Phonetics of V

Source: TW

विश्वास-टिप्पनी

The aim of this paper is not to prove that for Pānini v did not involve teeth in any capacity or that it was a pure bi-labial sound. It could have been a bi-labial ; but we have no positive proof to that effect.

So this paper is more about the vyAkaraNa sanJNA rather than actual pronunciation.

Intro

Bhandarkar Oriental Research Institute
PHONETICS OF V IN PĀṆINI
Author(s): M. M. Deshpande
Source: Annals of the Bhandarkar Oriental Research Institute, Vol. 56, No. 1/4 (1975), pp. 45-65
Published by: Bhandarkar Oriental Research Institute

1 The traditional Pāninian scholars of modern India rarely doubts as to the exact phonetic classification of the semi-vowel v. -standard texts like the kāśikā-vr̥tti and the Siddhānta-kaumudī classify v as being danty oṣṭhya , i. e. with teeth and lips as its points of articulation.1 This double point of articulation classification is also supported by the Pāninlya-śikṣā 2, and hence is rarely questioned by the traditional Pāninian circles. Some of the modern scholars have treated this question from a historical perspective. Jakob Wackernagel ( 1914 ) claims that for Pāṇini v was only an oṣṭhya “labial” sound, and it cannot be included in the class of dantya " dental " sounds as intended in P. 7. 3. 73 ( lug vā duha-diha-liha-guhām ātmanepade dantye ). His argument is based upon the fact that v is included in the class of oṣṭhya “labial” sounds in P. 7. 1. 102 ( ud oṣṭhya - pūrvasya ). This rule prescribes replacement of a verb-final r by u , where r is preceded by an oṣṭhya “labial " sound, i. e. mumūrṣati<mr . In order to derive forms like vuvūrṣati < vr, the labial class must include v. Wackernagel concluded that if v is included in the labial class, it could not have been also dental at the same time. The traditional argu ment against such a position is that a dantyoṣṭhya " labio-dental " sound may be called dantya " dental”, oṣṭhya " labial " or dantyoṣṭhya " labio-dental “.3 Hence inclusion of v in the labial class by itself cannot indicate that v is not a dental sound also. We need independent proof to that effect.

  1. Cardona ( 1964 ) has discussed Wackernagel’s claims and finally he sides with the traditional view as upheld by the Kāśikā - vr̥tti that v was a dantyoṣṭhya " labio-dental " sound for Pāṇini. He also seeks general support for this conclusion in the śikṣās and Pratiśākhyas . His general impressions are expressed as follows 1 :

" That v could be labio-dental in Pāṇini’s system cannot be doubted. … The Pāninīyaśikṣā clearly states this : dantyosthyo vah smrto budhaiḥ ( 18 ) ; the VPr. and the TPr also class v as a labio-dental : VPr. I. 70 uvospapā oṣṭhe ( sthāne ), I. 81 vo dantāgraih ( kriyate ) ; TPr . II. 43 oṣṭhāntābhyāṁ dantair vakāre. In the RPr. (1.47 śeṣa osthyo’pavādya nāsikyān ) it is true that v is included in a class characterized only as oṣṭhya , but this refers only to sthāna ( I. 49 iti sthānāni ). There is no definite statement classing v as uniquely oṣṭhya . Therefore, it is unreasonable to insist that this was true in Pāṇini’s system. "

sthāna vs karaṇa

Actually whether the teeth function as sthāna " point of articulation " or as karaṇa " articulator " makes a great difference in phonetic classifications of the ancient Indian texts. The term " dental " in English is sufficiently elastic to blur this distinction and include sounds with dental point of articulation and also sounds with dental articulator feature. However, it must be clearly understood that the Sanskrit term dantya refers only to those sounds which have dental point of articulation, and does not include sounds with dental articulator. Teeth function in two distinct capacities and these capacities cannot be mixed in one category of dentality. In the Prātikākhyas we find that oṣṭha " lips " and nāsikā " nose " also function in these two distinct capacities, and in the ancient descriptive system these capacities must be clearly distinguished.2

1 Ghosh ( 1938 )} p. 62, disregards the distinction between teeth functioning as the point of articulation and as the articulator and holds that for VPr and TPr v was a dantyoṣṭhya “labio-dental " sound. This is quite inappropriate especially when he aims at exclusively " Places of articulation " ( p. 62 ). A similar non-distinction is seen in Whitney’s Sanskrit Grammar , p. 56.
2 Under SGA 1.18 (mukhe viśeṣāḥ karaṇasya), we get SCA 1.25 ( oṣṭhyānām adharauṣṭham ), SGA 1.26 ( nāsikyānām nāsikā ) and SCA 1.28 ( rephasya danta-mūlāni ), where we have lips, nose and roots-of-teeth functioning as articulators. As can be gathered from the words dantya ( SCA 1. 24), oṣṭhya ( SCA 1. 25 ) and nāsikya ( SGA 1.26), teeth, lips and nose also function as points of articulation. We find that the VPr considers danta-mūla “root-of-teeth " to be the point of r in FPr'1.68 ( ro danta-mūle ), danta " teeth " to be the point of ḷ, l , s and the t-series in VPr 1.69 ( ḷ-la-si-tā dante ), and oṣṭha " lips " to be the point of u- vowels, v, o, ḥp and the p-series in VPr 1.70 ( u-v-o-ḥpa-pā oṣṭhe ). In VPr 1.81 ( vo dantāgraiḥ ), the tips-of teeth are said to be the articulator of v, while VPr 1.80 ( samāna-sthāna-karanā nāsikyauṣṭhyāḥ ) says that pure nasal sounds and labials have nose and lips as both the point and the articulator. On the other hand, VPr 1.75 ( mukha-nāsikā-karaṇo’nunāsikaḥ ) says that nasalized sounds, in contrast to pure nasals, have an oral point of articulation and nasal articulation and nasal articulator.

