Datasets:
Tasks:
Automatic Speech Recognition
Modalities:
Text
Formats:
json
Languages:
Kamba (Kenya)
Size:
< 1K
License:
File size: 1,029 Bytes
6a25a9d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 | # Kamba orthography and normalization
**Status: not yet written.** Roadmap item v0.1 requires examining real transcripts from the
candidate corpora first; the normalization rules themselves are v0.2+ work and must involve
native Kamba speakers before anything beyond `normalize.py` v0.1's language-agnostic base layer
is implemented.
This document will hold, once that work happens:
- the linguistic justification for any spelling variants the normalizer treats as equivalent
(the kind of citation-backed reasoning a normalizer needs — not rules invented without native-
speaker or literature support),
- documented accepted spelling variants and how they were identified,
- the relationship between this document and `normalize.py`'s `NORMALIZATION_CHANGELOG`, so every
normalization version is traceable to the linguistic reasoning behind it, and
- open questions about dialect-specific spelling differences, if any are found.
See `kambabench/scoring/normalize.py` for the current (v0.1, base-only) implementation.
|