1 3 Modern Perspectives on the Functioning of the Aṣṭādhyāyī

Purpose of vyākaraṇam

Before we explore how modern scholarship perceives the Aṣṭādhyāyī, let us very briefly consider what the tradition, and more specifically Kātyāyana and Patañjali say, about the meaning and purpose of vyākaraṇa. In vt. 14 of the Paspaśāhnika1, Kātyāyana says: lakṣya-lakṣaṇe vyākaraṇam ‘grammar (stands for the combination of) lakṣya i.e., words (and sentences)’ and lakṣaṇa ‘rules’. This is true of any grammar, not just the Aṣṭādhyāyī. But does the Aṣṭādhyāyī have certain mechanistic properties which set it apart from conventional grammars? Below we will look at modern perspectives on this topic. According to Patañjali2, vyākaraṇa serves the following purposes:

14

rakṣohāgama-laghv-asandehāḥ “rakṣā ‘protection of the Vedas’, ūha ‘adapting inflected forms in Vedic mantras as required during rituals’, āgama ‘following Vedic injunctions’, laghu ‘brevity i.e., easy of learning the language’, and asandeha ‘resolution of doubts’”. These certainly are some of the factors that must have motivated Pāṇini to write his grammar. But was Pāṇini also aiming to build a somewhat mechanistic model for deriving Sanskrit words (and subsequently, sentences)? Let us look at what modern scholarship tells us about topics like rule conflict and order of rule application in Pāṇinian derivations, and therefore, about the status of the Aṣṭādhyāyī as a ‘machine’.

Bronkhorst

Let us start by looking at Bronkhorst’s work on this topic. Bronkhorst (2004) shows that Patañjali prefers a linear+++(→memory-less)+++ reading of the Aṣṭādhyāyī, that is, Patañjali believes that in order to decide which rule should apply at any step in a derivation, one need not know the outcomes of previous or following steps. He says, “It is clear from the above that Patañjali tries both to avoid looking back and looking ahead in explaining grammatical derivations.”3 Bronkhorst also points out that the Paribhāṣenduśekhara teaches the metarule pūrva-para-nityāntaraṅgāpavādānām uttarottaraṁ balīyaḥ (paribhāṣā 38) ‘Of [these five kinds of rules, - viz.] a preceding [rule], a subsequent [rule]4, a nitya [rule]5, an antaraṅga [rule]6, and an apavāda [rule]7, - each following [rule] possesses greater force [than any one of, or all, the rules which in this paribhāṣā are mentioned before it].’8

15

He concludes:

“…(this)9 clearly shows that, according to the traditional view, decisions concerning the continuation of a grammatical derivation at any particular point are taken on the basis of the situation at hand. More specifically, no information about the earlier or later phases of the derivation is required to make a correct decision at any stage.”10

Bronkhorst states that he is unconvinced by Patañjali’s evidence suggesting that the Aṣṭādhyāyī functions linearly+++(→memory-less)+++. He thinks that Pāṇini did not intend for the Aṣṭādhyāyī to be approached linearly+++(→memory-less)+++, and attempts to establish that at least for some derivations, the knowledge of the derivation’s history and/or its future course is essential to select the right rule at a given step.11

One of the reasons Bronkhorst thinks looking ahead into the derivation is required is to determine the order in which two rules should apply with respect to each other.12 Roodbergen has a different opinion on this subject. 13 He recommends some changes to the traditional order in which the following processes occur: the replacement of lakāras ‘tense and mood proxies’ with tiṄ ‘verbal endings’ and the introduction of vikaraṇas ‘affixes placed between verbal roots and lakāras/the endings that replace lakāras’. This shows that Roodbergen does believe in reading the Aṣṭādhyāyī linearly+++(→memory-less)+++ but disagrees to some extent with the tradition’s order of rule application. And he thinks that this topic is not related to rule conflict and its resolution:

‘this ordering principle has nothing to do with a feeding relation between rules in which the application of one rule is made dependent on the effect of the application of another rule. It has nothing to do either, with the question of conflict of rules. To solve a conflict, other principles apply: paratva, siddha/asiddha14 and utsarga/apavāda.’

16

Cardona etc..

Scholars working on rule conflict have peripherally addressed the topic of linearity. Cardona says that ‘the derivational prehistory of a form is pertinent to the operations which apply to it.’15 Joshi and Kiparsky think that it is important to look ahead into a derivation. They propose the extended siddha principle which they claim governs Pāṇinian derivations and which

‘scans entire candidate derivations…’16 thanks to its ‘global (trans-derivational) “lookahead” condition on derivations’17 ‘…and chooses the one in which siddha-relations (bleeding and feeding)18 are maximized’[^42].

So, both Cardona and Joshi & Kiparsky, do not support an exclusively linear reading of the Aṣṭādhyāyī.

