Laryngeals and Vedic Metre

Achtung! Dies ist eine Internet-Sonderausgabe des Aufsatzes „Laryngeals and Vedic Metre“ von Jost Gippert (1996). Sie sollte nicht zitiert werden. Zitate sind der Originalausgabe in Sound Law and Analogy. Papers in honor of Robert S.P. Beekes on the occasion of his 60th birthday, ed. A. Lubotsky, Leiden: Rodopi 1997, 63-79 zu entnehmen. Attention! This is a special internet edition of the article “Laryngeals and Vedic Metre” by Jost Gippert (1996). It should not be quoted as such. For quotations, please refer to the original edition in Sound Law and Analogy. Papers in honor of Robert S.P. Beekes on the occasion of his 60th birthday, ed. A. Lubotsky, Leiden: Rodopi 1997, 63-79. Alle Rechte vorbehalten / All rights reserved: Jost Gippert, Frankfurt 1998-2011 Laryngeals and Vedic metre Jost Gippert (Universität Frankfurt) Within the study of Indo-European laryngeals — a field that has owed so much to Robert BEEKES —, phenomena relating to Vedic metre deserve of special attention. It has for long been stated that vestiges of laryngeals that once existed can be traced in the metrical behaviour of word forms such as rátha"wagon", the first syllable of which has often to be measured as long because of the "position" created by h2 in the underlying structure *rot.h2o- (where . marks the original syllable boundary between t and h2, later "merging" into an aspirate -th-, probably via a consonant cluster *-thh-)1. Investigations into this type of material have gained a new basis eversince the text of the Rgveda-Sa h itā (RV) exists in a reliable digitalized form. At present, three versions are available to the scholarly world that can be used complementing each other: The edition recently published by Barend VAN NOOTEN and Gary HOLLAND (hereafter: vNH) contains, in electronic form, both the plain Sa h itā text (according to AUFRECHT’s edition) and a metrically restored version of it (mostly following the proposals made by E.V. ARNOLD)2. Additionally, a sandhi-free (pada-pātha-like) version was prepared by A. LUBOTSKY for his Rigveda concordance3. Although none of these versions contains any metrical data as such, they can easily be adapted to a special computer program which parses the text in order to reveal the metrical shapes present in each verse4. In a forthcoming article5, I have reported about the results of an investigation using such a metrical "parser" with respect to a crucial laryngealistic question that had been raised by the late Jochem SCHINDLER on the occasion of the 1993 Copenhagen meeting of Indo-Europeanists. During the panel discussion about ṁ ˚ ṁ 1 Phenomena of this type were noted passim in H. OLDENBERG’s "Metrische und textgeschichtliche Prolegomena", Berlin 1988, E.V. ARNOLD’s "Vedic metre" (Cambridge 1905), and OLDENBERG’S Textkritische und exegetische Noten", Berlin 1909-1912. A first thorough discussion in laryngealistic terms was offered by J. KURYŁOWICZ ("Les effets du en indoiranien" ("Effets"), in: Prace filologiczne 11, 1927, 201-243). 2 Rigveda: A Metrically Restored Text with an Introduction and Notes, Harvard University Press 1994. 3 My thanks are due to A. LUBOTSKY for contributing this version to the "TITUS" collection of texts relevant for Indo-European studies ("Thesaurus Indogermanischer Text- und Sprachmaterialien"; cf. the WWW page under http://titus.uni-frankfurt.de/texte/texte.htm) and for discussing quite a lot of problematic passages with me. 4 The parser which was developed as a means for establishing a metrical and a sandhi index to the RV will be made available to the public after completion of the project in question ("AUREA"). — On the basis of an automatical electronic comparison, the different versions were also used to locate typing errors persisting in them. Besides, a list of typing errors as present in AUFRECHT’s RV edition was prepared (cf. my forthcoming article "TITUS — Alte und neue Perspektiven", to appear in SIMA 2, Prague 1996). 5 "Neue Wege zur sprachwissenschaftlichen Analyse der vedischen Metrik"; to appear in the Gedenkschrift für Jochem SCHINDLER. e 64 J. GIPPERT the phonetics of the laryngeals, SCHINDLER proposed to see a threefold representation of h2 in cases where words like duhitár- "daughter" have to be measured with a long first syllable: Here, h2, besides causing aspiration (of *-g-) and invoking a shewa (> -i-; "vocalization"), obviously created metrical position (schematically: PIE **dhugh2tér- > Proto-Indo-Aryan *dhuǐh.h tár-). Investigating the behaviour of word forms containing the structure -V̆(T)hi- (i.e. a short vowel followed by either a plain -h- or by an aspirate stop, plus -i-), in the most decisive metrical constellation where a long syllable is required, viz. the ictus within the cadence of both dimeter and trimeter verses (syllables 6:8, 10:11 and 10:12), astonishing results were achieved: Of 38 matching examples to be found in the whole RV, 17 belong to the type discussed by SCHINDLER (besides duhitár-, cp. e.g. p t hiv´ı̄- "earth": **plth2 i h2- > *p ht .h i h-). Of the remaining 21 cases, 15 still show position created by laryngeals: cp. e.g. mahini, voc.sg.fem. of mahín- "great", < *maíh.hin-, with original -i- (not i < schwa), or avahitá-: *a a dh.’ tá-, √dhā- < √dheh1-, where the aspiration is not caused by the laryngeal. In 6 cases only, the lengthening cannot be assigned to laryngeals with certainty (this is less than a sixth part). The following table lists the results6: e ˚ ṙ e ŭ *p t .h . í h- 9 mahínı̄avyathí- *maíh.hí.ni- 1 *a= . h a t .híh ŭ *a. a =d .’ .táe avahitá- ĭ ṙ ŭ h 4 * á .su=d .’ .ti- 8 áprahitadevayúbhi *á=p.ra=.íhi.ta- 1 3 ŭ a =. ú .bhiš ĭ ŭ *da . 4 1 vásudhiti- e ĭ ḣ none 4 p t hiv´ı̄- h ṙ twofold: position, "vocalization": *dhuǐh.h .térŭ e twofold: position, aspiration: ´ duhitā- verse type: syllables: 8 12 11 e h2 (plus i) h1,3 ŭ word type: h2 ṙ reflexes threefold: position, aspiration, "vocalization": ŭ e Laryngeal 1 2 As can be seen from the table, there seems to be a striking difference between 11-syllable verses and 8- or 12-syllable verses in that the former do not show any laryngeal reflexes of the types discussed. This may easily be explained by assuming that the tri u bh cadence with its ictus on the paenultima leaves hardly room enough for the word forms in question: We must consider that in this case, the syllable containing -hi- has to be the last one; a constellation which is hardly to be met with where laryngeals are involved (but cp. mahí, pathí-). Nevertheless it seems worth while investigating whether there are principal divergences in the structure of cadences between "even" and "odd" verses in the RV, as far as metrical position created by laryngeals is concerned. ṫ ṡ If we want to take unexpected "short" syllables bearing the ictus as a criterion, we first have to define what syllables we regard as "short" from a formal áprahita- belongs to √hi- "to send", not to √dhā-. The results will slightly differ depending on the treatment of avyathíbhi in 7,69,7c (in a Tri u bh-cadence); for this and avyathí u cf. below. ṡ 6 ṫ ṡ ḣ LARYNGEALS AND VEDIC 65 METRE point of view. According to general assumption, short syllables are open syllables containing a short vowel, syllable-final consonant(s) normally producing position (with muta cum liquida as a special