Linguistic analysis and epistemology are always closely linked in Mīmāṃsā. In fact, the main concern of Mīmāṃsā is the Veda and the Veda is a linguistic entity and, insofar as it is one, it is a valid instrument of knowledge. Hence, analysis of language is preliminary in order to understand how can the Veda convey knowledge. But, Bhāṭṭas and Prābhākaras agree that the Veda is made of prescriptions. How can prescriptions convey knowledge? Since it has already been observed that the discussion on instruments of knowledge plays in Indian philosophy a role comparable to the debate about truth in the Western one, the problem amounts to the Western question about whether injunctive sentences may have a truth-value (see Copi-Cohen). It is, in fact, difficult to figure out how to understand "truth" in regard to a non-descriptive statement. Some thinkers (the deontic logician Stig Kanger and some Christian theologians, for instance) maintain that the definition of truth as correspondence still holds. One only has to compare the content conveyed by the prescription with what should occur according to an ideal paradigm, be it God's Will or Kant's "kingdom of ends". In this way, the oddity of a prescriptive truth value is solved.
Something similar is attempted by Maṇḍana. His view of prescriptions as assertions conveying the idea that the action to be undertaken is a means to a desired end makes the attribution of truth-claims (in Indian terms: the attribution of prāmāṇya, i.e. the capacity to convey valid knowledge) to them, smooth. It is indeed easy to say that the Veda is an instrument of knowledge insofar as it conveys the idea that the Full- and New-Moon sacrifice leads one to heaven (obviously enough, verifying it is still far from easy, but this has nothing to do with the logical oddity of the Veda being at the same time prescriptive and an instrument of knowledge). Hence, the prescriptive "One ought to sacrifice" is said to be tantamount to "Sacrifice is an instrument to something good", which is much easier to analyse.
In logical terms:
would be the same as
s is p
with O=ought, the deontic modal, s=sacrificing and p=being an instrument to something good.
But in this way the modal specificity of prescriptions is just cleared away. This is also the fundamental objection of Rāmānujācārya against Maṇḍana (see TR IV §3.2): that an action is the instrument to realise something desired is an assertion. But it does not entail that one ought to undertake the action, although Maṇḍana might object that everyone undertakes actions, if only they lead to something desired. Finally, one no longer obeys the Veda because the Veda ordered one to do it, but rather because the action enjoined are conducive to something desired. Hence, I guess, no other harm should derive if one does not undertake them, but the non-attainment of the desired thing. Since, however the desired thing is happiness itself, its non attainment is enough to make everyone strive for the opposite.
On the other hand, the (other) Bhāṭṭas and the Prābhākaras take seriously in account the specificity of exhortations as against statements about actual state of affairs. An exhortation, according to both schools, cannot be reduced to a descriptive statement. This is even more true in the case of the Prābhākaras, which claim that the Veda points at conveying something to be done. In this case, the validity-criterion for the Veda is that it has to convey valid knowledge in its specific field, that is, the sphere of what has to be done. Since this sphere is not attainable through any other instrument of knowledge (as all other instruments of knowledge refer to what is presently available to one's grasp), the Veda is the only possible instrument of knowledge about what has to be done. The Bhāṭṭas are somehow more moderate, insofar as they refer the Veda's validity rather to the sphere of future state of affairs –which are also beyond the grasp of our faculties.