‘Oettinger statives’ are securely established for OldIndic. As mentioned above, the Indo-Iranian and Anatolian families provide primaryevidence for reconstructing the whole category. The Indo-Iranian reflexes of ‘Oettingerstatives’ have undergone a detailed analysis in K€ummel (1996). It is observed that they can beformed both from intransitive and transitive forms, functioning in the second case as agentlesspassives. The distinct stative endings are limited to the third person singular and plural (Ved.3SG.PRES -e, 3PL.PRES -(i)re; Av. 3SG.PRES -e, 3PL.PRES -(�ai)re). In Vedic, besides the root stems,there is a significant group of stative forms formed from nasal presents, e.g. �srnv�e ‘is heard’from �sru ‘to hear’, but Avestan does not share this trait (cf. sruii�e ‘is famed’ from sru ‘to hear’)and it probably represents an Indic innovation. According to Jasanoff (2003: 155), ‘Oettingerstatives’ belong to the larger ‘stative-intransitive system’, which includes, among other forms,Vedic Class IV presents in -ya- with stative/fientive meaning (e.g. b�udhyate ‘he wakes up’).13

By contrast, the opinions on the Indo-Iranian reflexes of Indo-European ‘�e-statives’ differwidely. For Cowgill (1963: 265), ‘Indo-Iranian lacks a stative formation’. According toWatkins (1971: 88), the absence of statives characterizes a whole subgroup of the Indo-European languages, comprising Indo-Iranian, Greek and Armenian (Watkins does notconsider the Greek �e-aorists as members of the same category).

On the other hand, an elaborate hypothesis about the reflection of Indo-European stativesin Indo-Iranian is found in Harðarson (1998), who follows here Rasmussen (1993).Developing his reconstruction of stative presents in *-h1ie/o-, Harðarson suggests that it ishidden in those inherently telic verbs that display stative presents with the Class IV -ya- suffix(e.g. kr�udhyati ‘he is angry’ from the root krudh, supposedly ‘to become angry’). Hedistinguishes such cases from the fientive presents in -ya- which, in his opinion, contain theinherited *-ye/o- suffix (e.g. b�udhyate ‘he wakes up’ from the root budh ‘to wake up’).Additional reflexes of Indo-European statives are seen in Vedic passive presents in -y�a-, whichare widely regarded as the development of Class IV with accent shift. On the formal level,Harðarson strives to find support for his hypothesis in the Vedic passive stem t�ay�a- from theroot tan ‘to stretch’, which he reconstructs as *tnh1-ie- (*tn-ie- would have given tany�a- by

12 In the following exposition, I adopt the ‘laryngealist’ reconstruction of the stative suffix, which is used in themajority of recent works on the subject. I must stress, however, that the only internal reason for this decision is thepossible dissimilation *CHAH > *CAH discussed in section 6 below. If one should regard the evidence of gl-�aya-(ti)‘to be weary, exhausted’, py-�aya-(te) ‘to swell’, and �sv-�aya-(ti) ‘to swell, prosper’ as not probative, then thereconstruction *-�eie/o- can be substituted for *-eh1ie/o-.

13 Bruno (2005: 45–60) argues against the synchronic status of ‘Oettinger statives’ as a separate functional categoryin Old Indic. In her opinion, they should be regarded as a subset of mediopassive verbs. For the passive meaning ofsome ‘Oettinger statives’ see also Kulikov (2012: 745).

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 7

sound laws). He also observes that some Old Iranian passives in -ya- display active endingsand proposes to see in them reflexes of historical statives, presumably alluding to the fact thatthe Greek passive aorist of stative origin likewise has active endings.

Of Harðarson’s arguments cited above, only the stem t�ay�a- can be regarded as a genuineproblem for his opponents. The purported influence of *s�ay�a- the passive stem of Ved. sani-‘to win, gain’ < IE. *senh2-, which is advocated in Jasanoff (2002/2003: 142), is hard to justify,because the stem s�aya- is attested only in the works of grammarians, although one can notethat it occurs alongside the irregular sanya-. According to Kulikov (2012: 96–97), thesecondary quasi-root t�a- might have come into being due to the reanalysis of the action nounutt�an�a- interpreted as an adjective in -n�a-. But in any case we are dealing here with agrammatical irregularity, and the stem t�ay�a- could indeed be regarded as a possible isolatedvestige of Indo-European statives in Vedic if no other candidates were available.14

The other two arguments endeavour to explain phenomena that do not really requireelaborate diachronic explanations. It is typologically common that one and the sameformation can be used both with the stative and fientive meaning depending on its lexicalinstantiation and/or context. To take an example from among the assured ‘�e-statives’,Russian belejet ‘is white’ and zelenejet ‘is green’ are formed in the same fashion as krasnejet‘becomes red’ and zheltejet ‘becomes yellow’. To be more precise, the meanings cited for theRussian denominative verbs are prototypical, and all four of them can in principle occur asboth essives and fientives. But this conclusion could hardly have been made based on a limitedcorpus study. In a similar fashion, the fact that individual Vedic verbs in -ya- could refereither to the state or to its transformation does not constitute a sufficient reason forpostulating an historical merger of two distinct morphological types. The same is true for thevariation of active and medio-passive endings with -ya- passives in Iranian. If the -ya- passivesrepresent an extension of the Indo-Iranian primary verbs in -ya- (Class IV of Sanskritgrammarians), it is only natural that they could display both type of endings, since this wasthe case with Indo-Iranian primary verbs in -ya-. It is the generalization of middle voice withOld Indic passives that should be regarded as an innovation.

Additional arguments against Harðarson’s solution can be found in Jasanoff (2002/2003).But the only Vedic formation that Jasanoff himself treats as a likely reflex of the Indo-European stative in *-eh1-ie/o- is ‘the isolated Vedic participle san�ay�at ‘old’’.15 If one weighsup the Rigveda hapax san�ay�at (RV. 1.62.13) against the Vedic hapax t�ay�amana- ‘stretched, inprogression’ (RV 10.17.7.) it might appear that Jasanoff’s and Harðarson’s reconstructions ofIndo-European statives are balanced up against each other on the strength of Indo-Iranianevidence, and both arguments are too flimsy to contribute to the debate in a meaningfulfashion. Fortunately for Jasanoff’s argument, there are additional Vedic and Sanskrit formsthat yield support to his reconstruction. In the rest of this paper I will focus on the analysis ofsuch formations.

4. ESSIVES/FIENTIVES

It seems to be appropriate to begin with an old observation about the grammar of ClassicalSanskrit, which, to my knowledge, has failed thus far to attract the attention of Indo-Europeanists. ‘In the later classical language, most old denominatives made from consonantstems disappear. The thematic type remains living and takes two forms: (1) in the active, the

14 The grammatical irregularity of the stem t�ay�a- led to its occasional reanalysis as a derivative of the root t�aexisting alongside tan. This is presumably the origin of Vedic hapaxes 3sg.pf.med. tate (RV 1.83.5) and pass. -i-aoristpr�at�ayi attested in the Aitareya-Ar�aṇyaka (cf. Kulikov 2011: 314, fn. 19).

15 Here Jasanoff follows Watkins (1971: 88), where an additional participial form Ved. rt�ay�ant- is introducedwithout comments.

8 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

normal -ayati is used, kalus: ayati ‘makes turbid’, taruṇayati ‘rejuvenates’; (2) in the middle,with intransitive sense, -�ayate is used kalus: �ayate ‘becomes turbid’, taruṇ�ayate ‘is rejuvenated’’(Burrow 1973: 364). Both the active transitives in -aya- and the mediopassive intransitives in-�aya- form open classes in classical Sanskrit, but the clearest cases of contrastive pairs derivedfrom the same adjective are listed in Table 2.16 In addition, Sanskrit features a somewhatsmaller but also productive class of active -aya- factitives matching mediopassive -�aya-essives/fientives formed from substantives. The available minimal pairs are collected inTable 3.

The distribution of suffixes outlined above corresponds precisely to one of the optionsreconstructed for Proto-Indo-European. The Pan-Indo-European intransitive essives in*-eh1ie/o- (cf. Table 1) contrast with the dialectal transitive factitives in *-eie/o-, which can bereconstructed at least for Germanic and Slavic.17 If Classical Sanskrit were the earliestattested stage of Old Indic, the suffix -�aya- would have been surely accepted long ago as astraightforward reflex of Indo-European ‘�e-statives’.

