Help talk:IPA/Swedish

Diphthong /œɪ/

Are there any sources that /œɪ/ is a diphthong and not just /œj/, because that’s the transcription I would assume. --Lundgren8 (t · c) 22:43, 12 December 2016 (UTC)

@Lundgren8: I don't have any sources, but Swedish /ɪ/ is very front and IMO is better transcribed as /i/. The phonetic realization of /œɪ/ would actually be [œi̯]. So the question is: can native speakers turn the second element into a fricative [ʝ]? If so, it should be transcribed /œj/. If that's not the case, either transcription is fine. Mr KEBAB (talk) 23:21, 5 April 2017 (UTC)

@Mr KEBAB:/@Kbb2: Frication to [ʝ] normally only occurs in stressed onset position and is unlikely to occur as a coda, so it’s not a good test. I would like to remove this from the chart if there are no sources backing this up as a diphthong. --Lundgren8 (t · c) 17:12, 18 January 2020 (UTC)

That's a bad test. Even if the ending point of the supposed [œɪ] diphthong was different from coda /j/ that occurs after other vowels, that's still not a proof that it's a diphthong and not a mere /œj/ sequence. In Danish, the closeness of coda /j/ depends on the openness of the preceding vowel, so that it's closer to [e̯] after open vowels and closer to [i̯] the higher the preceding vowel is. It can even be fricated (as in Swedish) when it occurs in the word-final position after a consonant. Whether something like [œj] is counted as a diphthong or not depends solely on the phonology of any given language. Plus, even /œj/ is a valid transcription of a phonological diphthong, believe it or not. In Swedish, it is a /VC/ (vowel + consonant) sequence. Same with affricates - whether [ts] is a phonological affricate or not depends on the phonology of the language.

In Urban East Norwegian, there are compelling reasons to consider /æʉ/ a diphthong and not a /VC/ sequence. First, in order to consider it a /VC/ sequence, you need to posit an additional /w/ phoneme, which corresponds to the central /ʉ/ (not the back /ʊ/). Apparently it's a labial-postpalatal approximant, or something like that. The evidence for the phonemic status of /w/ is circular: it exists in the underlying representation of [æʉ] according to Kristoffersen (who considers it to be /ɛw/, with the vowel being allophonically lowered before /w/), making [æʉ] a /VC/ sequence in phonology, rather than a diphthong. This allows for the analysis of the remaining two native diphthongs /œʏ/ and /æɪ/ as /œj/ and /ɛj/, where /j/ is allophonically labialized after /œ/ and /ɛ/ lowered before /j/ (I've never heard of such a rule in Norwegian-Swedish dialectology, but anyway). However, UEN /œʏ/ and /æɪ/ are monophthongized to /øː/ and /eː/ in a number of non-UEN dialects (e.g. Trøndersk), which is far more consistent with monophthongization of historical diphthongs, rather than anything else. In addition, /æʉ/ monophthongizes to /øː/ as well, resulting in a three-way merger of /œʏ/, /æʉ/ and /øː/. For that reason, /æʉ, œʏ, æɪ/ are best considered diphthongs in UEN. In Swedish, there's AFAIK little reason to consider /œj/ anything other than a /VC/ sequence. Sol505000 (talk) 19:12, 24 August 2021 (UTC)

Schweiz and Reuter rare

We don’t have any native words spelt with ei or eu after each other so I can agree that the spelling is rare, although the sounds I would argue aren’t as. Both w and j are letters, really only describing the consonants in the ao and ai/ei/oi diphtongs. Since j is a letter and sound very much used in Swedish /eɪ/ and /oi/ = /ej/ and /oj/ which we already have and they aren’t rare sounds either. It would be like thinking ”Thai” is a word with rare sounds in English. But since there are plenty of words spelt with ay (ay, buy, my, might, site etc.) it is not, as far as I know, considered as a word with rare sound. Can I remove the /eɪ/ and /oɪ/ as rare sound? Since we don’t have the consonant w, Wales DOES contain a rare sound and should thus remain.Jonteemil (talk) 01:48, 12 January 2019 (UTC)

To me, the supposed diphthongal vowels /eɪ/ and /oɪ/ are as dubious as the supposed diphthong /œɪ/. All of them look like /Vj/ to me as far as the phonology is concerned, making them a sequence of a short vowel followed by a consonant (most likely /ɛj, ɔj, œj/). Sol505000 (talk) 23:09, 24 August 2021 (UTC)

Secondary stress

@IvanScrooge98: Regarding this edit: where else would you add the secondary stress mark if not before a syllable that, in fact, has secondary stress according to reputable sources? Kbb2 (ex. Mr KEBAB) (talk) 10:20, 12 September 2019 (UTC)

@Kbb2: what I meant, and you know it, is that one may think there is always a secondary stress in words with toneme 2, which is not true, or that they should always have a secondary stress mark, as I used to think too. Please, we don’t need to discuss it because we have already been through this topic. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 11:04, 12 September 2019 (UTC)

And WP:COMMONSENSE has very little, if anything, to do with my edition. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 11:11, 12 September 2019 (UTC)

@IvanScrooge98: That's what you thought. I'm not buying that's what your average reader will think.

We do need to discuss this. [1] gives off WP:OWNERSHIP vibes, which is a no-no on Wikipedia.

It does. [2] isn't a clarification, it's stating the obvious. "[W]hen more than one syllable follows" means that secondary stress is possible only in words with three or more syllables.

