Meanings of Words and Sentences in Mīmāṃsā

Mīmāṃsakas of both the Bhāṭṭa and the Prābhākara subschools refute the idea of a sphoṭa carrying the meaning and being different from what we experience, namely phonemes and words, since this contradicts the principle of parsimony and our common experience. Accordingly, they claim that phonemes really exist and that they together constitute words. They also subscribe to the idea that words convey word-meanings, and thus refute the Bhartṛharian holism, again because this idea is confirmed by common experience and common experience should be trusted unless there is a valid reason not to. In fact, human beings commonly experience that one needs to understand the words composing a sentence in order to understand its meaning.

Moreover, human beings also agree about the fact that words (and not complex texts only) are related to a distinct meaning. The relation between a word as meaningful unit and its meaning is fixed, as it is proved by our common experience of language. This experience cannot be denied in favour of a view focusing on the text as a whole and rejecting without compelling reasons our prima facie experience of words as meaningful units.

Given that one can thus establish that words are meaningful, what exactly do they convey? Mainstream Mīmāṃsā authors, departing from Śabara, claim, against Nyāya ones, that words convey universals (see ŚBh ad PMS 1.1.24: sāmānye padam ”the word conveys the universal”). This is, again, confirmed, by our common experience, in which words figure again and again denoting the same element recurring in several particular items, namely their underlying universal aspect. For instance, the wordcow” denotes in every sentence in which it occurs the universal “cowness”, which is shared by all individual cows. However, this thesis seems at first sight to imply that words would never be able to convey a complex state of affairs on their own accord, and would therefore be almost useless. Human language would be constituted almost of extremely general statements about universals and, which is even more important for Mīmāṃsakas, no specific actions could be enjoined. In fact, each order presuppose a specification (one cannot bring the universal cowness, but only a particular cow). In order to solve this difficulty, Mīmāṃsakas claim that a complex state of affairs (viśiṣṭārtha in the Mīmāṃsā jargon) is conveyed by a sentence (see again, ŚBh ad PMS 1.1.24: viśeṣe vākyam ”the sentence conveys the specific”). This means that the sentence-meaning is more than the sheer sum of word-meanings, insofar as at the level of sentence meaning one moves from one level (that of universals) to the other (that of specific meanings). This solution, however, leads to a further question, namely: How are these two different levels reached? Do the same words lead to the one and then to the next?

The process of sentence‐signification, leading from words to the sentence‐meaning, is distinctly explained by the two main Mīmāṃsā sub-schools, Bhāṭṭa Mīmāṃsā and Prābhākara Mīmāṃsā. Both subschools agree on the basic tenets seen so far, but they differ on the path leading from the words signifying universals to the sentence signifying a particular state of affairs. According to Bhāṭṭa Mīmāṃsā authors, words conclude their function in denoting their own universal meanings (they ground this view in a statement by Śabara, describing words as nivṛttavyāpārāṇi `having concluded their function’, ŚBh ad 1.1.25). Thus, it is the word-meanings, conveyed by words, which convey the sentence-meaning once connected together.

One might (as did Prābhākara Mīmāṃsā authors) object that in this case the sentence meaning is no longer conveyed directly by words, but rather by their meanings and that it is therefore no longer strictly speaking linguistic. Thus, the sentence-meaning would no longer be conveyed through linguistic communication as a distinct instrument of knowledge. This might be a sheer terminological problem, but for Mīmāṃsā authors it has a much deeper relevance. In fact, Mīmāṃsā authors explain that only the Vedas can convey knowledge of dharma. This means that any knowledge of dharma obtained through another source is invariably unreliable. Therefore, if the sentence-meaning were not linguistic, then even the sentence-meanings about dharma would no longer be directly conveyed by Vedic sentences, and would therefore end up being unreliable.

Bhāṭṭa authors reply that the sentence meaning is indeed a function of words, although via their meanings. Bhāṭṭas therefore distinguish a direct denotation (abhidhā) of words, through which universals are denoted, and a secondary signification (lakṣaṇā), through which complex sentence meanings are conveyed.

