Paninian Studies

Author(s): G Cardona
Publisher: G Cardona
Catalog link: https://jainqq.org/explore/269545/1

Page #1

Påninian Studies*
G. Cardona 1. Påņinian studies have continued to flourish in recent years. The work done spans all major aspects of the field. A new English translation of the Aşpadhyāyi has recently appeared (Katre 1987), and an excellent introduction to Panini's system has been published (Junnarkar 1977; see also Bhagavat 1985). In addition, the first parts of projected lengthy works on Pāṇini, his methodology, and traditions associated with his work have recently appeared (Sharma 1987, Cardona 1988; see also Filliozat 1988). Translations into English and French of the Mahabhasya and its major commentaries (Joshi -Roodbergen 1975, 1976, 1980, 1981, 1986; Filliozat 1975-1980, Yagi 1984) continue; M. S. Narasimhacharya's extremely valuable edition of commentaries on the Mahabhäşya and the Pradipa, most importantly the Prakāśa and Annambhatta's Uddyotana has been completed (1973-1983); recently, a series of insightful articles by Devasvarupa Misra (1978), in Sanskrit, on the Mahabhäşya was published. Further, a new edition, with English translation and notes, of Bhartshari's important Mahābhâsyadipika (Palsule 1983, Limaye-Palsule-Bhagavat 1984, Bhagavat-Bhate 1986, Bronkhorst 1987a) has almost been completed, and the text has been made available in a facsimile edition of the only available manuscript (see Dipika). Considerable valuable work continues to be carried out on Bharthari's Vakyapadiya. A critical edition of the Kartka text is now available (Rau 1977); the second kanda with Punyarāja's commentary and the extant part of the svopajslavịtti has been edited anew (Subramania lyer-Aklujkar 1983); Rau (1981) has edited and annotated the Vakyapadiyaprameyasargraha; K. A. Subramania Iyer's translation of the second kända has been published (1977); Raghunatha Sarma (1977, 1979, 1980) has completed his commentary on this major work, together with the first part of his discussion on variant readings, a new translation, into Gujarati, was recently published (Shukla 1984); and A. N. Aklujkar's critical edition of the svopa jna vitti together with an index, is nearing completion. Editions of other important texts in the Påninian traditions are also being made available. Thus, Haradatta's Padamañjari was recently edited anew (Ramachandrudu-Sarma 1981); Ramacandra's Prakriyakaumud has been reedited together with Srikssna's Prakasa (Muralidhara Misra 1977-1980); new editions of Kaundbhatta's Vaiyakaraṇabhūşaņa have appeared (M. Bhattacharya 1985, Misra 1987); Nagesa's Vaiyakarañsiddhantamanjaşa has recently been edited twice (Sukla 1977, Kapil Dev Shastri 1985); the Paramalaghumanjūşa has again been published, with Sanskrit commentary and Hindi translations (Kapil Dev Shastri 1975, Jaya Shankar Lal Tripathi 1985) as well as with a Marathi translation (Bhagavat 1984), and a fully critical edition of this work, with an annotated English translation, is soon to be completed (Cardona forthcoming b). In addition, a good annotated translation of an important section in Kaundabhatta's Vaiyakaraṇabhūsana has been published (Gune 1978). The active state of research in the field can readily be seen from such recent monographs as those of B. Bhattacharya (1985), Bronkhorst (1987) Cardona (1983), Deshpande (1982, 1985), Kiparsky (1979, 1982), Subramanya Iyer (1982), as well as a collection of articles such as in Joshi-Laddu (1982), and other works mentioned in the appended bibliography, which is by no means complete.
The work carried out in the last decade or so of course represents, in great part, a continuation

Page #2

New Horizons of Research in Indology
of work done by predecessors. It is also quite varied, both in scope and in quality, so that it would be neither possible nor advisable to attempt an appraisal of even a large portion of this work in a brief article like this'. Instead, I think it proper and useful to concentrate on some works that illustrate trends which I see continue to be reflected in recent work and to assess their results, their premisses and their promises for future research.
50
2. Scholars have long been interested in the history of the Aṣṭadhyayi, and a great deal has been written on interpolations in the text?. The tendency to find in Panini's work evidence of massive interpolations made by post-Paninfan grammarians has continued.
2.1. One of the major reasons claimed for considering that rules contained in the Aṣṭādhyāy! as known to Pāṇinīyas represent additions to an earlier version of the text is that the received textual verison shows inconsistencies. Of course, what is claimed to be inconsistent may well appear to be so only in the limited vision of the scholar putting forth the claim. It will be methodologically instructive to consider a recent claim of truly massive interpolation, according to which all the sartas concerning compounds and derivates with taddhita affixes are later additions to the AstädhyayT. As the authors who defend the thesis remark, this 'involves major surgery in the present text of the Astadhyayı
The principal reason for adopting this position is the authors' perception of inconsistency in terminology. Thus, rules providing for compound formation make use of terms which refer to the semantic sphere of a compound, and of terms referring to items that terminate in endings of particular triplets. For example, Aṣṭadhyay! 2.1.24: feder Brachaafdariakerasians; provides that a pada terminating in a second-triplet (accusative) ending (dvitiya) optionally combines to form a tatpuruşa compound with a related pada that includes one of the terms śrita resorted to', atta 'gone beyond, surpassed,' patita 'fallen', gata 'gone', atyasta 'thrown beyond', prăpta reached, attained', apanna 'reached', e.g. kasta-śrita 'one who has undergone hardship' is derived by combining the padas kasta-am and śrita-s, the former including the second triplet ending am; the compound alternates with a sentential expression kastam śritah, with two independent padas. Aştadhyayı 2.1.26: gar also provides for a compound with a pada containing a second triplet ending, namely one with the term khatva. In addition, the rule specifies that the compound is formed if the sense of pejoration, censure (kşepe) is conveyed, as in khatvaradhah 'one who has climbed into bed", used with reference to a student who, contrary to established practice, sleeps in a bed instead of on the ground during his studies or who marries before receiving proper permission. Again, Aṣṭadhyayı 2.1.37: 45 lets a nomimal with a fifth-triplet (ablative) ending (pañcam) optionally combine with a related form of bhaya 'fear' to form a tatpuruşa compound, as in caurabhayam (-caura-bhyas-bhaya-s) 'fear of thieves.' The nominal endings contained in padas such as kaşta-am, khatva-am, caura-bhyas are introduced on condition that particular participants in actions, as assigned to given syntacitic-semantic categories, are to be signified. For example, Astadhyayt 2.3.2: for fear introduces a second triplet ending after a nominal if a karman is to be signifed, and 2.3.18: introduces a fifth-triplet ending after a nominal when an apadana is to be signified. Other rules of the Aṣṭadhyayī assign participants to particular karaka categories under specified conditions. Thus, Aṣṭadhyay 1.4.24-25: ध्रुवमपायेऽपादानम्। भीत्रार्थानां भयहेतु: provide that a kāraka which functions as point of departure is

Page #3

51
G. Cardona : Pāṇinian studies
assigned the class name apădāna, which also applies to a käraka that functions as source of fear with respect to acts denoted by verbs with the meanings of bhi 'fear', tra 'protect;1.4.49-50: कर्तुरीप्सिततमं कर्म। तथायुक्तं चानीप्सितम् assign to the karman category a karaka which functions as the primary goal of one classed as agent as well as a karaka that is related similarly to an action but which an agent either actively seeks not to reach or towards which he is neutral. For example, gråmat ( - grāma - as) 'village' in AMIGMora ... 'is coming from the village' and caurebhyas (caura-bhyas) in a fa ... 'is afraid of thieves refer to a village and thieves' assigned to the apădâna category; kasta-am refers to difficulty, hardhship assigned to the karman category. Now, a compounding rule such as Aşțădhyāyī 2.1.24 or 2.1.37 refers to padas that terminate in paricular triplets of nominal endings, not to kärakas: The rules are formulated as shown, not as *Pata * 37416F . On the other hand, sūtras which have to do with introducing kệt affixes do use käraka category labels. For example, Aşțădhyâyf 3.4.67: R H concerns kļt affixes in general: These are introduced to singify an agent, as in karty-, Käraka - doer, maker, with the kịt suffixes tộc pvul. On the basis of such difference in terminology, the claim has been made that rules in the section of the Aştādhyāyī dealing with compound formation and those dealing with krt affixes need not stem from the same author.
Another sort of inconsistency is perceived in the organization of groups of sūtras in the Aşgadhyāyī. Thus, rules such as 1.4.24-25, 49-50 ... part of the block of sūtras 1.4.23-55--- serve to assign kärakas to particular classes, and these classifications serve to condition the introduction of nominal endings, by rules such as Aşțădhyāyt 2.3.2, 18. But the two blocks of rules are not contiguous. And this is considered to be an inconsistency.
2.2. The need to see inconsistency in rules such as those considered in section 2.1 reflects, I think, an exaggerated eagerness to find difficulties where there are none, on the part of scholars who do not come to grips with principles of organization that are not all that difficult to perceive.
Pāņini's entire derivation's system reflects well founded principles that are fairly patent. He operates with two kinds of bases: verbs (dhatu) and nominal bases (prātipadika). Each of these is either primitive or derived. For example, krdo, make and go‘cow, bull', are primitive bases, but the desideratives cikīrṣa 'wish to do, make' and the causative kāri'have ... do, have .... make' are derived verbs, and kartı-, kaşta-śrita-are derived nominal bases. Moreover, Påņini perceptively derives such bases from other elements throug! a formal machinery. Thus, he introduces postnominal endings after nominal bases to arrive at padas such as kașța-am and śrita-s. From these, through applying phonological rules, one can arrive at strings such as kastam śritaḥ Alternatively, the related padas can be bracketed together in a compound, the endings in which are then deleted: Kasta-am-śrita-s - kaşta-śrita. To form a pada such as kasta-am in the first instance, Pāņini introduces an ending (Astādhyāyi 2.3.2 (see 2.1)) after a nominal base; and a derived base such as kastaśrita can then enter into the same derivational process whereby an ending is also introduced after it under specified conditions. Now, endings such as am in kasta-am of kasta-am śrita-s are introduced on condition that particular kärakas assigned to given categories are to be signified. Once a pada like kaşta-am has been formed, it is of course quite proper to refer to it as an element that terminates in a second-triplet ending (dvitīyā). It is understandable that a rule such as Aștadhyāyt 2.1.24 (see 2.1) has dvitiya instead of karma Indeed, since Pāṇini's

Page #4

New Horizons of Research in Indology
52
compositional rules operate on padas, it is only proper that he formulates the rule in this manner. Far from reflecting any sort of inconsistency due to interpolation, then, the difference in terminology among the rules in question reflects a well wrought and organized system of derivation.
Similarly, it should not surprise one that rules which introduce nominal endings be separated from sūtras that assign kärakas to categories. The latter are put in a section of the grammar that deals with classes which are generally disjunct, and these classes involve not only karakas but also linguistic elements such as particles (nipata) and compounds. All such rules are placed in a single section of the grammar, governed by a heading which provides that a given entity may bear only one claşs name unless otherwise provided for. The sūtras which introduce vibhaktis after nominal bases do not belong in this block of rules. Again, therefore, there is no inconsistency pointing to interpolation.
Recent claims of massive interpolation in the received text of the Aşgadhyāys thus not only continue a long-standing tendency (see note 2) to find evidence of such interpolations on a large scale, they also continue to reflect on the part of scholars who insist on finding such interpolations an inability or unwillingness to perceive in the work clear and well conceived principles of organization. It would be well to concentrate on discerning and understanding more fully such principles and the bases on which they are founded rather than to go on in a quest for purging the Aştādhyāyt of perceived massive interpolations
3. Another aspect of Pāņini's grammar that has recently attracted considerable attention is the manner in which he relates rules to each other. Certain claims made, moreover, may fruitfully be considered not only in connection with Pāṇini's principles of organization but also with respect to our attitudes towards the commentatorial traditions associated with the Astādhyāyi
3.1 Let us consider the arguments presented in vārttikas 1-5 on Astādhyāyi 6.1.86: tagarria:, which concerns a relation between rules and operations of the section headed by 6.1.84: 90: uput: Rules of this section let a single substitue replace two contiguous substituends. Such a rule' is suspended (asiddha) with respect to rules that provide for ș replacement and for the insertion of a final augment tuk. In his first vårttika on 6.1.86. Katyāyana notes that this suspension serves two purposes : 1) To preclude the application of what would take effect as conditioned by a replacement (adeśalakşaņapratişedhartham); 2) To allow the occurrence of what is conditioned by an original element that is subject to replacement (uisargalaksanabhåvartham)" Standard examples for 1) and 2) as given by Patanjali and other Pāņinīyas, are: Real ( 3146) 'Who did the irrigating?!and adhitya ‘after studying' (adhi-iya-adhi-itva).
There are also two possible interpretations concerning what is suspended: 6.1.86 provides that (A): the result of an operation (karya... 'to be carried out') that has taken effect is suspended (käryasiddhatva); or it provides that (B): a rule (sastra) is suspended (sastrăsiddhatva). I shall speak of (A) operation suspension and (B) rule suspension.
For instances of (1), the distinction between the two types of suspension is not crucial. Thus assume that 6.1.86 provides, under (A), that the single substitute -o- in kosincat is suspended with respect to the possible replacement of -s-by - 13 All this means is that the replacement of au by- o takes effect, but then is suspended with respect to the substitution of - & for-S,

Page #5

53
G. Cardona : Paninian studies so that the latter does not take effect after -o has come about. Under (B), 6.1.86 provides that the rule which serves to replace -o a- by -o-1 is suspended with respect to the rule that has -replace- (Aṣṭadhyay 8.3.59, see note 13). In effect, then, rule 6.1.109 ceases to exist in relation to 8.3.59, so that, although it does indeed apply when its conditions are met, it does not thereby provide a context in which 8.3.59 can apply.
For instances of (2), on the other hand, the distinction between (A) and (B) is pertinent. Given -T-ya, in which -i- occurs before the affix yap, the augment fuk can be added (-T-ya-it-ya), by the rule which provides for that this augment be added to a short vowel preceding a kṛt affix marked with p.' If, however, the contiguous vowels of adhi-iya are replaced by the single long vowel -I- (adhi-iya-adhlyaf a condition for adding the augment tuk is eliminated, since there is no longer a short vowel before the suffix lyap. Moreover, even if the augment is added first, the long-vowel replacement can take effect (adh-i-it-ya- adhitya), but the converse does not hold, as shown. Long vowel substitution and addition of the augment tuk, then, are related as what Paniniyas call nitya and anitya operations. And a nitya operation regularly takes precedence over one which is anitya." Accordingly, unless a special provision is made, a general principle of the grammar dictates that one have a form (adhlya) in which undesiredly, the augment tuk cannot be added. Panini therefore provides for suspension in Aṣṭadhyay 6.1.86.
Now, under (A) (operation suspension), this sutra simply states that the result of replacing two vowels with a single long vowel such as I is suspended with respect to the addition of the augment. In his second värttika," Katyayana remarks that the operation supposed to be conditioned by the substituend in question does not succeed (utsargalakṣaṇāprasiddhiḥ), since the substituend is absent (utsargabhävät). And, if one should suppose that this operation does succeed simply by virtue of stating that single-vowel substitution is suspended (asiddhavacanāt siddham iti cet), this is declared improper, since merely stating that one thing is suspended does not result in the occurrence of the other, which it has eliminated (nanyasyasiddhavacanãd anyasya bhavaḥ)1 That is merely declaring that the -- of adhiya is suspended does not restore the original short í required for the augment to be added. 20 Hence, it is suggested, two things must be provided for: suspension for deriving sequences such as ko'sificat and letting a replacement have the status of its substituend, in order to account for derivates like adhitya.21
This argumentation is clearly based on the assumption that Aṣṭādhyāyī 6.1.86 provides for suspending the results of operations. For in the next värttika,22 Katyayana goes on to say that letting the replacement have the value of its substituend has no purpose (sthanivadvacananarthakyam), because what is provided for is the suspension of rules (sasträsiddatvät). As is patent from what Katyāyana has said, moreover, he cannot envision, under this alternative, that Aṣṭadhyay 6.1.101 (see note 16) should take effect to yield adhiya. On the contrary, having this rule suspended with respect to Aştädhyay 6.1.71. (see note 15) has the effect of letting 6.1.71 apply first (adhi-iya-adhi-itya-adhitya), as Paniniyas have recognized13 Katyayana alludes to what he said in his comments concerning 6.1.863 and in his Bhasya on this Patanjali illustrates how the two purposes of suspension are served. As examples for (2), he gives the forms amuṣmai, amuṣmāt, amusya, amușmin, dervied from the pronoun adas 'that' with the basic endings ne, nasi, nas,
3.2 In his eighth värttika on Aṣṭādhyay 8.2.1:

