Paninian Sutras of Type Anyebhyopi Drushyate

Author(s): George Cardona
Publisher: Z_Nirgranth_Aetihasik_Lekh_Samucchay_Part_1_002105.pdf and Nirgranth_Aetihasik_Lekh_Samucchay_Part_2
Catalog link: https://jainqq.org/explore/269039/1

Page #1

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
George Cardona
1. The Mahābhāsya on Astādhyāyī 6.3.109 (yuta sifa Jellyfes) has a well known discussion concerning model Sanskrit speakers, referred to as frien:. The question is posed : If the śistas are the authority with respect to correct speech forms, what purpose does the Astādhyāyī serve? The answer is : The Astādhyāyi serves to make one know the sistas, as follows. Someone who is studying the Astādhyāyi observes that someone else, although he is not studying this grammar, uses the correct forms that are provided for therein. The student then reasons : Due to the grace of fate that is his or to his nature, this person uses the correct forms provided for in the Astädhyāyi without studying it. I reason that he knows other correct forms too?. According to Patanjali, then, a śista is not only a model for the correct speech forms explicitly provided for (fafer: 976T:) through operations stated in sūtras of the Astādhyāyī but he also knows other correct speech forms (PIAT Frife), which are not explicitly provided for in this manner. In this way, the Astādhyāyī serves not only to account directly for correct speech forms but also to account indirectly for such forms, by referring to the usage of accepted model speakers. As Hari puts it': the grammar serves as a means for conveying the correctness of items of the type 997 by virtue of their being sista usage, because one knows who the śistas are.
What Paninīyas say concerning Astādhyāyī 6. 3. 109 thus has important implications for the concept of what a grammar is supposed to accomplish. To begin with, a grammar is a means of explaining (3431CHW) through derivation the accepted usage that is the object of description. In addition, it is taken for granted that this usage includes open sets of items the grammar does not account for directly through description“, so that Pānini recognizes a living native language that is developing. The distinction in question is

Page #2

92
George Cardona
Jambu-jyoti
comparable to the one between ordered sets of elements-ganas-- exhaustively listed with respect to certain operations and groups of elements—referred to as 31 cm...which only represent a part of an open set of items with some common characteristic(s). The question remains whether the view Patañjali espouses in the Mahābhāsya on 6.3. 109 can be considered to reflect Pāṇini's approach.
2. Consider now a group of sútras which have two features in common : they refer to something as 'seen' and speak of other elements, using 3R or scr.
2.1. &1 31 236 377914fa zpyà (atel: [2881)
2.1.1. 6. 3. 137 comes after a series of rules that provide for longvowel substitution. For example, el 31 984 og ut a&iupqf21843मणिभिन्नछिन्नच्छिद्रस्नुवस्वस्तिकस्य । ६। ३। ११६ नहिवृत्तिवृषिव्यधिरुचिसहितनिषु क्वौ । ६॥ ३॥ १२२ उपसर्गस्य घज्यमनुष्ये बहुलम्।
According to 6. 3. 115, before chof ear' in a compound, the final vowel of a term denoting a mark used as a brand (gu) is replaced by a long vowel, except for the vowels of विष्ट, अष्टन् 'eight', पञ्चन् 'five', मणि jewel', भिन्न 'broken', fa'cut', fece 'a cur', qa'sruva spoon', Faffle, e.g., GEN90 (5179) 'an animal with a sickle symbol as a brand on its ear' but 31009uf an animal with eight stripes on its ear'. By 6.3. 116 long-vowel replacement applies to a final vowel of a prior term in a compound if this is followed by a derivate in fara from one of the following verbs : 7'gird, put on', 'turn, occur', 'rain', qe 'pierce', 57'shine, please', 'bear', 7 'stretch'; e.g., 3916 (346) 'sandal', fica(f )'a district, arque ( oʻrain season'. 6. 3. 122 provides that the final vowel of a preverb is replaced by a long vowel before a derivate in except in derivates that refer to a human (374757); in addition, the sutra provides that this replacement applies variously (691). For example : alate (farcita) 'wetting particularly', 3741411 ( 31941) 'wiping away', with lengthening, and the prasāra 'spreading, expansion' without lengthening. Foc a Nisāda', which refers to a particular human being, also does not show lengthening.
None of the rules from 6. 1. 115 to 6. 1. 136 provides for the long vowels of an 11- in het 'fighting in which opponents grab each other by the hair', tell- in polfe 'fighting in which opponents grab each other by

Page #3

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
93
the hair of the head', (+31-in Gusicfis 'fighting in which opponents hit each other with sticks', &- in eufe 'fighting in which opponents hit each other with their fists'. Moreover, Pānini explicitly provides for such derivatess. 6.3. 137 allows for the required long-vowel replacement. The rule does not, however, specify any domain of application in the way that preceding sūtras do. It simply says that the replacement applies also to others (371414fa). In other words, this rule lets long-vowel substitution apply to elements for which this operation has not been provided for, terms which occur with long vowels in actual usage. This is said explicitly in the Kāśikā, which also notes that this accords with the usage of śistas.
I think there can be no doubt whatever concerning two points. First, what the Kāśikā says agrees with what Patañjali and Bhartrhari said about śista usage. Second, this is the only reasonable way to account for Pänini's formulation. The Astādhyāyī has a series of sūtras providing explicitly for long-vowel replacement as illustrated. This substitution applies in specified domains. In addition, in some domains it applies variously. That is not to say that the replacement takes effect optionally, for this would mean the optional substitution would apply in every instance of the specified domain. Instead, the replacement applies obligatorily in some instances, not at all in others, and optionally in still others (see note 26). The replacement by a long vowel applies also to elements for which this operation has not been provided explicitly. Moreover, the rule stating this does not specify a domain : Pāņini says simply 3zarufa ‘also of others'.
2.2. 31 PI 808 3poft erud
2.2.1. 3.2.101 is connected with the following sūtras : 31 PIEL सप्तम्याञ्जनेर्ड: । ३। २। ६८ पञ्चम्यामजातौ । ३। २१ ९९ उपसर्गे च सञ्ज्ञायाम् । ३॥ २॥ १०० अनौ कर्मणि ।
All these rules serve to introduce the suffix g after be born, arise, come into being construed with certain items, as follows :
3.2.97: vf construed with a term containing a seventh-triplet nominal ending; e.g., FHF: 'born in a lake' and Hogtut: born in a stable', equivalent to fe ; and HGRET:.
3.2.98 : 7 construed with a term containing a fifth-triplet nominal ending, provided the nominal term in question does not designate a generic class of substances (अजातौ); e.g., बुद्धिज: 'arisen from the intellect' and दुःखजः