case) and diphthongs regularily measuring as long. A further special case has to be observed where short vowels are followed either by the palatal aspirate, ch, or by the retroflex laterals, and h : It can easily be shown that these consonants produce positional length throughout, in accordance with their emergence from former consonant clusters (ch < *ść, < žd, íd etc.): cp. gachati < *gaśćati or d h á< *d ídhá-7. Such cases can therefore be ruled out before starting the investigation8. The same holds true for those instances where the second syllable of pāvaká- "pure" bears the ictus in cadences9: It has been for long generally adopted that pāvaká- is but a secondary substitution for an original pavāka- yielding "correct" cadences10. Lastly we can dismiss those instances ´ gen.pl. of nár-/n - "man", is in the ictus where the first syllable of n ām, position: Here, the has to be scanned as long11. l̇ l̇ ṙ l̇ l̇ ṙ ṙ ṅ ṙ ṙ Looking for the remaining instances that match the conditions, the metrical parser gives the following immediate results when asked for "irregular" cadences12: Verse type 8 syllables total occurences irregular cadences percentage 7,15 12 syllables 11 syllables 14973 6924 16939 1071 153 278 2,21 1,64 7 vNH mark as long in such cases; this gives the correct results although the marking seems not just adequate because we have to deal with positional length here, not with vowel lengthening. 8 All in all, this constellation is present in the cadence of 116 11-syllable verses, 50 12-syllable verses, and 95 8-syllable verses (based upon the vNH "metrically restored" text). As to figures from less usual verse types, cf. below. 9 14 occurrences in 11-syllable verses (3,5,7c; 17,1c; 4,5,6a; 4,6,7c; 6,1,8c; 4,3c; 10,4d; 7,3,1d; 9d; 7,9,1b; 9,97,7c; 10,46,4c; 7b; 8c); one occurrence in an 8-syllable verse (3,21,2a: gh t ávanta pāvaka te). 10 In vNH, pavāká- is written throughout in the printed text; in the electronic (metrical) version, however, three cases of pāvaká- have persisted in RV 10,46 (4c; 7b; 8c). 11 9 occurrences in 11-syllable cadences (3,52,8b; 4,25,4d; 5,30,12d; 7,1,11a; 19,10d; 62,4d; 10,29,2b; 99,9d; 111,1b – the last three not marked in VNH); 5 occurrences (reading n¯ āá m) in 8syllable cadences (1,48,4d; 3,16,4d; 5,18,5e; 7,32,11d; 8,66,5b) and one in a 12-syllable verse ´ in an 8-syllable cadence (ní (10,93,12b). In 1,43,7b, we seem to have an exceptional disyllabic n ām ´ dhehi śatásya n ām), in a verse of the so called "trochaic Gāyatrı̄" type. 12 Against the "metrically restored" electronic text as provided by vNH, only a few rearrangements were made concerning the separation of verses (especially with respect to the two types of B h atı̄ and U i metres as dealt with in OLDENBERG’s "Prolegomena", 111 sq.; cp., e.g., RV 3,10 or 8,12 and 13), and some 30 obvious typing errors were corrected. The twelve verses of RV 9,67,27a-d, 31a-d and 32a-d missing both in the printed text and the electronic (metrical) version were added. — It goes without saying that the statistical results obtained here agree to a large extent with the ones offered by E.V. ARNOLD’s study; nevertheless it seems worth while entering into the study of Vedic metres again. ṙ ḣ ṙ ṅ ṙ ṅ ṙ ṅ ṙ ṙ ḣ ṅ ṡ 66 J. GIPPERT Of course, this enormous bulk of material13 cannot be investigated entirely here, all the more since the less common (and mostly irregular) verse types containing 4 to 7, 9, 10, or more than 12 syllables (all in all 995 verses in vNH) would deserve of a special treatment too. On the basis of the considerations noted above, it seems reasonable to restrict the present study to the cadences of 11-syllable verses14. From the 278 verses in question, a relatively large number (63) can be excluded by assuming that they show a regular J-cadence although the number of syllables in the verse cannot be "stretched" easily to give 12. In most of these cases (52), vNH suggest such a solution themselves in their "Metrical notes" (p. 577 sqq.), denoting the verses as "Jg. 11 syll.", "Tr Jg cadence" or the like (often referring to ARNOLD or OLDENBERG); there is no special marking in the electronic text, however. The solution is acceptable if the verse in question is a single 11-syllable line in a J-environment and if it shows a long syllable in the antepaenultimate syllable; cp., e.g., 6,15,1d jyók cid atti gárbho yád ácyutam which even by reading jiyók contains but 11 syllables. In two hymns, viz. 10,77 and 10,78, this type of verse is even prevalent. According to OLDENBERG’s "Noten", it emerged from a replacement of the usual two short syllables after the caesura by one long syllable; cp. 77,1a abhraprú o ná vācā´ pru ā vásu, where vācā´ stands in the position of a regular J-centerpiece consisting of ∪∪–. Like this, we achieve the following list of instances in accordance with vNH: 1,89,4b; 2,13,1a; 24,5b; 36,1a; 5,51,13a; 6,15,1d; 6,48,17c; 8,26,24c; 52,2a; 98,7b; 9,71,7c; 79,1c; 10,23,2d; 4c; 32,4c; 46,7c; 49,1d; 2a; 50,5b; 64,10c; 77,1abcd; 2abcd; 3abcd; 4abcd; 5ab; 78,1abd; 2d; 3bd; 78,4abcd; 78,6cd; 93,7b; 14c. ṡ ṡ In the remaining eleven cases, a similar solution is suggested by the fact that the given cadences or the word forms they contain usually occur in the final position of J (or A) verses. This holds true for the following pādas15: 13 The remarkable difference between dimeter and trimeter verses as revealing itself in the statistics is mostly due to the well known tendency of the Anu u bh metre towards the "epic" ślokatype verse, the normal cadence of which is ––× in odd verses; cp. 10,72,1a and c as against b and d: ´ ´ devānā nú vayá jānā prá vocāma vipanyáyā / yá páśyād úttare yugé // ukthé u śasyámāne u The structure ––× is indeed met with in 375 (from 1071) verses, 306 (153 + 153) of which are a first or third pāda in Anu u bh strophes of this type. Another great bulk of cases (422) increasing the number of "irregular" cadences of eight-syllable verses is the so-called "trochaic Gāyatrı̄" type with the structure ∪–×; cp. 1,27,10a-c: ´ d´śı̄kam// stóma rudrāya járābodha tád vivi h i viśé-viśe yajñíyāya / Both these types have no counterpart in trimeter verses. 14 Abbreviations used hereafter: T: 11-syllable (normally Tri u bh) verses; J: 12-syllable (normally Jagatı̄) verses; A: 8-syllable (mostly Anu u bh or Gāyatrı̄) verses; O: opening; M: centerpiece (of trimeter verses); C: cadence; B: caesura (break). No before-hand distinction will be made between equal verse types occurring in different strophic arrangements (as, e.g., J in normal Jagatı̄-strophes, 12-12-12-12, or in B h atı̄-strophes, 8-8-8-12). Note that regularly, cadences of J and A verses have the same metrical shape (–∪×) 15 If nothing else is stated, all verses will be reproduced hereafter according to the electronic text of vNH where metrically "restored" vowels are marked by italicization (this principle is reversed ṙ ṫ ṡ ṁ ḣ ṁ ṡ ṡ ṫ ṡ ṁ ḋ ḋ ṫ ṡ ṫ ṡ ṙ LARYNGEALS VEDIC AND 67 METRE 1,61,11a asyéd u tve á sā ranta síndhava (síndhava 8× in JC, 17× in AC; read asyá/ā´ íd u16?); 1,121,8c hári yát te mandína duk á n v d hé (v d hé after a long syllable 8× in JC, 10× in AC); 1,162,22a sugáviya no vāj´ı̄ suáśviyam (sváśvyam, always to be scanned as suáśviyam, 3× in JC, 4× in AC17); 2,18,2d só anyébhi sacate jényo v´ ā // (read jéniyo v´ ā as