prescription because the reciters describe the regular form that would result from
uru followed bya ̄in their ordinary speech, that is, the contemporary Sanskrit
they spoke and were familiar with. The form urva ̄, which happens to be absent
from the finite corpus of the Vedas, is grammatical: it belongs to the infinitely
many expressions of language that a native speaker is able to produce. The so-
calledana ̄rs.aforms, then, which do not come from the Vedas, are Sanskrit and
descriptive. They are, in fact, the precursors of Pa ̄n.ini’s description of a form of
the spoken language that he regarded as exemplary.
2Pra ̄tis ́a ̄khya
The practice of the Padapa ̄t.ha was formalized in the Pra ̄tis ́a ̄khyacompositions,
in principle one for each s ́a ̄kha ̄or “Vedic school.” The earlier name pa ̄rs.adasug-
gests that they belonged to a community (paris.ad) which orally transmitted and
discussed the contents of their s ́a ̄kha ̄. The relationship between continuous recita-
tionandword-for-wordanalysis is expressed ambiguously by the Sanskrit sen-
tencepadaprakr.tih.sam.hita ̄which may be taken to mean: “The Sam.hita ̄ is the base
of the Pada” or: “The Sam.hita ̄ has the Pada as its base.” The Pra ̄tis ́a ̄khyagener-
ally adopt the second interpretation, puzzling from a historical perspective, but
signalling a return to the original composers who put words together in their
inspired speech just as any user of language puts words together when speak-
ing or writing. This illustrates that Indian grammarians regarded language as
unchanging, an erroneous assumption which led, however, to the linguistically
productivesynchronicanalysis that was advocated in Europe more than two mil-
lennia later by Ferdinand de Saussure.
ThePra ̄tis ́a ̄khyasare early because of their structure and function, not
because of the forms that survive and may have been influenced by Pa ̄n.ini or
other grammarians. Their original aim was completeness. Whitney noted on a
section ofTaittirı ̄ya Pra ̄tis ́a ̄khya that he could not discover any case of a retroflex
nasal arising in the Taittirı ̄ya Sam.hita ̄ from a dental nasal in the Taittirı ̄ya
Padapa ̄t.ha“that was not duly provided for.” Weber used the term “complete”
(vollständig) and a century later, Surya Kanta called the Rik-Pra ̄tis ́a ̄khya “entirely
free from all oversights.” To be complete is only possible when we deal with a
finitecorpus ofutterances, (copies of) unique events in space and time such as
the extent of a Vedic s ́a ̄kha ̄.A living language consists ofsentenceswhich cannot
be enumerated because they are infinite in number as Patañjali knew. Follow-
ing Suryakanta, we postulate that the Pra ̄tis ́a ̄khyas originally produced exhaus-
tive listings (gan.as) of examples, showing patterns of linguistic structure, and
proceeded only later to generalizations explained by su ̄trarules. This dichotomy
of ideal types does not always survive in the texts as they have come down to us
because these have been influenced by Pa ̄n.ini or other grammarians.
ThePra ̄tis ́a ̄khyas introduced the metalinguistic use of case-endings, at first
the Nominative and Accusative as formulated by the metarule:”the expression
‘this (Nom.) that (Acc.)’ means ‘becoming-that’ with reference to the sound
the science of language 351