1 5 Analysis of the Traditional Perspective

Let us look at 1.4.2 “vipratiṣedhe paraṁ kāryam” again. Pāṇini does not explain the meaning of vipratiṣedha in the Aṣṭādhyāyī. The Kāśikā claims that vipratiṣedha means tulyabalavirodha ‘conflict between two equally powerful rules.’ This is a plausible assumption because, in Sanskrit literature, the term has been used to mean ‘the opposition of two courses of action which are equally important, the conflict of two even-matched interests’1. But which conflicts qualify as tulyabala ‘having equal strength’? The Kāśikā says that rule pairs which are not nitya-anitya, antaraṅga-bahiraṅga, apavāda-utsarga are tulyabala ‘having equal strength’.

Let us try to understand why the tradition felt the need to come up with these tools. According to the tradition, para in 1.4.2 means ‘the rule that appears after another in the serial order of the Aṣṭādhyāyī’. Thus, in the case of a conflict (vipratiṣedha) between two rules, the operation prescribed by the later rule should prevail. However, if one assumes that any rule conflict can be called vipratiṣedha, and therefore applies 1.4.2 uniformly to every instance of such a conflict, in many cases one gets the wrong answer at the end of the derivation. Let us consider a few examples: tud is a 6th class root which can take both parasmaipada ‘active’ and ātmanepada ‘middle’ endings.

When deriving its present third person singular form, two rules become applicable at the step tud + tiP. One is 7.3.86 pugantalaghūpadhasya ca (sārvadhātukārdhadhātukayoḥ guṇaḥ)2, which teaches that the penultimate light vowel iK (i, u, r̥, l̥) is replaced with guṇa (a, e, o) when followed by a sārvadhātuka or ārdhadhātuka affix.

23

The other is 3.1.77 tudādibhyaḥ śaḥ (sārvadhātuke kartari), which teaches the addition of affix Śa after roots belonging to the class starting with tud in the Dhātupāṭha, when the root is followed by a sārvadhātuka affix in an active construction. Note that, since 7.3.86 comes after 3.1.77 in the serial order of the Aṣṭādhyāyī, according to the traditional understanding of 1.4.2 it should win, but applying 7.3.86 would give the wrong answer: tod + tiP (7.3.86) 🡪 tod + Śa + tiP (3.1.77) 🡪 *todati.

Notice that 3.1.77 is applicable after 7.3.86 applies, as seen in the derivation above. On the other hand, if 3.1.77 applies first, we get: tud + Śa + tiP. Since Śa is marked with a Ś, it is sārvadhātuka by 3.4.113 tiṅśit sārvadhātukam, and being a sārvadhātuka which is not marked with a P, it is treated as if marked by Ṅ by 1.2.4 sārvadhātukam apit (ṅit). By 1.1.5 kṅiti ca (na iko guṇavr̥ddhī), the guṇa replacement of u in tud by 7.3.86 is no longer possible. So 7.3.86 is not applicable once 3.1.77 has applied.

Thus, 3.1.77 and 7.3.86 are nitya and anitya respectively. If the nitya rule, i.e., 3.1.77 wins, we get: tud + Śa + tiP (3.1.77) 🡪 tudati, which is the correct answer. In this example, relying on paratva gives the wrong answer, but using nityatva gives the right answer. We shall come back to this after we look at a few more examples.

syona

Consider the next example: to derive syona ‘a sack, something stitched’, na is added to siv ‘to sew, stitch’: siv + na (3.3.1 uṇādayo bahulam).3 First, by 6.4.194 chvoḥ śūḍ anunāsike ca (kvijhaloḥ kṅiti), v of siv is replaced with ū: (siū) + na. Now, two rules are simultaneously applicable here: 6.1.77 iko yaṇ aci, which is caused by ū and prescribes the replacement of i with y, and 7.3.86 pugantalaghūpadhasya ca, which is caused by na and prescribes the replacement of i of si with its corresponding guṇa (i.e., e). Since 7.3.86 comes after 6.1.77 in the serial order of the Aṣṭādhyāyī, by 1.4.2 it should win. But applying 7.3.86 gives us the wrong answer.

24

According to the Paribhāṣenduśekhara, ‘antaraṅga is (a rule) the causes (of the application) of which lie within (or before) the sum of the causes of a bahiraṅga rule’. So, in the case of siū + na, ū, the cause of 6.1.77, lies before (i.e., to the left of) na, the cause of 7.3.86. Thus, 6.1.77 is antaraṅga whereas 7.3.86 is bahiraṅga. Using the antaraṅgatva tool, 6.1.77 wins. We get syū + na 🡪 syo + na (sārvadhātukārdhadhātukayoḥ) 🡪 syona, which is the correct answer. Relying on paratva gives the wrong answer, while using antaraṅgatva gives the right answer.

Let us look at one more example. Two rules, namely 1.4.16 siti ca (padam) and 1.4.18 yaci bham (svādiṣv asarvanāmasthāne) lie in the ekā saṁjñā section (1.4.1 ā kaḍārād ekā saṁjñā). 1.4.1 teaches that up to 2.2.38 kaḍārāḥ karmadhāraye, any item can take only one saṁjñā ‘technical designation’. 1.4.16 siti ca (padam) teaches that an item is called pada when an affix marked with S follows, and 1.4.18 yaci bham (svādiṣv asarvanāmasthāne) teaches that an item is called bha when a y- or vowel-initial, non-sarvanāmasthāna affix belonging to the class starting with sU follows. Consider the example ūrṇā + yuS.5 Here ūrṇā can potentially take two saṁjñās: pada by 1.4.16 and bha by 1.4.18. However, since both rules lie within the jurisdiction of 1.4.1, ūrṇā can take only one of the two saṁjñās. By 1.4.2, the para rule i.e., 1.4.18 should win. But if ūrṇā takes the bha saṁjñā, then ā of ūrṇā gets deleted by 6.4.148 yasyeti ca (bhasya lopaḥ taddhite), which teaches that the final i or a (both short and long) of an item which is termed bha are deleted when followed by an ī or a taddhita affix. This gives us the wrong taddhita stem *ūrṇyu. The Kāśikā says that 1.4.16 is an apavāda of 1.4.18 without justifying this claim.6 If the apavādatva tool is used, 1.4.16 wins, which gives the correct stem ūrṇāyu. Using paratva gives the wrong answer, while using apavādatva gives the right answer.