in 1,140,2c JC); 2,42,2c pítryām ánu pradíśa kánikradat (read pítriyām; kánikradat 7× in JC, 13× in AC); 8 103,5c tuvé devatrā´ sádā purūvaso (purūvaso 4× in AC); 10,30,13a ´ (práti yád āpo ád śram āyat´ı̄r (āyat´ı̄ 1× in JC, 4× in AC, cf. OLDENBERG, ´ "Noten" a.l.); 10,59,5c rārandhí na sūriyasya sa d ´śi (cp. 10,37,6c with ´ sūriyasya sa d ´śi in a regular JC; similarly sa d ´śa 2× in JC, sa d ´śam 1× in AC); 10,103,8a índra āsā nayitā´ b´ h aspátir (b´ h aspáti 3× in JC in a Tcontext [10,109,5c, 128,7c, 161,4c], 5× in AC; other case forms similarly); ´ 10,132,1b ı̄jānám bhūmir abhí prabhū á i (all other -sáni-infinitives appear in JC or AC: tarı̄ á i 4,37,7d; 5,10,6d; g ı̄ á i 6,15,6b; 8,12,19b; upast ı̄ á i 6,44,6b; sak á i 10,32,1a; śū á i 10,93,1d; ne á i 10,126,3c; par á i 10,126,3d; i á i 2,2,9d). Lastly, a J-cadence is likely to be seen in 6,25,3a índra jāmáya utá yé ’jāmayo (better to be read ájāmayo with OLDENBERG, "Noten"). A special case is provided by three 11 syllable verses that should equally rather be scanned as containing 12 syllables, viz. 6,44,11d jahy ásu ı̄v n prá ´ v āp h ´ a ta (read prá v āh áp a ta as in 5,7,10c in AC), 1 100,6b asmākebhir ´ n´ b hi sūriya sanat (to be read with a rest after the fourth syllable), and ´ ´ 10,50,5a ávā nú ka jyāyān yajñávanaso (read jiyāyān as in 3,38,5a; cp. yajñávanasam in JC in 4,1,2b). Here, we have no overt long antepaenultima to match the ictus of JC; but considering the etymology of the underlying verbs, we can assume that an original root final laryngeal left its traces here in metrical position: sanat is a 3rd person singular aor.inj.act. of sani- < *senh2"to gain" ("[Indra] gains the sun together with our men"); and connecting áp a ta "the not generous ones" (acc.pl.: "weed out the misers") with pari(pres. p āti, < *perh3-18) "to give, donate" and yajñávanas- "pleased by / longing for sacrifice" ("assist those who are pleased by sacrificing, being superior") with vani- (pres. vanate, < * e nH-19) "to be(come) pleased, to desire", we arrive at the underlying structures -prn.‘a.tas (< *-prnh3ntns), -van.Ha-sas (< *- e nHesns20), and san.Hat < *snh2et21. ḣ ṙ ṙ ḣ ṡ ṡ ṁ ṁ ṁ ṡ ṙ ṡ ṙ ḣ ṁ ṙ ṙ ṙ ṁ ṁ ḣ ḣ ṙ ḣ ṁ ṙ ṙ ṙ ṅ ṅ ṡ ṅ ṙ ṅ ṅ ṡ ṅ ṡ ṁ ṁ ṡ ṅ ṙ ṅ ṡ ṅ ṡ ṡ ṅ ṡ ṅ ṡ ṡ ḣ ṅ ṙ ṙ ḣ ṅ ṙ ṁ ṙ ḣ ṙ ṁ ḣ ṅ ṙ ṅ ṙ ˚ ˚ ˚ ŭ ˚ ˚ ˚ ŭ here). 16 For the hymn in question (1,61) cf. J.S. KLEIN, The particle u in the Rigveda, Göttingen 1978, 108 sq. 17 In 1,180,9c and 10,113,10a, sváśvyā and sváśvyam seem to appear in T-cadences; vNH read suáśvyā / suáśvyam. Here, the ictus is on a long syllable (-áś-) in any case. 18 Cf. M. MAYRHOFER, Etymologisches Wörterbuch des Altindoarischen ("EWAia"), Heidelberg 1986-, II, 90 sq. s.v. PARI2. 19 Cf. MAYRHOFER, EWAia II, 501 s.v. VANI- and below. 20 Or *- o nHesns, cf. below. 21 That the thematic root aorist had a zero grade root is made probable by clear cases such as (á-)vidat; cf. R. BEEKES, in: Die Laryngaltheorie, ed. A. BAMMESBERGER, Heidelberg 1988, 65 for *snH-o-m and T. GOTŌ, Die "I. Präsensklasse" im Vedischen ("Präs.Kl."), Wien 1987, 284 for the corresponding thematic root aorist stem * nH-é- from * e nH- "to desire", O.Ind. vaná- (cf. below). According to J. NARTEN, Sprache 16, 1968, 119, the formation of the thematic aorist of san- can be assigned to the Indo-Iranian stage. — For the phonetic development involved (*-nHV- > -an.’V-) cf. ˚ ŭ ˚ ˚ ŭ ŭ 68 J. GIPPERT This type of laryngeal traces is not restricted to 12-syllables verses though. Within irregular cadences of 11-syllable pādas, it can be found in a surprisingly high number of cases. Starting with the structure of the last mentioned examples, i.e. -V̆NV- < *-V̆N.HV- (with V̆ standing in the cadence ictus, N representing any resonant, H, any laryngeal) this can be made responsible for at least 20 cases. In 13 of these verses, the cadence contains either jánā or jánān, nom./acc.pl. of jána- "man", as in 4,38,9a utá sma asya panayanti jánā "and the men praise his (Dadhikra’s) (zeal ...)" (the other cases: 1,89,10c; 173,8d; 2,20,2b; 3,46,2d; 5,33,2d; 6,10,5d; 11,4d; 20,1b; 49,15c; 51,11b; 67,3c; 68,5d); this is nearly one fourth of all occurrences of the word (54)22. As jána- has to be connected with jani- < *ǵenh1- "to beget" (most probably *ǵónh1o-23), its treatment is equivalent to the one of -vanasa as noted above. A similar case is provided by śava "strength" (≈ śa . ’as < *ḱe h 1os24) as ´ vithurá ná śáva "they will make (things) easily in 1,186,2d káran su āhā defeatable like a wavering power" (two further attestations in TC: 6,19,6a; 10,148,4b25; cp. 1,27,2a; 127,11g; 3,37,1a; 5,7,3c; 20,2b; 52,2a; 87,1e; 6,14,4c; 44,3a; 8,1,21b; 2,30c; 9,20a; 68,4b; 70,6b; 92,14a; 10,105,6b, where three syllable case forms like instr. śávasā appear in irregular ACs). One further instance would have to be added if in 1,122,10b the substitution of gūrtáśravā "whose glory is praised" by *gūrtáśavā "whose power is praised", proposed by Chl.H. WERBA26, is right: śárdhastaro narā´ gūrtáśravā . We have to note, however, that the same constellation — a masculine -s-stem as the second member of a compound in an irregular TC — is attested three more times, viz. in 1,89,6a (v d dháśravā "whose glory grows"), 1,167,5b (n m á ā "having men’s mind"), and 10,61,1a gūrtávacā "whose speech is praised"); for none of these, a laryngealistic explanation is possible27. Instead, we should consider that this behaviour might be due to an ablaut difference between -s-stem simplicia and compounds, -śravā and the like having been "regularized" for an older *-śrāvā from -o-coloured *-ḱlo ēs as the compound variant of *ḱlé o s28. The "regularization" leading back to -śravā would then be equivalent to what happened in Greek eùmen<h@ if this replaced a former ¯´ A similar view is possible for suśrávasam in 1,91,21c, and *h1su-mones. suprayásam in 2,4,1b. Here, it is the suffix vowel which seems to bear the ḣ ḣ ŭ ŭ ḣ ḣ ṁ ṡ ḣ ḣ ḣ ḣ ṅ ṁ ṙ ḣ ṙ ḣ ḣ ŭ ḣ ḣ ŭ below. 22 This is why its behaviour was noted as early as in OLDENBERG’s "Prolegomena" (1988, 478); it was treated in KURYŁOWICZ, o.c., too. 23 Cf. R. BEEKES, o.c., 64, who notes this word as an "apparent exception to Brugmann’s law" produced by "a postconsonantal laryngeal"; further cf. MAYRHOFER, "EWAia" I, 566. 24 ŚAVI- < *ḱe h 1- according to MAYRHOFER, "EWAia" II, 623 sq. 25 This verse seems to lack one or two syllables, cf. the restoration dã´ n´ b hyo n ã´ śūra śáva in vNH; but it has as clear a T-cadence as the surrounding verses. 26 In: WZKS 36, 1992, 15. 27 But cf. below for -mahas-. 