The trouble is that the existing Vedic evidence is not as abundant or straightforward as theClassical Sanskrit data. In the Vedic language, the distribution between -�aya- and -aya-denominatives is more complicated and governed not only primarily by semantic but also by

Table 2. Contrasive pairs of deadjectival verbs in Sanskrit

Transitives in -aya- Intransitives in -�aya- Base adjectives

utkaya- ‘to make long for’ utk�aya- ‘to long for’ utka- ‘longing for’unmanaya- ‘to excite’ unman�aya- ‘to get excited’ unmanas- ‘excited’kalus: aya- ‘to pollute’ kalus: �aya- ‘to become turbid or unclean’ kalus: a- ‘turbid, dirty’capalaya- ‘to cause anyone to actinconsiderately’

capal�aya- ‘to move to and fro’ capala- ‘moving to and fro’

jaḍaya ‘to make without feeling,render weak’

jaḍ�aya ‘to be stiff’ jaḍa- ‘cold, stiff’

taralaya- ‘to make tremble’ taral�aya- ‘to tremble’ tarala- ‘trembling’taruṇaya- ‘to make young’ taruṇ�aya- ‘to become young’ taruṇa- ‘young’timiraya- ‘to obscure’ timir�aya- ‘to appear dark’ timira- ‘dark’dh�umaya- ‘to cover with smoke’ dh�um�aya- ‘to smoke’ dh�uma- ‘smoke’pratyaks: aya- ‘to make visible’ pratyaks: �aya- ‘to come clearly before the

eyes’pratyaks: a- ‘visible’

mandaya- ‘to weaken, lessen’ mand�aya- ‘to go slowly, linger, loiter’ manda- ‘slow, weak’vi�sadaya- ‘to purify, to make clear’ vi�sad�aya- ‘to become clear’ vi�sada- ‘clear, pure’�sithilaya- ‘to make loose, relax’ �sithil�aya- ‘to become relaxed’ �sithila- ‘loose, relaxed’�si�siraya- ‘to cool (trans.)’ �si�sir�aya- ‘to become cool’ �si�sira- ‘cool, cold’saralaya- ‘to make straight’ saral�aya- ‘to advance steadily’ sarala- ‘running on, straight’sthiraya- ‘to strengthen’ sthir�aya- ‘to become fixed’ sthira- ‘firm, fixed’

16 The forms in Tables 2 and 3 are all taken from Monier-Williams (1899). I would like to make clear that theproposed distribution reflects a tendency rather than strict rule. For example, there are instances where the derivativesin -�aya- are attested as factitives (e.g. abhr�aya- ‘to make clouds’ from abhra- ‘cloud’), or have both transitive andintransitive meanings (e.g. amrt�aya- ‘to be like nectar’ and ‘to turn into nectar’). The derivational character of the-aya- and -�aya- suffixes explains why only one member of the contrastive pair is attested with most roots. Even in thecases collected in Tables 2 and 3 the denominatives belonging to contrastive pairs are usually attested in differenttexts. But as a statistical generalization, the distribution outlined in Tables 2 and 3 is quite robust: I have failed to finda single example that could illustrate the opposite contrast between the -aya- essives/fientives and -�aya- factitives.

17 Cf. e.g. Goth. ga-qiujan ‘to make alive’ vs. qius ‘alive’ and OCS. o-�ziviti ‘to make alive’ vs. �zivъ ‘alive’ for whichone can reconstruct dialectal IE *gw�ıweye- ‘to make alive’ from *gw�ıwo- ‘alive’. It appears that the main model ofcreating deadjectival factitives in Early Indo-European involved a different suffix -ah2-(ie/o)-, e.g. *new-ah2-(ie/o)- ‘torenew’ from *newo- ‘new’ cf. Hitt. newaḫḫ- ‘to renew’, Lat. nov�are ‘to renew’ and Greek ne�a�o preserved in thetechnical meaning ‘I replough’. Nevertheless, the suffix *-eie/o- should be reconstructed as a common Indo-Europeanmarker of denominative transitive verbs, which further developed into causatives (e.g. *sodeie/o- ‘to bring into asitting position’ > ‘to make sit’). The development from denominative transitives to deadjectival factitives appears tobe trivial and could be independently implemented in several areas.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 9

phonetic factors. In a nutshell, its phonetic component can be described as follows: thevariant -aya- is more likely to be used if the syllable before or after the one with -�a- is long,otherwise the variant -�aya- is preferred. For example, the stem devaya- ‘to worship the gods’does not have the variant dev�aya- in the Rigveda on account of the long first syllable /de:/,18

but rt�aya- ‘to act according to the truth’ displays a contrast between 3PL.IMPF.MED rtayanta(the penultimate syllable /jan/ is long) and the participial forms rt�ayat�e, rt�ayat�as (the syllablesbefore and after /ta:/ are short). The phonetic distribution according to these guidelines is notstrict, since archaic forms without metrical restructuring are available, and a number ofanalogical changes led to the levelling of individual paradigms in favour of either -aya- or-�aya-. The most detailed synchronic outline of the Vedic facts is presented in Insler (1997).

Two different suggestions have been made with regard to the diachronic explanation of theVedic distribution between -�aya- and -aya- denominatives. According to Insler (1997) we aredealing with a rhythmic lengthening *-aya- > -�aya- aimed at avoiding the sequence of threeshort syllables (e.g. vr-s: a-y�a-te > vr-s: �a-y�a-te). Insler finds a parallel to this phenomenon in thebehaviour of Vedic adjectives endowed with the possessive suffix -vant-. In this formation,according to Insler, the stem vowel of the base noun tends to be long between two shortsyllables, otherwise it tends to be short, e.g. deva-vatas ‘related to gods (GEN.SG)’, �sat�a-vantam‘hundredfold (ACC.SG.M)’, but sut�a-vatas ‘possessing Soma (GEN.SG)’. Here we are also dealingwith many instances of analogical levelling across paradigms, but the residue can indeed bemore easily described as metrical lengthening rather than metrical shortening, since thethematic nominal bases of adjectives in -vant- were clearly more numerous than theircounterparts in -�a. A different reconstruction of the denominatives is found in Lubotsky(1989), who believes that verbs in *-�aya- shortened their suffix to -aya- before -nt-, avoidingthus the sequence of two heavy syllables. In support of his claim he cites the parallelbehaviour of verbal stems in -�aya- belonging to the roots that otherwise form the presentstems of Class 9. Thus the Vedic root grbh ‘to take’ forms present stems grbhn�a-, grbh�aya-,and grbhaya-, the first two of which can be projected at least to the Indo-Iranian level and arecommonly reconstructed as IE. *grbh-neh2- vs. *grbh-nh2-ye/o-. The third one is attestedthrough the Vedic participle grbh�ayant-, which displays metrical shortening according toLubotsky’s rule.

Table 3. Contrasive pairs of denominative verbs in Sanskrit

Transitives in -aya- Intransitives in -�aya- Base nouns

garvaya- ‘to make proud’ garv�aya- ‘to show pride’ garva- ‘pride’guṇaya ‘to multiply, advise, invite’ guṇ�aya- ‘to become or appear as a merit’ guṇa- ‘quality, merit’daṇḍaya- ‘to chastise, punish’ daṇḍ�aya- ‘to stand erect’ daṇḍa- ‘stick, punishment’p�a�saya- ‘to bind’ p�a�s�aya- ‘to become a rope’ p�a�sa- ‘snare, cord’maṇḍalaya- ‘to whirl round’ maṇḍal�aya- ‘to become a circle, coil oneself’ maṇḍala- ‘circle’mukulaya- ‘to cause to close or shut(the eyes)’

mukul�aya- ‘to shut like a bud, resemble aclosed bud’

mukula- ‘bud’

meghaya- ‘to cause cloudy weather’ megh�aya- ‘to become cloudy’ megha- ‘cloud’�salyaya- ‘to pain, torment, injure’ �saly�aya- ‘to become a thorn or sting’ �salya- ‘arrow, thorn, sting’�sekharaya- ‘to make into a chaplet,diadem’

�sekhar�aya- ‘to become a chaplet or diadem’ �sekhara- ‘chaplet, diadem’

sthalaya- ‘to make into dry land’ sthal�aya- ‘to become dry’ land’ sthala- ‘dry land’

18 Note, however, that the variant dev�aya- is attested in the Atharvaveda and in Vedic prose, while P�aṇini (7.4.38)specifically indicates that this form is typical of the K�aṭhakas. This supports the assumption that the rhythmic rulesaddressed in this section had limited dialectal scope. I am grateful to Martin K€ummel (pers. comm.) for thisobservation.

10 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

To which explanation should one then give preference, Lubotsky’s or Insler’s? Fortunately,we do not really have to choose at this point. Both metrical lengthening and metricalshortening could be conceived as simultaneous processes that affected the originalmorphological distribution between aya- and -�aya- denominatives. Short /a/ was sometimeslengthened between two light syllables, while long �a was occasionally shortened next to aheavy syllable on either side. Under this scenario, the Classical Sanskrit distribution of aya-factitives and �aya-essives can be ultimately projected to the Indo-European level, while theVedic rhythmically-driven changes would reflect the distortion of the original picture in acorpus the core of which consists of poetic hymns with quantitative metre. Having said this,the metrical shortening described by Lubotsky is more relevant to the concerns of the presentpaper, since it was likely to affect the behaviour of the etymological -�aya- formations.

But the Vedic data are special in one more respect. Tucker (1988) demonstrated that thewhole system of denominatives in -ya- underwent substantial remodelling in Early Vedic. Themorphological core of this process was the rising productivity of the present participles basedon denominative stems. Almost half of all the denominative verbs occurring in the Rigvedaare attested exclusively through such participles (Tucker 1988: 95). The new active participlesin -yant- were apparently synonymous with the adjectives in -yu-, likewise of denominativeorigin, which rarely occur after Early Vedic. The close association between the two formsextended to the semantic domain, as both could acquire the secondary desiderative meaning,e.g. vas�uy�ant- and vas�uy�u- ‘seeking goods’, from vasu- ‘goods.19 Tucker advances a plausiblehypothesis that the trigger for this development was a complex proportional analogyinvolving inherited desideratives in -su- and -sant- (e.g. d�ıps�ant- and d�ıps�u- ‘seeking to injure’).Overall, neither participles in -yant- without matching verbal forms nor denominatives withdesiderative meaning are typical of Old Indic outside Vedic.

