To paraphrase a contemporary philosopher: ‘examples stick around in philosophy in a way they don’t in other fields such as linguistics.’ There is some truth to this. I cannot imagine a philosopher today who doesn’t know Frege’s famous problem: ‘the morning star is the evening star’ or Russell’s: ‘The King of France is bald.’
In this blog post, I want to talk about a puzzle (or, more accurately: a set of problem-data) that played a crucial role in a debate on philosophy of language in premodern South Asia for nearly a millennium. This specific set of data was first attested in Maṇḍanamiśra’s Bhāvanāviveka (composed c. 680-690 CE?) and remained a key puzzle for Sanskrit intellectuals thinking about the semantics of verbs and theories of action for a millennium.
If we think about events expressed by verbs, such as: ‘one makes a pot’ (ghaṭaṃ karoti) or: ‘Theodore cooks’ (devadattaḥ pacati), we can identify two elements as candidates for the verbal suffix’s referent(s):
i) physical movement (parispanda)
ii) mental exertion (yatna), i.e., making a conscious effort
On the one hand, the very fact of these two potential candidates presents a challenge for a unified theory of the verbal suffix’s semantics. However, what about events where only one of these two candidates is present? How do we develop a unified theory of the semantics of verbs if there is no one consistent element across all verbs?
We first find this puzzle in Maṇḍana’s Bhāvanāviveka via the set of problem-data:
1. Verbs which express conscious effort, but not physical movement:
i) One sits [in place] (āste)
ii) One sacrifices (yajate) *For Mīmāṃsakas, the verbal root yaj- directly signifies the mental commitment/decision on the part of the sacrifice’s patron (yajamāna) to perform a sacrifice, not the performance thereof.
iii) One cognizes (jānāti)
2. Verbs which express physical movement, but not conscious effort:
i) The riverbank falls (kūlam patati)
ii) The chariot goes (ratho gacchati)
iii) The fire burns (agnir jvalati)
This puzzle/data set is presented early in the Bhāvanāviveka (vss. 4-5), but clearly informs the position of a real interlocutor, against whose position Maṇḍana articulates his own conclusion at the tail-end of his essay (vs. 47 ff.), as the interlocutor’s position addresses precisely this puzzle. I translate the interlocutor’s position:
On that, some say:
[Śabara] understood mental exertion (prayatna)
when he said: ‘one should strive’ (yateta) in his commentary [on 2.1.1].
Due to the frequent usage of verbs to [describe] insentient things, there is no
understanding of the linguistic usage as figurative. 47.
[Śabara] has shown that the verb’s referent is mental effort conducive to the occurrence of the result [when he said]: ‘one should strive (yateta) so that something comes into being.’ The usage, however, of a verb [to describe] an insentient thing is based on figurative usage. [But] we don’t understand it as figurative due to its frequent usage [in this way].
The interlocutor’s attempt at a unified theory of the semantics of the verb’s suffix is that it only directly signifies mental exertion (prayatna), and the interlocutor accounts for cases where we only understand physical movement (parispanda) is to treat these as indirect signification/figurative usage (such as in: ‘the chariot goes’). However, says the interlocutor, we don’t overtly cognize these cases as figurative, due to our high frequency usage of verbs in this way (think like a dead metaphor –> even though it’s not the literal meaning, we treat it as such).
To this account, Maṇḍana’s answer is:
However, this answer hinges upon the establishment of [mental exertion and not physical movement as the directly signified sense], and there is no proof of that when our cognition [of both] is the same and when we lack another reason, because it isn’t possible to preclude that we could take it the other way.
Therefore, [we] believe the meaning of the verb: ‘one makes’ (karoti) has as its nature the commonality of a cessation of inertia, and the arising of an awareness of that occurs in both cases. 48.
For the meaning of ‘one makes’ (karoti) is a cessation of inertia, due to the absence of an awareness of action in something that is inert. And an awareness of falling from inertia is twofold: due to conscious effort in the self (ātman) and due to physical movement in something other [than the self].
Maṇḍana’s solution to articulate a unified theory of the verbal suffix’s semantics is to find a commonality between mental exertion and physical movement, and present that as the suffix’s directly signified sense. This commonality is the cessation of inertia. Whether the chariot begins to physically move or I begin to mentally think about something, this constitutes a cessation of inertia. The only difference is the form that cessation of inertia takes, physical or mental. Thus, on Maṇḍana’s account, both mental exertion and physical movement can be treated as directly signified (these are the two ways inertia ceases), even though we’re only admitting the verbal suffix to have one directly signified sense (the cessation of inertia).