28 Cp. J. WACKERNAGEL, Altindische Grammatik ("AiGr") II/1, Göttingen 1905, 101, § 43 who mentions ví-hāyas- "having extended power" and sárva-hāyas- "having all power(s)" (after OLDENBERG). The process envisaged here would well be motivated if J. SCHINDLER’s (unpublished) assumption of a regular ablaut shift leading from proterodynamic to hysterodynamic stems in derivation and composition is right. ḣ ŭ ṁ ṅ ṙ ṙ LARYNGEALS AND VEDIC 69 METRE ictus; cp. bhare u jā´ suk i tí suśrávasa "(Soma), born in battles, (gaining) ´ agním átithi suprayásam "Agni, good abodes, having good reputation"; viśām the guest of the houses, the one granting good favour(s)". If we suppose, however, that these pādas are "shortended" Jagatı̄ verses, suśrávasam can be ´ accounted for in the way indicated above (*su-śrāvasam < *h1su-ḱló e s-m). For -prayás-, there is an easier solution even because this word contained a root final laryngeal29 (*su.pra . há.sam < *h1su-pro H és-m); indeed, its simplex also occurs once in a comparable TC environment, viz. 2,19,1b mánı̄ i a suvānásya práyasa "(there is a drink), exstatic ones, (made) from the pressed delightment" (cp. *prá . Ha.sas < *pré/ó . Hes.os). We must not neglect, however, that this analysis too presupposes the "stretching" of the given 11syllable verses to Jagatı̄ pādas (by adding a "rest" before the caesura). The same surface structure as in śávas- is present in urú "wide" (6,50,3a utá dyāvāp t hivı̄ k a trám urú "and (you), heaven and earth, (bestow upon us) wide suzerainty") and in two derivatives of tari- < *terh2- "to pass, transgress, overcome", viz. táturi- "overcoming" (4,39,2d dadáthur mitrāvaru ā táturim "(Dadhikra, whom ye,) M., gave (to the Pūrus) as the winning one"; 6,68,7d táturi ) and v t ratúram "overcoming the foe" (6,20,1d daddhí sūno sahaso v t ratúram "son of might, give (us wealth) to overcome the foe"). Here, the vowel bearing the ictus most probably developed in a sequence of a syllabic H -u-; cp. sanat < sonant plus a laryngeal, i.e., *-RHV- > -uR’V- (*-t 2h - / * *snh2et as discussed above). Some further cases are less clear. First, we may seek vani- in two compounds appearing in 5,41, viz. j uváni- (5,41,15d smát sūríbhir j uhásta rjuváni ) and ´ upamātivani- (5,41,16e asmākam bhūd upamātiváni ) . Usually, these compounds are connected with the root van- meaning "to win, to defeat" (cp. GELDNER’s translations, "die redlichen Gewinn bringt" and "der Gewinner von Schenkungen"). But in the following verses (17bc), we read dévāso vánate mártiyo va | ā´ devāso vanate mártiyo va 30, which, according to T. GOTŌ’s study31, could easily be understood as containing vani- "to desire": "Gods, the mortal one desires you; the mortal one, Gods, desires you (to come) to (him)". If this is correct, we may interpret j uváni- as "whose desire is straight" ´ (speaking about "mother Rasā"), and upamātiváni as "the one longing for 32 allotments" (Ahi Budhnya) , even though vasuváni- (7,1,23c) and v i váni(10,98,7c), both occurring in TM, are rather derived from van- "to win" ("gaining good(s)", "winning rain"); cp. 7,15,4c vásva kuvít vanāti na "he (Agni) will for sure win us some good" and 10,98,3c yáyā v í śá t anave vánāva "(the speech) by which we shall win rain for Śa t anu"33. If we accept ṁ ṁ ṡ ṁ ṡ ṁ ˚ ŭ ˚ ḣ ṅ ĭ ĭ ṡ ḣ ĭ ĭ ṡ ṙ ṅ ṙ ḣ ṙ ṙ ŭ ˚ ṙ ḣ ˚ ṙ ṙ ḣ ḣ ṙ ḣ ṫ ṡ ṙ ḣ ṁ ḣ ṁ ṫ ṡ ṙ ṁ 29 Cf. MAYRHOFER, "EWAia" II, 181 s.v. PRAYI-. 30 ´ in 17c, indicated in the printed edition of vNH, is not present in the The "rest" after devāso electronic text and should be dismissed. 31 "Präs.Kl.", 284; GOTŌ mentions verse 17b only. 32 No other forms of van-/vani- appear in any context together with Rasā or Ahi Budhnya in the RV. 33 Cf. J. NARTEN, Die sigmatischen Aoriste im Veda ("Sigm.Aor."), Wiesbaden 1964, 235 with n. 716. 70 J. GIPPERT J. WACKERNAGEL’s proposal that in the present type of compounds, the -ibelongs to the root whenever it is not accented34, we should expect that -vánirepresents a se root in both cases. But the differentiation of van- "to win" and vani- "to love, to desire" into one ani and one se root35 does not work in all cases36: Note, e.g., that the accusatives vasuvánim and v i vánim are also located in a position where their third syllable should be long, viz. in the dadhāti position after the caesura in T verses: sá devátā vasuváni (∪–∪–/∪∪––\∪–∪: 7,1,23c) and devaśrúta v i váni rárā o (––∪–/–∪––\∪–∪: 10,98,7c). So this cannot disprove the hypothesis that -váni- in j uváni- and upamātiváni- represents an underlying *- é nH-37, i.e., a root noun from a se root as a second member in a compound. Regarding the measuring of -va- as long in 5,41, we can offer two different explanations then: either, this is a case of a "twofold" reflex of a laryngeal, manifesting itself by position + "vocalization" (*- á n.Hiš < *- é nH s < *- é nHs), or it is the trace of a former lengthened grade to be expected in the nominative of this type of compound ´ The first solution seems to be preferable if we (cp. the type of v t rahā). consider that the formations in -i- were usually remodelled as -i-stems showing no ablaut in the root syllable at all38 (cp. the plural genitives mahi v á ı̄nām and tuvi v á ı̄nām appearing in 8,46,18cd, both belonging to svani- "to roar"). ṫ ṫ ṫ ṫ ṡ ṙ ṁ ṅ ṁ ṫ ṡ ṙ ṁ ṙ ṫ ŭ e ŭ ŭ ŭ ṙ ṅ ṡ ṅ ṡ The same structure can then be seen in suśámi in 5 87,9a gántā no yajñá yajñiyā suśámi "come to our sacrifice with good execution, venerable ones", if this belongs to śami- < *ḱemh2- "to finish; get finished, get tired, calm down / be finished with"39: Although the form is not completely clear morphologically, the -i can be an immediate reflex of the root final laryngeal, which additionally left its traces in metrical position. If we adopt to this view, we can further assume that in avyathí- "not wavering, steadfast" too, the -í- is a reflex of the root-final laryngeal (* e th2-), although it was (secondarily) accented40. This would add avyathí u (8,2,24a in AC) to the group of cases showing threefold laryngeal effects (*a. a th.hi.šu < *a. a th.h .šu < *n- e th2su); and the same assumption could be made for avyathíbhi , occurring in 7,69,7c in TC (patatríbhir aśramaír avyathíbhir "(you, Aśvins, saved Bhujyu) with (your) winged (horses) which never get tired, which never stumble"), provided that we can "stretch" the verse to have 12 syllables, either ṁ ḣ ĭ ŭ e ṡ ĭ ŭ ĭ ŭ ĭ ŭ ˚ ḣ 34 WACKERNAGEL argued in reverse direction ("AiGr" II/1, 224, § 92d): "wo dagegen i wurzelhaft, also das Hinterglied eigentlich suffixlos ist, gilt .. Paroxytonese". 35 For the separation of VAN- and VANI-, cf. MAYRHOFER, "EWAia" II, 499 and 501. 36 It seems that van- "to win" developed more and more into vani- under the influence of both its quasi-homonym, vani-, and its synonym, sani-; cp. NARTEN, "Sigm.Aor." 235, n. 712. 37 Cf. KURYŁOWICZ, "Effets", 233: "L’accentuation des tatpuru a ’s en -sáni, -váni etc., prouve que l’i < n’est pas traité comme un i originaire". 38 This derivation seems to be preferred by St.W. JAMISON, in: Die Laryngaltheorie, ed. A. BAMMESBERGER, Heidelberg 1988, 222 with n. 16. 