What are the principles by which we divide texts into portions, and why would we want to do that in the first place?
What Elisa has referred to as “the prescriptive portion of the Veda” is what Mīmāṃsā authors call brāhmaṇa. For Mīmāṃsā, this was the most important portion of the Vedic text, since this is where our knowledge of dharma in the first instance comes from. The brāhmaṇas are where our dharmic obligations originate.
It is thus the interpretive agenda of Mīmāṃsā, which contends both that knowledge of dharma comes exclusively from the Veda and that the Veda eventuates exclusively in knowledge of dharma, that motivates the priority of the brāhmaṇas over other portions as well as the very category of brāhmaṇa itself. This agenda has, over the millennia, shaped both the way that people think about the Veda and the way the Veda has been transmitted.
What I find most interesting about the division of the Veda into brāhmaṇa and non-brāhmaṇa portions is that Mīmāṃsā grounds the distinction in the modalities of language: the language of the brāhmaṇas is generally injunctive (vidhāyaka), and that of mantras (which will represent the non-brāhmaṇa portion for my purposes) is generally non-injunctive (avidhāyaka). The definition of mantra and brāhmaṇa in the sūtras themselves (2.1.32, 2.1.33) suggests this treatment, since it has the structure of a rule and exceptions (the Veda itself is brāhmaṇa, except for parts that are denotative rather than injunctive, which are called mantra).
This is not an obvious or intuitive way of talking about the division. I say this from a “western” perspective, where the criteria for distinguishing between texts, or between portions of texts, generally fall into the following categories:
- Historical: portion x was composed before or after portion y, or by different people, etc. This is by far the most commonly-invoked class of distinctions in the Western tradition (think of, e.g., the Bible, which though it was “revealed” was nevertheless authored by historical individuals), but it’s completely off-limits to Mīmāṃsā, since it holds the Veda to be an authorless text that was not fashioned in history. But this is how Indologists generally think of the Veda: the saṃhitā text is the earliest (and hence most valuable from the perspective of historical linguistics and Indo-European studies), and the brāhmaṇas belong to a later period.
- Formal: the formal or stylistic characteristics of portions x and y differ. In the case of the Veda, we can say that brāhmaṇas mainly consist of prose, and the saṃhitās of verse. Or, we could say that the saṃhitās use a certain kind of language—say, the one that Pāṇini refers to as chandas—that distinguishes it from the brāhmaṇas. The difference between injunctive and non-injunctive portions partly relates to formal differences, that is, between injunctive and non-injunctive verbal endings, but only partly.
- Transmissional: the Yajurveda, for example, exists in two versions: the White Yajurveda transmits the saṃhitā text and the brāhmaṇa text separately, while the Black Yajurveda contains both interspersed.
From the perspective of Mīmāṃsā, these differences are either specious, or they result from the more fundamental distinction between injunctive and non-injunctive mentioned above.
The problem is: how can we possibly motivate the idea the mantras are non-injunctive and brāhmaṇas are injunctive? Where, in other words, does the injunctive (or non-injunctive) character of a particular passage of text come from?
The answers to this question can be classified into language-internal and language-external accounts. The pūrvapakṣin in the mantrāvidhāyakatvādhikaraṇa (2.1.30–31) relies on a language-internal account: in mantras we sometimes encounter the very same verbal forms by virtue of which the brāhmaṇas are said to be “injunctive.” The uttarapakṣin employs a language-external account: mantras are used in rituals, and therefore they cannot tell us to perform additional rituals. Basically this is a dispute about the role of context (in the wider sense, not in the sense of prakaraṇa). The pūrvapakṣin does not include context at all in his evaluation of the modality of these passages; the uttarapakṣin takes the context for granted and uses it to argue for a certain linguistic modality. In Kumārila’s framing of the problem, the difference is whether the injunctive force of a text arises before or after the “assignment” (viniyoga), the process by which one determines what elements of a ritual performance are primary and secondary, in consequence of which mantras are always determined to be instrumental to, and thus subordinate to, a sacrificial act enjoined by a brāhmaṇa text.
Kumārila takes aim at both positions: if injunctive force is internal to language (svato vidhiśaktiḥ), then it’s impossible to see why it would recede in some contexts and reappear in other contexts; if it is not there in the first place (prathamam eva nāsti), then no amount of contextual assignment could make it appear. His answer is to allow that language has a modality that the pūrvapakṣin does not acknowledge, namely denotation (abhidhāna).
Many questions about denotation, and especially about the early history of this concept and its application in Mīmāṃsā, remain to be worked out. But the general idea is that when the injunctive force of a text is weakened or cancelled out, as the contextualist wants, then what remains is the denotative power. Except Kumārila insists that it is the text itself, and not contextual assignment, that cancels out the injunctive force. Specifically, there are certain linguistic forms that can neutralize the injunctive power (Kumārila gives examples of conditional clauses, the relative pronoun, vocative forms, and first-person verb forms). This neutralization happens prior to assignment; in fact, it is the diminished injunctive capacity of such texts that give us a clue about the role they are to play in the performance of a sacrifice. “Denotation” as a modality applies to things that, in Mīmāṃsā terms, “obtain” (prāpta)—that is, things that are epistemically available to be referred to. Hence Kumārila arrives at a classic case of hermeneutic codependency: we need to recall or remember (smṛti) certain elements at the time of performance, and this is precisely and exclusively the purpose which mantras serve.
In this way, Kumārila argues that mantras and brāhmaṇas are distinct not because of their formal features, or because they are transmitted separately, or least of all because of the historical circumstances of their production, but because there are language-internal features of mantras that lead us to ascertain the different modes of operation between mantras and brāhmaṇas and hence the different purposes that each serves in the integrated complex of dharma.
We could talk about whether this classification is compelling or well-grounded, or how it maps against the classifications mentioned previously. But this section is a good specimen of the interpretive angle of Mīmāṃsā more generally. We are not supposed to take the division of the Veda into prescriptive and non-prescriptive portions for granted; rather, this division is supposed to arise from general principles of interpretation, which are themselves rooted in a philosophy of language. Even if Kumārila often gives the impression of making up the finer points of that philosophy as he goes along, he is very sensitive to the philosophical and hermeneutic problems generated by both a strongly language-internal account and a strongly language-external account, and he tries to find a viable third path.