This also isn't the best wording as it suggests that wherever you see ⟨ˌ⟩ it means that the secondary stress is optional. That's not how we should use that sign. Kbb2 (ex. Mr KEBAB) (talk) 11:26, 12 September 2019 (UTC)

"[S]econdary stress is possible only in words with three or more syllables." may be BS though: försal [²fœːˌʂɑːl] "entrance hall" and spärrnät [²spærːˌnɛːt] both have secondary stress, per Swedish phonology. Kbb2 (ex. Mr KEBAB) (talk) 11:32, 12 September 2019 (UTC)

Duh, I’m claiming no ownership of anything, I simply stated that I knew the reason why I had added that and what I meant by it. And I still believe we should be clearer as it could be misinterpreted. I could reply to other stuff you are disputing, but I am just tired of continuously discussing every single thing here. Do as you like and let’s stop it here. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 11:38, 12 September 2019 (UTC)

@IvanScrooge98: Then you should've stated that reason in the WP:EDITSUMMARY. The function of edit summaries is to summarize your edits to other users. If you write them like this one it's like not providing an edit summary at all. It's supposed to be meaningful. That, again, is WP:COMMONSENSE.

This wording shouldn't be misinterpreted.

Nobody has the luxury of having their edits go unchallenged on WP (or at least expecting them not to be challenged). That includes you, me and all other editors. Kbb2 (ex. Mr KEBAB) (talk) 11:52, 12 September 2019 (UTC)

I stated that in this edit summary: to clarify; but you didn’t agree there was any need of clarification, I tried to edit differently but it resulted into a worse wording. And here we are. But I won’t go on any further. Keep the page as you like it, with an unclear consensus behind when we should have secondary stress marks in Swedish transcriptions. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話）

@IvanScrooge98: On second thought, I can see why For words with the second toneme, ⟨²⟩ will be used instead of the primary stress mark, and ⟨ˌ⟩ to indicate the secondary stress when more than one syllable follows could've been understood the way you did. I apologize if my reference to WP:COMMONSENSE was uncivil because your interpretation wasn't as illogical as I thought.

But we've already established that the previous wording was wrong. Per Swedish phonology, försal [²fœːˌʂɑːl] "entrance hall" and spärrnät [²spærːˌnɛːt] both have secondary stress (that is, if those transcriptions are correct). Also, neither of us has delved into reputable sources to actually find out when we should use the secondary stress mark. That's where our answer is. We're not supposed to build a pseudo-consensus (which is not a consensus at all, actually) based on assumptions and guesses. We're not even native speakers of Swedish. Kbb2 (ex. Mr KEBAB) (talk) 12:07, 12 September 2019 (UTC)

@Kbb2: (that is, if those transcriptions are correct); that is the point and one of the things in the discussion of which I didn’t want to plunge into. Is toneme 2 closely related to secondary stressing, as I imagine? If so, how do the sources transcribe it and how do we want to transcribe it? I mean, are [²fœːˌʂɑːl], [²spærːˌnɛːt] etc. transcribed with ⟨ˌ⟩ because that is how the sources deal with the syllables that follow the primarily stressed one (in which case it would be superfluous to mark it for bisyllables, as I thought we were doing), or is secondary stress unrelated to tonemes? As you said, neither of us can deal, but since we realize there is an issue, we should try to build consensus anyway, which I could not do completely because I do not know any Swedish users on the English Wikipedia that may help us with the topic. Again, I tried to have as much common sense as I could, based on what I could see on the project. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 12:27, 12 September 2019 (UTC)

@IvanScrooge98: I've already told you that this is too hard for me, as it clearly is for you. We need WP:COMPETENT users to deal with this, based on WP:RS and not assumptions. We're out of our depth, how many times do I need to say this?

WP:RS must be the foundation of any consensus here, which should be build by editors that are knowledgeable enough to know what they're doing. The correct course of action for you seems to be to stop transcribing Swedish into IPA. Kbb2 (ex. Mr KEBAB) (talk) 12:37, 12 September 2019 (UTC)

For what it's worth, the secondary stress found in Urban East Norwegian has no bearing on the phonetic realization of tonemes. It's likely the same in Swedish. And it's important to indicate it where it's appropriate. Unstressed mid front vowels don't lower like the stressed ones do and there's no gemination in unstressed syllables. But it's not a terribly big deal either. Wherever you indicate the lowering of mid front vowels or gemination, it clearly points to the fact that the syllable bears some stress (and so does the presence of long vowels, AFAIK). It'd be a bigger deal not to indicate secondary stress in phonemic transcription, which would make some compound nouns look non-compound. Sol505000 (talk) 09:12, 25 August 2021 (UTC)

stress on monosyllabic names

I started adding stress to Karl, Lars, Hans and Jans, when transcribed as a separate word, but some of these might be pronounced as compound personal names (like the Carl Gustaf recoilless rifle) with no stress on the first. If so, they shouldn't be written as separate words, and presumably retain at least secondary stress, so something's wrong. Anyone know? — kwami (talk) 07:49, 11 December 2019 (UTC)

BTW, '1' and '2' aren't IPA. For '1', the normal stress mark is sufficient. For '2', there's an old IPA convention for Swedish compound tone that we can use. — kwami (talk) 07:49, 11 December 2019 (UTC)

Where's the consensus for this? I thought we used ⟨²⟩ to avoid a bias towards a particular dialect. ⟨¹, ²⟩ are used by Riad (2014). Also, why adopt a recommendation from 1900 when newer publications by the IPA recommend different symbols (⟨ˆ⟩ in 1921 and "arbitrarily chosen mark" like ⟨˟⟩ or ⟨ˇ⟩ in 1949) for accent 2? Nardog (talk) 17:42, 11 December 2019 (UTC)

Those sources don't discuss Norwegian, but indeed are better for here.