Thus early Vedic denominatives have been shown to exhibit a whole set of innovations thatare not continued in later Sanskrit. This may have to do with dialectal or stylistic differencesbetween individual varieties of Old Indic, and their precise nature remains beyond the scopeof the present paper. But the morphological peculiarities of early Vedic could haverepercussions for its phonology. Lubotsky (1989: 96ff.) provided evidence that the participlein -yant- represent the most frequent locus for the metrical shortening -�ay�a- > -ay�a-. Thereforeit is perfectly possible that their prevalence in the Vedas contributed to the weakening of thefunctional contrast -�ay�a- vs. -ay�a- within this corpus. This explanation does not, of course,preclude the influence of the Vedic metre on finite verbal forms. Whether or not participlesplayed a special part in Vedic rhythmic shortening, the functional distribution between -�aya-and -aya- denominatives in later Sanskrit can be described as an archaism.

I would like, however, to underscore that even if one were willing to ignore altogether thedata of later Sanskrit, the Vedic data alone would be compatible with the reconstruction ofstatives in -�ay�a-. Table 4 below contains a list of synchronic denominative essives with thestress on the last vowel of the suffix, which are derived from thematic stems in the Rigveda (allof them, except for san�ay�a-, are treated in Lubotsky 1989: 101).20 It is easy to see that themajority of these verbal stems contain the suffix -�ay�a. The ratio between forms with long suffixvowel, short suffix vowel, and alternating suffix vowel is 6:2:1. There is, furthermore, a directcorrelation between the choice of a suffix and the quantity of the preceding syllable: if theinitial syllable is metrically long, -�ay�a- may be shortened. Thus all the instances of shortening

19 Note that the lengthening of �u in vas�uy�ant- etc., has nothing to do with the sporadic lengthening -aya- -> -�aya-entertained by Insler. The lengthening i > �ı_ #y and u > �u_#y in Vedic denominative verbs and related formations is acompletely automatic process, which can be fairly described as synchronic sandhi (Burrow 1973: 363).

20 Not all the intransitive denominatives listed in Lubotsky (1989: 101) qualify as essives. The necessary conditionfor a denominative X-aya- to be called essive is the availability of a loose paraphrase ‘to be X’ for such a denominative(where X is the base of derivation).

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 11

-�ay�a- > -ay�a- within this group can be fairly described as manifestations of a tendency to avoidspondaic sequences.21

The opposite explanation, namely the lengthening -ay�a- > -�ay�a- in order to break thesequence of three short syllables, would be less attractive for the verbs in Table 4, even on thestrength of synchronic Vedic evidence. It would require more analogical levelling and offer noformal motivation for the essive character of verbs under consideration. In addition to essives,Vedic also featured factitive denominatives (e.g. devay�a- ‘to worship’ from dev�a- ‘god’) anddesiderative denominatives (e.g. v�ajay�a- ‘to seek spoils’ from v�aja- ‘spoils’). Although thelimited Vedic corpus does not appear to supply us with lexical minimal pairs, the Indo-European evidence, as mentioned above, speaks in favour of the original contrast between-ay�a- factitives and -�ay�a- essives in Old Indic. This hypothesis has an obvious advantage overthe assumption that -�ay�a- denominatives were originally formed only from nominal -�a- stems,while -ay�a- denominatives were restricted to thematic stems (thus Burrow 1973: 364).Burrow’s account does not correspond to the synchronic state of affairs in Vedic, and impliesan analogical conspiracy as the explanation for the functional distribution of -�aya- and -aya-classes in Classical Sanskrit, which would then mirror the Indo-European distribution by amere accident.

We have seen that Ved. san�ay�a- ‘to be old’ is not an isolated verbal form but rather amember of a whole class of denominative essives, whose productivity only increases in theclassical Sanskrit period. Since only indray�a- and probably amitray�a- qualify as assureddenominatives in Table 4, while rt�ay�a- is ambiguous, it would be reasonable to assume thatOld Indic denominative essives were secondary to deadjectival essives, which would squarewell with the Indo-European data. I have also tried to demonstrate that the stem-formingsuffix -�ay�a- can be reconstructed for this class based on Vedic evidence alone, and that thisreconstruction receives additional support if one takes into consideration the Sanskrit forms.Although Lubotsky (1989: 102, fn. 5) suggests reconstructing -�ay�a- as Indo-European *-oie/o-,with lengthening by Brugmann’s Law, the stative suffix *-eh1ie/o- emerges as a morestraightforward alternative.22

Table 4. Intransitive denominatives/deadjectival verbs in the Rigveda

agh�ay�a- (ti) ‘to plan mischief’ i.e. *‘to be evil’ agh�a- ‘evil’ajir�ay�a- (te) ‘to be swift’ ajir�a- ‘swift’amitray�a- (PTC) ‘to act as enemy’ i.e. *‘to be inimical’ am�ıtra- ‘not a friend’indray�a- (te) ‘behave as Indra’ �ındra- ‘Indra’rt�ay�a- (PTC) ‘to act according to truth’ i.e. *‘to be true’ rt�a- ‘true, truth’tilvil�ay�a- (te) ‘to be fertile’ t�ılvila- ‘fertile’priy�ay�a- (te) ‘to become friends’ priy�a- ‘beloved, dear’rathir�ay�a- (PTC) ‘to be conveyed in a chariot’ rathira- ‘furnished with a chariot’san�ay�a- (PTC) ‘to be old’ s�ana- ‘old’sumn�ay�a- (PTC) ‘to be pious, benevolent’ sumn�a- ‘pious, benevolent’

21 In Table 4, verbs attested only through Vedic participles are marked as (PTC), whereas the others are providedwith information about their voice. It is worth mentioning that Ved. amitray�a- corresponds to amitr�aya- in classicaltexts in conformity with the syntactic distribution outlined above for later Sanskrit. On the Vedic participle rt�ayant-with accent retraction, see section 5.

22 Alexander Lubotsky (pers. comm.) regards Pre-Indo-Iranian denominatives in *-o-ie/o- as derivatives ofthematic stems in *-o- provided with the -ie/o- suffix, which was common to all the denominative verbs in Indo-Iranian. He further hypothesizes that the specific association of -�aya- with the Sanskrit intransitive denominatives is alate phenomenon, which was due to the fact that the Sanskrit factitives had generalized the -aya- suffix throughassociation with the causatives. Under this scenario Ved. san�ay�a- should be reconstructed as seno-ie/o- and itssimilarity with Lat. sen�ere, Lith. senėti is fortuitous. Compare a similar scenario discussed in section 2 with referenceto the Germanic third class weak verbs, where it is provisionally adopted as a possible solution faute de mieux.

12 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

5. CALAND STATIVES

The Vedic grammar makes a synchronic distinction between the denominatives of thematicstems in -ay�a- (and �ay�a-) and the stems with the suffix -�aya-, most of which are derived fromverbal roots. With regard to the second group, it is commonly observed that the derivatives in-�aya- formed from the zero grade of the root are frequently intransitive and rarely causative,whereas their counterparts attached to the full or lengthened grade of the root have thecausative semantics (see e.g. Burrow 1973: 331). The causative formations can be safelyprojected back on the Indo-European level, with numerous lexical correspondences. Thus IE.*sodeie/o- ‘to make sit, set’, causative formed from the root *sed ‘to sit’, is attested not onlythrough Ved. s�ad�aya- ‘to make sit, set’, but also through English set, German setzen ‘to set’,and numerous cognates in Celtic and Slavic languages (cf. Rix 2001: 513).

By contrast, the origin of -�aya- verbs with the zero grade of the root remains problematic.This topic was seriously engaged with for the first time in the Yale doctoral dissertation ofStephanie Jamison. As the author aptly observed, one can formally attempt to reconstructIndo-European forms with the -�eie/o- suffix attached to zero-grade roots, but ‘there are simplyno intransitive formations in other IE languages that should be clearly derived from such aformation’ (Jamison 1977: 293). Therefore Jamison justly assumed that this formation mustreflect the restructuring of a different category, and specifically mentioned statives in *-�e-,which she reconstructed as *-eH-. With regard to the function of the intransitive -�aya-formations, she noticed that many of them have stative semantics, and a group of them areformed from the same roots as adjectives in -ra- (more rarely -u-), belonging thus to the so-called ‘Caland System’ of lexemes denoting properties and states (Jamison 1977: 302). Apeculiarity of this group in comparison with the denominatives treated in the precedingsection, is that here the adjectives and the stative verbs represent independent derivatives ofabstract roots (e.g. cit�aya- is not the derivative of citr�a-, but both are derived from the rootcit). The somewhat expanded list of Vedic -�aya- verbs synchronically belonging to the CalandSystem23 is introduced in Table 5.24

Unfortunately, Jamison’s formal account, adopted at the advice of her academic supervisorStanley Insler, did not stand the test of time. Her dissertation endeavoured to argue that thesuffix -�aya- in intransitive verbs can be reconstructed as -eH-e/o-, where -eH- is the stative

Table 5. Intransitive Vedic verbs in -�aya- matched by adjectives in -ra-

is: �aya- (te/ti) ‘to be strong’ is: ir�a- ‘strong’cit�aya- (te/ti) ‘to appear/reveal’ citr�a- ‘clear’

dhvas�aya- (ti) ‘to smoke/cover with smoke’ dhvasir�a- ‘darkened (by clouds of dust)’dhvasr�a- ‘dissipating (like smoke)’

mrd�aya- (ti) ‘to be merciful’ mrd�u- ‘soft’

v�ıḍ�aya- (te/ti) ‘to become/make firm’ v�ıḍ�u- ‘firm’

�suc�aya- (PTC) ‘to gleam’ �sukr�a- ‘white, bright’

�subh�aya- (te) ‘to be beautiful, splendid’(also �subh�aya-)

�subhr�a- ‘splendid, pure’

23 Here and below, I refer to the constrained Caland system introduced in Nussbaum (1976), although myconclusions would also hold a fortiori with reference to its expanded version described in Rau (2009). Note that theVedic adjectives in -ra- are normally replaced with their counterparts in -u- if their root contains a rhotic element (asin mrd�u- ‘soft’).