In all three examples discussed above, using paratva gives the wrong answer, but using nityatva, antaraṅgatva and apavādatva respectively leads to the correct answer. Below, I present an abridged version of how I think the current method of solving rule conflict has gradually evolved.

Having realized that treating all rule conflicts as vipratiṣedha and applying (4.1.76: taddhitāḥ), included among the affixes referred to in 1.4.17-8, and all also begin with y or a vowel. Hence, items occurring before these are eligible for being bha by 1.4.18.” With the help of this information, we can infer that 1.4.18 is applicable wherever 1.4.16 is applicable, but 1.4.16 is not always applicable where 1.4.18 is. 1.4.16 is more specific than 1.4.18 and thus wins.

25

1.4.2 uniformly to every instance of such a conflict gives the wrong answer in many cases, the Pāṇinīyas:

(1) claimed that they found jñāpakas ‘hints or clues’ in Pāṇini’s sūtras which authorised them to devise new tools like nityatva, antaraṅgatva, anavakāśatva, etc., for the purpose of solving rule conflicts;

(2) restricted the jurisdiction of rule 1.4.2 by declaring that vipratiṣedha implies only tulyabala conflicts, i.e., conflicts between equally powerful rules; and

(3) declared that rule pairs like nitya-anitya, antaraṅga-bahiraṅga, and anavakāśa-sāvakāśa were to be called atulyabala ‘not equally powerful’.

This allowed them to exclude the atulyabala rule pairs, namely nitya-anitya, antaraṅga bahiraṅga etc., from the jurisdiction of 1.4.2, thereby containing the problems caused by their interpretation of 1.4.2 to a smaller number of cases. Gradually, the Pāṇinīyas also constructed the hierarchy taught in paribhāṣā 38 of Paribhāṣenduśekhara above to determine which tool takes precedence over which other tools.

However, these post-Pāṇinian tools are not without flaws, to compensate for which umpteen other paribhāṣās have been written by Pāṇinīyas. Many of these paribhāṣās address very specific cases7 or even single examples of conflict, thereby defeating the entire purpose of writing metarules, which is to arrive at broad generalizations that can govern the application of and interactions between the whole body of rules. And even after this, the Pāṇinīyas are not able to solve every case of conflict correctly: every time they falter, they find one tortuous explanation or the other to justify that ‘exception’.

I do not think that all paribhāṣās taught by the Pāṇinīyas should be rejected. Many post Pāṇinian paribhāṣās accurately capture how the Pāṇinian machine functions, and thus they are of great importance to us. They are mostly descriptive in nature and make insightful observations about the Aṣṭādhyāyī. However, we also find post-Pāṇinian paribhāṣās that teach us tools for rule conflict resolution, such as nityatva and antaraṅgatva, which Pāṇini would certainly not have left unstated if he actually wanted to teach them, and which impose post Pāṇinian ideas onto the Aṣṭādhyāyī.+++(4)+++ Thus, the validity of this set of paribhāṣās is questionable.

26

  1. See the entry on vipratiṣedha in Apte’s Sanskrit dictionary. ↩︎

  2. The terms in brackets are anuvr̥tta ‘continued’ from previous sūtras. ↩︎

  3. This is one of only two sūtras that refer to an ancillary text known as Uṇādisūtras, which provide for introducing certain affixes after verb roots to derive nominal bases (Cardona 1976: 170). There is no clear consensus about whether or not Pāṇini himself wrote the Uṇādisūtras (Cardona 1976: 174). I personally do not think he did, and so I do not consider this derivation ‘Pāṇinīya’. But because the commentarial tradition uses this as an example in the present context, I discuss it nonetheless. The relevant Uṇādi sūtra here is 289 siveṣ ṭer yū ca. ↩︎

  4. I am aware that the tradition reads this rule as cchvoḥ… and not as chvoḥ…. However, I think that the original version must have been chvoḥ. See Kiparsky 1982: 89. ↩︎

  5. 5.2.123 ūrṇāyā yus ‘The taddhita suffix yuS occurs to denote the sense of matUP after syntactically related nominal stem ūrṇā “wool”’. ↩︎

  6. I agree with Cardona’s (1970: 46) explanation for this: “Consider now 1.4.16. There are only four affixes marked with S in Pāṇini’s grammar: ghaS (🡪 iya by 7.1.2) introduced by 5.1.106, chaS (🡪 īya, 7.1.2) by 4.2.114-5, yaS (ya) by 5.2.138, and yuS (🡪 aka, 7.1.1) by 5.2.123, 138, 140. For example, r̥tviya- ‘tempestivus’ (<r̥tu ‘appropriate time, season’) contains ghaS. All such affixes are taddhita ↩︎

  7. For example, consider Pbh 52 of the Paribhāṣenduśekhara, antaraṅgān api vidhīn bahiraṅgo lug bādhate ‘A bahiraṅga rule teaching LUK deletion defeats an antaraṅga rule [in case of conflict]’, which is an exception of Pbh 50 antaraṅge bahiraṅgam asiddham ‘An antaraṅga rule treats a bahiraṅga rule as suspended.’ ↩︎