Page #6

New Horizons of Research in Indology
54
di (respectively, dative, ablative, genitive, locative singular). These endings should be replaced by smai, smat, sya smin: adas-e-> adaa-e -ada-es ada-smai - amu-smai → amuşmai, adas-as... ada-as → ada-smat... → amuşmát, adas-as... → adas-as... ada-sya ... → amuşya, adas-i... → ada-i... → ada-smin... → amușimir? Now, Aştadhyayi 8.2.80 (see note 26) provides for two substitutions in the pronominal adas once the final -s of this item has been replaced: an u-vowel substitutes for the sound following -d, and -d-is replaced by -m-. If these substitutions were to take effect, giving amu-e and so on, the endings following the pronominal stem would not be subject to replacement by smai and so on, since these substitutions apply after stems in short-a Hence, Pāṇini puts 8.2.80 in the section of rules headed by 8.2.2. A rule of this section is suspended (asiddham) with respect to a preceding rule (pūrvatra), so that 8.2.70 is suspended with respect to 7.1.12 and so on. Under rule suspension, this means that rules of the first seven and three quarter adhyāyas of the Astădhyay apply before rules of the Tripadt, so that the order of applying rules in deriving amuşmai and so on is as I have presented it.28
Since, as shown, in 8.2.1 vårttika 8, Katyayana alludes to what he said earlier, it is to be assumed that, as eariler he concluded that Așțădhyāyf 6.1.86 provides for rule suspension, so also must he invoke rule suspension for 8.2.1. And later Paņinīyas explicitly say that Aștādhyāyr 8.2.1 provides for rule suspension. For example, in its explication of the sūtra, the Kašikāvștti remarks that the three-fourths of an adhyaya headed by this rule is suspended with respect to the preceding seventh and three fourths adhyayas and that in the Tripadt itself each subsequent rule is suspended with respect to a preceding one; moreover, the text explicitly uses the term yoga 'rule'.29 Commentaors on the Kaśikāvstti point out that the very wording used here shows that rule suspension is invoked. 30
In addition, commentators consider the implications of the very term asiddha used here. As Jinendrabuddhi remarks, siddha means which has come about, is an accomplished entity (nişpanna) and asiddha thus refers to something that has not come about. How, then, one may ask, is it possible to say that a rule (śästra) which has indeed been formulated and stated ir, the Astādhyāyi be referred to as asiddha, since no number of statements that something thus formulated has not come about can make it not to have come about. It is because of this, both Jinendra and Haradatta note, that the Kaśikā goes on to say that a rule is treated as asiddha (asiddavat), that is, does not cause an operation that is brought about.?!.
In sum, a sūtra of the Tripādt is indeed an accomplished entity, a rule which has been formulated and is part of the corpus. By saying that such a rule is asiddha with respect to a preceding statement, then, Påņini simply means that the sūtra in question is declared nonexistent with respect to that other rule although it is indeed a real part of the grammar and cannot be denied as such. Hence, I have used 'suspended as a rendition of asiddha. Of course, once a sūtra is thus suspended, its effect is also suspended.
3.3. In vårttika 4 to Astādhyāyt 6.1.86 (see 3.1 with note 21), Katyāyana shows that he draws a clear distincition between the result of an operation being suspended (asiddhatva) and a replacement's having the status of its substituend (sthanivadbhava). The same clear distincition is drawn in another context.

Page #7

55
G. Cardona : Påņinian studies
Astădhyāyi 1.1.57: 377: rydfat (PIAGIERT: 56) provides that the substitute for a vowel replaced in a right context has the status of its substituend with respect to an operation relative to an element that precedes the original vowel. In the second vārttika on this sūtra, it is suggested that a special provision must be made to forbid ar operation that is conditioned by a replacement for a vowel in a right context.32 For example, the -u- of vayu-os'winds' (gen. -loc. du) is replaced by -v- before the vowel of the dual that follows." There is also a rule whereby v, y are deleted before a consonant other than y. Kátyāyanas argument here is that, although the extension rule 1.1.57 lets a substitute such as the -v- of våyu-os have the status of the original vowel, it does not also deny this semivowel its actual status, so that this could indeed condtition the deletion of the preceding -y- undersiredly producing forms like 'vāvoḥ instead vāyvoń. The suggestion is then made that the desired results are established if one provides that a vowel replacement which is conditioned by a right context (ajadeśaḥ paranimittakaḥ) is suspended (asiddhaḥ) with respect to an operation on what precedes.3s Clearly, vayvos is an instance where suspension is supposed to prevent what would be conditioned by a replacement.36 Equally clearly, Katyāyana and Patanjali make a sharp, well defined distinction between an extension rule, whereby a property of an original element is extended to a replacement for that element, and a suspension.
3.4. Consider now another sūtra that provides for suspension. Astādhyāyi 5.4.22: 35R1696 HTC whereby whatever results from applying a sütra of the section begining with this rule and extending through the section headed by 6.4.129: HRT is tcated as suspended 37 with respect to any other operation provided by a sūtra of this group, provided the operations in question have the same conditioning element.
Three examples will suffice to illlustrate the procedure. śādhi, the second - singular imperative of śas instruct, command' is derived from śās-hi,38 and the third-plural imperfects asan 'they were' ayan 'they went' derive from as-ant, i-ant." Given śás-hi, two operations are to be consdiered: The ending hi conditions the replacement of śās by sa (6.4.35. STT E (STRA: 34)); hi is replaced by dhi after hu 'offer oblations' and verbs that end in non-nasal stops or spirants (6.4.101: CHATT STEF:) Similarly, pairs of operations are to be considered for as-ant, i-ant. a of the stem as -is subject to deletion (6.4.111 TRA14:1); and the stem iis subject to replacement by y before the vowel of an ending (6.4.81: JUTT JATI). In addition, since the forms in question have an ending that derives from the L-assix lan (see note 39), the stems are subject to receiving an initial augment: Before endings that derive from lur, lan, Iță, consonant-initial stems are augmented with high pitched á (ad udattah), vowel -intitial stems (ajādinām), with a (at)6.4.71-72:
FG TEETGGIT: 316UTCH I In each instance, one of the operations is nitya (see note 17) with respect to the other, so that it takes precedence. Thus,if the ending of śās-hi'is first replaced by dhi, the preceding stem is still subject to replacement by śa, since the substitute is given the status of the original element. 40 On the other hand, once śa substitutes for śās, the replacement of hi by dhi can no longer apply: This is conditioned by a sound, so that Aştadhyāyt 1.1.56 (see note 40) does not give śå the status of the original śas with respect to this operation. Similarly, whether or not the augment át is added to the stems of as-ant, i-ant, as- is subject to having its vowel dropped, and semivowel replacement applies to i, but once these operations take effect, the stems no longer begin with a vowel, so that åt cannot be added. Hence suspension is brought
amples will suffice to illustratchishi8 and the third-plural operations are to be

Page #8

New Horizons of Research in Indology
into play. Moreover, all the examples I have given are clearly of the type (2), where suspension serves to allow an operation conditioned by an original element that is subject to replacement. As I noted earlier (see 3.1) in such instances Katyayana considers letting a replacement have the status of the substituend and then says that the Papinian rules in question provide for rule suspension. In the present instances, Aṣṭādhyāyl 1.1.56 cannot apply to give the substitutes -, y- the status of their substituends with respect to the required operations, since the latter are conditioned by sounds.
56
Now, let us suppose that here too one should operate with rule suspension. One would then let 6.4.35 be suspended with respect to 6.4.101, which would apply first: s-his-dhi→ sadhi To derive asan, ayan from as-ant, i-ant on the other hand, one would have to let 6.4.111, 81 be suspended with respect to 6.4.72, which would apply first: as- ant a→ as- ant → äsan (8.2.23: :1), -i-anti-antay-antayan. Clearly, in his set of rules as ordered, Panini cannot formulate a suspension rule for all such instances in a single way comparable to Aṣṭadhyay 6.1.86 or 8.2.2. In order to specify what rule is suspended with respect to what other rule, he would have to bring all instances of the types sahi and ayan together in separate sections, headed in one case by a rule stating that a rule is suspended with respect to a following rule (uttaratra), in another case by a heading letting a rule be suspended with respect to a preceding rule11
It is understandable, then, that in Aṣṭādhyay 6.4.22 Panini says asiddhavat, not asiddham. That is, we have here instances of operation suspension. The operations which take precedence by virtue of general principals are allowed to take effect: śās - hi→ sa - hi, as-ant → s-ant, i-ant → y - ant. The effects are then suspended, so that sa, s-, y- are given the statues of sās, as-, i- in order to allow the replacement of hi by dhi after sa and the addition of the initial augment af to sy. This is obviously comparable to the procedure whereby Panini lets a replacement have the status of its substituend (Aştädhyay 1.1.56, note 40). In both instances, Panini uses terms with the suffix vati: asiddhavat, sthänivat. Morevorer, just as Aṣṭādhyāyl 1.1.56 provides for operations to be extended to substitutes (käryätideśa), so 6.4.22, under the interpretation given here, lets opeations be suspended.
3.5 It is noteworthy that Paniniyas regularly interpret asiddhavat of Aṣṭadhyay 6.4.22 in the same manner as asiddhaḥ of 6.1.86 and asiddham of 8.2.1. In his first värttika on 6.4.22 (зffen 311), Katyayana merely alludes to what was said in the first värttika on 6.1.86, and he does not go on to discuss the question of rule suspension versus operation suspension, thus leaving the impression that he accepts here the conclusion drawn carlier. Other Paniniyas explicity remark that rule suspension is accepted here. Of course, Panini does not specify here which rule is suspended with respect to which other rule. Consequently, under the assumption of rule suspension, one would have to assume that, for example, 6.4.35 and 6.4.101 are both suspended with respect to each other. This recognizes that the operations provided for in these rules are mutually dependent. It also has the undesirable consequence that neither rule could now interact with the other. To avoid this consequence, one might assume that, by the general principle whereby a nitya operation takes precedence over an anitya operation, 6.4.35 does indeed have to apply first, and that rule suspension in this instance means that this rule is suspended with respect to

Page #9

57
G. Cardona : Paņinian studies
48
6.4.101, though in a different sense: Now, since the rule is suspended, it does not allow for ones taking cognizance of its result with respect to 6.4.101, so that relative to this sūtra, one understands the original element śás to occur and, on the basis of this, the replacement of -hi by -dhi succeeds. Such a complication, however, is not necessary. "S
3.6 One may, of course, claim that, contrary to what Pāņniyas generally maintain, 46 all instances of suspension should be treated as involving operation suspension, so that, for example, in deriviing adhitya, one does indeed let the single-vowel replacement take place beore the introduction of the augment tuk. Suspension then serve to let -I- be treated as though it were still -i: adhi-iya adhfya adhitya."? It would appear only just and proper, however, that any scholar taking such a stand and claiming at the same time to judge what Pāṇiniyas say should consider all the evidence available and try to decide how far the views held by Pāņiniyas accord with the actual statements and wordings of the Astādhyāyt. Merely to claim, contrary to evidence such as I have discussed and without thoroughly considering all such evidence, that Pāņinian grammarians from very early on created a confusion about the concepts in question, is unjustifiable.48
• 4. The term siddha is, of course, not a defined technical term of grammar but simply a term of ordinary Sanskrit, the past participle of sidh (siddhyati) 'succeed, come about, be established, be accomplished'. Although forms of this verb, with these meanings, are attested from early Vedic on, perhaps the most pellucid instance showing its meaning, at least in so far as concerns us here, is the first kärikä of the Sadhanasamuddeśa of the Vakyapadiya,"9 where BhartȚhari says that something is undrstood to be an activity (kriya) by virtue of its taking on temporal sequence (aśritakramarupatvat), whether it has been accomplished or not (siddham asiddham vă), so long as it is spoken of a something that is brought to accomplishment (sådhyatvenābhidhsyate). That is an activity, the meaning of a verbal base dhātu), is something sequential in time, that is brought to accomplishment (sādhya(kriya) (actio) perficienda)'.
Now, the result of an operation can be referred to as siddha, by which is meant that the operation in question has applied, so that the result is established. For example, the first two värttikason Astādhyāyt 6.4.11450 say that the sof daridrashould be deleted when an ardhadhatuka affix is to be used with this verb, and that this deletion is established siddhaś ca) with respect to the provison of the affix. Thus, one should allow for the agent-noun daridara-, with the affix ad! The final vowel of daridrá should be deleted, and the result of this operation is to remain established with respect to the introduction of the affix, so that the affix introduced after daridr-is ac, not ap which follows verbs in -2.52
A rule (sastra) also is obviously spoken of as siddha. Given what siddha signifies, this means nothing more than that a sütra of the Astādhyāyī is a finished entity, a rule that has been formulated and set forth. Stipulating that a sūtra is asiddha with respect to some other sūtra, then, means that it is declared not to be a finished entity with respect to this sūtra, that is, in effect, that it is treated as nonexistent relative to the rule in question. In accordance with what I have noted earlier, let me speak of rules as either in force or suspended.
4.1 Pāṇiniyas regularly consider that what is done is to superimpose absence (abhåvaropa)

