Template talk:ko-conj-verb

Forms to show
A great idea, but where does it end? ;-) My learner's dictionary of Korean endings (어미조사 사전, Hankook Munhwasa, 2006), in its appendix of verb conjugations, lists the following:  dictionary (가다), present attributive (가는), rational (가니까), causative (가서), conditional (가면), conjunctive (가고), intentional (가려고), contrastive (가지만), polite-stem (가), plain indicative (간다), formal indicative (갑니다), plain past (갔다), spoken plain interrogative (가냐), plain interrogative (가니), exclamatory (가는구나!), plain imperative (가라), and suggestive (가자) forms... but even this is clearly just a small sample, and the dictionary itself lists some hundreds of verb endings (many of which can be combined).  And that's just a learner's dictionary... -- Visviva 15:56, 24 May 2007 (UTC)


 * Yes, where does it end? The too-vague-to-be-helpful answer is to show the most common grammatical categories (tenses/moods/etc.) in the most frequent honorific/speech level combinations.  E.g. I left out the lesser-used speech levels and the honorific versions of the non-polite speech levels because I get the impression that they aren't so commonly used as the others.  There's plenty of vertical room left in the chart, so I hope we can at least show the forms that would be expected in a learner's dictionary.  I'll see if my library has a copy of 어미조사 사전. Rod (A. Smith) 16:49, 24 May 2007 (UTC)


 * I believe the full, proper title is 한국어 학습 학습자용 어미 조사 사전 (with that pesky raised dot between "어미" and "조사"). ISBN 89-5726-393-4, for what it's worth; authors are 이희자 and 이종희.
 * A thorny issue here is what to do with those forms (like 가지만, 가면, etc.) which are found at all speech levels. Do we need a second table for these?  -- Visviva 18:05, 24 May 2007 (UTC)


 * I added some of the forms from your learner's dictionary to, but I don't know the actual rules for the others, so I'll wait until I have a decent reference for them. Hopefully I'll find one soon. Rod (A. Smith) 01:40, 25 May 2007 (UTC)


 * Along the lines of "where does it end?", are there any opinions about whether to show romanizations in the conjugation template? I've shown two different examples of how 하다 (hada) could display, above. Enabling romanization requires extra parameters.  Is romanization worth the verbosity?  Rod (A. Smith) 06:49, 25 May 2007 (UTC)


 * I think the policy is pretty clear that we should have romanizations, what with this being the English Wiktionary and all... it does seem like a godawful pain to generate them all for each verb, though. -- Visviva 06:59, 25 May 2007 (UTC)

Korean names for Japanese conjugation categories
The following form names were at 하다 (hada) before I removed them. They seem to have come directly from so they probably apply more to Japanese grammar than to Korean grammar, but here they are for temporary storage until we determine the proper forms to display: Rod (A. Smith) 20:51, 24 May 2007 (UTC)
 * Imperfective 미연형 (未然形)
 * Continuative 연용형 (連用形)
 * Terminal 종지형 (終止形)
 * Attributive 연체형 (連体形)
 * Classical imperfective 이연형 (已然形)
 * Hypothetical 가정형 (仮定形)

Usage
Would it be reasonable to state that this template should be used only on the main ("dictionary form"/"infinitive") entry for each verb? This is a lot of information to keep track of, and I think we would do best to have (for example) 하자 simply direct users to 하다 and/or to Appendix:Korean verb conjugations for conjugational info. -- Visviva 06:59, 25 May 2007 (UTC)


 * Absolutely. The same applies to all the conjugation and declension templates, synonyms, antonyms, translations, and etymology in all languages here.  It's been one of those unstated pseudo-policies: as much information as reasonable is kept in the lemma entry only. I suppose WT:AK should say something to that effect. Rod (A. Smith) 07:05, 25 May 2007 (UTC)

Alternative approach
Looking through my various learner's dictionaries and grammars, I found about 30 forms that are frequently included in verb conjugation charts. Interestingly this doesn't include most regular speech-level conjugations, but I think there are good reasons to include those as well... anyway, I found that these can be represented (including romanizations) for vowel-stem verbs with 11-14 parameters, or a few more in the case of consonant-stem verbs; see User_talk:Visviva/Conjugate for the demo. It would be possible to recode the template so that fully regular verbs like 가다 would only require 7 parameters.