The consensus is that we use IPA for IPA. [1] and [2] aren't IPA and have no set value. Using them is like using local conventions for tone in Chinese or other languages, despite the fact that the digits are internationally ambiguous. Digits work fine for Riad, since Riad is only dealing with Swedish and can define the symbols any way they like, just as Marshallese vowels have been transcribed /♠ ♣ ♥ ♦/, but they don't work for us, where [2] could be high tone, low tone, mid tone, tone #2 or something else entirely. And here our transcriptions are supposedly phonetic, so should use phonetic symbols. (And Swedish is already weird in using <ɧ>, which is a special symbol with no specific phonetic value just for Swedish.) This is a problem when not transcribing a specific pronunciation, but at least these symbols have some definition within the IPA, unlike digits. — kwami (talk) 08:37, 12 December 2019 (UTC)

Can you point to the discussion or policy/guideline where that consensus is made explicit? And where's the consensus that ⟨ˇ⟩, specifically, be used for Swedish accent 2? Also, I have news for you: ⟨ˇ⟩ isn't IPA either. Only changes to the alphabet or chart that have been approved by the Council can be considered part of the official IPA. One could argue the 1949 Principles was Council-approved, but so was the 1999 Handbook, which superseded Principles and does not recommend any symbol for accent 2. By your logic we might as well start using ⟨ɼ⟩ for [r̝], which used to be IPA-approved, or start eliminating ⟨ʱ⟩, which does not and has never appeared on the chart. You're clearly being selective as to what counts as IPA and what does not—which is fine, but what gives you the prerogative to act upon it and edit ~1.5K pages without prior consultation with the community?

@Kbb2, Jonteemil, Lee Vilenski, IvanScrooge98, Ardalazzagal, Jeppiz, Surfo, Lundgren8, Aeusoes1, and Gilgamesh~enwiki: (1) Do we condone this change? (2) Do we condone the way the change was implemented, i.e. unilaterally? Can I go around replacing, say, ⟨ʱ⟩ on the grounds that "we use IPA for IPA" with no consequences (not that I would)? Nardog (talk) 02:45, 16 December 2019 (UTC)

(Ping reply.) I have no particular thoughts on this matter. - Gilgamesh (talk) 04:02, 16 December 2019 (UTC)

@Nardog: Norwegian should use ⟨V̀⟩ and ⟨V̂⟩ instead and the IPA should be limited to Urban East Norwegian (hence ⟨V̀⟩ and ⟨V̂⟩) unless we have a RS that provides the local pron (or an editor that knows what they're doing - many don't and don't provide stress, tonemes, vowel length, etc.). When it comes to Swedish, I prefer ⟨¹⟩ and ⟨²⟩ because regional varieties of Swedish are much more similar to one another (because they're based on written Swedish) than Norwegian "dialects" that could just as well be classified as closely related languages. Using ⟨ˈ⟩ in the tonemic system might suggest that it's fine to provide stress instead of tonemes in the case of Swedish, and it's not (not when we're dealing with Swedish Swedish). Kbb2 (ex. Mr KEBAB) (talk) 09:05, 16 December 2019 (UTC)

If the transcriptions were phonemic and if the help page was meant to represent Swedish as a whole, I would prefer superscript 1/2 for sure. But since it states that the transcriptions are in Central Swedish, I suppose it’s fine to use Central Swedish tonal accents as well. The transcriptions also use retroflex consonants for instance, which is another geographically specific non-phonemic process, and consonant length, which is also non-phonemic. I do however think that the accent marks should be placed on top of the stressed vowel. It’s a tonal marker and not a stress sign. --Lundgren8 (t · c) 15:36, 18 December 2019 (UTC)

@Lundgren8: Thanks for the reply (and apologies for the belated response). The current transcription of retroflexion does indeed make a strong case for basing the whole thing on Central Swedish. If we were to use the tone diacritics, how would "the duck", "the spirit", and Oxenstierna be transcribed? Nardog (talk) 18:19, 12 January 2020 (UTC)

@Nardog: (1) Consensus is against these changes, and Kwamikagami's differing opinion does not change the consensus. (2) Like any other changes that go against consensus, these, too, must be reverted. And that can be done unilaterally. Ardalazzagal (talk) 20:23, 1 January 2020 (UTC)

I have raised the issue regarding the changes at ANI, so the way they were introduced and implemented may be discussed there. Discussion on what symbols to use to mark tone/stress in Swedish transcriptions are still welcome here. Nardog (talk) 06:48, 2 January 2020 (UTC)

We have Lundgren8's POV that we should use the tone diacritics for Central Swedish, since everything else is Central Swedish. That would be fine, since it follows the IPA conventions that we claim to follow. '1' and '2' are not acceptable for what claims to be IPA, since they mean nothing in that context. If we're going to use digits, then we need to change the template and key to no longer claim we're using IPA. The fact that some sources on Swedish violate the IPA is not reason for us to do so as well, since unlike them WP does not transcribe only Swedish. If we're going to use IPA tone diacritics, what should they be? — kwami (talk) 02:09, 8 January 2020 (UTC)

But, again, we're not removing ⟨ʱ⟩ from transcriptions of South Asian languages just because it's not part of the official IPA. I beg you to refrain from making undiscussed blanket changes like this in the future. Nardog (talk) 18:54, 12 January 2020 (UTC)

I’m not a prosodist, therefore I’m a bit wary giving a view that is simplified at best and incorrect at worst. For Central Swedish, I think the help page is correct that accent 1 is characterised by a rising-falling contour. We could write this either as [ˈa᷈ndɛn] or [ˈǎndɛ̂n] I suppose. It might be simpler and easier to write to mark it stressed syllable only, and it’s possibly more correct if pitch accent is restricted to stressed syllables? On the other hand, what’s characteristic about accent 2 is the double-peak accent, and I guess ”falling-falling” works as a description. There is no falling-falling diacritic however, and I think the double-peak usually spills over outside of the stressed syllable as well, so in that case we would need to write [ˈândɛ̂n], since only a falling diacritic *[ˈândɛn] would miss the characteristic double peak.

