World Sanskrit Conference (WSC) (17th : 2018)

On the question of circularity in A 1.3.3 hal antyam Bonino, Beatrice 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


70440-Bonino_Beatrice_Question_circularity_2019.pdf [ 722.72kB ]
JSON: 70440-1.0380178.json
JSON-LD: 70440-1.0380178-ld.json
RDF/XML (Pretty): 70440-1.0380178-rdf.xml
RDF/JSON: 70440-1.0380178-rdf.json
Turtle: 70440-1.0380178-turtle.txt
N-Triples: 70440-1.0380178-rdf-ntriples.txt
Original Record: 70440-1.0380178-source.json
Full Text

Full Text

 On the Question of Circularity in A 1.3.3 hal antyam  Beatrice Bonino Proceedings of the 17th World Sanskrit Conference, Vancouver, Canada, July 9-13, 2018, Section 3: Vyākaraṇa.  Section Convenors: Malhar Kulkarni and Peter Scharf
General Editor: Adheesh Sathaye Published by the Department of Asian Studies, University of British Columbia, on behalf of the International Association for Sanskrit Studies. DOI: 10.14288/1.0380178.
URI: Suggested Citation Format: MLA:
Bonino, Beatrice. “On the Question of Circularity in A 1.3.3 hal antyam.” Proceedings of the 17th World Sanskrit Conference, Vancouver, Canada, July 9-13, 2018, Section 3: Vyākaraṇa. Edited by Malhar Kulkarni and Peter Scharf, 2019. DOI: 10.14288/1.0380178. APA:
Bonino, B. (2019). On the question of circularity in A 1.3.3 hal antyam. In M. Kulkarni & P. Scharf (eds.) Proceedings of the 17th World Sanskrit Conference, Vancouver, Canada, July 9-13, 2018, Section 3: Vyākaraṇa. DOI: 10.14288/1.0380178. Chicago:
Bonino, Beatrice. 2019. “On the Question of Circularity in A 1.3.3 hal antyam.” In Proceedings of the 17th World Sanskrit Conference, Vancouver, Canada, July 9-13, 2018, Section 3: Vyākaraṇa, edited by Malhar Kulkarni and Peter Scharf. DOI: 10.14288/1.0380178. Proceedings of the 17th World Sanskrit Conference, July 9-13, 2018 University of British Columbia, Vancouver, CanadaCopyright © 2019 by the author. Content is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license (CC BY-NC-ND 4.0).वैधुसव ्मकबुंटुकअ ारा यसं तृा यनसमवायःINTERNATIONAL ASSOCIATION OF SANSKRIT STUDIESTHE   17TH    WORLD   SANSKRIT  CONFERENCEVANCOUVER, CANADA • JULY 9-13, 2018 THE 17TH WORLD SANSKRIT CONFERENCE, VANCOUVER, CANADA, JULY 9-13, 2018 On the Question of Circularity in A 1.3.3 hal antyam Beatrice Bonino Paris 3 Sorbonne Nouvelle,  Paris, France. Abstract The fault of mutual dependence, according to the grammatical tradition, starting with the Kāśikāvṛtti, affects Pāṇini’s rule A 1.3.3 hal antyam. Candotti (2006, 2016) exhaustively treated the cases of mutual dependence that involve denomination in the Pāṇinian grammar, namely the rules that prescribe an element by means of its name (saṃjñāsūtra), since they present this element as something not existing, and yet they treat it as already existing by the very fact that they use the same name to name it. However, the particular case of the itaretarāśraya concerning A 1.3.3 hal antyam and A 1.1.71 ādir antyena sahetā doesn’t figure among those. On the other hand, it is briefly mentioned in other important grammatical works (Renou 1942, Cardona 1976). However, even if Kātyāyana and Patañjali seem to recognize the logical fallacy of itaretarāśraya in the present rule, they never explicitly employ this term to describe it, as they normally do in a context when such an error is in fact identified. In the present text, I will retrace the general features of this flaw, trans-late the main passages treating the mutual dependence and analyse the possible reasons behind the absence of the expected terminology for the defect affecting A 1.3.3. Keywords: itaretarāśrayadoṣa, mutual dependence, circularity, Pāṇini, hal antyam, Patañjali In this paper I shall try to retrace the process of attribution of a fault to Pāṇini by his commentators, a practice which has either been considered as legitimate by scholars who have accepted these traditional reparsings or, on the other hand, as post-Pāṇinian – if not un-Pāṇinian. In particular, I will focus on the logical falla-cy of mutual dependence, the itaretarāśrayadoṣa, affecting the rule A 1.3.3 hal antyam “the final consonant [is an it]”. Proceedings of the 17th World Sanskrit Conference, Vancouver, Canada, July 9-13 2018, Section 3: Vyākaraṇa, edited by Malhar Kulkarni and Peter Scharf, 2019. DOI: 10.14288/1.0380178. BONINO 2Maria Piera Candotti, in Interprétations du discours métalinguistique (137-145), distinguishes trivial cases of this error from others which are more complex and require an ad hoc solution. Concerning the former, this flaw often arises, as pointed out by Candotti (2006: 139), on a metalinguistic level, when a rule that enjoins an element by means of its name (saṃjñā), on the one hand, presents this element as something which did not exist before, while still treating it as already existing by the very fact that it uses the same name to name it; in other words, as Candotti (2016: 109) states, “when the name plays the role of vidheya and enjoins that same denotatum that it names.” The classic example of this first variant of itaretarāśrayadoṣa is A 1.1.1 vṛddhir ādaic, where the attribution of the technical term vṛddhi applies to non-existing elements, namely the phonemes ā, ai and au, while they depend on the technical term itself.  