Help talk:IPA/Danish/Archive 2

Archive 1

Archive 2

When to transcribe schwa-assimilation, and how

Clearly rane, gade, mæle, etc. can be realized with syllabic consonants, but are not transcribed so in the key. Whether they should be or not, there definitely needs to be some guidance on when to use ⟨C̩, ɪ, ʊ⟩ and when not to on the page IMO, because, AFAICT, in not all positions can sonorant + /ə/ or /ə/ + sonorant be realized as a syllabic consonant, and not all of those which can be are assimilated as often as others.

I turned to Basbøll, Grønnum, and SDU for conditions and frequency/formality of schwa-assimilation, but they were all over the place, surprisingly scarce, or above my head. So I looked for other sources for a change, and I found Jensen's (2001) summation most accessible yet rather comprehensive, according to which:

• /Vːə/ → [V:V] (e.g. pege) is "more or less obligatory in all speaking styles" (this also needs to be mentioned on the page, whether we transcribe it or not);
• /əC/ after an obstruent → [C̩] (katten) and /j, v, ð/ + /ə/ → [C̩] (bade) are "very common";
• /Cə/ → [C̩] is "very common" after two semivowels, which are "(almost) always" [ɐ̯w] (værge), and is "less common" after a sonorant (obsruent + /Cə/ isn't even mentioned so I have to assume assimilation never occurs in that context).

A problem with Jensen, though very helpful, is that it cites neither existing literature nor empirical study, and it's not really a descriptive work as much as about speech synthesis, so we shan't put all the eggs in this basket.

Then I finally found that Schwa dk had written theses on schwa-assimilation (figures!) based on the DanPass corpus. His primary finding, summarized here and here, is that schwa almost always assimilates:

• before a tautosyllabic sonorant;
• after a heterosyllabic vocoid;
• in the penultimate syllable in an antepenultimate-stressed stem

and is particularly retained:

• after two consecutive sonorants;
• when final in an antepenultimate-stressed stem;
• when phrase/utterance-final

which is a much simpler and more widely applicable set of rules.

It still calls for some caveats though, like: (1) as the author acknowledges, the study was based on a contemporary corpus, so it may not be representative of the varieties the reference works we already rely on describe; (2) although it might be more phonetically accurate, it would mean losing some lexical information (cf. DDO, Molbæk Hansen, and Politikens, which don't even bother with it), so it probably comes down to what we want to encode, lexical vs. phonetic (a similar problem to the one I raised above).

I'm also confused as to when vowel + /jə, və/ should be transcribed with [-ɪ, -ʊ] and when [-jɪ, -wʊ]. Looking at Grønnum (2005: 186–7, 269), I thought [ɪ, ʊ] could only follow long vowels and [Vjɪ, Vwʊ] were merely innovative variants, but SDU gives only [ˈlajɪ] for lege, for example. [-jɪ, -wʊ] are used when the preceding vowel is short, as is [CC̩]. Nardog (talk) 05:56, 28 March 2020 (UTC)

and is particularly retained (...) when phrase/utterance-final - that's not my experience when it comes to the recordings from Den Danske Ordbog, which obviously should represent citation forms. On those recordings, the final schwa is sometimes assimilated, yet at other times it's not. Or perhaps Danish schwa is that short after sonorants that sometimes I can't hear it. For instance, here, does the speaker say [ˈʁɑːnə] or [ˈʁɑːn̩]? Kbb2 (ex. Mr KEBAB) (talk) 06:27, 28 March 2020 (UTC)

It's released while still voiced, that's for sure (sorta like French final [n]). In ran, [n] is released after the voicing has stopped, so you hear a [t]-like plosion. Maybe that's how the contrast manifests. Nardog (talk) 07:43, 28 March 2020 (UTC)

This[1] short overview by Lars Brink (essentially a follow-up of Brink&Lund 1975) has a short section (No 5.) about schwa-assimilation. He writes that it makes a major difference whether schwa occurs before pause of not. If not, as in /əC/ (C=sonorant), it is obligatory. For /V:ə/, the "1955"-generation has obligatory schwa-assimilation in sponaneous utterance. The "younger" generation has obligatory schwa-assimilation following the vocoids /ð j w/.

