Christian Seidl. Wiesbaden 1994, 201-206.
RV. a
vidhat
ALEXANDER LUBOTSKY
1. The form avidhat, synchronically the 3sg. aor. of the root vidh- `to allot, apportion', is the only augmented form of this root in the RV, where it occurs ten times. In six instances, avidhat stands at the end of a jagat or anustubh pda and its first syllable must be scanned long, cf. (the comma stands for the caesura):
2.1.7d tuvampyur,dameyaste (a)vidhat 1
2.1.9c tuvamputro,bhavasiyaste (a)vidhat1
2.26.4a yoasmaihavyair, ghrtavadbhiravidhat
8.23.21b hutimmarto (a)vidhat1
8.27.15d yovodhmabhyo (a)vidhat1
8.61.9a aviprovyadavidhat
In the remaining four instances, the length of the augment cannot be determined: either
the initial vowel is combined with the preceding -, cf.:
1.136.5a yomitrya,varunyvidhajjanah
6.54.4a yoasmaihavisvidhat
9.114.1d yastesomvidhanmanah
or the position of the verb in the verse does not allow a conclusion, cf.:
10.83.1a yastemanyo,'vidhadvajrasyaka
The fact that the poets by preference used avidhat in the cadence is a strong indication
that the first syllable of this word was long. In other words, it seems certain that, in the speech of
the Vedic bards, avidhat had a long augment. Therefore, Arnold (1905: 129) proposed to read
vidhat everywhere in the text, but a few years later Oldenberg (Noten ad 2.1.7, etc.) dismissed Arnold's restitution with the remark: "das ist recht zweifelhaft". As Oldenberg did not give any
reasons for his doubts, we may only guess at them. I can think of two possible
counter-arguments. First, a short tenth syllable in a jagat verse and a short sixth syllable in an anustubh
verse are very rare in the RV, but not unparalleled. According to the counts of Arnold (1905:
149, 159f, 175, 204, cf. also Kuryɫowicz 1927: 239), there are 252 jagat or tristubh verses with a
short tenth syllable out of the total amount of nearly 24.000 verses composed in these metres
(1,05%) and 406 anustubh verses with a short sixth syllable out of the total amount of
approximately 15.000 anustubh verses (2,71%). On the other hand, Kuryɫowicz has
demonstrated (1927: 239ff, 1928: 208f) that the vast majority of these short syllables is due to a
laryngeal which originally made position (e.g., in avase < *HavHase, jana- < *janHa-, sakhi- <
*sakHi-, etc.). These cases excluded, avidhat becomes even more exceptional. Another possible
counter-argument may concern the fact that the long augment is not written in the case of avidhat
as opposed to the forms like nat (√nas- `attain'); yunak, yukta, yukstm (√yuj- `yoke');
rinak, raik (√ric- `leave'); vah (√vas- `shine'); var (√vr- `cover'); vrnak (√vrj- `turn'); vidhyat (√vyadh- `wound'), etc. The Samhit text of the RV is not free of inconsistencies, however, and this argument has hardly any weight.
It is probably due to the authority of Oldenberg that avidhat has been omitted in all later
discussions of the problem of long augments in the RV (cf., for instance, Kuryɫowicz 1928: 210ff., 1935: 31), unjustly, in my opinion.
2. Long augment normally points to an initial laryngeal in the root. Of the forms cited
above, only the lengthening in rinak, raik is secondary, since √ric- does not have an initial
laryngeal, cf. Gr. . An initial larygeal in the other roots is either certain (√nas-, √vas-,
√vrj-) or at least probable. Consequently, the long augment of avidhat is most likely due to an
initial laryngeal in the root vidh-.
As Thieme (1949: 36f) and Hoffmann (1969 = 1975: 238ff) have shown, √vidh- is a
secondary root, based on the root aorist forms of √dh- with the preverb vi-. For the philological
justification of this analysis I refer to the article of Karl Hoffmann. As √vidh- has an initial
laryngeal, we must assume that the preverb vi `apart, asunder' reflects Proto-Indo-Iranian *Hui
or *ʔui, since laryngeals most probably merged into a glottal stop in this language branch (cf.
Polome 1972: 241, 244, Lubotsky 1981: 138). The form avidhat is a precious relic of
lengthening before vi, as this is the only case where the augment precedes the preverb. The other
examples of metrical lengthening in the position before vi are of no value.