v as ōṣṭhya

3 Thus the sound v can be included in the dantya class only if we can say that teeth function as the point of articulation for v. Unfortunately we do not get this impression from the VPr. 1.81 (vo dantāgraih) refers to dantāgra “tips of teeth” as karaṇa “articulator” of v , while VPr. I. 70 ( uvoḥpapā oṣṭhe) " lips " as the sthāna " point of articulation " of v. The RT. II. 7 ( dante tslāḥ ) says that teeth serve as the point of articulation for t-series of stops, s and l. The RT II 8 ( repho mūle vā ) says that either teeth or root-of-teeth are the point of articulation ( oṣṭhe voḥpū ) says that lips are the point of articulation for v, o, lp, p series and u-vowels. There is no reference to teeth in the context of v in any capacity. The Tribhāṣya-ratna and other commentaries on the TPr. II. 43 ( oṣṭhāntābhyām dantair vakāre ) say that v has teeth as its point of articulation while lips are its articulator.1 Validity of this interpretation has already been doubted by Whitney.2 M. M. Paṇḍita śivadatta Kuddāla, the learned editor of the Nirṇaya-sāgara edition of the Mahābhāṣya , interprets the TPr. II. 43 to mean that teeth are the articulator of v , while lips are its point of articulation.3 In the TPr, the karaṇa " articulator " is normally expressed in instrumental,4 and hence there is a greater possibility of dantaiḥ standing for articulator of v. The Yājnavalkya-śikṣā enumerates " labial " sounds including v, and it also enumerates eight " dental " sounds without mentioning v.5 The Varṇaratnapradīpikāśikṣā also clearly says that v is produced in the lips, with teeth as its articulator.6 The same picture is seen in the work of Patañjali. 7 Thus majority of the texts seem to consider danta “teeth” to be karaṇa " articulator " of v, and thus supporting the pure labial point of articulation classification for v.

  1. This point is elaborated later in this article.

Meaning of dantyoṣṭhya

4 The Pāninīya-śikṣā, the āpiśali-śikṣā-sutras and the Cāndra-varṇasūtras explicitly classify v as a dantyoṣṭhya " labio-dental " sound. 5 It is hard to say that this represents any substantive difference.

The possibility of a terminological shift seems to be more probable. For instance, the Pāniniya-śikṣā does not explain the articulator of v, which is described to be a " labio-dental " sound. Paṇḍita Sivadatta Kuddāla interprets the compound dantyoṣṭhya " labio-dental " in such a way that teeth could still be the articulator.1 Though we may entirely agree with this view, still there is the undeniable possibility of teeth which were classified by the older texts as functioning in two distinct capacities, i. e. sthāna “point of articulation” and karaṇa " articulator” being reclassified as functioning only in one capacity, i. e. sthāna " point of articulation”. The Pāninīya-śikṣā includes danta " teeth " in its list of eight points of articulation.2 It does not discuss karaṇa " articulator " of any sounds, nor does it mention articulator as a distinctive feature of sounds. It says that sounds are classified according to five principles : svara " accent “, kāla " time “, sthāna” point of articulation”, prayatna “effort” and anupradāna “emission”.3 Since karaṇa is not included here, the author of the śikṣā might have elevated teeth entirely to the function of the point of articulation. The Āpiśali-śikṣā-sútras consider v to be " labio-dental " in the sthāna-section,4 but the karaṇa-section of these sūtras presents some problems. It is said that the jihvya “velar”, tālavya " palatal “, mūrdhanya " retroflex " and dantya " dental " sounds have different parts of the tongue ( jihvā ) as their articulators, remaining ( śeṣāḥ ) sounds have the same organ functioning as sthāna " point of articulation " and karaṇa " articulator” 6. This leaves the description of v somewhat unclear. If dantyoṣṭhya as a separte category be included in the śeṣa " remaining” then teeth and lips would be both the points and articulators of v . Otherwise, as a dantya sound would have jihvāgra " tip of the tongue " as the articulator and the point of articulation, while, as an oṣṭhya " labial " sound, v would have lips as both the point and the articulator. The exact intentions of the text are unclear. The same difficulty is also to be seen in the Cāndra-varṇasūtras .7 Thus these texts do not form a clear on the classification of v.

5 For the Pāniniya-śikṣā , see : Note 2 on p. 45 vakāro dantyoṣṭhyaḥ , Āpiśali-śikṣāsūtras , sthāna-prakaraṇa 11, śikṣā-sūtrāṇi , p. 2. The same rule is found in the Pāninīya-śikṣā-sútras . śikṣā-sūtrāṇi , p. 10, 20. The Cāndra-varṇa-sūtra 12 says dantauṣṭhaṁ vakārasya ( sthānam ), śikṣā-siitrāni , p. 24. Among other ancient texts, only the śaiśirīya-śikṣā considers v to be dantyoṣṭhya “labio-dental " [ ed. by Tarapada Chowdhury, the Journal of Vedia Stu- dies, Vol. IL, No 2, Lahore, 1935, p. 3 ]. Ghosh ( 1938 ) presents five versions of the versified Pāniniya-śikṣā and reconstructs an original which he ascribes to Pāṇini. Raghu Vīra ( 1931 ) claims that the versified Pāniniya - śikṣā is a late work, while the śikṣā-sūtras published by Swami Dayananda represent the real work of Pāṇini. This claim has been rejected by Ghosh ( 1938, pp. xlvi ff. ), while it has been upheld by Yudhiṣṭhir Mīmāṁsak ( saṁvat 2030, pp. 62 ff). Madhukar Phatak ( 1972, pp. 14 ff. ) argues that both the versions are late works and cannot be Ascribed to Pānini. In my view, the present version of the āpiśali-śikṣā-sūtras belongs to post- Patañjali times. For details : see my New Material On The KauUa- Vyākaraṇa , appearing in the Journal of the Oriental Institute , Baroda, note 38.