Houben

According to Houben19,

‘a comparison between Pāṇini’s grammar and “a machine” may be useful in demonstrating some of the features and procedures it incorporates, but the comparison has now and then been carried too far.’

He continues:

‘in fact, in the practice of Pāṇinīyas through the ages up to the present, no-one can ever have produced a correct form through Pāṇini’s system that was not already his starting point, or among his starting options … the system is therefore not well characterized as “synthetic”, even if synthetic procedures are central and most visible; rather the system is to be called “reconstitutive” - which implies the presence of a user, a preliminary statement, and the application of both analytic and synthetic procedures to the words in it … aiming at the best possible, saṁskr̥ta form of his preliminary statement.’20

17

He attributes the reception of Pāṇinian grammar as a machine to Bhaṭṭojī Dīkṣita’s Siddhāntakaumudī and Nāgeśa’s Paribhāṣenduśekhara:

‘in order to provide the desired solid authoritative basis to Sanskrit grammar it was moreover necessary to posit it as a closed system of rules and metarules – something it had never been in a true sense of this term for around two millennia, although Kātyāyana’s and Patañjali’s investigations on selected sūtras had prepared the ground for such an approach. The culmination in this trend came only a few generations later with Nāgeśa Bhaṭṭa’s Paribhāṣenduśekhara…’21

Summary

Let us summarize what we have surveyed so far.

Houben is not in favour of perceiving the Aṣṭādhyāyī as a derivational machine, thereby also implicitly dismissing both the concept of linearity+++(→memory-less)+++ and consistent conflict-resolution procedures.

Roodbergen believes that the Aṣṭādhyāyī is a derivational machine, proposes his own version of a linear reading of the Aṣṭādhyāyī. Roodbergen also argues that the order of rule application and resolution of rule conflict are not related or associated with each other.

Bronkhorst claims that the existence of paribhāṣā 38 of the Paribhāṣenduśekhara, which creates a hierarchy of conflict resolution tools (in addition to Patañjali’s statements), indicates that the tradition prefers a linear+++(→memory-less)+++ reading of the Aṣṭādhyāyī. In doing so, Bronkhorst establishes a correlation between consistent rule conflict resolution procedures and a linear+++(→memory-less)+++ reading of the Aṣṭādhyāyī. Bronkhorst rejects the linear approach.

On the other hand, Joshi & Kiparsky and Cardona seem to think that their rejection of a strictly linear+++(→memory-less)+++ reading of the Aṣṭādhyāyī does not substantially undermine the mechanistic prowess of the Pāṇinian system, and devote much of their scholarly attention to solving rule conflict.

Structure

While the functioning of the Aṣṭādhyāyī remains the primary focus of this thesis, we shall also look at its interactions with the structure of the Aṣṭādhyāyī. Let me first outline how the Aṣṭādhyāyī is structured. The rules of the Aṣṭādhyāyī are organized on the basis of their purpose: rules teaching certain saṁjñās are grouped together, rules about a certain substitute are placed together and so on and so forth. In most such groups, the apavāda sūtras ‘exception rules’ are listed immediately after the utsarga sūtras ‘general rules’. These groups of rules are themselves placed in one of the eight books depending on their role: saṁjñā sūtras ‘definition rules’ and paribhāṣā sutras ‘metarules’ are generally placed in the first two adhyāyas, rules teaching affixation in the following three, and rules teaching morpho-phonological changes in the last three.

[[18]]

The structure and organization of the Aṣṭādhyāyī, that is, the general arrangement and serial order of rules in the Aṣṭādhyāyī, have an influence on its functioning in different ways. In the opinion of the tradition, 1.4.2 vipratiṣedhe paraṁ kāryam teaches that in the case of conflict between two equally powerful rules, the rule that appears later in the Aṣṭādhyāyī’s serial order wins, which implies that the serial order of rules in the Aṣṭādhyāyī has a direct impact on rule conflict resolution.

asiddha sections

Pāṇini has ingeniously composed three asiddha sections, headed respectively by 6.1.86 ṣatvatukor asiddhaḥ22, 6.4.22 asiddhavad atrābhāt and 8.2.1 pūrvatrāsiddham. 6.4.22 teaches us that two rules treat each other as asiddhavat ‘as if suspended’ when both lie within 6.4.22- 12947, which helps avoid certain undesirable instances of rule conflict. 8.2.1 teaches us that from there onwards, a preceding rule treats any following rule as asiddha ‘suspended’, which helps facilitate or avoid the application of certain rules. Here too, the position of one rule with respect to other rules has a significant impact on Pāṇinian derivations or the functioning of the Aṣṭādhyāyī.