In compounds, I believe the second peak is delayed to the syllable which has the last secondary stress of a word, so a word like ˈsommarˌledigˌheten (Riad’s favourite example) has a high tone (or falling I guess?) on the first syllable, and then low tones on all the syllables up to -heten which is has a rising-falling tone like a normal accent 1 word, I guess this could be written something like [ˈsɔ̂mːarˌleːdɪˌhe᷈ːtɛn]? Riad himself writes H* <-- L*H L% with the ToBI system, where H* is the high tone of the main stress, <-- is a plateau of low tones spread leftwards from L*H, which is the accent of -heten with a rising tone, followed by the boundary tone L% at the end of the word, which seems equivalent to rising-falling in words pronounced in isolation (with nothing following them). --Lundgren8 (t · c) 14:47, 22 January 2020 (UTC)

@Lundgren8: [ˈânːdɛ̂n] for accent 2 sounds good, but [ˈa᷈nːdɛn] for accent 1 makes little sense to me given the falling doesn't happen in the syllable with the diacritic. So I think [ˈǎnːdɛ̂n] makes more sense, although [ˈǎnːdɛn] might be better for the sake of simplicity. Do you think stressed monosyllables should also be transcribed with a diacritic for accent 1? Also, what do you think of Engstrand's transcription included in the IPA Handbook? If I understand correctly, he's attaching the grave to the first vowels with accent 2, which I guess makes sense if one wants to transcribe prosody with as few symbols as possible (it feels like it should be the circumflex rather than the grave though).

Pinging @Kbb2, Jonteemil, Lee Vilenski, IvanScrooge98, Ardalazzagal, Jeppiz, Surfo, Aeusoes1, Erutuon, LiliCharlie, J. 'mach' wust, Peter Isotalo, and Glades12: for input. Nardog (talk) 11:32, 23 January 2020 (UTC)

Sorry, but I don't think I have much to add here. Being a speaker of a language doesn't mean that you know all the intricacies of its phonology. Glades12 (talk) 12:44, 23 January 2020 (UTC)

There's nothing to apologize for! Nardog (talk) 13:06, 23 January 2020 (UTC)

The tone contour of unstressed syllables is certainly predictable and can be dispensed with in a phonemic notation. Auditorily however it is very conspicuous and should be written in a notation that tries to reflect the actual speech sounds, I think. — I don't see the need for transcribing level (mid) tone, as Swedish is a contour tone language rather than a register tone language. And I don't see the benefit of marking tone of monosyllabic words either. Love —LiliCharlie (talk) 18:52, 23 January 2020 (UTC)

@LiliCharlie: How do you think we should be transcribing stressed polysyllables then? Nardog (talk) 19:41, 23 January 2020 (UTC)

Good question. There was a time I wrote [ˈ$$] for accent 1, and [ˈ$ˌ$] for accent 2, where $ means syllable. This also shows that monosyllables can't have contrasting accents. I'm not sure if anyone else has used such a notation that doesn't rely on tone at all, but on perceived stress levels. Love —LiliCharlie (talk) 20:32, 23 January 2020 (UTC)

Wiktionary previously used this system. It’s inaccurate since IPA ⟨ˌ⟩ is reserved for secondary stress, and only compounded accent 2 words carry secondary stress. --Lundgren8 (t · c) 02:37, 24 January 2020 (UTC)

So you mean a symbol for tertiary stress (or maybe "secondary weak stress") would be accurate if there was one? Love —LiliCharlie (talk) 18:43, 24 January 2020 (UTC)

@LiliCharlie: But fader [ˈfɑ̂ːdɛr] can't be /ˈfɑːˌdɛr/ in phonology. The /ɛ/ isn't lowered before /r/ (which is suggested by the phonemic transcription /ˈfɑːˌdɛr/; it suggests a compound noun [ˈfɑ̂ːˌdærː]) due to the unstressed nature of the syllable. Per Riad 2014, CSS doesn't have a schwa, but even if it had one, the transcription /ˈfɑːˌdər/ would be just as bad (if not actually worse) because the schwa is said to be unstressed-only even by those authors who consider it to be a part of CSS. Sol505000 (talk) 07:25, 25 August 2021 (UTC)

@Nardog: I don’t think Engstrand’s system is phonetic. He’s using a grave accent since ”grave accent” is a common synonym for accent 2 in Swedish. That is to say it’s equivalent to writing a superscript 2. --Lundgren8 (t · c) 19:53, 23 January 2020 (UTC)

@Lundgren8: I'm trying to figure out what we can realistically convert the existing transcriptions to and realistically tell editors and readers to use or interpret. I'm leaning toward [ˈǎnːdɛn] for accent 1 and [ˈânːdɛn] for accent 2 as these would require the minimum number of additional symbols for people to keep track of, free us from having to figure out where the second peak occurs in compounds, and achieve a harmony with Norwegian transcriptions if we used ⟨◌̀, ◌̂⟩ in them as Kbb2 suggested (also the second diacritic would always be falling regardless of the accent type if we marked it, which renders it sort of redundant). Is this a bad idea? (Also I'd like to know what you think we should do with monosyllables.) Nardog (talk) 22:20, 25 January 2020 (UTC)