3. The only Indo-European branch where the preverb *ui was productive is Indo-Iranian
(Skt. vi, Av. v, OP vi(y)). Elsewhere we only find isolated nominal formations. Of the forms
listed by Pokorny (1175f.), only Goth. wira, OHG widar `with, against, opposed', OCS vьtorъ,
202
3 vъtorъ2 `second', corresponding to Skt. (RV) vitaram `farther', Av. vtarəm `aside',
vtara-`following, further' are certain. Less clear are Lat. vitium `mistake', vtricus `step-father', and
OIc. vir, OHG wt `wide' if from *ui-(h1)ito- (Pok. 295).
It is remarkable that outside IIr. the preverb is only attested before dentals in the next
syllable, which provides additional confirmation for the old etymology, explaining *ui- from
*dui- `entzwei'. The apparent preverb *ui- is thus due to dissimilatory loss of the initial *d- in
forms like *dui-tero-. The dissimilated form has been generalized only in Indo-Iranian, where it
eventually came to mean `apart, asunder'. Greek, Italic and Germanic have used the adverb *duis
instead with approximately the same range of meanings, cf. Gr. -, Lat. dis-, OHG ze(r)-.
In view of our reconstruction of Skt. vi as *Hui, we must assume that the initial d- of
*dui- was not simply lost, but merged with the reflex of the laryngeals. This sound change is not
so haphazard as it looks at first glance. We have a perfect parallel in Gr. `20' < *
< *h1uidkmti < *duidkmti, where the initial *d- yielded *h1- in the position before a dental in the
following syllable (Kortlandt 1983: 100). This phenomenon finds a natural explanation in the framework of the glottalic theory, according to which the Indo-European mediae were glottalic
(or, rather, preglottalized, cf. Kortlandt 1985). The development *d- > *h1- can be seen as a
partial dissimilation of *d- = *ʔd- to *h
1- = *ʔ-. The buccal part of d- disappears, whereas the
glottalization, which was a concomitant feature of d, yields a glottal stop.3
Moreover, the same development took place in the position before *k, as can be seen
from Gr. `100' < *h1kmtom < *dkmtom, `50' < *penkʷe-h1komt- <
*dkomt-, < *septm-h1komt- < *dkomt-, etc. (Kortlandt 1983). Also the Armenian
numerals point in this direction, cf. k`arasown `40', ewt`anasown `70' with -ara- and -ana- from
CrHC and CnHC.
After the presentation of my paper in Zrich, several scholars pointed out to me that Skt. dsvms- `devout, pious', which is the original perfect participle of the root das- and reflects *dedk-uos-, may be yet another example of the same rule (so already Klingenschmitt 1982: 129,
2 An alternative etymology connects OCS vъtorь with the word for `the other' (Lith. an~t(a)ras, OPr. antars, Skt.
antara-, OHG andar, etc.). The well-known objection against this etymology (cf. Vasmer s.v.) is the zero-grade, which must be reconstructed for the Slavic word and which is not attested elsewhere.
3A typological parallel to this development is the Sanskrit sound change dh,bh>h: after the loss of the buccal part
of dh and bh, the (voiced) aspiration merges with the existing phoneme h. The conditions of this change are mani-fold. First, intervocalic dh,bh became h before a final short vowel, cf. the endings of 2sg. impv. -hi and 1pl. -mahi, adverbs in -ha (iha `here', kuha `where', saha `together with' beside sadha in compounds, visvah `always' at the end of a pda next to visvadh within a pda, etc.), 1,3 sg.pf. ha `I, he said' < *dha. A second factor was dissimilation, which explains hita-, hiti- (beside dhita-,dhiti- still found in fine compositi) and rohit-, rohita- `red' as opposed to rudhira-. A different type of development is probably responsible for grha- `house' < *grdha- and grh- `to take, grasp' < grbh-. I hope to present further details in a future article. [[See now A. Lubotsky, Sanskrit h < *dh, bh. Sthāpakasrāddham, Professor G.A. Zograf Commemorative Volume, ed. by N.V. Gurov and Ja.V. Vasil'kov, St. Petersburg (Centr "Peterburgskoe Vostokovedenie"), 1995, 124-144.]]
fn. 4, who explains this form "durch Dissimilation mit gleichzeitiger Ersatzlngung", which
contradicts the Greek and Armenian facts). In view of Gr. , we must assume that the
dissimilation *duiD- > *h1uiD- was anterior to *dk > *k.
4. Let us now return to Skt. vidh-. The proposed analysis of this root as *h1uidh- < *h1 ui-dheh
1- < *dui-dheh1- has several additional advantages.
