THE ORDERING OF THE RULES IN PANINI'S GRAMMAR By S. D. JOSHI In the Astadhyayi, excluding the One-name rule+ and the Asiddhatva rules? Panini did not mention any additional principle of ordering, to decide which of two rules, applicable at a particular stage of derivation, should precede the other. This does not mean that the rules in the Assadhydyi, where a mechanical procedure for deciding the order is not available, apply in an arbitrary order. Tn such cases also, the order of application is consi- dered to be basically predictable, The rules are applied in the specific order because in many cases the application of one rule feeds the application of the other rule, That is to say, the rules often become automatically ordered because the application of one is dependent on the application of other. The special devices mentioned by P&nini for ordering imply that there are some general conventions for the ordering of the rules, which are assumed by Panini himself, but are not stated in the body of the Grammar. The first basic simple convention, assumed by Panini* with regard to the ordering of. the rules, appears to be that the general rule (utsarga or sdmanya) and the specific rule ( apavdda or visesa), stated in the Asradhyayl, are disjunctively ordered, in the sense that the application of the apavdda rule precludes the application of the uisarga rule, for instance,! P. 8. 2. 31, and 1. A kadarad cha sathjiié; vipratisedhe parath Raryam (P. 1.4, 1-2) : ‘up to kadara ( P, 2.2.38 ), one technical term { is applied ); ( and ) in case of a conflict, apply the later (technical term)’, The traditional interpretation of P. 14.2 assumes that in case of a conflict the later rule prevails, and this priaciple is valid uhoughout the Grammar B, Faddegon (Studies om Pazini’s Grammar, Amsterdam, 1963, pp. 26-27) first pointed out that this is overgencralization. This principle 4s relevant for the One-name Section only. Normally, rules introducing technical terms are not ordered. Only in the Section referred to, the rulen concerning technical terms are disjunctively ordered. 2, Asiddhavad atra bhat (P. 6.4.22); parvatrasiddham (P, 8.2.1): ‘Inthe A bhat Seco tion two rules do not interact; and in the Triptd? Section, the earlier rules prevail over the subrequent rules, and the earlier rules cannot be applied after the application of the subsequent rules,’ 3. Pinini offers a clue to this principle in his rule P. 3.1.94 where he states that in this particular section the afavade does not cancel the ufsarga, i.e. the apavada and the wrarga are conjunctively ordered. ‘This offers a clue that in the remaining section of the Asfa- dhy8yt the opauida and the utsarga rules are disjunctively ordered. 4, George Cardona, “ Some principles of Pinini’s Grammar", Journal of Indian Philo- soply, Vol. 1 (1970 }, Holland, pp, 39-42,668 ABORI : Diatnond Jubilee Volume P.8.2.32, The modification h gh prescribed (P. 8. 2.32) for the final h of a root beginning with d in certain environments, is a special operation, where- as, the modification hk > dh (P. 8.2.31) prescribed for the final / of any root in the same environment is a general operation. Therefore, the h -> gh rule and the dh rule are disjunctively ordered. The h->gh rule is first applied, and in the remaining domain the h > dh rule is applied. The Asiddhatva principle stated by P. 8. 2. 1 implies that, after apply- ing the rules mentioned in the first seven books and in the first quarter of the eighth book, we should proceed to apply rules stated in the last three quar- ters of the eighth book. Even in the last three quarters of the eighth book an earlier rule takes precedence over a subsequent rule, An earlier rule can- not be applicable after the application of a subsequent rule, Here rules are ordered according to the number in which they are stated. If the rules are applied in the same sequence in which they are given, they become conjunc- tively ordered. But, at a particular stage of derivation, if a later rule is first applied, it precludes the application of an earlier one. In such cases, the later rule and the earlier rule become disjunctively ordered. For instance, in rdjabhih after dropping -n of the stem rajan by P. 8.2.7 we should not proceed to apply the modification a->ais (P. 7.1.9). Here P. 8.2.7 and P. 7.1.9 are disjunctively ordered. In pakva, the modification c+ & condi- tioned by the jhal environment (P. 8.2.30) takes precedence over the modi- fication ta va ( P. 8.2.52). At the stage pac + 1a, P, 8.2.30 and P, 8.2.52 are simultaneously applicable. If P. 8.2.52 is allowed to take effect first, P. 8.2.30 cannot be applicable at the stage pac + va, because the jhal environ- ment is lost. Therefore here P. 8.2.30 and P. 8.2.52 are conjunctively ordered.