Although Maṇḍana’s specific position here doesn’t gain a huge following, Maṇḍana sets the stage here for many subsequent intellectuals to carve out their own positions against his (more on this in relation to the Naiyāyikas in my dissertation, I suppose).
This puzzle remains relevant for a millennium. Sanskrit intellectuals nearly one thousand years later such as Raghunātha Śiromaṇi and the Grammarian revivalist Koṇḍabhaṭṭa thought about this very puzzle in articulating their own new theories — although perhaps through second-hand sources.
Turning away from this specific case, the use of philosophical puzzles in premodern South Asia raises further intellectual-historical questions, such as: how are philosophical puzzles approached methodologically by these Sanskrit intellectuals? What is the role and status of these puzzles in relation to the positions that these Sanskrit intellectuals develop?
A good answer is probably nuanced and context-dependent. In the case-study explored above, Maṇḍana develops his own new position as a unitary theory of the semantics of verbs in relation to the problematic data set. However, Maṇḍana’s solution is still carved out within the fold of certain Bhāṭṭa commitments (e.g., the archetypal Vedic injunction: svargakāmo yajeta has the semantic structure: yāgena svargam bhāvayet ‘one should bring heaven into being via sacrifice’). Likewise, Maṇḍana’s Bhāvanāviveka is the first extant hostile engagement with Prabhākara’s theory of the command (niyoga): Maṇḍana is not simply developing his theory of the semantics of verbs in a vacuum, but within a very specific context while navigating many other goals and agendas that I didn’t touch upon in this post.
However, some centuries after Maṇḍana, in the context of the Prābhākara and Naiyāyika debate on philosophy of language, the game is not one of developing a new position in relation to puzzles, but rather of deploying puzzles and problematic data against the opponent to undercut their positions and to show the superiority of one’s tradition’s established doctrine by squaring it with the data.
This shift in how puzzles are used appears to be tied to the nature of the debate: we’re now in a context where Prābhākaras and Naiyāyikas are refining established doctrines, not developing radically new theories.
Anyways, these latter thoughts probably need some further refinement … don’t quote me on this …
If anyone has any thoughts about this or other philosophical puzzles they’ve encountered and thought about, do let me know 🙂
Hi Patrick,
Very interesting post, thanks!
I was just thinking, and I know very little about Śabara and Maṇḍanamiśra, what’s a literal meaning as opposed to figurative meaning? Frequency of use?
Hi Szymon,
Thanks for reading it and for the response.
Hm, suppose I say: ‘There’s a donkey on the farm’ vs. ‘Bob is a donkey.’
In the former, I’m using the term donkey to refer to the animal, the equus africus asinus. This is the term donkey’s literal referent.
In the latter, however, I’m using the term donkey figuratively: I’m not purporting that Bob has four legs, brays, and so on, rather I’m attributing qualities of the donkey to Bob: stubbornness, stupidity, etc.
So in the context at hand, the interlocutor wants to claim that only one referent of the verbal suffix — mental exertion (prayatna) — is directly signified, but when we use verbs and there isn’t a conscious agent (e.g. ‘the chariot goes’ [ratho gacchati]), this is a figurative usage.
So as per the interlocutor, when I say: ‘Theodore thinks’ (devadattaḥ jānāti), I’m using the verb literally, because Bob is making an effort to have an awareness event. However, when I say: ‘the chariot goes’ (ratho gacchati), the chariot is not making a conscious effort, rather it’s engaged in physical movement. Thus, when I use a verb to refer to physical movement, I’m shifting from a literal meaning (mental effort) to a non-literal but connected meaning (physical movement).
Thanks Patrick, that’s very interesting.
I’ve spent some time reading Dharmakīrti against Mīmāṃsakas recently, so I have some questions. It’s a bit out of the imitated context of verb suffixes’ semantics but about meaning in general, if that’s okay 🙂
Reference, as I understand it, is a relationship between a speech act and an object. You say that ‘There’s a donkey on the farm’ has the literal referent: the animal. I take it that you mean a particular donkey that actually lives on some farm. Is that right? In contrast, ‘Bob is a donkey’ refers to properties that a particular Bob shares with some donkey. Is the difference between literal and figurative speech the object of reference? For literal speech, a particular and, for figurative speech, properties a particular share with some other particular?