Prābhākara authors object in three ways: 1. They claim that lakṣaṇā is possible only once the direct denotation is impossible (for instance, in the case of “The village on the Ganges”, one comes to understand that the village is on the Ganges’ bank because the primary meaning would be impossible). But what exactly is incongruous in the word meanings once connected? 2. How do word-meanings connect to each other? If they do it because the words bestow into them the capacity to connect to each other, then it is more economical to just postulate that the words themselves convey the sentence-meaning, without the intermediate step of the sentence-meaning. 3. If word-meanings can automatically connect among themselves, then why don’t they do it unless once in a sentence (in this connection it is important to recollect that artha means both a linguistically conveyed meaning and a cognitively acquired one)? A plausible answer to 1. would point to the fact that the connection of various universals leads in fact to an impossibility since, as in the above example, one cannot bring the universal cowness. One might also suggest that lakṣaṇā in the Bhāṭṭa account acquires a technical meaning, different from the one it assumes in accounts of implicature etc. As for 2. and 3., Kumārila Bhaṭṭa answers that word-meanings do in fact connect automatically and this this does actually occur even outside of sentences. The example Kumārila mentions will be discussed by generations of authors and will remain the only one discussed in this connection: A person sees an indistinct white shape, hears a neighing and perceives the sound of hooves. These three unconnected meanings automatically connect into the complex meaning ”A white horse is running”.

By contrast, Prābhākara Mīmāṃsā authors, and especially Prabhākara’s main commentator, Śalikanātha, state that words first get connected and then denote the specified sentence meaning only once connected. This assures that the sentence meaning can be said to be linguistically conveyed, since there is not the intermediary step of word-meanings, a conclusion which is very important for the Mīmāṃsā epistemology, regarding linguistic communication as a distinct instrument of knowledge (see the section above). However, this explanation altogether skips the role of word-meanings. Thus, Prābhākara authors have to explain the fact that the own meanings of single words appear to do have a role to play in the process, since there is an invariable concomitance between knowing the words’ individual meanings and knowing the sentence’s one. This tension between the opposing risks of atomism and holism is dealt with differently by various authors. Prabhākara seems to present the most basic version of the theory, where word-meanings just don’t play a role in the apprehension of the sentence-meaning. Śālikanātha and his Bhāṭṭa opponent Sucarita start discussing the role the memory of the individual word-meanings plays in the process. Words would accordingly cause one to remember their own meanings, then get related to one another and then denote the complex sentence-meaning. The word-meanings would therefore be recollected, but not denoted by words.

Words get connected into a complex sentence meaning through proximity, semantic fitness and syntactic expectancy. These three criteria correspond to the requirement of being uttered one after the other with no intervening time (unlike in the case of the words ”a cow” and ”runs” pronounced on two different days), being semantically fit to connect (unlike the words ”watering” and ”with fire”) and being linkable through syntactic expectancy (as in the case of a verb and its arguments).

Do you think the Mīmāṃsā theories only make sense in their own context? Or do they look convincing even for people like us? Why (not)?

(cross-posted in my personal blog)

About elisa freschi

My long-term program is to make "Indian Philosophy" part of "Philosophy". You can follow me also on my personal blog:, on Academia, on Amazon, etc.

10 thoughts on “Meanings of Words and Sentences in Mīmāṃsā

  1. Elisa,
    Thanks for this informative post, which helps someone like me who has not kept up on some of these topics for some time now. I have a rather simple and perhaps embarrassingly naive question (well, two questions): so, is it the case that neither Bhāṭṭa nor Prābhākara Mīmāṃsā authors recognize the relevance of what today we would call the distinction between semantics and pragmatics (in other words, denying the relevance of the latter)? Second, and relatedly, is it thus true that our Mīmāṃsā authors would likewise not recognize what Paul Grice termed “shared speaker meaning” (distinct from ‘sentence meaning’)? As an example by way of illustration, contract law has traditionally been dependent on the “common meaning (one might insert ‘intention’ here) of the parties,” which provides normative warrant, legally speaking, for the contract. I say “traditionally,” because it seems contract law has evolved to the point where, in the words of Robin Bradley Kar and Margaret Jane Radin, “agreement,” “consent,” and/or “assent,” no longer have their conventional meaning (i.e., denoting the intention of the respective parties coming to an actual agreement). In short, contract, “which was once a legal regime rooted in actual agreement, has become pseudo-contract—or a system of private obligations with expanding contents that are created unilaterally by one private party [usually the one with far superior bargaining power, as recognized by courts in ‘adhesion contracts,’ which are legal, but said to deserve closer scrutiny should trouble arise between the parties and may raise issues like ‘unconscionability’] without any actual agreement over many ‘terms.’ The fake ‘terms’ in a regime of pseudo-contract invite burgeoning forms of deception that are difficult for courts to discern because they are hidden under the mantel of ‘contract.'” Thus today, and for example, boilerplate language as “sentence meaning” has, generally, come to dominate contractual interpretation and enforcement, while Gricean “speaker meaning” has fallen by the wayside. As our two authors note, “contract meaning has always depended on shared speaker meaning, not sentence meaning. Shared speaker meaning depends, in turn, on not just on what the sentences mean of any boilerplate text or other language delivered during contract formation but also on how parties use language cooperatively in interpersonal conversation during contract formation.”