§ The Asiddhatva principle stated by P. 6.4.22 assumes either (i) that the rules in the 4 bhae section are unordered and that they apply simultaneo- usly, or (ii) that any two rules in this section do not interact at all. The latter statement implies that the application of rule A does not feed or bleed the application of rule B, and that the application of rule B does not feed or bleed the application of rule A. In other words, in connection with the rules mentioned in this section, the output of one rule cannot be the input of another one. The example for the simultaneous application is sadhi, at the stage Sas + hi, the modification is > S@ (P. 6.4.35) and the modification hiadhi (P. 64.101) are simultaneously applied. If it is assumed that §as—> $@ (P. 6.4.35) and hi- dhi(P. 6.4.101) are applied in a fixed order we will not be able to derive the form sadhi. If the hi dhi rule applies first, the sas—> $¢ operation cannot become effective, as its application is condi- 5, For more examples, see H. E. Buiskool, The Tripadi, PArealrasiddham, Leiden 1939,JosHI { The Ordering of the Rules in Panini’s Grammar 669 tioned by the presence of hi. On the other hand, if we apply the S@s > Sa operation first, the hi dhi operation has no chance to apply, because the preceding jhal (i, ¢. s—) environment is no longer present. Since these conditioning environments cannot become sequentially available, P. 6.4.22 allows simultaneous application. That is to say, in this example the bleeding order is avoided. The example for avoiding the feeding order of rules is jahi. At the stage han + hi, the han-> ja rule (P. 6.4.36) is applied first. After its application, the a + hi>a +O rule (P. 64.105) becomes applicable: But the Asiddhatva principle in the A bhaz Section states that any two rules in this section do not feed or bleed each other. Therefore, although P. 6.4.36 provides the conditioning environment for the application of P. 6.4.105, it is not applied. But we see no justification for the simultaneous application of rules. The hypothesis that rules apply in sequence and in a fixed order would solve the problem. If we phrase the rule 4 dhau instead of {4 hau (P. 6.4.35) and make the conditioning environment —dhi instead of —Ai, the linear prin- ciple of ordering would explain the form. First P. 6.4.101 is applied, and then P. 6.4.35 (i. e. S@ dhau) where the former feeds the latter. For this we will haye to place P. 6.4.101 earlier, and P. 6.4.35 later. That is to say, we need not assume mutual non-interaction of the rules in the A bhdt section. Forms like jahi can be derived by the simpler hypothesis of linear ordering. If we strictly order the rules P. 6.4. 105 and P. 6.4.36, we could say that after P. 6.4.36 we are not supposed to apply P. 6.4.105. To achieve this desired result we will have to place P. 6.4.105 earlier and P, 6.4.36 later in the text of the Asrddhyayi. Instead of changing the place in the Asrddhyayi Pani gives the principle of reverse ordering. The rules P. 6.4.36 and P. 6. 4. 105 are reversly ordered, in the sense that after the application of P. 6.4.36, P. 6.4.105 is not to be applied. But we do not get any cogent example for mutual non-interaction or mutual reverse ordering of rules in the sense that tule A does not feed or bleed rule B, and rule B does not feed or bleed rule A. Instead of drawing help from the strange hypothesis of simultaneous appli- cation or that of mutual reverse ordering, we can manage with the simpler principle of linear ordering. But for that we will have to change the numbering of the rules in the text of Asradhyayi. The later overgenaralised paribhasas offer complex principles for deci- ding the ordering of rules. Paribh@sés XXXVII-LVUI give nityafanitya, and antarariga/bahirariga principles in determining the order of application of the rules in Panini's Asgadhydyi. These are overgeneralised conventions. To 6. F. Kiclhom, Paribhapendutehhara of Nagojibhatta, part 11, second ed, Poona, 1960-670 ABORI : Diamond Jubilee Volume counter the over-application of these conventions, many artificially sounding ad hoc principles, which again have many counter-examples, are added in the paribhdsd literature. All these principles’ taken together appear to propose : (i) Word-building rules precede phonologicol rules, and (ii) phonological rules are applied first to the innermost layer of constituents, then to the next layer, and so on until the largest constituent, the whole sentence, is reached. From the preceding discussion it will be clear that Panini's Astadhyayi introduces some special devices for the ordering of the rules. The later paribhasas increase the number of formal devices which are not fully justified. These later principles handte only marginal cases. The organisa- tion of rules in the Sapddasaptadhyayi appears to be arbitray in the sense that sequential ordering is not given in it. Yet, the ordering of the rules which are unordered is basically predictable, because the application of one rule is dependent on the other. In the case of two unordered rules we see that one rule introduces the environment for the application of the other rule. The derivation of forms assumes maximum