Page #10

New Horizons of Research in Indology
on a sūtra declared to be asiddha, and this interpretation goes right back to the earliest Pāṇiniyas.
In this connection, let us consider part of a discussion in the Mahabhäşya on Astādhyāyr 8.2.1. In answer to an objection that need not detain us here, it is declared that, although the three quarters of a chapter that constitute the Tripadi are asiddha with respect to the preceding seven and one-quarter chapters of sūtras, the latter is siddha with respect to the former. In particular, samjñāsūtras and paribhāṣās included in the latter take their place along side the former, by vitrue of taking full effect only at the time that operational rules they serve to interpret come into play. Thus, for example, the ablative and locative forms in Aşțădhyāyt 8.2.26:
S istare interpreted 3 as referring respectively to left and right contexts: -s- is deletd if it follows and precedes a consonant other than a semivowel or nasal. The paribhāṣās 1.1.66-67 take their place along side 8.2.66, which they serve to interpret, although this sūtra is indeed suspended with respect to preceding rules, so that it cannot take its place along side them. That is, this rule suspension is not reciprocal. The position thus taken meets with a possible objection: Assuming that Așțădhyāyi 1.4.2. fagfaget op Af also is a general metarule, it too should take its place along side rules of the Tripādī. Accordingly, it will allow what is provided for by a subsequently stated rule, even in the Tripadi, to take precedence over what is provided for by a prior rule.54 In his first värttika on 8.2.1.,55 Kätyāyana meets this objection, remarking that there is no conflict in the case of any rule of the section, headed by 8.2-1 since a subsequent rule in question, included in the section is absent (abhāvăd uttarasya). As Patañjali explains, 56 a possible conflict (vipratişedhaḥ) can obtain only between two rules that have independent scopes of application (savakāśayoḥ) and come into consideration simultaneously as possibly applying in a single context (samavasthitayoh). But in the present instance, a rule of the section headed by 8.2.1, though it is subsequent to another, does not exist with respect to that prior rule (na ca pūrvatråsiddhe paraṁ pūrvaṁ prati bhavati).
4.2 Of course, if a rule R+1, by virtue of being declared suspended with resspect to a rule R, is treated as non-extistent in respect of that rule, this means that R+1 cannot serve to supply elements on which R can operate and that, in contexts where both the rules could tentatively apply if one of them had not been suspended, only the one that has not been suspended can take effect first. Thus, consider again the examples given in section 3.2. Of 7.1.14 and 8.2. 80., only the former can apply to ada-e; 8.2.80 cannot apply to replace elements here, since it is treated as non-existent relative to 7.1.14, so that only the latter can take effect. Once this rule has applied, however, since the supension is not reciprocal, 8.2.80 docs indeed apply. Moreover, this suspension must be provided for, since, by virtue of other general principles, 8.2.80 would indeed apply instead of 7.1.14. The latter provides an operation that is nitya with respect to the operation provided by 7.1.14, which is anitya: If -da -of ada-is replaced by -mu-the replacement of -e by -smai cannot take effect, since this applies after stems in -a; but even if the ending of ada-eis replaced by -smai, the substitutions allowed by 8.2.80 can indeed take effect. In order to preclude this, then, rule suspension is called into play.
4.3 What I have said in section 4.2 is predicated on a straightforward interpretation of the Sanskrit terms siddha, asiddha, in accordance with what the terms mean in ordinary Sanskrit. In the context of the grammar, siddha refers to sūtras considered as finished entities, real stated

Page #11

G. Cardona : Paninian studies
⚫57
rules included in the grammar, which are in force with respect to other such statements; and asiddha refers to sorra declared to be the opposite of this in respect of particular other statements, rules which are suspended in respect of others. Once could take another position, giving the terms in question much more techinical values. Thus, one could say of a 'siddha principle' that, 'It simply tells us that interacting rules in a given derivation must acknowledge their results. More techiniquely, having adopted a straightforward and innocent looking principle that all sutras of the Aṣṭādhyay! are siddha with respect to each other unless otherwise specified, one might go on to say that in particular instances letting rules be' siddha with respect to each other allows them to interact in particular ways. For example, consider the following derivations:
58
59
a) ayaja-i → ayaje, ayaje indramayajay indram -ayaja indram 'I venerated Indra.' b) tud-aka tod-aka 60 'one who goads, wounds'
c) tud-ti tud-a-th ->
Derivations such as (a) involve what modern western grammarians commonly call feeding order, that is, the result of applying one rule is subject to another rule, and so on. Derivations such as (c) involve bleeding, that is, applying one rule precludes the application of another. Thus, once the affix sa is introduced to give tud-a-ti, guna substitution (by 7.3.86, see note 60), which would tentatively apply in the sequence tud-ti (cf. (b), is precluded, since the -u- of tud-a-, which is a stem relative to ti, is not a penultimate vowel in this stem, and a is given a speical marking to show that it does not condition guna substitution in its stem.62 One might wish to suggest that the principle according to which all rules are siddha with respect to all rules predicts the interaction of rules in derivations of the type (a) and that the same principle also says rules should apply in bleeding order for particular derivations."3
4.4. The first claim amounts to a truism. Obviously, if applying a given rule R results in the conditions being met for another rule Ŕ and only this is the case, the latter rule should indeed apply, given that it is in force in the grammar. Thus, once Aṣṭādhyay 6.1.87 has applied to give ayaje in ayaje indram, the conditions are obviously met for applying Aṣṭādhyāyr 6.1.78, and once this has applied to given ayajay indram, the conditions are met for applying Aṣṭädhyāyr 8.3.19 (see note 59). Since 6.1.87 is in force with respect to 6.1.78 and the latter is in force with respect to 8.3.19, all the substitutions provided for in these sutra take effect. On the other hand, 8.3.19 is placed in the Tripăd, so that it is suspended with respect to prior rules. Hence, the contiguous vowels - a i- which result from applying 8.3.19 are not subject to replacement by the guna vowel -e- in accordance with 6.1.87.
The claim that principle (2) says that rules are to be applied in bleeding order, however, is not so obvious. If we take term siddha in its usual meaning, saying that rules such as Aṣṭādhyāy! 3.1.77 and 7.3.86 are siddha with respect to each other amounts to saying that they are in force with respect to each other, so that they can interact if the proper conditions are met, nothing more. Thus, given tud-ti, the conditions are met for both the rules to apply; and if 7.3.86 were to apply first giving tod-ti3.1.77 would still apply, since todhas the value of a verb in the ordered set beginning with tud, just as does the actual primitive base tud. Of course, if 7.3.86 did apply

Page #12

New Horizons of Research in Indology
first, one would allow an undesired form, "tod-a-ti instead of tudati. But it is not at all obvious that the mere fact of rules being in force with respect to each other predicts the proper order of application in such instances.
Nor is it immediately acceptable to say that this claim is justified by virtue of Pāņini's having declared certain sūtras to be suspended with respect to others. Consider again two examples dealt with earlier: ayaja indram, amuşmai (see 4.4, 3.2). The first requires suspension in order to preclude an operation which could apply to the result of substitution (adeśalaksanapratişedhartham, see 3.1) suspension is needed in the second example in order to allow the proper ordering of two rules that come under consideration for a single sequence. Thus, once -y in ayajay indram has been dropped to give ayaja indram, only one operation comes under consideration: guna substitution. Suspension precludes this. On the other hand, given ada-e, two rules come under consideration. Providing that 8.2.80 is suspended with respect to a preceding rule simply means that this sūtra is kept from interacting with any preceding sūtra “s Conversely, then, letting rules be in force with respect to each other simply means that they may interact with each other if their conditions are met, it does not mean that, if the conditions for both are met at once, one rule must bleed the other. This can neither be understood from the ordinary meanings of siddha and asiddha, nor justifiably implied from what Pāṇini says.66.
4.5 It follows that the so-called 'siddha principle' should not be made to do the work of other principles that Pāṇiniyas have operated with and the bases for which can be seen in the Astadhyayt itself.67 Another claim made recently is that the application of the principle concerning internally and externally conditioned operations (antarangal bahiranga) by Paņiniyas to word-internal operations is unjustified, and that Pāņini himself made use of this only across word boundaries.68
It will suffice to consider one of the major examples, since this brings up the basic issue and the weaknesses of the arguments involved. The verb kşi'dwell' has a present of the type kşiyati, with the vikarana sa (see note 61) (kşi-a-ti) contains, as shown, two stems (anga:po: ((kși-)a-) is a complex stem relative to the ending tip, which replaces the L-suffix lat first introduced after the verb, and (kşi-) is a simple stem relative to the affix sa. Since the complex stem has a light penultimate vowel and the verb ending tip is not marked with g, k, or n, 7.3.86 (see note 60) could apply here to replace this with the guņa vowel -e-, which would undesiredly result in 'kṣayati instead of kşiyati. On the other hand, since śa has imposed on it a marking with i(see note 62), it does not condition guna substitution. Instead, -sis replaced by iy kși-a-ti
→ kşiy-a-ti- The complex stem which results is -kşiy-a-, so that 7.3.86 cannot apply to replace -i-, which is now not a penultimate sound, with a guna vowel. The proper derivation is accounted for by a principle that an internally conditioned operation takes precedence over one that is externally conditioned., and the same principle serves to assure the proper derivation of dhiyati 'holds, maintains', riyati 'flows'.72 Similarly, given (ayaja-i) (indra-am), two operations are to be considered concerning -i: first, -a-i is subject to guņa replacement (ayaja-i ayaje, see note 59); on the other hand, -i i-could be subject to replacement by the single long vowel -1.The former takes precedence, since it is internally conditioned. 74
It has been claimed that only the latter type, involving word boundaries, is valid, and that

Page #13

G. Cardona : Paninian studies
Paniniyas have wrongly applied the antaranga-paribhasa to instances like kṣiyáti, dhiyáti. The argument involves the interpretation of the metarule 1.1.5 (see note 62): 'In reality the argument is illusory because guna never has a chance to apply to any of these forms. It is prohibited by 1.1.5 kniti ca, which states that guna and viddhi replacement (a, ai, au, a, e, o, 1.1.1-2) for i, u, r, i (ikah, 1.1.3) do not apply before kit and nit suffixes.... There is no question of applying guna at all in dhi Sa + tip or sru + CaN +tip. Thus, if the most straightforward interpretation of the guna rules is accepted, the apparent motivation for the antaranga-paribhasa simply disappears." There follows a simplified version of what is said in part of the Mahabhasya on 1.1.5 (see below), after which the argument continues: Actually there is no problem when we consider precisely the formulation of the operative guna rule 7.2.86 pugantalaghapadhasya ca. The sthänin of guna replacement is qualified as ikah (i, u, r, by 1.1.3 iko gunavṛddhi and as angasya (a stem) by 6.4.1 angasya. We can take the prohibition 1.1.5 kniti ca as applicable whenever either one of these items, the ik vowel or the anga, is directly followed by a kit or nit item. Then guna is blocked in bhid + Kta because the anga bhid is directly followed by a kit item (although the vowel /itself is not), and guna is blocked in dhi SatiPbecause the vowel i is directly followed by a nit item (although the anga dhi+ Sa is not). Of course, with regard to the other relevant guna rule 7.3.84 sarvadhatukärdhadhatukayoh, both the ivowel and the anga dhi are directly followed by a nit suffix Sa, so that guna is blocked twice over. 76
61
This seemingly innocuous solution is not without problems. Before confronting these, however, let us consider briefly some of the arguments presented in the Mahabhäṣyä on 1.1.5. if 1.1.3 and 1.1.5 are consdidered by themselves, without bringing them together immediately with rules they serve to interpret (cf. note 60), then one can invoke also another metarule directly to interpret the locative kniti: This denotes a right context before which something applies to an immediately preceding element (Astädhyay 1.1:66, see note 53). Under this possibility, 1.1.3,5 could serve to preclude guna substitution in an item like ci-ta'gathered, heaped', where stem-final -i immediately precedes Kta, but one could also argue that they would not preclude guna substitition in bhid-ta (bhinna "broken"), where the stem vowel does not immediately precede the affix. Assuming this, Katyayana proposes" to modify 1.1.5 providing that guna and viddhi replacements which would be conditioned by items marked with g. k, n (tannimittagrahanam) do not obtain. This suggestion meets with an objection: In cases such as riyáti, dhiyáti, the negation of guna substitution thus provided for does not obtain." That is, guna substitution which apply (by 7.3.84) to the stem ri, dhias determined by sa is indeed precluded, since the suffix is marked with n, but guna substition (by 7.3.86) on the penultimate vowel of the stem ri-a-, dhi-a- as determined by tip is not now precluded, since this suffix is not so marked. Hence, in his third värttika" Katyayana argues that the correct results are guaranteed, since replacement of iby. -iy is internally conditioned (antarangalakṣaṇatvar).
On the other hand, Patanjali does not leave things there, and a series of alternatives is presented, as follows. The first suggestion involves understanding dhatu of 1.1.3 (see note 63) to recur in 1.1.5 and understanding that a verb is followed by an element marked with g, k, n. This immediately takes care of instances such as bhinna, since a verb (bhid) is now indeed immediately followed by an affix with the appropriate marking. On the other hand, in a stem such as cinu of cinutas 'they two are gathering..., heaping...', the affixed marked with follows

Page #14

New Horizons of Research in Indology
62
the vikaraṇa śmu, not the verb ci. Hence, it is decided that the negation should apply when an element is followed by an affix which has been introduced to a verb: In the derivation of cinutas, the ending tas replaces lat, introduced after the verb, so that all is in order. 80 As an alternative (athava or, alternatively) to the above, it is suggested that, in accordance with the general prinicple according to which samjñasūtras and paribhaşas take their place with rules they serve to interpret, 1.1.3,5 are brought in together with rules like 7.3.84, 86. This of course means that no modification is required in 1.1.5 and that kniti also need not be taken as a nimittasaptami. Thus, together with the metarules 1.1.3,5 and the headings 6.4.1, Aștādhyāyi 7.3.84 provides that guna substitution applies to the final vowels i and so on of stems that end in these sounds when there follows a sărvadhatuka or an ardhadhatuka suffix, but that this does not apply if the suffix is marked as shown; similarly, 7.3.86 provides that guna substitution also applies to a penultimate light vowel -i- and so on of a stem followed by a sārvadhatuka or ardhadhatuka affix, but that this does not apply if the affix which follows the stem is marked as shown. Another alternative (athava) is brought up: 1.1.5 states kriti, and the negation provided here would lack its full scope of application unless it applied to all possible instances, including those where a vowel i and so on is separated from the appropriately marked affix.82 As a final alternative (athava), Patanjali suggests that Pāṇini's own procedure serve to make known that the negation of guna replacement applies also to penultimate vowels. Thus, he derives gļdhnu'greedy', dhrşnu'daring', kşipnu'one who habitually throws' by introducing an affix marked with k (knu), and he provides that the unaugmented desiderative suffix san is marked with k not only after verbs that end with i and so on but also after verbs in which these vowels precede final consonants.83 Marking these affixes with k would serve no purpose unless 1.1.5 could indeed serve to negate guna substitution also applies to a penultimate vowels. Similarly, 7.3.86 provides that guna substitution also applies to penultimate light vowel -i- and so on of a stem followed by a sarvadhatuka or ardhadhatuka affix, but that this does not apply if the affix is marked as shown.84.
Of course, not all these alternatives are equally good. Thus, since 1.1.4-5 are two separate negative rules complementing 1.1.3, it is not immediately obvious that dhatu- of dhatulope in 1.1.4 should be understood to recur in the following rule and be construed there as denoting a qualificand. In addition, although any - sūtra must, of course, have its proper scope of application, this can indeed be obtained for 1.1.5 if it precludes guna substitution in cases like cita, so that the third alternative is not truly cogent. Moreover, though Panini marking knu directly with k and imposing this marker on san in particular contexts certainly does show that 1.1.5 should serve to deny guna substitution to penultimate vowels, it does not follow that one should have his marking the affixes in this manner serve by itself to show that this is the case, it would be preferable to have this marking required by the application of 1.1.5. Whatever else one might say, then, the second proposal seems most appropriate. For, 1.13-5 are clearly a block of related metarules that serve together with operational rules in order to govern how these shall apply to ensure only desired results. Thus, since 1.1.3 applies with 7.3.84, so that ikahis understood here as a qualifier of the heading angasya, by a general rule of reference, 8s one understands that guna substitution applies to the final sounds of stems that end in vowels denoted by ik. and 1.1.5 comes into play to provide that this does not apply if the affix that follows such a stem is marked with g, k, or n. In effect, the locative kniti is to be construed with sārvadhatukardhadhatukayoh, hence is