Page #4

94
George Cardona
Jambu-jyoti
fresulting from suffering', equivalent to Tut: and g: Orala:, but no comparable formation corresponding to 3781* : 'born from a horse'.
3.2.99 : 59 construed with a preverb, provided the derivate formed is a term referring to a particular thing ( 1414) and not to just anything born; e.g., "1 'living creature'.
3.2.100 : t used with the preverb 317 (311 59 'be born after, follow...in succession') and construed with a term signifying an object ( for); e.g., पुमनुजः 'born after a male', equivalent to पुमांसमनुजातः.
3.2.101 then provides that follows construed also with coöccurring terms other than those specified in the preceding rules (Taifa). 3. 2. 97 specifies a term with a seventh-triplet ending, but there are derivates like 375 'not born', such that 37 is construed with a term that contains a firsttriplet ending : 04. 3. 2. 98 requires a term that does not refer to a class of individual substances characterized by a generic property, but there are derivates such as Fleurs 'which has arisen from Brāhmanas'. 3. 2. 101 requires not only a complex 371 but also a coöccurring item signifying an object. On the other hand, there are derivates such as 37557 fone born later, younger sibling, not construed with an object-signifying term. All such cases are taken care of by 3. 2. 101?.
Once more, a sūtra provides for an operation with a domain that is simply additional to the particular domains stated in related rules but not specified any more. Consequently, one must observe usage to know just what is allowed and what is not
2.3. 3121809C 3721f9 Epudi
2.3.1. This sūtra is related to 3171866 Em a itstuga: fenyi which introduces the affix faq after 2757 'shine', 974 ‘shine, appear', ya 'harm', a 'flash', 35wt be strong', ['fill', 'speed', and I ‘praise' construed with a stone'. The derivates refer to agents which perform the acts in question habitually, as part of their nature, or wells. For example, fart "lightning'. 3. 2. 178 states that faay is seen to occur also after other verb bases under the same conditions, thus accounting for derivates like fage "...which cuts...
2.4. ५।३।१४ इतराभ्योऽपि दृश्यन्ते । concerns taddhita affixes optionally introduced after padas formed from a 'much, many and pronominals

Page #5

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
95
other than those of the subset beginning with free but including the interrogative feranto
2.4.1. This sūtra is related to the following: 41316 49PFEN 141310 तसेश्च । ५।३।९ पर्यभिभ्याञ्च । ५।३।१० सप्तम्यास्त्रल। ५।३।११ इदमोह: । ५।३।१२ किमोऽत् । ५।३।१३ वा ह चछन्दसि ।
5.3.7-9 provide that a pada with a fifth-triplet ending is followed by तसिल, which also replaces तसिil and foliows padas with परि, अभि; e.g., ततस् (-तद्-अस्-तस्) 'from that', यतस् 'from which', परितस् 'round about in all directions', 37f4GTA 'from both sides'. 5.3.10-13 introduce affixes after padas with a seventh-triplet ending. In general, 38 follows any such pada formed with a base of the group stated in 5. 3. 2 (see 2.4 with note 10), but follows such padas containing C4. A pada with fat takes the suffix 37. In Vedic usage, moreover, optionally follows such a pada. For example : 75 (AC-5-7) in that, there', 7 'in which, where', E (4-564-5-6) 'in this, here', a ( 064-3-37) 'in which, where ?', 6'in which, where ?'. According to 5. 3. 14, the affixes introduced by preceding rules also are seen to occur after padas with other vibhaktis ( 1927:). That is, for and w can follow padas with endings of triplets other than the fifth and seventh. One can say, for example, 7401 and 4 , where is coreferential with a nominative and an accusative, so that it is derived with it after padas with endings of the first and second triplets.
Moreover, one must appeal to usage in order to know what particular nominal bases enter into such formations. From what is said in the Mahābhāsya, it is clear that, at Patañjali's time, the affixation provided by 5. 3. 14 applied for padas with the bases 49€ 'you', qays 'long lived', Tarifi beloved of the gods', and 3r TL'long lived', all used as equivalents of a second person pronoun">. It is also clear from the way Pāṇini formulates 5. 3. 14 that at his time the domain of the affixation was not definite.
2.5. 3131830 3722921f gerai
2.5.1. This sūtra is related to 3131876 BTN YE ($wE:YN FENTi zeneig [epe]) 1 ३।३।१२९ छन्दसि गत्यर्थेभ्यः ।
According 3. 3. 128, 19(437) is affixed to verbs in -a construed with $46, 54 and to form derivates meaning 'easy to...' and difficult to...'; e.g., ईषत्पान 'easy to drink', दुर्दान 'hard to give', सुदान 'easy to give'. By 3. 3. 129, युच्

