Boundaries and sacredness

One of the characteristics of whatever is "sacred" or "religious" seems to be its being segregated from the corresponding normal behaviour (see M.S.'s point about it in this post). This might mean that walking, eating, cooking, etc., might all be secular as well as religious activities and that the difference lies first and foremost in their context. If the context sets a precise boundary, secluding them from normal experience, then they acquire a religious meaning. This is what would happen within a Vedic ritual, which encompasses all sorts of "normal" activities (such as the one listed above).
Does this apply to language as well? Indian grammarians seem to imply that the correct use of language bestows religious merit. But how does this happen? The topic has been dealt with recently by Paolo Visigalli and Marco Ferrante at the second CBF. I am sure I misunderstood most of their papers, but long summaries might be downloaded here. An intesting way to start the discussion is the first vārttika by Kātyāyana on Pāṇini. The vārttika says:

siddhe śabdārthasaṃbandhe lokato 'rthaprayukte śabdaprayoge śāstreṇa dharmaniyamaḥ yathā laukikavaidikeṣu.

In George Cardona's translation:

It is given from every day communication in the world that there is an established relation between words and meanings; it is also given that the use of a word is prompted by a given meaning in that one uses words in order to convey meanings. This being so, a restriction intended for merit (dharmaniyamaḥ) is established by the grammar, like in common and Vedic words.

The last point might imply that Grammar imposes a restriction into language and that only a so-restricted language may convey a meaning. Interestingly enough, the clause yathā laukikavaidikeṣu might mean that no language is intrinsically "sacred" (not even Vedic) and that, rather, sacredness depends on the way one adds special constraints to it. These constraints might be those of correctness, but perhaps also of a conscious usage (one which is made correct by Grammar and not just by the fact that one is a native speaker).
The main point seems to be that religious merit (dharma) has to do with a selection among equally possible options (niyama). Do readers know of other contexts of usage of the compound dharmaniyama?

I never blogged specifically on Indian Grammar, but if you look for "Grammar", you will find several posts dedicated to special issues (such as the minimal significant units according to Indian Grammarians, the interactions between Grammar, Mīmāṃsā and Ritual Sūtras, etc.).