Module talk:fi-IPA

j and v after diphthongs and "gn"

Could I get consensus to do the following changes:

1. "gn" -> /ŋn/, like in words "signaali" and "magneetti". (Are there exceptions?)

   Yes, "diagnoosi" is pronounced with /gn/ and not /ŋn/. 93.106.3.188 18:55, 28 May 2019 (UTC)

   I'd argue the opposite, albeit /ŋn/ is probably a faster and less formal pronunciation. — surjection ⟨?⟩ 21:06, 28 May 2019 (UTC)

   Well, I can't myself remember hearing anyone saying /diɑŋnoːsi/ (but it may of course be that I've not always listened carefully enough) but I do remember a friend of mine from Vantaa saying /siɡnɑːli/ or /mɑɡneːtːi/. I can't remember for sure which one and in which grammatical case, but she did once pronounce one of those two lexemes (signaali or magneetti) with a /ɡn/. 93.106.3.188 08:12, 29 May 2019 (UTC)

   Weird, because when I look up how the word is pronounced in various YouTube videos for instance, in none of them can I hear a clear plosive /ɡ/ and for me it sounds more like how it is in magneetti. — surjection ⟨?⟩ 08:21, 29 May 2019 (UTC)

   @Surjection I second 93.106.3.188. "Diagnoosi" is without the preceding "n" as well as "prognoosi", "stagnaatio", and probably some others. You can check here [1] if you like. As "gn" would be expected, they say nothing about pronunciation under "diagnoosi" but in the entry for "signaali" they mention that it is usually "ngn". Thus "gn" would be acceptable even there. --Hekaheka (talk) 23:01, 30 May 2019 (UTC)

   Well, if even Kotus disagrees with me, so be it. I'll probably remove the <gn> -> <ŋn> and make it manual. — surjection ⟨?⟩ 07:45, 31 May 2019 (UTC)

2. Make j /j/ and v /ʋ/ after their respective diphthongs: j is long /jː/ after diphthongs ending in i, while v is long /ʋː/ after diphthongs ending in u. (Again, are there any exceptions?)

SURJECTION _{·talk·contr·log·} 20:22, 26 June 2018 (UTC)

The second change should probably only take place if a pronunciation hint isn't directly given, or at least have some way of not triggering it - using the dot perhaps? SURJECTION _{·talk·contr·log·} 20:29, 26 June 2018 (UTC)

I'd note that first applies only to intervocalic gn, while initial gn as in gnostikko is /gn/. --Tropylium (talk) 19:20, 19 May 2019 (UTC)

Raised h

@Surjection: What does [h̝] mean? The raising diacritic means the tongue or the jaw is higher, according to Relative articulation § Raised and lowered on Wikipedia, but [h] is glottal, so the tongue and jaw aren't involved in making the sound. — Eru·tuon 21:02, 11 August 2018 (UTC)

It might be the wrong diacritic; I took it from w:Finnish phonology#Consonants (which could very well have a mistake), which describes it as "...a voiceless fricative [h̝]." SURJECTION _{·talk·contr·log·} 21:05, 11 August 2018 (UTC)

@Surjection: Okay, so apparently it means there is some friction in the mouth, but in an unspecified place. But I don't see that transcription given in the reference for that paragraph, Suomi, Toivanen & Ylitalo (2008:28), so it must be original research or from another source. — Eru·tuon 22:13, 11 August 2018 (UTC)

It might simply be a generalization of all the special cases that come after - in which case, the diacritic would be meaningless on its own, which is probably sufficient grounds to just get rid of it. SURJECTION _{·talk·contr·log·} 22:19, 11 August 2018 (UTC)

* in narrow transcription

Would it be a good idea to convert all instances of * in the narrow transcription to actually show their effect, like the code

		-- handle * in narrow transcription
		IPA = mw.ustring.gsub(IPA, "ˣ(ˌ?)(.[" .. diacritics .. "]?)",
			function (stress, after)
				if consonants:find(mw.ustring.sub(after, 1, 1)) then
					return stress .. after .. long
				else
					return stress .. "ʔ" .. after
				end
			end)		
		IPA = mw.ustring.gsub(IPA, "ˣ$", "(ʔ)")

would do? SURJECTION _{·talk·contr·log·} 19:06, 14 August 2018 (UTC)

(diacritics would contain the used IPA diacritics) SURJECTION _{·talk·contr·log·} 19:07, 14 August 2018 (UTC)

@Surjection: I think that would be a good idea. Could you add some examples to the testcases? — Eru·tuon 19:38, 14 August 2018 (UTC)

Will do, and I'll also add the feature itself. SURJECTION _{·talk·contr·log·} 19:41, 14 August 2018 (UTC)

/ts/

This should not be rendered with dental [t̪] in close transcription, but rather alveolar [t]. --Tropylium (talk) 19:12, 19 May 2019 (UTC)

Fixed. — surjection ⟨?⟩ 21:19, 19 May 2019 (UTC)

"Tertiary stress"

While not in use right now (they are just mapped to secondary stresses), there seems to be a case for a "tertiary stress" in Finnish when it comes to compound words with more than two components. Verkkokielioppi 1.7.2 argues that such words should have secondary stress according to how the word should be divided into sections, such as juhlajumalanpalvelus becoming /ˈjuhlɑˌjumɑlɑnpɑlvelus/ and not ×/ˈjuhlɑjumɑlɑnˌpɑlvelus/ (because it is juhla + jumalanpalvelus, not *juhlajumalan + palvelus). Yet still the component words would have secondary stresses of their own as well as the other component still not being completely unstressed (thus the initial syllable of palvelus would have tertiary stress).

FSI's Conversational Finnish (1987) also states that "the stress on the first syllable of the second element in a compound word [is] slightly stronger than the secondary stress in a multisyllable word", which also would imply that there are two distinct types of secondary stress. Of course, it's possible that the sources or my interpretations of such are simply wrong. — surjection ⟨??⟩ 22:53, 12 April 2020 (UTC)

[w]?

Suomi, Toivanen & Ylitalo (2008, p. 31) argue that <v> is [w] after a diphthong ending in /u/. This seems wrong to me, though - I would classify it, at least based on my own speech, as a voiced bilabial approximant [β̞]. Am I just wrong? — SURJECTION ^{/ T / C / L /} 15:46, 10 January 2023 (UTC)