As for the DDO recordings, I consistenly hear schwa after /l m n ŋ/. After /V j w/ I always hear schwa-assimilation, after /ð/ mostly. –Austronesier (talk) 13:45, 29 March 2020 (UTC)

Thanks for the resource. So what should we do with the transcription? FWIW, I'm leaning towards keeping ⟨ə⟩ in all environments and relegating the assimilation to footnotes on the guide (and/or listing ⟨əC, Cə⟩ as separate segments much like the syllabic consonants in the English key) so that we could at least be consistent—especially if it's not a permanent diachronic change towards fewer [ə] but a synchronic variation where [ə] that disappeared may show up again in the same lexeme, which I understand is the case. It wouldn't be unprecedented either (cf. DDO and Molbæk Hansen). Nardog (talk) 04:44, 30 March 2020 (UTC)

@Nardog: I'm fine with ditching the syllabic consonants and [ɪ ʊ]. Kbb2 (ex. Mr KEBAB) (talk) 08:17, 30 March 2020 (UTC)

I'm fine with leaving /Cə/ unchanged, but would prefer to transcribe /əC/ with syllabic [ð̩ l̩ n̩] (but without further assimilation in [pn̩ ~ kn̩]). Especially with [ð̩], this is obligatory, since [ð] behaves like (or better: is) a semivowel, and on p. 108 the authors of the SDU write "ə findes aldrig foran halvvokal (ð...)". I flipped through the entries of the SDU, AFAICS they always use [l̩] for /əl/, and mostly syllabic nasals for /əN/. With /Cə/, the first entry usually is [Cə]. –Austronesier (talk) 09:52, 30 March 2020 (UTC)

That seems consistent with Schwa dk's findings, but is there reason to disregard the place assimilation in /ən/? That strikes me as half-baked, at least on the face of it. Nardog (talk) 10:03, 30 March 2020 (UTC)

@Nardog: Ok, [pm̩ ~ kŋ̍] then. The SDU has many entries with [pm̩ ~ kŋ̍] as primary pronunciation, but a few unassimilated entries too. –Austronesier (talk) 10:36, 30 March 2020 (UTC)

If it was up to me, I would then use the freed-up ⟨ɪ⟩ for /e/, which I've found most confusing (without the knowledge that ⟨ɛ⟩ is also used for the same language, one expects ⟨e⟩ to represent a mid vowel as that's the case most of the time) yet we haven't replaced! Nardog (talk) 09:57, 30 March 2020 (UTC)

@Nardog: If we're not gonna use it for /jə/ than that's not a bad idea - especially given the fact that the overwhelming vast majority of instances of the short /e/ are spelled with ⟨i⟩. Kbb2 (ex. Mr KEBAB) (talk) 10:08, 30 March 2020 (UTC)

@Nardog: I don't think using ⟨ɪ⟩ for /e/ is a good idea, since it is AFAICS against native speaker intuition. In case we use ⟨ɪ⟩, how would we spell shortened /e:/ in monosyllables with stød, or det and til? –Austronesier (talk) 10:49, 30 March 2020 (UTC)

What's wrong with ⟨ɪˀ, ˈdɪ, ˈtɪ(l)⟩? Nardog (talk) 11:04, 30 March 2020 (UTC)

I'm not aware of a source that transcribes /e:/ and /e/ differently (and uses short ⟨ɪ⟩ for that matter). Jespersen described short /e/ a little bit more open than /e:/; in that case, actually long /e:/ would be the better candidate for being transcribed with ⟨ɪ:⟩ (at least height-wise). –Austronesier (talk) 11:15, 30 March 2020 (UTC)

@Austronesier: I think Nardog is talking about both short and long versions, not just short. Kbb2 (ex. Mr KEBAB) (talk) 11:17, 30 March 2020 (UTC)

Yes. Nardog (talk) 11:21, 30 March 2020 (UTC)

