Template talk:ko-conj-adj/auto

About this module and template
It looks very promising and educational. Does it do only adjectives (it has "adj") or verbs as well? Sorry, if it's a silly question. My knowledge of the Korean grammar is very basic and I only know forms I come across often. I couldn't find good Korean grammar references - my textbooks only give very common forms, perhaps less than a half of what is in the template. So I don't even know their usage. I don't feel confident right now but I will try t use later with your guidance. --Anatoli (обсудить/вклад) 00:10, 11 April 2014 (UTC)


 * It's just adjectives currently; I'm working on verbs now (Module:User:Wyang/ko-conj). Verbs are slightly more complicated than adjectives. I'm not changing the basics of the conjugation templates - everything the new module Module:ko-conj is doing is it generates all the parameters and the table automatically. The forms in these templates are basically correct, although occasionally some forms of some adjectives/verbs may not exist in the actual language due to semantic constraints (these defective ones have in part been accounted for in the templates).
 * Module:ko-conj uses Module:ko-translit to generate the transliterations, which is quite a crappy transliteration module at the moment. A lot of words are not transliterated correctly. I don't think the transliteration approach used there is the most convenient approach (it would require a 28*16 matrix to be filled in every cell to make it work correctly), and I'd like to rewrite it. I'd also like to modulise (automatic IPA, and RR, MC, Yale, RRV transcriptions) so that  at 한국어 is able to generate everything that is there now. Wyang (talk) 00:23, 11 April 2014 (UTC)


 * Thanks. I'm happy to look at Korean transliteration again but perhaps we should be using User:TAKASUGI_Shinji recommendations - the phonetic RR transcription of the National Institute of the Korean Language. Is that what you have in mind? (I can be more helpful with the transliteration module than with grammar). Perhaps it's not as bad as you describe and someone might help. We can discuss corner cases. --Anatoli (обсудить/вклад) 00:39, 11 April 2014 (UTC)


 * Yes. That is the default used here too. I'd say the transliteration module is quite bad... if you have a look at the table for 노랗다 at the bottom, the "sentence-final forms" - less than half of it is correct. eg.

indicative 노랗다 (norahda) past: 노랬다 (noraetda) 노래다 (noraeda) past: 노랬어 (noraeteo) 노래요 (noraeyo) past: 노랬어요 (noraeteoyo) 노랗습니다 (norahseumnida) past: 노랬습니다 (noraetseumnida) interrogative 노라냐 (noranya) past: 노랬냐 (noraetnya) 노래 (norae) past: 노랬어 (noraeteo) 노래요 (noraeyo) past: 노랬어요 (noraeteoyo) 노랗습니까 (norahseumnikka) past: 노랬습니까 (noraetseumnikka) assertive 노랗겠다 (norahgetda) 노랗겠어 (norahgeteo) 노랗겠어요 (norahgeteoyo) 노랗겠습니다 (norahgetseumnida)

The correct ones are:

indicative 노랗다 (norata) past: 노랬다 (noraetda) 노래다 (noraeda) past: 노랬어 (noraesseo) 노래요 (noraeyo) past: 노랬어요 (noraesseoyo) 노랗습니다 (noraseumnida) past: 노랬습니다 (noraetseumnida) interrogative 노라냐 (noranya) past: 노랬냐 (noraennya) 노래 (norae) past: 노랬어 (noraesseo) 노래요 (noraeyo) past: 노랬어요 (noraesseoyo) 노랗습니까 (noraseumnikka) past: 노랬습니까 (noraetseumnikka) assertive 노랗겠다 (noraketda) 노랗겠어 (norakesseo) 노랗겠어요 (norakesseoyo) 노랗겠습니다 (noraketseumnida)

Wyang (talk) 00:58, 11 April 2014 (UTC)


 * All incorrect cases are to do with jamo ᇂ and a case with ᆻ, which should be "ss" before vowels? Do you think they are really hard to fix? I'll add them to Module:ko-translit/testcases. The finals with consonant clusters need some attention as well (not all current cases are correct). --Anatoli (обсудить/вклад) 01:06, 11 April 2014 (UTC)

Basically everything case outside this 12*7=84 table is potentially incorrect.

Which means there are 364 remaining cells (out of 448) in the matrix to be fixed. Palatalisation is excluded from this too. Wyang (talk) 01:13, 11 April 2014 (UTC)


 * User:TAKASUGI_Shinji mentioned that a French template(s) for Korean transliteration does it well. Is it worth looking at its logic? I'll try to find it. --Anatoli (обсудить/вклад) 01:21, 11 April 2014 (UTC)


 * It is fr:Modèle:ko-pron, fr:Modèle:ko-translit and fr:Modèle:ko-roman. All require decomposed Hangul syllables and some further specifications. The template zh:Template:朝新 does the same too (but substitution only), but does not require characters to be decomposed. All of these have extraneous features. Wyang (talk) 01:24, 11 April 2014 (UTC)


 * We have a Hangeul decomposition function, don't we? The approach can be simpler, IMO. The romanised text could be fixed instead. E.g. 1. Convert all to default readings. 2. Fix incorrect cases, like hg(+ vowel)->k (+ vowel), (final) -ss,-ch, -j, s->-t, ti->chi, etc. Do you think it may work? --Anatoli (обсудить/вклад)


 * Yes, but I think it would be better to achieve decomposition and transliteration all in Lua. These templates were written before Hangul syllable analyses were possible. Wyang (talk) 01:44, 11 April 2014 (UTC)
 * So, that means it will take a lot of typing based on the table above? Could you give me an example of what you need that's missing? The conversion table could go into a separate data module. --Anatoli (обсудить/вклад) 01:50, 11 April 2014 (UTC)

There are more cases than this. For example, kh-final: 부엌 (eg. 부엌간, 부엌방), ph-final: 숲길, 엎누르다, 덮다, th-final: 낱말, 붙잡다, 끝나다, ch-final: 꽃가게, 옻나무, 숯검정, 낯모르다, kk-final: 뜻밖에, 볶음밥, 낚시, j-final: 젖먹이, plus all the consonant cluster cases: 닭장, 산기슭, 삶기다, 굵다랗다, 밞히다, 늙은이, 넓삐죽하다, 찰흙, 몫몫이, 뱃삯, 많아지다, 끊기다, 얹혀살다, 옰, 홅이다, 읊조리다.... The approach should not be trying to exhaustively list the consequence of each possibility, but follow principles as to why the changes occur. Eg. It should search for a intersyllable sequence of [k,p,t,j,ch,kh,ph,th,kk,pp,tt,jj][mnr], and convert the preceding stop into the homorganic nasal, and the second into 'n' if 'r'. Eg2. It should search for a sequence of two plosives: [k,p,t,j,ch,kh,ph,th,kk,pp,tt,jj][k,p,t,j,ch,kh,ph,th,kk,pp,tt,jj], and convert the first plosive into the corresponding unreleased stop [k,p,t] and keep the second unchanged (transliteration) or convert it into a tense consonant if unaspirated (IPA). Wyang (talk) 02:28, 11 April 2014 (UTC)