Page #15

63
G. Cardona : Pāṇinian studies
cotextually modified appropriately, and both these locatives are interpreted in the context of 7.3.84, as referring to elements that serve as right contexts for operations on preceding items.
This still does not obviate the need for the antaranga principle: Given dhi-a-ti, the condition is still met for guna substitution by 7.3.86. Moreover, the solution Kiparsky proposes in arguing against 'the antaranga-paribhasa' is not apt. The author fails to show that truly 'we can take the prohibition 1.1.5 in the manner proposed, saying that the rule simply denies guna and vȚddhi replacement for an ik vowel that is directly followed by an appropriately marked element. Certainly, this is not the way paribhăşās normally operate: They are not independent statments to be interpreted by themselves; instead, they do indeed constitute complements to operational rules. Moreover, even granting the propsed interpretation, Kiparsky's claim is not free from objection. Under his interpretation, 1.1.5 would preclude guna substitution in any anga that is immediately followed by an item marked with k, n. Consider now the derivation of kurutas 'they two make, do', from kr-u-tas. The affix v86 is an ardhadhātuka element not marked with g, k, ri, so that it does condition guna substitution: KỊ-u-tas -kar-u-tas (7.3.84) kurutas. 87 On the other hand, kr-u- itself is an arga directly followed by the ending tas, which is indeed marked with ń (1.2.4, note 62), and the stem in question has -!- as its penultimate vowel. Accordingly, even under the proposed interpretation of 1.1.5, guņa substitution should be precluded in the stem kr-u. Of course, this substitution must apply to give the sequence kar-u-tas, which Pāṇini operates with (sce above with note 87) as in intermediate stage. To be sure, one could still argue that 1.1.5 precludes guna substitution for the final vowel of an anga preceding an affix or for the penultimate vowel of an anga before such an affix. Hence, tas fails to condition guņa only for the final vowel of the stem in -u, not also for the penultimate vowel of the stem kr-u. But obviously this is tantamount to admitting the antaranga principle. And indeed, there is no cogent reason to deny that Panini operated with this bracketing priciple not only at word boundaries but also within words, at stem boundaries.
4.6 It is necessary to consider a bit more in detail just what this bracketing principle consists in, since the authors who deny that the antaranga-paribhășa is valid within word boundaries cite apparent evidence for their view from Pāņinian derivations where, they say, the application of this prinicple would go contrary to what Pāṇini actually does. The authors can say this because they choose to ignore a distinction between two formulations of the principle in question.88 One of their example will illustrate the point. The form vstrahabhyam (instr-dat-abl. du.) derives from vītrahan-bhyåm, with the compound vstra-han- 'one who has slain Vstra', which itself is an obligatory compound formed with the agent noun - han, derived with the affix kvip.89 vộtrahanin the term vstrahan-bhyam is classed as a pada,” so that its -n is dropped. vȚtrahan-bhyam vītrahabhyam. Since an operation conditioned by an affix can take effect even after the affix has been deleted, this opens the possibility for having the final augment tuk added (6.1.71, note 15). Now', goes the argument, if the antaranga-principle were valid in such a case, it would have the desired effect of preventing the addition of t, since this is an antaranga cote relative to the rule of n deletion which feeds it. Nevertheless, Pāņini took special measures to block the insertion of t: he added a provision to the grammar that the deletion of n is asiddha for purposes of t- insertion before kst suffixes (8.2.2 nalopah supsvarasamjñātugvidhişu krti)... In any case, the special mention of UK in 8.2.2 shows that Paņini did not have in mind the antarea principle

Page #16

New Horizons of Research in Indology
in composing the Aṣṭadhyayt." What the author fails to mention here is that there are two versions of the antaranga principle. First, there is a straightforward bracketing principle according to which an internally conditioned operation (antarangam) takes precedence (ballyah (is) 'stronger') over an externally conditioned one in sequences of the type ((a)b)c (see note 72). It is this sort of antarangatva that Katyayana appeals to in connection with instances such as (ri-) a-) ti) (1.1.5 vt. 3, see note 79). In addition, there is a version according to which a rule/operation which is externally conditioned is suspended with respect to an internally conditioned one." Obviously, the use of examples such as vṛtrahabhyam in the manner illustrated depends crucially on assuming that if Panini had operated with any antaranga principle it was the second version. This is certainly not justified. Nor is it justified merely to ignore the other version of the principle, which is quite in harmony with Panini's system.95
64
5. One of the most interesting and far reaching claims made in recent scholarship concerning Panini and Paniniyas again concerns the use of particular terms in the Aṣṭādhya and their supposed misunderstanding at the hands of Paninlyas.
5.1 Aştadhyay 1.1.44: fer far has to do with terms denoting options. Interpreted at its strictest, the sutra states that vibhāṣa is a term denoting the meaning of na va (or not). It is well known that Panini uses various terms in rules providing for options, such as anyatarasyam, ubhayatha, va, vibhāṣa, vibhāṣitam. In the traditions of Paniniyas, such terms are usually treated as equivalent. Kiparsky (1979), however, has argued that three of the terms in question had strictly distinct meanings. He proposes (1979:1) that they are to be translated as follows: v 'or rather', 'usually', 'preferably', vibhasa 'or rather not', 'rarely', 'preferably not', 'marginally', anyatarasyam 'either way', 'sometimes', 'optionally', 'alternatively', In this manner, he justifies (1979:9) the need for the sutra 1.1.44 as formulated: If, as the tradition maintains, vibhāṣā and va are equivalent, then the negation na is entirely redundant, and the simpler rule veti vibhāṣā would have been preferable." On our interpretation, however, the definition given by Panini is the only possible definition. Vibhasa preferably not is exactly equivalent to va+ na and quite different in meaning from plain va
5.2 Kiparsky's study has the merit of dealing, in some detail, with the use of the concerned terms in the Astädhyayt and of doing so with a certain rigor. Nevertheless, as the author himself has to recognize, albeit obliquely and with much reluctance, his major claim remains only that, a claim. For one thing, he has attempted to demonstrate a three tiered scale of preferences relative to the three terms noted, but he must accept (1979:205) that instead of anyatarasyam, Panini uses ubhayatha in rules concerning Vedic usage: Ubhayath occurs only in Vedic rules, so that it could be viewed as filling the gap left by the non-use of anyatarasyam in Vedic rules (p.196). This despite the following declaration (Kiparsky 1979: 233): we have found that Panini uses his theoretical terms with practically complete consistency. Out of 200 theoretical terms introduced formally in the Astädhyay, of which many occur dozens of times there, none has a synonym, either theorcical or non-theoretical, except for the single case of vṛddha for gotra in rule 1.2.65. We cannot explain this consistency in any other way than by assuming that the author deliberately designed his grammar that way. Whatever earlier pre - Paninian clements were taken over into it must have been terminologically adapted to standard Paninian usage, and whatever
iqring som

Page #17

G. Cardona : Paninian studies
supplementary rules were later interpolated into it must have followed the conventions of the original. In addition, Kiparsky has to admit that not all cases involving even the three terms to which he devotes most of his energy conform to his thesis: While anyatarasyam is frequently used for intermediate case, which look like truly "free" variation.. it definitely ranges over the whole scale. Why Panini did not use va and vibhāṣa in all cases where a preference is discernible is hard to say. It may well be that these preferences were not so clear in Panini's own speech as they now in retrospect appear from the Sanskrit corpus taken as a whole. A relatively modest margin of preference (let us say, a ratio of three to two) will reveal itself to the statistics of a philologist but to the introspection of the native speaker of a living language it may be scarcely distinguishable from absolutely equal frequency. (1979:195). More disturbing still is the fact that, as the author hismelf has to admit, there is at least one instance which absolutely goes against his main thesis. Consider the rules 7.2.35, 56, 15 आर्धधातुकस्ये वलादेः उदितो वा यस्य विभाषा. These concern the addition of the intial augment it to ardhadhätuka affixes that begin with consonants other than y. 7.2.35 is the general rule that provides for this augment under these conditions. The next rule lets the suffix ktvä receive the augment optionally (va) after a verb that is marked with u. Finally, 7.2.15 precludes the augment for the suffixes kta, ktavatu after particular verbs: those with respect to which an option (yasya vibhasa) is provided in subsequent rules. Now, under the view that vibhāṣa and va are not absolutely opposed terms signifying respectively marginal and preferred options, all is in accord with what we know of Sanskrit; for example, the verbs rdh (t dhú) 'grow, thrive', krp (kr pu) 'be in order' kram (krámú) stride, tan (tánú) 'stretch', danbh (dánbhú) 'deceive', dhvans (dhvánsu) 'fall, disappear', vit (vr tu) occur, be', vrdh (vr dhu) 'grow, syand (syando) 'flow, run' all have past participles, well attested from Vedic times on, without it: rddha, kipta, kranta-, tata-, dabdha, dhvasta, vtta, vṛddha, syanna. If, on the other hand, vibhāṣa and va were truly opposed terms so that the use of the former in 7.2.15 should not include reference to the latter in 7.2.56, the general rule 7.2.35 would have to apply; one would, then, have to admit that the Astädhyay would serve to derive participles like "ardhita"kalpita, "kramita from the simple (non-causative) verbs in question. This is, of course, intolerable. Moreover, try mightily as he may to escape the problems posed by 7.2.15 and related rules, Kiparsky (1979;156-159) cannot avoid concluding (1979:158) that, 'In sum, there appears to be an inconsistency in the system which is not removed by either reading (1) or reading (2) or, of course, by the traditional interpretation.
65
5.3 Given such factors," it is reasonable to ask whether, after all, despite its initial intuitive reasonableness, Kiparsky's thesis is worthy of being accepted. A fundamental problem, I think, lies at the basis of his position, and that is his insistance that Astädhyayi 1.1.44 is simply a definition, comparable, say, to rules that introduce phonological terms such as hrasva, dirgha, laghu, guru or sütras that serve to assign entities to particular categories, such as karty, karma, karana, adhikarana. Thus, one of his major arguments (1979:9-10) is that, 'Apart from these technical anomalies, it is quite incomprehensible, on the traditional interpretation, why there should be three words for "optionally", to begin with, when one would do. It is true that Panini occasionally uses two words for the same thing without any apparent structural motivation. But it does not elsewhere happen on such a massive scale -- three synonyms used about one hundred times each and, a very important point, it does not happen with terms which are defined

Page #18

New Horizons of Research in Indology
66
in the Aşțădhyay. Having introduced a term into his system by means of rules, Pāṇini does not then proceed to use another, una fined term -- let alone two -- in seemingly random alternation with it, as he apparently does in the case of the words for "optionally". The well known fact however, is that not all saṁjñāsūtrasare definition in the same sense. Consider the follwing: 1.2.27, 1.4.10, 1.4.45: Achtsstraugai: ero agl 3ERTSTECHRUTHI and 1.1.68 94 9TEGRITETEGYİŞTI. All of these are what the tradition calls sam jñasūtra, and they do all indeed concern conventions of terminology. Nevertheless, they do not all serve the same purpose. Sūtras of the first group are what I call classification rules. They serve to assign class names to specified entities, thereby classifying them: vowels with the duration of u ū, u3, respectively, have the names hrasva, dirgha, pluta; a short vowel (hrasva) has the name laghu; a participant that funcitions as locus with respect to an action has the name adhikaraņa (see also 2.1). And Päņini of course uses these terms with reference to short vowels and so on. 1.1.68, on the other hand, serves only to establish a convention, not to assign any entity to a class: a term (svas rūpam 'own form') other than a technical term of grammar (aśabdasamjna) is a sajña for the term itself (sabdasya), not for its meaning. This obviously establishes the convention that non-technical terms are regularly self-referring. Moreover, there are instances where, for particular reasons, Panini deviates from this convention of self-reference, allowing terms like vȚksa to signify trees instead of the term itself.98 Equally clearly, 1.1.44 does not serve to classify anything. This too concerns a covention. In this connection, consider the rule 4.1.82: T FT CRAIGT. In normal Sanskrit usage, va 'or' can be construed with various constitutents of sentences. Accordingly, a native speaker of Sanskrit, confronted with an utterance such as 4.1.82, could interpret this as stating that a taddhita affix follows the first (prathamat) or some other one (anyasmād vā) of related elements given in subsequent rules; he could make a required suppletion to understand the statement in this way. However, this is obviously not the desired interpretation. Instead, one should understand the sutra in the same way that one understands GTI oft al. That is, vă should be construed, as equivalent to na va, with the supplied verb from asti : The rule should provide that a taddhita affix follows the first of specified related elements or does not do so; that is, it provides for the optional introduction of taddhita affixes. One might suggest, then that the purpose of Pāṇini's rule 1.1.44 is precisely to establish such a convention for all possiblc Sanskrit words that have to do with optionality. It would then be understandable why Panini uses vibhasa in this sūtra, as the most general term for option, derived from bhas 'speak' with the preverb vi just as vikalpa derives from krp with this preverb. This also makes understandable that in Astādhyāyī 7.2.15, Pāṇini uses vibhāṣā: As the general term for optionality, this covers also terms like vå. Under this interpretation, of course, we do not have the problems noted above concerning 7.2.15.99 We also do not have to assume for vā or a highly special meaning that is not at all the usual meaning of the particle in Sanskrit of any cpoch.100
5.4 In sum, although Kiparsky's thesis is both intelligently conceived and generally well argued, in the end I find it neither cogently maintained nor acceptable. The commentarorial tradition, which he is so willing to disparage may, in this case, be perfectly right in not having any inkling of the supposed systematic semantic distinction among the three terms for optionality.
6. Recent years have witnessed an increasd interest in questions concerning Panini's syntactic theory and method of deriving Sanskrit sentences and relating them to each other.