It’s a side question but I’m curious what you think. How about a non-referential speech? Can I say something literally about nymphs? Or all talk about nymphs is metaphorical because I only can refer to properties other things have?
Hey Syzmon,
You’re on the right track about reference, in that there is a difference of object referred to, but as to what you ask: ‘does literal speech refer to particulars, but figurative speech to a shared property?’ In short: close, but not exactly. But to be more specific, what words refer to also depends who you ask (Mīmāṃsaka, Buddhist, Naiyāyika, etc.)
So the example I gave (‘Bob is a donkey’) was just one type of indirect signification, based on a shared quality. There are, however, other types. I could say: ‘The suits are angry’. Here, it’s more like metonymy. Put very succinctly, in all cases of indirect signification (lakṣaṇā) there is at least 1 (but potentially more) shifts from the term’s literal referent to a connected referent (on the part of the listener in computing the meaning). So if I say: ‘the suits are angry’ the first meaning of the term ‘suit’ that would appear to your mind — the literal referent (i.e., the physical piece of clothing) — won’t fit the context (b.c clothes can’t have emotions), and so you’ll shift to a connected, pragmatically appropriate referent (the guy wearing the suit).
So it doesn’t always have to be a quality.
Also, different traditions — and even different thinkers within a given textual tradition — have different views on what the objects are that words refer to.
For example, the standard Mīmāṃsā doctrine is that terms only directly signify eternal universals such as class-properties (cow-ness), qualities, and so on. There’s a complex defense of Vedic authority as eternal in the background (let’s not go there, but let’s have in sight their motivation for going this route). What Kumārila, for example, would (reluctantly) say is that words refer to universals, but sentences refer to particulars, and so all ‘sentence-meaning’ i.e., meaning at the level of the utterance is indirect.
In contrast, a Buddhist would say terms refer to exclusions (apohas) or excluded referents (apoḍha). The term cow would refer to whatever is not a non-cow. And so on.
As to the question of non-referential speech, again I think it depends who you ask. For Dharmakīrtians, all concepts (e.g. ‘cow’ or ‘tree’) are mental constructs that don’t correspond to external realities, and so I don’t think there would be much of a problem for nymphs or the sky-flower (khapuṣpa) –> it might just be a degree of difference (as regular language never refers to anything real, as only unique momentary particulars are real). I’m sure Dharmakīrti or some other thinker in his tradition must deal with this, but I’m not the best guy to ask for a reference on that …
Hm, for a Mīmāṃsaka or Naiyāyika, I suspect non-referential speech would refer to an upādhi, a constructed property, but I’d be curious to hear someone else confirm or deny this. So for example, pot-ness (ghaṭatva) is not a naturally occurring class property (unlike cow-ness or something) and moreover pot-ness doesn’t fit properly under one bigger class property (jāti), as you can make pots out of both clay (i.e. earth) and gold (i.e. fire –> for Naiyāyikas gold comes from fire, but let’s put that aside). The problem is that you’d have pot-ness cutting across class properties, were you to admit pot-ness as a naturally-occurring universal (jāti). The solution to this is a constructed property, an upādhi. And so the term ‘pot’ would refer to an upādhi. My suspicion is that non-referential speech could be treated as referring to upādhis too, but I can’t recall seeing this spelt out …
If you want further readings, my recommendations would be:
i) For a general conceptual treatment, I’d say poke around Kunjunni Raja’s ‘Indian Theories of Meaning’. It must be available on archive.org.
ii) For indirect speech specifically, I’d recommend Malcolm Keating’s ‘Language, Use and Meaning in Indian Philosophy’ if you want to access the materials in English, or Mukulabhaṭṭa’s Abhidhāvṛttamātṛkā if you want to access the Sanskrit.
iii) For historical treatment of a wider range of topics, I’d recommend Alessandro Graheli’s Bloomsbury Research Handbook on Philosophy of Language
PS: Thanks for the recommendation of my book! I’d add Tillemans, Tom. 1999. “What is the svadharmin in Buddhist logic.” In Scripture, Language, and Logic: Essays on Dharmakīrti and His Tibetan Successors, by Tom Tillemans, 172-199. Boston: Wisdom Publications, for discussion of referential failure in Buddhism.
“for a Mīmāṃsaka or Naiyāyika, I suspect non-referential speech would refer to an upādhi, a constructed property, but I’d be curious to hear someone else confirm or deny this.”
