Let us begin by looking at the fundamental justification given by Joshi and Kiparsky (1979) for their siddha principle and will present my ideas on the same. In ‘The Ordering of Rules in Pāṇini’s Grammar’ (1982), Kiparsky gives a detailed explanation of the siddha principle. I will quote from this paper here.
Kiparsky proposes the siddha principle on the basis of a vārttika on 6.1.86 ṣatvatukor asiddhaḥ ‘a single replacement in place of the preceding and the following sound segments is suspended1 with respect to any potential replacement with ṣ or insertion of augment tUK’. Kiparsky explains: “Kātyāyana says that making a rule asiddha has two functions: (ṣatvatukor)2 asiddhavacanam ādeśalakṣaṇapratiṣedhārtham utsargalakṣaṇabhāvārthaṁ ca (6.1.86, vt. 1).
Utsarga here means sthānin, the element which undergoes substitution in a rule.”3
I translate this vārttika as follows: ‘the statement that ṣ [replacing s] and [the insertion of the augment] tUK are asiddha [has been made] for the purpose of preventing the operations that are due for application to the substitute, and facilitating the operations that are due for application to the substituendum (original item)’.
Kiparsky then says: “to use terms common in linguistics, asiddhatva blocks bleeding and feeding between rules.” Before going further, let us understand what he means by feeding and bleeding: A feeds B if the application of A facilitates the application of B, and P bleeds Q if the application of P obstructs the application of Q.
Kiparsky concludes: “it can be said that asiddha and the other devices are restrictions (niyamas) on a general paribhāṣā that determines how rules interact when no special statement about their ordering is made in the grammar. This paribhāṣā is not stated in the grammar itself but it is presupposed by the correct operation of rules in it and implied by the various restrictions on it that are stated in the grammar. It is to be formulated as ‘sarvatra siddham’ and we refer to it as the siddha principle.”4 He continues: “[W]hat the siddha principle says is that in the general case we have ādeśalakṣaṇabhāva and utsargalakṣaṇapratiṣedha…in short, the siddha relations
238
of bleeding and feeding are given free by the underlying theory of the Aṣṭādhyāyī and if we must not have them in some particular case, then only something must be said in the grammar itself.”5
Further, he says, “As far as feeding is concerned, this really goes without saying. In almost any derivation, the application of one rule creates scope for another rule to apply, that rule applies creating scope for a third rule and so on. That all rules in such a chain of rules are to be applied is taken for granted in the tradition.”6 He adds, “By this point anyone familiar with the topic will already have recognized that the principle of bleeding order is simply equivalent to the nitya-principle in the traditional inventory of the paribhāṣās.”7
Thus, we can say feeding and bleeding together are simply equivalent to nityatva in the Pāṇinian tradition. And the siddha principle, which means the maximization of feeding and bleeding, is tantamount to the maximization, wherever possible, of the use of nityatva for rule conflict resolution, that is, in all cases involving unidirectional blocking.
239
Now, using diagrams, I will explain why I think Joshi and Kiparsky have made a logical error in interpreting the aforementioned vārttika. I will focus on bleeding and not on feeding, because as Kiparsky himself says, what he calls ‘feeding’ is built into the Pāṇinian system, and there is no controversy about it.
What Kātyāyana’s vārttika implies:
A
(asiddha = does not allow
any recognition of the
suspended i.e., asiddha
rules and as a result
bleeding can never take
place
A’
(siddha = allows
recognition of all rules and as a result bleeding can potentially take place. It may or may not take place)
Kiparsky takes the liberty to interpret this as:
A
(asiddha = bleeding
never takes place)
A’
(siddha = bleeding always takes
place)
240Let us use an analogy to understand this, just like Patañjali often does. Imagine that a young boy, who is obedient to his parents, can be given one of two possible instructions by his parents about going near the fire:
Parental instruction
What this instruction actually entails
What Kiparsky assumes it entails
You are not allowed to go near the fire. (These rules are asiddha.)
The child will never burn his hand. (Bleeding will never take place).