Ok, then it's: I'm not aware of a source that transcribes /e(:)/ with ⟨ɪ⟩. Apart from that, [e̝] could well be transcribed by semi-narrow ⟨ɪ⟩. It will confuse our Danish readers, but for other readers who are familiar with general IPA, it serves as a easily recognized target pronunciation. –Austronesier (talk) 11:36, 30 March 2020 (UTC)

I know, but I'm not aware of a source that uses ⟨æ⟩ for /a(ː)/ either. (And the IPA defined ⟨ɪ⟩ simply as "lowered [i]", i.e. without centralization, until 1979.) Nardog (talk) 11:48, 30 March 2020 (UTC)

That's a point. One clear benefit is, we'll have less diacritics for the unrounded front series. –Austronesier (talk) 14:20, 30 March 2020 (UTC)

Add @Nardog: The SDU does, on p. 85. Short /a/ is given as [a̝], også [æ]. –Austronesier (talk) 17:38, 2 April 2020 (UTC)

By "use" I mean throughout a work. We have switched to using ⟨æ⟩ not because SDU mentions it on one page but because linguistic descriptions show that the phonetic quality of the phone is, in standard IPA, closer to [æ]. Nardog (talk) 00:41, 3 April 2020 (UTC)

Well, this key doesn't use diacritics except for syllabicity to begin with so the benefit is rather that now both ⟨e⟩ and ⟨ɛ⟩ would be used for the phones that better match the IPA values. If you're talking about using ⟨ɪ⟩ for /e(ː)/ in Danish phonology, I wouldn't push for it, because that article has to deal with phonemics too, which IPA-xx transcriptions do not. Nardog (talk) 14:48, 30 March 2020 (UTC)

@Nardog: Don't forget that sooner or later transcriptions linked to this guide will match the system used in Danish phonology (as they should). Kwami will take care of that. I see no problem with using ⟨ɪ⟩ for the near-close front unrounded vowel (that's phonemically close-mid) both here and there, just as there's no problem with using ⟨e⟩ for the close-mid vowel that's phonemically open-mid. Basbøll makes it clear that [ɪ] for /jə/ is short-only, unstressed-only and non-phonemic. (BTW, should pige be transcribed [ˈpʰiːə] or [ˈpʰiːi]?) Kbb2 (ex. Mr KEBAB) (talk) 08:45, 31 March 2020 (UTC)

But we're not introducing ⟨ɛ̝, œ̝, ɔ̝⟩ to the guide, are we? And the phonology article has to mention the assmilatory variant of /jə/ anyway, so I don't see much benefit in replacing ⟨e̝⟩ with ⟨ɪ⟩ there.

[ˈpʰiːə], I would say. To be clear, I'm still not convinced that we should show any assimilated forms in IPA-da transcriptions. Sources say some of them are obligatory, granted, but there seems to be too much contextual and generational variation while DDO transcribes none of them. I don't think there's any harm in following DDO and explaining it in a footnote. Nardog (talk) 20:06, 31 March 2020 (UTC)

 Comment: Per DDO, I'd also say it should be [ˈpʰiːə], though it might be worth adding adding a note about it being pronounced [ˈpʰiːi] in everday usage (if a reliable source can back this claim up, of course). —Biscuit-in-Chief :-) ^{(/tɔːk/ – /ˈkɒntɹɪbs/)} 19:10, 3 April 2020 (UTC)

I'm circling back to using ⟨e̝⟩ rather than ⟨ɪ⟩. Even for those who are used to seeing ⟨e⟩ representing a mid vowel as in many languages, ⟨e̝⟩ at least signals it's not a "usual" [e], while ⟨ɪ⟩ may indeed be confusing to some already familiar with Danish.

