cff-version: 1.2.0
title: "uyghur-normalizer: 2009 imla orthographic and de-calquing normalization for Uyghur"
message: "If you use this software or its map data, please cite it."
type: software
# Author is an entity with `name:` — Zenodo's CFF parser does not accept
# a person with only `alias:`, and fails silently (release publishes, no DOI).
authors:
  - name: UyghurReferenceTexts
    website: "https://github.com/UyghurReferenceTexts"
repository-code: "https://github.com/UyghurReferenceTexts/uyghur-normalizer"
abstract: >-
  A dependency-free normalizer for Uyghur text in the Arabic script. Applies
  three ordered passes: terminology calqued from Chinese, pre-2009 spellings,
  and loanword shape under the 2009 orthographic reform. An optional pass
  corrects common misuses of words, from the corrections marked in the
  6-volume Uyghur edebiy tilining izahliq lughiti (1990-1999). Suffixes are
  re-derived against the replacement with vowel harmony, voicing assimilation
  and raising, and the tool declines rather than guesses where the derivation
  is not safe. Includes the map data, arbitrated word by word against the 2009
  Hazirqi zaman Uyghur edebiy tilining imla lughiti dictionary of the Milletler
  Til-Yeziq Komiteti and 2011 Hazirqi zaman Uyghur tilining izahliq lughiti
  (qisqartilmisi), and checked against every headword of both, with
  hand-verified headwords required to pass unchanged. Every rule is traceable
  to the dictionary row and scan image that attest it.
keywords:
  - Uyghur
  - orthography
  - imla
  - text normalization
  - low-resource languages
  - machine translation
  - linguistics
license: MIT
license-url: "https://github.com/UyghurReferenceTexts/uyghur-normalizer/blob/main/LICENSE"
doi: "10.5281/zenodo.22000836"
version: 2.0.1
date-released: "2026-10-17"