The triviality of this first variant 1consists in the fact that a single answer can be used for all the cases affected by this circular reasoning: the nityaśabdatva,  the permanency of linguistic elements 2“intimately linked with the themes of the precedence of linguistic usage over grammar and of the restrictive function of the latter,” is given as the solution to this problem. This is not the case when one of these serious variants of itare-tarāśrayadoṣa is examined. The circularity involved in these cases, “unlike what happened in the trivial cases, develops within the grammatical system itself,”  3and concerns the application of two rules. Moreover, in the case of a non-trivial variant of this fault, grammarians never have recourse to the permanence of lin-guistic forms to solve the difficulty, nor to the notion of bhāvinī saṃjñā, names that do not denote objects in the context described by the rule, but objects that will become indicated by the name only at the end of the process described by the rule in question. Candotti lists eight occurrences found in the Vārttikas of this type of mutual dependence.   4Now, if one looks through the Vārttikas of Kātyāyana and the commentary of Patañjali, it can be observed that each time the two grammarians detect circu-lar reasoning, whether trivial or non-trivial, they methodically employ the term  Vt. 8 ad A 1.1.1: sato vṛddhyādiṣu saṃjñābhāvāt tadāśraya itaretarāśrayatvād aprasiddhiḥ. 1 Vt. 9 ad A 1.1.1: siddhaṃ tu nityaśabdatvāt. 2 See Candotti 2016: 110.3 See Candotti 2016: 110-112.4 On the Question of Circularity in A 1.3.3 hal antyam 3itaretarāśraya.  This consideration finds its reason in the fact that one case seems 5to elude this recurrence: when one attempts to redraw the history of this fault pertaining to A 1.3.3 hal antyam in the grammatical tradition, one is confronted with the absence of this term until the Kāśikāvṛtti and Kaiyaṭa’s Pradīpa, who di-rectly mention the mutual dependence while commenting on vt. 3 (itare-tarāśrayaṃ manyate). After this point, all the commentators will endeavour to find a solution to the established circularity of this Pāṇinian rule. The saṃjñāsū-tra A 1.3.3, being one exception to the uniformity observed in the Vārttikas and in the Mahābhāṣya, was the subject of the traditional debate at least until Nāgeśa. So, why wouldn’t Kātyāyana and Patañjali employ the term itaretarāśraya, for the sake of consistency, in the discussion relating to A 1.3.3? While no one else after them hesitated in identifying a circularity between A 1.3.3’s attribution of the technical term it to the last consonant of a grammatical element and the forma-tion of the pratyāhāra by A 1.1.71 ādir antyena sahetā, although they seem to rec-ognize such a flaw in this case, they never use the term itaretarāśrayadoṣa, even though all the three conditions for considering it as a non-trivial variant of itare-tarāśrayadoṣa are there: the circularity is internal to the grammatical system, in-volves two rules, and it can’t be solved by resorting to the permanence of linguis-tic forms. As a matter of fact, it has already been pointed out that the authors of the Kāśikāvṛtti are the first to state the logical vice of circularity affecting the rela-tion between these two rules, somehow taking for granted that Kātyāyana had in mind an itaretarāśraya, namely that “the attribution of the technical term it to l depends on the pratyāhāra hal and the pratyāhāra hal depends on the attribution of the technical term it to l.”  6Instead, we observe that neither Kātyāyana nor Patañjali directly mention this fault as one of mutual dependency. They rather discuss the morphological, syntactic and semantic interpretation of A 1.3.3 hal antyam, refraining from changing the text as it is without any integration or emendation. At first, the Vārttikakāra draws attention to the fact that “the rule does not contain any speci-fication”  concerning the word antyam: 7 In fact, Patañjali regularly marks this fault by the almost fixed formula itaretarāśrayāṇi 5ca kāryāṇi na prakalpante “operations affected by mutual dependence are not correct.” Pradīpa, vol. II, p. 204 ad vt. 3 pratyāhārāśrayetsaṁjñā, tadāśrayaś ca pratyāhāraḥ.6 See Joshi-Roodbergen 1994: 8.7 BONINO 4(vt. 1) halantye sarvaprasaṅgaḥ sarvāntyatvāt. 
“In the sūtra hal antyam we would have the undesired consequence that it applies to every [consonant] because all the elements are final.” The term antyam is not defined narrowly enough, according to Kātyāyana, to avoid the chance for it to apply to every consonant since, as stated by Patañ-jali, sarvo hi hal taṃ tam avadhiṃ praty antyo bhavati “for every consonant is final with respect to each limit.”  Kaiyaṭa further specifies that each consonant is po8 -tentially connected with a pause (sarvasya halo ’vasānena sambandhāt) and that the purpose of mentioning the term antyam in the sūtra is “to exclude what is initial” (antyagrahaṇaṃ tv ādinivṛttyarthaṃ syāt). Hence, it can be observed that, at this stage, the commentators are preoccupied above all with what antyam  refers 9to, because otherwise the consequence would be that any consonant, because final in the sequence of sounds that ends in it, may be considered final and therefore be termed it and deleted by A 1.3.9 tasya lopaḥ.  In the second vārttika, Kātyāyana proposes to modify the rule in order to disambiguate the term antyam: (vt. 2) siddhaṃ tu vyavasitāntyatvāt. 