@Austronesier: Do you still think ⟨C̩⟩ should be kept just for /əC/? My concerns, in addition to those I've already articulated, are that schwa-assimilation in other environments seems equaliy common so it would be inconsistent if we only transcribed it in /əC/, and that, on the flip side, transcribing all would entail determining the exact environments where it is obligatory, which strikes me as formidably complex. Nardog (talk) 13:28, 7 April 2020 (UTC)

@Nardog: It depends on how much the unassimilated forms sound stilted and over-pronounced. For /Cə/, unassimilated forms are still possible in careful yet natural speech; but as for /əC/, Basbøll says unassimilated forms only occur "in highly distinct speech forms". To me it appears that [-əð] is in fact hard to realize with two segments, and consequently the SDU even says that [-əð] is a disallowed sequence. [-əl -əm -ən -əŋ] are certainly much easier to produce, but what is the native speaker's intuition? Is [ɔːpən] instead of [ɔːpm̩] on par with [ˈpʰiːə] vs. [ˈpʰiːi]? Or does [ɔːpən] sound more "unnatural" than [ˈpʰiːə]? @Biscuit-in-Chief: What do you think? –Austronesier (talk) 20:43, 7 April 2020 (UTC)

Transcriptions need not always accurately reflect the pronunciation. I don't think any speaker of a standard variety of English pronounces nation, level, etc. with a schwa, but we transcribe them with ⟨ə⟩ because an understanding between the transcriber and the reader is made at Help:IPA/English#Notes that ⟨ən, əl⟩ denote syllabic consonants in those positions. I don't see why it can't be like that for Danish as well. Nardog (talk) 21:44, 7 April 2020 (UTC)

@Austronesier: [ɔːpən], although the formal pronouncation (see DDO), definitely sounds very unnatural to me—and if one were to pronounce it that way without context, I probably wouldn't understand what they were saying. [ˈpʰiːə] does sound slightly odd, but nowhere near as odd as [ɔːpən]. —Biscuit-in-Chief :-) ^{(/tɔːk/ – /ˈkɒntɹɪbs/)} 00:40, 8 April 2020 (UTC)

Keep in mind that I don't know about all the dialects of Danish, so more obscure ones may pronounce it [ɔːpən] (I doubt it, but I can't know for certain). —Biscuit-in-Chief :-) ^{(/tɔːk/ – /ˈkɒntɹɪbs/)} 00:43, 8 April 2020 (UTC)

@Biscuit-in-Chief: Thanks a lot, that sort of confirms what our sources say. @Nardog: It's not really comparable to English, where syllabic sonorants are not as ubiquitous as in Danish. English has blocking rules (e.g. after clusters), whereas in Danish there are no such constraints e.g. lampen [lampm̩] (cf. English [dæmpʰən]). And then there's unstressed dem, den [pm̩ tn̩], which appear in formal natural speech, unless carefully pronounced. So syllabic [C̩] for /əC/ really adds to the "Danishness" of the transcriptions: we already use it in [kʰøpm̩ˈhɑwˀn], and [kʰøpənˈhɑwˀn] does not look like an improvement. –Austronesier (talk) 08:18, 8 April 2020 (UTC)

I find your argument a bit hard to follow. To reiterate, I'm only opposed to limiting the use of syllabic consonants to /əC/ because Jensen, Schachtenhaufen, and Brink suggest schwa-assimilation in other environments is equally common. If there's reliable evidence that sets /əC/ apart from the rest of contexts in terms of frequency or markedness, I'll be down for that.

How about we partially follow Schachtenhaufen's rules? That is, we only transcribe /əC/ where C is a sonorant and /V(ː)Cə/ where C is a vocoid ([ð, j, w, ɐ̯]) as syllabic. The rules are set out in detail here, and I haven't read the whole thing so correct me if I'm wrong, but he seems to say that the syllabification rule is that the syllable preceding one that contains a schwa maximizes its coda, which makes "one vocoid following a vowel" a good rule of thumb that we can practically follow. The rule about proparoxytone stems seems a bit too complex, but we can also follow it if we want to be thorough. Nardog (talk) 01:21, 9 April 2020 (UTC)

Now that I think about it, the second rule about vocoids only applies word-internally, because schwa is resistant to assimilation when phrase/utterance-final, and we mainly transcribe citation forms. Nardog (talk) 01:38, 9 April 2020 (UTC)