    The brilliant (by my lights) philosopher Elisabeth Camp has likewise illustrated the importance of such shared speaker meaning when she writes that “speakers’ communicative purposes and intentions often encompass more than just the exchange of information, and that illocutionary force and expressive attitudes interact in systematic ways with the presentation of truth-conditional content.” But Camp has shown, in a brilliant discussion and analysis of “sarcasm,” that even focus on “what is said” as well as “what is meant” is not sufficient. Hence her conclusion: “sarcasm, like many other uses of language, intertwines the communication of information with the expression of, and exhortation to, evaluative attitudes. The exclusion of norms and emotions from the realm of meaning becomes increasingly unpalatable as the range of cases widens to include not just figurative speech like sarcasm and metaphor, but also sincere, direct, and literal speech, as in the case of slurs and moral terms.” Furthermore, while it may be increasingly granted that semantics cannot be sharply distinguished or “encapsulated” from pragmatics, this fact is typically understood by “even the most permissive contextualists” to simply mean that “‘what is said’ … [is] intuitively tied to the modulation or enrichment of lexical meaning, … [and] that only the enriched or modulated meaning has any psychological reality.” Camp’s point is greater, as it were, for “sarcasm does not intuitively belong within ‘what is said’ understood in this way: on the contrary, it’s a commonplace, and not a rarefied theoretical postulate, that a sarcastic speaker means the opposite of what they say.” Therefore, “lexical sarcasm demonstrates even more clearly that non-encoded processes can help to determine the content of the speaker’s primary assertion or other illocutionary act without intuitively belonging to ‘what is said’. Thus, the relationship between conventionally-encoded, compositional meaning and the speaker’s primary speech act—let alone her overall communicated content— is considerably more complex than contextualists typically allow.”

    Again, thanks for provoking this reader to at least think about this stuff!

    • Thank you for this extremely interesting comment, Patrick. I am not sure why you call it “a rather simple and perhaps embarrassingly naive question”, in fact!
      Now, in order to sketch an answer let me preliminarly say that one never finds 1:1 correspondences. If one asks questions like “is there Aristoteles’ syllogism in India?”, the answer should therefore only be “no”. However, one can look for interesting insights, new questions and parallel developments on a given topic in one or the other philosophical tradition.

      As for your question, the distinction between “what is said” and “what is meant” is dealt with in different ways in Mīmāṃsā and Alaṅkāraśāstra, as you probably know already. I wrote an article about illocutionary and perlocutionary speech acts according to Mīmāṃsā (to be found here) and the ultimative must-read about the topic is Larry McCrea’s first book investigating on the Mīmāṃsā’s influence on Alaṅkāraśāstra.

      To put it short: While Nyāya focuses on the speaker’s perspective, Mīmāṃsā focuses on the hearer’s perspective and is therefore extremely sensitive to what happens to the hearer, independently of whether it conforms with the speaker’s intention. The Prābhākara subschool is particularly sensitive to the topic of how language is prescriptive (i.e., has effects going beyond what is “said”) even while not including prescriptive endings (optative, imperative and the like).

      Last, let me remind the distinction between abhidhā `denotative function of language’ (conveying “what is said”) and lakṣaṇā `indirect function of language’. Some authors add a separate third and fourth level, others include them within the second. All agree that you have not got the whole sentence-meaning if you stop at the first or second level (or at the first part of the second level for authors like Mukula).

      I have to admit that I do not know of applications of Mīmāṃsā principles to the jurusprudence of contracts, but it is an interesting topic and I am sure there have been such intersections (I will keep my eyes open looking for them).

      • Would you also agree, Elisa, that since Mīmāṃsā is mainly focused on Vedic hermeneutics, their emphasis is not going to be on speaker intention or prayoga–in addition to the point about the hearer’s perspective? They do talk about prayoga when dealing with laukika speech, but we shouldn’t expect it to be their primary focus. Alaṃkāra, instead, is a better place for that inquiry, as it deals with human speech.

        And, as you mention Mukula, his characterization of the conditions for indication (lakṣaṇā, by which he means to include all secondary meaning, such as gauṇavṛtti) are strikingly “Gricean” in flavor (and he is not the only one to put things this way). He says indication occurs:

        Due to an inconsistency in primary meaning—that is, its being obstructed by another knowledge source;
        And due to the meaning being indicated having dependence on the primary meaning;
        And due to grasping another meaning because of a motive.