kāśikā view

5 Unfortunately Pāṇini has left us no phonetic treatise. All the different versions of the Pāninīyaśikṣā seem to be post-Pāninian, or even post-Patañjali, additions to this school.1 Thus to investigate what Pāṇini’s intentions were, we must use the commentators to start with, and then critically examine their statements with internal and external evidence. We shall start with the argument of the Kāśikā-vr̥tti and then examine it. The background of the argument is explained below.

The rule P. 7. 3. 73 (lug vā duha-diha-liha-guhām ātmanepade dantye ) says that after the roots duh, dih, lih and guh the aorist infix Ksa (= sa ) is optionally zeroed by luk- deletion, if followed by a middle affix beginning with a dental sound. For instance the sequence a-duh-sa-ta results in the alternation : adugdha ( with loss of sa ) and adhukṣata ( with sa ). There is no doubt that this optional luk-deletion takes place before the affixes ta , thās and dhvam , because t , th and dh are undoubtedly dantya sounds. The aorist first person dual affix vahi is the problem case. Wackernagel ( 1914 ) argued that vahi does not begin with a dantya " dental " sound, and hence we cannot have the alternation adhuksāvahi aduhvahi.4 The Kāśikā-vr̥tti on the other hand, insists that v is a dantyoṣṭhya " labio-dental " sound, and hence can be included in the class of dantya “dental " sounds. Therefore, the alternation mentioned above does result. The Kāśikā-vr̥tti tries to find proof to this effect in the particular formulation of P. 7. 3. 73 which specifically speaks of the luk-deletion of Ksa , resulting in zeroing of the eniire sa affix at once. If P. 7. 3. 73 was not intended to apply before vahi , the Kāiikā-vr̥tti argues, Pāṇini could have formulated P. 7. 3. 73 without luk-deletion. He could have continued the lopa - deletion from P. 7. 3. 70 into P. 7. 3. 73. That could take care of the optional loss of sa before ta , thās and dhvam , but not before vahi. Thus, the specific use of the luk-deletion in P. 7. 3. 73 is taken as an indication that Pāṇini wanted optional loss of sa before vahi as well. The passage in the Kāsikā-vr̥tti runs as follows1 :

lopa iti varttamāne lug-grahanam sarvādeśārtham / tac ca vahy~artham / anyatra tv antyasyaiva lope kr̥te * jhalo jhali* iti sakāra-lopena siddhyati / sthānivad-bhāvō’py akāra-lopasya nāsti, pūrvatrāsiddhe na sthānivad iti / dantyoçthyo’pi vakāro dantya iti gr̥hyate / yadi sa na gr̥hyeta , tatas tau-grahaṇam evātra kr̥tam syāt / , Kāsikā-vr̥tti , Vol. VI., ppé 82-3.

Though the word lopa " deletion " continues [ in P. 7. 3. 73 •from P. 7. 3. 70 ], still the specific prescription of the luk deletion is for the [ zero- ] substitution for the entire [ affix Ksa ]. This is to account for the affix vahi . In the case of the other [ dental-initial aorist affixes, i. e. ta, thās and dhvam ], when there is lopa-deletion of the final [ a of sa , by P. 1. 1. 53 ( alontyasya ) ], the [remaining] s [of sa] could be deleted by P. 8. 2. 26 ( jhalo jhali ), and the correct forms could be established. The lopa-deletion-substitute of a [ in sa ] cannot be treated like the original substituen^m [ a, by P. 1. 1. 57 ( acah parasmin pūrva-vidhau ) to prevent the application of P. 8. 2. 26 {jhalo jhali ) to s ], because it is said that the process of sthānivad-bhāva " treatment of the substitute like the substituendum " does not apply with respect to a rule in the last three padas of the Astādhyāyī. Though v is dantyoṣṭhya “labio-dental”, still it is covered by dantya " dental”. If it were not to be covered, then [just to refer to t , th and dh ] Pāṇini could have simply said tau “when tu” [ = t-series ] follows

The argument is that if Pāṇini did not intend v to be covered by the dantya " dental " class, he could have done away with the specific luh - deletion of Ksa in P. 7. 3. 73. The lopa-deletion could continue in P. 7, 8; 73 from P. 7. 3. 70. Then the rule would mean that after the roots duh etc. the lopa-deletion applies to sa, if followed by a dental-initial-middle aorist affix. The lopa-deletion is considered to be a zero-substU tute and is governed by the meta-rule P. 1. 1. 53 ( alontyasya ). This rule says that, unless otherwise indicated, a substitute replaces the final sound [aL ] of the substituendum. Thus the lopa-deletion would apply to a of sa in a sequence like a-duh-sa-ta yielding a-duh ( s+φ) ta. Then a lopa-deletion can be applied to 8 by P. 8. 2. 26 (jhalo jhali). This rule says that an s is deleted, if it is preceded by a jhal sound and followed by a jhal sound. This would yield a-duh-φφ-ta, and finally adugdha. Similarly we can also derive adugdhāh and adugdhvam because t, th and dh are all jhaL sounds. Then the kāśikā vr̥tti considers a possible rpoblem and offers a way out.

