RV 1.164.23-24 and Bhartrhari's philosophy of language

Jan E.M. Houben, “Rgveda 1.164.23-24 and Bhartrhari’s Philosophy of Language.” In: Expanding and Merging Horizons : Contributions to South Asian and Cross-Cultural Studies in Commemoration of Wilhelm Halbfass (ed. by K. Preisendanz), Wien : Verlag der Österreichischen Akademie der Wissenschaften, 2007, p. 711-719. With an *update* on p. 7. gveda 1.164.23-24 and Bharthari's philosophy of language* Jan E.M. Houben, Leiden and Paris 1.1 In a rich and insightful paragraph on Bharthari and the Veda, Wilhelm Halbfass remarked that "Bharthari does not draw a strict border between the uncreated Veda and the traditions of human thought and exegesis" (Halbfass 1991:37). Nevertheless, we find that there is a significant contrast between the way the grammarian and philosopher Bharthari deals with these two types of texts. He frequently supports his discussion with precise references to the traditions of human thought and exegesis – first of all the Pāṇinian grammatical tradition, but also various philosophical schools in his time (Mīmāṁsā, Vaiśeṣika, buddhist schools). However, it is only to exemplify grammatical points and not on account of the thoughts expressed that he gives quotations from Vedic texts.1 Bharthari's own work, as is well known, has a direct exegetical relationship with the Pāṇinian tradition: his Mahābhāṣya-Dīpikā (MBhD), to the extent it is available, is a running commentary on Patañjali's Vyākaraṇa-Mahābhāṣya, while his magnum opus the Vākyapadīya (VP) is a topical commentary on major philosophical issues in the same text of Patañjali. Apart from the occasions where the Veda is a transcendent "entity" nearly identical with Brahman,2 and apart from Vedic expressions cited only by way of grammatical example or illustration, are there any direct links with Vedic texts, and especially with the oldest and in several respects most important Vedic text, the gveda? 1.2 While direct references to the gveda, only to exemplify grammatical points, are rare in Bharthari's MBhD and VP,3 commentaries including the ancient Vtti do find reason to occasionally cite from the gveda in order to support an idea. It is not surprising that one of the verses cited is from the gvedic "Riddle Hymn" * The research on which this paper is based was supported by the KNAW (Royal Dutch Academy of Sciences) when the author was KNAW-research fellow at Leiden University. I profited much from Frits Staal's studies on the practice of Sāmavedic chanting, his collection of material in the form of films and recordings, and from a few email-exchanges with him on various problems. 1 That is, Vedic texts in the strict sense: the Saṁhitās and Brāhmaṇas of the various Śākhās (cf. the dictum mantra-brāhmaṇayor veda-nāmadheyam, ĀpŚS 24.1.31, and mantrāś ca brāhmaṇaṁ ca vedaḥ, Śabara on Mīmāṁsā-Sūtra 2.1.33). Such quotations are predominantly from Yajurvedic texts, and among these especially from the Maitrāyaṇīya-Saṁhitā: cf. Rau 1980 and Bronkhorst 1981, 1987. 2 VP 1.5ab: prāptyupāyo 'nukāraś ca tasya (viz., brahmaṇaḥ) vedo ... "Of this (Brahman) the Veda is the means of attainment and the image"; VP 1.172ab anādim anavacchinnāṁ śrutim āhur akartkām "It is said that the authorless Śruti (revealed text, i.e., the Veda) is beginningless and uninterrupted"; cf. also VP 1.173 avibhāgād vivttānām abhikhyā svapnavac chrutau "Those evolved from the undivided (i.e., the primeval ṣis evolved from brahman), (had) a perception with regard to the Śruti as in a dream." See further Houben 1997: 331-336 and cf. Aklujkar 1991. 3 Cf. V 10.85.33c cited in MBhDīp 1: 4.18 (ed. A.-L. 5.8); V 9.12.3c in MBhDīp 5: 18.17 (ed. A.-L. 176.21), 19.28 (ed. A.-L. 178.16); V 1.108.10ab in 6a: 27.3 (ed. A.-L. 214.20), V 8.75.9a in 6b: 19.6 (ed. A.-L. 240.14). J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 1 (V 1.164.454): "Speech" is a major theme in this hymn (cf. Brown 1968; Houben 2000) as it is in Bharthari's philosophy of language. On closer study it turns out that a few other enigmatic verses in this hymn express thoughts (712:) which are remarkably relevant to crucial ideas in Bharthari's philosophy of language. One of these ideas5 was of special importance to Bharthari and later grammarians, but it seems largely absent in the preceding Sanskrit tradition as far as available, especially in Patañjali's Mahābhāṣya and Kātyāyana's Vārtikas. 