application of rules in the conjuctiye order, although Panini does not give the formal principle governing it. With regard to a strict ordering of rules in Panini's Grammar Ghatage* amakes the following observation. A strict ordering of rules, however, is something different. In it, the application of the second rule depends on the first in deciding its scope of application which is what is left over with the application of the preceding tule. A change in the order of the rules will give a different result as the ranges of the reordered rules become different. Only in such cases we can speak of strict ordering of rules. It is a matter of some doubt whether Pinini's grammar permits such an ordering... A genuine case of a strict ordering of rules is, however, found in the Varstika on P. 1.4.32 and Pataii- jali’s discussion on it. To account for the dative in connection with intransitive verbs, it is suggested that, in addition to the word Aarmand in P. 1.4.32, the word Kriyayd should be stated. This gives us two rules? : 7. SD. Joshi and Paul Kiparski, “The word-building rules in Pauini’s grammar”, unpublished, 8. A. M. Ghatage, “Panini 1.4.92,” IL ( Vol. 29, 1968), Katre Felicitatien Volume, part I, pp. 150-54. ‘9. 8, D. Joshi and J. A. F. Roodbergen, Pataijali's Uydkarana-Mahabhagya, Karakalni- ka (P. 1,4.23-1.4.55), Publications of the entre of Advanced Study in Sanskrit, Univer- sity of Poona, 1976, pp. 119-21,JOSHI : The Ordering of the Rules in Panini’s Grammar on (i) karmand yam abhipraiti sa sarpradanam : ‘what one has in view through the karman, that is called sartpradana; and Gi) kriyayé yam abhipraiti sa sarspraddnam : ‘what one has in view through the action is called saripradana’. The first rule (i) covers the use of the dative in connection with tran- sitive verbs, and the second in connection with intravsitive verbs. But the second rule would be applicable to give the designation sarpraddna to the Karman : ‘ direct object * also, because the direct object also one has in view through the action expressed by the verb concetned. Now if we assume that rules (i) and (ii) are to be applied in any order, and if rule (ii) had been allowed to apply before rule (i), we will have such unacceptable usages as upadhyaydya gave dadati : ‘he presents a cow to his teacher’, permitted as grammatical. This difficulty, according to Ghatage,"* is removed by taking recourse to the procedure of a strict ordering of rules. What Ghatage means is as follows : Rules (i) and (ii) are ordered in the sense that if rule (i) is appli- cable, rule (ii) is not permitted to apply. Rule (ii) is restricted to the domain left over by rule (i). Once these are disjunctively ordered the difi- culty noted above does not arise. Once rule (i) has been applied in the transitive environment, the designation saripradana is prevented from apply- ing to the karman go by rule (ii), because after the application of rule (i) we should not proceed to apply rule ( ii). Ghatage™ assumes that this strict ordering of rules (i) and (ii) is probably indicated by the word api in the Varttika, kripagrahanam api kartavyam. The first thing that I want to point out here is that kriyagrahanam api kartavyam is not a Varitika.™® The Keilhorn text does not give it as the Vérttika at all. Only the Banaras edition has inserted it, on what ground being not clear. Further, the word api appearing in the Bhdsya-text cannot indicate a strict ordering of rules, as Ghatage claims. No other example is known to assume that the particle api indicates a strict ordering. The parti- cle api*® probably hints that the rule has marginal cases. Thus we cannot claim that any condition is introduced in the rules (i) and (ii) to indicate precedence of rule (i) over rule (ii). In other words, 10. A. M. Ghatage, of. cit, p. 154. LL, Bid, p. 154. 12, S, D. Joshi and J. A. F. Roodbergen, op. cit, p. 121. 13, See P, 3.1.84; 2.61; 2.75; 2101; 2.178; 3.130; 5.3.14 ete,672 ABORI : Diamond Jubilee Volume a strict ordering of rules (i) and (ii) cannot be attributed to Katyayana or Patafijali, The undesired consequence of the over-application of the desig- nation sarapraddna to a karman is not removed by a strict ordering of rules (i) and (ji), as Ghatage claims, but by the procedure prescribed in case of a conflict between two designations as laid down by P. 1.4.1-2* In case of a conflict, according to P. 1.4.2, the designation prescribed by the later rule prevails. In upadhyéyaya gan dadati, the designation sampradana is appli- cable to go by rule (ii) on P. 1.4.32, and the designation karman is applica- ble by P. 1.4.49. The designation karman prevails over the designation sarhpradéna because it is prescribed by the later rule. Here P. 1.4.2 deter mines the priority, and we cannot say that rules (i) and (ii) are formulated with great care so that the application of the second rule depends on the first in deciding its scope for application. In fact, rule (i) is redundant, because tule (ii) can cover the use of the dative in connection with both transitive and intransitive verbs. We can dispense with the additional Bhdsya rule by interpreting the word karmand in the sense kriyaya."