“But this is established because of the fact that it is at the end of deter-mined phonemes”. What does the Vārttikakāra mean by the term vyavasita? The answer follows in the Mahābhāṣya: (MBh. I, p. 261, 9-10 ad vt. 2) vyavasitāntyo hal itsaṃjño bhavatīti vaktavyam. ke punar vyavasitāḥ? dhātuprātipadikapratyayanipātāgamādeśāḥ. sidhyati. “‘The final consonant of determined elements receives the technical term it’ has to be added. But what are these ‘determined elements?’ The verbal roots, the nominal stems, the affixes, the particles, the augments and the substitutes. This is established [when vyavasitāntya has been added].” Patañjali considers that Kātyāyana, by the term vyavasita, means a set of elements taught in the Pāṇinian grammar that is elaborated in the enumeration received by the author of the Mahābhāṣya. Kaiyaṭa further clarifies the term by  MBh. I, p. 261, 7 ad vt. 1.8 Pradīpa, vol. I, p. 202 ad vt. 1: ante ’vasāne bhavam antyam “that which exists at the end, at 9the pause”. On the Question of Circularity in A 1.3.3 hal antyam 5the following definition: vyavasitāḥ paricchinnā  ye samudāyā dhātutvapratyaya-10tvādinā dharmeṇa, teṣāṃ yo 'ntya ity arthaḥ. “The final [element] is what belongs to vyavasitas, i.e. groups [of phonemes] that are determined by the property of be-ing verbal roots, affixes, etc.” To the addition suggested by Kātyāyana is closely connected another ques-tion, that is the concern regarding the authenticity of the mention of upadeśe in A 1.3.2 upadeśe 'j anunāsika it and then repeated by means of anuvṛtti in A 1.3.3. It is not quite clear whether the meanings of vyavasita and upadeśa precisely overlap. There are several different opinions about the definition of upadeśa:  Patañjali, 11which knows the sūtra as containing the word upadeśe, begins with pratyakṣam ākhyānam upadeśaḥ and ends up with it meaning śāstram in his commentary ad A 1.3.2. But, as for the sūtra A 1.3.2, the author of the Vārttikas seems to feel the need to add this element. In his vārttikas ad A 1.3.2 he never refers to the word upadeśe and then, in vt. 2, he mentions we can manage reading the term upadeśane in the rule. In my opinion, Kātyāyana does the same thing in his com-mentary to A 1.3.3 where he maybe would not have recourse to the modification of the rule into vyavasitāntyam hal, if upadeśe was there from the beginning. This seems to me a hint corroborating the opinion that originally the term upadeśe  In the Uddyota, Nāgeśa comments on this term as follows: śāstrakṛtā paricchinnatvena 10bodhitā ity arthaḥ “the meaning is that they are taught as something/as elements de-termined by the author of the śāstra.” The interpretation of the technical term upadeśa is a rather involved topic and would 11deserve to be treated in a separate article. As we know, this term broadly conveys whatever is pronounced by the first teacher Pāṇini, the original enunciation of what was previously unknown, i.e. the Sūtrapāṭha, Dhātupātḥa, Gaṇapāṭha, Liṅgānuśāsana, Uṇādisūtras. It also refers to grammatical elements such as roots, nominal stems, af-fixes, augments, substitutes etc.  It seems that the first person to define upadeśa as ādyoccāraṇa is Haradatta ad A 6.1.45: anirjñātasvarūpasya svarūpajñāpanārtham ādyam uccāraṇam upadeśaḥ. Similarly, Nāgeśa in the Uddyota gives the definition ajñātajñāpanam. As it has been stated, we find dif-ferent interpretations of this term. For example, the following verse: dhātusūtragaṇoṇā-divākyaliṅgānuśāsanam, āgamapratyayādeśā upadeśāḥ prakīrtitāḥ. To the best of my knowledge, this first appears in the Rūpāvatāra ad P. 1.3.2. It is not found in the Kāśikāvṛtti or its two published commentators, but is quoted by Rāmacandra in the Prakriyākaumudī ad loc. cit. Also in the Mādhavīyadhātuvṛtti ad A 1.3.2 we find a similar definition of upadeśa: upadeśo dhātusūtraprātipadikapratyayāgamādeśalakṣaṇaḥ. We can therefore conclude that, to some extent, the two meanings attributed to upadeśa and vyavasita correspond. BONINO 6was lacking in A 1.3.2 and was then inserted in the rule in the time between Kā-tyāyana and Patañjali. This question deserves certainly more attention. As for the word upadeśane in A 1.3.2, Patañjali rejects Kātyāyana’s proposi-tion, showing his well-known reluctance to change the original wording  of 12Pāṇini’s rules: yathānyāsam evāstu “Let [the rule] be just as it has been originally phrased”. At this point, the Bhāṣyakāra states the necessity to resort to a prakarṣagati, a special understanding, which allows to grasp the full meaning of the sūtra without any further addition. (MBh I, p. 261, 11-14 ad vt. 2) nanu coktaṃ halantye sarvaprasaṅgaḥ sarvānty-atvād iti, naiṣa doṣaḥ. āhāyaṃ halantyam itsaṃjñaṃ bhavatīti sarvaś ca hal taṃ tam avadhiṃ praty antyo bhavati. tatra prakarṣagatir vijñāsyate. sādhīyo yo ’ntya iti. kaś ca sādhīyaḥ? yo vyavasitāntyaḥ. 
“[Objection:] But it has been stated that ‘In the sūtra hal antyam we would have the undesired consequence that it applies to every [conso-nant].’ This is not a fault. Pāṇini says that the last consonant receives the technical term it, and every consonant is final with respect to each limit. An exceptional understanding will be ascertained there. ‘That which is final in a higher degree’ is understood. And what is [final] in a higher degree? That which is at the end of determined elements.” In this passage, Patañjali explains that in order to properly interpret what Pāṇini meant by antyam, a prakarṣagati, a superior degree of understanding, is required. The idea of pre-eminence is specified by the term sādhīyas, the com-parative degree of the adjective sādhu-: what is ultimately final is what is at the end of a determined sequence of phonemes, namely after dhātus (“verbal roots”), prātipadikas (“nominal stems”), pratyayas (“affixes”), nipātas (“particles”), āgamas (“augments”) and ādeśas (“substitutes”). By stating this, the Bhāṣyakāra prevents the undesired consequence that the technical term it would apply to every con-sonant. At this point, Kaiyaṭa points out that here the term antyam must have a broader scope than simply excluding an initial element: as a matter of fact, every phoneme is necessarily final since even the initial phoneme of a word is final immediately after being uttered. Therefore, prakarṣa āśrīyate “an exceptionality is  MBh. I, p. 261, 11 ad vt. 2 sūtraṃ tarhi bhidyate “In that case the rule is altered.”12 On the Question of Circularity in A 1.3.3 hal antyam 7required.”  The author of the Pradīpa also adds that the finality of beginnings 13and middles is only imagined (ādimadhyānāṃ ca kālpanikam antyatvaṃ), but the one of the final element, meaning at the end of a vyavasita, is absolutely nityam “invariable” (sarvāntyasya  tu nityam eva). In conclusion, according to both Patañ14 -jali and Kaiyaṭa, the addition of vyavasita, that is, the sūtra then resulting as vyavasitāntyam hal, would be redundant since already by the word antyam alone one should understand that only a consonant situated at the end “of [a group of] determined elements” is susceptible to be termed it.   Once the element antyam has been discussed in the third vārttika, Kātyāyana, concerned with the analysis of hal, detects a more important defect in the formulation of the rule hal antyam: (vt. 3) lakārasyānubandhājñāpitatvād dhalgrahaṇāprasiddhiḥ. 