Page #19

G. Cardona : Paninian studies
Although some of the work is repetitive and at times simply confusing due to authors insistence on reading into texts what is not there, 101 some very interesting and perceptive research has been done.
67
6.1. A series of studies by Deshpande merits special mention both because they show a solid command of and respect for the texts and because they present some very challenging ideas, linking formal grammatical factors with sociolinguistic and historical developments.102 example will suffice to illustrate.
Pāņini's grammar has rules that serve to derive and relate utterances like (a) gf and (b), both meaning.. 'wants to go'. Aṣṭadhyay 3.3.157-158: early fargeci i antraging
I provide the following: A verb is followed by either of the L-affixes lin, lof if it is used in construction with verbs meaning wish; a verb is followed by the affix tumun if, in addition, the acts signified by the cooccurring verbs have the same agent. 3.3.158 accounts for derivates such as gantum (-gam-tum-s) of (a). In addition, according to Astädhyayт 3.1.7: : for:
mandantecai at, the suffix san optionally follows a verb if this denotes an act which is the object of wishing and has the same agent as wishing. This accounts for derived verbs such as jigamisa (gam-sa) 'wish to go' in jigamișati. Since 3.1.7. introduces the desiderative suffix san only optionally (va), utterances of types (a) and (b) are accounted for as alternants.
103
6.2 Katyayana's first nine värttikas on Astädhyay 3.1.7 center, as his värttikas do on other occasions also, around the reasons that require each of the terms in the sutra to be included therein. That is, in accordance with what värttikas are said traditionally to do, Katyayana questions the need for things said in a sutra and thereby considers possibly more concise formulations. In his tenth värttika, Katyayana then suggests an altenaive way of deriving verbs like cikīrsa 'wish to do make': Let san be introduced after a term ending with tumun and let this suffix be deleted." Concerning this and another derivation, which need not be considered here ---- Deshpande (1985:81-82) says: Katyayana's derivations of the karma-karty construction and the desiderative may now be termed PARAPHRASE DERIVATIONS, i.e. that instead of deriving a sentence or a form directly from the semantic and the syntactic structures, Katyāyana creates an underlying structure, which is a more explicit and a more contemporary PARAPHRASE. Deshpande also suggests that one possible motivation for Katyayana's suggesting his alternative derivations is that, *.. these paraphrases have direct parallels in Middle Indo-Aryan languages; and hence are more explanatory and transparent that the inherited synthetic form jigamișati. "He concludes (Deshpande 1985:82) on a cautious note: 'Of course, the most difficult part of such an investigation is that any traditional grammarian such as Katyayana stands at the intersection of the meticulously preserved academic traditions on the one hand and the inevitable impact of the contemporary linguistic situation on the other. Hence, it is not always possible to find a uniform influence of just one of these factors. Therefore, in Katyayana's work we do not get a complete Midddle Indic grammar of Sanskrit. But we can indeed trace Middle Indic impulses.'
There can indeed be little doubt that both Katyāyana and Patanjali used Middle Indic dialects as means of communication for everyday purpose, so that it is also possible for some of Katyayana's suggestions to be based on 'Middle Indic impulses. By the very nature of the material, nevertheless and as Deshpande admits explicitly -- it is quite difficult to demonstrate this with

Page #20

New Horizons of Research in Indology
68
certainty. For example, what Katyayana suggests in his ninth and tenth varttikas on Aştädhyay 3.1.7 can also be understood immediately in the same vein as the preceding värttikas. Still, proposals such as Deshpande's based on a fair appreciation of what Panini and Katyāyana say, are valuable as stimuli to considering the Astådhyāyt, the language with which it deals, and the relations of these to historical developments all with fresh eyes.
6.3 The following claim, on the other hand, is subject to objection and indeed can serve to illustrate how such issues should not he appoached:
It may well be that Katyāyana felt more at home with gantum icchati than with jigamișati, but I don't think that is relvant here. The fact is that Katyāyana does not use the phrase gantum icchati at all. And he had good reasons for doing so, for again Paņini had left a real problem unsolved.
We may assume that for both Pāṇini and Katyāyana gantum icchati and jigamisati meant the same thing. This should entitle that these two expressions were formed on the basis of the same meanings. The fact is that in Pāṇini's garmmar they aren't. In order to derive jigamisati we have to know that there are two actions, the one being wishing (iccha), the other the object of that wishing; both actions, moreover should have the same agent (p.3.1.7: dhatoh karmapaḥ samanakartȚkad icchayam va.) For the derivation of gantum icchati, on the other hand, we must know that there is an action, in this case wishing, for the sake of another action (p.3.3.10: tumunņvulau kriyāyām kriyarthayam). These conditions are not identical. Yet the two expressions are synonymous. That is to say, the meanings underlying gantum icchati should be able to give rise to jigamisati. How can this be accomplished? Katyāyana chose the simplest solution. He allowed P.3.3.10 to add tum to gam, then he adds saN to gam-tum, followed by the deletion of tum. Note that Katyāyna propses this derivation as an alternative to Pāņini's, but an alternative which solves a real problem.104
The real problem of course, is that the author of these lines is dealing with a nonissue on the basis of ignorance, unaware of -- or at least not mentioning -- Panini's rule 3.3.158.105
7. Interest continues to be shown in Pāṇini's background and his relation to possible antecedants.
7.1 For many years, the issue of whether Yaska antedated Pāṇini or not was hotly debated (see Cardona 1976a/1980: 270 - 273). This topic is now somewhat in the background, but recently a claim was made concerning Yaska's derivations, namely that he would allow simultaneous derivations for a single noun. 106 It has also been show that this claim is not justified by the evidence.107
7.2. As is well known, Pāņini refers to Šakalya in the Aşțădhyäyt. On the basis of Panini's sūtra 1.1.16: Re SCHURIAChlef and the procedure supposed to have been followed in the padapatha associated with the Rgveda, it was recently claimed108 that for Pāṇini this padapatha was not really a separate text linked with the samhitāpātha. This claim, however, is without foundation, since it is based on a misunderatanding of just how Sakalya proceeded in his padapatha.109

Page #21

G. Cardona : Paninian studies
8. Quite recently, there has been a spurt in scholarship concerning the history of the text of the Mahabhäşya. Rau's fundamental work of Vedic citations in the Bhasya (1985) is the basis for a series of interesting studies concerning the history of this text; see Witzel 1986, Bronkhorst 1987b: 14-42.
In a similar vein, and again thanks to the solid scholarship of Rau (1980) there is now serious interest in the Vedic citations found in Bharthari's work (e.g. Bronkorst 1987c).
Al such studies have important implications for our knowledge concerning the state of Vedic texts at the times of Patanjali and, later in Bhartshari's era.
Summary. In this brief -- and admittedly personal -- view of recent work in Paninian studies, I hope to have touched on what are, at least in my opinion, some of the major emphases and trends. There has been a marked tendency on the part of a group of scholars towards finding in the recived Aştādhyay text evidence of massive interpolations. In addition, some scholars have recently placed great emphasis on the proposition that Påņiniyas such as Katyāyana and Patanjali not only did not inherit a single direct tradition but also were ignorant of certain crucial usages known to Pāṇini and ideed misrepresented Pāṇini's intentions. Although none of this recent scholarship is without some merit, the major premisses from which the scholars in question proceed and, to a great extent, the manners in which they attempt to maintain their theses are neither wholly cogent nor acceptable. To finish, let me say a few words about where I think fruitful work remains to be done.
The detailed study of Paņini's Așțadhyay is a task which can hardly ever be considered complete. I would emphasise, contrary to what some recent scholars have tended to do, that, instead of seeking to find interpolations or denigrating traditional scholarship of both ancient and modern Påņiniyas, it is well to treat all the materials in their full details, without any prejudgement about whether the claims and arguments found in them are ultimately to be accepted. This is not because texts such as the Mahabhāsya or the Vakyapadiya in and of themselves have any special claim to a truth, or are based on a single continuous tradition, or because none of the conclusions reached by their authors should be rejected. It is just that, if we wish to understand the full history of this subject, we cannot, without defeating our purpose, afford to leave out the arguments and ideas presented in these works. Moreover, whatever we may conclude concerning particular aspects of Pāņini's theory or method, the fact remains that Kätyāyana, Patanjali, and other Pāṇintyas, with their unequalled mastery of the grammar and its techniques, took up and argued issues that are both worth considering and indispensable for a full understanding of Pāṇini and his work.
Of course, there must be a reliable basis for all such research, which means that reliable editions of all the important works must be made available. Some very good work has been done (see I), but much remains to be done.
In sum, Panini and his traditions still beckon us.

Page #22

New Horizons of Research in Indology
Bibliography
(UPTER
(For bibliographic details not shown in particular entries see Cardona 1976a/1980) Aklujkar, A. N. 1983a. "The Aştädhyāyf as a case in textual criticism'. Jorhi - Laddu pp. 1-10.
... 1983b. Forward to K. A. Subramaina lyer 1983 Ananthanarayana, H. S. 1976. Four lectures on Pāṇini's Aştadhyayr. Annamalainagar: Annamalai University.
... 1981. Pratyähäras in Paņini's Astādhyāyi. Hyderabad: D. K. Dakshayani. Balasurbrahmanyam, M.D. 1981. The system of krt accentuation in Panini and the Veda. Tirupati: Kendriya Sanskrit
Vidyapeeth Bandini, G. 1980. Die Erörterung der Wirksamkeit: Bhartharis Kriyasamuddeka und Helärājas Prakāka, zum ersetn Mal
aus dem Sanskrit übersetzt, mit einer Einführung und einem Glossar versehen. Wiesbaden: Steiner.
Bh: Patanjali's Mahabhāşya, edited by F. Kielhorn, revised edition by K. V. Abhyankar. Bhagavat, V. B. 1984.PHOTHUAT (HOT Roureitaifa | HİGUIRT Theffera jer) TT ( PRIO 31745).
(THE HIIT go4 9921) Pune: Philosophy Department, University of Poona. ... 1985. narty 2 U for NIMASIH. Bombay: HERIE RIV4 Hiery 3 hala isos. Bhagavat, V. B and S. Bhate 1986. Mahabhāşyadipika of Bharthari, fascicle m: Ahnika VI Part 1, critically edited. Pune:
Bhandarkar Oriental Research Institute. Bhate, S. 1980. Treatment of roots didh and vevt in Pāņinis grammar, Centre of Advance Study in Sanskrit, Studies
V. pp. 7-12. ... 1987. The meaning adhikäras in the taddhita section of the Aşțădhyāyr: an analysis, Indo-Iranian Journal
30:81-92. Bhattacharya, B. 1985. Bharthari's Vákyapadiya and linguistic monism. (Professor K. V. Abhyankar Memorial Lectures,
third series.) Pune Bhandarkar Oriental Institute. Bhattacharya, M. 1985. Bệhadvaiyakaranabhūsaņam of Sri Kaunda Bhata.. edited with Rūpălt notes and appendix.
Varanasi: Chaukhamba Amarbharati prakashan. Bronkhorst, J. 1980. Asiddha in the Astadhyāyr: A misunderstanding among the traditional commentarors? Indo-Iraninan
Journal 8:69-95. 1982. On the padapātha of the Rgveda. Indo - Iranian Journal 24:181-189.."
1984. Review of Kiparsky 1982. Indo-Iranian Journal 27; 309-314.. ... 1986. Tradition and argument in classical Indian linguistics. The bahirangaparibhasa in the
Paribhaşendusekhara. (Studies of Classical India 6.) Dordrecht : Reidel. 1987a. Mahabhāşyadipika of Bharthari, fascicle IV: Ahnika I critically edited. Pune : Bhandarkar Oriental Research Institute. 1987b. Three problems pertaining to the Mahābhāsya. (The. Pandit Shripad Sastri Deodhar Memorial Lectures, third series.) Pune: Bhandarkar Oriental Research Institute. 1987c. Further remarks on Bhartharis Vedic affiliation. Studies in Indian culture (Sahitya Siromaņi Professor S. Ramachandra Rao Felicitation Volume. Bangalore), pp. 215-223. 1987d. Review of M. M. Deshpande 1985. Indo-Iranian Journal 30:296-301.
1987e. Review of Rau 1985. Kratylos 32:52-51. Cardona, G. 1970. 'Some principles of Pāņini's grammar. Journal of Indian Philosophy 1: 40-74.

Page #23

71
G. Cardona : Paņinian studies
... 1976a/1980. Paņini, a survey of research. The Hauge: Mounton; Indian reprint: Delhi: Motilal Banarsidasa.
1976b. "Some features of Pāņinian derivations. History of linguistic thought and contemporary linguistics, edited by H. Parret, pp. 137-158. . 1977. Review of Joshi-Roodbergen 1973. Journal of the Oriental Institute, Baroda 26: 328-342. 1983a. Linguistic analyisis and some Indian traditions (The Pandit Shripad Sastri Deodhar Memorial Lectures, first series). Pune : Bhandarkar Oriental Research Institute. 1983b. On the formulation of Astadhyay 8.3.4: anunåsikat paro'nusvårah Surabhi, E. R. Sreekrishna Sarma felicitation volume, editd by M. S. Narayana Murti, pp. 199-205. 1988. Pāņini's grammar and its traditions. Volume I: Background and introduction. Delhi: Motilal Banarsidass. forthcoming a. 'On Pāṇini, Šakalya, Vedic dialects and Vedic exegetical traditions.' Grammatical studies dedicated to S. D. Joshi, edited by M. M. Deshpande forthcoming b. The Vaiyakaraṇasiddhäntaparamalaghumati jūşā attribued to Nagekabharta, critically edited with
an annotated English traslation. Deshpande, M. M. 1975. Critical studies in Indain grammarians, I: The theory of homogeneity (sivarnya). (The Michigan
studies in South and Southeast Asian Languages and Linguistics, 2) Ann Arbor: University of Michigan, Center for South and Southeast Asian Studies. 1979. Sociolinguistic attitudes in India: An historical reconstruction. Ann Arbor: Karoma. 1980. Evolution of syntactic theory in Sanskrit grammar: Syntax of the Sanskrit infinitive tum - UN. Ann Arbor: Karoma 1985. Ellipsis and syntactic overlapping: Current issues in Paninian syntactic theory (The Pandit Shripad Sastri Deodhar Memorial Lectures, second serics.) Pune: Bhandarkar Oriental Research Institute: 1986. 'From uttarapadalopa to madhyamapadalopa: implications for theoretical change. Annals of the Bhandarkar Oriental Research Institute LXVII: 251-257.
. . ., 1987a. Pāņinian syntax and the changing notion of sentence. Annals of the Bhandarkar Oriental Institute LXVIII (R. G. Bhandarkar 150th birth anniversary volume) 55.98. 1987b. Panini's syntax of Sanskrit gerund constructions: an alternative view. Adyar Library Bulletin 51
(Festschrift for Ludo Rocher:) 242-266. Devasthali, G. V. 1983a. Apropos of vibhasa. Joshi Laddu. 1983 pp. 157-165
t ... 1983b. Pāņini and Vedic - a critique. Annals of the Bhandarkar Oriental Rescarch Institute LXIV:137-148.
Dipika. Mahābhāşyadspikā of Bhartphari. Pune: Bhandarkar Oriental Research Institute, 1980. Filliozat, P. S. 1975-1980 (tr) Le Mahābhāşya de Patañjali avec le Pradipa de Kaiyata et l'Uddyota de Nageka. (Publications
de l'Institute Français D'Indologie No. 54,1 - 4) Pondichéri: Institut Français d'Indologie (54,1 (1975): adhyâya 1 pâda 1 âhnika 1-4, 54:2 (1976): adhyâya 1, pada 1, ahnika 5-7; 54,3 (1976): adhyâya 1, pada
1, ähnika 8-9; 54,4, (1980): adhyâya 1, pada 2, 54: 54, 5:) ... 1988. Grammaire sanskrite pāninéenne. (Collection Connaissance des Languer.) Paris: Bocard. Gune, J. A. 1978. The meaning of tenses and moods: The text of Kaundabhatta's Lakararthanirnaya, with introduction,
and explanatory notes. Pune: Deccan College Postgraduate and Research Institute Joshi, S. D. 1982. Presidential address, Indian linguistics section. Proceeding of the All-India Oriental Conference, thirtieth
session, Visva Bharati, Santiniketan (Pune: Bhandarkar Oriental Research Institute), pp. 91-111. Joshi, S. D. and S. Bhate 1984. The fundamentals of anuvrti. (Publications of the Centre of Advanced Study in Sanskrit,
. Class B. No. 9) Pune: University of Poona. Joshi, S. D. and P. Kiparsky. 1979. Siddha and asiddha in Pāṇinian phonology. D. Dinnsen, editor, Curons approaches to phonological theory (Bloomington: Indiana University Press), pp. 223-250.
. .
. . 17.