@Nardog: I'm fine with using ⟨◌̀, ◌̂⟩ for both Norwegian and Swedish. AFAICS, the stress marks are redundant in the case of (Urban East) Norwegian (Trondheimsfjorden would be pronounced [trɔ̂nhæɪmsfjúːɳ] even if the second syllable had secondary stress), which mirrors the situation with Serbo-Croatian. That frees us to write fast as [fɑst], without either a stress mark or a tone diacritic. I don't know whether it's the same with Swedish. In UEN I'd simply write [bœ̀nnəɾ, bœ̂nnəɾ, fɑst]. Kbb2 (ex. Mr KEBAB) (talk) 23:32, 25 January 2020 (UTC)

@Kbb2: For Swedish accent 1 you mean ⟨◌̌⟩, not ⟨◌̀⟩, right? Nardog (talk) 23:39, 25 January 2020 (UTC)

@Nardog: Yes, my bad ([ˈǎnːdɛn, ˈânːdɛn] - so perhaps it's better to write [bœ̀nːər, bœ̂nːər, fɑst] (still without the stress mark, which is redundant in Norwegian) in the case of UEN in order to achieve harmony between transcriptions of the two languages). Kbb2 (ex. Mr KEBAB) (talk) 08:31, 26 January 2020 (UTC)

@Nardog: Oops, I forgot that Norwegian and Swedish, unlike Serbo-Croatian (standard SC at least) allow final stress on polysyllabic words. In that case forget what I said about the stress marks (I'd write [ˈǎnːdɛn, ˈânːdɛn, ˈfasːt] in Swedish and [ˈbœ̀nːər, ˈbœ̂nːər, ˈfɑsːt] in Norwegian - the /s/ is phonetically long by the way). Kbb2 (ex. Mr KEBAB) (talk) 13:05, 26 January 2020 (UTC)

I'll implement this in a few days if no one has a problem. Nardog (talk) 16:55, 31 January 2020 (UTC)

I will mainly paraphrase Glades12, as I cannot really add much here. Being a native speaker of Swedish, I must admit there are at least 40-50 languages I've read more about than Swedish. Swedish never had that exotic lure ;-) While still probably more competent to discuss Swedish linguistics that most fellow non-linguist Swedes, due to a knowledge of linguistics overall as well as of course having read some 40-50 books on Swedish linguistics (though mainly the development of Swedish dialect), I must admit the specialist conversation here, while certainly good and relevant, and interesting to follow, is outside my competences. Jeppiz (talk) 18:44, 25 January 2020 (UTC)

Transcription of the rhotic

As we know, Swedish /r/ isn't (normally) trilled in CSS. It's more commonly a fricative or an approximant, and Engstrand (1999) transcribes it with ⟨ɹ⟩. We already use that symbol in Faroese phonology and we'll use it in Help:IPA/Faroese, should we ever feel the need to create such a guide. Swedish phonology#Narrow transcription uses ⟨ɹ⟩ a lot, and on the recording there are even more instances of it ([mis]transcribed ⟨ɾ⟩). AFAICS, the approximant is used even where /r/ is geminated. Should we use ⟨ɹ⟩ then? If not, wouldn't ⟨ɾ⟩ still be a better choice than ⟨r⟩? After all, those fricatives and approximants sound like a weakened [ɾ] to me (Spanish /ɾ/ has a similar allophony).

What about Finland Swedish? What kind of an /r/ do speakers of that variety use? Kbb2 (ex. Mr KEBAB) (talk) 08:46, 26 January 2020 (UTC)

Self-response: ⟨ɾ⟩ isn't better because it yields [ˈstâɾːkastɛ] for starkaste, which is strange ([ˈstâɾɾkastɛ] is just as weird, IMO). This is also why we should use ⟨r⟩ in Norwegian. But I'm still open to the possibility of using ⟨ɹ⟩ in this guide. Kbb2 (ex. Mr KEBAB) (talk) 10:11, 26 January 2020 (UTC)

Is the fricative/approximant articulation common even in intervocalic and /Vrj/ positions (in CSS)? I think I always hear a tap in those positions in recordings by Peter Isotalo and PJ (which may not be representative of what we want to represent in our transcription, notwithstanding).

I would be reluctant to switch to ⟨ɹ⟩ if [ɾ] was still common. (If the distribution of [ɹ] and [ɾ~r] was completely complementary (i.e. with no overlap), using both ⟨ɹ⟩ and ⟨ɾ⟩, alternating depending on position, would seem like a plausible option, but that doesn't seem to be the case AFAICT.) Nardog (talk) 12:45, 26 January 2020 (UTC)

One thing you should keep in mind is that both myself and Engstrand are very much Stockholmers. The accent we speak isn't even in the majority here any more. The differences aren't huge, but when it comes to /r/, there's plenty of trills and taps. People who speak "Rinkeby Swedish" use taps and trills almost exclusively as far I can tell. Outside of Stockholm, this varies plenty, even if you exclude the southerners. You're not going to get a useful overview of Swedish realizations of /r/ by looking at only our speech patterns.

For a phoneme that is all over the place, I see no benefit in switching symbols. To me, it seems like one of those really practical conservative compromises worth keeping.