24 All the forms in Table 5 can be found in Jamison (1977) and Jamison (1983), and most of them were recentlysummarized in a table in Rau (2009: 139), albeit without explanatory account. Some verbs listed in Table 5 are diffuse,i.e. have both transitive and intransitive meanings, but the former is invariably more rare. It seems reasonable toassume that the transitive forms are secondary and probably represent a product of realignment to the causative typein -�aya-.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 13

suffix, while -e/o- is the thematic vowel. She remained agnostic as to whether -y- represents adirect reflex of the disappearing ‘laryngeal’ or rather a secondary hiatus-breaker (Jamison1977: 299), but neither of the two alternatives would be in agreement with the mainstreamviews of the contemporary Laryngeal Theory. The overwhelming majority of scholars whowould now reconstruct the stative suffix as *-eh1- would also assume that that the sequence*-eh1-e should yield *-aʔ-a- and then contract to -�a- in Old Indic (cf. e.g. Mayrhofer 1986:124). This is presumably the reason why Jamison withdrew her formal reconstruction ofstatives in the revised version of her dissertation published as a book. She takes there anessentially agnostic stance: ‘[T]he origin of the -�aya- intransitives cannot be certainlydetermined. It may be an innovation in IIr., or it may continue a PIE formation of some type:either one identical to the transitive type with *-�eye-suffix, except for its zero-grade vocalism,or a more radically remodelled form, such as the *�e-stative (as suggested to me by S. Insler,and argued for in my 1977 dissertation)’ (Jamison 1983: 67).

Thus the problem remains unsolved. From the functional perspective, it is obviouslytempting to compare the derivational model in Table 5 with such pairs as Lat. ruber ‘red’ vs.rub�ere ‘to be red’ or OCS. b�udr�u ‘alert’ vs. b�ud�eti ‘to be awake’ and see here an extension of theessive derivation pattern described in the previous section. As befits statives belonging to theCaland System, verbs in Table 5 invariably contain the zero grade of the root. On theother hand, it appears to be simply impossible to derive is: �aya- and similar verbs from the*-�e-ie/o- statives by sound laws. Under such circumstances, it seems natural to enquirewhether the intransitives in -�aya- belonging to the Caland System might be explained as aresult of analogically driven morphological transformation.

I would like to suggest that the productive class of -�aya- was progressively spreading in Pre-Vedic and Vedic at the expense of less transparent formations. This process is postulated on alimited scale already in Lubotsky (1989: 103): ‘It seems … plausible to assume that those -�aya-intransitives, which are attested only as participles or a 3PL.MED are of denominal origin andeventually connected with denominatives in �ay�a-. This presupposes that the participles in-�ay�ant- of some denominative verbs shortened their �a for metrical reasons and then retractedthe accent’. This solution would directly account for part. �subh�ayant- existing alongside thefinite form �subh�ayate ‘is splendid, shines’ (RV 9.28.3a), the only derivative �aya- stem in Table 5.It would likewise explain the accent retraction in the participles rt�ayant- and v�aj�ayant- formedfrom the transparent denominatives rt�ay�a- ‘to act according to truth’ and v�ajay�a- ‘to seekbooty, compete’. But how should one treat stems like cit�aya-, which is reasonably well attestedin the Rigveda through both participial and finite forms (Lubotsky 1989: 98)? In such cases I seeno alternative to assuming that the earlier *cit�ay�a- fell victim to morphological restructuring.The process of vowel shortening could indeed begin first in the participles and other metricallysuitable formations, but it could not be completed without analogical levelling.25

There are two additional considerations that support the analogical replacement-�ay�a- > -�aya-. First, the distribution of accent between suffix vowels in denominatives is notphonetic in any case, since transparent Vedic denominatives normally display the accent in-�ay�a- also in the participial formations (cf. the list in Lubotsky 1989: 101). We have only asmall number of non-Caland denominative verbs, such as kul�ay�aya- ‘to nest’ vs. kul�aya- ‘nest’,v�ır�aya- ‘to act like a man’ vs. v�ır�a- ‘man’, and perhaps sus: v�aya- ‘be fertile’ vs. sus: �u ‘well-bearing’, which always appear with accent on the first vowel of the suffix in the Rigveda, and ahandful of others, mentioned above, which display accent variation. The simplest way to

25 It is worth noting in passing that that the ability to trigger analogical restructuring appears to constitute a crucialdifference between the metrical shortening -�aya- > -aya- postulated in Lubotsky (1989) and metrical lengthening -aya-> -�aya- advocated in Insler (1997). While, as argued above, the two processes need not be regarded as mutuallyexclusive, only the first one generated systemic changes.

14 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

describe this state of affairs is the lexical diffusion of analogical levelling.26 On the other hand,transparent transitive denominatives, such as arth�aya- ‘to wish, request’, n�ıḷ�aya- ‘to bring torest’, mantr�aya- ‘to advise’, or mrg�aya- ‘to chase’, regularly display the same accentuation asthe causatives (Lubotsky 1989: 101, fn. 5).

Second, unlike the transparent essives, which appear to have preserved and even sometimesrestored the middle forms in -�aya- in later Sanskrit, the Caland statives simply disappear inthe post-Vedic period. The reason for this was probably their insufficient formal differencefrom the robust class of -�aya- causatives: forms like cit�aya- ‘to appear’ were lost in acompetition with cet�aya- ‘to perceive’ and cintaya- ‘to think’. Already in the Vedic period, theetymological Caland statives are morphologically and semantically unstable. Thus of the 21occurrences of cit�aya- in the Rigveda only 13 unquestionably preserve the etymologicalintransitive meaning, while the remaining 8 are transitive or doubtful (Lubotsky 1989: 99).The alternations between active and middle voice recorded in Table 5 are also likely to reflectthe gradual adjustment to the active transitives, dominant among the -�aya- formations. Butthese explanations are easier to envisage if forms like *cit�ay�a- had been eliminated throughanalogical levelling by the Vedic stage, not merely shortened in certain metrical conditionsand preserved in others.

Thus the hypothesis that the Vedic Caland statives contained the restructured reflexes of theIndo-European *-eh1ie/o- suffix simplifies rather than complicates an account of this class.Since the verbs of this category were morphologically more opaque than the denominativesdealt with in the previous section, they could be among the first to undergo levelling to the-�aya- type. The only reason why one may wish to abandon the ‘stative’ link would be analternative satisfactory etymology for this small class. In this connection one must discuss thesuggestion of Got�o (1987: 60), who casually remarked that cit�aya- and the other verbs of‘shining’ in -�aya- can be regarded as iteratives. In order to evaluate this idea one has to cast alook at the Indo-European iteratives in *-eie/o- and their most likely reflexes in Old Indic.

It is generally accepted in the Indo-European studies that both causatives and iteratives couldbe formed by adding the *-eie/o- suffix to the o-grade of the root. The choice between the twomeanings would then depend on a language and a lexeme. Thus iterative semanticpredominates in Greek (e.g. phor�e�o ‘I carry around’ vs. ph�er�o ‘I carry’), both types are wellrepresented in Slavic (OCS. nositi ‘to carry around vs. nesti ‘to carry’, but po-lo�ziti ‘to place’ vs.le�sti ‘to lie down’), while the causative meaning is near-generalized in Old Indic (Ved. p�at�aya-‘make fall, make fly’ vs. p�ata- ‘fall, fly’). Whether the iterative type left any traces at all in Vedicis a point of disagreement in Indological literature, and the panoply of opinions on this subjectis neatly summarized in Kulikov (2008: 325–326). According to Kulikov’s own views, the onlythree Vedic verbs in -�aya- that qualify as synchronic iteratives are pat�aya- ‘to flutter, movearound’ vs. p�ata- ‘fall, fly’, dravaya- ‘to run around’ vs. drava- ‘to run’, and dyutaya- ‘to flash’vs. dy�ota- ‘to shine’. Kulikov observes that semantics of this Vedic derivational category can bebest described in terms of the opposition telic/atelic: the -�aya- derivatives refer to a random setof actions without a designated goal or point of completion.27

26 Here I am leaving aside the issue to what extent the accentual distinction between Vedic causatives/iteratives in-�aya- and denominatives in -�a-y�a can be projected back into Proto-Indo-European or Proto-Indo-Iranian. To myknowledge, there is no comparative evidence for a similar contrast anywhere outside Indic, which makes it possiblethat it reflects secondary grammaticalization. As one of the anonymous reviewers observes in his/her comments,it would be easier to explain analogical levelling in favour of -�aya- if there were inherited denominatives in -�aya- <IE -�e-ie/o-. This hypothesis, however, deserves a separate pleading.