P. 1. 1. 55 (sthānivad ādeso’nal-vidhau ) says that a substitute can be treated like the original substituendum, except in an operation involving phonemic features of the substituendum. The next rule P. 1. 1. 56 (acah parasmin pūrva-vidhau ) says that a substitute of vowel ( a ) conditioned by the following element is treated like the original vowel substituendum with respect to an operation applying to the preceding element. For example, let us say that the vowel V in the sequence xVy is substituted by S, and that this substitution is conditioned by the following y. Then the substitute S may be treated to be the vowel V, with respect to an operation applying to the preceding. For the rule that applies to x, the sequence looks like a Vy, rather than like xSy, though in fact the sequeuene is xSy. In the present case, the substitute [ =zero ] for a in a-duh-sa-ta [ >a-dhu-sφ-ta ] may be treated to be the original vowel a, with respect to the operation of deletion of by P. 8. 2. 26 (jhalo jhali). In that case, the sound is as if it is followed by a, and not by the jhaL sound t. Thus P. 8. 2. 26 will not take effect. If this happens, the desired forms with the deletion of s and a both cannot be obtained. The Kāśikā-vrtti says that this does not happen, because of the maxim that this procedure of sthänivad-bhäva " treatment of the substitute like the substituendum” does not operate with respect to a rule in the last three padas of the Aptadhyaye. Thus after φ is not considered to be the original a, because P. 8. 2. 26 belongs to the last three padas of the grammar. In this way, we can derive the forms adugdha, adugdhvam and adugdhah, with the deletion of sa, without having the specific luk-deletion in P. 7. 3. 73.

If Pāṇini intended only these forms, he could have certainly omitted the specific luk-deletion in P. 7. 3. 73. However, by omitting this specific luk deletion we cannot derive aduhvahi from the sequence a-duh-sa-vahi. Even though we could get lopa-deletion of a in a-duh-sd-vahi, still s cannot be deleted before vahi by P. 8. 2. 26 (jhalo jhali), since v is not jhal sound. Thus by not having luk-deletion in P. 7. 3. 73, we are unable to derive aduhvahi as an aorist form. But by putting the luk deletion in P. 7. 3. 73, Pāṇini indicates that he does intend the form aduhvahi. Thus v is included in the dantya “dental " class, and hence before va hi in a-duh-sa-vahi, the entire sa is zeroed by lok-deletion. P. 1. 1. 61 ( pratyayasya luk-śtu-lupah, adarśanam from P. 1. 1. 60 ) says that luk, ślu and lup refer to the deletion of the whole affix.

The supplementary argument of the Kāśikā-vr̥tti is that if only t , th and dh are to be included in the dantya " dental " initials of the middle affixes, Pāṇini could have saved a syllable by using the locative of tU, which stands for the t-series of stops, i, e. tau. Since Pāṇini uses the larger class of dantya " dental " sounds, it includes at least some dental sounds other than the t-series of stops. The only possible member is v. This is the entire argument of the Kāiikā-vr̥tti to prove that v was a labio-dental sound for Pāṇini.

Patañjali’s view

6 When we search the works of Kātyāyana and Patañjali, we find that the argument as presented by the Kāśikā-vr̥tti is a mosaic of various elements in these older texts, and many parts of this argument were not intended by either Kātyāyana or Patañjali. Some of the postulates in the argument of the Kāiikā-vr̥tti are not even intended by Pānini.+++(5)+++ Before we undertake a thorough examination of these older sources it must be noted that there is a clear indication in Patañjali’s Mahābhāṣya that he does not consider v to be a dantya " dental " sound. On P. 6. 1. 64 ( dhātv-ādeḥ ṣaḥ saḥ ), Patañjali talks about ṣopadeśa verbs. These verbs begin with s, but are presented in the Dhātu-pātha " root-list " with ṣ, for meta-linguistic purposes. P. 6. 1. 64 says that ṣ in the initial of these roots is replaced with s . Patañjali says that the ṣopadeśa verbs are those which begin with s, followed by a vowel ( aC ) or by a dantya " dental " sound, and also the verbs smi , svad, svid , svañj and svap , and excluding sr̥t sr̥j, str̥, styā , sek and sr̥.1 This statement is extremely important. If the sound v were a dental sound, Patañjali would not have listed verbs like svad separately.+++(5)+++ They had to be listed separately, because they were not covered by the general description (aj-dantya-parāḥ sādayaḥ). Kaiyaṭa who after the Kāśikā-vr̥tti is so much influenced by it that he is puzzled with the statement of Patañjali. He says that Patañjali has listed these roots separately, simply because he wants to remove the doubt that v might not be dental.2 This is obviously farfetched. Patañjali does not consider v to be a dantya " dental " sound.

7 If v is not a dental sound for Patañjali as demons then P. 7. 3. 73 cannot derive forms like aduhvahi from a-duh-Ksa-vahi. Patañjali gives the examples adugdha and adugdhāh (MB, Vol. I., Sec. I., pp. 318 and 328. ), but he has never given the example of aduhvahi , nor is there any indication that he was aware of any such form. The evidence in Kātyayana and Patañjali actually tends to show that they did not think of deriving this form, and thus were not thinking of v as a labio-dental sound.

Patañjali does not directly comment upon P. 7. 3. 73, and there are no vārttikas under this rule. But this rule has been discussed twice by Kātyayana and Patañjali in the context of rules discussing questions of sthānivad-bhāva " treatment of the substitute like the substituendum In what follows, we shall discuss these two contexts and their implications for phonetics of v.