“The mention of hal is not well known because the phoneme l hasn’t been taught as a marker.” Patañjali further elucidates what is meant in the second vārttika, stating that the mention of hal in the sūtra is not communicative because “when it has been stated that the last consonant receives the technical term it, at the outset the technical term it has not applied to the phoneme l.”   15It is at this stage that Kātyāyana seems to lay the foundations for the attri-bution of the itaretarāśrayadoṣa.  Or at least this is the interpretation of Kaiyaṭa 16on this passage: after him, all commentators will explicitly deal with the flaw of mutual dependence between the sūtras A 1.3.3 hal antyam and A 1.1.71 ādir antyena sahetā. However, if we stick to the Mahābhāṣya’s text, there is no trace of the term  Pradīpa, vol. II, p. 203 ad vt. 2 tatra prakarṣagatir iti. ādinivṛttyarthatvād antyaśabdasya 13kathaṃ prakarṣagatiḥ? ucyate – āder apy uccāraṇānantaram avasānasambandhād antyatvam anivāryam iti nāsti tannivṛttir iti sāmarthyāt prakarṣa āśrīyate “[MBh.] tatra prakarṣagatiḥ: how could there be an exceptional understanding of the word antya- given that it has the purpose to exclude what is initial? We answer that there is no exclusion even of what is initial because, since there is a connection with a pause immediately after the pronunciation of an initial phoneme, its finality is unavoidable. On the strength of this, exceptionality is required.” Here the word “sarva” refers to the “entire (form of a vyavasita)” and not to any “all.” 14 MBh I, p. 261, 17-18 ad vt. 3 hal antyam itsaṃjñaṃ bhavatīty ucyate lakārasyaiva tāvad it15 -saṃjñā na prāpnoti. A general definition of itaretarāśrayatva is anyajñaptyadhīnajñaptiviṣayatvam “the fact 16that the object of knowledge depends on an other knowledge” (M. Kulkarni). BONINO 8itaretarāśrayadoṣa. Before skipping to the two solutions provided by Kātyāyana in the following vārttikas, Kaiyaṭa’s formulation of this error is worth some atten-tion, also because it had success among the later commentators, to the point that it is quoted nearly word for word by Haradatta, for example, in the Padamañjarī  and by Sāyaṇa in the Mādhavīyadhātuvṛtti.  17 18(Pradīpa, vol. II, p. 204 ad vt. 3) lakārasyeti. itaretarāśrayaṃ manyate, śaṣasar hal ity atra lakārasya satyām itsaṁjñāyāṃ hal iti pratyāhāraḥ, sati ca pratyāhāre laṇ ity atra lakārasya haltvāt tasyaiva śaṣasar hal ity atrānte nirdiṣṭatvād itsaṁjñeti pratyāhārāśrayetsaṁjñā, tadāśrayaś ca pratyāhāraḥ “[Vt.] lakārasya: [The author of the vārttika] has in mind [the logical falla-cy of] mutual dependence. When the technical term it is assigned to the l here in śaṣasar hal there is the pratyāhāra hal, and since l has been taught here at the end of śaṣasar hal it itself receives the technical term it be-cause the l in laṇ is hal when there is the pratyāhāra. Therefore, the attri- In the Padamañjarī, the fault of mutual dependence is formulated as follows, when 17Haradatta comments on Kāśikāvṛtti ad A 1.3.3: iha śaṣasar hal iti yo lakāras tasyet-saṃjñāyāṃ satyāṃ hal ity ayaṃ pratyāhāra upapadyate, sati ca pratyāhāre laṇ ity atra lakārasya haltvāt tasyaiva śaṣasarhal ity atrānte nirdiṣṭatvād itsaṃjñā, tadāśrayaś ca pratyāhāra iti itaretarāśrayatvāt pratyāhāro nopapannaḥ. “Here when the technical term it is attributed to the l in śaṣasar hal, then the pratyāhāra is logically justifiable, and since l has been taught here at the end of śaṣasar hal it itself receives the technical term it because the l in laṇ is hal when there is the pratyāhāra. And the pratyāhāra depends on this (the attribution of the technical term it); therefore, the pratyāhāra is not possible since there is mutual dependence.” In the Mādhavīyadhātuvṛtti, the formulation of the itaretarāśrayadoṣa is found in the first 18part of the text that deals with the prakriyā of the verbal root bhū sattāyām, “to be, in the sense of being, existence,” precisely under the sūtra 3.2.123 vartamāne laṭ, when the author is preoccupied with defining why the phonemes a and ṭ in laṭ will undergo ellip-sis. Sāyaṇa mostly limits himself to summarizing, partially reformulating and rear-ranging the interpretations already formulated by the previous commentators. For example, from the Pradīpa: nanu śaṣasar hal ity atra lakārasya ittve tena ādir antyena sahetā iti pratyāhāraḥ, sati ca tasmin laṇ ity atra lakārasya haltvāt tasyaiva śaṣasar hal ity atra ante nirdeśāt itsaṃjñā, tadāśrayaś ca hal iti pratyāhāra iti itaretarāśrayatvād ayam eva tāvat pratyāhāro ‘nupapannaḥ. “[Objection:] When the technical term it is assigned to the l here in śaṣasar hal, the pratyāhāra is formed by virtue of ādir antyena sahetā, and since l has been taught here at the end of śaṣasar hal it itself receives the technical term it be-cause the l in laṅ is hal when there is the pratyāhāra. And the pratyāhāra hal depends on this, therefore precisely this pratyāhāra is not possible at the outset because of the mutual dependence.” On the Question of Circularity in A 1.3.3 hal antyam 9bution of the technical term it to l depends on the pratyāhāra hal and the pratyāhāra hal depends on the attribution of the technical term it to l.” As it has previously been highlighted, according to Kaiyaṭa, when Kā-tyāyana formulates the third vārttika, he’s tackling the flaw of mutual depen-dence in the discussion. This is because hal antyam prescribes that the last con-sonant of a phonetic group has to be considered as an it, a marker that will dis-appear according to A 1.3.9; on the other hand, ādir antyena sahetā explains the formation of a pratyāhāra, the combining in a single syllable of two phonemes, the first being a real phoneme, the second a marker, an it, and the set designat-ing, besides the phoneme in question, everything that is intervening. However, in hal antyam, the element ‘consonant’ is represented by a pratyāhāra, hal, which refers to all consonants (from the fifth pratyāhārasūtra to the fourteenth). Now, if it is true that the formation of hal requires this last sūtra, it is also true that ādir antyena sahetā presupposes the rule hal antyam, since it is precisely in this sūtra that we come to know what an it is. Therefore, in order to assign the tech-nical term it to hal, the pratyāhāra must be already formed, but to form a pratyāhāra we need to know what an it is. As a matter of fact, this corresponds precisely to what is considered to be the non-trivial variant of an itaretarāśraya-doṣa, as it is described by Candotti. Both Kātyāyana and Patañjali never affirm clearly that the sūtra A 1.3.3 is affected by this defect, although the two solutions offered in the fourth and in the fifth vārttika seem nonetheless to deal precisely with it, positing in the for-mer the existence of a ‘ghost marker’,  in the latter of an invisible word. Regard19 -ing the first proposal, they say: (vt. 4) siddhaṃ tu lakāranirdeśāt. 