2.1 Among the verses in the "Riddle Hymn" connected with the theme of "speech" are verses 23-24. In the subdivision of the hymn based on ritual correspondences, these verses belong to the first of three parallel, alternative liturgies (cf. Houben 2000). Within this first liturgy they come towards the end of the section which accompanies the fanning of the pot, and just as the verse at the end of this section in the middle liturgy (1.164.39) and two verses in the third liturgy (1.164.45-46) it refers to the knowledge and application of (metrical) speech employed in the ritual. In the case of verses 23-24 (as well as the subsequent 25) the subject is apparently metrical speech employed in ritual songs or Sāmans. Verse 23 is as follows: yád gāyatré ádhi gāyatrám hitaṁ traíṣṭubhād vā traíṣṭubhaṁ nirátakṣata / yád vā jágaj jágaty hitaṁ padáṁ yá ít tád vidús té amtatvám ānaśuḥ // An approximate translation does not seem very difficult. 6 The statement must be in some way paradoxical or enigmatic but it is not precisely clear how. Interpreters have remained relatively silent about it, and so far did not present a convincing interpretation.7 Geldner (1951: 231f note) observes: "Hier steigt zunächst der Dichter von den Höhen der bisherigen Spekulation in die Niederungen der dichterischen Technik und der rituellen Praxis hinab. Er geht von den metrischen Elementen aus. Das Paradoxon liegt wohl darin, daß das metrische Element (das 4 This famous verse on the four padás of speech, and on the availability of only one part of speech to normal men, is cited in the Vtti on VP 1.159 (VP I:221.1-2) which mentions three forms of speech, vaikharī (the elaborate, explicit one; cf. Mayrhofer 1996: 587 for its possible etymology via Prakrit), madhyamā (the intermediate one), and paśyantī (the "seeing" or visionary one). 5 An important idea expressed in V 1.164 and of pervading relevance for Bharthari's philosophy which I do not discuss here is "unity in diversity": V 1.164.46c ékaṁ sád víprā bahudh vadanti "One reality that is – the inspired ones speak of it as many" may be compared with VP 3.3.87cd eko 'rthaḥ śabdavācyatve bahurūpaḥ prakāśate "the one Thing-meant (i.e., the ultimate object of linguistic expressions) manifests itself as manifold when it is expressed by words." 6 We may take Geldner's translation as starting point: "Nur die haben die Unsterblichkeit erlangt, die wissen, daß der Gāyatrī-(fuß) auf dem Gāyatrī(lied) beruht, oder daß der Triṣṭubh(fuß) aus dem Triṣṭubh(lied) herausgebildet wurde, oder daß der Jagatfuß auf dem Jagat(lied) beruht." 7 Renou 1967, in his notes on V 1.164, has nothing to remark on verses 23-24 (no translation is given in his Études Védiques et Pāṇinéennes); Oldenberg 1909: 158f discusses some matters of sandhi in 23 and 24 and remains doubtful on a proposed emendation in 24 because of the "Unbestimmtheit des Sinnes." J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 2 padám oder die Reihe) auf dem Lied beruhen soll statt umgekehrt. Doch bleibt der tiefere Sinn verborgen." 2.2 In the group of verses 23-25, several terms clearly point in the direction of Sāmavedic chanting: jágat, rathantará and gāyatrám.8 If we concentrate on verse 23, here too the terms gāyatrám and jágat can be very well connected with Sāmavedic chanting. The same applies to (713:) the term tráiṣṭubham, even though it is not equally well known as name of a Sāman.9 But all three terms may just as well refer to a line or verse in the gāyatr resp. the tríṣṭubh or jágatī metre (which may provide the yóni or underlying text of a Sāman), and the use of the term padám in connection with jágat in 23c suggests indeed that this metrical aspect is intended rather than the melodious song.10 Unfortunately, we do not have any precise knowledge of the practice of chanting