Page #24

New Horizons of Research in Indology
72
Joshi, S. D. and S. D. Laddu 1983. Proceedings of the international seminar on studies in the Appadhyay of Panini
(held in July, 1981) Pune. University of Poona. Joshi, S. D. and J. A. F. Roodbergen 1973. Patanjali's vyakarana-mahabha$ya tatpurushnika (P.2.2.-2.2.23). Pune:
University of Poona. 1975. Patafijali's vyakarana-mahabharya kärakähnika (P.1.4.23-1.4.55). Pune: University of Poona. 1976. Patanjali's vyakarana-mahabhagya anabhihitahnika (P.2.3.1.-2.3.7.) Punc. Univerisity of Poona. 1980. Patanjali's vyakarana -mahabhäsya vibhaktyähnika (p. 2.3.18-2.3.45). Pune: University of Poona. 1981. Patanjali's vyakarana-mahabhäşya pratipadikärthasesdhnika (P. 2.3.46-2.3.71). Pune: University of Poona. 1983. The structure of the Astädhyay in historical perspective. Joshi Laddu 1983, pp. 59-95. 1986. Patafi jali's vyakarana-mahabhāşya paspaśáhnika. Pune: University of Poona. 1987. 'On siddha, asiddha, and sthānivat. Annals of the Bhandarkar Oriental Research Institute LXVIII
(Ramakrishna Gopal Bhandarkar 150th birth anniversary volume): 541-549. Junnarkar, P. J. 1977-1988. An introduction to Pâņini. 4 volumes (1977,1980, 1983,1988). Baroda: Shanti Dighe. Kapil Dev Shastri 1975. TITAE-K tomu RIGHTRAY ( p 31746 H 18). Kurukshetra:
Kurukshetra University. ... 1985. Nageka Bhatta's Vaiyakarana-siddhantamafijüşa. Kurukshetra: Vishal Publications.
Kak: Käsikävytti, edited by Aryendra Sharma, Khanderao Deshpande, D. G. Padhye Katre, S. M. 1981. A glossary of grammatical elements and operations in the Asfadhyäyt. Myrore: Central Institute of
Indian Languages. ... 1987. Asfadhyay of Paņini. Austin: University of Texas Press. Kiparsky, P. 1979. Panini as a variationist. (Current Studies in Linguistics, 7). Cambridge,
Pune: MIT and The Poona University Press. 1982. Some theoretical problems in Paņinis grammar. (Professor K. V. Abhyankar Memorial Lectures, second series). Pune: Bhandarkar Oriental Research Institute. 1987. What is Siddha? Annals of the Bhandarkar Oriental Research Institute LXVII (Ramakrishna Gopal
Bhandarkar 150th birth-anniversary volume): 295-303. Limaye, V. P., G. B. Palsule and V. B. Bhagavat 1984. Mahābhāşyadipika of Bharthari, Fascicle II: Ahnika 5, critically
edited. Pune: Bhandarkar Oriental Research Institute. Mehendale, M. A. 1986. 'On the Brhaddevata 2 .102-103.' Indo-Iranian Journal 29:117-120. Misra. Devaswarupa 1978. Mahabhāşyanigüdhakutayaḥ. (Sarasvati Bhavana Studies XXII). Varanasi: Sampurnanand
Sanskrit Vishvavidyalaya. Misra. Muralidhara. 1977-1980. Prakriyakaumudi by Ramacandracarya with Prakaka by Srtkrana. (Sarasvati Bhavana
Granthamala 112.) Varanasi: Sampurnanand Sanskrit Vishvavidyalaya. (Three volumes: I-II: 1977, III: 1980). Mifra. Vidyaniwas 1987. Kaunda Bhatta's Vaiyakaraṇabhūşanam (A treatise on philosophy of grammar). Delhi: Bharatiya Vidya Prakashan. Narasimhacharya, M. S. 1973-1983. Mahabhäşyapradipavyakhyanani. (Publications de l'Institut Français d' Indologie,
No. 51, 1-10). Pondichéry: Institut Francais D'Indologie
Nyāsa: JinendraBuddhi's Kafikāvivaraṇapanjika, in Dwarikadas Sastri's edition of the Kakika Palsule, G. B. 1983. Mahabhäşyadiptka of Bhartshari, Fascicle I: Ahnika II, critically edited. Pune: Bhandarkar Oriental Research Instiute PM: see Ramachandrudu, P. Srl and V. Sundara Sarma

Page #25

73
G. Cardona Paninian studies
Pr: Kaiyata's Pradipa, in Vedavrata's edition of the Mahabhāṣya.
Raghunatha Sarma. 1977. Vakyapadiyam (part III) Padakanda Vṛttisamuddeka) with the commentary Prakāka by Helārāja and Ambakartri by Pt. Raghunatha Sarma. (Sarasvati Bhavana Granthamala 91.) Varanasi: Sampurnananand Sanskrit Vishvavidyalaya.
1979. Vakyapadiyam (part III, Vol. II) (Bhoyodravya-guna-dik-sadhana-kriya-kala-puruşa sankaya-upagraha and lingasamuddeka) with the commentary Prakāka by Helārāja and Ambakartri by Pt. Raghunatha Sarma. (Sarasvati Bhavana Granthamālā 91). Varanasi: Sampurnanand Sanskrit Vishvavidyalaya.
1980. Vakyapadiyapathabhedanirnayaḥ (part I). (Sarasvati Bhavana Granthamālā 116.) Varanasi: Sampurnand Sanskrit Vishvavidyalaya
Ramachandrudu, P. Sri and V. Sundara Sarma. 1981. Padamafjarf by Sri Haradattamisra. A commentary on Käsikä of Vamana and Jayaditya, edited by Dr. P. Sri Ramachandrudu and V. Sundara Sarma. 2 volumes. Hyderabad: Sanskrit Academy, Osmania University.
Rau, W. 1977. Bharthari's Vakyapadtya. Die Malakärikäs nach den Handschriften herausgegeben und mit einem Päda-Index versehen. (Abhandlungen für die Kunde des Morgenlandes XLII, 4.) Wiesbaden: Steiner Verlag. 1978. Vakyapadīyaprameyasamgraha. Ein anonymes Scholion zum zweiten Kända des Vakyapadiya. (Ahandlungen der Marburger Gelehrten Gesellschaft, Jahrgana 1978, Nr. 2) München: Fink verlag. 1980. Bhartrhari und der Veda. Studien zur Indologie und Iranistik 5/6 (Festschrift P. Thieme): 167-180 1985. Die vedichen Zitate im Vyakarana-Mahābhāṣya. (Akademie der Wissens chatten und der literatur, Mainz, Abhandlungen der Geistes und Wissenschaftlischen Klasse Jahrgang 1985, Nr. 4) Wiesbaden: Steiner Verlag
Scharfe, H. 1977. Grammatical literature A history of Indian literature, part II, fascicle 2). Wiesbaden: Harasowitz. Sekhara: Nageka's Laghusabdendusekhara, edited by Guruprasad Shastri.
Sharma, R. N. 1975. Referential indices in Panini. Indo-Iraning Journal 17:31-39.
1976a. पाणिनिव्याकरण में प्रजनक प्रविधियां Agra: Central Hindi Institute.
1976b. 'Word derivation in Panini.' Folia Linguistica IX:215-228
1987. The Aṣṭadhyay! of Panini, Volume I: Introduction to the Aṣṭādhyāyf as a grammatical device. Delhi: Musnshiram Manoharlal.
Sharma, Virendra. 1977. वाक्यपदीय - सबन्धसमुद्देश हेलाराजीय व्याख्या के प्रकाश में एक विवेचनात्मक अध्ययन. Vakyapadiya Sambandhasamuddeka, a critical study with special reference to the commentrary of Heläraja. (Panjab University Indological Series 9.) Hoshiarpur: Panjab University. (In Hindi)
Shukla, J. M. 1984. Bharthari's Vakyapadiya with Gujarati translation and notes. (L. D. Series 98.) Ahmedabad: L. D. Institute of Indology.
Speijer, J. S. 1886. Sanskrit syntax. Leiden: Brill (reprinted 1968, Kyoto: The Rinsen-Shoten Bookstore.) Subramania Iyer, K.A. 1977. The Vakyapadiya of bhartṛhari, Kända II; English translation with exegetical notes. Delhi:
Motilal Banarsidass.
1982. The Vakyapadiya: some problems (Professor K. V. Abhyankar Memorial Lectures, first series.) Pune: Bhandarksar Oriental Research Institute.
1983. Vakyapadiya of Bhartṛhari (An ancient treatise on the philosophy of Sanskrit grammar) containing the lika of Punyaraja and the ancient vrtti, edited with a forward by Ashok Aklujkar. Delhi: Motilal Banarsidass
Sukla, K. P. 1977. Vaiyakaraṇa-siddhantamanjoşa by Nägela Bhatta. (M.M. Śivakumarasastri Granthamālā 3) Varanasi:
Sampurnanand Sanskrit Vishvavidyalaya.
Ud. Nägela's Uddyota (Rohatak edition).

Page #26

New Horizons of Research in Indology
74
Tripathi, Mhagiratha Prasada. 1980. Dhatvarthavijfånam. (Sarasvati bhavana Studies 28.) Varanasi: Sampurnanand
Sanskrit Vishvavidyalaya. Tripathl, Jaya Shakla Lal 1985. Parama-laghu-manijma of Sri Nageka Bhatta edited with Bhävaprakakika and Balabodhin
Sanskrit-Hindi commentaries. (Krishnadas Sanskrit Series 50) Varanasi: Krishnadas Academy. Varma, Siddheshwar 1978. Pāņini and clision: being an analytic study of Paņini's sūtras on lope (elision) in Sanskrit,
Hoshiarpur: Panjab University. Vedala klira, Bhimasinhha. 1988. ISIC 46141073746 941, Delhi: Parimal Publications. Vedalankära, Raghuvir.
... 1977 a astat hietas 37E2R44. Delhi: Nag Publishers. Venkatacharya, T. 1982. Review of Deshpande 1980. Indo-Iraninan Journal 24:132-135. Wealer, A. 1976. Some observations on the grammatical terminology of Panini (Marginalia to Panini's Astadhyay II).
German Scholars on India. Contributions to Indian Studies, edited by the Cultural Department of the Embassy
of the Federal Republic of Germany, New Delhi. Volume II, pp. 361-379. 1977. Zum Problem der "negativen Appositionen" bei Panini (Marginalien zu Pāņini's Asgādhyāyt ). Studien zur Indologie und Iranistik 3.35-70. 1980. "Studien zu Patañjalis Mahabhäşya I Der Term Svärtha und paribhasa CXII. Studien zur Indologie und Iranistik 5/6 (Festschrift Paul Thieme): 279-309. 1986a. "Zum Verständnis des Bhäşya zu Paņini 3.3.18 (Studien zu Patanjalis Mahābhāşya I).' Wiener Zeitschrift für die Kunde Südasiens Archiv für indische Philosophie 30:91 108. 1986b. Cattle, field and barley: A note on Mahabhäsya I 337.24-27 (Studies in Patanjalis Mahâbhâsya II). The Adyar Library Bulletin 50:431-477. 1986c. A note on Mahabhagya II 366.26. gunasandravo dravyam. Studies on Mallavadin's Dvadakaranayacakra II. Buddhism and its Relation to other Religions: Essays in Honour of Dr. Shozen Kumai
on his Seventieth Birthday, pp. 1-33. Witzel, M. 1986. On the archetype of Patafjali's Mahabhâsya. Indo-Iranian Journal 29:249-259. Yagi T. 1984. Le Mahabhāşya ad Panini 6.4.19 (Publications de Institut de Civilisation Indienne, série in 80, fascicule
50.) Paris: Institut de Civilisation Indienne.
FOOTNOTES
I am grateful to A. M. Ghatge, J. A. Gune, S. D. Laddu, and M. A. Mehendale for discussions concerning points dealt with in this article. Of course, none of these scholars necessarily shares all my views and any errors which might appear are mine alone.
In connection with the 25th anniversary of the Centre of Advanced Study in Sanskrit, a bibliograhpic survey of recent work is projected. For a fairly detailed critical bibliography of work done in the field up to 1975, see Cardona 1976a/1980. The recent history of Sanskrit grammatical literature by Scharfe (1977) leaves much to be desired (see Cardona 1979). 2 See Cardona 1976/1980: 153-160 (Section III. 1.3.3.) for a critical survey of major claims in this area. 3 Joshi-Roodbergen 1983:93: Our claim in the present article is that we have found a number of such inconsistencies, and that they can be traced to a single cause, namely the addition, at a later date, of the taddhita and samasa sections. This is our hypothesis which involves major surgery

Page #27

75
G. Cardona : Påninian studies
in the present text of the Astadhyayf. In an earlier work (Joshi-Roodbergen 1973), the same authors claimed that a group of compounding rules was to be considered interpolated because, in their opinion, the compounds in question could be derived otherwise and the analysis reflected in the rules is supposedly clumsy. Their arguments are not well founded (see Cardona 1976b/1980: 159-60, 1977). Bhate 1980, 1987 are in the same interpretational vein as Joshi -Roodbeergen's work. For example Bhate says (1987:91), Although it is difficult to show different layers in chronological hierachy, the observations made so far do not prevent us from concluding that the kaleidoscopic view of the taddhita section in the Aşțădhyāyf is due to multiple authorship.
Astadhyayr 3.1.133: ugegati 5 Joshi-Roodbergen 1983: 61: The point made here is that as regards the pharsing of the rules in the kịt section and in the taddhita and samāsa sections on the other hand there is no evidence to indicate that they necessarily stem from the same authcr. Joshi and Roodbergen's apparent caution is subsequently abandoned, when they reach the conclusion cited in note 3 above. It is not necessary or fruitful to consider here what they have to say about sûrtas introducing taddhita affixes..
Joshi-Roodbergen 1983:65: The first sequence forms the Karaka-section. The section logically expected to come immediately after this sequence is the vibhakti -section. 'I have presented the general features of this derivation system in a brief article and in a section of the first volume of a large work (Cardona 1976b, 1988: 159-468). 8 Astādhyayf 1.4.1: 3T ERIC FISITI 'This is not to claim that the text of the Astädhyāyi received by Pāņiniyas represents absolutely faithfully what must have been the original text. For example, I consider that the original version of Aşțădhyāyi 8.3.4. was 37991 IRISTER:; and that anunasikat of the received text represents a haplological shortening of ananunåsikat; see Cardona 1983b. Aklujkar 1982 also deals with possible modifications in the original text of the Astädhyay. I consider it wholly possible and indeed probable not only that Pāṇini took over terms of earlier teachers (pārvācāryasam jña) but also that he took over whole sûtras from predecessor without modification. This would accord well with the attitude reflected in Patañjalis statement that teachers do not rescind sätras once they have formulated them (Bh. I. 12.9.-10: 7 entrarf: enfor anar Asturt)see also volume I, pages 44-45 of the Rohtak edition of the Mahābhāşya with the editors first note on page 45. 10 In saying this, I anticipate the conclusion to be drawn below. 11 6.1.86. vt. 1: KagamiTTAISAH derfynaurarare I 12 kas asincat - kaü asiñcat - ko asincat - kosincat. See Cardona 1988:490 (654) for details concerning the rules for each step in this derivation and the derivation of adhitya. . 13 Aștadhyay 8.3.59: Beyrut :1 14 Aștādhyāyi 6.1.109: 5: Clefai 15 Așțădhyâyr 6.1.71: Ari Aa fat gi