Page #6

96
George Cardona
Jambu-jyoti
follows verbs of movement (गत्यर्थेभ्यः) to form comparable derivates in Vedic (छन्दसि). For example : सुतरण ‘easy to cross' (e.g., RV4.19.6d : सुतरणाँ अकृणोरिन्द्र सिन्धून् । 'Indra, you made the rivers easy to cross...'), दुश्च्य वन 'hard to shake' (e.g., RV 10. 103. 2bc : युत्कारेण दुश्च्यवनेन धृष्णुना । तदिन्द्रेण जयत तत्सहध्वम् 'Win, prevail...with Indra who does battle, is hard to shake, is daring.'); see also Debrunner 188 (section 83). According to 3. 3. 130, युच् is seen to occur in Vedic usage also after other verbs, under the same conditions. For example, सुवेदन 'easy to find', as in RV 10. 112. 8d : सुवेदनामकृणोब्रह्मणे गाम् | 'You made the cow easy to find...'
From what Pānini says, then, in the Vedic traditions known to him, such derivates with y were generally formed from verbs of motion and occurred also with other verbs, but he could not determine any definite distribution as to the semantics or formal properties of the other bases in question.
2.6. ३।२।७४ आतो मनिन्क्वनिब्वनिपश्च (विच् छन्दसि [७३]) 1 ३1२।७५ अन्येभ्योऽपि दृश्यन्ते । ३।२७६ क्विाच।
2.6.1. 3. 2. 74 allows विच् and also मनिन्, क्वनिप् and वनिप् to follow verb bases that end in -ā, provided these verbs are used in construction with a coöccurring item terminating in a nominal ending. The derivates formed,
moreover, occur in Vedic. For example : सुदामन् (nom. sg. सुदामा) 'one who gives well' (+-दा-मन् : मनिन्), सुधीवन् (सुधीवा) 'one who places, Imakes well' (-धा-वन् : क्वनिप्), भूरिदावन् (भूरिदावा) 'one who gives much' (e-दा-वन् : वनिप्), कीलालपा (कीलालपाः) 'one who drinks kilala' (-पा-: विच). 3.2.75 states that these affixes occur (दृश्यन्ते 'are seen') after other verbal bases also (अन्येभ्योऽपि); for example, सुशर्मन् (सुशर्मा) one who destroys well' (--'श-मनिन्), प्रातरित्वन् (प्रातरित्वा) 'one who goes in the morning' (-इ-वन् : क्वनिप्), विजावन् (विजावा) 'one who is born, one who procreates' (-जन्-वन् : वनिप्), रेष् (रेट्) 'one who suffers harm3' (रिष्-विच्). 3. 2. 76 then allows क्विप् also to follow any verbal base; e.g., उखास्त्रस (उखास्त्रत्) '...which falls from the ukha pot' (सन्स-क्विप).
2.6.2. 3. 2. 76 is related to two other sūtras : ३।२।६१ सत्सूद्विषQहदुहयुजविदभिदछिदजिनीराजामुपसर्गेऽपि क्विप् । ३१ २। ८७ ब्रह्मभ्रूणवृत्रेषु विवप् (कर्मणि हनः [८६] भूते [८४])।
According to 3. 2. 61, foray is introduced after certain verbal bases if these are used in construction with a coöccurring item terminating in

Page #7

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
97
a nominal ending, which can also be a preverb(347075f9). The bases are : 'sit', give birth to', fac 'hate', i 'wish ill to someone', e 'milk, F join, yoke', farç 'know', 'be', fire 'break', Fesc'cut', for 'conquer, win, be victorious', नी 'lead', राज् 'shine, rule'; e.g., शुचिसद् 'one who sits in a pure place', शतसू 'one who gives birth to a hundred', wigy one who hates exceedingly', one who wishes ill to, deceives a friend', GE cow milker', 37847 horse yoker', asfaç 'one who knows the Veda', $18fte ‘one who breaks a piece of wood, रज्जुछिद् 'one who cuts a rope', शत्रुजित् 'conqueror of an enemy', सेनानी leader of an army'. 3. 2. 87 introduces feay after 57 kill' if this is construed with any of the terms 16 'Brāhmana', u'foetus', Vstra' signifying an object of killing, provided also that the act is referred to the past : 16167 (nom. sg. बह्महा) 'one who has killed a Brahmana', भ्रूणहन्, वृत्रहन्. This sutra provides a two-fold restriction with respect to 3. 2. 76. There is a restriction regarding the base 5 used with reference to past killing, such that this takes foay only if it occurs with one of the terms given in 3. 2. 87, thus, to refer to a person who has killed a man, one uses हतवत् (nom. sg. masc. हतवान्), as in पुरुषं हतवान्. Moreover, there is a restriction regarding past time reference, such that only forang is allowed to occur after 6 used with reference to past killing and construed with the terms aut and so on; this accounts for the use of a derivate like पितृव्यघातिनः पितृव्यघाती 'one who has killed his paternal uncle'14 but not TELEIF.15 3. 2.87 thus restricts the scope of 3. 2. 76 so that cannot take Terary without limitation. 3. 2. 61 applies to introduce toay only to bases construed with a coöccurring nominal (3446), which may be a preverb, but 3. 2. 76 does not have this limitation 16.
2.6.3. As Kātyāyana and Patañjali go on to remark, if one considers 3. 2. 87 to apply in the manner described, then the formulation of the next sūtra, 3. 2. 88: 570ESGH I, serves a purpose. This sūtra provides for Vedic usage (छन्दसि), wherein हन् takes क्विप् variously (बहुलम्). Thus, -हन् occurs with upapadas other than those specified in 3.2.87, as in ATTEI FUTET ATLET 'who has killed his mother, his father, his brother 18'; and 374016: 'slayer of enemies!", contains - E1G-, derived with the suffix 370720.
2.6.4. Now, consider 312108 HEPTARIR fuck / 3171092 31 TG: BIR103 foto Eraf According to 3. 2. 71-72, in usage found in mantras fuori follows श्वेतवह, उक्थशस्, and पुरोडाश् as well as यज् used with the preverb अव. 3. 2. 73 introduces farq after 4 used with 34 to form derivates found in general Vedic usage.