at the time of the gveda, nor do we know what the precise relation was between a Sāman and its underlying verse, usually from the gveda.11 In the classical system, in any case, "[T]he connection between 8 For an example of the application of Gāyatra and Rathantara Sāmans cf. Parpola 1969: 13-14; Staal 1983, vol. I: 538. Gāyatra as name of a Sāman is well known (cf. Howard 1977: 514f). Also Rathantara is well known (Howard 1977:534). In the classical system Jagat as name of a Sāman seems to refer to any Sāman based on a verse in Jagatī metre: cf. jagatsāman "having the Jagat as Sāman" in ĀpŚS 12.14.1 next to rathantarasāman and bhatsāman. As such it is equivalent to Jāgata Sāman. R. Simon's index to PS lists the Jāgata-Varuṇa Sāman, LŚS 7.3.11, 7.10.12 know in addition of a Jāgata-Soma Sāman. Cf. further ÆaèvBr 1.4.12, and ŚB 4.2.5.20 with Eggeling's notes on the employment of metrical verses in the three Soma-pressings (morning, noon and third). 9 The index of R. Simon's edition of the Puṣpasūtra does list Traiṣṭubha (occurring only in some of the commentaries on the PS), Traiṣṭubha-Vātsapra (variant name in some mss. for Vātsapra) and Traiṣṭubha-Śyāvāśva. It was demonstrated long ago by Oldenberg (1884) that the Triṣṭubh (as well as its extended version, the Jagatī) is the preferred metre in hymns intended for recitation (by the Hot and his group), whereas Gāyatrī and Pragātha are the main metres for chant (by the Udgāt and his group). 10 The relation between gāyatrá etc. as metrical term (as in the frequent Yajurvedic expression gāyatréṇa chandasā ... etc.) and gāyatrá as name of a Sāman is complex. The famous Gāyatra Sāman (melody, way of singing) is indeed based on a verse in gāyatrī metre (V 3.62.10) (cf. Howard 1983, 1987), but the same melody may be applied to other verses or parts of other verses as well (cf. LŚS 1.8.9 where the Gāyatra way of singing is applied to a verse in the Paṅkti metre). In the case of tráiṣṭubham a linguistic derivation from tríṣṭubh seems likely; in the case of gāyatrám the historical derivation is rather 'song' directly from gā 'to sing', with gāyatr as the feminine indicating the metre. If gāyatrám, tráiṣṭubham and jágat appear together, however, a reference to the three major gvedic metres gāyatr, tríṣṭubh and jágatī seems natural, even if there is a Sāmavedic context. Cf. SV 2.9.2.7.2-3 sung at the beginning of the Soma-pressing (cf. Caland & Henry 1906: 134): 3 1 2r 3 1 2 3 1 2 3 1 2 3 1 2r 3 1 2 yuñje vāca śatapadīṁ gāye sahasravartani / gāyatraṁ traiṣṭubhaṁ jagat // 2 // 3 1 2r 3 2 3 1 23 2 3 1 2 3 1 2r 3 2 gāyatraṁ traiṣṭubhaṁ jagat / viśvā rūpāṇi saṁbhtā / devā okāsi cakrire // 3 // 11 On the other hand, it is known and accepted that the authors of the gveda were familiar with the practice of Sāman-chanting, and that there was also an intimate relation between the texts of the gveda and this practice. Apart from the mentioned names of specific Sāmans, the hymns contain numerous references to the sman 'song', and the priest specialized in singing, the Udgāt, has been mentioned (V 2.43.2) and was referred to (V 10.71.11). A considerable number of hymns in the gveda was apparently from the outset intended to be sung. Cf. Oldenberg 1884: 441 (Kleine J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 3 a verse and its melody is a loose and mechanical one, e.g. the caesura divides in the prosody the triṣṭubh into two parts of 5 and 6, or 4 and 7 syllables, but the sāman- techniques never care about this essential metrical construction and generally divide the triṣṭubh into three parvans" (van der Hoogt 1929: 51).12 We thus seem justified in distinguishing, also in the time of the gveda, the metrical aspect of the line as it appears in a recitation (with a pattern of heavy and light syllables) and the line as employed in chanting. 