* Let us take the word karmand in its normal meaning : * direct object *. Now, by mentioning the condition karmand in the dative rule (P. 1.4.32), Panini indicates that P. 1.4.49 and P. 1.4.32 are conjunctively ordered. That is to say, P. 1.4.49 is applied first and in the domain left over with the appli- cation of P. 1.4.49 we should proceed to apply the dative by P. 1.4.32. This strict ordering of two general rules dealing with the direct and the indirect object is indicated by word karmand mentioned in P. 1.4.32. Normally, the general rule applies in the domain left over with the application of the special rule. But P. 1.4.27 offers the example for the reverse order between the general and the special rules. P. 1.4.27 which prescribes the special ablative in connection with the verbs meaning ‘ keep- ing back ’ ( vdrandrtha ) applies in the domain left over with the application of the general karman rule (P. 1.4.49). ‘This is suggested by the original reading, varanarthanérh karmana ipsitah® P. 1.4.27 prescribes the designation apadana for the ipsita item in connection with the verbs meaning ‘ keeping back from’. In this rule Patafi- jali takes ipsita in its literal sense, namely, an object reached by the action. Pataiijali offers the example, agner mdnavakarn varayati : ‘he keeps the boy away from the fire’. Pataiijali then raises the difficulty that in the above 14, $. D. Joshi and-J. A. F. Roodbergen, cit, p. 109, 15, Panini also uses the word Karman ( P. 1.3.14,3.3.43 ) ; karana ( P. 3.1.17), adhikaraza (P.2.4.13), ete. in the non-technical sense. See F. Kielhorn, op. cit, paribhard ix. 46, 8, D, Joshi and J. A, F, Roodbergen, of. cit., p, 81. =JOSHI : The Ordering of the Rules in Panini’s Grammar 673 mentioned example both agni and manavaka would be called apadana beca- use both are reached by the action of preventing. He removes this difficulty by rephrasing P. 1.4.27 as vdrandrthandm karmano yad ipsitam : ‘in connec- tion with the verbs meaning “keeping back from doing something”, the thing desired to be reached by the direct object is called apadana’. The effect of the rephrasing is that the rules P. 1.4.49 and P. 1.4.27 become conjunctively ordered. First P. 1.4.49 is applied, and then in the remaining domain we apply P. 1.4.27. Thus in agner manavakart varayati the designa- tion karman is applied to manavaka first, and in the remaining domain the special designation apadana, prescribed in connection with the verbs meaning “keeping back’ ( vdrandrtha), is applied. Varttika | on P. 1.4.27 says that the word karman in this rule is re- dundant because the demarcation between the domains of the apadana (P. 1.4.27) and the karman ( P. 1.4.49) has been already made by tama in kartur ipsitatamam karma.” What Katyayana means is that the designation karman applies to the items which are directly ( positively ) reached, whereas the de- signation apadana applies to the items which are indirectly reached (barely reached ) by the action of preventing. From this Vattika statement we get the impression that Panini’s original rule contained the word karman (i. e. karmanah). The original rule should be vdrandrthdnarh karmana ipsitah. ‘The present rule which has come down to us without the word karmanah in- corporates Katyayana’s later suggestion. ‘The designation karman has scope where the apaddnasarhjfid prescribed in connection with the varandrtha verbs is not applicable. The designation prescribed by P. 1.4.27 is apavada : ‘exception’. Thus, the special designa- tion apavada in connection with the vdrandrtha roots would prevail over the general designation karman. In the normal circumstances of the utsargapa- vada relations, the ordering of the rules will be disjunctive, i. e. the applica- tion of the special rule would preclude the application of a general rule, However, in connection with the varanartha roots Panini indicates departure from the normal principle of disjunctive application of the utsargapavada by adding the word karman in P. 1.4.27. Panini makes special devices governing special cases of ordering. He does not give a general principle for the ordering of the rules. But these special statements are of great interest because they will eventually give us knowledge of some underlying principles for the ordering of the rules. ‘There might be some general conventions which are assumed by Panini him- 17, Ibid., p. 84. 85 Annals [D. J.]674 ABORI : Diamond Jubilee Volume self but about which we are stil! in the dark. We cannot take it for granted that the traditional interpretation is nearly infallible. Our task should be to re-examine the rules of Panini in a critical way. It will give us different insights about the metatheory assumed by Panini which we might have missed in the traditional interpretation.