A Grammar Built Like a Machine

The Ashtadhyayi describes the whole of Sanskrit in just under four thousand rules, and the reason it is still studied is the engineering rather than the coverage. It opens with a set of sound groupings arranged so that any needed class of letters can be named by taking its first member and a marker, which lets a rule refer to a category in two syllables instead of listing it. It has rules about rules, governing how far a heading extends and which rule wins where two apply. Its default is that a more specific statement overrides a more general one, and that a later one overrides an earlier one, so the ordering of the text is itself part of the content.

Nothing about it is written to be browsed. It is composed to be memorised and recited, and brevity is pursued to the point where saving a single syllable is a recognised achievement in the commentary tradition. The result behaves less like a description than like a procedure: given a meaning and a set of elements, the rules apply in order and a correct form comes out. Linguists in the twentieth century, meeting it after developing formal grammars of their own, found a system they could recognise, composed at least two thousand years before them, and its structure is still argued over as a technical object rather than only as a historical one.