Anyway, my draft is still rather rough but I'd like to see about how we can combine these two approaches... Cheers, -- Visviva 19:29, 20 June 2007 (UTC)


 * Yes, it would be ideal to combine the two approaches. ko-conj-verb allows the editor to override any individual form, hence the large number of parameters.  In most cases, only a few parameters are required.  One confusion I have is about whether to consider the honorific versions of verbs to be part of the conjugation set for that verb.  E.g. should we show the various forms of  within the conjugation table for, or are those forms better considered outside of the scope of conjugation?  Another point to clear up is whether to use transcription or transliteration.  (If we use transliteration, the romanized forms will be easier to specify.)  Rod (A. Smith) 19:48, 20 June 2007 (UTC)


 * Hmm... I think it will save us a lot of grief if we file distinct roots separately. (For one thing, the line between honorific and non-honorific forms is not as clear as one might expect; 먹으십시오 gets 27k hits.)  In fact I left out honorifics from my draft entirely -- not so much because I think they don't belong, as because I got a headache trying to figure out where to put them.  Technically the honorific stems (하시, 하셔) of a verb could generate a whole conjugation table by themselves (minus a few forms such as the hortative).  It wouldn't really be a problem to code this, but it would leave us with a table covering close to two screenfuls of text.  A bit much, even if it's auto-hidden.
 * Surprisingly -- with the predictable exception of certain consonant-stem verbs -- standard RR actually seems to agree perfectly with the three-stem system (먹,먹어,먹으). I hadn't really noticed this before, but stem1 (meok in standard RR) occurs as such only before consonants; otherwise stem2 or stem3 are always used.  And since word-final stops and stops occurring before another stop are romanized in the same way, the various k/g, r/l, etc. issues don't really arise.  Using RR transliteration would simplify things somewhat, but actually wouldn't make a big difference here.  So I'm not sure where I stand on that.
 * Here are some other picky technical issues that occurred to me as I was fooling around with this:
 * 1. Although having parameter names in hangul is very elegant, it is likely to create maintainability issues. Not everyone who comes through to debug or update the template is going to be able to read (or even display) Korean text.  This was why I opted for "hae," "hae_r," etc. rather than "해," "hae."
 * 2. When using a stem-based approach, it is imperative to distinguish the hani stem (e.g. 먹으) from the hada stem (e.g. 먹). The problem is that we can't use hada in this case, because for hada (and most if not all vowel-stem verbs) these stems are identical.  This is why I opted for "stem1," "stem2," "stem2a," "stem3," although obviously it would be nice to have names that are more transparent than these.  This order does, at least, correspond to the standard dictionary order of principal parts.
 * 3. All in all, the conjugation of Korean verbs is surprisingly regular.  If it weren't for the fact that hangul characters are stored in Unicode as unified glyphs, we would only need six arguments (for regular vowel-stem verbs) to fill out a conjugation table of arbitrary size.
 * I've been thinking about this far too much, as you can see. ;-) -- Visviva 03:15, 21 June 2007 (UTC)

Update: I've tried to combine the two approaches in the current form of the template. Feedback sought, especially if there are errors in the output. More coding is still in order -- in particular need a more elegant way to handle the ㅅ in 먹습니다, 받습니다, et al. -- but I'd like to use something like this here. Cheers, -- Visviva 16:26, 21 June 2007 (UTC)

Oof... If I had it to do over again, I think the ㄹ-stem verbs would get their own template. The "stem1a" workaround I came up with is pretty clunky... However, I'm fairly confident now that there are no major errors remaining in the output. If you find any please let me know... I think this is basically ready for prime-time, so if nobody screams I will boldly go forth and replace the current template with the new version shortly (after adding wikilinks and language coding). -- Visviva 15:02, 22 June 2007 (UTC)


 * I will review the output within a couple of hours, but your new approach now seems better than the original in several ways. Good job!  Rod (A. Smith) 16:19, 22 June 2007 (UTC)

Updated
I have updated the template from User:Visviva/Conjugate, per above discussion (some relevant discussion can also be found at the draft discussion page). Thanks to Rodasmith for his very helpful feedback.

