Wiktionary talk:Finnish inflection types

If you're interested in making more inflection templates, look how I've done: for example, in Template:Fi10a 'Fi' refers to Finnish, '10' to inflection type 10, 'a' to back-vowel version (the other is 'b' for front-vowel version ). Word stem is the first parameter ({{{1}}}), strong grade second, weak grade third, and the long vowel in illative fourth (naturally not needed in this example, because every word ends with 'a' in type 10). Some inflection types may need more parameters, some less. And above all, don't hesitate to correct any errors you may (read: will) encounter!--Jyril 21:41, 5 December 2005 (UTC)Reply

Well, I didn't read this note thoroughly before I made Template:Fi48a. There the parameters are two stems (without the final e as they would be in traditional grammar), as in ote: ote and otte. Naturally, this is a different case, so called inverse gradation (weak grade in the nominative/partitive singular, strong in others), and that's how I justify that creation... The templates for group 7 are like templates you did.

Inverse consonant gradation can happen in groups 32, 33, 34, 35 (are there other words than lämmin? I'd say hapan), 41 and 48. They have common that they end in consonants, except 48 (which ended in k or h in the past).

Also, should these templates be renamed? The English templates are in the form Template:en-noun-reg and Template:en-noun-reg-es. Could something like fi-noun-48-a be better for parsing visually? Of course, it doesn't make much sense before one has looked at the list of types. At least those a/b templates could be merged using an additional parameter for the vowel (a/ä) in the suffix. ¦ hyark digyik 11:48, 6 January 2006 (UTC)Reply

If there is a template naming standard, it should be followed. Also, when the templates seem good enough, they should be subst:ed. Now they're using a metatemplate, which slows down the servers.--Jyril 22:06, 6 January 2006 (UTC)Reply

As the templates also apply to adjectives, numerals, and pronouns, fi-decl- might be better than fi-noun-. An a/ä parameter would be a good idea if it sufficed, but several templates (03, 09, 11, 12, 19, 34) insert an o/ö too, and having to add yet another parameter for that would get too annoying. I thought the expansions of templates were cached; how much does subst really speed things up? I'd rather avoid subst because then the formatting of Template:FiNounCases could no longer be changed. 213.216.199.2 15:55, 13 July 2006 (UTC)Reply

Mostly right. With few exceptions, we do not "subst" main namespace templates here. The main part of speech templates (confusingly also called "inflection" templates) are used after the part of speech header. They show only a few key inflections of the headword and have a naming convention where the part of speech follows the ISO language name. A different set of templates is used to show the complete declension (for nouns, pronouns, and adjectives) or conjugation (for verbs) of the headword. Those templates appear in a separate "====Declension====" or "====Conjugation====" section and have "-decl-" or "-conj-" in the template name. Template:Fi10a is a declension template because it shows not just the key forms but the full declension of the term, so "fi-decl-10a" might be an appropriate name. Does that sound right to you? Rod (A. Smith) 20:32, 15 July 2006 (UTC)Reply

I would like to open discussion about numbering the inflection templates. The dictionary "Nykysuomen sanakirja" (NSK in continuation) is regarded as authoritative source for correct Finnish. NSK lists 85 declensions for nominals and 45 for verbs. Wouldn't it make sense to use the same numbering and inflection models in Wiki? Currently only the first three inflection tables correspond with NSK. I have also combined the front- and back-vowels in a single template in the first three tables. I think the resulting structure is quite easy to use and comprehend. Hekaheka 19:56, 21 January 2007 (UTC)Reply

85 + 45 declensions?! Many of them must be unique like tuhat or mies which probably don't need own templates (though they might be useful in compound nouns, where the only parameter is the first word, cf. Template:Fi42). But I agree, we should use NSK numbering since it is the "authory" on the subject. Furthermore, since I have no access to NSK I had to do a lot of guesswork, so many of the definitions are probably incorrect. If you have examples for all the cases, could you please include them to the project page? (Done! Hekaheka 21:32, 18 February 2007 (UTC))Reply

fi-decl-xx sounds good. Since Template:FiVerbDeclensions is used only on this project page, I took the liberty to rename it to fi-conj. Future verb sub-templates could be named accordingly fi-conj-xx. There is no point to rename the templates, however, before the numbering issue is settled.