39 Cf. MAYRHOFER, "EWAia" II, 610 sq. 40 Cf. JAMISON, o.c., 222, who mentions the six occurrences of "nom./acc. (-)vyáthis" in the RV. ṡ e LARYNGEALS AND VEDIC 71 METRE by inserting a rest at the caesura, or by replacing aśramaír by the concurrent instrumental, aśramébhir41. The same type of laryngeal residue could further be assumed in 5,33,7a the cadence of which contains the imperative ava "help" according to the arrangement proposed in vNH: evā´ na indara ūtíbhir ava "(come) hither to us like this, Indra, with your help(s); help (, protect the poets) ...". There can be no doubt that the root present in ava contained a laryngeal that could be responsible for the first syllable to be measured as long (*a . Ha < *h2e H e); cp. the -s-stem dative ávase (*h2é H os-) which appears 22× in AC and 6× in JC. As to the analysis of the verse in question, however, there is room for serious objections. Given that the following pāda has 9 syllable as it stands (pāhí g a tá śūra ´ ava seems rather to belong to this than to 7a, all the more since ūtíbhi kārūn), would produce a perfect 8 syllable cadence (as it does in 43 other A verses). Of course we would have to expect ava to be accented in this case (áva) so that a correction of the traditional text would be necessary in any way42. Another word that is not easy to account for is śáru- "arrow, weapon" which appears two times in TC; cp. 1,186,9c ádha yád e ā sudíne ná śárur "when (the Maruts’ armies attack) like a thunderbolt on a clear day", and 10,99,7b ā´ ´ śárum "he (Indra) will create the weapon for Arśasāna" (for sāvi a d arśasānāya the formula, cp. 7,45,3b ā´ sāvi a d vásupatir vásūni where Savit is meant). Reconsidering an etymological connection of śáru- with the se root śarimeaning "to dash, to smash"43, these cases could easily be explained by assuming position invoked by the root final syllable again: śáru- < *śár.Hu-44. Note that although both instances belong to allegedly late hymns, the first one at least has a clear formulaic character45. One further type that might belong here is attested all in all 20 times. This type consists mostly of superlatives showing the suffix -tama- as in 5,41,4d ājí ná jagmur āśúaśvatamā "(they have come to the sacrifice) just as the ŭ ŭ ŭ ḣ ṅ ṙ ḣ ṁ ṡ ṡ ṙ ṡ ṫ ḣ ṁ 41 But we have to concede that within 7,69, this would be the only 12-syllable verse in a pure Tri u bh environment. Should we rather suppose that avyathíbhi was restored from an original avyáthı̄ showing the deletion of the ending -bhi in an environment where group inflection can be assumed (cf. L. RENOU, Grammaire de la langue védique, Paris 1952, 83, § 105, or C. WATKINS, Indogermanische Grammatik ("IGr") III/1, Heidelberg 1969, 143, § 130 for examples)? In this case we would still have to "fill" the verse in order to arrive at 11 syllables. 42 Cf. OLDENBERG, "Noten" a.l.; according to GELDNER (translation a.l.), "das Metrum ist nicht in Ordnung" in 7ab. Instead of áva, we could even think of substituting déva. The corruption could ´ áva (inserted as a gloss), and déva. be due to a mutual influence between evā, 43 This etymology was upheld by WACKERNAGEL-DEBRUNNER ("AiGr" II/2, 475) but styled as "semantisch wenig befriedigend" by MAYRHOFER, "EWAia" II, 618. 44 Like this, the old comparison with Greek keraun<o@ "thunderbolt, flash" has to be reconsidered too (cp. GELDNER’s translation of 1,186,9c: "wenn dann wie ein Blitz bei heiterem Himmel ..." with F 199-200: àlla> ka>i o] @ de<idoike Di>o@ megaloio < keraun>on | dein<hn te bront<hn, o) t’ àp’ oùran<o\en smarag<hs9.). Should śáru- — pace FRISK — reflect the same original - / n-stem as the Greek word does? A nom. śáru / śárur < *ḱérh2 r, could indeed have been metanalysed as a fem. -u-stem, śaru. The development of *-VrH r > *-Vr.Hur might have been supported by some oblique case forms showing *-VrHunV- > *-Vr.Hu.nV-. 45 The instr.sg. śárvā is trisyllabic (śáruvā) throughout in the RV (5×: 1,100,182b; 2,12,10b; 4,28,3d; 7,85,2d; 10,87,6d: always in TM). ṙ ŭ ḣ ṫ ṡ ḣ ˚ ŭ ˚ ŭ ḣ ḣ 72 J. GIPPERT (riders) having the fastest horses have come to the race" (the other examples: 1,61,1d rātátamā; 1,158,5a māt´ t amā; 1,169,5a tośátamā ; 2,20,6b dasmátama ; ´ i tamo; 6,4,7d n´ t amā ; 6,29,4d devávātatamā ; 6,33,3d n t ama; 4,4,3a tūr 6,33,5d go á tamā ; 6,37,4b tuvikūrmítama ; 6,38,1a citrátamo; 6,50,7c ´ 6,63,3b suprāya á tamam; 6,67,1a māt´ t amā; 6,62,5a puruśākatamā; jyé h atamā; 10,3,6c devátamo; 10,6,6c índravātatamā). The same structure may be seen in adjectival -amá-stems designating extreme local positions like paramá- "the most distant one"; cp. 6,25,1a yā´ ta ūtír avamā´ yā´ paramā´ "whatever help from you, the nearest one and the most distant one" (another example with paramá- in 1,167,2c). The derivation from a former *-tmmo- / *-mmo- as proposed by WACKERNAGEL / DEBRUNNER ("AiGr" II/2, 609 / 752) could easily be motivated if we assume an underlying *-tmHo- / *-mHo-, implying again that the laryngeal was able to invoke metrical position after a sequence of short vowel + consonant resulting from an original syllabic sonant (schematically: *-mHo- > *-amHo-). We cannot exclude though that the metrical behaviour in these cases is due to an expressive lengthening of either the vowel or the consonant involved. A special case is provided by three instances where a gen. or loc. dual stands in TC; cp. 6,29,2c ā´ raśmáyo gábhastiyo sthūráyor "(the charioteer,) the reins in (his) firm hands ...", similarily, with pronominal ayór, 6,25,6a sá patyata á m ayór "he (Indra) disposes of the manly courage of both these ubháyor n m (sc. armies)". Discussing forms such as pitrós which has always three syllables in the RV, K. HOFFMANN proposed to reconstruct the ending of the gen.-loc.du. as *-h1o s (in his notation, *- 1o )s 46, which would well account for these two instances (-ayo < *-a . ’a š ). The same holds true for bhuríjo in 4,2,14c rátha ná kránto ápasā bhuríjor: No matter whether bhuríj- means "arm" here (cf. GELDNER: "mit der Arme Arbeit"; MAYRHOFER, "EWAia" II, 266: "Arme, Hände") or, rather, a two-part tool like scissors (cf. GRASSMANN: "Scheere; Schnitzbank"; "like the ones making a chariot with the endeavour of their tongs"), this form will contain the same dual ending, *-h1o s . But note that here, the laryngeal must have produced metrical position in combination with š . a stop: bhu.rij.or < *bh . ’iǐ/í.’a 47 This same development can be assumed for the two irregular T-verses the cadence of which contains forms of dhā-, viz. 1,103,4d yád dha sūnú śrávase ´ nāma dadhé "the name ‘son of power’ which he (Indra) has acquired (for ´ himself)" and 6,62,9a yá ı̄ rājānāv t uthā´ vidádhad "the one (of the) two kings (Mitra and Varu a ) who (recognizes the demon), ruling according to the seasons". Here, we have to deal with a sequence of an inherited media aspirata plus h1 which produced metrical position: dadhé < *dadh.’á < *de-dhh1-é , -dádhat < *-dádh.’at < *-dé-dhh1nts. Much better attested are those word forms where the second laryngal left its traces in both the aspiration of a preceding stop and in metrical position. This type comprises first 15 instances of rátha- and its derivatives (*rátha- < ḣ ṙ ṙ ḣ ḣ ṙ ṅ ḣ ḣ ṡ ṅ ṙ ṫ ṡ ˚ ˚ ˚ ˚ ˚ ṅ ĕ ŭ ḣ ṙ ŭ ŭ ĭ ḣ ṁ ŭ ŭ ṙ ḣ ṙ ṁ ṅ ĭ ˚ 47 In: Aufsätze zur Indo-Iranistik ("Aufs.") II, Wiesbaden 1976, 561, n. 2. The etymology of bhuríj- remains unclear; cf. MAYRHOFER, "EWAia" II, 266. ĭ ḣ 46 LARYNGEALS AND VEDIC 73 METRE *ráth.ha- < *róth2o-), and 6 instances of sákhi- and compounds (nom. *sákhā < *sákh.hā < *sékuh2ō( )) ; cp., e.g., 4,16,20b bráhma akarma bh´ g avo ná rátham "we have made (now) a prayer (for Indra) just like the Bh g us (made) a chariot" (the same cadence is repeated in 10,39,14b átak āma bh´ g avo ná rátham), or 4,4,10c tásya trātā´ bhavasi tásya sákhā "you will be his protector, his friend" (the other attestations: rátha-: 1,77,3b; 100,16c; 141,12b; 186,8c; 2,20,1b; 5,31,5c; 33,5b; 6,65,2b; 66,7b; 99,4c; sákhi-: 1,63,4a; 4,17,18a; 7,34,24b; 7,96,2c; 9,96,2c). The same type can further be seen in vadhaí in 1,121,9d ´ vadhaí "(when) you (Indra) overcome Śu a with chú a m anantaí pariyāsi 48 endless strokes" . As vadhá- obviously belongs to the verbal root vadh- "to kill, to strike" and as this root is most likely