2.3 Against this background we can interpret verse 23 on the assumption that the terms refer to the metrical aspect of text-parts underlying Sāmans. The verse says that something is based on something, and something is created from something. More precisely, pāda a of 23 informs us that "something of Gāyatra character" is based on "something of Gāyatra character," and pāda b that "something of Traiṣṭubha-character" is created from "something (714:) of Traiṣṭubha-character." Initially, however, it is not clear exactly what is based on what, and what is created from what. From the statement in pāda c, which is just a bit more explicit, we understand that it is the padá 'foot' or 'metrical line' (namely a padá 'foot' or metrical line of Jagat-character) that is based on something (namely, on something else having a Jagat-character). The term padá 'line' is already in the gveda used with reference to the metrical lines which make up versified speech (cf. e.g. 1.164.45 catvri vk párimitā padni). Taking into account the structural parallelism in pādas a, b and c, we may fill in this notion of padá 'line' at the corresponding open places in pāda a and b. In pāda b, we then arrive at the statement that a metrical line (padá), namely one of Traiṣṭubha-character, is created from something else, still undefined, but in any case having a Traiṣṭubha-character. In pāda a we arrive at the statement that a metrical line (padá), namely one of Gāyatra-character, is based on something else, still undefined, but in any case having a Gāyatra character. In line with our preceding considerations, we may assume that "to have a Gāyatra character" amounts to "to be in the Gāyatrī metre."13 It is to be noted that the metrical line is the smallest unit that can still have a Gāyatra character. Hence, we may infer that the other thing of Gāyatra character is another, larger unit, e.g., a verse, hymn or song. Because neither this verse nor 24-25 refers to the unit of the verse (c), but references are made to larger units (arká 'song of praise', vāká 'recitation'), it is preferable to take 'song' or 'hymn' as the unexpressed unit in the remaining places in 23a, b and c. With this we have solved the "riddle" of verse 23 in the sense that we have inferred which units remained unexpressed. But the resulting statement may still be regarded as enigmatic. The verse now says that in Schriften vol. 1: 515): "Wäre eine Ausdrucksweise erlaubt, bei welcher der Unterschied von Sāman, d.h. Sangweisen, und Sāman-Texten ignorirt wird, könnte man geradezu sagen: der Rigveda ist zugleich der älteste Sāmaveda." 12 While the Triṣṭubh would have been preferred for recitation rather than for singing according to the argument of Oldenberg (1884, 1915), van der Hoogt (1929: 51 note 2) points out that this threefold structure in Sāmavedic application suits the name Triṣṭubh better than the twofold structure in metrical recitation. 13 That is, minimally it will refer to the eight-syllabic Gāyatrī-sequence, which normally forms part of a verse of three lines. J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 4 the case of Gāyatrī-lines and -hymns, Triṣṭubh-lines and -hymns, and Jagatī-lines and -hymns, it is the smaller unit, the line, which is based on, or fashioned out of, the larger unit – contrary to what one would expect from a common sense point of view. The underlying idea has no doubt to do with an experience at the basis of the inspired creation of a hymn or chant. Poets reflecting on this process (cf. in general Gonda 1963) occasionally speak of a "milking" which suggests a complete object, the hymn or Sāman, given in advance, albeit in an indiscriminate, vague form.14 TĀ 5.10 lets mantras and songs be "milked" from Prajāpati or the sun. The subsequent differentiation into smaller units is expressed by the verb vi-k (TĀ 5.10.1: tád agnír vy ákarot), elsewhere with vy-ā-k (TS 6.4.7.3; cf. Thieme 1983 on vyākaraṇa). The gvedic poet's creation of hymns has also been referred to as a process of "seeing",15 which likewise suggests a complete object given in advance. 