S'abda or word is regarded as a separate means of proof by most of the recognized Indian systems of thought excepting the Jaina, Buddhist, Carvaka and Vais`e@sika. A discussion on this topic however has but little philosophical value and I have therefore omitted to give any attention to it in connection with the Nyaya, and the Sa@mkhya-Yoga systems. The validity and authority of the Vedas were acknowledged by all Hindu writers and they had wordy battles over it with the Buddhists who denied it. Some sought to establish this authority on the supposition that they were the word of God, while others, particularly the Mima@msists strove to prove that they were not written by anyone, and had no beginning in time nor end and were eternal. Their authority was not derived from the authority of any trustworthy person or God. Their words are valid in themselves. Evidently a discussion on these matters has but little value with us, though it was a very favourite theme of debate in the old days of India. It was in fact the most important subject for Mima@msa, for the _Mima@msa sutras_ were written for the purpose of laying down canons for a right interpretation of the Vedas. The slight extent to which it has dealt with its own epistemological doctrines has been due solely to their laying the foundation of its structure of interpretative maxims, and not to writing philosophy for its own sake. It does not dwell so much upon salvation as other systems do, but seeks to serve as a rational compendium of maxims with the help of which the Vedas may be rightly understood and the sacrifices rightly performed. But a brief examination of the doctrine of word (_s'abda_) as a means of proof cannot be dispensed with in connection with Mima@msa as it is its very soul.
S'abda (word) as a prama@na means the knowledge that we get about things (not within the purview of our perception) from relevant sentences by understanding the meaning of the words of which they are made up. These sentences may be of two kinds, viz. those uttered by men and those which belong to the Vedas. The first becomes a valid means of knowledge when it is not uttered by untrustworthy persons and the second is valid in itself. The meanings of words are of course known to us before, and cannot therefore be counted as a means of proof; but the meanings of sentences involving a knowledge of the relations of words cannot be known by any other acknowledged means of proof, and it is for this that we have to accept s`abda as a separate means of proof. Even if it is admitted that the validity of any sentence may be inferred on the ground of its being uttered by a trustworthy person, yet that would not explain how we understand the meanings of sentences, for when even the name or person of a writer or speaker is not known, we have no difficulty in understanding the meaning of any sentence.
Prabhakara thinks that all sounds are in the form of letters, or are understandable as combinations of letters. The constituent letters of a word however cannot yield any meaning, and are thus to be regarded as elements of auditory perception which serve as a means for understanding the meaning of a word. The reason of our apprehension of the meaning of any word is to be found in a separate potency existing in the letters by which the denotation of the word may be comprehended. The perception of each letter-sound vanishes the moment it is uttered, but leaves behind an impression which combines with the impressions of the successively dying perceptions of letters, and this brings about the whole word which contains the potency of bringing about the comprehension of a certain meaning. If even on hearing a word the meaning cannot be comprehended, it has to be admitted that the hearer lacks certain auxiliaries necessary for the purpose. As the potency of the word originates from the separate potencies of the letters, it has to be admitted that the latter is the direct cause of verbal cognition. Both Prabhakara and Kumarila agree on this point.
Another peculiar doctrine expounded here is that all words have natural denotative powers by which they themselves out of their own nature refer to certain objects irrespective of their comprehension or non-comprehension by the hearer. The hearer will not understand the meaning unless it is known to him that the word in question is expressive of such and such a meaning, but the word was all along competent to denote that meaning and it is the hearer's knowledge of that fact that helps him to understand the meaning of a word. Mimamsa does not think that the association of a particular meaning with a word is due to conventions among people who introduce and give meanings to the words [Footnote ref 549]. Words are thus acknowledged to be denotative of themselves. It is only about proper names that convention is admitted to be the cause of denotation. It is easy to see the bearing of this doctrine on the self-validity of the Vedic commandments, by the performance of which such results would arise as could not have been predicted by any other person. Again all words are believed to be eternally existent; but though they are ever present some manifestive agency is required by which they are manifested to us. This manifestive agency consists of the effort put forth by the man who pronounces the word. Nyaya thinks that this effort of pronouncing is the cause that produces the word while Mimam@sa thinks that it only manifests to the hearer the ever-existing word.
The process by which according to Prabhakara the meanings of words are acquired maybe exemplified thus: a senior commands a junior to bring a cow and to bind a horse, and the child on noticing the action of the junior in obedience to the senior's commands comes to understand the meaning of "cow" and "horse." Thus according to him the meanings of words can only be known from words occurring in injunctive sentences; he deduces from this the conclusion that words must denote things only as related to the other factors of the injunction (_anvitabhidhana vada_), and no word can be comprehended as having any denotation when taken apart from such a sentence. This doctrine holds that each word yields its meaning only as being generally related to other factors or only as a part of an injunctive sentence, thus the word _gam_ accusative case of _go_ (cow) means that it is intended that something is to be done with the cow or the bovine genus, and it appears only as connected with a specific kind of action, viz. bringing in the sentence _gam anaya_--bring the cow. Kumarila however thinks that words independently express separate meanings which are subsequently combined into a sentence expressing one connected idea (_abhihitanvayavada_). Thus in _gam anaya_, according to Kumarila, _gam_ means the bovine class in the accusative character and _anaya_ independently meansbring; these two are then combined into the meaning "bring the cow." But on the former theory the word _gam_ means that it is connected with some kind of action, and the particular sentence only shows what the special kind of action is, as in the above sentence it appears as associated with bringing, but it cannot have any meaning separately by itself. This theory of Kumarila which is also the Nyaya theory is called abhihitanvayavada [Footnote ref 550].
Lastly according to Prabhakara it is only the Veda that can be called s'abda-prama@na, and only those sentences of it which contain injunctions (such as, perform this sacrifice in this way with these things). In all other cases the validity of words is only inferred on the ground of the trustworthy character of the speaker. But Kumarila considers the words of all trustworthy persons as s'abda-prama@na.