Template talk:izh-decl/kana

@BengkelBerkah05, @Thadh, I suggest to consider consonant gradation as "2-step gradation + gemination" instead of "3-step gradation" — at least this is how Mehmet Muslimov describes this, but I've seen this approach elsewhere, though can't share a reference off the top of my head. In this case for kana it would be "no gradation, gemination".
Also a good description for kana-type words is given in Junus:

Täi viisii painuttaijaat (saavat monikon pohjaa a-n sihalle oi):

a) a-loppuiset kakslogaiset sanat, kummiis ensimäisees slogas ono a, i, e.

b) Monislogaiset a-loppuiset sanat.

I.e. these are 2-syllable words with a/i/e in the first syllable or 3+ syllable words. This is similar to how Finnish language textbooks put it for the very beginners. We could see later if this definition should be refined. --KirillW (talk) 18:51, 28 April 2021 (UTC)

I like the definition, but have one question: do polysyllabic words with front harmony also use the kana declension, or still the koira?

I'm not so sure what to do with gemination. I suspect it's regular (so only k,p,t,m,n,l (is that it?) geminate, but do so always in this declension), and if that's true, we could omit this information and just autogenerate it with something like Bengkel did at {{krl-grad-form}}. Thadh (talk) 19:27, 28 April 2021 (UTC)

@Thadh: I was thinking about omitting gemination information, because it is supposed to be quite consistent, but is still quite a difficult question (see "Two phonological rarities in Ingrian dialects/Две фонологические редкости в диалектах ижорского языка" for details). Overall, it's the weak grade that doesn't geminate. But I will check with Mehmet on this and on ä-words. --KirillW (talk) 19:58, 28 April 2021 (UTC)

@Thadh, Mehmet has confirmed that this "a>o" rule looks correct. He mentioned though, that it varies among informants, but that's expected.

I still don't have a good answer on gemination. Let's follow this up. --KirillW (talk) 19:07, 1 May 2021 (UTC)

On gemination a good source should be Laanest (1986) and his thesis. I haven't checked it yet, but I'm going to. --KirillW (talk) 20:57, 22 May 2021 (UTC)