“But it is established by the mention of the phoneme l.”  (MBh. I, p. 261, 9-10, ad vt. 4) siddham etat. katham? lakāranirdeśaḥ kar-tavyaḥ, hal antyam itsaṃjñaṃ bhavati lakāraś ceti vaktavyam. 
“This is established. How? The mention of l should be made. And it should be stated that ‘the final consonant receives the technical term it as well as the phoneme l.’”  See Kiparsky 2007: 8. In this article, he argues that the fact of positing “ghost” markers 19is an un-Pāṇinian workaround and a later reparsing, since “consonantal markers are attached to the vocalic edge of a morpheme if possible, and unpronounceable clusters are wholly eschewed.”  BONINO 10Once again, Kaiyaṭa helps to clarify what the two commentators mean by the elliptical lakāranirdeśa.  In fact, what they both suggest is to interpret the 20sūtra hal antyam as hal antyam lakāraś ca  and to consider the term hal as a 21samāhāradvandva, a neuter, singular dvandva compound (hal ca l ca). The second member of the compound being part of conjunct consonants would be elided by A 8.2.23 saṃyogāntasya lopaḥ, and therefore be invisible, in hal. This solution would allow the assignment of the technical term it to the final l of hal, which corresponds here to the fourteenth pratyāhārasūtra. Once this is established, this l can be connected with any beginning sound by means of A 1.1.71 to form a pratyāhāra: hence, the pratyāhāra hal mentioned in A 1.3.3 can be obtained with-out creating any circularity. The second solution is explained in the fifth vārttika and commented on by Patañjali as follows: (Vt. 5) ekaśeṣanirdeśād vā. 
“Or rather [it is established] by the mention of the ekaśeṣa.” (MBh. I, p. 261, 23 ad vt. 5) athavaikaśeṣanirdeśo ’yam. hal ca hal ca hal. hal antyam itsaṃjñaṃ bhavatīti. 
“Or rather this [hal] is the mention of the single remainder: hal is the single remainder in [the dvandva compound] hal and hal.” Katyāyāna proposes here to consider the pratyāhāra hal in A 1.3.3 as an ekaśeṣa, the single form remaining of the original two words hal and hal, which is prescribed in A 1.2.64 sarūpāṇām ekaśeṣa ekavibhaktau. As stated by Patañjali, from hal and hal we derive hal. According to the Pradīpa, the former hal would be a ṣaṣṭhītatpuruṣa, a determinative compound with the first member in the geni-tive case, the traditional analysis of which is hasya l hal “the l next to ha,”  the 22 Pradīpa ad Vt. 4: lakāraś ceti. pūrvaṃ lakārasyetsaṁjñā vidheyā, tena hal iti pratyāhāra upa20 -padyate. tatra hal ca l ceti samāhāradvandvaṃ kṛtvā ‘saṃyogāntasya lopa’ iti lakāro lupyate “[MBh.] lakāraś ca: first the technical term it has to be prescribed for the phoneme l, then the pratyāhāra hal is logically justifiable. In this pratyāhāra, after forming the dvandva compound neuter and singular (samāhāradvandva) hal ca l ca, the l is elided by A 8.2.23 saṃyogāntasya lopaḥ.” Uddyota ad Vt. 4: hal antyam laś ca.21 The Padamañjarī ad Kāśikāvṛtti ad A 1.3.3 explains that the relation expressed by the 22genitive is one of sāmīpya, proximity: samīpasamīpisambandhe ṣaṣṭhīsamāsa ity arthaḥ “a genitive tatpuruṣa compound is meant because there is a connection between what is close and what has the closeness.” On the Question of Circularity in A 1.3.3 hal antyam 11latter the pratyāhāra hal. Again, if we first apply the technical term it to final l of the fourteenth pratyāhārasūtra hal, here indicated by the tatpuruṣa, the l follow-ing the consonant ha, then we can form the pratyāhāra hal without the reasoning being affected by the flaw of circularity.   23The Vārttikakāra seems to be satisfied with this last proposal. And it is in-teresting to notice that even Nāgeśa, commenting on this vārttika in the Uddyota, affirms that the solution of considering hal as an ekaśeṣa is the only possible con-clusion in order to prevent the itaretarāśrayadoṣa (tasmād ekaśeṣanirdeśād vety eva samādhānam) in the case of A 1.3.3. However, Patañjali examines two other alter-natives: (MBh. I, p. 262, 1-3 ad vt. 5) athavā ḷkārasyaivedaṃ guṇabhūtasya grahaṇaṃ, tatropadeśe 'j anunāsika it itītsaṃjñā bhaviṣyati. athavācāryapravṛttir jñāpay-ati bhavati lakārasyetsaṃjñeti yad ayaṃ ṇalaṃ litaṃ karoti. 
