pischel alignments
Prakrit forms from Pischel's grammar, aligned to the Sanskrit equivalents he gives. Each table is one Sanskrit word. The top row is the Sanskrit, and each row below gives, under every Sanskrit segment, the Prakrit segment that continues it, or - where it was lost. Segments a Prakrit form added are shown in green before the segment they precede. The alignments were made by a neural CRF aligner and have not been reviewed. See also the reviewed alignments.
Source. R. Pischel, Grammatik der Prakrit-Sprachen (Strassburg, 1900), read from the English translation by Subhadra Jhā, A Grammar of the Prākrit Languages. Page numbers are those of the translation. The forms were transcribed from page images by AI.
Accuracy. On eight random pages, two independent readings agree on 96.4% of 589 forms and on 95% of 262 Sanskrit equivalents, and their variety labels differ on 5.4% of rows. Expect about 2% of forms to carry a wrong diacritic, most often a missed underdot.
Left out. 15,106 citations have no Sanskrit equivalent printed, so they have nothing to align to. Many are paradigm forms. A form with no equivalent beside it that has exactly one elsewhere in the book uses that one.
Labels. D or Ḍh marks rows that could not be sorted between Dākṣiṇātyā and Ḍhakkī. Pkt means Pischel names no variety. Sets are built on the Sanskrit as printed, so inflected forms of one word fall in separate sets.