8 As we have seen above, P. 1. 1. 55 (sthānivad ādeso’nal - vidhau ) says that a substitute is treated like the original substituendum, except in reference to operations which are based on the phonemic properties of the substituendum. For instance, a substitute of an affix, can be considered to be an affix, and a substitute of a verb-root can be treated to be a verb-root. But a consonantal substitute of a vocalic element cannot normally be considered to be vocalic. The next rule P. I. 1. 56 (acaḥ parasmin pūrva-vidhau ) gives an exception to the previous rule. It says that a substitute of a vowel, conditioned by the following element, may be considered vocalic, with reference to an operation applying to the preceding element. What is not clear in the rule* is whether we need this operation applying to the preceding element conditioned by the vocalic substituendum ( āśrīyamāṇāyām prakr̥tau ), or could it be conditioned by any element ( aviśeṣeṇa ). Kātyāyana discusses merits and demerits of both of these alternatives, and Patañjali explains these arguments.

9 ( aviśeṣeṇa sthānivad ) : If we say that to the previous sound could be conditioned following substitute of a vowel would be con says Katyftyana, we would have to have the 3. 73. We cannot simply do with constituting t previous rules. (2 ksa-lope lug~vacanam, Vārttika on P. 1. 1. 57 adugdha adugdhāḥ / lug vā duha-diha-liha-guhām ātmanēpadē dantyē iti MB, Vol. I., Sec. I., p. 318t ) The argument is as follows Kas-ta , suppose we apply lopa- deletion by luk-deletion from that rule, then by P. 1.53 (alo’ntyasya) get lopa-deletion only of the final a of Ksa , thus yielding the sequence a-duh-sφ-ta . Since this lopa-deletion is a 0-substitute for a, conditioned by the following ta, it would be treated like a, with reference to P. 8. 2. 26 ( jhalo jhali ), which prescribes lopa-deletion of s followed and preceded by a jhaL sound. If the φ-substitute is treated to be the substituendum a, then s is not followed by the jhaL sound t , but by a, and hence P. 8. 2. 26 would not be able to delete s, and we would not be able to get the correct form adugdha. To get at this form, we have no choice but to have luk-deletion in P. 7. 3. The luk- deletion of Ksa deletes the entire affix Ksa at once, and w the form adugdha. Thus, Kātyāyana argues that with the aviśeṣeṇa sthānivad " the previous operation may be conditioned element” we must have lōpa-deletion in P. 7. 3. 73 to derive correct forms, and we cannot just utilize the continuation of the lopa- deletio from the previous rules.

10 What the above discussion implies is that if we do not accept the postulate aviśeṣeṇa sthānivad , we could derive all correct forms simply by continuing the lopa-deletion in P. 7. 3. 73. The other postulate is that the previous operation must be conditioned by the following original substituendum, in order to consider the substitute to be like the substituedum ( āśrīyamāṇāyām prakr̥tau). With this postulate, if we have lopa-deletion of a of Ksa ( = sa ) in the sequence a-duh-sa-ta yielding a-duh-sφ-ta, we can easily apply P. 8. 2. 26 ( jhalo jhali ) to delete s. P.8.2.26 is conditioned by the following jhaL sound and not by the following original substituendum of φ, i.e. a . Therefore, the φ substitute cannoť be treated to be the original a. Hence s in a-duh-sφ ta is, immediately followed by ta, and thus s is deleted by P.8.2.26, yielding the desired form adugdha .

11 It is important to remember that Kātyāyana is trying to see if we can omit or must retain the luk-deletion in P.7.3.73, but we must assume that in both the alternatives he is bound to be able to derive the same correct forms. Thus the difference is only in the process of derivation, but no correct examples are expected to be sacrificed. However, if we look at the second alternative above, we find that there is no way we oan derive aduhvahi from a-duh-sa-vahi. Even if we include v in the dantya “dental” class for the sake of the argument, still when we bave the lopa-deletion of a of sa yielding a-duh-sφ-vahi, we cannot further delete φ by P.8.2.26. The sound v is not a jhal sound. Thus if we cannot derive aduhvati with the lopa-alternative, it must be assumed that Kātyāyana never thought of deriving it in the luk-alternative either. Patanjali clearly says that only in the postulate aviśeṣeṇa sthānivad, do we need luk-deletion in P.7.3.73, implying that otherwise we can derive all correct forms with lopa-deletion. 1 Thus aduhvahi was not in the picture.

12 There is another discussion in Kātyāyana’s vārttikas, which suggests the same conclusion. On P.1.1.57 ( na padānta.. ) Kätyāyana offers another postulate and examines its merits and demerits. The postulate is : “The procedure of treating the substitute like the substituendum does not apply with reference to an operation in the last three pädas of Panini’s grammar”. It must be noted that there is no direct rule in Panini’s grammar to this effect. This is a postulate offered by Kātyāyana for discussion. As a benefit of this postulate, Kätyāyana says that we can get rid of luk-deletion in P.7.3.73, and get by with continuation of the lopa-deletion from previous rules.

From the sequence a-duh-sa-ta, by continuation of the lopa-deletion in P.7.3.73, we get the sequence a-duh-sφ-ta. Now we need to apply the lopa-deletion to φ by P.8.2.26 (jhalo jhali). Since this rule belongs to the last three pādas of Pāņini’s grammar, the φ-substitute for a cannot be treated to be a with reference to P.8.2.26. Thus 8 is immediately followed by ta, and is deleted by P. 8.2.26, yielding a-duh-oo-ta. Kātýāyana says that with this postulate we can derive all the correct forms without having the luk-deletion, and this is echoed by Patāñjali.