The child will never burn his hand. (Bleeding will never take place).
You are allowed to go near the fire.
(These rules are siddha.)
The child can potentially burn his hand. He may or may not burn his hand. (Bleeding can potentially take place. It may or may not take place.)
The child will always burn his hand. (Bleeding will always take place).
I conclude that it is not logically possible to infer the siddha principle from vt. 1 on 6.1.86.
Regardless of that, let me briefly comment on the following question: how useful is the siddha principle in dealing with cases of SSRI? As stated in chapter 6, the siddha principle rejects the antaraṅga tool and essentially resorts to the nitya tool to solve not only those cases which the tradition solves using nityatva, but also those which it solves using antaraṅgatva. Of course, this means that the siddha principle is able to tackle cases of unidirectional blocking but not of mutual blocking – which is one of its drawbacks.8 Another drawback of the siddha principle is that it pays little attention to and offers no solutions for those cases of SSRI which do not involve any blocking at all (‘non-conflict’).
How useful is the siddha principle in dealing with cases of unidirectional blocking? Since the siddha principle is no different from the nitya principle, albeit with a wider scope of application than the traditional one, the answer to this question is the same as the answer to the question about the potency of the nitya principle, which I have given in footnote 62 of chapter 4, and which I reproduce here: “This is exactly why the traditional nitya tool which teaches that the nitya rule defeats the anitya rule, always correctly resolves cases of DOI involving
241
unidirectional blocking: the nitya rule is applicable to the RHS operand and the anitya rule to the LHS operand. By (my interpretation of) 1.4.2, the RHS rule (which is also the nitya rule) defeats the LHS rule (which is the anitya rule).” However, I do not know if the nitya / siddha principle is always correctly able to solve cases of SOI involving unidirectional blocking. A majority of the examples discussed in Kiparsky (1982) involve DOI and not SOI.
A major shortcoming of the nitya, and therefore the siddha principle, is its propensity to look ahead into the derivation: one needs to know what will happen at the next step if, hypothetically speaking, a certain rule is applied at the presentstep. I think this very much qualifies as ‘looking ahead’, even though it involves considering merely the potential – and not the actual – future course of the derivation. Joshi and Kiparsky (2005) take this a step further by proposing the extended siddha principle which ‘scans entire candidate derivations…’9 thanks to its ‘global (trans-derivational) “lookahead” condition on derivations’10 ‘…and chooses the one in which siddha-relations (i.e., bleeding and feeding)11 are maximized’12.13 In simple words, they ask us to choose, from amongst all possible derivational paths, that derivational path in which the nitya tool has been applied the highest number of times.
Why does the derivational path in which siddha relations are maximized lead to the correct answer though? It is easy to explain this with respect to DOI. In case of DOI, Pāṇini teaches us (according to my interpretation of 1.4.2) that we must pick the RHS rule. But as we know (see the footnoted reproduced above), it is the RHS rule which is also the nitya rule in cases of DOI involving unidirectional blocking. So, it is natural that, of all the possible derivational paths, the correct one has the highest number of instances in which the nitya (RHS rule) defeats the anitya (LHS) rule – in cases of DOI involving unidirectional blocking. It is difficult to verify if Joshi and Kiparsky’s extended-siddha principle holds true with respect to SOI.
Now let us ask: how useful is the extended siddha principle in resolving cases of SSRI? If one has to chart out all possible derivational paths to make a decision, how is choosing the derivational path in which siddha-relations are maximized any better than simply choosing the derivational path which gives the correct grammatical form – which we know thanks to our
242
knowledge of Sanskrit? And in the latter case, why perform derivations at all if we have to rely on the correct final form to choose the correct derivational path?
Joshi and Kiparsky have discussed several examples in the aforementioned papers, a number of which I have solved using my method in this thesis. Please see Appendix D for relevant tables of concordance. While it is not within the scope of this thesis to discuss in detail Joshi and Kiparsky’s solutions for individual examples, we ought to study the work produced by them in greater depth in the future to gain new insights into the functioning of Pāṇini’s grammar.
243