        (mukhyāsya arthasya pramāṇântara-bādhitatvena asaṃbhavāt. lakṣyamāṇasya ca arthasya mukhyârthaṃ prati āsannatvamāt. sāntarârtha-grahaṇasya ca sa-prayojanatvāt.)

        His text analyzes the different conditions under which the primary meaning encounters a cognitive obstacle, the different relationships between the primary and secondary meaning, and the different motives. I’ve written on this elsewhere, so I won’t bore everyone with rehearsing the details.

        On another note, this year in reading the Arthasaṃgraha alongside some Western speech act theory (JL Austin’s lecture series along with more recent contributions) one thing that struck me anew is that Mīmāṃsā pushes us to consider to what degree we need speaker intention to account for pragmatic effects. This question is hardly settled in Western literature–even if Camp’s work is excellent!–and Mīmāṃsā analysis of Vedic speech may act as a limit case. How far can we go, explanatorily, without speaker intention? (By the way, Elisa, I had my students read your paper along with Taber’s 1989 “Are Mantras Speech Acts?” as they grappled with this question.)

        • Thank you for this interesting comment, Malcolm. I basically agree with you on everything. Only on the translation of Mukula, I would have a different emphasis:

          Due to an inconsistency in primary meaning—that is, its being obstructed by another knowledge source;
          And due to the meaning being indicated having dependence on the primary meaning;
          And due to grasping another meaning because of a motive.

          [There is secondary signification] 1) because the primary meaning is impossible, since it is blocked by another instrument of knowledge, 2) because the secondary meaning is proximate to the primary one (e.g., insofar as one is metonymically related to the other), 3) and because the grasping of the other meaning has a purpose (e.g., in the case of the hamlet on the Ganges, conveying its purity).

          (mukhyāsya arthasya pramāṇântara-bādhitatvena asaṃbhavāt. lakṣyamāṇasya ca arthasya mukhyârthaṃ prati āsannatvāt. antarârthagrahaṇasya ca saprayojanatvāt.)

          (I did not check the edition and just emended the text as I deemed it right, please correct it if I got it wrong).

          p.s. Which article of mine? Thanks, it is a honour to be read together with John!

          • Elisa, yes, in fact I was trying to edit my comment last night, but I guess that’s not possible? (I wonder if we can add a preview function to the comments, too?) The last line should be “And due to grasping another meaning because of *its having* a motive” (taking the “sa” in “sa-prayojanatvāt” into account).

            As for the second requirement, Dvivedi’s edition has “āsannatvamāt” (which is what my typed version duplicates) but Venugopalan corrects to “āsannatvāt” as does Avasthi, though neither makes a mention of it. Probably it is a typo introduced in Dvivedi’s edition? In any case, I think “proximate” is too strict for what Mukula is arguing in this particular context (if you change your parenthetical i.e. to an e.g. I would be more in agreement). He wants to show that indication has some close relationship–but he goes on to identify quite a few in his discussion a quote from Bhartṛmitra: there is close connection (saṃbandha), association (samavāya–I think the context makes clear this isn’t inherence in the Naiyāyika use of the term), opposition (vaiparītya), and relationship to an action (kriyāyoga). For that reason, I was taking “āsanna” as trading on its primary sense of “seated,” thinking it shows some broad sense of dependence. But just “close” might work–as long as we are clear that this closeness can take many forms. For Mukula, at least, lakṣaṇā is not only metonymy, but includes all secondary meaning.

            Re the article, it’s the one you mentioned in the comment. My students didn’t do anything with Alaṅkāra, really, other than read that it was indebted to Mīmāṃsā, but I had one student get really excited, saying, “Couldn’t we apply Mīmāṃsā hermeneutics to literature?” It’s fun to see rediscoveries like that.

          • Malcolm, yes, sure, it should have been “e.g.” since the beginning and I just corrected it, thanks for pointing it out!

            Re. editing comments, this is possible if you log in as an author.

            And, congratulations for your smart students!

  2. Thanks so much Elisa. It seems I need to read a few things before confirming or disconfirming some tentative beliefs and intuitions. (I knew there were no direct correspondences but was more wondering along the lines of similarity or analogy.) The contractual example was simply meant to illustrate some Gricean ideas (I could have chosen other examples but I recently read their article so it came quickest to mind!).

Leave a Reply to elisa freschi Cancel reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>