“Or rather this is the mention of nothing other of the vowel ḷ that is gu-nated; the technical term it will be attributed to it by the rule A 1.3.2 upadeśe 'j anunāsika it. Or rather the method of the teacher [Pāṇini] makes us understand that the technical term it is attributed to l, because he makes the suffix ṇal furnished with the marker l.” The first option takes into account the rule A 1.1.51 ur aṇ raparaḥ.  In both 24the pratyāhārasūtra as well as the rule hal antyam, hal results from ha + ḷ. Both a + ḷ will then be replaced by the guṇa of ḷ (a) by A 6.1.87 and l will be added by A 1.1.51: in this case ḷ is the it. Kaiyaṭa rejects this workaround, objecting that again there would be itaretarāśraya between the rules A 1.3.2 and A 1.1.71 since also in the sūtra upadeśe 'j anunāsika it, the element “vowel” is represented by a pratyāhāra, ac, the formation of which depends on ādir antyena sahetā, and vice versa. However, if we instead understand that Patañjali has in his mind the rule A 3.1.55 puṣādidyutādyḷditaḥ parasmaipadeṣu, where Pāṇini has left a jñāpaka that ḷ  It is worth noticing that the solution invoked in vt. 5, the ekaśeṣa, is deeply connected to 23the idea of polysemy and it shows that the commentators were at this stage already aware of this concept, as Pāṇini probably was. The device of the single remainder among forms identical to each other is nothing but the technical mechanism underly-ing the idea that words can express more than one meaning at a time and it is only formally different from the solution provided later by the Kāśikāvṛtti, namely tantra (see p. 10 of this article).  This rule applies to vocalic l as well as vocalic r, either by the vt. lakāreti vaktavyam, or 24by reference to the vt. on A1.1.9 ṛḷvarṇayor sāvarṇyaṁ vācyam. BONINO 12is an it by using “ḷdit.” In other words, Pāṇini could only say “ḷdit” if ḷ is an it. Therefore, there is no longer mutual dependence.   25Otherwise, in the second option, the solution provided by reference to the jñāpaka ḷditaḥ is precisely echoed. It is based on the fact that in A 3.1.55 there is mention of hal: hal is only relevant for excluding ṇal. If the l in ṇal were to be pronounced, i.e. were not an it, there would be no point in mentioning hal in A 3.1.55. As it has been shown, unlike the other cases of itaretarāśrayadoṣa found in the Aṣṭādhyāyī, Kātyāyana and Patañjali never directly addressed the problem perceived in the formulation of the rule A 1.3.3 hal antyam by this term, and yet they seem to offer solutions precisely for it, to the point that Kaiyaṭa in the Pradīpa doesn’t hesitate to mention the mutual dependence affecting this sūtra and A 1.1.71 ādir antyena sahetā. However, from a chronological point of view, he’s not the first one to identify such a flaw in the rule hal antyam. The Kāśikāvṛtti, directly commenting on the Pāṇinian sūtra, provides an-other solution to the recognized fault of mutual dependence in A 1.3.3:  (KV ad A 1.3.3) hasya l hal iti dvitīyam atra halgrahaṇaṃ tantreṇopāttaṃ draṣṭavyam, tena pratyāhārapāṭhe hal ity atra lakārasya itsaṃjñā kriyate. tathā ca sati, ‘hal antyam’ ity atra pratyāhāre netaretarāśrayadoṣo bhavati. 
“Here the second mention of hal, meaning the [ṣaṣṭhītatpuruṣa] hasya l has to be considered as obtained by tantra; therefore, here in hal as it is read in the pratyāhārasūtra, l is termed it. And if that is the case, there is no fault of mutual dependence concerning the pratyāhāra in hal antyam.”  Pradīpa ad Vt. 5: athaveti. śaṣasar hal ity atra hal antyam ity atra ca ḷkārasyaikādeśo lapara-25tvaṃ ca kriyate. nanv evam apītaretarāśrayam eva, ḷkārasya satyām itsaṁjñāyāṃ hal iti pratyāhāraḥ, sati ca pratyāhāre aiauj iti cakārasyetsaṁjñā, tasyāṃ satyām aj iti pratyāhāre sati ḷkārasyetsaṁjñā. evan tarhi ḷd itaḥ iti jñāpakāl ḷkārasyetsaṁjñākāryaṃ bhaviṣyatīti nāstītare-tarāśrayatvam iti bhagavato bhāṣyakārasyā ‘bhiprāyaḥ. “[MBh.] athavā: both in the pratyāhārasūtra hal and in the rule hal antyam the single substitute of the vowel ḷ (a) and the fact that this is followed by an l are made. [Objection:] Even in this way the reasoning is absolutely affected by mutual dependence: when the technical term it is attributed to the vowel ḷ there is the pratyāhāra hal, when there is the pratyāhāra the technical term it is attributed to the phoneme c of aiauc, when this is attributed, when there is the pratyāhāra ac, the technical term it is attributed to the vowel ḷ. Then in this manner, by the evidence of puṣādidyutādyḷditaḥ parasmaipadeṣu (A 3.1.55), there will be the result of the attribution of the technical term it to the vowel ḷ, therefore the reason-ing is no longer affected by mutual dependence – This is the intention of Patañjali.” On the Question of Circularity in A 1.3.3 hal antyam 13The need of a second hal in the sūtra, namely the ṣaṣṭhītatpuruṣa hal rep-resenting the fourteenth pratyāhārasūtra, is still stated, but instead of suppos-ing an elision of one of the two hals, this would be obtained by tantra, namely the bringing about of several meanings through the denotative power of a single word, uttered once.  