As I noted on the draft discussion page, I think this will eventually need a trim in order to provide optimal value. But I am one of those who prefer to have (and provide) too much information rather than too little. :-)

If there are any lingering errors in the template, please note them here. Cheers, -- Visviva 15:43, 23 June 2007 (UTC)

ㅅ in formal past
Hm, I think that 했습니다, et al., are the correct forms. I should have time to look this up later today... but Google seems to agree with me; while 했읍니다 gets a respectable 547k hits, 했습니다 gets nearly 80 times that number. The situation is somewhat complicated by romanization, since in MR and RR the third "s" is omitted (just as it is in pronunciation). -- Visviva 03:49, 26 June 2007 (UTC)


 * http://www.hangul.or.kr/12b-1-7.htm and http://www.nzeo.com/bbs/zboard.php?id=content_korean&page=7&sn1=&divpage=1&sn=off&ss=on&sc=off&select_arrange=headnum&desc=asc&no=306 discuss ㅆ습 vs. ㅆ읍. With my poor Korean reading comprehension, I think those pages say that 1988 hangeul orthography changes (한글 맞춤법) caused the shift from ㅆ읍니다 to ㅆ습니다. Rod (A. Smith) 19:09, 26 June 2007 (UTC)


 * That does seem to be the case; at least, before the 1989 표준어 규정, it appears that both forms were accepted, but the 습 form was in more widespread use and was therefore selected as standard based on the principle that when two spellings are current for the same pronunciation, the more widely-used one is selected as standard. Which does raise the question of what North Korean orthographers think... -- Visviva

Interstitial s
I'm not clear about when to use the "interstitial" s parameter. Is it usually when the stem ends in a stop batchim (ㄱ,ㄲ,ㅋ,ㄷ,ㄸ,ㅌ,ㅅ,ㅆ,ㅂ,ㅃ,ㅍ)? Rod (A. Smith) 16:48, 28 June 2007 (UTC)


 * Yes, also ㄴ (안습니다), ㅁ (감습니다), and ㅎ (닿습니다, 않습니다)... it may be the case that s=yes whenever stem3=stem1+으, although I have an odd feeling that's not always true. -- Visviva 09:37, 7 July 2007 (UTC)


 * OK. I was confused about its purpose because in the template code it selects both -는- vs. -ㄴ- and -습- vs. -ㅂ-.  According to Seok Choon Song in Barron's 201 Korean Verbs (1998), -ㄴ- and -ㅂ- are used if and only if the verb stem ends in a vowel.  It says -는- and -습- are used otherwise (that is, it says that after taking into account the orthography change from -ㅆ읍- to -ㅆ습-). So, unless "interstitial s" is the common term for that selection, perhaps it would be better to replace the "s" parameter with a "consonant" parameter, indicating whether the stem ends in a consonant.  We could usually omit the parameter and derive it from stem3 (i.e. if the parameter is missing, the stem is assumed to end in a consonant if and only if stem3=stem1+으):


 * Better? Rod (A. Smith) 18:22, 7 July 2007 (UTC)


 * Works for me... I was thinking of perhaps a "cstem" parameter or some such, which would be set to the actual batchim if any (so cstem=ㅎ would stand in for h=yes and s=yes).
 * Straying slightly from the topic, it's not strictly true that a stem (meaning the true 어간, stem1) ending in a consonant always conjugates this way; the ㄹ-stem verbs like 몰다 (몬다, 몹니다) are the primary exception (but they form their stem3 differently -- 몰다, 몰아, 모니-- so wouldn't be affected by this). -- Visviva 03:51, 8 July 2007 (UTC)


