Romanisation scheme
Also: IAST · ALA-LC · Transliteration standard
A specific convention for writing one script in Latin letters. IAST, ITRANS and Harvard-Kyoto all romanise Sanskrit, and all disagree.
Last updated
The same Sanskrit word is saṃskṛta in IAST, saMskRta in ITRANS and saMskRta in Harvard-Kyoto with different case conventions. None is wrong; they optimise for different things. IAST is diacritical and reversible, and dominates Indological publishing. ITRANS and Harvard-Kyoto are ASCII-only, which mattered enormously before Unicode and still matters for typing. ALA-LC is what library catalogue records use, which is what makes it the scheme to match if you want your citations to resolve against a catalogue.
The practical consequence: a transliteration whose scheme is not stated cannot be searched for reliably, compared with other scholarship, or converted back to the original script with confidence. It reads fine and is academically inert.
Archēglyph picks a documented default per language — IAST for Indic, ALA-LC for Arabic and Persian, Hepburn for Japanese, Hanyu Pinyin for Chinese, BGN/PCGN for Cyrillic — names it in the instruction the model receives, and displays it beside the text it produced.