We should also decide in which order the parameters are given. A better way would to name them, like luonto -> {{fi-decl-01|r=luon|s=t|w=n|v=o|h=a}} (r = root, s = strong gradation, w = weak, v = last vowel, h = vowel harmony {a|ä}).--Jyril 17:18, 13 February 2007 (UTC)Reply

Yep, that might be a logical order. When I formulated the current formula, I thought that it is easier to remember a formula where the word appears first in its uninflected form. But, if we stop including tables into the articles, it is not anymore necessary to remember the template syntax. Actually the templates become altogether unnecessary in their current formHekaheka 22:16, 18 February 2007 (UTC)Reply

We should also figure out what verb forms will be included in the template; I'd like to include conditional, imperative and potential, but that would mean that the already large table would become over twice as large...--Jyril 17:40, 13 February 2007 (UTC)Reply

Here we might follow NSK. It uses present, imperfect, conditional, imperative, 2nd participe and passive imperfect. The problem we have in our hands is so large that we might as well forget the verbs at this stage, concentrate in the nouns, and return to verbs when we have the nouns in order. Hekaheka 22:16, 18 February 2007 (UTC)Reply

See my ideas here. It might be better if we don't include inflection tables in the articles at all (except for irregular cases). In that case we need to include all possible consonant gradations and back/front vowel versions. This page is already getting too large, so the examples could be located in subpages (e.g. Wiktionary:Finnish declension classes/1 for declension class 1 etc.) Each consonant gradation type requires a subheader ===1-A===, ===1-B===, ===1-C=== and so on.--Jyril 20:08, 18 February 2007 (UTC)Reply

I like the first three, because they are quite clear and readily understandable. Specifying the stem and the deflection type in the article are good ideas, but shouldn't we specify front/back vowel also? That would keep the inflection tables nice and easy. Only problem is that the person writing an article should be able to figure out the stem and the declension type, and it is not always obvious for a "layman". But on the other hand, Finnish is a complicated language and we cannot make it simple. It is always possible to write an article without specifying declension type and leave that to an "expert"Hekaheka 22:16, 18 February 2007 (UTC)Reply

Yes, front/back vowels must have own examples, either by putting them next to each other like on this project page, or by putting them under their own subheaders. Declensing is not simple, but except for a fairly small number of words it is fairly regular. One way is to use for example Excel or OpenOffice Spreadsheet and check if the resulting word is sensible. But first of all, we need correct declensions for all those words...

I'll create the template, and add an option that the "declension type #" doesn't show up if it has not been defined.--Jyril 01:29, 19 February 2007 (UTC)Reply

Hi! I went through the model noun inflections and noticed that a couple of them are really superfluous. In particular, tiili is inflected exactly like hiili, uni like tuli, and valmis like kauris. For clearness' sake, I would suggest removing one of each pair. At the moment e.g. hiili is being linked to from kivihiili and aktiivihiili, while tiili is being linked from tiili, radiohiili, and tulitiili. Is there any preference on which to leave? Are these words derived from somewhere or just chosen randomly by one of you guys? Malhonen 09:58, 16 December 2007 (UTC)Reply

You've spotted errors: tiilten (23) is not a valid form according to KOTUS although hiilten/unten (24) is. Kauris and valmis are indeed the same.--Jyril 13:24, 18 December 2007 (UTC)Reply

That's interesting. Is tiilten a new analogical formation then? It does occur even in formal texts (, e.g. see hits on the city of Lahti and Tilastokeskus), and doesn't sound bad to my ear. I'm not saying I know better than Kotus, just wondering... Tulten occurs already in 1907 (Dityrambeja by Aarni Kouta). Malhonen 22:21, 18 December 2007 (UTC)Reply

Actually, revontulten is almost as common as revontulien. Malhonen 22:25, 18 December 2007 (UTC)Reply

But on the other hand, tulten is much less common than tulien. It may be an outdated form. Too bad I don't have Nykysuomen sanakirja available so I can't check if tulten has ever been a valid form.--Jyril 23:14, 18 December 2007 (UTC)Reply