to have been an ani root vadhi- on account of its -i a- orist49, vadha- will represent an underlying * a dh.há- < * o dhh2ó-. Here, we have to note a difference as against rátha- and sákhi- in that we cannot decide whether the aspiration was triggered by the laryngeal alone or whether there was an original aspirate media (cp. *roteh2, sekueto with plain stops). A special case is created again where the aspirate emerging from a sequence of a stop plus h2 developed into Old Indic h. This case seems to be present in four verses the cadence of which contains either mah´ı̄m "big" (acc.sg.fem.) or compounds with mahas- "largeness" as their second member; cp. 10,74,4c ´ mah´ı̄ "who (want to milk the cow) that has born sak t súva yé puruputrām once, that has many calves, the big one"; 6,37,5b índro gı̄rbhír vardhatā v d dhámahā "may Indra grow by the songs, (he) whose largeness has increased" (another instance of v d dhámahā in TC is 6,20,3b); and 6,3,6b śocí ā rārapı̄ti mitrámahā "he (Agni) mutters with (his) flame, (he) who has the largeness of Mitra". These cases can easily be accounted for if we suppose an underlying *máíh.h- < *méǵh2-50. A word strongly reminding of máh- "large" with respect to its metrical behaviour is áhar/n- "day". In TC, we meet its locative áhan three times; cf. 4,12,1b trís te ánna k á vat sásminn áhan "three times preparing food for you (Agni) on the same day" (the same cadence in 10,95,11c) and 10,68,11c ´ támo ádadhur jyótir áhan "they (the fathers) installed darkness in night rātryā and light in day". A fourth instance of áhan- in TC may be seen in 1,140,13c ´ "(aurorae), going (up) for (abundance in) cows gávya yávya yánto dı̄rghāhā (and) corn, for long days ...", if this has to be scanned as proposed in vNH: gávya yáviya yánto dı̄rghā´ áhā51. If we consider that in 8-syllable verses too, the word figures three times in irregular cadences as in 1,132,1d nédi h e ṙ ĭ ṙ ṙ ṡ ḣ ṅ ṡ ḣ ḣ ṅ ṡ ṫ ŭ ṡ ŭ ĭ ṁ ṁ ṙ ṁ ḣ ḣ ṙ ṙ ḣ ṡ ṅ ṙ ṁ ṁ ṁ ṁ ṁ ṁ ṫ ṡ 48 Cf. K. HOFFMANN, Der Injunktiv im Veda ("Inj."), Heidelberg 1967, 191 with n. 156 for the passage in question. 49 Cf. NARTEN, "Sigm.Aor.", 233. 50 According to J. SCHINDLER (in: Festschrift for Henry HOENIGSWALD, 345 with n. 50) the -s-stem máhas- (as contained in v d dhámahas- and mitrámahas-) was only secondarily remodelled after the adjective máh- < *meǵ-h2, its original form being *meǵ-es- with no laryngeal. If this is right, we have to assume that either the remodelling took place early enough to be able to produce metrical position (*meǵ-es > *meǵ-h-es) or we have the reflex of a compound -o-grade here (*-moǵ-es, cp. the examples discussed above). 51 Cf. OLDENBERG, "Noten" a.l. for a discussion of this and other possibilities; according to him, we should expect a shortening of the final syllable (dı̄rghā´ > dı̄rghá) in this case. ṙ 74 J. GIPPERT asmín áhani "on this next day" (the other instances, always with áhani: 5,82,8a; 10,18,4a), this can hardly be accidental. As the etymology of áhar/n- is far from being clear52, we remain free to analyze áhan and áhani as *áíh.han and *áíh.ha.ni, resp., the aspirate deriving from a former sequence of a palatal stop plus h2 (*éǵh2en?). A similar problem is posed by b h án "high" in 6,24,3a ák o ná cakríyo śūra b h án "like the high axle (protruding) over the two wheels, hero, (you, Indra, by your largeness, protrude over the two worlds)". But as this metrical behaviour remains unique within the attestations of b h ánt-, we should not search for a laryngealistic solution, all the more since the syntactical construction is not totally clear53 and the verse looks defective (note the "rest" inserted after the third syllable in vNH)54. A different type of laryngeal treatment responsible for irregular cadences can be seen in a set of five instances where the second syllable of i i rá- "lively, frisky, vigorous" has to be measured as long; cp. 5,37,3b yá ı̄ váhāte máhi ı̄m ´ "(a husband) who will marry her as (his) lively spouse" (the other i i rām instances: 2,29,1a; 9,96,15c; 10,68,3a; 98,3b). If the supposed derivation from a preform *(H)ish1ró- connecting i i rá- with Greek "ier<o@ is right55, we may attribute the metrical behaviour to a variant *i ı̄rá- (later removed by the redactors) where the laryngeal was reflected by -ı̄- as, most probably, in ga(m)bhı̄rá- "deep". This solution remains questionable, of course, as long as the conditions under which -ı̄- emerged have not been established56. But the view that the second -i- derives from a former laryngeal seems to be supported by two instances where i é , dat.sg. of the verbal root noun i - "to urge, push forward, send out, further", bears the cadence ictus; cp. 10,50,3a ké té nára indara yé ta i é "who are the men, Indra, who (care) for your refection" (the second example: 6,13,2a). If this is the same root as the one present in i i rá-57, we might well suppose that it was the root final laryngeal here which produced metrical position (*iš.Há ) . The same development as with i ¯˘ırá- can perhaps be assumed to explain the irregular cadence of 10,108,8a éhá gamann ´ a ya sómaśitā "the seers will come hither, (their senses being) ‘whetted’ by Soma". According to this interpretation, the compound sómaśita- contains the verbal adjective śitápertaining to the root śā-/śı̄-58 which can be derived from *ḱeh3-/ḱh3-. Although ḣ ṡ ṙ ṙ ṙ ṡ ṡ ṁ ṡ ṡ ṡ ṡ ṡ ṡ ṡ ĭ ṡ ḣ ṡ ṙ 52 Cf. MAYRHOFER, "EWAia" s.v. who remains sceptical as against a connection with the Germanic word for "day" (OHG tag etc.). 53 Cf. OLDENBERG, "Noten" and GELDNER’s translation a.l. 54 ´ Cf. the proposal by H. BERGER according to whom Should b h án be influenced by *mahān? Pāli brahant has to be explained in a similar way (Zwei Probleme der mittelindischen Lautgeschichte, München 1955, 21). 55 Cf. BEEKES, o.c., 62; MAYRHOFER, "EWAia" I, 199 is more cautious. 56 Cf. JAMISON, o.c., according to whose results *i ı̄ra- is hardly motivated. — Another solution for *i ı̄rá- could consist in assuming an influence of its synonym, jı̄rá- where -ı̄- most probably resulted from i + laryngeal (*guih3ro-). 57 Cf. MAYRHOFER, "EWAia" I, 198 and 271 sq. s.v. e 2-; for the verbal root, MAYRHOFER thinks of h2. 58 Cp. GELDNER: "durch Soma scharf gemacht"; GRASSMANN: "durch Soma geschärft". ṡ ṙ ṡ ṡ LARYNGEALS AND VEDIC 75 METRE a variant *śı̄ta- seems not to be a regular outcome from *ḱh3tó- (> *ḱ¯ tó-?), it might well have emerged by influence of other zero grade forms in the paradigm such as śiśı̄te59. If it had, it could have been removed by the redactors because of the conflict with its homonym, śı̄tá- "cold"60. A similar case seems to be provided by 6,50,2a sujyóti a sūriya dák a pit¯ n "sun (god), (ask) the beautifully shining ones, whose father is Dak a ...". But although pitā´ "father" derives from *ph2te(r), ¯´ it would very hard to restore -pitā as *-pı̄tā (< *p¯ tā) in this unique case. The second attestation of pitā´ in an irregular cadence, in 7,55,5a sástu mātā´ sástu pitā´ "may mother sleep, may father sleep", is not comparable and does not help. In 9,88,1c, the cadence is represented by a finite verbal form, vav é : tuvá ha yá cak é tvá vav á "(the Soma drink) which you (Indra) have made and chosen for yourself". If the attribution of this form (a 2nd person sg.perf.med.) to the root var(i)- "to