This is a G2P (grapheme to phoneme) translator for Turkish that uses orthography and phonotactic rules. It additionally syllabifies the translated words. There are two .csv files encoding underlying and allophonic sounds. There are 37 phonemes for consonants and 15 phonemes for vowels.
- Consonants IPA: b, d͡ʒ, t͡ʃ, t͡ʃʰ, d, f, ɸ, g, g̟, ɣ, h, ç, x, ʒ, k, kʰ, k̟, k̟ʰ, l, ɬ, ʎ, m, n, p, pʰ, ɹ, r̥, r, s, ʃ, t, tʰ, v, β, ʋ, j, z
- Vowels IPA: a, a:, a̟, e, ɜ, ɯ, i, i:, u, u:, u̟, y, o, o̟, ø
-
Vowel fronting
Short back vowels (a, u, o) are fronted when they are in the environment of a liquid (j) or preceded by a lateral (l, ɬ, ʎ). -
Open “e”
The vowel “e” gets more open (ɜ) preceding a sonorant (l, m, n, r, j). -
Consonant fronting
Velar consonants (g,h,k) are fronted when preceding a front vowel (e, i, i:, y, ø) -
Consonant aspiration
Voiceless stops (k, p, t) and affricate (t͡ʃ) are aspirated word finally, preceding a vowel and in complex onsets. -
V changes
The underlying voiced labiodental fricative (v) becomes an approximant (ʋ) in between vowels and if one of the vowels are round (u, u:, u̟, y, o, o̟, ø). It becomes a bilabial (β) if it is in the environment of a round vowel and a consonant. -
F changes
The underlying voiceless labiodental fricative (f) becomes a bilabial (ɸ) if it is in the environment of a round vowel (u, u:, u̟, y, o, o̟, ø). -
R changes
The underlying alveolar rhotic (r) becomes a tap (r̥) at the end of a word. It becomes a trill (ɹ) in the beginning of a word.
Alpha. Further development points:
- Improve the rule profile
- Turn the script into a function with toggleable rules
- Include a dictionary for exceptional underlying forms
- Overhaul (version shift Beta)
- Provide a rule writing scheme
- Optimize for bulk translation
- Come up with a learning algorithm for selecting epenthesis in consonant clusters (version shift 1.0)
- Provide acoustic profile for underlying forms