[01:46:17] *** Quits: wmacmil (~wmacmil@c83-252-138-144.bredband.tele2.se) (Ping timeout: 258 seconds) [02:17:32] for general tips on working with bidirectional text, I have a blog post :-D https://inariksit.github.io/gf/2019/04/06/rtl.html [02:17:56] > -could it have something to do with the writing backwards that it is trying to inflect the word but taking the first letter instead of the last?? [02:19:54] no, I'm pretty sure it isn't that! there are a couple of different paradigms, the one that's just 1-argument mkV for a string expects 3rd person masculine imperfect [02:24:24] then you've got all these constructors with numbers: v1-v11 which take just a string, exception v1 which also needs to take a vowel [02:24:42] and the mkV is a special assumption about v1, where the vowels are i and a [02:26:07] the numbers are explained in the API here http://www.grammaticalframework.org/lib/doc/synopsis/index.html#toc87, and also here with actual arabic fonts https://en.wiktionary.org/wiki/%D9%81_%D8%B9_%D9%84#Arabic [02:28:29] here's the implementation of 1-argument mkV: [02:28:31] regV : Str -> V = \wo -> [02:28:31] let rau : Str * Vowel * Vowel = [02:28:31] case wo of { [02:28:31] "يَ" + fc + "ُ" + l => ; [02:28:31] "يَ" + fc + "ِ" + l => ; [02:28:32] "يَ" + fc + "َ" + l => ; [02:28:32] f@? + "َ" + c@? + "ِ" + l => ; [02:28:33] _ => Predef.error ("regV not applicable to" ++ wo) [02:28:55] so the first 3 cases are to check if it's a weak stem, which starts with a semivowel [02:30:37] if it doesn't match that, then it tries to match any character, followed by the diacritic fatha (that marks a short "a"), any character, a diacritic kasra (marks the short "i"), and ending in any character [02:31:07] and if it's not any of these, then it just tells that regV is not applicable [02:31:47] I don't personally know why it doesn't continue pattern matching like radical 1, , radical 2, , radical 3 and just give the vowels forward [02:32:56] 100% speculation, but class 1 seems to be the only one where verbal nouns (called "masdar" in the documentation) are unpredictable, could it be that the version of Form 1 with vowels i-a is the only one where the masdar is predictable= [02:32:58] *? [02:33:48] if you want to forward your student to talk directly to me, I'm fine with that! they can email me at inari@digitalgrammars.com [02:43:29] my advice is to avoid as much as possible writing concrete strings with the diacritics, it's easy to get wrong -- at least for me who doesn't speak any arabic! there is this collection of patterns https://github.com/GrammaticalFramework/gf-rgl/blob/master/src/arabic/PatternsAra.gf [02:43:52] the patterns are applied to the roots to produce appropriate word forms for the different syntactic functions [02:44:58] it may be harder to debug, because if you want to make a lexicon entry, your input will not be any of the actual forms that are inserted into the inflection table [02:45:38] but given how the arabic morphology works, I have hard time thinking of a better solution [03:50:43] *** Joins: drbean (~drbean@TC210-63-209-184.static.apol.com.tw) [08:17:40] *** Joins: wmacmil (~wmacmil@c83-252-138-144.bredband.tele2.se) [10:07:18] of course there is always something useful in inari's blog :) how could i forget about it [11:04:20] *** Joins: dobrovskaya (~dobrovska@139.red-83-44-195.dynamicip.rima-tde.net) [11:04:43] *** Quits: drbean (~drbean@TC210-63-209-184.static.apol.com.tw) (Quit: ZNC 1.8.2+cygwin2 - https://znc.in) [12:13:34] daherb: hehe, that's not as obvious thing as more directly GF-related posts [12:13:46] (re: the post on bidirectional text) [12:18:24] but now that you mentioned it, i remembered its existence [12:19:35] inariksit: btw did you hear about denny's talk at gu/chalmers? [12:56:21] *** Quits: wmacmil (~wmacmil@c83-252-138-144.bredband.tele2.se) (Ping timeout: 258 seconds) [13:41:32] *** Quits: dobrovskaya (~dobrovska@139.red-83-44-195.dynamicip.rima-tde.net) (Ping timeout: 252 seconds) [13:58:02] *** Joins: drbean (~drbean@TC210-63-209-205.static.apol.com.tw) [14:56:53] daherb: yes, aarne told me! [14:57:03] I didn't get a link though yet, he just said that it will happen [14:57:09] is there a link and time for it? [15:05:36] i can forward the mail [15:07:00] you should have mail now [15:47:20] *** Quits: drbean (~drbean@TC210-63-209-205.static.apol.com.tw) (Quit: ZNC 1.8.2+cygwin2 - https://znc.in)