2.3 Those familiar with Indian linguistic philosophy and grammar will recognise in the above formulation the position for which the 5th century philosopher- grammarian Bharthari became famous: sentences or larger units rather than words or smaller units are primary.16 To state it with more precision – since a sentence may consist of a single word, a word of a sin(715:)gle phoneme – the units of a higher organizational order are primary, those of a lower organizational order are secondary. Bharthari did not claim originality for this view, but both for him and for the author of the Vtti it has been very difficult to find convincing authoritative antecedents for it in the work of those claimed as predecessors, especially Patañjali and Pāṇini. Pāṇini gives no explicit definition of the sentence at all, but we find two definitions in Patañjali's Mahābhāṣya, viz., ekatiṅ vākyam (deriving from Kātyāyana) and ākhyātaṁ sāvyaya-kāraka-viśeṣaṇam vākyam (MBh 1:367.10-17). Both definitions point to words as the basic units which together form a sentence. Bharthari can further refer to a more theoretic statement implying the primacy of the word, and this statement occurs a few times in the Mahābhāṣya (yad atrādhikyaṁ vākyārthaḥ saḥ, MBh 1:462.4, 1:464.10-12). But neither Bharthari nor the author of the Vtti can cite an equivalent authoritative statement from the Mahābhāṣya which would indicate the primary status of the sentence. Bharthari refers to the Mahābhāṣya statement on the primacy of the continuous version (saṁhitā-pāṭha) of Vedic texts over the word-by-word version (pada-pāṭha) (VP 2.58-59), and he refers to a rather obscure thinker mentioned in the Nirukta, viz., Audumbarāyaṇa (VP 2.344, cf. Nir 1.1-2). If the author of the Vtti comes to 14 According to V 5.44.13 sutambharó yájamānasya sátpatir víśvāsām dhaḥ sá dhiym udáñcanaḥ Sutambhara (name of a seer in Sāyaṇa's explanation which is accepted by Geldner) ‘milks’ the udder of all dh, which is both abstract ‘vision’ and its concrete expression in a verse or hymn. 15 Cf. ṣibhir mantradgbhiḥ in the opening verse of the gvidhāna. 16 Since linguistics includes semantics I refrain from using an expression such as "linguistic- semantic unit." It is to be noted, however, that in Bharthari's arguments the semantics of the units under discussion is always of crucial importance, even where the focus is on the linguistic form rather than the meaning it is supposed to have. J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 5 referring to traditional support for the sentence as basic unit (Vtti on VP 1.24-26) he cites only from the lost pre-Patañjali work the Saṁgraha.17 The citation is a verse that appropriately emphasizes the dependence of the word on the sentence in its semantic aspect, in other words, the sentence meaning: na hi kiñ cid padaṁ nāma rūpeṇa niyataṁ kva cit / padānāṁ rūpam artho vā vākyārthād eva jāyate // Nowhere, indeed, is any word fixed as to its form; form and meaning of words arise from the meaning of the sentence only. 2.4 Taking into account the conspicuous absence of references to the view that the sentence is primary in older Pāṇinian works, the view which we find expressed in V 1.164.23 resonates remarkably well with Bharthari's theoretical preference. This gvedic verse is hence a perfect candidate for a direct source of inspiration for Bharthari's position. Otherwise, the verse may very well have had its influence on works of grammarians and other language-oriented thinkers now lost to us, such as the author of the Saṁgraha, and hence have indirectly supported or confirmed Bharthari in accepting the sentence as the main linguistic unit. While we should not exclude the possibility that contemporaneous philosophical discussion may have stimulated Bharthari to search theoretical solutions to language- philosophical problems in a certain direction, 18 Bharthari's own self-presentation as someone continuing the tradition of Vedic and Brahmanical grammarians is to be taken seriously. The verse may in any case be translated as follows (the crucial word that solves the "riddle" in capitals): 23. That the Gāyatrī(-line) is based on the Gāyatrī (-hymn) (smaller on the larger unit, rather than the other way round), and that the Triṣṭubh(-line) is fashioned out of the Triṣṭubh(-hymn), and that the Jagatī-LINE is based on the Jagatī (-hymn): only those who know this have attained immortality. Our translation is finally not that much different from Geldner's; but the latter's solution of the "riddle" remained a good guess, no arguments were adduced, and he admitted not to have (716:) any idea about the intention behind the enigmatic statement. His "Lied" could refer to several aspects of a ritual utterance. Here we assume that the other unit which remains unexpressed is first of all a unit of metric linguistic utterance – its melodious character is not in focus. 