Page #8

98
George Cardona
Jambū-jyoti
3.2.74-76 apply as shown in 2. 6. 1. It is reasonable, I think, to accept what the Kāśikā says about 3. 2. 75. The Kāśikā remarks that 3. 2. 75 uses दृश्यते in order that usage be followed (प्रयोगानुसरणार्थम्)21. This accords with what Patañjali said much earlier (see 1 with note 2), since the accepted usage is that of model speakers. That is, the Vedic usage known to Panini was such that he could determine particular domains in which fua and fare occurred. Moreover, verbs in -ă could form derivates with off and so on (3. 2. 74 [2.6.1]) in compounds. In both Vedic and the spoken language Pānini describes, the same suffixes occur with verbs in ā and also other verbs, as could feary, too, both in compounds and independently. In other words, for Vedic f and so on had a restricted domain of usage, which could be described formally. In Panini's bhāsā, on the other hand, the domain of these affixes had spread, but the development was still at a stage where Panini could not formally describe this with precision. In fact, these affixes were being generalized fully. He therefore provides that these affixes, and 499, are seen to occur elsewhere also. The exact extent of their usage, then, is still in flux within the community of model speakers.
2.6.5. One final point has to be taken into account, concerning 3. 2. 1 (see note 20). This is formulated as a very general rule, whereby 3t could occur with any verb whatever. Accordingly, the sūtra allows derivates such as 3 frugs 'one who looks at the sun'. The discussion of this sütra in the Bhāsya ends with Kāryāyana's statement that such derivates are not used to signify the meanings in question, so that the grammar does not have to state an exhaustive listing of possible derivates in order to preclude these and some others22. That is, formally derivates like frigs are comparable to derivates like it pot maker', yet in the spoken language of Katyāyana's time and place they were not used to express what one expressed using a phrase 371fcrerrogafa. Panini recognizes the fact that derivates with 3m have a very general domain. He therefore formulates a general rule. In effect, he understands that 37 can indeed be used with any verb construed with an object. The resulting derivate might not immediately appear the best usage to a native speaker, but such a speaker has to admit its grammatical correctness and acceptibility. It is as though a modern speaker of English used a derivate like axer referring to someone who wields an axe. This is not the best of usage, but one has to admit it is grammatically well formed and acceptable.

Page #9

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
99
The difference between 3. 2. 1 and 3. 2. 75, then, is that fit and so on, used after verbs construed with upapadas, did indeed have a restricted specifiable domain at one stage of the language and were being generalized, so that one has to observe the usage of model speakers to determine grammatically appropriate usage.
2.7. 31318 30cEtam (adent (3.2.123] 1411 [3.2.85]) 1 31 31 pasfo दृश्यन्ते । ३॥ ३॥ ३ भविष्यति गम्यादयः ।
2.7.1. 3.3.1 states that the affixes of the set beginning with 30 occur variously (579) when an action signified by verbal bases to which they are introduced is referred to current time (aria). For example : sols 'artisan', वायु 'wind', पायु 'anus', जायु 'medicinal herb, physician', मायु 'bile', स्वादु 'sweet',
M 'noble person, sādhu', 3119 'fact23. 3.3.2 then provides that the same affixes also are seen to occur when an action is referred to the past. For example, art, vartman 'path' refers to something that has been gone on (74)24. Further, 3.3.3 states that the derivates 47 and so on refer to the future : T117 one who will go, a traveller', 3T-0117 one who will come', प्रस्थायिन् 'one who will depart', प्रतिरोधिन् 'one who will oppose', प्रतिबोधिन् 'one who will awaken', प्रतियोधिन् 'one who will oppose', प्रतियोगिन् 'one that will be contrary to', uforefuz'one who will come back”, 3121647 'one who will come', Hifah 'which will be 25%
2.7.2. A ślokavārttika on 3.3.1 gives reasons why Pānini formulates this sūtra with an instead of simply saying 301164:, thereby providing that derivates with the affixes of the set beginning with 34 are acceptable correct usage. The property of applying variously (aller ) is taken into account because the affixes 34 and so on appear, in derivations actually provided for, after a small number of the possible bases, not all the bases after which such affixes could occur, and because only most of the actually possible affixes are actually introduced in derivations provided, not all of the possible affixes. In addition, there is a residue of operations that is not taken care of in the derivations as actually provided26. The Kāśikā captures the major emphasis of this by noting that the affixes in question appear also with bases other than those after which they are explicitly provided for and some, though they are not explicitly provided for, are inferred from usage27,
2.7.3. In 3.3.2, Pānini uses a punt, and the Kāśikā remarks that this is meant in order that one follow usage in determining which particular

Page #10

100
George Cardona
Jambū-jyoti
affixes of the golf set are used when an action is referred to the past28,
though, according to the Kāśikā, both Astādhyāyi 3.3.1 and 3.3.2 thus involve a necessary recourse to usage for determining what the rules are to allow, there is nevertheless a difference between the two, which justifies their different formulations. Now, if instead of 3.3.1 as formulated, Panini had stated 341164 a, thereby providing that unādi affixes occur optionally (ar), this sūtra would provide that all unādi affixes apply optionally under conditions stated. As shown in 2.7.2, this is clearly different from what 3.3.1, with 6674, allows29. Moreover, even given the very broad variation that 3.3.1 thus allows for, it does not obviate the need for a separate rule 3.3.2. The former provides for unādi affixes under a single condition of time reference: an action is referred to current time (a ). 3.3.2, on the other hand, allows for some unādi affixes when an action is referred to the past. In addition, 3.3.1 applies in general (l 'for the most part), but 3.3.2 applies with respect to only a subgroup of unādi affixes : some of these also are used with past-time reference30. In addition, there is an enumerable subgroup of items used with future reference, covered by 3.3.3.
2.8. &18193 11910€ 7GRf Epid provide that certain operations are seen to apply also in Vedic
2.8.1. 6.4.73 has to do with augments added to verbal stems followed byendings that replace the L-affixes लुङ्ल ङ्लु ङ्. By६४।७१ लुङ्लङ्लुझ्वडुदात्तः। such stems receive high-pitched 3rę as an initial augment. 3T occurs only with consonant-initial bases, however, since this general rule has an exception: 618192 37TSIGT i provides that high-pitched 3r is added as an initial augment to vowel-initial stems. In addition, 6.4.73 states that 347. is also seen to occur in Vedic usage. Now, 6.4.71-72 do not specify that the augments in question are added in the currently spoken language alone, so that it would make no sense to say that 6.4.73 simply provides for the use of 31€ also in Vedic. This sūtra must, on the contrary, provide for a broader use of this augment than is the norm in the currently spoken language. That is, the rule provides that occurs also with stems other than those for which specific provision is made in 6.4.7231, thus accounting for the fact that in Vedic 317 occurs with consonant-initial as well as vowel-initial stems. For example : आयुनक् । अयुनक् 'yoked'.