Ok. Here's another question: Are the model inflections supposed to be there for every existing declination/consonant gradation/vowel harmony combination? There's for instance vety, which belongs to declination 1, has consonant gradation t > d and front vowel harmony. At the moment its declension is linked to that of lyhty, which is otherwise the same but with consonant gradation ht > hd. Mechanically it produces the same result (t > d), but traditionally t > d between vowels is considered a different gradation from ht > hd. For the back vowel variant, there's a model inflection for katu (t > d) but not for rahtu (ht > hd). Should I add the missing combinations with whatever sample word I come up with, or is there a better way, or a good reason not to do that? Malhonen 06:48, 22 December 2007 (UTC)Reply

1) Yes, that's the idea. All words that have different endings should have own examples. This because if you replace the "stem" in the example with the one in {{fi-noun}} you get immediately correct inflections without needing to worry about vowel harmonies, correct vowels etc. 2) But how ht → hd is different from t → d? If they're identical, there is no reason to add different examples.--Jyril 13:18, 22 December 2007 (UTC)Reply

Well, the short answer is that that's just the way all the grammars are written, i.e. the source and the target forms of consonant gradation are taken to be the whole consonant cluster between vowels. This way, for instance, lp > lv (halpa > halvan) is considered different from single consonant p > v (lapa > lavan), although they both produce the same result. Actually, even kk > k (kukka > kukan) and k > Ø (käki > käen) are the same from the computer's point of view... The longer answer tells that ht and single t are often handled differently in Finnish dialects, e.g. modern Helsinki colloquial has vetää > vedän, but lähteä lähtee > lähen. Malhonen 14:34, 22 December 2007 (UTC)Reply

Hm. Considered by whom? Interesting point about kk: k. But how about rakka → rakan, raaka → raa'an? Looks like we're missing the apostrophed words... Forget dialectal and colloquial inflections, they're so unstable that we'll never get them right. Only the most basic non-standard inflections for the most important words are worth including (e.g. mun, sun etc.)--Jyril 20:41, 22 December 2007 (UTC)Reply

By people writing Finnish grammar for more than a century now... Take a look at any elementary Finnish textbook or grammar, and you will see something like this. The first and pedagogical reason is that it's easier for learners to remember that "t between vowels always becomes d, ht becomes hd, st doesn't change, lt becomes ll etc." than to say that "t changes to d except if it is preceded by s, l, r, n, or another t, in which case etc. etc." The second and linguistic reason is that all of these different consonant clusters tend to change independently of each other in diachronic evolution. That's why I mentioned Helsinki colloquial: it happens there now, and has happened many times before and will still happen in the future. In Finnish literary language we have a remnant of the same distribution—there is kato- > kadottaa, but lähte- > lähettää, not *lähdettää. In Olonets Karelian the distribution is more or less the opposite: they have kodi : kois (Finnish koti), but kohtu : kohtas (Finnish kohta). So, even if we describe just standard literary Finnish, in the end it's just historical coincidence why t between vowels behaves the same as ht. Malhonen 17:37, 30 December 2007 (UTC)Reply

I see. Although this is interesting from etymological point of view, let's not make the inflection table any more complicated than it already is.--Jyril 18:01, 30 December 2007 (UTC)Reply

Ok. That's fine with me. It just didn't even come to my mind, that you might have such a system here. I went through the noun inflections and compiled a comprehensive table at Wiktionary:Finnish inflection templates. I added there some noun inflections which are still missing with mention (missing). An example word of all of them can be found by editing the source. According to Kotus (see the link you gave above), the declination you have numbered 15 should actually be divided in two: there's harakka, which inflects harakoihin or harakkoihin in ill.pl. (Kotus list 14 solakka), and sitruuna which inflects just sitruunoihin (Kotus list 13 katiska). The references to the model inflections are good, since all words ending in -kka/-ppa/-tta/-kkä inflect the same (like harakka). It's just that the models harakka, ulappa, navetta and kännykkä should be linked to something else than fi-decl-15. Malhonen 07:43, 3 January 2008 (UTC)Reply

Shouldn't these all be in the Appendix: namespace? Conrad.Irwin 13:54, 15 April 2008 (UTC)Reply