However, we can derive only the form adugdha, adugdhāḥ, and adugdhvam. We cannot derive aduhvahi by the same process, because in the stage a-duh-sφ-vahi, s is followed by v which is not a jhaL sound. Therefore, P.8.2.26 ( jhalo jhali) will not apply to 8 before vahi. Since Kātyāyana and Patañjali say that we can drop the luk-deletion with the above mentioned postulate, they obviously were not interested in deriving the form aduhvahi. Another off-shoot of this discussion is that since Pāṇini does have the luk-deletion in P. 7. 3. 73, we may conclude that he did not intend the postulate pūrvatrāsiddhe na sthānivat. This postulate would have made th specific luk-deletion in P. 7. 3. 73 purposeless.

Thus from the above discussion we may conclude that Kātyāyana and Patañjali did not intend to derive the aorist forms like aduhvahi , and they did not consider v to be a dantya , “dental " sound. In fact Kātyāyana has listed many defects of the postulate mentioned above. ( tasya doṣaḥ saṁyogādi-lopa-latva-ṇatveṣu , Vārttika on P. 1. 58) If he thought that the form aduhvahi was a correct aorist form, he would hav certainly come up with a statement to that effect in listing the defect of that postulate. It is necessary to contrast this conclusion with th conclusion of the Kāśikā-vr̥tti. It is interesting to see how the Kāśikā vr̥tti constructs a mosaic of these very arguments to prove the opposite. The arguments in the Kāsikā-vr̥tti are in a sense taken out of the context and do not represent the intention of Kātyāyana and Patañjali. Thus closer we get to Pānini, swimming backwards in the tradition of the Pāninian commentators, more and more we are convinced that v was not a dantya " dental " sound for Pāṇini, and that he most probably did not intend to derive the aorist form aduhvahi by P. 7.3.73. This form nowhere attested in the Vedic or classical Sanskrit literature, and if at all someone wants to insist on this form, it can certainly be derived a past imperfect ( laṅ ) form, where deletion of sa is not involved.

13 We may also examine the compulsive force of the other argu- ment given by the Kāsikā-vr̥tti. It is said that since Pānini says dant " followed by a dental sound “, it must include sounds other than tho covered by the ř-series of stops. If Pānini intended to refer only to t affixes ta , thās and dhvam in connection with P. 7. 3. 73, and not include vahi , he would have simply said tau , i. e. locative singular of tU whic stands for the whole ¿-series of stops. In doing that he would ha saved a full syllable. Since he does not do that, it indicates that wanted to include dentals other than the ¿-series. The only other suc dental sound in the initial of the middle aorist affixes could possibly v of vahi.

The argument is well worded and yet it is based on an implicit assumption that Panini has never used over-extensive classes, when he could have used a smaller appropriate class. This assumption is true in general, and yet it is not universally valid. There are several cases of unnecessary over-extensive classes in Pāṇini. One glaring example is P. 1. 1. 48 (eca ig ghrasvādese). This rule says that when a short substitute is prescribed for eC, i. e. e, o, ai and au, iK sounds, i. e. i, u, r̥ and ḷ, are to be substituted. In fact i is the short substitute for e and ai, while u is the short substitute for o and au, while r̥ and ḷ are never the short substitutes for these sounds. Though r̥ and ḷ are unnecessary, still Panini has used the larger class. There are other cases of unnecessary use of the marker T in Pāṇini which have puzzled Päņinian scholars for centuries. 1 Again the argument offered by the Kāśikāvr̥tti involves circularity. In order to prove that v is included in the class of dantya “dental” sounds, we have to prove that the rule applies to vahi, and to apply the rule to vahi, we must know that v is dental. Thus, we need independent proof to justify that v is a labiodental sound.

Pāṇini’s view

14 So far we have discussed the views of the commentators on Pāṇini. Here I plan to present an argument which shows that Pāṇini himself thought of v as a labial sound, rather than as a labio-dental sound. By feeding both the phonetic alternatives in Pāṇini’s rules, we can see which alternative creates problems. P. 1. 1. 50 ( sthāne’ ntaratamah ) says that while choosing a substitute for a given substituendum from among a host of candidates, the one which is most similar to the substituendum should be selected. Traditionally it is understood that priority is given to the point of articulation classification of a sound in deciding this " most similarity “. This can be proved by studying the application patterns of Pāninian rules, though the traditional interpretation of the rule may create some interpretational problems.2

Some rules do not create problems, whether v be dental or labio-dental. For instance, let us consider the rule P. 6. 1. 77 (iko yaṇ aci). This rule literally says that iK sounds are replaced by yaN sounds, if followed by aC sounds, i. e. vowels. Then we expand the pratyāḥāras “short-forms” Thus we get : [ u, r , I ] are replaced by [ y, v, r, I ]. One may think that one could use P. 1. 3. 10 ( yathā-saṅkhyam anudśiaḥ samanām ) to effect here a sequential substitution. This rule says that given the same number of the substitutes and substituenda the replacement takes place one-to-one in the same order. However, we cannot use this rule to effect substitution in P. 6. 1. 77, because we would want this rule to apply to homogeneous sounds of i etc. as well. The only case we want to consider here is the choice between v and l as a substitute for ḷ-vowels. The sounds ḷ and I are both dentals. If v is labio-dental, still the dental ḷ is " most similar " to the dental ḷ, than the labio- dental v. Thus there is no conflict either way.

15 Now we shall consider a case where the labio-dental classification of v creates problems, while the pure labial classification avoids problems. In P. 1. 1. 45 ( ig yaṇaḥ samprasāraṇam ) we have the term samprasāraṇa defined for the substitution of yaN sounds by iK sounds. Expanding the short-forms we can state the substitution as : [y,v, r and l ] are substituted by [ i, u , r and I ]. As explained earlier, these sounds also stand for their savarṇa " homogeneous " varieties and hence we cannot have the simple sequential substitution ( P. 1. 3. 10 ). Here if v is pure labial, then there is only one pure labial choice among the substitutes, i. e. u. But if v is labio- dental, then there are two equally " most similar " choices, i. e. labial u and dental l. There is no featural basis for making a one-way choice in this case. Since Pāṇini never has l as the samprasārana substituteřfor v , we may clearly infer that v was a pure labial sound for Pāṇini. This example has been pointed out by M. M. Paṇḍita śivadatta Kuddāla, who firmly argues that Pāṇini classified v as a pure labial sound. ( 1 See : Note 8.) This example and its significance has also been discussed by my student Mr. James Bare in tion which will soon be submitted to the Linguistics Depar University of Michigan.