Page #11

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
101
2.8.2. According to 018194 freeft U TEGIT: (nufq[98]37f [93]), अनsubstitutes for the final -इof अस्थि 'bone', दधि 'curds', सक्थि 'thigh' and 37fe 'eye' before a vowel-initial ending of triplets starting from the third; e.g., 37-7-391 (instr. sg.: third-triplet ending a ) 3T4T-371 37932, 123:41 ...247, fare-37 +... HOTI, 37729-377 ...378011, but 3472774 (instr.-dat.abl.du), दधिभ्याम्, सक्थिभ्याम, अक्षिभ्याम्, with the endings भ्याम् of the third, fourth, and fifth triplets. In addition, 61819€ Syrenia apud i states that 3F is also seen to occur as a replacement for the - of these stems in Vedic usage. The situation here is comparable to that of 6.4.73 (2.8.1): Since 7.1.75 does not apply specifically in the spoken language alone, thus excluding Vedic usage, 7.1.76 cannot be meant to provide for Vedic usage involving the same endings as covered by the preceding sutra. On the contrary, 7.1.76 has to allow the replacement in places where this is not provided for by the preceding rule : before consonant-initial endings, before endings other than those of the triplets starting with the third, and before elements other than endings. For example : :, 379174: (instr. pl.); green (acc. pl.); 378 oant (instr. sg.) 'which has eyes', 340 (acc.sg.) 'corporeal', with 37871 and 37797- before the suffix act, which is not a vibhakti.
3. The evidence considered here is best accounted for on the thesis that Pānini34 has described a living spoken language of his time35, a language that exhibits traits found in any living language. In particular, Pānini has to take into account that there are relations among variants which can be described in a determinate fashion and others which cannot. Thus, there are instances where elements or operations A and B are complementary, each occurring in a specifiable domain. There are also instances where A and B vary within a stateable domain. In addition, there are instances such that, in a determined domain, there is indeterminate variation. This is the situation for which Panini says that certain operations apply variously (564) in a stated domain. Conversely, there are instances where the operations are determined but the domains in question remain partly indeterminate. This is the situation for which Panini states that given operations are seen to occur under conditions other than those already stated. To reiterate the first example considered : Pānini can indeed specify certain places where long-vowel replacement applies in prior members of compounds (2.1.1). He must also accept, however, that this substitution applies to vowels of items which he cannot specify. Accordingly, he says

Page #12

George Cardona
Jambu-jyoti
(6.3.137 [2.1]) that such long-vowel substitution is seen to apply to items other than those he has previously specified. Similarly, having gone as far as possible in stating the domains of certain affixes, he has to admit that they are seen to occur under additional, indeterminate, conditions.
102
Clearly, Panini does not consider it his task to put a strait-jacket on some state of a language to have it go on as a frozen fossil. On the contrary, he describes a living language used by native speakers who were carrying out innovations and generalizations. Moreover, Panini's rules recognize, as should any grammar, that the final results of developments which have started cannot be predicted. Hence, the Aṣṭādhyāyī appropriately includes what I call 'escape rules', statements which, I think, testify to the acumen and insight of Panini.
Annotations:
1. A short version of this paper was presented at the Xth World Sanskrit Conference in Bangalore (January 3-9, 1997), and an abstract under the title 'Escape rules in Panini: sūtras of the type "anyebhyo' pi drsyate" was published in the proceedings of this conference :Xth World Sanskrit Conference... English Abstracts (New Delhi : Rashtriya Sanskrit Sansthan, 1997),. pp. 412-413. I have used the following abbreviations: RV: Rgveda, Kāś.: Käsikävṛtti: Aryendra Sharma, Khanderao Deshpande, D. G. Padhye 1969-70, PM: Padamañjarī: Dwarikadas Shastri and Kalika Prasad Shukla 1965-1967, Paddhati: Vṛsabhadeva's Paddhati: K. A. Subramania Iyer 1963a, Pr.: Padīpa: Vedavrata 1962-63, Bh.: Mahābhāṣya : Abhyankar 1962-72, LSS: Laghuśabdendusekhara with Bhairava's Candrakalā (edited by Narahari Sastri Pendse, reedited by Gopala Šāstrī Nene, Varanasi : Chaukhambha Sanskrit Sansthan, 1987), Vas: Värarucasangraha with Nārāyaṇa's Dīpaprabhā (edited by D. N. Pandey, Varanasi: Chaukhambha Sanskrit Sansthan, 1986), vt. vārttika, VPVṛ. Vṛtti on the Vakyapadiya: K. A. Subramania Iyer 1963a, VS Vājasaneyisamhitä, SK Siddhantakaumudi Giridhara Sarmā Caturveda and Parameśvarānanda Sarma Bhaskara. For full bibliographic references, see Panini, A Survey of Research (The Hague: Mouton, 1976, Delhi: Motilal Banarsidass, 1980 [reprinted 1997]). In addition, 'Debrunner' refers to Albert Debrunner, Altindische Grammatik von Jakob Wackernagel, Band II, 2: Die Nominalsuffire (Göttingen: Vandenhoek & Ruprecht, 1954) and 'Cardona 1997 refers to Panini, his Work and its Traditions, Volume I: Background and Introduction, 2nd edition, revised and enlarged (Delhi: Motilal Banarsidass).