Page #28

New Horizons of Researcn in Indology
"Astadhyay 6.1.101: अक: सवर्णे दीर्घः। 17 See Cardona 1988:448-449 186.1.86.vt. 2: तत्रोत्सर्गलक्षणाप्रसिद्धिरुत्सर्गाभावात्॥ 196.1.86.Vt3: असिद्धवचनात्सिद्धमिति चेन्नान्यस्यासिद्धवचनादन्यस्य भाव: Thesame is said in varttika4on Astadhyay 1.1.57 (see note 36). 20 Patanjali puts it, killing Devadatta's slayer does not result in Devadatta's reappearance. Bh.
165.19-20. न ह्यन्यस्यासिद्धवचनादन्यस्य प्रादुर्भावो भवति न हि देवदत्तस्य हन्तरि हते देवदत्तस्य प्रादुर्भावो भवति। Patanjali says this also in his Bhasya on 1.1.57 vt 4 (cf. note 19) 216.1.8414:तस्मात्स्थानिवदचनमसिद्धत्वं च | Bh III.66.2-3 तस्मात्स्थानिवद्मावो वक्तव्योऽसिद्धत्वं च । अधीत्य प्रेत्येति स्थानिवद्भावः। कोऽसिञ्चत् योऽसिञ्चदित्यत्रासिद्धत्वम्। 226.1.84vt 4: स्थानिवद्वचनानर्थक्यं शास्वासिद्धत्वात्। Bh. II.66.5-7: स्थानिवद्वचनमनर्थकम्। कि कारणम्। शास्त्रासिद्धत्वात्। नानेन कार्यासिद्धत्वं क्रियते। कि तर्हि शास्त्रासिद्धत्वमनेन क्रियते एकादेशशास्त्रं तुक्शास्वेऽसिद्धं भवतीति॥
For example, Pr. IV. 415-416. एकादेशशास्त्रमिति। ततश्च पूर्व तुक्शास्त्र प्रवर्तते पश्चादेकादेशशास्त्रमित्युक्तं भवति। Pr. V. 358. अधीत्येत्यादौ पूर्व तुग्भवति पश्चादेकादेशः। 248.2.1vt.8: असिद्धवचन उक्तम्॥ 25Astadhyay1 7.2.102, 6.1.97: त्यदादीनामः । अतो गुणो। By the former a substitutes for the final sound of a stem that is a base of the pronominal subset begining with tyad, before any ending; the latter lets the second of two contiguous vowels replaced both these sounds if the first is short a that does not occur at the end of a pada and the second is a guna vowel. 26Astadhyayr7.1.12,14,15,8.2.80, 8.3.59: टाङसिङसामिनात्स्याः (अत:९)। सर्वनाम्नः स्मै (डेर्य:१३)। डसिक्यो: स्मात्स्मिनौ। अदसोऽसेर्दादु दो मः। आदेशप्रत्यययोः। "Bh. II.386. 18-20 उत्सर्गलक्षणभावार्थ च अमुष्मै अमुष्मात् अमुष्य अमुष्मिन्नित्यत्र मुभावे कृतेऽत इति स्मादयो न प्राप्नुवन्ति। असिद्धत्वाद्भवन्ति। 28Cr. Pr. V.358. उत्वशास्त्रस्यासिद्धत्वादप्रवर्तनात्पूर्व स्मादय: पश्चादुत्वम्। 29KasII.904:... यदित ऊर्ध्वमनुक्रमिष्याम: पूर्वत्रासिद्धमित्येवं तद्वेदितव्यम्। तत्र येयं सपादसप्ताध्याय्यनुक्रान्ता एतस्यामयं पादोनोऽध्यायोऽसिद्धो भवति । इत उत्तरं च उत्तरोत्तरो योग: पूर्वत्रासिद्धो भवति असिद्धवद्भवति सिद्धकार्य न करोतीत्यर्थ: । The Kasika's येयं 'ऽसिद्धो.. is directly taken from the very beginning of the Mahabhasya on 8.2.1. "For example, Nyasa VI. 329-30: पादोनाध्यायग्रहणेनोत्तरयोगग्रहणेन च शास्त्रसिध्दत्वमिहाश्रीयत इति दर्शयति। INyasa VI.330 सिद्धशब्दो निष्पन्नवचनः । सिद्धं निष्पन्नमित्यर्थः। न सिद्धमसिद्धम् अनिष्पन्नमित्युच्यते। परञ्च शास्त्रं यच निष्पन्न तन्न शक्यं क्चनशतेनाप्यनिष्पन्नसत्तायामापादयितुम् तस्माद्यथा ब्रहादत्तोऽयमित्युक्तेऽतिदेशोऽयं गम्यते तथेहाप्यसिद्धत्वमित्युक्तेऽतिदेशोऽयं गम्यत इत्याह असिद्ध वदिति। एतदेव स्पष्टीकर्तुमाह सिद्धकार्यामित्यादि। एवं हि तदसिद्धवद्भवति यदि सिद्धस्य यत्कार्य तन्न करोति। PMII.668: कथं पुनरुचरितं शास्वमसिद्धमित्युच्यते न हि क्चनशतेनापि सिद्धमप्यसिद्धं भवति। अत आह सिद्धकार्य न करोतीत्यर्थ इति। वचनादतिदेश आश्रीयत इत्यर्थः। It is not necessary to discuss here details of extension rules (atideśasūtra) in which forms with or without vati are used. 321.1.57 vt. 2. तत्रादेशलक्षणप्रतिषेधः। As Kaiyata remarks, this is said on the assumption that a rule of extenson allows what would obtain by something other than an element that is present but does

Page #29

77
G. Cardona : Päninian studies
not thereby disallow what is conditioned by that element itself. Pr. I.436: 3 न तु स्वाश्रयं निवर्त्यते ।
33
Astādhyāyī 6.1.77 इको यणचि ।
Aṣṭadhyay 6.1.66:
35 Bh. 1. 436: असिद्धनात्सिद्धम् अजादेशः परनिमित्तकः पूर्वस्य विधिं प्रत्यसिद्धो भवतीति वक्तव्यम् from what is said in the following värttika (see note 36), it is patent that this is intended to be an alternative to Aṣādhyāy 1.1.57 as formulated and that Patanjali correctly presents what Katyayana intended.
Katyayana goes on to consider the other purpose served by suspension. In his third värttika (असिद्धवचनात्सिद्धमिति चेदुत्सर्गलक्षणानामनुदेश), he notes that if suspension is invoked, one must nevertheless also have an extension rule to allow for operations conditioned by original vowels that have been replaced. The examples Patanjali gives are patvya, mrdvya (patu-1-4, mrdu-I-a): Assuming that semivowel substitution (6.1.77, see note 33) applies first to F, the replacement -y must be allowed to have the status of the original vowel, in order that semivowel substitution apply also to the preceding -u. It is then proposed that this too is taken care of by suspension, so that, in effect, a rule of suspension is assumed to replace the original rule 1.1.57, and to this the same objection is brought up (1.1.57 vt. 4: fgfhin dezenfience) as is made in värttika 3 to Aṣṭādhyāyī 6.1.86 (see note 19). Hence, it is concluded that one needs both an extension rule and a statement of suspension (1.1.57 vt 5:1). In värttika 6 (3), finally, Katyayana alludes to what was said in varttika 5 on 6.1.86 (see note 22). Note that elsewhere (Bh I. 145.5-9), Patanjali argues for having semivovel replacement apply first to the -u-of patu-T-a, mrdu-1-4. The details regarding the alternatives of which only the latter is truly acceptable -
need not be considered here.
"Here again I anticipate the conclusion to be drawn below.
With hireplacing a basic ending sip (Aṣṭadhyay 3.4.87).
39The ending antderives from jh(7.1.3: ss), which is gotten from the basic ending jhi. This in turn is a substitute, in the present instance, for the L-affix lad. The iof a verb ending that replaces an L-affix marked with n is dropped (3.4.100:
).
40By Aṣṭadhyay 1.1.56: (neretsafe, a substitute is treated as though it were the original element (sthanivat) except with respect to an operation that would be conditioned by an original sound (analvidhau).
41Sce Cardona 1988:495-496 (660-661), where examples of suspension covered by Aşṭādhyāy 6.4.22 in order to achieve the first purpose noted earlier are also considered.
*He could not simply provide for replacements to have the status of their substituends, since both types 1) and 2) are involved (see note 41). I do not consider here additional details concerning the use of terms with or without vati in extension rules.
43For example, Pr. IV. 687 : इतरो व्यापकत्वाच्छास्त्रासिद्धत्वं प्रदेशान्तर एव स्थापितं मन्यमान आह असिद्धवचनमिति।
4Cf. Ud, IV. 687688 शास्त्रासिद्धत्व एत्वशाभावास्याभ्यां तत्कार्ययोरप्रमितेर्विशास्त्रं प्रति स्थानिबुद्धिरेवेति तन्निबन्धनवार्यसिद्धिरिति भावः । Nagesa brings in also 6.4.119: Et in connection with the derivation of edhi'be' which

Page #30

New Horizons of Research in Indology
78
I have not considered here. 4SFull details concerning rule and operations suspension were considered in an unpublished chapter of my large work on Panini's grammar, distributed in 1976, which will be incorporated as volume II of Panini, his work and its traditions. Bronkhorst 1980 also emphasises the distinction between asiddha and asiddhavat. However, he appears to hold that Aşțădhyāyt 6.4.22 nevertheless provides for rule suspension: 'It is here that P. 6.4.22 comes to our rescue. To see how we may note that both the rules considered in connecion with the derivation of sadhi, both P. 6.4.101 and P. 6.4.35 belong to the Abhiya-section, i.e. to the domain of P. 6.4.22. This has as a result that, that one of the two rules, P. 6.4.101 and 6.4.35 is as if had not taken effect with regard to the other. (1980:70)' In addition, Bronkhorst does not seem to be aware of -- or at least does not make explicit -- how one of the rules in question is indeed allowed to apply before the other. Moreover, he says (1980:70) that each rule... when taking effect, removes the conditions for the application of the other. He thus seems unaware that 6.4.35 can indeed still apply even if 6.4.101 has applied to let dhi replace hi. Finally, Bronkhorst seems not to realize the reason for Pāņinīyas opting for rule suspension even in the case of Așțădhyāyt 6.4.22, namely that, as established elsewhere, rule suspension takes care of both types 1) and 2) hence is more encompassing than operation suspension. cf. Pr. IV.687; $ct 2114 | Braire Ropa pe RH 3116. Consequently, although I agree with Bronkhorst that the position generally accepted by Pāņiniyas concerning 6.4.22 is not acceptable, I cannot agree with him in saying that this is simply a gross misinterpretation on their part. Kiparsky (1982;107-108) essentially agrees with Bronkhorst concerning 6.4.22. Kiparsky also (1982:105), again like Bronkhorst, seems unaware that the two rules concerned in śas-hi do not equally preclude each others applying.. 461 qualify my statement, saying generally because not all Pāṇiniyas in fact adopt rule suspesion in all instances. Thus, Rāmacandra interprets Așțădhyāyi 8.2.1 as providing for operation suspension: PK 1.56: #4CHHEZIT Rafect or weite Anrei fast start 47Thus Joshi-Roodbergen 1987:545-546. 48 Joshi-Roodbergen 1987:548: 'The conclusion can only be that in Sanskrit grammatical tradition the concept of asiddhain Pāņini's grammar, sthånivadbhava and the antaranga pb. have been badly mixed up, starting from a relatively early (pre-Patanjali) date, because to all three of these the same function has been ascribed'. This major claim is made after discussing very little (546-7): in particular, they do not mention places where Pāņinīyas clearly distinguish between asiddhatva and sthànivadbhāva. 49VP.3.8.1. qrafieHRIST AT HTEZIfrated, 3 rei afonifa wetu 506.4.114 vt. 1-2 Ratkrefurgos ats: 11911 Peuruulautail The sūtraitself provides that the -dof daridra'be poor, in straits' is replaced by i before a consonat-initial sarvadhatuka affix that is marked with k. . For example, the past participle daridri-ta-has daridri- before the suffix kta. STAşțădhyay 3.1.134: afrafeefeat ayforrera:1 52Aştadhyāyt 3.1.141: Quarga anturasraamwerk . If aņwere introduced, the final augement yuk would be added to the verb in a (7.3.33: 316 yo furpat:)

Page #31

79
.
G. Cardona : Pāṇinian studies
By Așțădhyay 1.1.66,67: afAMA After ferihityTRI S4Bhasya III.385.2-11: af AGRİ HATEZPIPARI TEMATSzirt shared E HIAT: 981&atuten seit: स्युः। तत्र को दोषः। झलो झलि इत्येतेषां निर्देशानामसिद्धत्वात्तस्मिन्निति निर्दिष्टे पूर्वस्य तस्मादित्युत्तस्य इत्येता: परिभाषा न प्रकल्पेरन्। नैष दोषः। यद्यपीदं तत्रासिद्धं तत्त्विह सिद्धम् कथम्। कार्यकालं संज्ञापरिभाषम् यत्र कार्य तत्रोपस्थितं द्रष्टव्यम्। झलो झलि. उपस्थितं भवति तस्मिन्निति निर्दिष्टे पूर्वस्य तस्मादित्युत्तरस्य. इति। यदि कार्यकालं संज्ञापरिभाषमित्युच्यते इयमपि परिभाषास्ति विप्रतिषेधे परं कार्यमिति। सापीहोपतिठेत। 558.2.1. vt. 1: date for PaufadetsHTAISTIRTI 56 Bhasya III. 385:15-17: In fa paurada: P RUTH THIAGUIRIGIE HIcarr: Haraldufar det hafal न च पूर्वत्रासिद्धे परं पूर्व प्रति भवति। 57 Joshi-Roodbergen 1987:542. 58 Joshi-Kiparsky 1979:228: The most general principle governing the order in which rules are to be applied is that of transparent odering. 2). All rules are siddha w.l.t. all rules (sarvatrasiddham).' 59a-i-eby Aştādhyāyi 6.1.87: 3175 TOT: which provides that a guna vowel is the single substitute for an a-vowel and a following vowel; -ei- ay i- by 6.1.78: galsarita;, whereby ay avây ăv- repsectively substitute for e o ai au - before vowels; 8.3.19: : URI states that, according to Sakalya, word-final -y - v are deleted before vowels. 6 According to Aşpadhyayi 7.3.86: galeria, a guna vowel replaces a penultimate light vowel of a stem followed by a sārvadhatuka or ardhadhatuka affix. See note 62. 61 -tiis an agent-signifying sārvadhātuka affix; according to Aşțădhyāyi 3.1.77: gart: :, the suffix sa is introduced after a verb of the ordered set which starts with tud, if this occurs before an agent signifying sārvadhatuka affix. 62By Asțădhyay 1.2.4: Hidangafta a sārvadhâtuka affix that is not marked with p is marked with . This serves to keep the affix from conditioning guna substitution. Astådhyāyt 1.1.3-5 sant Jorge न धातुलोप आर्धधातुके । विङति च । is a series of metarules associated with operational rules which provide for guna and vrddhi substituiton. If the terms guna, vȚddhi are used to signify replacement and the sound subject to substitution are not specified, one is to understand that the replacements apply to sounds denoted by ik, that is to itu ūļ tl. Guna and vȚddhi replacements that would otherwise apply as stipulated to elements followed by particular affixes do not apply if: the affix in question is an ardhadhatuka that determines deletion in a verb or is one marked with g, k, n. These metarules (paribhasa) apply in conjunction with rules of substitution; thus; together with 1.1.5, Așțådhyāyt 7.3.86 (see note 60) provides for guna substitution in stems followed by affixes that are not marked with g, k, n. 63 Joshi-Kiparsky 1979:231: 'It will be seen that all crucial ordering relations in the derivation of both words (atra - etad-tra, iya ja -yaj-a) involve feeding. The interaction of rules here is in its entirely predicted by (2). There are other types of ordering, not involving feeding, which also full under (2) if the siddha relation is defined in the general manner of (1), as we have suggested. Numerically, the next most important set of cases involve the bleeding relation. Here principle (2) says that rules are to be applied in bleeding order. Their prinicple i) is formulated as follows