16 It is hoped that the above argumentation has clearly shown that Pāṇini believed v to be a pure oṣṭhya “labial " sound, is widely supported by the Prātikākhyas . That this was Kātyāyana and Patañjali has also been demonstrated. Somehow the Pāninīya-śikṣā came to elevate danta " teeth " from their usually accepted karaṇa " articulator " status to the status of sthāna “point of articulation " in the case of v . This created the image of double point of articulation classification for v, while at the same time creating unexpressed confusion about its articulator. Probably the Pāninīya śikṣā, a creation of post-Patañjali era, influenced Candragomin, the Cāndra-varṇa-sūtras. Influence of Candragomin on the a Kāiśkā-vr̥tti is an accepted opinion.1

1 " The authors of the Kāśikā-vr̥tti knew that grammar ( i, e. th vyākaraṇa ) and used it in the compilation of their own work. ( 1886 ), p. 184. The Cāndra-vr̥tti on Candra 6.1.101 (lug vā liha-guhām taṅi dantye ) does include the alternation aduhvahi / adhuksāvahi . Cāndra-vyākaraṇa , Pt. II., ed. by K. C. Chatterji, Deccan College, Poona, 1961, p. 293. I have discussed another case of borrowal by the Kāsikā-vr̥tti from the Cāndra-Vyākaraṇa in " A Note On Kāka-peyā Nadī”, Journal of the Oriental Institute, Baroda, Vol. XXIII, No. 3, March 1974, p. 160.

This explains why vr̥tti accepted v to be labio-dental. Once this was accepted, the justification of the aorist form aduhvahi in the Kāsikā-vr̥tti is a fight for the phonetic classification. However, this form is not intended by either Kātyāyana or Patañjali and is nowhere attested in the Vedic literature. The argument of the Kākikā-vr̥tti is blindly followed by the later tradition of the Pāninian scholars, and this was accelerated by the growing reverence for the Pāniniya-śikṣā. With due respect to the great teachers in the Pāninian tradition, it is necessary to misconceptions about Pāṇini’ s own phonetic classifications.

17 The aim of this paper is not to prove that for Pānini v did not involve teeth in any capacity or that it was a pure bi-labial sound. It could have been a bi-labial ; but we have no positive proof to that effect.

Like the Prātiśākhyas , he probably considered teeth to be the articulator of v , thus leading to a pure labial point of articulation classification for v. The Pāninīya-śikṣā elevated teeth uniformly to the status of a point of articulation, and thus presented v as a labio-dental, with double point of articulation classification. The result is not a substantive difference, but a significant classificatory difference which affects formulation of rules in grammar.+++(5)+++ A comparable case is that of the pronunciation of short and long a. The RPr considers all vowels, spirants and m as having continuous non-contact as their internal effort ( asprstam sthitam ) meaning that a and ā both have the same internal effort of non-contact.1 On the other hand, Pāṇini considers a to be a closed ( samvrta ) sound and ā to be an open ( vivrta ) sound, a classification which is based more on the difference in the size of a smaller and a larger gap, rather than on the denial of a gap or non-contact in the case of a. The śikṣā quoted by Uvata on the VPr says that vowels involve non-contact ( asprṣṭa ), short a is closed ( saṁvr̥ta ) and other vowels are open ( vivr̥ta ). This clearly shows that closed and open internal efforts are both subsumed under the single category of non-contact.3 However, if Pāṇini had accepted non-contact ( aspr̥ṣṭa ) classification for vowels, it would have accorded the same effort to a and ā, and it would not have been necessary for Pāṇini to write his controversial rule P. 8.4.68 ( a a ).4 Thus a pure classificatory difference can also be significant in the formulation of a grammatical system, and hence in the interpretation of a given grammatical system.

1 svarānusvāroṣmanām aspr̥ṣṭaṁ sthitam , RPr , 13.3
2 saṁvr̥tāsya-prayatnaḥ akāraḥ , vivr̥tāsya-prayatnā itare svarāḥ / aspr̥ṣṭāsya-prayatnāḥ svarāḥ / , Uvata on the VPr 1.72, Madras edn., p. 29.
3 For a detailed treatment of this question, see my " New Material On The Kautsa-vyākaraṇa " , appearing in Journal of the Oriental Institute , Baroda, 1975 issues.
4 K. C. Chattopadhyaya (1974 ) argues that for Pāṇini a , like ā, was a vivrta " open " sound. He claims that P. 8. 4. 68 {a a) was not a part of the original Aṣṭādhyāyi , but was inserted in later times, when a became a closed sound under the influence of the Dravidian contacts. I disagree with this view. For my detailed arguments, see “.Phonetics of Short A in Sanskrit”, appearing in the 1975 issues of the Indo~Iranian Journal , and also my forthcoming monograph : Critical Studies in Indian Grammarians , /, Theory of Homogeneity [ sāvarnya ], Centre of South and Southeast Asian Studies, University of Michigan, Ann Arbor, ( 1976 ).