Page #13

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
2. Bh. III. 174.10-15 यदि तर्हि शिष्टाः शब्देषु प्रमाणं किमष्टाध्याय्या कियते शिष्टज्ञानार्थाष्टाध्यायी। कथं पुनरष्टाध्याय्या शिष्टा: शक्या विज्ञातुम् अष्टाध्यायीमधीयानोऽन्यं पश्यत्यनधीयानं ये चात्र विहिताः शब्दास्तान्प्रयुञ्जानम् । स पश्यति नूनमस्य देवानुग्रहः स्वभावो वा योऽयं न चाष्टाध्यायीमधीते ये चास्यां विहिताः शब्दास्ताँश्च प्रयुङ्क्ते । नूनमन्यानपि जानाति। एवमेषा शिष्टाज्ञानार्थाष्टाध्ययी । The Bhāsya on 6.3.109 has been discussed often; see most recently Cardona 1997: 551-553 (834). 3. VPV 1.12 (43.8-9 ) शिष्टज्ञानाच्च पृषोदरप्रकाराणां शिष्टप्रयोगत्वात्साधुत्वप्रतिपादने निमित्तं व्याकरणम् । 4. Vrsabhadeva refers to the two types as अन्वाख्यात and अनन्वाख्यात: Paddhati 1.12 (43.22, 24-26) एवं तदन्वाख्यातानां व्याकरणमुपायः । अनन्वाख्यातानामपीति दर्शयति... शिष्टज्ञानाच्य इति । तैर्य उपदिष्टा लक्षणेन नान्वाख्याताः पृषोदरप्रकारास्ते तत्प्रयुक्तत्वात्साधव इति ज्ञायन्ते एवं पारम्पयत्तिषां व्याकरणं साधुत्वप्रतिपत्तौ लघुरुपायः ।
103
5. Bahuvrihi compound by 2.2.27 तत्र तेनेदमिति सरूपे । Samāsānta इच् by 5.4.127 : इच्कर्मव्यतिहारे ।
6. Kas 6.3.137 अन्येषामपि दीर्घो दृश्यते स शिष्टप्रयोगादनुगन्तव्यः । यस्य दीर्घत्वं न विहितं दृश्यते च प्रयोगे तदनेन कर्तव्यम् ।
I
7. The Kasika includes all these and more. In particular, it considers thar Panini's use of 3ft is intended to allow all the conditions given in previous rules to be superseded, so that, for example, a derivate with can refer to a käraka other than an agent, as in fat 'moat', which is derived from 'dig' and refers to something that has been dug अन्येष्वप्युपपदेषु कारकेषु जनेर्डप्रत्ययो दृश्यते सप्तम्यामित्युक्तम् असप्तम्यामपि दृश्यते । न जायते इत्यजः । द्विजता द्विजाः । पञ्चम्यामजातावित्युक्तम् जातावपि दृश्यते । ब्राह्मणजो धर्मः क्षत्रियजं युद्धम् । उपसर्गे च सञ्ज्ञायामित्युक्तम् । असञ्ज्ञायामपि दृश्यते। अभिजाः परिजाः केशाः । अनौ कर्मणीत्युक्तम् । अकर्मण्यपि दृश्यते अनु जातः अनुजः । अपिशब्दः सर्वोपाधिव्यभिचारार्थः तेन धात्वन्तरादपि भवति कारकान्तरेऽपि । परितः खाता परिखा आखा । Other Paniniyas interpret अपि similarly, eg., K 3011 (IV.83).
1
1
8. 3.2.134 आक्वेस्तच्छीलतद्धर्मतत्साधुकारिषु । is a heading valid through 3.2.177, providing that the affixes introduced by rules of this section are introduced to signify agents thus qualified. 'up to and including' is used, so that ferq by 3.2.177 is introduced 'under these conditions, as is also the same affix introduced by 3.2.178.
9. Kas 3.2.178 अन्येभ्योऽपि धातुभ्यस्ताच्छीलिकेषु क्विप् प्रत्ययो दृश्यते युक् भित्I cannot take up here the reasons-given in the Bhasya on 3.2.178 and elsewhwere--for stating this sūtra.
10. 5.3.14 is part of the section of rules headed by 5.3.2 किंसर्वनामबहुभ्यो ऽद्वयादिभ्यः । 11. ५.४.४४-४९ प्रतियोगे पञ्चम्यास्तसिः । अपादाने चहीयरुहो । अतिग्राव्यधनक्षेपेष्वकर्तरि तृतीयायाः । हीयमानपापयोगाच्च । पह्या व्याश्रये रोगाच्चापनयने ।
:
I

Page #14

104
George Cardona
Jambu-jyoti
12. Bh. II.405.14-15 : इह कस्मान भवति सः तौ ते । भवादिभिर्योग इति वक्तव्यम् । के पुनर्भवादयः । भवान्
दीर्घायः देवानाम्प्रिय: आयुष्मानिति । Later Paniniyas consider that दृश्यते in the sutra has the effect of so delirniting its application;e.g., Kas. 5.3.14 : दृशिग्रहणं प्रायिकविध्यर्थम् ।
तेन भवादिभिर्योग एवैतद्विधानम् । 13. VS 6.18 : रेडस्स्यग्निष्ट्वा श्श्रीणातु... (Fat,) you are harmed; may the fire cook you...' 14. With the affix णिनि bv3.2.86: कर्मणि हनः (णिनिः ७ि८).
15. In his first värttika on 3.1.87, Kātyāyana says this sūtra serves to formulate
a restriction : ब्रह्मादिषु हन्तेः क्विब्वचनं नियमार्थम् ॥ In the Bhasya, Patanjali notes that the restriction is two-fold, with respect to verbal base (479fm) and the time reference (कालनियम) : Bh. II. 112.9.12 : नियमार्थोऽयमारम्भः । ब्रह्मादिष्वेव हन्तेः क्विब्यथास्यात् । किमविशेषेण । नेत्याह। उपपदविशेष एतस्मिंश्च विशेषे। अथ ब्रह्मादिषु हन्तेणिनिना भवितव्यम् । न भवितव्यम् । कि कारणम्। उभयतो नियमात्। उभयतो नियमोऽयम्। ब्रह्मादिष्वेव हन्तेर्भूते क्विब्भवति क्विबेव च ब्रह्मादिष्वति । As Kaiyata points out in his Pradīpa on this passage, Patañjali accepts only these two restrictions, not the four-fold restriction - such that the upapada and affix are also restricted-which others accept. Kaiyata also remarks that this four-fold restriction is not to be granted acceptance, since it conflicts with the Bhāsya : Pr. III. 255: एतदेव नियमद्वयं भाष्यकारेणाश्रितम् ब्रह्मदिषु हन्तेरेव भूते क्विब्भवति ब्रह्मादिषु हन्तेः क्विब्भूत एवेत्येतत्तु नियमद्वयं नाभ्युपगतम्। अन्यस्त्वाहोपुरुषिकया चतुर्विधो नियमो व्याख्यातः । स भाष्यविरोधान्नादरणीयः । The source Kaiyata ridicules here is the Kāśikā, which, in its commentary on 3.1.87, does indeed say the rule involves four restrictions. Additional details need not be discussed here, since they do not affect my main topic.
16. Here too there is a difference of opinion. The Kāśikā interprets 3.2.76 as introducing
faqafter all verbs, whether or not they are construed with an upapada, and to form derivates used both in Vedic and in the spoken bhāsa of Panini's time : सर्वधातुभ्यः सोपसर्गेभ्यो निरुपसर्गेभ्यश्च छन्दसि भाषायां च क्विप्प्रत्ययो भवति। The Dipaprabha on Vas 23 (p. 58), on the other hand, says 3.2.76 introduced the affix only to bases construed with an upapada :क्विप्चेति सोपपदेभ्यः सर्वधातुभ्य: सामान्येन विहित: क्विप्|If a derivate likeगिर् (nom. sg. :) 'speech' is derived with feqyby 3.2.76— thus, e.g., SK 2983 (=3.2.76 [IV.70]) - then obviously the sūtra does not require a verb to be used with an upapada. On the other hand, if the akrtigana सम्पदादि along with varttika9on 3.3.108 (सम्पदादिभ्यः f99) is invoked to account for derivates like fire (thus Bhānuji on Amarakośa 1.6.1), then one can adopt the alternative view. The consequence of this is that such derivates as fire, which must have been known to Panini, since they were used from earliest Vedicon, have to be accounted for by an addition to the grammar. I therefore consider the Kāśikā's interpretation preferable.