Page #32

New Horizons of Research in Indology
80
(Joshi-Kiparsky 1979:225): (la) A is siddha w.a.t. B - For all such that BIA (Ø))4B, A(R), A is applied before B to ø. (lb) A is asiddha w.r.t. B- For all such that B(A(6)) + B, A (6), A is not applied before B to Ø where B(A(Ø)) denotes the result of applying to a given input rules A and rule B in that order and B, A (Ø) denotes the result of applying, to a given input (6) rules A and B simultaneously. Kiparsky 1982:77-95 maintains the same principles, using much the same material, though some slight modifications have been made. 64By Astādhyāyī 1.1.56 (see not 40). The status of tud, tod is comparable to that of ti and tu: the latter is gotten through replacing -i with -u and, by virtue of 1.1.56, tu has the status of ti, so that a form like pacatu... 'should cook'is as much a pada as a form like pacati... is cooking.' 65As I pointed cut in section 2.1, the distinction between operation suspension and rule suspension is not crucial in the case where suspension serves to preclude an operation whose conditions are produced by the application of a rule to be suspended. In this regard, it is worth noting that, in his Laghuśabdendusekhara, Nägeśa considers the possibility that, for instances of type 1) a rule's being suspeded could mean it is treated as not having taken effect (anispannam -apravsttam), although its application is dictated by another principle, but emphasises that, where there is the possibility of two rules applying to the same sequence, suspension means simply that a given rule does not apply at all: Sekh. I.109-111: Re Alluri uqrry fie scufufruit de atic (cf.3) तदिन्नमनिष्पन्नमप्रवृत्तम् शास्त्रसिद्धत्वे तु यत्राप्यन्तरङ्गत्वात् त्रैपादिकं प्रवृत्तं तत्रापि तत्तच्छास्त्रे एवाभावारोपेण तन्निवृत्तिबुद्धेरभाव आदेशबुयभावो लक्ष्ये जातत्वबुद्ध्यभावश्च यत्र तु सहप्रसङ्गस्तत्राप्रवृत्तिरेव अभावारोपात्। 66Note that neither in Joshi-Kiparsky 1979 nor in Kiparsky 1982: 77-87 is any attempt made to justify the siddha principle through a thorough investigation of what siddha means in ordinary usage. I agree with Bronkhorst (see Bronkhorst 1984:311 with reference) that Kiparsky and Joshi have not fully understood the meaning of the terms they make use of. In order to vindicate his and Joshi's views, Kiparsky (1987) answers some of the objections made by Bronkhorst, but the fundamental problem of what siddha means is still not dealt with. 671 have taken up the evidence in the chapter alluded to in note 45. 68 Joshi-Kiparsky 1979:235-240, Kiparsky 1982:95-102. 69Aştādhyāyt 1.4.13. Ruffertereife use! 7° Aşțădhyāyī 7.3.84: Hildurgarfurgat (goT: 82) lets gunareplacement apply to the final sounds of stems which end in i-u -r- vowels, if a sărvadhătuka or an ardhadhatuka affix follows the stem. "Aştādhyāyi 6.4.77: 31fe egurgyat tanrugasti 721.4.2 vt 8, 15:3R$TI GUSTI JUC i Bhāşya 1.306.12:3FTRTacitut wanita hamil.308.24-25:4514 गुणावत्यन्तरङ्गत: प्रयोजनम्। धियति रियति। इयडादेशश्च प्राप्नोति गुणश्च इयडादेशो भवत्यन्तरङ्गत:। 13Așțădhyāyī 6.1.101: 377: Aquf af:1 741.4.2 vt. 10: sfusait HICY: Hurlerara

Page #33

81
G. Cardona : Paninian studies
FKiparsky 1982:96
Kiparsky 1982:97 71.1.5.vt. 1: विडति प्रतिषेधे तनिमित्तग्रहणमुपधारोरवीत्यर्थम्। 7%Bhasya 1.54.3-4: यदि तन्निमित्तग्रहणं क्रियते शचङन्ते दोष: रियति' 791.1.5vt.3:शचङन्तस्यान्तरजलक्षणत्वाता 8°Bh. 154. 13-17: उपधार्थेन तावन्नार्थः। धातोरिति वर्ततेो धातुं विङत्परत्वेन विशेषयिष्यामः। यदि धातुर्विशेष्यते विकरणस्य न प्राप्नोति चिनुत: सुनुत: लुनीत: पुनीत इति। नैष दोषः। विहितिविशेषणं धातुग्रहणम् धातोर्यो विहित इति। धातोरेव तर्हि न प्राप्नोति। नैवं विज्ञायते धातोविहितस्य विडतीति। कथं तहि। धातोर्विहिते विडतीति।। 8"Bh. 1.54.17-19:अथवा कार्यकालं हि संज्ञापरिभाष यत्र कार्यतत्र द्रष्टव्यम् पुगन्तलघूपधस्यगुणो भवतीत्युपस्थितमिदं भवति क्ङितिनेति।। 82Bh. 1.54.19-20. अथवा यदेतस्मिन्योगे क्डिाहणं तदनवकाशं तस्यानवकाशत्वाद् गुणवृद्धी न भविष्यतः। BIAstadhyayi 3.2.140: त्रसिगृधिधृषिक्षिपे: क्नुः। 1.2.9-10 इको झला हलन्ताचा 84Bh. 1:54.19-24: अथवाचार्यप्रवृत्तिापयति भवत्युपधालक्षणस्य गुणस्य प्रतिषेध इति यदयं त्रसिगृधिषिक्षिपे: क्नुः इको झल्हलन्ताचेति क्नुसनी कितौ करोति। कथं कृत्वा ज्ञापकम्। कित्करण एतत्प्रयोजनं गुणः कथं न स्यादिति। यदि चात्र गुणप्रतिषेधो न स्यात्कित्करणमनर्थकं स्यात्। पश्यति त्वाचार्य: भवत्युपधालक्षणस्य प्रतिषेध इति तत: क्नुसनौ कितौ करोति।। 8SAstadhyay 1.1.72: येन विधिस्तदन्तस्य। BGAstadhyay 3.1.79: तनादिकृभ्य उः। 87Astadhyay 6.4.110: अत उत्सार्वधातुके। 881 say 'choose to ignore' becasuse the evidence is not only in the Mahabhāşya for all to see but was also considered several years ago (Cardona 1970:54). 8 Astadhyayi 3.2.87: ग्रह्मभ्रूणवृत्रेषु विप्! The affix kvip is deleted (6.1.67: वेरपृक्तस्य). 9DAstadhyay 1.4.17: स्वादिष्वसर्वनामस्थानो "Astadhyayn: 8.2.7: नलोप: प्रातिपदिकान्तस्या
Astadhyay 1.1.62: प्रत्ययलोपे प्रत्ययलक्षणम्। AKiparsky 1982:102-103. 94असिद्ध बहिरङ्गमन्तरले 9sNor again is it justifiable to use this second version of the principle, silently omitting mention of the other, as Joshi and Kiparsky do (1979:235), in order to show that 'the antaranga principle' does not apply properly in deriving aksadyū- As was pointed out some time ago (Cardona 1970:54), Patañjali himself recognized the weakness of the version in connection with such examples. Although selectively ignoring what others say may well be a favoured method of polemical argumentation, it is, of course, not justified. Kiparsky seems to favour this, nevertheless, not just with respect to ancient Pāṇiniyas but also in connection with modern authors. Thus, he misrepresents (1982:117-118) what was said in Cardona 1970:57: It is certainly true

Page #34

New Horizons of Research in Indology
82
that, if one considers operands and contexts independetly, problems arise, but I also did not do this; instead, I cosidered the totality of contexts and operands at issue: In the case of the single shared operand (r) then, 7.1.100 will counter 7.3.84, since all the contexts of the former are included in those of the latter. Kiparsky also chooses to ignore that, in the capter alluded to in note 45 above, the formulation in question was made amply clear for quite a few instances of what I call 'limited blocking'. The same kind of attitude towards evidence is shown when Kiparsky (1982:35) qualifies as 'bizarre' the example madyam pāyayata atmanam and goes on to say (1982:35 n 37), 'I cannot recall seeing it in the grammatical literature or elswehere. Obviously, Kiparsky's reading in the literature is limited; the example is a slight modification of Bhattikävya 8.41: HAURRET FTOS Ateaict sprato il 31164rt grarutyanici Il 9Note the formulation of Taittirfyaprātisakhya 22.7 afarh4; which, in its traditional interpretation, says that the term và has to do with option, has the meaning option(al) (Tribhāşyaratna: a stast
a t afect wafari Vaidikābharaṇa: arrat PANUTA: Peref sfa 1190). 97 As well as other problems which have been pointed out; see for example, Devasthali 19836. The arguments Devasthali gives against Kiparsky's main thesis in another paper (1983a) are, in my opinion, not cogent. 98Instances of this are discussed in commentaries on 1.1.68. The context of a rule generally indicates when the self-reference convention is discarded. 9'In this context, note the use of vikalpa in Katyāyanaśrautasūtra 1.4.14: Pater geti suferais where it is stated that, when an option is allowed (vikalpe), if a substitute is to be used due to the absence of the material provided for the first instance, this should be like the one with which the rite has begun (pravịttam) and not the altrnative material provided for in the statement allowing the option, since the latter would result in a distinct ritual act. For example, Katyāyanaśrautasūtra 1.9.1:
d a ar af states that rice (vrthin) or barley (yavăn vă) is to be used in offering oblations. If, then rice is not available, one should perform the rite in question using something similar to it, such as nīvāra grains, but not barely. Just as vibhāşă in Aştadhyāyf 7.2.15 can include in its reference options provided for using vă, so does the general term vikalpa include reference to options provided for using vă.
100 That và can indeed be used in statements which concern options preferable to what was stated earlier is well known from traditional sources, though Kiparsky does not mention this. For example, Sábara frequently notes that va is used in particular contexts to set aside a previous view (e. g. Sabarabhāsya on Mimāṁsāsūtra 1.3.2,7,9, (pp. 74,118, 141 of vol.97.2 of the Ānandāśrama Sanskrit Series edition of 1970: 3f afar wat onartai sufo ata quella:1 ARK: na andefat). That is, va is like exclusive or inclusive 'or' and another reason --complementary to the one suggested above ---- Pāṇini felt constrained to formulate 1.1.44 may be to show that the va in the Așțădhyāyf is not exclusive. Of course, it is not proper to conclude from passages such as those cited above from the Jaiminisūtras that preferably is the lexically basic meaning of the term. It is in the nature of argumentative discourse that several options be considiered and that one introduced with vă may be the prefered option, though not necessarily; for example in a Mahābhāşya passage discussed earlier (see 3.5 with notes 81, 82, 84) athavă is used several times,

Page #35

83
G. Cardona Păninian studies and one would be hard pressed to say that each of the alternatives presented is viewed as definitively preferable to the preceding. Indeed, in the Mahabhäṣya on 3.1.7 (Bh. II. 12.19
enfam fidget) Patanjali characterizes as an agatik gatiha position which Katyayana introduces as an option, using va (3.1.7 vt. 5: far). Similarly, when Jaimini says f Fagent wel gef in Mimämsäsütra 1.2.19, 26 he is definitely not using va to state a preferred option; the statements in question are parvapaksas of arguments. Again, in Manusmrti 2.224: धर्मार्थावुच्यते श्रेयः कामार्थी धर्म एव चा अर्थ एवेह वा श्रेयस्थिवर्ग इति तु स्थितिः॥ a series of different views is given concerning what constitutes śreyas, and one of these is stated with va, yet Manu then goes on to give what he considers the correctly established position, so that here too va is clearly not used in connection with a prefered alternative. One quite important thing Kiparsky fails to do is to justify his attribution of the meaning 'preferred option' to v which he treats as a primitive, on the basis of any extensive textual investigation of early Sankrit literature --- preferably Vedic, since, in Kiparsky's opinion, the use of val in the sense he attributes to it was lost by the time of earliest Panintyas. He chooses instead to be satisfied with a very brief refernce (1979:3 note 1) to a section of Speijer's Sanskrit Syntax (Speijer 1886:333-334 (426) where two exmaples, from the Pañcatantra and the Rāmāyaṇa, are given. At least, however, Kiparsky admits (1979: 60) that vain Sankhayanasrautasütra 1.1.34-35: daw simply means 'or' and.. 'does
not have its Paninian sense.'
101 For example, attempting to justify a view he had taken carlier, Kiparsky (1982:28) insists that 'The abstract tenses IAȚetc. are comparable to the kärakas in mediating at the level of abstract syntax between semantics and Lorphology, to both of which they stand, like the kärakas, in a many-many relation.' As R.N. Sharma (1987:57-58) notes, Kiparsky's view is certainly confused, in that, unlike Panini's kāraka categories, which are categories that mediate between semantics and grammatical expressions, the L-affixes are just that, affixes, units to which the class name pratyaya is assigned, just as nominal endings like su au jas and so on are affixes. L-affixes are introduced on condition that karakas--agent (karte) and object (karman)- are to be signified, just as nominal endings are introduced under such conditions. Kiparsky's argument assumes that he is entitled to give to terms whatever status he himself finds convenient, regardless of how Panini used them.
102Naturally, not all the views expressed are acceptable. Thus, in his excellent review of Deshpande 1980, Venkatacharya demonstrates that Deshpande's major thesis is at least not demonstrated with abosulte cogency. Deshpande has more recently (1987b) again taken up the issues surrounding sentences with gerunds and attempted to rebut Venkatacharya.
1033.1.7 vt. 10: तुमुनन्ताद्वा तस्य च लुग्वचनम् । bh. II 149 तुमुनन्ताद्वा सन् वक्तव्यस्तस्य च तुमुनो लुग्वक्तव्यः। In the text vārttika Katyayana sugests also introducing san after a term with lin.
104 Bronkhorst 1987d:300. Incidentally, since Bronkhorst is one of those scholars who sounds the clarion call to 'revolutionary' science in lieu of 'normal science' and chides those scholars who are not sufficiently speculative (Bronkhorst 1987e:57), let me add that my objection to his claim is not meant to discourage speculation. It is only reasonable to ask that speculation be of an intelligent nature, based on reliable evidence that has been fully understood.

Page #36
New Horizons of Research in Indology 84 105 In connection with sentences like goed ont ragu on the other hand, Paninsyas do use Astadhyayi 3.3.10 to account for gantum; see, for example, Padamanjart on 3.4.26, Helaraja on Vakyapadiya 3.7.84. But this is connected with a particular syntactic problem, which cannot be considered here. 106 Bronkhorst 1981:6-7. 107 Mehendale 1986. 108Bronkhorst 1982. 109See Cardona forthcoming a.