27 Kulikov (2008) objects against using the term ‘iteratives’ for pat�aya- and similar verbs in a synchronic context,arguing that it is descriptively inaccurate. This may well be, technically speaking, the case, but reserving the label‘atelic formation’ for such verbs in a historical context would be even more confusing, since this amounts to ignoringtheir derivational oppositions. Under such a definition, for example, it would be impossible to distinguish between-�aya- iteratives and -�aya- statives, since both groups of verbs are synchronically atelic.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 15

Of these three pairs, dy�ota-/dyutaya- clearly reflect secondary denominative formationsbased on a derivative of IE. dieu- ‘daytime sky (vel sim.)’ (cf. Rix 2001: 125). This pair clearlycannot be projected into Proto-Indo-European. The peculiar feature of dravaya- is the rootvocalism a, which precludes the reconstructions of both the zero and o-grades of this iterativefrom a synchronic aniṭ-root. My own solution would be treating it as analogical to thesynchronic iterative from a seṭ-root pat�aya- (*poth1-eie/o-, Rix 2001: 597), as well as thehistorical iteratives from seṭ-roots, such as stan�aya- ‘to thunder’ (*stonh2-eie/o-, Rix 2001: 479)and svan�aya- ‘to sound’ (*suonh2-eie/o-, Rix 2001: 611).28 Given the scarcity of the data, itwould be unwise to go further, but one thing is clear: if the etymological *-eie/o-iterativesindeed exist in Vedic, they do not require the zero grade of the root, nor do they belong to theCaland System. On the other hand, none of the zero-grade formations treated in Table 5displays the loss of telicity vis-�a-vis the competing present stems. Cf. the remark by Roesler(1997: 161, fn. 290) who likewise observes that the -�aya- verbs with the zero grade of the rootnormally do not display iterative meaning.

Thus the synchronic -�aya- intransitives in Early Vedic can be described as originallybelonging to at least two distinct historical classes: statives and iteratives. The segregationbetween the two historical classes needs to be made on a case by case basis, taking intoconsideration the morphological and semantic properties of individual verbs. The mostcommon features that tip the scales in favour of the stative interpretation of a particular �aya-formation are the combination of the zero grade of its root and its membership in the CalandSystem manifested through the presence of cognate adjectives in -ra-/-u-.29

6. DEVERBATIVE STATIVES

We have seen thus far that the Indo-European ‘�e-statives’ left traces in Old Indic both asstraightforward denominative/deadjectival formations and members of the Caland System,which lies somewhere in between the denominative and deverbative derivational systems. Thisall agrees well with Jasanoff’s basic hypothesis, according to which the original locus of thestative suffix was outside the stems of primary verbs. Nevertheless, the majority of ancientIndo-European languages feature a number of verbal stems that can be described as ‘stativesproper’ (cf. Table 1). Under the hypothesis developed in the present paper they can beregarded as a result of formal convergence between the paradigms of primary anddenominative verbs. Below I will try to demonstrate that Indo-Iranian was not immune tothis process, and a number of ‘statives proper’ can be identified or conjectured in Old Indic.

It is, of course, necessary to set up strict criteria for identifying such formations. On theformal side, they must contain reflexes of the stative suffix, which, as we saw in sections 4 and5, can take the shape of -�aya- or -aya- in Vedic and Sanskrit. Since the Indo-European‘�e-statives’ (including Caland statives) attach the suffix to the zero grade of the root, thiscondition has also to be fulfilled in our case. On the functional side, the necessary conditionfor treating a stem as a deverbative stative should be the availability of a transformative stemderived from the same root, either in Indo-Iranian or elsewhere in Indo-European. Just as

28 In spite of the lack of synchronic evidence for reconstructing Ved. stan�aya- and svan�aya- as eie/o-iteratives, asopposed to eh1ie/o-statives, such a reconstruction follows from the formal properties of their Latin cognates,respectively ton�are, tonitum and son�are, sonitum, which undergo ‘laryngeal colouring’.

29 A case that need be considered separately is the Rigveda hapax rucaya- ‘to shine’ (3.6.7a). Although Jamison(1983: 62) prefers to take it as a nonce form built on the pattern of other verbs of shining collected in Table 5, this ishardly a compelling solution, because the comparison with Lat. l�uc�ere ‘to shine’ (with analogical full grade of theroot) and Hitt. lukkes- ‘to become bright’ vindicates the reconstruction of IE *lukeh1(ie/o)- ‘to shine’. In this case weare lacking a matching adjective in -ra-, but the Avestan abstract noun rao�c-ah- ‘light’ is another ‘smoking gun’ thatbetrays the membership of this stem in the Caland System. For additional derivatives implying the membership of thisroot in the Caland System, see Lowe (2012).

16 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

denominative ‘statives’ treated in section 4 may have both essive and fientive meanings, so thedeverbative formations treated in this section can function as both statives proper andingressives. This is, of course, a natural development of the stative category in a language thatdoes not feature a productive inchoative suffix.

Based on the above criteria, the following stems from the Rigveda qualify as historicalstatives belonging to the verbal system:

1 ks: -�aya-(ti) ‘to burn (intr.)’ < *dhgwh-eh1ie/o- can be contrasted with the simple thematicpresent d�ah-a-ti ‘to burn (trans.)’ < *dhegwh-e/o-. The etymological connection betweenthe transitive and intransitive stem is deemed likely in Mayrhofer (1986–2001, I: 430) andexplicitly recognized in Rix (2001: 133-4), where the suffix of the intransitive stem isreconstructed as fientive. There is, however, no empirical support for the contrived claim ofRix (2001) that this suffix originally belonged to the aorist stem, since the present stemks: -�aya-(ti) does not have an aorist counterpart. Thus this verb can be regarded as aparticularly strong argument against the treatment of statives by Harðarson and Rixaddressed in section 2. Ved. ks: �ar�a- ‘biting, acrid’ and Ved. ks: �am�a- ‘burnt’ suggest that thestative stem under consideration was also on the way to being integrated into the CalandSystem.

2 gl-�aya-(ti) ‘to be weary, exhausted’ (only the secondary causative gl�apaya- is attested in theRigveda) is tentatively reconstructed as *gwl-eH and compared with German Qual ‘pain’ inMayrhofer (1986–2001, I: 510). Further cognates of the same root include the reflexes ofthe base conjugation attested as Lith. g�elti ‘to sting, ache’ and Middle Welsh vel ‘strikes’, aswell as the Church Slavic stative derivative �zel-�e-ti ‘to be sad’ (Rix 2001: 207). From thefunctional point of view, it is obviously tempting to treat gl-�aya-(ti) as anotherinstantiation of the stative formation, but there is a formal problem. The acute intonationof Lith. g�elti suggests the reconstruction *gwelH for the root in question, so the stative stemshould be reconstructed as *gwlH-eh1i e/o-, whose expected outcome would be Ved.*gul-�aya-. But the only formal alternative in this case would be reckoning with aSchwebeablaut present *gwleH-ie/o-, which would leave without explanation the semanticproperties of the derived stem. Therefore I prefer to maintain the stative reconstruction*gwl

˚H-eh1ie/o- on the assumption that the reflexes of a ‘laryngeal’ are invisible before the

stative suffix in old Indic. On the phonetic level, this rule can be explained as laryngealdissimilation, e.g. *gl

˚H-aHya- > *gl-aHya- Another recent proposal involving a similar

dissimilation of Vedic laryngeals is the *aHiHa> aiHa rule promulgated in Jasanoff (2003:102).

3 chad-�aya-(ti) ‘to appear, please’, can be contrasted with the transitive stem chand-aya-(te)‘to enjoy’ already in the Rigveda (Jamison 1983: 54, 73). Both present stems find respectiveparallels in Avestan sadaiia- ‘to appear’ and səṇdaiia- ‘to enjoy’, and therefore must goback to Proto-Indo-Iranian (cf. Cheung 2007: 332–324). According to Rix (2001: 554), weare dealing here with a rare Indo-European present type characterized the suffix *-eie/o-attached to the zero grade of the root, but other (prototypically) intransitive verbs assignedto the same type in Rix (2001) (cit�aya-, dhvas�aya-) have already been explained asre-formed Caland statives in the previous section, and the stative explanation (with animplied shift to the ingressive) appears to be preferable also in this case.30 While thisparticular stative is poorly integrated in the Caland System, this is possibly due to the

30 It is worth a parenthetical note that the reconstruction of zero-grade *-eie/o- deverbatives on the Proto-Indo-European level generally remains highly doubtful and seems to have been introduced faute de mieux. The intransitivesassigned to this type according to Rix (2001) are all suspect of being adaptations of historical statives, while thetransitive formations assigned to it are einzelsprachlich.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 17

secondary association of the crucial nominal derivatives with the transitive stemchand-aya- (cf. e.g. ch�andu- ‘pleasing’, instead of the expected *ch�adu-, or ch�andas- ‘hymnof praise’, not *‘appearance’).