First, the explanation of vidh- from vi-dh- has often been met with some sceptisism
because we find the same root in Lat. dvidere and probably Toch. AB wtk- (caus.) `to divide'.
Since there is no preverb vi in these languages, we have to assume that the Univerbierung took
place in PIE already. At that early stage, however, there were no preverbs, only loose adverbs, and it is difficult to explain how such a syntagm could have arisen. Since we now assume a
compound *dui-dheh
1-, this objection disappears because this compound was of the same type as
*kred-dheh
1- or *mns-dheh1-, which are certainly of PIE date.
Furthermore, we may better account for the connection of the root vidh- with the IE word
for `widow'. As Beekes 1992 has pointed out, `widow' must be derived from a PIE adj.
*h1uidhu- `single (unmarried)'. The reflex of the initial h1 can be found in Gr. `unmarried
young man', which is likely to reflect * (with metrical lengthening) < *h1uidhh1euo-.
For the details of the reconstruction I refer to Beekes' article. Since there are strong reasons to
believe that √vidh- also has an initial *h1, the connection between this verb and the word for
`widow' is impeccable from the phonetic point of view. The adjective `single, bereft of its half',
which was the basis of the `widow', may go back to *dui-dhh
1-u-.
It seems to me that there are no semantic objections either. We know that adjectives of
this type existed already in the Indo-European proto-language, cf. *mns-dhh
1-ro- (Skt. medhira-,
OCS modrъ, etc.). An adjective *dui-dhh
1-u- would probably have meant `divided in two, bereft
of its half', which is exactly the meaning postulated for the original `widow' word. The original
idea of division in two parts may also be reflected in Lith. vidus, Latv. vidus m. `the middle'.
5. Finally, we may add some considerations on the Rgvedic hapax vidhu-. It most
probably is a Rtselwort for `moon' (cf. Mayrhofer 1968: 104):
10.55.5 vidhumdadrnamsamanebahnm
yuvnamsantampalitojagra| devasyapasyakviyam
mahitv--adymamrasahiyahsamna||
"The grey-haired one has swallowed the one who is young, vidhu-, running in the
company of many (= stars). Behold the wisdom of the god in its greatness: he is dead today, yesterday he took its (first) breath".
5
It has generally been assumed that vidhu- means `lonely, solitary' here, while Mayrhofer
(op.cit.: 105) has suggested the original meaning `einteilend, Einteiler' because the moon is often called in the RV `the divider of the months'. However, in the context of the new-born moon,
being swallowed by the sun, vidhu- is more likely to mean `a crescent', which is exactly what we
expect it to mean: `divided in two parts'.
References
Arnold, E.V. 1905:Vedic metre in its historical development. Cambridge. Beekes, R.S.P. 1992: 'Widow'. KZ 105, 171-188.
Hoffmann, K. 1969: Ved. vidh, vindh. Die Sprache 15, 1-7.
Hoffmann, K. 1975: Aufstze zur Indoiranistik, Band 1. Wiesbaden. Klingenschmitt, G. 1982: Das altarmenische Verbum. Darmstadt.
Kortlandt, F.H.H. 1983: Greek numerals and PIE glottalic consonants. MSS 42, 97-104.
Kortlandt, F.H.H. 1985: Proto-Indo-European glottalic stops: The comparative evidence. Folia Linguis-tica Historica 6, 183-201.
Kuryɫowicz, J. 1927: Les effets du ə en indoiranien. Prace Filologiczne 11, 201-243.
Kuryɫowicz, J. 1928: Quelques problemes metriques du Rigveda. Rocznik Orjentalisticzny 4, 196-218. Kuryɫowicz, J. 1935: Etudes indoeuropeennes. Krakow.
Lubotsky, A. 1981: Gr. pgnumi: Skt. pajra- and loss of laryngeals before mediae in Indo-Iranian, MSS 40, 133-138.
Mayrhofer, M. 1968: Vedisch vidhu- "vereinsamt" – ein indogermanistischer Mythos? Studien zur
Sprachwissenschaft und Kulturkunde (Fs. W. Brandenstein). Innsbruck, 103-105.
Oldenberg, H. Noten: Rgveda: Textkritische und exegetische Noten. 2 vols. Berlin. 1909-1912.
Polome, E. 1972: Reflexes of laryngeals in Indo-Iranian with special reference to the problem of the voiceless aspirates. Saga og sprak, FS L.M. Hollander. Austin, 233-351.
Thieme, P. 1949: Untersuchungen zur Wortkunde und Auslegung des Rigveda. Halle.