3.1 In verse 24 we find an explicit reference to a sma or chant (in pāda b). The verse may be taken as a discussion of, again, sacred metrical speech in the first place. The word sma in pāda b may refer to the song as a whole, i.e., to the text of the song to which the specific melody is applied. In 24 it is said that several 17 Bharthari's preferred theory of the sentence as linguistic unit has received considerable attention (cf. Cardona 1976: 300-302), but the contours of its originality were for the first time systematically explored in Houben 1993, 1995, and 1999. 18 Cf. Bronkhorst 1998 arguing that developments in Buddhist thought were crucial for Bharthari's view of the sentence as primary linguistic unit. J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 6 things are "made according to" several other things (práti mimīte, mimate, plus word in instr.). As in 23, it is not immediately clear which things have a specific relation to which other things. In 23 a solution was found by applying the one explicit reference to a unit of speech (the padá or LINE) towards the end of the statement, to structurally parallel previous parts of the statement. In 24 we find explicit mention of two units of speech towards the end of the statement: seven vṇī or 'voices' are made in accordance with the syllable (akṣára). In contrast with the syllable, the vṇī or 'voice' is likely to be a larger unit, and I propose to take it as the complete recitation (set of recitations) of one priest.19 The verse then states the dependence of the larger unit upon the smaller one, and presents a viewpoint opposite or complementary to the view expressed in 23. This view, too, can be linked to the way poets have seen the process of their poetic creation: the metaphors employed include the construction of a chariot by the poets, 20 which would imply the piecing together of parts given in advance. The verse and its translation are as follows (crucial words that solve the "riddle" in capitals): 24. gāyatréṇa práti mimīte arkám arkéṇa sma traiṣṭubhena vākám / vākéna vākáṁ dvipádā cátuṣpadā- -akṣáreṇa mimate saptá vṇīḥ // According to the Gāyatrī (-line) one makes the song of praise (arká); according to the song of praise a chant (sma), according to the Triṣṭubh (-line) the recitation; according to the two- and four-lined recitation (again a larger) recitation; according to the SYLLABLE they make the seven VOICES. *update* 28-7-2019: With “According to” I revert to my 2000 translation of this verse, taking the verb in the first line, práti mimīte ‘make in accordance with; copy’ (Grassmann, 1875, p. 1023-1024: práti mā, wonach [I.] ein Lied [A.] bilden) to be valid in the two subsequent lines, and even in the last line (simplex pro composito) where the verb is explicitly given as mimate (viz. práti), but the construction with instrumental is continued from the previous lines. The "seven voices" (vṇī, cf. Mayrhofer 1996 s.v. vāṇá: "nicht geklärt") are not infrequently referred to in the gveda.21 Still, a precise identification of these "voices" is difficult: interpretations vary from "rivers" (Sāy. on V 3.1.6 and 19 Cf. Monier-Williams 1899 s.v. vṇī: speech, language, words; literary composition. The "seven voices" or complete sets of utterances will belong to the seven main priests of the gvedic ritual, referred to elsewhere in the present hymn, viz. in verses 2 and 3. One may also compare the seven hotrāḥ or vaṣaṭkartāraḥ of the Śrauta ritual, cf. Caland and Henry 1906: 3. "Seven voices" are also referred to in V 3.1.6, 3.7.1, 9.103.3, 8.59.3 (Vālakhilya). 