BIBLIOGRAPHY

Primary Sources

  • Dantyoṣṭhyavidhi, the fourth Lakṣaṇa treatise of the Atharvaveda , ed, by Ramagopala Sastri, Dayānanda Mahāvidyālaya Samskrta Granthamālā , No. 4, Lahore, 1921.
  • Kāsikā-vr̥tti , with the commentaries Nyāsa and Pada-Manjarī , in 6 vols. Pracya Bharati Prakashan and Tara Publications, Bañaras, 1967.
  • MB Mahābhāṣya of Patañjali, in 3 vols. Motilal Banarasidass, Delhi 1967. [ with the Pradīpa of Kaiyata and Uddyota of Nāgeāabhatta. ]
  • RPr Rgveda-prātiśākhya , text only, ed. by M. D. Shastri Bañaras, 1959.
  • RT Rktantra , by Sākatāyana, ed. by Surya Kanta, Lahore, 1933, reprinted from Delhi, 1970.
  • SCA śaunakiyā Caturādhyāyikā , ed. and tr. by W. D. Whitney, New Haven, 1862.
  • Siddhānta-kaumudī of Bhattoji Diksita, ed. and revised by Wasudev Laxman Sastri Pansikar, 11th edn., Nirnaya Sagara Press, Bombay, 1938.
  • śikṣā-samgraha , a collection of various śiicsa texts, Bañaras Sanskrit Series , Bañaras, 1893.
  • śiksāsutrāni9 by Āpiśali, Pāṇini and Candragomin, ed. by Yudhiṣṭhir Mīmāṁsak, Ajmer, samvat 2024.
  • TPr Taittirīya-Prātiśākhya , [ 1 ] with the commentary Tribhās - yaratna , ed. and tr. by W. D. Whitney, New Haven, 1868 ; [ 2 ] with the commentaries Tribhāsyaratna and Vaidikabharaṇa ed. by K. Rangacharya and R. Rama Sastri, Government Oriental Library Series , Bibliotheca Sanskrita9 No. 33, Mysore, 1906 ; [ 3 ] with Padakramasadanabhāṣyat Madras University Sanskrit Series , No. 1, ed. by V. Venkatarama Sharma, Madras, 1930.
  • VPr Vājasaneyi-prātisālchya, with the commentaries by Uvata and Anantabhatta, ed. by V. Venkatarama Sharma, Madras University Sanskrit Series f No. 5, Madras, 1934,

Secondary Sources

  • Allen, W. S. (1953) Phonetics in Ancient India , London . Oriental Series , No. 1, Oxford University Press, 1953.
  • Bare, James ( 1975 ) Phonetics and Phonemics in Pāṇini , Ph. D. Dis- sertation, Department of Linguistics, University of Michigan, Ann Arbor, 1975 ( unpublished ).
  • Cardona, George (1964) ‘The Formulation of Pāṇini 7. 3. 73’, Journal of the Oriental Institute , Baroda, 1964.
  • ChAttopadhyaya, K. C. ( 1974 ) ’ Did Pāṇini Envisage " A " as a Closed (samvrta ) Sound Î ’ Oharudeva Shastri Felicitation Volume , Delhi 1974.
  • Deshpande, Madhav ( 1972 ) c Pāniniarç Procedure of Taparakaraṇa : A Historical Investigation ‘, Zeitschrift fūr Vergleichende Sprachforschung , Band 86, Heft 2, 1972. tion in Pāṇini ‘, Silver Jubilee Research , University of Madra ir anian Journal. Journal of the Oriental Institu I, Theory of Homogeneity , T and Southeast Asian Langua Centre of South and Southeast Michigan, Ann Arbor, ( 1975 ). Ghosh, Manmohan ( 1938 ) Pāninly five recensions , with Intr., Calcutta, 1938. Kielhorn, Franz ( 1886 ) c The C vr̥tti ’ The Indian Antiquary , Luders, Hermann ( 1894 ) Die Vyāsa ltnis zum Taittirīya-prātisākh MīmāṁSAK, Yudhishthir ( samvat Itihāsa , vol. III., Ajmer, samv
  • Phatak, Madhukar ( 1972 ) Pāninīya-śikṣāyāḥ śi Samīlcsā, Bañaras, 1972.
  • Raghu Vīra ( 1931 ) ‘Discovery of the Lost Phonetic JRAS , 1931.
  • Thieme, Paul ( 1935 ) Zur Datierung des Pāṇini, ZD
  • Varma, Siddheshwar ( 1929 ) Phonetic Observations of Indian Grammarians, London, 1929, reprinted from Delhi, 1961.
  • Wackernagel, J. ( 1914 ) Zur Bildung des 7, Aorists im Altindischen, KZ , 1914.
  • Whitney, W. D. ( 1888 ), Sanskrit Grammar , 2nd end., New Haven, 1888, reprinted from Delhi, 1962.

  1. Vakārct8ya dantoṣṭham ( sthānam ), Siddhānta-Kaumudi od P. 1.1.9, p. 2 ; dantyosthyoypi vakāro dantya iti grhyate /, Kāḥikā-vr̥tti on P. 7.3.73, Vol. VI, p. 82. ↩︎

  2. Manmohan Ghosh ( 1938 ) includes dantyosthyo vah smrto budhaih into the so called Reconstructed Version of the Pāniniya-śikṣā ( p. 2 ) as verse 12b. It is found in all the versions published by Ghosh. ↩︎

  3. See : Kāsikā-vr̥tti on P. 7. 3. 73, Vol. VI, p. 82. A similar discussion is found in Patañjali on P. 1.-1.8 ( mukha-nāsikā-vacano’nunāsikah ) : ye ubhayavacanāḥ , grhyanta èva te mukha-grahanena nāsikā-grahanena ca / MB, Vol4 I., Sec. I., p. 151. ↩︎