choose" is correct, we should expect a reflex of the root final laryngeal (* e lh1-) in a form like *vavūr e (*vaūr e ?) < * e - lh1-só . This could have been remodelled as various other forms of the paradigm were61; in the present verse, the cooccurrence of the "rhyming" cak é could have supported the substitution. The same effect — the substitution of a se zero grade by an ani one — is ´ nom.sg. of the perf.act. participle of sani- "to gain" which visible in sasavān, ´ appears three times in irregular T cadences; cp. 9,74,8b kār m ann ā´ vāj´ı̄ akramı̄t ´ sasavān "the racer, having won, has reached the (finishing) mark" (the other ´ the regular two instances: 7,87,2b; 10,29,2d). The process by which *sasāvān, ´ was clearly outcome of *se-snh2- o(n)s, ¯´ was secondarily replaced by sasavān, 62 demonstrated by K. HOFFMANN ; in the attestations in TC, we have a clear vestige of its former existence. In 3,20,5d the cadence is represented by ihá huve: vásūn rudrā´ m̆˙ ādityā´ m̆˙ ihá huve "I call the Vasus, the Rudras, (and) the Adityas (to come) hither". The ´ as in 1,45,1b) was proposal to see a Jagatı̄ cadence here (reading āditiyām rejected by OLDENBERG ("Noten" a.l.) who stated that we should expect a lengthened *ihā´ in this case. The present TC could be motivated if we were to assume that huvé, 1st person sg. of an old root present63, contained a sequence of two laryngeals originally (*íhuH.há < *ǵhuH.h2á ) ; but such a laryngeal treatment would remain unparalleled. A comparatively large set of cadences is characterized by short word final vowels bearing the ictus. Here we have to differentiate two types, depending on whether the following word has an initial consonant or not. The latter case, represented by verses like 7,42,4c súprı̄to agní súdhito dáma ā´ "Agni, the well satisfied one, well installed in the (sacrificer’s) house", occurs five times in TC (the other instances: 1,60,4c; 149,1b; 2,4,3d; 10,105,4a). It can easily be e ṙ ṡ ḣ ṡ ṡ e ṁ ṡ ṙ ṡ ṙ ĭ ˚ ŭ ŭ ṡ ṁ ṡ ṙ ṡ ṁ ŭ ṡ ṙ ṫ ṫ ṡ ŭ ĭ ˚ ĭ ḣ 59 Cf. JAMISON, o.c., 223 sqq. for a discussion of this type of presents. This belongs to ŚYĀ "to freeze", cf. MAYRHOFER, "EWAia" II, 660 sq. 61 Cf. HOFFMANN, "Aufs." 1, 247 sq., according to whom the remodelling was probably due to the unusual shapes the root gained when v was lost in sequences of v + u/ū. 62 In: "Aufs." II, 544 sq. 63 Cf. NARTEN, "Sigm.Aor." 288. 60 76 J. GIPPERT accounted for by assuming that here, the final vowel, reduced to short a by later application of sandhi rules, has to be restituted in its original form (e.g., dame ´ This solution, envisaged but not carried out by vNH64, has no bearing on ā). laryngeal reconstruction, of course. The same holds true for most of the examples representing the second case, i.e. word final short vowel before initial consonant. This comprises two verses ending in hí á (cp. 5,2,4c ná tā´ ag b hrann ájani a hí á "they did not seize him, for he was just born"; the second example is from the same hymn, 5,2,7b); four verses have úpa na (cp. 4,16,1b drávantu asya háraya úpa na "his dun (horses) are to run towards us" (the other examples: 4,21,1a; 7,92,1a; 93,6a); one case each is provided by ú na and u na (2,20,1a vayá te váya indra viddhí ú a "we (bring) you, Indra — recognize us well! — a refreshment"; 6,51,10a té hí śráyi h avarcasas tá u nas "for they are the most beautifully brilliant ones, and they (lead) us .."); and in 5,41,5b, the verbal ending -ta bears the ictus: rāyá é e ávase dadhı̄ta dh´ı̄ "searching for wealth, may the song be used for help". In all these cases the problem would be solved if we could substitute a lengthened variant for the word forms in question; we have to admit though that while ū and sū are as well attested as the lengthened variant of the medial ending, -tā65, *úpā and *h´ı̄ do not occur in the Sa h itā text. A further example of hí bearing the TC ictus can be excluded with certainty: In 1,36,12ab rāyás pūrdhi svadhāvó ásti ´ hí té | ágne devé u āpiyam, the verse boundary should rather be inserted before (á)sti, given that the sequence of ásti plus hí plus enclitic appears only verseinitially elsewhere. A laryngeal can be involved in just one such case, viz. in 1,62,5a g ānó áṅgirobhir dasma ví var "praised by the Aṅgirases, you, the master, opened (the darkness)". var is an aorist injunctive of var- "to enclose", the indicative of which has a long augment throughout (āvar, e.g. in 1,92,4d; 1,113,4b; 13b; 4,52,6b; 5,45,1d; 6,44,8d; 7,95,6b; 8,9,16c; 9,97,38b). The only counter-example would be 5,31,3c where támo ’va is written; but this should rather be read as támo va again (with an injunctive as in 1,62,5a)66. If the long augment is a reflex of a root-initial laryngeal (*e-H e l-t)67, this same laryngeal could be responsible for the lengthening of ví in the cadence of 1,62,5a (*v´ı̄ var < * í H e lt), assuming a word internal development for the sequence of preverb + finite verbal form. Yet another type of laryngeal development can be seen in 1,173,12b ásti hí ām te śu m inn avayā´ "for there is, ardent one (Indra), an apology (going on) to you (by the sacrificer)". If we are right in assigning four syllables to the ´ "apology", as was proposed by OLDENBERG68, this word in question, avayāḣ ṡ ḣ ṡ ṫ ṡ ṙ ḣ ḣ ṡ ḣ ṅ ṡ ṁ ḣ ḣ ṫ ṡ ṡ ḣ ṁ ṡ ṅ ṙ ḣ ḣ ŭ ŭ ŭ ḣ ṡ ṡ 64 Cf. the introduction, p. vi. Cp., e.g., śiśı̄tā in 8,40,10a and 11a; av ı̄tā in 2,33,13c and 7,33,2d. 66 Note that both cases refer to the same (Vala-) myth; for the injunctive, cf. HOFFMANN, "Inj.", 172 and 150. 67 Cf. M. PETERS, Untersuchungen zur Vertretung der indogermanischen Laryngale im Griechischen, Wien 1980, 46 sq. for a discussion of the possible Greek evidence for assuming *h1 e l-. 68 "Noten", a.l.; vNH have avayã´ in the electronic text but avayā´ in the printed one. 65 ṅ ṙ ŭ ḣ ḣ LARYNGEALS AND VEDIC 77 METRE ´ could be due to a root final laryngeal (a.va.yā.’as < *a a āHas, yā- < * a H- "to 69 pray") although it seems unusual that in this type of "Zerdehnungen" of long vowels, the first element should be long70. The assumption is supported, however, by a formula which is repeated four times in RV 1, viz. é ā´ yāsı̄ a tanúve vayā´ "may it (the song) pray for apology for (the poet’s) body, (going) forth with refreshment" (1,165,15c; 166,15c; 167,11c; 168,10c). If we accept ´ here (the initial aOLDENBERG’s view again who proposed to read avayām being elided by true abhinihita sandhi), we arrive at the following scansion: é ā´ ´ 71. This would match exactly with avayāa ´ yāsı̄ a tanúve ’vayāa in 1,173,12b. A last verse that would fit into this picture is 7,40,5a asyá devásya ´ ) "an apology for this gracious god": Reading ’vayāa ´ mı̄ h ú o vayā( with OLDENBERG again, we should achieve the same type of tri u bh cadence as in the examples dealt with before72. ´ ) in irregular TC in the RV, viz. There are two more verses containing vayā( 6,13,1b ágne ví yanti vaníno ná vayā´ and 6,24,3c v k á sya nú te puruhūta ´ This is a different word, however: vayā´ f. means "twig"; cp. the vayā. translations "Agni, (from you all goods) emerge like the twigs (branch off) from a tree"; and "like the twigs of a tree, your (graces grow), often called one (Indra)!". Are we to suppose that vayā´ reflects a former * a H áh2, with a 73 laryngeal motivating its use in TC? In 1,122,10d, a "Zerdehnung" of the syllable bearing the cadence ictus is ´ proposed for a long ū: víśvāsu p t sú sádam íc chũra . Although there is no doubt that the ū must be due to a laryngeal (*ḱúh1ro-74, cp. śávas- as dealt with above), there is hardly any reason why this might have to be dissolved into two syllables, all the more since in the given cadence, the second syllable ought ´ ´ to be the long one (*śu.