dmort27 / epitran

A tool for transcribing orthographic text as IPA (International Phonetic Alphabet)
MIT License
625 stars 120 forks source link

Add Finnish #169

Closed jessicah25 closed 5 months ago

jessicah25 commented 5 months ago

Pronunciations are mostly taken from Finnish: A Comprehensive Grammar [1], and The world's writing systems [2]. Additional rules for pronunciation are taken from Finnish sound structure: Phonetics, phonology, phonotactics and prosody [3]. We choose rules that mostly conform to the pronunciations given in [1] and [2] - namely we do not use the dental t, the flap ɾ, and we also choose s -> x / r instead of r -> ɹ̝ / s . We consult Wikipedia [4] for an initial survey. Additional resources have slight variations in rules [5]. Rules and pronunciations are supplemented or supported by [6].

Rule for sr: s -> x / _ r

Rules for /h/: h -> ç / (::highfrontvowel::) (::consonant::) h -> x / (::backvowel::) (::consonant::) h -> ɦ / (::vowel::) _ (::vowel::)

Rules for /m/ and /n/: m -> ɱ / f n -> n̪ / t n -> n̪ / t _

Rules for /l/: l -> l̪ / t l -> l̪ / t

Rule for /ʋ/: ʋ -> w / (::u_enddiphthongs::)

Rules for /ŋ/: n -> ŋ / k g -> ŋ / n

Test cases are taken from a combination of the resources and words that we found to address edge cases. Wiktionary [7] has the pronunciations for these words, but does not fully follow the rules we list, so we edit these words for the test cases.

[1] Karlsson, Fred. Finnish: A comprehensive grammar. Routledge, 2017. [2] Daniels, Peter T., and William Bright, eds. The world's writing systems. Oxford University Press, 1996. [3] Suomi, Kari, Juhani Toivanen, and Riikka Ylitalo. Finnish sound structure: Phonetics, phonology, phonotactics and prosody. University of Oulu, 2009. [4] https://en.wikipedia.org/wiki/Help:IPA/Finnish [5] https://www.jkorpela.fi/suomi/fon.html [6] Karlsson, Fred. Finnish: An essential grammar. Routledge, 1999. (https://books.google.com/books?id=NqEEvGzH0mwC) [7] https://en.wiktionary.org/wiki/Wiktionary:Main_Page

dmort27 commented 5 months ago

Looks good.