4 py-�aya-(te) ‘to swell’ can be contrasted with the diffuse stem p�ınva-, having both theintransitive meaning ‘to swell’ and the transitive ‘to make swell’ (Kulikov 2011: 315 withref.), while the thematic stem p�aya- is probably secondary (Mayrhofer 1986–2001, II: 84).The root itself is reconstructed as peiH- in Rix (2001: 464–465), while the stem py�aya- istentatively analysed in the same source as reflecting a Schwebeablaut present *pieH-ie/o-.But in this instance, just as in the case of gl�aya- above, it would be hard to offer anymotivation for the semantic properties of the derived variant. A possible alternative issegmenting the stative suffix -�a-ya- in this stem, which would be fully consistent with itsfunction. But if one starts with the Indo-Iranian reconstruction *piH-aH-ya-, it is necessaryonce again to assume laryngeal dissimilation, otherwise the expected outcome would be**piy�aya- with trisyllabic scansion.

5 �sr-�aya-(ti) ‘to become cooked’ can be segmented through comparison with the participle�sr-t�a- ‘cooked’ (Mayrhofer 1986–2001, II: 669). A plausible lexical cognate to �sr-�aya-(ti),which is tentatively adduced in Rix (2001: 323), is Lat. cal-�ere ‘to be hot’. The Vedic form,however, speaks strongly against the ‘Harðarson essive’ *kl-h1i e/o- reconstructed for Lat.cal-�e-re in Rix (2001), especially in the absence of the matching ‘Harðarson fientive’. Iwould like to suggest that the transitive counterpart to the stative present under discussionis attested in Ossetic særyn ‘to temper (trans.)’. This comparison was, in fact, alreadyadduced in a nutshell in Abaev (1958–1995, III: 89–90), but the author failed at the time todistinguish between the three different Vedic roots �sar ‘to boil, cook’, *�sari ‘to mix’, and�sray ‘to beautify’, which have since been correctly separated by Narten (1987). The Osseticform can be reconstructed as IE. *kel-e/o- ‘to heat, cook’, a simple thematic stemrepresenting the transformative counterpart to the stative *kl-eh1-ie/o- ‘to be heated,cooked’. For the semantic shift ‘to heat’ � ‘to temper’, cf. typologically Russian ras-kalit’‘to heat’ vs. za-kalit’ ‘to temper’.

6 �sv-�aya-(ti) ‘to swell, prosper’ cannot be separated from Gk. ky�e�o ‘I am pregnant’ and Lat.inci�ens ‘pregnant’, but contrasts with Av. s�auuaiia- ‘to make prosper’ (probably asecondary formation). The hypothesis that we are dealing with a historical Caland stativefinds support in the Vedic individualized adjective �s�ura- ‘strong, powerful, heroic; hero’,which also speaks for the seṭ character of the root. Its Indo-Iranian full grade is attestedthrough another Caland derivative, Ved. �s�avas ‘power’ = Av. sauuah- ‘profit’, so it can bereconstructed as *keuH on the Proto-Indo-European level (cf. Cheung 2007: 341).31 Themorphologically mediated shortening from the earlier *�sv-�aya-, generally expected ofCaland statives in Vedic, finds here additional support in the secondary stem �sv�a-, which ispatent in Ved. �sv�a-tr�a- ‘swelling, prosperous; power’ and possibly �sv�a-nt�a- ‘prosperous,powerful (?)’ (cf. Mayrhofer 1986–2001, II: 677–678). This speaks against the attempt ofRix (2001: 339) to present �sv-�aya- and its cognates as historical *-eie/o- presents derivedfrom the zero grade of the root (cf. above on chad-�aya-). But the derivation of *�sv-�aya-from the Indo-European stative *kuH-eh1ie/o- would again require the assumption oflaryngeal dissimilation, otherwise one would expect **�suv-�aya-.

31 Differently Rix (2001: 339–340), where the root under discussion is reconstructed as kueh1, while the variantkeuh1 is explained through Schwebeablaut. Nevertheless, the data collected in Rix (2001) imply that the ‘primary’ stem�sv�a is restricted to Old Indic, whereas its ‘Schwebeablaut’ variant is at least Proto-Indo-Iranian.

18 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

Summing up, there are six Old Indic present stems that match our criteria for segregatingthe reflexes of Indo-European deverbative statives. Two of these stems, chad-�aya- and �sv-�aya-can also be described as historical members of the Caland System, which therefore shortenedtheir suffix vowel. The remaining four do not shorten it, presumably because their long vowelbegan to be perceived as parts of secondary roots ks: �a, gl�a, py�a, and �sr�a. But none of the verbsbelonging to this group can be described as transparent denominatives, and therefore theiraccent is consistently placed on the initial vowel of the etymological suffix.

The hypothesis that the stative presents belonging to this group are somehow secondary tothe ‘fientive’ aorists in *a-CC-�a- has little to recommend itself, since two verbs, ks: -�aya- and�sr-�aya- lack Vedic aorists altogether, chad-�aya- shows the original sigmatic aorist ach�an etc.,while the thematic aorist �a�svat is attested in late Vedic prose. Only in the case of gl-�aya-,py-�aya-, one must acknowledge that the secondary roots gl�a- and py�a- penetrated into theaorist system, hence the secondary sigmatic aorists like Vedic py�a-sis: - and Epic gl�as�ıs. Ifthese two verbs are indeed stative in origin, one must assume that their present stems werere-segmented at an early point.

In three cases the new analysis is formally unproblematic, but in three other cases, namelygl-�aya-, py-�aya-, and �sv-�aya-, it comes at a cost, as one has to introduce a newmorphophonological rule deleting root-final ‘laryngeals’ before the stative morpheme inOld Indic. This rule can be interpreted as the Proto-Indo-Iranian laryngeal dissimilation*CHaH > *CaH and compared with the *aHiHa > *aiHa rule proposed in Jasanoff (2003).This parallel cannot be regarded as a probative argument, because both Jasanoff’s ‘AHIHA-rule’ and the dissimilation proposed in this paper are based on controversial morphologicalreconstructions. On the other hand, even the general rule prescribing the dissimilation of‘laryngeals’ separated by a vowel (*HVH > *VH) would appear to lack obviouscounterexamples in proto-Indo-Iranian, except for those which can be explained as a resultof paradigmatic levelling (e.g. NOM.SG *sakHaH ‘friend’ instead of *sakaH by analogy withGEN.SG *sakHas or LOC.SG *sakHi). Furthermore, the perceptually driven dissimilation oflaryngeal stops/fricatives would be typologically trivial and parallel to the effects ofGrassmann’s Law in Old Indic (cf. Blevins 2004: 149 with references).

The existing account for the three problematic stems in question relies on the‘Schwebeablaut’ alternations, which cannot predict the stative properties of the derivedstems. In addition, the original segmentation gl�a-ya- and py�a-ya- begs the question about thenature of secondary full grades in formations where a zero grade would be expected.32

Therefore I believe that the new morphological solution is superior to its existing alternative.But even if one is willing to discard some or all of the three problematic cases, we are still leftwith the residue of stative presents ks: -�aya-, chad–�aya- and �sr-�aya-.

7. CONCLUSIONS

In the sections 4–6 I have attempted to present three-pronged evidence for the preservation ofIndo-European statives in Old Indic. The goal of section 4 was to demonstrate that theintransitive denominative and deadjectival essives/fientives represent a productive class inboth Vedic and Classical Sanskrit, are sometimes formally confused with the factitive class in-ay�a- the Rigveda, but display the special suffix -�aya- in later texts. Section 5 is devoted to theVedic statives integrated into the Caland System, whose synchronic morphological marker

32 The secondary roots ending in -�a recently underwent a summary treatment in Kulikov (2011). The author comesto the conclusion that in the majority of cases the secondary derivatives in -�a-, whatever be their origin, display ahigher degree of transitivity than the base roots. The cases treated in this section, which represent a minority classfrom the perspective of Kulikov (2011), are a priori likely to have a different origin from the bulk of the materialtreated in this paper.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 19

-�aya- is explained through grammaticalization of the shortened variant of the stative suffix.Section 6 treats the few likely or assured cases of Old Indic statives that penetrated into theverbal system, where they had the anticausative function. The likely candidates feature thesuffixes -�aya- and -�aya-.

To the lexical comparison of Ved. san�ay�a- ‘to be old’ with Lat. sen�ere ‘id.’, adduced byWatkins and Jasanoff, one can now add Ved. ruc�aya- ‘to shine’ vs. Lat. l�uc�ere ‘id.’, Ved.�sr�aya- ‘to become ready, cooked’ vs. Lat. cal�ere ‘to be hot’, and Ved. �sv�aya- ‘to swell, prosper’vs. Gk. ky�e�o ‘I am pregnant’. Nevertheless, these four pairs are not much in comparison withthe rest of the stative forms treated in this paper, especially given that three pairs implyanalogical restructuring of one or the other form. To be sure, the mere numbers of lexicalmatches does not constitute a sufficient reason for accepting or rejecting an etymology of aderivational morpheme: thus only one Hittite ‘�e-stative’, lukkes- ‘to dawn’, appears to have alexical match outside Anatolian (Watkins 1971: 87), and nevertheless, the hypothesis ofWatkins has gained wide acceptance. But the proposed morphological comparison could bemade more attractive if it were possible to explain the relative paucity of supporting lexicaletymologies.