ū.ra ) . Therefore we should rather suspect that śūra was substituted for śávı̄ra here, i.e. a full grade variant (*śá ¯ ra- < *ḱé 1h ró-)75. We have to consider, however, that this latter word is rarely attested, appearing as a feminine only (1,32,2b and 1,30,17b śávı̄rayā instr.sg.). A last set of irregular cadences that might be explained in laryngealistic ´ gen.pl. of áp- "water" as in 3,1,5c śocír vásāna pári terms contains apām, ´āyur apā´ "(Agni,) dressing himself in glowing fire and in the vitality of waters" (the other examples: 1,181,1b; 2,19,3a; 2,35,11b; 6,13,1d; 6,17,12b; 7,101,2a). But although áp- certainly derives from a preform with h2-, *h2ep-, ĭ ṫ ṡ ĭ ŭ ṡ ṁ ṡ ḣ ṁ ḣ ṫ ṡ ḣ ṡ l̇ ṫ ṡ ḣ ṡ ĭ ṙ ŭ ḣ ḣ ḣ ṙ e ŭ ḣ ḣ ŭ ḣ ṁ ´ with the root yā- cf. NARTEN, "Sigm.Aor.", 210 with n. 632; For the connection of avayāMAYRHOFER, "EWAia" II, 408 sq. s.v. YĀ2. 70 Cf. OLDENBERG, "Noten", 165, who states: "Allerdings ist die Setzung dieses ā˜ (∪∪) im Tri u bhausgang (––˘) eine Lizenz, die eben hingenommen werden muß". 71 vNH read ā´ iśā´ yāsı̄ a tanúve vayā´ which has the disadvantage that there are both an unusual caesura ("rare caesura 3|") and an unmotivated cadence. 72 In vNH, the verse is treated as "Tr. 10 syllables" so that it was not counted in the present study; there is no indication of the necessary "Zerdehnung". 73 Is the word connected with the numeral "two" as German Zweig suggests? Can it contain the dual formant, -ih1-? 74 Cf. MAYRHOFER, "EWAia" II, 650 sq. 75 Cf. MAYRHOFER, "EWAia" II, 624. ṁ 69 ṫ ṡ ṫ ṡ 78 J. GIPPERT these cases are not easy to account for. Considering that in compounds, the zero grade of the (root) stem was used yielding word forms such as dvı̄pá- "island" < *d i -h2p-ó- or anūpá- "marshland" < *(h1)enu-h2p-o-76, we might be tempted ´ came to replace a former ı̄pām ´ < *¯ pām ´ to assume that (non-ablauting) apām < *h2pom. ¯´ But of course, such a development would be unique for a laryngeal in word initial position, and so we should look for another explanation. On the ´ "(what are the goods) that basis of 1,181,1b adhvaryántā yád unninı̄thó apām ´ you bailed out from the waters by sacrificing", we could propose to read apām ´ as *pāam with a "Zerdehnung" typical for the genitive plural ending, here used ´ discussed above. In 1,181,1b metrically in a similar way as the one of avayā´ (and 2,19,3a; 7,101,2a) where a word final -o precedes, *pāam could well have ´ been substituted by apām because it was erroneously taken for an abhinihita sandhi variant. The substitution would then have been extended to all instances ´ ´ Note, however, that trisyllabic apãm of apām. is assumed two times in AC and once in JC by vNH (1,46,4a and 8,25,14a / 9,108,10c), thus producing another ´ type of irregular cadence (cp. utá na síndhur apãm). As the trisyllabic scansion seems necessary several times in the centerpiece of both T and J verses too (1,122,3b; 4c; 168,2c; 186,5c; 2,4,2a; 4,21,8b; 5,41,10b; 6,13,3d; 52,15b; ´ requires some further study. 10,49,2b; 93,5a77) the case of apām All in all, we arrive at a number of 100 cases (from 215) at least where irregular cadences of 11-syllable verses can be explained by assuming laryngeal residues. This is nearly one half, an astonishing amount even if a few examples remain doubtful. The remaining irregular cadences cannot be dealt with in extenso here; a short notice will suffice to demonstrate the problems involved. 1,62,3d; 2,19,1d; 5,33,5a; 6,47,31c; 7,3,5b: nára "men"; influenced by synonym jánā ? ´ and 3,18,4a: stutó, "passive" participles of stu2,19,5b; 20,5c; 6,24,8b: stavān "to praise"; cf. WATKINS, IGr III/1, 143 sq. (who does not account for the metrical problem, however); 6,67,11d yunájan, 10,12,4c áyan: subjunctive forms; * u .náǐ.’an, *á . ’an? 10,50,1b viśvābhúve compound "present to all": read viśva.ābhúve, without contraction of vowels in the compound seam, thus yielding a Jagatı̄ cadence? 1,121,1a: devayatā´ ; 1,122,11d ráthavate; 1,174,9a and 6,20,12a dhúnimatı̄ ; ´ ´ irregular full grade (-ante 50,11a vājavato; 9,96,4b b h até, 10,3,1c b h atā: etc.) instead of zero grade? 1,100,8c and 2,42,1d vidat inj.aor.; both verses with Jagatı̄ cadence (jyótir vidat / víśviyā vidat)? But cp. 3,29,7c viśvavídam acc.sg. "all-knowing"; 1,33,9a ródası̄ ubhé; 10,1,7a agna ubhé: both verses with Jagatı̄ cadence? 9,93,4a: imperative rada "scratch out"; restitute *rāda as in YAv. rādaiti? ŭ e ḣ ḣ ḣ ĭ ĭ ḣ ṙ ṁ ṙ 76 Cf. MAYRHOFER, "EWAia" I, 81 for further examples. To this list, 1,149,4c may be added at least from verses containing 10 syllables. For 1,61,12d, 7,34,15a, 10,46,1b and 10,46,2a, this would be possible too, but these hymns contain several other 10 syllable verses. 77 LARYNGEALS AND VEDIC 79 METRE 10,105,7c; 149,2c: rája nom.acc.sg.ntr. "darkness / atmosphere"; 3,31,20c; 6,24,10bc; 63,2c: ri á gen.abl. of the root noun "damage"; in 3,31,20c and 6,63,2c there is possibly a Jagatı̄ cadence (pāhi no ri ó , yāatho ri ó ) but not in 6,24 (pāhi ri ó ); ´ ´ 3,58,6b: jahnāvyām loc.sg.fem. "in the Jahnu clan"; Jagatı̄ cadence jahnāviyām? 8,46,17a jágmaye dat.sg. "the going one": different verse splitting should be envisaged, cf. OLDENBERG, "Noten" a.l.; 7,93,5c devayúbhi : substitute devayádbhi as in 10,69,8d? Further unique cases: 1,59,4a: sūnáve ródası̄; 1,104,3b: phénam udán; ´ 1,117,22b: práti airayatam; 1,121,15a: ví dasad; 1,126,1c: ámimı̄ta savān; ´ 1,173,11b: mánasā pariyán; 2,19,6a: sadíva sārathaye; 2,19,6c: ca náva; 2,30,6a: yá vanuthó; 4,13,3a: támase vip´ c e; 4,26,6b: mandrám mádam; 5,2,1d: níhitam arataú; 5,33,10c: sa v ára a sya ´ e r; 5,33,3b: yád ásan; 5,41,10d: ri āti vánā; 6,1,12c: b h at´ı̄r āré-aghā; 6,4,7c: śávasā devátā; 6,10,6a: ´ ´ agna uśán; 6,11,1c: mitrāváru ā nāsatiyā; 6,12,6c: yāsi duchúnā; 6,24,4b: ´ srutáya sa c ára ı̄ ; 6,24,7c: asya tanū; 6,40,5a: yád ´ d hag; 6,49,12d: ´ 6,68,2a: devátātā tujā; ´ vacanásya vípa ; 6,64,5b: jó a m ánu; 6,66,5a: cid ayā; 7,76,1d: víśvam u ā´ ; 7,103,10a: ajámāyur adāt; 8,46,28a: yá suvarā´ ; 9,89,3b: ´ asyá pátim; 9,94,1a: vājínı̄va śúbha; 9,94,1b: ná víśa ; 9,94,3a: kāviyā bhárate; 9,97,1b: ap k ta rásam; 9,97,32c: b h atá sudíva ; 10,3,6a: dad śānápaver; ´ ´ ´ 10,16,6b: śvāpada ; 10,22,4a: vãtasya dhúnı̄; 10,22,11a: indara dānāpnasa; 10,74,2d: k á vanta suvaí ; 10,79,5b: juhóti pú y ati; 10,95,4b: vá y ántig āh t; 10,110,5d: bhavata suprāya ā´ ; 10,164,3a: ni śásā abhiśásā. Note that some hymns occur several times in this list as, e.g., 9,94 with three irregular cadences. Many further investigations (into cadences of all verse types) are required before we can draw our conclusions from this fact. ṙ ḣ ḣ ṡ ṡ ṡ ṡ ḣ ḣ ḣ ṙ ṡ ṙ ṁ ṅ ṁ ṙ ṅ ṅ ṙ ḣ ṅ ṁ ṡ l̇ ḣ ḣ ḣ ḣ ṡ ḣ ṙ ḣ ḣ ṙ ṙ ḣ ṫ ṡ ṡ ḣ ḣ ḣ ṅ ṙ ṅ Jost Gippert Digital unterschrieben von Jost Gippert DN: cn=Jost Gippert, o=Universität Frankfurt, ou=Vergleichende Sprachwissenschaft, [email protected], c=DE Datum: 2011.12.28 18:01:38 +01'00'