Page #15

Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते
105
17. 3.2.87 vt. 2 : तथा चोत्तरस्य वचनार्थः !! Bh. II. 112.14-15 : एवञ्च कृत्वोत्तरस्य योगस्य वचनार्थ
उपपत्रो भवति बहुलं छन्दसीति । यो मातृहा पितृहा भ्रातृहा । न च भवति अमित्रघातः। 18. The Paippalada Atharvaveda has (19.46.14) : यो मातृहा पितृहा स्वसृहा, with स्वसृहा'one ____who has killed his sister'. 19. Atharvaveda Paippalada 2.88.lab : शास इत्था महाँ अस्यमित्रघातो अद्भुतः | The Saunakiya
Atharvaveda (1.20. 4ab) has शास इत्था महाँ अस्यमित्रसाहो अस्तुत: I and the Rgveda
(10.152.1ab) has शास इत्था महाँ अस्यमित्रखादो अद्भुतः । 20. 3.2.1. कर्मण्यण । See 2.6.5. 21. Kas 3.2.75 : छन्दसीति निवृत्तम् । अन्येभ्योऽपिधातुभ्योऽनाकारान्तेभ्यो मनिन् क्वनिप् वनिप् इत्येते प्रत्यया
दृश्यन्ते विच्च । सुशर्मा । क्वनिए प्रातरित्वा प्रातरित्वानौ । वनिप् विजावा अग्रेगावा । विच् खल्वपि रेडसि पर्ण न वेः । अपिशब्दः सर्वोपाधिव्यभिचारार्थः । निरुपपदादपि भवति धीवा पीवा । दृशिग्रहणं प्रयोगानुसरणार्थम् ।
22. 3.2.1 vt. 4-5 : अपरिगणनं वा ।। अनभिधानात् ।।
23. These are derivates with the affix उण् by Unadisutra 1.1 : कृवापाजिमिस्वदिसाध्यशूभ्य
39Note that has a general reference, to anything that is sweet to the taste (स्वद्यते), but that कारु refers specifically to an artisan (शिल्पिन्), notjustanyone who does something ( fe). Pāņiniyas are therefore doubtless correct when they say that सझायाम्is understood in 3.3.1. Accordingly, the sutra provides that the affixes in question are introduced to form derivates with specific referents and this variously,
so that some of the derivates can have general referents. 24. This is a derivate with मनिन् by Unadishtra 4.144 : सर्वधातुभ्यो मनिन् !, which allows
this affix to follow all verbal bases. Of is the first example given in the Kāśikā on 3.3.2, which also gives the examples चर्मन् 'skin' and भस्मन् 'ashes', derived from चर् 'go about' and भस् 'shine'. In his comments on Unadisutra 4.144, Ujjvaladatta
also gives these derivates and refers to Astādhyāyī 3.3.2. 25. These are the derivates given in the Kāśikā, which treats the set gamyādi as an
exhaustive listing (परिगणनम्). Of these, only गमिन्, आगामिन, भाविन, and प्रस्थायिन् are directly derivable by known Unadisutras:4.6-9: गमेरिनिः 1 आडि णित् । भुवश्च । प्रेस्थः । The Padamanjari on Kasika 3.3.3 refers to these - but with variants आङि णिच्च and प्रात्स्थ: for Unadisutra 4.7,9-and notes alternative ways of accounting for the other examples : प्रतिरोधिन, प्रतिबोधिन्, प्रतियोधिन् and प्रतियायिन् are derived with णिनि by Astadhyayi3.1.134: नन्दिग्रहिपचादिभ्यो ल्युणिन्यचः।। प्रतियोगिन्is derived with घिनुण by Astādhyāyī 3.2.142: APEo; others say some of the derivates are formed with forfa by Astadhyayi 3.2.78 : सुष्यजातौ णिनिस्ताच्छील्ये।; and still others say that these derivates are simply given as ready-made in the list accompanying Astādhyāyī 3.3.3. PM