I believe that the reason for the small number of lexical overlaps between ‘�e-statives’ in OldIndic and the rest of Indo-European is the transfer of productive statives to the area wherethey had originally not been productive, namely to the simple denominative essive/fientives(cf. section 4). It is, of course, likely that a similar type of functional extension independentlyoccurred in other Indo-European languages. On the other hand, what distinguishes thesituation in Vedic from its counterpart in Latin or Slavic is the drastic reduction in thenumber of the Caland statives. There are considerably more Vedic adjectives in -ra-/-u- orabstract nouns in -as than their matching -�aya- statives, and the latter become completelyextinct in post-Vedic period. As a characteristic illustration of this mismatch one can mentionLat. ruber ‘red’/rub�ere ‘to be red’, Church Slavic rĭdĭr�u ‘red’/r�ud�eti ‘to be red’, and Ved.rudhir�a ‘red’ that is not matched by the stative **rudh�aya- ‘to be red’. Competition withcausatives in -�aya- was no doubt one of the factors that reduced the lifespan of Calandstatives in -�aya- in Old Indic.

A close parallel to the transfer of the inherited stative suffix to new lexical material is, infact, available in Greek. Although the intransitive denominatives in *-�e- are abundantlyattested in Homer, none of them has direct lexical parallels outside Greek, and only in thecase of bar�e- ‘to be heavy’ this is likely to be an accidental gap (Tucker 1990: 71 with fn. 82).Tucker argues at some length in her monograph that the reason for this state of affairs is therealignment of the Caland System in early Greek, where the denominative statives lost theirties with the historical Caland adjectives and came to be associated with the second elementsof possessive compounds in -�es (e.g. *alg�e- ‘to be in pain’ vs. thum-alg�es ‘heart-grieving’). Thisis, of course, a rather less trivial development than the realignment of statives with thematicadjectives, which started already in Vedic but was fully accomplished in Classical Sanskrit (asin taruṇ�aya- ‘to become young’).

Table 6. Merger between statives and performatives in the history of Greek

Stage I Stage II

Pres. *thars�e-mi Pres. kosm�e-�o Pres. thars�e-�o kosm�e-�o

Aor. (e)th�ars�e-sa Aor. *(e)k�osme-sa Aor. (e)th�ars�e-sa ? (e)k�osm�e-sa

‘to have courage’(*�e-stative)

‘to adorn’(*eie-performative)

‘to have courage’ ‘to adorn’(one class of -�e�o-verbs)

20 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

But the analogy between the situations in Indic and Greek does not end here. By the stageof Homeric Greek one can observe the complete formal merger between the etymological‘�e-statives’ and ‘eie/o-performatives’. The scenario reflected in Table 6 summarizes theargument of Tucker (1990: 181–199), also taking into consideration the results of Rujigh(2004). In prose, the present stems generalized the *-eie/o- suffix, while the secondary aoriststems extended the stative suffix to both types. At least in the aorist stems, the proposedchange was not phonologically mediated and can be described as pure analogical levellingbetween paradigms. In the case of presents stems one may argue that the intermediate form*thars�e�o violated a low-level phonological constraint on a long vowel before hiatus, and thiswas a trigger for the analogical change. But this argument would not be that different fromLubotsky’s line of argumentation, which accounts for the denominative intransitives in -aya-in terms of rhythmic shortening, with the addition of my own scenario implying thegrammaticalization of this phonological shortening in the instance of the Caland statives. Themain difference between the reconstructed developments in Vedic and Homeric Greekconcerns their scope: in Vedic only the Caland statives consistently passed to the -�aya-type,whereas in the Homeric Epic a similar change was extended to all the stative stems.

It follows from the above account that the Indic reflexes of the Indo-European ‘�e-statives’developed in the same direction as their Greek counterparts. Although few lexical etymologiesare capable of supporting the proposed scenario in either Greek or Old Indic, the scarcity oflexical comparanda emerges as a predictable by-product of morphological restructuring inboth cases. If the hypothesis of Jasanoff (2002/2003: 147) is to be followed, ClassicalArmenian has likewise preserved the inherited *-�eie/o- suffix, but modified its scope, turning itinto a generalized passive present marker of thematic verbs (berim ‘I am brought, sirim ‘I amloved’ etc.). Thus the functional re-alignment of the inherited ‘�e-statives’ is not restricted toOld Indic, but likely constitutes a common ‘negative innovation’ of the well-known Indo-European dialectal area comprising the ancestor dialects of Greek, Armenian and Indo-Iranian.

Provided that my conclusions are correct, they have far-reaching consequences for thereconstruction of ‘�e-statives’ on the Indo-European level. They confirm the hypothesisdefended by Jasanoff that the original locus of these formations was situated outside thesystem of primary verbs. More specifically, they strengthen the case for the association of‘�e-statives’ with the Caland System in Late Indo-European (cf. section 5 and Nussbaum 1976:50–56). We have seen in section 6 that Old Indic ‘�e-statives’ that formally qualify asdeverbatives are not numerous, while some of them also show synchronic connection to theCaland System. This contrasts with the larger number of ‘Oettinger statives’ within thisdomain, which are described in K€ummel (1996) and were apparently at home in the system ofIndo-Iranian primary verbs (cf. section 3 above). On the other hand, as indicated above, thesimple denominative essives/fientives treated in section 4 would represent the naturalextension of the Caland statives in a period when the complex morphological alternationswithin the Caland System stopped being productive. For example, the Caland stativecorresponding to Skt. timira- ‘dark’ could have been *tim�aya-, but the attested timir�aya- iseasy to explain as a result of trivial analogical levelling. By contrast, the secondary back-formation of Caland statives from simple denominatives would be hard to justify inmethodological terms.

On a formal level, the collected set of data falsifies Harðarson’s claim that the essivesincorporated into the present system are historically secondary to fientive aorists in *-eh1-.None of the Old Indic presents treated in this paper, with the exception of the two otherwiseproblematic deverbative statives gl�aya- and py�aya-, displays aorist counterparts in -�a-. Asmentioned in section 3, Old Indic denominative essives are not integrated into the system ofaccentual oppositions. The Caland statives appear in some cases to find a match in passive

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 21

aorists (Ved. �aceti, �a�soci, Skt. a�sobhi), but whatever one thinks about the origin of these forms,they clearly do not represent a formal match to the Greek passive aorists in -�e-. The impossibilityof reconstructing the aorists of the Caland statives even for Late Indo-European is compatiblewith the assumption that they simply did not have aorists at that stage.

The Sanskrit essives/fientives in -�aya-, whose existence can be now traced all the way backto early Vedic poetry, represent a formal match to the Hittite e-fientives, whose suffix can bereconstructed as *-eh1- or *-eh1ie/o-. Since the two suffixes represent primary ways of buildingintransitive deadjectival verbs in both Hittite and Old Indic, all speculations about thesecondary character of the *-eh1ie/o- suffix, which allegedly replaced the earlier *-h1ie/o-, losetheir raison d’etre. We have seen that the only serious argument for the traces of *-h1ie/o- inVedic is t�ay�a- ‘to be stretched’, but if the intransitive verbs in �ay�a- enjoyed productivity atsome early stage in Old Indic, this might have actually been the reason for the restructuring ofthe expected *tany�a-, which would render unnecessary an alternative account of this formalong the lines proposed in Harðarson (1998).

If my thesis is accepted, then we can safely reconstruct the ‘�e-statives’ for PIE and notmerely as a dialectal phenomenon restricted to a particular Indo-European subgroup. As longas one stays within the boundaries of comparative Indo-European linguistics, it appears nolonger necessary to derive this category from something else. The emphasis should rather bemade on studying the changing formal and distributional properties of the stative suffix inindividual Indo-European languages.

Moscow State UniversityInstitute of World CultureMoscow 119991GSP-1,Vorobjevy Gory, MGU, 1 uchebnyjkorpus, komn. 854Email: [email protected]

References

ABAEV, VASILIJ I. 1958-1995. Istoriko-etimologicheskij slovar’ osetinskogo jazyka, Moscow: Nauka.BLEVINS, JULIETTE, 2004. Evolutionary phonology: The emergence of sound patterns. Cambridge: Cambridge UniversityPress.

BURROW, THOMAS, 1973. The Sanskrit language, 3rd revised edn, London: Faber.BRUNO, CARLA, 2005. Forme della sintassi media: Due studi sulla lingua del Ṛgveda, Perugia: Guerra.COWGILL, WARREN, 1963. Review of laryngeals and the Indo-European verb by Jaan Puhvel. Language 39, 248–70.CHEUNG, JOHNNY, 2007. Etymological dictionary of the Iranian verb. Leiden: Brill.DISHINGTON, JAMES, 1976. Functions of the Germanic �e-verbs: A clue to their formal prehistory. Language 52, 851–65.DISHINGTON, JAMES, 2010. The Caland System and the Germanic third weak verb class. Historische Sprachforschung123, 297–317.

