What an Editor Does With a Variant Spelling

The temptation, reading an inscription in a language one knows only from its classical form, is to correct. A Prakrit word that looks like a misspelled Sanskrit one invites emendation, and the practice of Indian epigraphy is firmly against it. What the stone shows is evidence about the language actually spoken and written in a place at a time, and normalising it destroys exactly the information the inscription uniquely carries.

The alternation between this letter and the labial semivowel is the standard case. It is not a scribal error but a real merger in the dialects concerned, and its distribution across the sites where the Ashokan edicts were cut is one of the pieces of evidence used to group those dialects. Correcting the spellings would erase a dialect boundary. The same holds for the eastern substitution of l for an inherited r, for the varying treatment of consonant clusters and for the reduction of the three sibilants to one.

A modern edition therefore prints the text as cut, with the Sanskrit equivalents in an apparatus, and the apparatus is often longer than the text. That is not fussiness. The Ashokan corpus is the earliest substantial written material in any Indian language, and the variations in it are close to the whole of what is known about how the language of the third century BCE actually differed from place to place.