Peter ^Isotalo 20:18, 27 January 2020 (UTC)

I agree with Isotalo, this phoneme has huge variation within Swedish ranging from [r~ɾ~ɹ~ɻ~ʐ~ʁ~ɣ~ʀ~∅] and probably more, depending on the speaker, dialect, position within the word, etc. I think /r/ is a good and neutral symbol for the phoneme. --Lundgren8 (t · c) 12:51, 28 January 2020 (UTC)

[ʁ~ɣ~ʀ] are non-Central Standard Swedish, though. And they most often imply no retroflexion, which we do transcribe. But I agree that the best symbol to cover other, CSS variants is ⟨r⟩. Sol505000 (talk) 08:39, 26 August 2021 (UTC)

Rov example

The example pronunciation of "rov" sounds with a rolling r, not at all similar to the American/Scottish examples, but I have certainly heard Swedes pronounce some R's in a more similar way (also in northern Norway, where it in more extreme cases even seems to have evolved into a "zh" sound). A better example is needed, and it should probably be added that this pronunciation is not ubiquitous (for any word at all?). Elias (talk) 08:12, 20 September 2021 (UTC)

It's nothing unusual for a tapped R to vary with a trilled R (and a weakly fricative R vaguely similar to "English" ZH, which isn't so English as it's used mostly in loanwords) in languages in which the three sounds do not contrast. Sol505000 (talk) 16:59, 2 October 2021 (UTC)

Secondary stress in two-syllable compounds

@Aeusoes1, Nardog, Lundgren8, Peter Isotalo, Kbb2, and Jonteemil: I’ll open a discussion here since Sol505000 seems unwilling to do so and keeps coming at my personal talk. The editor decided to add a secondary stress mark in the transcription for stormvind at Fångad av en stormvind, something I have never seen (and therefore never used) in Swedish IPA transcriptions. I thought it was implied there is no difference between the pronunciation of a bisyllabic compound any that of any other bisyllable with toneme 2 (e.g. kvinna), and therefore no difference in transcription.

Now the questions are, 1. is there an actual difference in realization? 2. is there a source backing this up? (Sol505000 cited transcriptions from Riad 2014, does the source state it clearly?) 3. if there isn’t any difference, should we still transcribe compounds of two syllables with a secondary stress mark? 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 14:13, 24 September 2021 (UTC)

If you've never seen that then I guess you haven't read that much of the relevant literature. Riad (2014) is available on Google Books, you can go there and read the source (specifically the chapter called "Stress") yourself.

I'm not sure about non-compound words. We should follow the literature. But to be honest, Swedish phonology is sorely lacking the information on the phonetic realization of tonemes, especially in Central Standard Swedish. Sol505000 (talk) 14:16, 24 September 2021 (UTC)

Given that there is question about this issue and the Google Books preview is limited, it would be in good order for whoever can access this book to quote the relevant passages to enlighten the rest of us on what the source says. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:44, 24 September 2021 (UTC)

What I meant was that Riad marks secondary stress in all types of compounds, including the disyllabic ones. The secondary stress clearly can immediately follow the primary stress. Given that's what he does, I didn't bother looking for an explanation in prose. There should be one in the book, though. Sol505000 (talk) 15:49, 24 September 2021 (UTC)

The thing is tone and stress are intertwined. As far as I understand, compounding always triggers the use of the second toneme in Swedish, hence my questions above. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 15:58, 24 September 2021 (UTC)

Per Phonology of Norwegian (I'll try looking for the page later), the more weakly stressed member of a compound noun has no impact on the phonetic realization of the pitch accent (apart from the fact that compounding tends to automatically trigger Accent 2 in most types of Swedish, which it *doesn't* in many types of Norwegian). I imagine this is true of all types of Norwegian and Swedish. Sol505000 (talk) 16:02, 24 September 2021 (UTC)

@IvanScrooge98: I thought it was implied there is no difference between the pronunciation of a bisyllabic compound any that of any other bisyllable with toneme 2 (e.g. kvinna), and therefore no difference in transcription. There isn't, as far as I know. How would Sol505000 transcribe for example kvinna and vinglas, a compound? What would be the difference, except the individual phonemes?Jonteemil (talk) 23:20, 24 September 2021 (UTC)

They would transcribe them as [ˈkvɪ̂nːa] and [ˈvîːnˌɡlɑːs], though as I said there is no point not to go for [ˈvîːnɡlɑːs] for the latter unless there is an actual phonetic difference. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 08:42, 25 September 2021 (UTC)

And I say that this makes no sense as compounding automatically results in secondary stress, according to Riad. We still lack any evidence that secondary stress influences pitch accent in any way. In Norwegian, it doesn't. At least in Oslo. This makes the secondary stress a purely phonological marking, the way I see it anyway. Sol505000 (talk) 08:45, 25 September 2021 (UTC)

Then what makes kvinna NOT have secondary stress? I agree with Ivan. If "the secondary stress [is] a purely phonological marking" then that's all the more reason not to include secondary stress after accent 2 at least in disyllables, for our transcription is phonetic. Nardog (talk) 09:18, 25 September 2021 (UTC)

The final short vowel, which proves it is not a compound noun. It's my understanding that secondarily stressed syllables in Swedish have to be build like the primarily stressed ones, i.e. they have to have a long vowel or a long consonant. Also, if the mid front vowels are lowered, that's also a sign of secondary stress, at least in the case of the unrounded vowels. The unstressed instances of /ɛ/ are always mid. Sol505000 (talk) 09:47, 25 September 2021 (UTC)

The final short vowel, which proves it is not a compound noun. You just answered yourself. There is no need to mark the second syllable when it is the second part of a compound, especially since the syllable required to distinguish accent 2 from 1 can only be that one, and there is apparently no phonetic difference between /ˈV̂V/ and /ˈV̂ˌV/. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 10:23, 25 September 2021 (UTC)