 * I believe he exception I was thinking of was the irregular ㄷ-stem verbs like 묻다 in the sense "to ask" (묻는,묻습니다, but stem3=물으; see ). Of course there aren't many of these, so a simple manual override is probably sufficient.  -- Visviva 08:14, 8 July 2007 (UTC)

Optional parameters
Some of the previously required parameters (hal, ham, han, hap, and <tt>haet</tt>) are not really necessary for regular verbs and adjectives whose stems end in a consonant, so I modified the verb template to derive them when stem2 or stem3 gives enough information to do so. The change seems pretty straight-forward, but let me know if I overlooked something or if you don't think such automation is desirable. If it seems OK, I'll do the same with. Rod (A. Smith) 18:17, 28 June 2007 (UTC)


 * Looks great to me. (somehow I missed these posts before, sorry).  -- Visviva 09:38, 7 July 2007 (UTC)

Stem2
If the optional parameter <tt>stem2</tt> is omitted, its value is assumed to be the same as <tt>stem1</tt>. It seems, though, that <tt>stem2</tt> is rarely the same as stem1 (i.e. only when <tt>stem1</tt> ends in ㅏ or ㅓ). It's much more frequently <tt>stem1</tt>+어. Should we make <tt>stem2</tt> default to <tt>stem1</tt>+어 or perhaps make <tt>stem2</tt> a required parameter? Rod (A. Smith) 18:25, 28 June 2007 (UTC)
 * In practice, I daresay stem2 will usually be required no matter what we do; there are far too many irregularities involved. Where the final vowel in stem1 is 아 or 오, the thematic vowel (can I use that in talking about Korean?) is usually 아 (감아, 닮아); but then again it is often assimilated into 와 or some such... or where the thematic vowel is 어 this may also be assimilated into 워, etc.
 * I guess it seemed most intuitive to me that where a verb has only one stem form (e.g. 가다, 가, 가니), only one stem form would need to be entered; that was why I structured it that way. If this doesn't suit, I would lean towards just making stem2 mandatory, as you suggest. -- Visviva 09:48, 7 July 2007 (UTC)
 * Ah, you're right. It's more intuitive to drop stem2 only when it matches stem1, i.e. it's best as is.
 * Also, "thematic vowel" seems applicable for the choice of -어- vs. -아-, but the vowel harmony description seem simple enough: stems whose last syllable is bright (i.e. 아 or 오) take a bright initial suffix when there is a choice. Rod (A. Smith) 18:22, 7 July 2007 (UTC)

Stem2 and Stem3
You should have named them oppositely… Linguists including Rokuro Kono and A. A. Kholodovich have long been calling them stem III and stem II. For example, in the case of 돕다, 돕- is stem I, 도우- is stem II, and 도와- is stem III. The following pages may be interesting: - T AKASUGI Shinji (talk) 05:41, 10 December 2009 (UTC)
 * C HO Eui-sung (2000), 현대 한국어 용언 어간의 조정에 대하여.
 * C HO Eui-sung (2000), 어기.

Honorific verbs
As a learner of Korean, I'd like to suggest some improvement in Template:ko-conj-verb. Now I've learnt that there are verbs which are used as honorifics only and therefore should not be conjugated like normal verbs. However, the template's current set-up doesn't reflect this, as I cannot safely produce 계세요 (the usual polite form of 계시다). Perhaps the "honorific" parameter should be modified, where a 'yes' answer will open only the honorific rows of the table. – Fanatix (talk) 06:19, 22 October 2012 (UTC)
 * It's OK; I have created a new template for this: Template:ko-conj-verb-sida – Fanatix (talk) 11:11, 26 October 2012 (UTC)