In the Padamañjarī, Haradatta defines the term tantra as “a 26common support”  (sādhāraṇa) and says that it consists in a prayatnaviśeṣa, a 27particular kind of effort, the same kind that we find in śveto dhāvati “a white (horse) is running [or/and] a dog runs from this point,” where a double meaning is conveyed with a single effort: in the same way, by the single element hal in the sūtra, one understands two meanings, on one hand the pratyāhārasūtra hal and on the other the pratyāhāra hal.  In this way, the mutual dependence identified 28in the sūtra hal antyam is averted.  As it has been pointed out by Joshi and Roodbergen (Paspaśāhnika, fn. 894), the origins 26of the concept of tantra draw back to the ritualistic context. Śabara, in his commentary on Mīmāṃsāsūtra 11.1.1, explains it as yat sakṛt kṛtam bahūnām upakaroti tat tantram ity ucyate “that which is performed once for many [people/purposes], is called tantra.” In the Nyāyakośa or Dictionary of Technical Terms of Indian Philosophy by Jhalakīkar (p. 318), revised and re-edited by Abhyankar, among many definitions, we find the following one, which seems to be closer to the idea of polysemy: sakṛduccāritasyaikasya śabdasya śaktyā anekārthapratipādakatvaṃ tantram iti śābdikā vadanti “the grammarians say that tantra is the fact of conveying multiple meanings with the power of a single word pro-nounced once.” Cf. the famous example from daily life given by Haradatta to illustrate the notion of 27sādhāraṇa: tulyakakṣyayor bhuñjānayoḥ pradīpaḥ “a lamp for two people eating in the same room.” The same linguistic metaphor is employed by Jinendrabuddhi in the Nyāsa: pradīpaḥ suprajvalito bahūnāṃ chātrāṇām upakāraṃ karoti – “a lamp that, shining bright in the room, is useful to many students.” Padamañjarī ad Kāśikāvṛtti ad A 1.3.3: sa ceha prayatnaviśeṣaḥ, yathā – śveto dhāvatīty atra 28sādhāraṇenaikenaiva prayatnena dvayor vākyayor uccāritayoḥ phalaṃ sampadyate, tathehāpy ekena tantreṇa prabalena dvitīyaṃ halgrahaṇam upāttaṃ parigṛhītaṃ veditavyam. “And in this context this is a particular kind of effort, for example when we say śveto dhāvati the (semantic) fruit of two sentences pronounced is attained with a single common effort, in the same manner with a single strong tantra a second mention of hal should be un-derstood as accepted.” Similarly, in the Nyāsa ad Kāśikāvṛtti ad A 1.3.3: iha tu prayatna-viśeṣas tantraśabdena vivakṣitaḥ. tena tantreṇa dvitīyam atra halgrahaṇam upāttaṃ parigṛhī-taṃ veditavyam. yathā śveto dhavatīty ekena prayatnena dve vākye uccārite bhavataḥ, tathe-hāpy ekenaiva prayatnena dvau halśabdāv uccāritāv ity abhiprāyaḥ. BONINO 14As previously highlighted, the case under scrutiny is particular since it represents an exception to an univocally attributed fault found in the Pāṇinian grammatical system: Kātyāyana and Patañjali demonstrate that they are aware of a deficiency in the formulation of the rule A 1.3.3, but, unlike what themselves do in other occasions (a little less than 80 times) and unlike what other commen-tators whose texts have been the object of our analysis do,  they never name this 29defect as itaretarāśrayadoṣa. It has also been argued that, in the case it was Patañ-jali’s intention to mark a circularity in this sūtra, his practice would here com-pletely diverge from the general way of identifying this type of error, namely by stating that there is no accomplishment because of the mutual dependence (itaretarāśrayatvāt aprasiddhiḥ) and that operations affected by this mutual de-pendence do not attain their outcome (itaretarāśrayāṇi ca kāryāṇi na prakalpante).  I’m not able to offer a final solution to this matter, but I think that it would be appropriate to consider it within the metalinguistic context, to which belong the notions of it, “marker,” and pratyāhāra, the “condensed expression of sounds.” As we know, the phonemes characterized as it are attached to mor-phemes to encode their grammatical properties and to allow classes to be formed by the pratyāhāra technique, but then they are deleted and they are not part of the morpheme at any stage of the derivation. It is this simultaneous presence and absence of the it elements in the metalanguage which earns them the reputation of “les éléments au statut le plus difficilement définissable.”  In the 30Mahābhāṣya, they are often mentioned in relation to the teaching about the sounds used in the grammar and in relation to the pratyāhāras: in fact, it is clearly stated that the list of sounds in the pratyāhārasūtras is posited for the purpose of teaching the markers, it, besides being necessary for the sake of the whole grammatical system; and the markers are established in order to form the pratyāhāras, this operation being taught in the saṃjñāsūtra A 1.1.71 ādir antyena sahetā.  In another passage, Patañjali more affirms that: 31 The solution provided by Bhaṭṭoji Dīkṣita, namely the āvṛtti, the repetition of the sūtra 29hal antyam, is very well known, and on account of this it has not been wholly explained in this paper. See Candotti 2006: 156.30 MBh. I, p. 13, I. 8-12 ad vt. 16: anubandhakaraṇārthaś ca varṇānām upadeśaḥ kartavyaḥ. 31anubandhān āsaṅkṣyāmīti. na hi anupadiśya varṇān anubandhāḥ śakyāḥ āsaṅktum.  saḥ eṣa varṇānām upadeśo vṛttisamavāyārthaś cānubandhakaraṇārthaś ca. vṛttisamavāyaś ca anubandhakaraṇaṃ ca pratyāhārārtham. pratyāhāro vṛttyarthaḥ. On the Question of Circularity in A 1.3.3 hal antyam 15(MBh. I, p. 64, 1. 12-14 ad A 1.1.10 vt. 4) varṇānām upadeśas tāvat, upadeśot-tarakāletsaṃjñā, itsaṃjñottarakāla “ādir antyena sahetā” iti pratyāhāraḥ. 
