Module talk:tzm-IPA
Add topicLatest comment: 3 years ago by Malku H₂n̥rés in topic Ready module
Ready module
[edit]@Metaknowledge the module is now ready for actual use! Thus I'm asking you to move the module page back to its regular name (ie. without my username) (1) and the deletion of the sandbox page and its associated sandbox template (2), as well as your authorization to use the module in all the Central Atlas Tamazight entries (3).
For the latter, I tested, in preview mode, every single element of the module, and then every single phonological rule applied in it, adding the subsequent rule when the precedent version of the module was working (as can attest the 150 edits in the sandbox module).
Eventually, just so you know, given that the phonotactics of the aforesaid language are quite complex, a wrong pronunciation might unexpectedly happen from time to time, but I'll fix this when it happens. Malku H₂n̥rés (talk) 11:51, 7 April 2021 (UTC)
- @Malku H₂n̥rés: It looks good, but why is there no testcases subpage to demonstrate that the more complex rules work as intended? —Μετάknowledgediscuss/deeds 16:09, 7 April 2021 (UTC)
- @Metaknowledge thank you it's the thing I wanted to mention but I forgot what it was. I had the intention of saying that I'd be glad to leave thus the proof it works, however I don't know how to manage it. The module returns a string which contains the dialect and standardized IPA pronunciations (named f). Since there's currently only one dialect, the best solution would be to somehow get back the string with the phonetic pronunciation (named n) at the end of the module but I ignore how to manage it. As for the examples, Penchoen provides them in his grammar, but they are there to show how it works and not whether it works. Therefore it is a good basis but not sufficient to ensure that all the possible cases are actually handled by the module, so new cases will have to be added stepwise, in keeping with new examples come. Malku H₂n̥rés (talk) 17:58, 7 April 2021 (UTC)
- @Malku H₂n̥rés I have reworked a bit the module to make it testable and also added the infrastructure for tests. See Module:User:Malku_H₂n̥rés/tzm-pron/testcases. — Fenakhay (تكلم معاي · ما ساهمت)
- @Fenakhay thank you very much for the testcases page! First of all you made changes to the module, I can only notice it still works fine, but I can't follow the changes you made to the rules (variable r). Did you change something in the strings ? Moreover, why testing script whereas this is already done in Module:ber-translit ? I wanted to make a minimalist module (so as not to bother anyone with it and because it's enough big and complex for a simple pronunciation module in a Berber language) but anyway. The use of the variables (whence their name for most of them, so that we could replace their names to make them 'meaningful') are the following:
- s: substitution (for gsub which means global substitution)
- c: correspondences [from transcription to phonemic]
- p: phonetic [from phonemic]
- t: tense [consonants]
- o: order [of rules application]
- C: consonant
- V: vowel
- r: rules
- a: arguments
- tr: transcription
- f: final [ie. returned string]
- i, j, k: sequence of loops
- n: temporary (reset after each dialect)
- N: alternative of n (for 'while' loops, it allows to test if n is different now and before the substitution)
- x, y: just two consecutive unused letters (first used in sandbox so I didn't care about their name sorry)
- d, g: forgot what it was supposed to mean
- Whence another question: why did you replace 'o' by 'rules' knowing 'r' ? It's just variables names, not very important. As for the testcases, there seems to be a bug on the first test, the syllabic diacritic is displayed under 'fl' (as if it were a ligature) instead of being just under 'l'. This never happened during the tests I made in preview mode. Malku H₂n̥rés (talk) 11:24, 8 April 2021 (UTC)
- @Fenakhay I'm glad to tell you the module is finally ready! All the bugs are fixed now, it's ready to be used in entries, if you allow it and move the page to its regular name (without my username).
- The rules of Penchoen for schwa insertion weren't consistent, not very logic (from the end of the word forward), with exceptions and very complex and not respected in his own grammar. Therefore I made new rules for this and tested it on a corpus of roughly 250 words whose schwa positions are known (provided at the end of his grammar as well as the examples of the 'Phonology' part) and had only 3 exceptions (instead of a dozen with Penchoen's rules). Also, he doesn't say what about <ⴰⵓ> /au/ and <ⵓⵓ> /uu/ (ie. if a yod is inserted or one of both deleted) but it seems that these sequences or vowels simply don't exist in this language.
- Eventually, is it possible to add a parameter to add another pronunciation and Lua error in Module:parameters at line 376: Parameter 1 should be a valid language or etymology language code; the value "Arabism" is not valid. See WT:LOL and WT:LOL/E. before it? Indeed there are plenty of borrowings from Arabic into Central Atlas Tamazight which kept the stop consonants or other features not native from the Ayt Ndhir dialect, and it is needed to have a parameter to put directly the phonetic pronunciation. Malku H₂n̥rés (talk) 11:55, 19 April 2021 (UTC)
- @Fenakhay thank you very much for the testcases page! First of all you made changes to the module, I can only notice it still works fine, but I can't follow the changes you made to the rules (variable r). Did you change something in the strings ? Moreover, why testing script whereas this is already done in Module:ber-translit ? I wanted to make a minimalist module (so as not to bother anyone with it and because it's enough big and complex for a simple pronunciation module in a Berber language) but anyway. The use of the variables (whence their name for most of them, so that we could replace their names to make them 'meaningful') are the following:
- @Malku H₂n̥rés I have reworked a bit the module to make it testable and also added the infrastructure for tests. See Module:User:Malku_H₂n̥rés/tzm-pron/testcases. — Fenakhay (تكلم معاي · ما ساهمت)
- @Metaknowledge thank you it's the thing I wanted to mention but I forgot what it was. I had the intention of saying that I'd be glad to leave thus the proof it works, however I don't know how to manage it. The module returns a string which contains the dialect and standardized IPA pronunciations (named f). Since there's currently only one dialect, the best solution would be to somehow get back the string with the phonetic pronunciation (named n) at the end of the module but I ignore how to manage it. As for the examples, Penchoen provides them in his grammar, but they are there to show how it works and not whether it works. Therefore it is a good basis but not sufficient to ensure that all the possible cases are actually handled by the module, so new cases will have to be added stepwise, in keeping with new examples come. Malku H₂n̥rés (talk) 17:58, 7 April 2021 (UTC)