Page #16

106
George Cardona
Jambu-jyoti
III.7 : गमी आगामीति । गमेरिनि: आङि णिच्चेतीनिः । भावी प्रस्थायीति । अस्मिन्नेवाधिकारे भुवश्च प्रात्स्थ इतीनिः । णित्त्वाद् वृद्धिर्युक्च । रुधिबुधियुधियातिभ्यः प्रतिपूर्वेभ्यो ग्रहादिणिनिः । अस्मादेव निपातनादित्यन्ये । सुप्यजातावित्यन्ये । प्रतियोगीति। सम्पृचादिसूत्रेण घिनुणि चजोः कु धिण्यतोरिति कुत्वम्। णिनिप्रत्यय एव
न्यड्क्वादिपाठादस्मादेव निपातनाद्वा कुत्वमित्यन्ये । 26. 3.3.1 vt. 1 : बाहुलकं प्रकृतेस्तनुदृष्टेः प्रायसमुच्चयनादपि तेषाम् । कार्यसशेषविधेश्च तदुक्तम्... This
ślokavārttika and Patañjali's comments on it appear to assume, as later Pāņiniyas explicitly state, tha: the derivations are according to sūtras of another grammar, the Uņādisūtras. I do not enter into this question here, except to remark that, in the Indian context, a mere list of affixes without rules introducing them to bases, would indeed be anomalous. Jinendrabuddhi (Nyāsa III. I) cites a traditional verse stating on what basis the property of applying variously is attributed to operations : because an operation sometimes applies, sometimes does not apply, sometimes applies optionally, and because sometimes some other operation applies: Wafacsafa: क्वचिदप्रवृत्तिः क्वचिद्विभाषा क्वचिदन्यदेव । विधेविधानं बहुधा समीक्ष्य चतुविधं बाहुलकं वदन्ति ।। This verse is well known, cited elsewhere in full (e.g., Anantabhatta on Vajasaneyipratisakhya 3.18, Hemacandra on Siddhahemasabdanusāsana 8.1.2) or in part (e.g., PM III.I) as well as simply alluded to indirectly (e.g., Uvata on
Vājasaneyiprātiśākhya 3.18). 27. Kas 3.3.1 : उणादयः प्रत्यया वर्तमानेऽर्थे सज्ञायां विषये बहुलं भवन्ति । यतो विहितास्ततोऽन्यत्रापि भवन्ति ।
केचिदविहिता एव प्रयोगत उन्नीयन्ते । 28. Kas 3.3.2 : पूर्वत्र वर्तमानाधिकाराद् भूतार्थमिदं वचनम् । भूते काल उणादयः प्रत्यया दृश्यन्ते । वृत्तमिदं
वम । चरितमिदमिदि चर्म। भसितं तदिति भस्म । दृशिग्रहणं प्रयोगानुसारार्थम् ।
29. This is brought out in the Bălamanoramă on Siddhāntakaumudi 3169 (= 3.3.8
[IV.307) explaining Bhattoji's statement that some affixes not explicitly provided for are to be inferred (केचिदविहिता अप्यूह्याः) उणादयो वेत्यनुक्त्वा बहुलग्रहणस्य प्रयोजनमाह
केचिदविहिता अप्यूह्या इति। 30. Siddhāntakaumuditattvabodhini IV.308 : नन्वेवं वर्तमानग्रहणं च उणादयो बहुलमित्यत्र
नानुवर्त्यताम् । एतच्चोत्तरसूत्रं च त्यज्यताम् । अविशेषेण कालत्रयेऽपि प्रत्ययलाभादिति चेदत्राहुः । बाहुल्येन वर्तमाने भवन्ति भूतभविष्यतोस्तु क्वचिदेवेति विवेक दर्शनार्थमिति । Candrakala on Laghusabdendusekhara 3.3.1 (II.816): न च वर्तमाने इति सूत्राद्वर्तमान इति नानुवर्तनीयम् । एवञ्च भूतादौ सिद्धिर्भविष्यत्येवेति भूतेऽपीत्यादि न कार्यमिति वाच्यम् । प्रायेण वर्तमान एव क्वचिदेव तु भविष्यति वेति व्युत्पादनाय पुनः सूत्रारम्भात् । Similarly, Nyāsa and Padamanjari on Kasika 3.3.2
(III.6). 31. Kas. 6.4.73 : यतो हि विहितस्ततोऽन्यत्रापि दृश्यते। 32. 6.4.134 : अल्लोपोऽनः ।

Page #17
Paninian Sutras of the Type अन्येभ्योऽपि दृश्यते 107 33. Kas 7.1.76 : A fafectionszufu Ered ! statym4 i 3115cafu Erud I T LIET Strefy: (RV 1.84.13a) 496 pierdutartiefstal: 11 (VS 25.21=Rv 1.89.85) i catanदिष्वित्युक्तम् / अतृतीयादिष्वपि दृश्यते / अस्थान्युत्कृत्य जुहोति / विभक्तावित्युक्तम् अविभक्तावपि दृश्यते / Baaa! Hinta 372-4 vidret fatta (RV 1. 164.40) || 34. I have considered only the Paninian evidence because of space limitations. Comparable statements are found elsewhere in early Paniniya works : 2.1.33 vt. 1 (Sintetice R$250f9 Trza I), 5.1.57-58 vt. 6(372927STER GRETA 10), 5.2.112 vt. 1 (area USR uzsfq apud II), 5.2.120 vt. 1 (42 ustafo exud 11), Bhasya on 3.2.48 (II. 103. 14:3797 37EUR 58 Russafer Erga ispirato at vanillo ano 1). 35. Of course, he also takes earlier archaic usage into account, as has been noted in sections above. 36. This is connected with the issue of how, in ancient India, one viewed the usage we refer to as Sanskrit, including both the spoken language of Panini's period, with its dialect variations, and earlier Vedic : as manifestations of a single eternal language or a product of human beings. Although the Astadhyayi does not say anything explicit concerning this, Katyayana and Patanjali do consider the issue whether speech is eternal or produced. I cannot enter into a discussion of these questions here. They have been dealt with recently by Deshpande; see his Sanskrit and Prakrit : Sociolinguistic Issues (Motilal Banarsidass, Delhi 1993), pp. 53-74. Let me say only that I see no contradiction between Panini's describing a state of language such that certain developments had begun but not yet been carried out and Paniniyas' maintaining theoretically that all manifestations of Sanskrit speech represent an eternal language. There can be, in the Paninian scheme of things, a perennity of a flow (giftrat) such that what appear to be newly created elements are merely manifestations of entities always there in the eternal flow. Obviously, I do not think that the conclusions I have reached on the basis of the evidence considered here require one to consider Panini a linguist who claims to be describing a language historically. 000