“At the beginning, there is the instruction of the sounds. After this, [the introduction] of the technical term it. After the introduction of the term it, [the explanation] of the pratyāhāras by ādir antyena sahetā.” In this passage which establishes the correct sequence of the grammatical instruction – first the sounds taught in the pratyāhārasūtras (along with the fi-nal it concluding each string), then the markers it, and, only after this, the for-mation of the pratyāhāras – Patañjali presents an ideal situation where the sounds would be independent available and therefore there would be no itare-tarāśrayadoṣa. But both Kātyāyana and Patañjali are aware that this is not quite reflected in the Pāṇinian system, where the Pratyāhārasūtras seem, in most of the cases, to have to be somehow instructed. It is then clear that the issue at stake, from the point of view of Kātyāyana and Patañjali, is the circularity be-tween the attribution of the technical term it and the formation of a pratyāhāra. The two commentators recognize the flaw of mutual dependence affecting the sūtra hal antyam and try to offer certain solutions in order not to modify the original enunciation. What remains problematic and deserves further investiga-tions is the reason why, while Kātyāyana clearly lays the argumentations to point this error out in the system of the Aṣṭādhyāyī, none of the formulaic language associated with the other instances of itaretarāśraya are used in this context. Bibliography Primary sources Böhtlingk, O. 1887. Pāṇini’s Grammatik. Leipzig: Verlagvon H.Haessel.  Joshi, S. D., Roodbergen, J.A.F. 1986. Patañjali’s Vyākaraṇa-Mahābhāṣya Paspaśāh-nika – Introduction, Text, Translation and Notes. Pune: Publications of the Centre of Advanced Study in Sanskrit. Joshi, S. D., Roodbergen, J.A.F., eds. 1994. The Aṣṭādhyāyī of Pāṇini with Translation and Explanatory Notes, Volume III (1.3.1-1.3.93). Delhi: Sahitya Akademi. Katre, S. M. 1987. Aṣṭādhyāyī of Pāṇini in Roman Transliteration and English Transla-tion by Sumitra M. Katre. Austin: University of Texas Press.
 BONINO 16Kielhorn, F. 1962. The Vyākaraṇa-mahābhāṣya of Patañjali, edited by F. Kielhorn. Third Edition revised and furnished with additional readings, references and select critical notes by K. V. Abhyankar. Vol. I. Pune: Bhandarkar Oriental Research Insti-tute. Misra, S. 1985. Kāśikāvṛtti of Jayāditya-Vāmana (Along with commentaries Vivaraṇa-pañcikā-Nyāsa of Jinendrabuddhi and Padamañjarī of Haradatta Miśra), Part I. Varanasi: Ratna Publications. Renou, L. 1948-54. La Grammaire de Pāṇini. Traduite du sanskrit avec des extraits des commentaires indigènes. Paris: Klincksieck. Shastri, D. 1964. The Mādhavīyadhātuvṛtti - A Treatise on Sanskrit Roots Based on the Dhātupāṭha of Pāṇini by Sāyaṇācārya. Varanasi: Tara Publication.  Vasu, C. 1891 [1988]. The Aṣṭādhyāyī of Pāṇini. Delhi, Varanasi, Patna: Motilal Ba-narsidass.  Vedavrata. 1962-1963. Śribhagavatpatañjalivicaritaṃ vyākaraṇamahābhāṣyam. Śrī-kaiyaṭakṛtapradīpena nagojībhaṭṭakṛtena bhāṣyapradīpoddyotena ca vibhūṣitam, 5 Vols. Gurukula Jhajjar (Rohatak): Harayāṇā Sāhitya Saṃsthāna.  Secondary references Abhyankar, K. V. 1961. A Dictionary of Sanskrit Grammar. Baroda: Oriental Institute of Baroda.  Aussant, É. 2005. “L’autonymie dans la tradition grammaticale sanskrite.” His-toire, Épistémologie, Langage 27: 73-92. Bronkorst, J. 2004. From Pāṇini to Patañjali: the Search for Linearity. Post-Graduate and Research Department Series; 46. Pune: Bhandarkar Oriental Research Institute. Candotti, M. P. 2006. Interprétation du discours métalinguistique – La fortune du sūtra A 1 1 68 chez Patañjali et Bhartṛhari. Firenze: Firenze University Press. Candotti, M. P. 2016. “Natural and Grammatical Zero: The Case of Indeclinable.” In G. Cardona & H. Ogawa, eds., Proceedings of the Vyākaraṇa Section of the 16th World Sanskrit Conference. New Delhi: DK Publisher Distributors, 99-138. Cardona, G. 1976. Pāṇini. A Survey of Research. The Hague, Paris: Mouton.  On the Question of Circularity in A 1.3.3 hal antyam 17Cardona, G. 1997. Pāṇini. His Work and its Traditions. Volume One. Background and Introduction, Second Edition, revised and enlarged. Delhi: Motilal Banarsi-dass. Cardona, G. 1999. Recent Research in Pāṇinian Studies.Delhi: Motilal Banarsidass. Devasthali, G. 1967. V., The Anubandhas of Pāṇini. Poona: Publications of the Ad-vanced Centre Of Research, University of Poona. Jhalakīkar, M. B. 1928. Nyāyakośa or Dictionary of Technical Terms of Indian Philoso-phy, revised and re-edited by M. V. Abhyankar. Poona: Bhandarkar Orien-tal Research Institute. Kiparsky, P. 1979. Pāṇini as a Variationist.Poona: Poona University Press. Kiparsky, P. 1982. Some Theoretical Problems in Pāṇini’s Grammar. Poona: Bhandar-kar Oriental Research Institute. Kiparsky, P. 1991. “Economy and the Construction of the Śivasūtras.” In Desh-pande M. M., Bhate, S. (eds.), Pāṇinian Studies: S.D. Joshi Felicitation Volume. Ann Arbor: University of Michigan Center for South Asia Studies. Kiparsky, P. 2007. “Pāṇini is slick, but he isn’t mean.” Nagoya Studies in Indian Culture and Buddhism: Sambhasa 26: 1-28. Kunjunni Raja, K. 1963. Indian Theories of Meaning. The Adyar Library Series vol. 91. Madras: The Adyar Library and Research Centre. Ogawa, H. 2005. Process & Language – A study of the Mahābhāṣya ad A 1.3.1 bhūvādayo dhātavaḥ. Delhi: Motilal Banarsidass. Palsule, G. B. 1961. A Sanskrit Dhātupāṭha – A critical study. Poona: Deccan College Dissertation Series.  Renou, L. 1942. Terminologie grammaticale du sanskrit, 3 Vol. Paris: Librairie An-cienne Honoré Champion. Roodbergen, J. A. 2008. A Dictionary of Pāṇinian Grammatical Terminology. Poona: Bhandarkar Oriental Research Institute. Scharfe, H. 1971. Pāṇini’s Metalanguage.Philadelphia: American Philosophical So-ciety. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items