20 Cf. Gonda 1963: 110-111, and, in addition to the places discussed there, V 5.73.10 im bráhmāṇi ... y tákṣāma ráthā iva. 21 Cf. V 3.1.6d ékaṁ gárbhaṁ dadhire saptá vṇīḥ; 3.7.1b  mātárā viviśuḥ saptá vṇīḥ; 8.59.3b mádhva ūrmíṁ duhate saptá vṇīḥ; 9.103.3c abhí vṇīḥ ṣīṇāṁ saptá nūṣata . J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 7 3.7.1) to "metres" (Sāy. on V 1.164.24).22 In the context of the present hymn, as in the context of V 9.103.3 and the similar expression in 9.104.4 (abhí vṇīr anūṣata), the voices can be taken as the seven complete sets of utterances, including recitations and chants, one set for each of the seven main priests (717:) or the seven primeval priest-seers of a gvedic ritual. The seven main priests are referred to elsewhere in V 1.164,23 whereas 9.103.3 speaks of the seven voices of the seers (vṇīḥ ṣīṇāṁ saptá). 3.2 We thus see that V 1.164.23-24, two enigmatic statements in the "Riddle Hymn," express two complementary viewpoints on the relationship between smaller and larger units of metrical speech employed in ritual chanting. As such, they provide antecedents for two complementary views which play a major role in Bharthari's Vākyapadīya, one according to which the units of a lower organizational level (especially the word and its meaning) are primary, the other according to which units of a higher organizational level (especially the sentence and its meaning) are primary. Especially the latter view so far seemed to have had hardly any predecessors in the Sanskrit tradition, although Bharthari did not claim originality for it. 22 Cf. Grassmann 1875 on saptá vṇīḥ s.v. vṇīḥ: "die sieben Stimmen des Soma, die sieben rauschenden Somagüsse, persönlich gefasst auch mit sieben Stimmen der Sänger verglichen ... die sieben Tonweisen oder Liedformen." 23 Cf. V 1.164.2ab saptá yuñjanti rátham ékacakram éko áśvo vahati saptánāmā "Seven (priest-seers) yoke (employ in their sacrifice) the one-wheeled chariot (the year); one horse (the sun) with seven names draws it." In pādas cd of the next verse "the seven sisters" may very well refer to the same seven voices vṇīḥ (fem. plur.) of the priests which we have in 24: saptá svásāro abhí sáṁ navante yátra gávāṁ níhitā saptá nma"Seven sisters (the seven voices of the seven priests?), in whom the seven names of the cows are deposited, collectively sing songs of praise." J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 8 1) Primary sources ĀpŚS Āpastamba-Śrauta-Sūtra, ed. R. Garbe (Bibl. Indica 92), Calcutta 1882- 1902. LŚS Lāṭyāyana-Śrauta-Sūtra, ed. and tr. H.R. Ranade, Delhi 1998. MBh Patañjali's Vyākaraṇa-Mahābhāṣya, ed. F. Kielhorn, third ed. rev. by K.V. Abhyankar, Pune 1962-1972: ref. to volume, page, line. MBhD Bharthari's Mahābhāṣya-Dīpikā, ed. J. Bronkhorst, G.B. Palsule et al., Pune 1985-1991: ref. to Āhnika, page and line; ed. Abhyankar and Limaye, Pune 1970: ref. to page and line. Nir Yāska's Nirukta, ed. and tr. L. Sarup, 1920-1927; reprint Delhi, 1967. PS Puṣpa-Sūtra, ed. and tr. R. Simon, München 1909. V gveda-saṁhitā, ed. F. Müller, Third ed. Varanasi 1965. ṢaḍvB Ṣaḍviṁśa-Brāhmaṇa, ed. B.R. Sharma, Tirupati 1983. ŚB Śatapatha-Brāhmaṇa (Mādhy.): ed. Weber, London 1855. SV Sāmavedārcika, ed. Th. Benfey, Leipzig 1848. VP Bharthari's Vākyapadīya, ed. W. Rau, Wiesbaden 1977. (718:) 2) Secondary sources Aklujkar, Ashok N. 1991 "Bharthari's concept of the Veda." Pāṇini and the Veda. (Panels of the VIIth World Sanskrit Conference: Kern Institute, Leiden: August 23-29, 1987, Vol. V, ed. by M. Deshpande): 1-18. Leiden: E.J. Brill. Bronkhorst, Johannes 1981 "On some Vedic quotations in Bharthari's works." Studien zur Indologie und Iranistik 7: 173-175. 1987 "Further remarks on Bharthari's Vedic affiliation." Studies in Indian Culture (S. Ramachandra Rao Felicitation Volume): 216-223. Bangalore: Prof. Ramachandra Rao Felicitation Committee. 