Function and structure

Interestingly, the functioning of the Aṣṭādhyāyī may have had an impact on its structure too. Roodbergen argues that ‘the word building process proceeds in what is visually a left-to-right direction’.23 According to Roodbergen, this direction of word-building which underlies the functioning of the Aṣṭādhyāyī, impacts its structure, that is, the positioning of rules in different books and chapters of the Aṣṭādhyāyī: ‘rules dealing with left-side elements are introduced earlier [in earlier sections of the Aṣṭādhyāyī]24 than rules dealing with right-side elements’[^50].

Conclusion

We have seen what the existing literature on the subject says about the functioning of the Aṣṭādhyāyī and its connection with its structure. In this thesis, I share my research on rule interaction, and then go on to show how these findings shed light on the functioning of the Aṣṭādhyāyī. I conclude that Pāṇini did intend for the Aṣṭādhyāyī to be interpreted linearly and as a closed grammatical machine. Before I share my understanding of rule interaction, let us first look at the tradition’s views on this subject.

[[19]]

  1. Mbh I.12.15. Note that Mbh I.12.15 stands for Volume I of Mahābhāṣya edited by Kielhorn, page number 12, line number 15. ↩︎

  2. Mbh I.1.14. ↩︎

  3. Bronkhorst 2004: 37. ↩︎

  4. 1.4.2 vipratiṣedhe paraṁ kāryam ‘The rule that comes later in the serial order of the Aṣṭādhyāyī wins the rule conflict between two equally powerful rules.’ ↩︎

  5. Let us say that there is a conflict between rules A and B. A is called nitya with respect to B if A is applicable (both before and) after the application of B (cf. Pbh 117 kr̥tākr̥ta-prasaṅgī yo vidhiḥ sa nityaḥ, Vyāḍiparibhāṣāpāṭha). B is called anitya with respect to A if B is applicable before, but not after the application of A. The nitya rule A is stronger than, and defeats the anitya rule B. ↩︎

  6. Paribhāṣenduśekhara, describes antaraṅga as follows: antar-madhye bahiraṅga-śāstrīya-nimitta-samudāya-madhye ’ntar-bhūtāny aṅgāni nimittāni yasya tad antaraṅgam. Kielhorn translates it as follows: ‘antaraṅga is (a rule) the causes (of the application) of which lie within (or before) the sum of the causes of a bahiraṅga rule’. See Abhyankar’s reprint (second edition) of Kielhorn’s work (1960: 221-222). ↩︎

  7. An apavāda ‘exception’ is stronger than, and thus defeats, the utsarga ‘general’ rule in case of conflict. ↩︎

  8. Abhyankar (ed.) 1960: 185. ↩︎

  9. The contents in brackets have been added by me. ↩︎

  10. Bronkhorst 2004: 6. Patañjali says that para may mean iṣṭa ‘desirable’ in his commentary on 1.4.1 (iṣṭavācī paraśabdaḥ. vipratiṣedhe paraṁ yad iṣṭaṁ tad bhavati; Mbh I.306.9-10). According to Bronkhorst, by iṣṭa, Patañjali means ‘the rule that he thinks should be applied’.

    I disagree with Bronkhorst’s interpretation. I think by iṣṭa, Patañjali means ‘the rule that should be applied so as to get the correct final form.’ This means that, in order to determine which rule is iṣṭa, one is required to know the final form.+++(4)+++ And to know the final form, one needs to look ahead into the derivation. So, in my opinion, this is an instance where Patañjali repudiates his linear reading of the Aṣṭādhyāyī. ↩︎

  11. Bronkhorst 2004: 6. ↩︎

  12. Ibid., 16-17. ↩︎

  13. Roodbergen 1991: 313. ↩︎

  14. A is siddha with respect to B if B recognizes the existence of A. Likewise, A is asiddha ‘not siddha’ with respect to B if B does not recognize the existence of A. ↩︎

  15. Cardona 1970: 41. ↩︎

  16. Joshi and Kiparsky 2005: 7. ↩︎

  17. Ibid. ↩︎

  18. The contents in brackets have been added by me. Rule A bleeds rule B if B, which was applicable before the application of A, is no longer applicable after the application of A. A feeds B, if B, which was not applicable before the application of A, becomes applicable after the application of A. 42 Joshi and Kiparsky 2005: 7. ↩︎

  19. Houben 2003: 50. ↩︎

  20. Ibid., 53. ↩︎

  21. Houben 2015: 6. ↩︎

  22. A single replacement of the preceding and the following sounds is suspended (asiddha) with respect to rules teaching replacement with ṣ (ṣatva) and the introduction of augment tUK. 47 According to the Kāśikā, and broadly, the tradition, the scope of 6.4.22 continues up to the end of 6.4. I will discuss this in detail in chapter 5. ↩︎

  23. Roodbergen 1991: 313. ↩︎

  24. The contents in brackets have been added by me to clarify what the author means. 50 Roodbergen 1991: 313. However, note that the positioning of rules teaching compounds in the Aṣṭādhyāyī poses a challenge to Roodbergen’s proposition. ↩︎