GARC�IA TRABAZO, JOS�E VIRGILIO, 2009. €Uber die Vertretung der indogermanischen Aktionsarten ‘Fientiv’ und ‘Essiv’in Hethitischen und Lateinischen. In Rosemarie L€uhr & Sabine Ziegler (eds.), Protolanguage and prehistory: Aktender XII. Fachtagung der Indogermanischen Gesellschaft. 149–158. Wiesbaden: Reichert.

GOERING, NELSON. 2012. Reconstruction and background of the Germanic Class III weak verbs. Oxford: University ofOxford M.Phil. thesis.

GOT�O, TOSHIFUMI, 1987. Die ‘I. Pr€asensklasse’ im Vedischen: Untersuchungen der vollstufigen thematischenWurzelpr€asentia, Vienna: €Osterreichische Akademie der Wissenschaften.

INSLER, STANLEY, 1997. Vedic denominatives to thematic a-stems. In Alexander Lubotsky (ed.), Sound law andanalogy: Papers in honor of Robert S. P. Beekes on the occasion of his 60th birthday. 103–110. Amsterdam: Rodopi.

JAMISON, STEPHANIE W. 1977. Function and form of the -�aya- formations of the RigVeda and AtharvaVeda. New Haven,CT: Yale University Ph.D. Dissertation.

JAMISON, STEPHANIE W. 1983. Function and form of the -�aya- formations of the Rig Veda and Atharva Veda. G€ottingen:Vandenhoeck & Ruprecht.

JASANOFF, JAY, 1978. Stative and middle in Indo-European. Innsbruck: Institut f€ur Sprachwissenschaft der Universit€atInnsbruck.

JASANOFF, JAY, 2002/2003. ‘Stative *-�e- revisited’. Die Sprache 43, 127–70.

22 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 0, 2013

JASANOFF, JAY, 2003. Hittite and the Indo-European verb, Oxford: Oxford University Press.HARðARSON, J�ON A. 1998. Mit dem Suffix *-eh1- bzw. *-(e)h1-io- gebildete Verbalst€amme im Indogermanischen. InWolfgang Meid (ed.), Sprache und Kultur der Indogermanen. Akten der X. Fachtagung der IndogermanischenGesellschaft (Innsbruck, 22–28 September 1996). 323–339. Innsbruck: Institut f€ur Sprachwissenschaft derUniversit€at Innsbruck.

KORTLANDT, FREDERIK, 1990. The Germanic third class of weak verbs. North-Western Indo-European LanguageEvolution 15, 3–10.

KULIKOV, LEONID, 2008. The Vedic type pat�ayati revisited: Semantic oppositions, paradigmatic relationships andhistorical connections. In Alexander Lubotsky, Jos Schaeken & Jeroen Wiedenhof (eds), Evidence and counter-evidence: Essays in honour of Frederik Kortlandt, vol. 1: Balto-Slavic and Indo-European linguistics. 323–42.Amsterdam: Rodopi.

KULIKOV, LEONID, 2011. The Vedic root variants of the type CaC/C(C)�a: Morphological features and syntacticpatterns. In Thomas Krisch & Thomas Lindner (eds.), Indogermanistik und linguistik im dialog: Akten der XIII.Fachtagung der Indogermanischen Gesellschaft vom 21. bis 27. September 2008 in Salzburg. 310–320. Wiesbaden:Reichert.

KULIKOV, LEONID, 2012. The Vedic -ya-presents: Passives and intransitivity in Old Indo-Aryan, Amsterdam: Rodopi.K €UMMEL, MARTIN, 1996. Stativ und Passivaorist im Indoiranischen, G€ottingen: Vandenhoeck & Ruprecht.LUBOTSKY, ALEXANDER, 1989. ‘The Vedic -�aya- formations’ (Review article of Jamison 1983). Indo-Iranian Journal 32,89–113.

LOWE, JOHN, 2012. Caland adjectives and participles in the Ṛgveda: The case of -�ana-. In Stephanie W. Jamison,H. Craig Melchert & Brent Vine (eds), Proceedings of the 23rd Annual UCLA Indo-European Conference 83–98.Bremen: Hempen.

L €UHR, ROSEMARIE, 1999. Der slawische Imperfect: Chronologie einer Periphrase mit dem Instrumental. In J€urgenHabisreitinger, Robert Plath & Sabine Ziegler (eds.), Gering und doch von Herzen: 25 indogermanistische Beitr€ageBernhard Forssman zum 65. Geburtstag. 167–182. Wiesbaden: Reichert.

MAYRHOFER, MANFRED, 1986. Lautlehre. In M. Mayrhofer (ed.), Indogermanische Grammatik, vol 1, 75–181.Heidelberg: Winter.

MAYRHOFER, MANFRED. 1986–2001. Etymologisches W€orterbuch des Altindoarischen. Heidelberg: Winter.MONIER-WILLIAMS, MONIER. 1899. Sanskrit-English dictionary, with contributions by Ernst Leumann and CarlCappeller. Oxford: Clarendon.

NARTEN, JOHANNA, 1987. ‘Ved. �sr�ın�ati, gr. jqeίxm, jqέxm’. Zeitschrift f€ur vergleichende Sprachforschung 100, 270–96.NUSSBAUM, ALAN, 1976. ‘Caland’s law’ and the Caland system. Cambridge, MA: Harvard University Ph.D. dissertation.OETTINGER, NORBERT, 1976. Der indogermanische Stativ. M€unchener Studien zur Sprachwissenschaft 34, 109–49.RASMUSSEN, JENS E. 1993. The Slavic i-verbs with an excursus on Indo-European �e-verbs. In Bela Brogyanyi & ReinerLipp (eds.), Comparative-historical linguistics: Indo-European and Finno-Ugric, Papers in Honor of OswaldSzemer�enyi III. 475–487. Amsterdam: Benjamins.

RAU, JEREMY, 2009. Indo-European nominal morphology: The decads and the Caland System. Innsbruck: Institut f€urSprachwissenschaft der Universit€at Innsbruck.

RAU, JEREMY, 2013. An athematic inflection of the Greek contract verb revisited. Handout presented at the 32nd EastCoast Indo-European Conference, Pozna�n, June 2013.

RINGE, DONALD, 1996. On the chronology of sound changes in Tocharian. Volume 1: From Proto-Indo-European toProto-Tocharian. New Haven, CT: American Oriental Society.

RINGE, DONALD, 2006. A linguistic history of English, volume 1. From Indo-European to Proto-Germanic. Oxford:Oxford University Press.

RIX, HELMUT, 1998. Lexicon der indogermanischen Verben: Die Wurzeln und ihre Prim€arstammbildungen, incooperation with Martin K€ummel, Thomas Zehnder, Reiner Lipp & Brigitte Schirmer. Wiesbaden: Reichert.

RIX, HELMUTH, 2001. Lexicon der indogermanischen Verben: Die Wurzeln und ihre Prim€arstammbildungen, 2nd revised ed.,in collaboration with Martin K€ummel, Thomas Zehnder, Reiner Lipp & Brigitte Schirmer. Wiesbaden: Reichert.

ROESLER, ULRIKE, 1997. Licht und Leuchten in Ṛgveda: Untersuchungen zum Wortfeld des Leuchtens und zur Bedeutungdes Lichts. Swisstal-Oldendorf: Indica et Tibetica.

RUIJGH, CORNELIUS J. 2004. The Stative Value of the PIE Verbal Suffix *-�eh1-. In John H. W. Penney (ed.), Indo-European perspectives. Studies in honour of Anna Morpurgo Davies. 48–64. Oxford: Oxford University Press.

STANG, CHRISTIAN S. 1966. Die vergleichende Grammatik der baltischen Sprachen. Oslo: Universitetsforlaget.TUCKER, ELIZABETH F. 1988. Some innovations in the system of denominative verbs in Early Indic. Transactions of thePhilological Society 86, 93–114.

TUCKER, ELIZABETH F. 1990. The creation of morphological regularity: Early Greek verbs in -�e�o, -�a�o, -�o�o, -�u�o and –�ı�o.G€ottingen: Vandenhoeck & Ruprecht.

TUCKER, ELIZABETH F. 2004. Denominative forms in Avestan: Derivatives from thematic stems. In John H. W. Penney(ed.), Indo-European perspectives. Studies in honour of Anna Morpurgo Davies. 549–561. Oxford: Oxford UniversityPress.

WATKINS, CALVERT, 1971. Hittite and Indo-European studies: The denominative statives in *-�e-’. Transactions of thePhilological Society 70, 51–93.

YAKUBOVICH, ILYA S. 1999. Stative’ suffix /-�ai-a-/ in the verbal system of Indo-Iranian. In Karlene Jones-Bley, MartinHuld & Angela della Volpe (eds.), Proceedings of the Tenth Annual UCLA Indo-European Conference, Los Angeles,May 21–23, 1998. 65–75. Washington, DC: Institute for the Study of Man.

YAKUBOVICH – ‘�e-STATIVES’ IN OLD INDIC 23