For Kumārila’s view on this, one key place to look is the TV ad MS 1.4.23ff, in which he talks about the meaning of kha-puṣpa (sky-flower), pradhāna (the Sāṃkhya primordial nature), and śaśa-viṣāṇa (horned hare). He gives a whole series of alternatives, and we might think that the last in the sequence is his definitive view, though that is just a heuristic–it need not be that he has a settled view.
If it is his view, then it sounds to me like a kind of paraphrase view, on which we can identify a series of properties which would enable us to speak about non-existent entities. So a hare’s horn is what grows on the flat part of the head and grows higher. Pradhāna is jagatkāraṇādi. (He doesn’t explicitly go back to kha-puṣpa.)
This is important for avoiding the fallacy of āśrayāsiddhi, which is where the Sāṃkhya example comes in, I take it, since this is one of the examples that Dignāga takes up in the Nyāyamukha–though his strategy is to appeal to what is conceptually constructed (kalpita). This text is discussed in Kunio Harikai 2017 (“Mīmāṃsā theory of gauṇa or metaphor from Śabarasvāmin to Kumārilabhaṭṭa” in Dieux, génies, anges et démons dans les cultures orientales & florilegium indiae orientalis Jean-Marie Verpoorten in honorem) though he does not discuss the non-referential cases, sadly. I have a draft that I need to polish and get out on this, at some point!
Thanks for the reference to Tillemans, I found the whole book on his academia.edu page (just FYI in case anyone else wants to access it …)
And thanks for the reference in Kumārila –> I glanced over p. 359-60 just now. Good to know!
Oh and for anyone potentially reading, definitely reject my hypothesis that non-referential speech would refer to a constructed property for a Mīmāṃsaka.
As Malcolm points out, for Kumārila at least, the way we can make a proposition such as: ‘the rabbit’s horn doesn’t exist’ is tied to the existence of the constituent parts (horn + rabbit). These things actually do exist, and are referenced in the proposition above; all we’re doing is asserting that these two real properties don’t have a connection (i.e. that they aren’t located in the same entity), and so the language of this proposition is all within the scope of direct signification. In other words, there’s no indirect or figurative usage at play in ‘non-referential’ speech, because the morphemes are in fact, referential.
*At least this is my cursory reading of the issue …
Thanks for your post Patrick!
As someone only recently aware of language theories/thinking in India, I can only pose very generic, basic and broad questions. I seek forbearance on that count.
Reading about the debates on points of grammar, on points of ākhyāta/आख्यात/verb-endings and effort/yatna/volition/vṛtti etc, I wonder if the teaching of grammar was different in different sampradāya grammar schools? Was there not one standard way or curriculum of teaching grammar, esp. to young learners?
Else, I wonder, how did so many interpretations of basic grammar rules come about, esp. if someone like Pāṇini had laid down the (non-sectarian) rules of language? Did the early grammarians not fix the meanings/referents of something as basic as verbal-suffixes? How did everyone develop their own way of explaining verbal endings, say, given that the concepts of kriyā, kāraka etc were dealt with in so much detail? Why were conditions of a puzzle even created? Or is that the nature of language in general, that certain constructions are open to differing views?
I am genuinely curious and would appreciate any literature/pointers in this regard.
Dear Ananda,
Thanks for your response.
Great question. In short, I would say that even within the field of Vyākaraṇa, things are not so fixed by Pāṇini as one might suspect from the outside. Kātyāyana, for example, openly supplements, modifies and criticizes Pāṇini’s rules. Commentators on Pāṇini also find many ways to monkey around with the system (but in ways that are more exegetically concealed). This is to say nothing of other non-Pāṇinian grammatical traditions (e.g., Cāndravyākaraṇa).
Mīmāṃsakas find many ways to subvert the rules of grammar. A great example is Kumārila on the kartradhikaraṇa (‘The Topic of the Agent’ in Book 3 Chapter 4 of the Mīmāṃsā system): Kumārila subverts the position the Pāṇinian grammarians, including Patañjali, that the finite verbal suffix directly signifies the agent (e.g., –ti in gacchati or –eta in yajeta). For Kumārila, the verbal suffix directly signifies only the agent’s number, and the agent is entailed by the bhāvanā (the ‘bringing into being’ i.e., action intended to produce an outcome, itself also expressed by the verbal suffix).
To give something of a general answer: while there are accepted parameters (e.g., Mīmāṃsakas never chuck out kāraka theory as such), there is ample scope for innovation within the fold of language analysis: different motivations come into play and underpin the different positions these intellectuals take …