I have no big problem with removing the secondary stress mark from our guide and then adjusting transcriptions to match it. What I do have a problem with is inconsistency. Secondary stress in Swedish seems to be literally a long vowel/consonant in a non-primarily-stressed syllable. Either we mark it or we don't. Sol505000 (talk) 10:43, 25 September 2021 (UTC)

But isn't secondary stress necessary to mark where the second falling tone occurs in compounds with more than two syllables? Or is it predictable too? Nardog (talk) 11:20, 25 September 2021 (UTC)

I've got no idea about the phonetics of the Swedish pitch accent. The "double falling tone" description of Accent 2 may be applicable only to disyllabic words. Sol505000 (talk) 11:41, 25 September 2021 (UTC)

Lundgren8 says In compounds, I believe the second peak is delayed to the syllable which has the last secondary stress of a word above. Notations for sommarledigheten appears on p. 128 of The Phonology of Swedish as well as in Riad (2006, 2009) and Riad & Segerup (2008). Nardog (talk) 13:45, 25 September 2021 (UTC)

Right, this probably has something to do with Stockholm Swedish being a two-peaked dialect. I believe Stavanger Norwegian is the same in that regard. Their rendering of sommarledigheten would probably be [ˈsɞ̂mːɔʁˌleːdɪˌhêːdn̩], more or less. Sol505000 (talk) 15:52, 26 September 2021 (UTC)

@Nardog: Maybe we should just start transcribing the second peak in IPA? After all, it's one of the things that distinguishes Central Standard Swedish from Urban East Norwegian. Sol505000 (talk) 17:50, 26 September 2021 (UTC)

AFAIK in compounds with more than one syllable after the stress we need to mark it. Which we absolutely don’t need to do with disyllables. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 12:01, 25 September 2021 (UTC)

What are you basing that on? Sol505000 (talk) 12:06, 25 September 2021 (UTC)

Svenska Akademiens Ordbok which uses a superscript “4” after the stressed syllable with accent 1 but pairs “3” and “2” for the two main syllables in words with accent 2. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 12:16, 25 September 2021 (UTC)

Can someone who knows Swedish verify that? The relevant subpage seems to be here. Sol505000 (talk) 12:22, 25 September 2021 (UTC)

Yeah, that’s correct. They use 4-0 for accent 1 and 3-2 for accent 2 generally. --Lundgren8 (t · c) 08:19, 4 October 2021 (UTC)

@IvanScrooge98: Okay, and how would you transcribe tomten (definite of tomt, accent I) and tomten (definite of tomte, accent II), respectively?Jonteemil (talk) 12:48, 26 September 2021 (UTC)

@Jonteemil: I think we both would transcribe those as [ˈtɔ̌mːtɛn] and [ˈtɔ̂mːtɛn] respectively, without secondary stress marks. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 13:28, 26 September 2021 (UTC)

I think we've already established that declension doesn't result in secondary stress in the way compounding does, as the former results in the addition of an unstressed syllable (or not even that, in the case of tomten, with Accent 2), which is a syllable without long vowels and consonants (I think only /a/ and /ɛ/ occur in unstressed syllables, no?). Compounding by definition adds more stressed syllables (those containing a long vowel or consonant) to the word. Sol505000 (talk) 15:34, 26 September 2021 (UTC)

Again, would a secondary syllable for accent 2 sound any different if that syllable had a secondary stress from compounding? Or are the two things almost one and the same? Would e.g. a native speaker hear a difference between the -ing in viking and the -ing in guldring? These are the questions needing a proper answer to decide on the matter. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 16:59, 26 September 2021 (UTC)

I think that the ng is longer in guldring. That's secondary stress. Sol505000 (talk) 17:30, 26 September 2021 (UTC)

@IvanScrooge98: Okay, now I understand. The pronunciation of the -ing in viking and guldring are no different, unless the ring part of gulring is deliberately stressed, which then of course longens it. I didn't really get what you were arguing about but now I do. I too don't think there is a difference between non-compound words with accent II and compound dittos, so no secondary stress mark.Jonteemil (talk) 22:26, 26 September 2021 (UTC)

But compounding often results in a long vowel in a non-primarily stressed syllable, and AFAIK long vowels and consonants are banned in unstressed syllables. Guldgruva /ˈɡɵ̂ldˌɡrʉːva/ has /ʉː/, not /ɵ/. For this reason guldring is /ˈɡɵ̂ldˌrɪŋ/ in phonology - because it is made of guld and ring, even if /ŋ/ is deaccented to the point of being perceived as being the same as the corresponding velar nasal in the non-compound viking - but what if that depends on the variety of Swedish? Maybe in some varieties there is a length difference. I'm not sure whether the second syllable of gullring /ˈɡʉ̂ɭˌrɪŋ/ has the exact same duration as viking /ˈvîːkɪŋ/ in Urban East Norwegian. The fact that guldring is (according to one native speaker anyway) pronounced as if it were a non-compound has no impact on the fact that the boundary between guld and ring in guldring is not just a morpheme boundary (I believe viking is monomorphemic in modern Swedish, even though it wasn't in Old Norse - in which case it behaves almost like a loanword) but a word boundary in a compound noun, not unlike the boundary between guld and gruva in guldgruva. I'm trying not to ramble, I hope you get my point.