1998 "Les éléments linguistiques porteurs du sens dans la tradition grammaticale du sanskrit." Histoire – Épistémologie – Language: Journal published by the SHSL and the PUV (Paris), Tome 20, fasc. 1: 29-38. Brown, W. Norman 1968 "Agni, Sun, Sacrifice, and Vāc: A Sacerdotal Ode by Dīrghatamas (Rig Veda 1.164)." Journal of the American Oriental Society 88: 199-218. Caland, Wilhelm, and Henry, Victor 1906-07 L'Agniṣṭoma: description compl`ete de la forme normale du sacrifice de Soma dans le culte védique. Tome premier - Tome second. Paris: E. Leroux. Cardona, George 1976 Pāṇini: A Survey of Research. The Hague: Mouton. J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 9 Geldner, K.F. 1951 Der Rig-Veda aus dem Sanskrit ins Deutsche übersetzt und mit einem laufenden Kommentar versehen. Harvard Oriental Series vols. 33-35. Cambridge, Mass.: Harvard University Press. Gonda, Jan 1963 The Vision of the Vedic Poets. The Hague: Mouton & Co. Grassmann, Hermann 1875 Wörterbuch zum Rig-Veda. Leipzig: Brockhaus. Halbfass, Wilhelm 1991 Tradition and Reflection: Explorations in Indian Thought. Albany: State University of New York Press. Hoogt, van der 1929 The Vedic Chant, Studied in its textual and melodic form. Dissertation Amsterdam. Wageningen: Veenman & Sons. Houben, Jan E.M. 1993 "Who are Bharthari's padadarśins? On the development of Bharthari's philosophy of language." Études Asiatiques / Asiatische Studien 47.1 (Proceedings of the World Conference on Bharthari, Pune, 1992): 155-169. 1995 "Bharthari's Perspectivism (2): Bharthari on the Primary Unit of Language." History and Rationality: the Skoevde Papers in the Historiography of Linguistics (Acta Universitatis Skovdensis, Series Linguistica, Vol. 1. Ed. by Klaus D. Dutz and Kjell-Ake Forsgren): 29-62. Münster: Nodus Publicationen. (719:) 1997 "Bharthari's Perspectivism (1): The Vtti and Bhartrhari's perspectivism in the first Kāṇèa of the Vākyapadīya." Beyond Orientalism: the impact of the work of W. Halbfass on Indian and cross-cultural studies (ed. by K. Preisendanz and E. Franco): 317-358. Amsterdam: Rodopi, 1997. 1999 "The Theoretical Positions of Bharthari and the Respectable Grammarian." Rivista degli Studi Orientali, vol. LXXII, Fasc. 1-4 (Roma, 1998): 101-142. 2000 "The Ritual Pragmatics of a Vedic Hymn: The 'Riddle Hymn' and the Pravargya Ritual." Journal of the American Oriental Society 120: 499-536. Howard, Wayne 1977 Sāmavedic Chant. New Haven and London: Yale Univ. Press. 1983 "The music of Nambudiri unexpressed chant (aniruktagāna)." In: Staal 1983 vol. II: 311-342. 1987 "The body of the bodiless Gāyatra." Indo-Iranian Journal 30: 161-173. Mayrhofer, Manfred J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 10 1996 Etymologisches Wörterbuch des Altindoarischen. II. Band. Heidelberg: Universitätsverl. C. Winter. Monier-Williams, M. 1899 Sanskrit-English Dictionary. Oxford: Clarendon Press. Oldenberg, Herman 1884 "gveda-Saṁhitā und Sāmavedārcika. Nebst Bemerkungen über die Zerlegung der gveda-Hymnen in Theilhymnen und Strophen sowie über einige andere Fragen." Zeitschrift der deutschen morgenländischen Gesellschaft 38: 439-480. [= Oldenberg 1967: 513-554.] 1909 gveda: textkritische und exegetische Noten. Bd. I: Erstes bis sechstes Buch. Abh. der Königlichen Gesellschaft der Wissensch. zu Göttingen, Phil.-hist. Klasse, N.F., Bd. 11, no. 5. Berlin: Weidmannsche Buchhandlung. 1915 "Zur Geschichte der Triṣṭubh. Mit einem Exkurs: Zur Behandlung des auslautenden –i und –u im gveda." Nachrichten v.d. Kgl. Gesellschaft der Wissenschaften zu Göttingen, aus dem Jahre 1915: 490-543. [= Oldenberg 1967: 1216-1269.] 1967 Hermann Oldenberg: Kleine Schriften. 2 Teile. (Herausg. v. Klaus L. Janert.) Wiesbaden: Franz Steiner. Parpola, Asko 1969 The Śrautasūtras of Lāṭyāyana and Drāhyāyaṇa and their commentaries. An English translation and study. Vol. I: 2: The agniṣṭoma (LŚA I-II, DŚS I- VI). Helsinki: Helsingfors. Rau, Wilhelm 1980 "Bharthari und der Veda." Studien zur Indologie und Iranistik 5/6: 167-180. Renou, Louis 1967 Études Védiques et Pāṇinéennes. Tome XVI. Paris: E. de Boccard. Staal, Frits (ed.) 1983 Agni: The Vedic Ritual of the Fire Altar. Volume I-II. Berkeley: Asian Humanities Press. Thieme, Paul 1983 "Meaning and form of the 'grammar' of Pāṇini." Studien zur Indologie und Iranistik 8/9: 3-34. J.E.M. Houben, RV 1.164.23-24 and Bhartrhari 11