As I've already said, maybe let's just remove the secondary stress mark and start transcribing the second peak instead. We can keep the secondary stress mark in phonemic transcription. The phonetic output of /ˈɡɵ̂ldˌɡrʉːva/ would then be transcribed as [ˈɡɵ̂lːdɡrʉ̂ːva]. Sol505000 (talk) 02:19, 27 September 2021 (UTC)

@IvanScrooge98 and Sol505000: Interesting discussion. Phonologically, it is relevant to mark secondary stress, since secondary stressed syllables get an obvious tonal peak in longer compounds. Using Riad’s examples, sjukdom (illness) and tidning (newspaper) have similar prosodic patterns, but underlyingly, -dom carries secondary stress, but -ning does not. This shows up in longer compounds: magsjukdom (stomach illness) and kvällstidning (evening newspaper). Here -dom get tonal prominence but -ning does not, which shows that there is a phonological secondary stress.

In shorter compounds however, there is stress clash resolution. So in compound words like stormvind, the secondary syllable is reduced. Riad writes: ”If the second syllable of a compound or derivation with stressed suffix contains a long vowel, the vowel quality and to some extent quantity are retained under stress reduction”. Hence, there is a difference between e.g. tillta /²ˈtilːˌtɑː/ [²ˈtɪl(ˌ)tɑˑ] vs. tilta /²ˈtilta/ [²ˈtɪlːta]. Note Riad’s marking with half-length and secondary stress in brackets. He continues: ”If however, the vowel is short, the second syllable virtually neutralizes with unstressed syllables, in normal speech.” His examples are e.g. kvarskatt (with phonological secondary stress) vs. forskat (with none), or lokatt vs bokat. Here he does not mark consonant length in the second syllable, since the length is ”neutralized”. In another piece of work he writes that the tonal accent structure overrides the secondary stress and makes it ”hard to hear”.

In another chapter which is not about stress clash resolution, he marks secondary stress + consonant length phonetically in compounds such as [drømːˌtʰɔrːpɛ̠t] ’the dream cottage’ or [ˈɧɛ̝mːtˌɔrːɡɪ̯ɛ̠] ‘joke orgy’, even though these also have stress clash. On another page he also gives full phonetic vowel length rather than half-length in a secondary stressed syllable: [²ˈjʉːlˌloːv] ‘Christmas break’.

In summary, he is not consistent in his marking of length in secondary stressed syllables, and it seems like we can deduce that phonological length is generally preserved phonetically in some cases, but may be reduced or neutralized in other cases. I don’t know of any studies that have looked at this unfortunately. Since we use some kind of broad phonetic transcription, I think it’s safe to mark secondary stress even phonetically, even though it can be reduced in cases like stormvind or guldring. --Lundgren8 (t · c) 15:19, 30 September 2021 (UTC)

Hmm, I understand. The inconsistency of the source, this might indeed be a reason to keep it, but also one to get rid of it for disyllables. I still wonder whether the difference is relevant enough to the reader. Though at least I got the point now. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 07:44, 1 October 2021 (UTC)

Thanks, this helps a lot. Does this stress clash occur in all varieties of Swedish, or just those that are two-peaked? I know that this guide is for Central Standard Swedish alone (as far as Sweden Swedish is concerned, of course), I'm just curious whether this process is extended to all tonal varieties of Standard Swedish.

I disagree that we're using a broad transcription. We explicitly transcribe consonant length, which is non-phonemic according to the analysis in Swedish phonology. We also transcribe the tones with the corresponding IPA symbols, rather than the superscript variants of 1 and 2. So let me repeat my question: isn't it better to ditch the secondary stress symbol and transcribe the second peak instead? Then the last secondary stress in any given phonemic transcription (plus any unstressed syllable in disyllables with accent 2) would automatically yield the second falling tone in phonetic transcription. I'd agree that using both in either type of transcription would be an overkill (even more so in phonemic transcription, as one falling tone diacritic is enough to distinguish Accent 2 from Accent 1, which is transcribed as a rising tone), but our phonetic transcriptions (those that link to this key anyway) aren't broad, at least in my opinion. Sol505000 (talk) 16:32, 2 October 2021 (UTC)

Yes, in fact according to Riad (1998), stress clash is the origin of accent 2. Riad claims that the two stresses that clashed were preserved as a double-peak contour, but that’s another discussion. In accents like the one in Malmö, which often have accent 1 in compounds (unlike C.Sw.), stress clash generally results in accent 2 (Bruce 1973). If you don’t have access to those papers, this handout from 2007 contains most of his points.

Since Swedish length generally only occurs in stressed syllables, doesn’t it make more sense to mark secondary stress phonetically since there’s no full neutralization in vowel length in cases like [²ˈjʉːlˌloːv], and [²ˈtɪlːˌtɑː]? Neutralization does occur in some cases (with consonant length), but it seems to me like a secondary rule. But I see the point that phonetically it’s enough to mark the secondary peak of the pitch accent in longer compounds, and maybe the vowel length in cases like jullov and do away with secondary stress altogether if one wanted to. I don’t know what a proper phonetician would say about it. --Lundgren8 (t · c) 08:02, 4 October 2021 (UTC)

The absence of the secondary stress mark in phonetic transcription doesn't necessarily mean that the syllable is unstressed (in the case of German, Duden does away with it entirely). And in CSS, not only are /iː yː ʉː uː eː ɛː øː oː ɑː/ banned in fully unstressed positions, the vowel quality of [oː] and [ɒː] (more commonly written with ⟨ɑː⟩, which I guess is fair enough) and some other vowels (and some short vowels I guess?) is not allowed as well. In [ˈjʉ̂ːlˌloːv] and [ˈtɪ̂lːˌtɑː] you have not one but three indicators of the secondary stress on the final syllable: the secondary stress mark, the vowel quality and the vowel quantity (which will be read as half-length also according to English IPA conventions, so there's no problem). Sol505000 (talk) 09:39, 9 October 2021 (UTC)