Well, there's Basbøll (2005:293): "schwa-assimilation is obligatory, except in highly distinct speech forms, when schwa occurs before a sonorant", while other cases are "strongly favoured" or just "favoured". Brink (2013) says the same: "nobody has ə nowadays in types like marked, fakkel, koen, spanden" because there is no option of a pause after schwa. But he also mentions that schwa-assimilation is also obligatory in other cases (after long vowels and semi-vowels) for people born after 1955 or later. That's essentially the situation described by Schachtenhaufen. My idea was to apply schwa-assimilation only the obligatory cases in Basbøll's standard. But both cut-off points are well-sourced, so I'm fine to notate the contemporary situation described by Schachtenhaufen, by using your rule of thumb. –Austronesier (talk) 07:35, 9 April 2020 (UTC)

Well, the standard we follow is SDU, which is based on the speech of people born circa 1930 (which is arguably old but there is no newer source that's as comprehensive), and Basbøll is a strong source. I'm persuaded, let's transcribe schwa-assimilation only in /əC/. It's much simpler than Schachtenhaufen's rules too, saving our hassle. Nardog (talk) 07:53, 9 April 2020 (UTC)

frisk = [ˈfʁæsk]?

I wonder if this is some kind of mistake. Basbøll, DDO, and Molbæk Hansen all transcribe /e/ after /ʁ/, as in frisk and brist, with ⟨ɛ⟩, and Grønnum with ⟨æ⟩, so it cannot possibly be lower than the cardinal [ɛ]. It's [ɛ̝] in standard IPA according to Basbøll and Grønnum, if it has the same quality as the neutralized /eː/–/ɛː/ after /ʁ/ (which they appear to suggest given their appendices are supposed to illustrate the whole sound inventory).

We do seem to transcribe /e(ː), ɛː/ after /ʁ/ with ⟨æ⟩ in names like Christian, Margrethe, Frederik, etc. Indeed, the frisk example was already in the key before we switched to non-normalized IPA. Coupled with what I said in the above sections, I've begun to think we might want to re-transcribe the whole thing all over. Nardog (talk) 12:23, 29 March 2020 (UTC)

frisk should be [ˈfʁɛ̝sk], also bær should be [ˈbɛ̝ɐ̯], since Grønnum has ⟨æ⟩ in half-narrow transcription corresponding to ⟨ɛ̝⟩ in narrow transcription (although she only has examples with long ⟨ɛ̝:⟩ for the narrow transcription. –Austronesier (talk) 14:11, 29 March 2020 (UTC)

I don't know, Basbøll and Grønnum's ⟨æ⟩ corresponds to [ɛ] in male, have, jage according to the appendices. How do you know it's the same quality as in grene rather than [ɛ]? Nardog (talk) 04:03, 30 March 2020 (UTC)

@Nardog: Yeah, I was wrong about bær. Both use ⟨æ⟩, so it must be plain [ɛ]. As for frisk, I presume that the short and long vowels transcribed as ⟨ɛ ~ ɛ:⟩ by Basbøll and ⟨æ ~ æ:⟩ by Grønnum after R have the same quality—at least for long grene, both write narrow [ɛ̝:], from which can safely infer [ɛ̝] for the short vowel. –Austronesier (talk) 08:11, 30 March 2020 (UTC)

I now see why ⟨æ⟩ was still used for [ɛ~ɛ̝]: ⟨ɛ⟩ now means sometimes [e] and sometimes [ɛ~ɛ̝]. But to keep using ⟨æ⟩ for [ɛ~ɛ̝] would have meant it sometimes meant [æ~a] and sometimes [ɛ~ɛ̝] (and when long, it still does).

I think it was a mistake to start using ⟨æ⟩ for [æ] without introducing another symbol for [e̝] and thus freeing up ⟨e⟩ for [e] and ⟨ɛ⟩ for [ɛ~ɛ̝]. Let's introduce ⟨e̝⟩ already. Nardog (talk) 13:28, 7 April 2020 (UTC)

Yes, most non-high front vowels have to be moved one step up, except for the vowel in kat, which is already narrowly transcribed as [ˈkʰæt] (NB not [ˈkʰɛt]! That would be the stigmatized and obsolete Low Copenhagen "flat a"). Short [ɛ] can still be used in [ˈfʁɛsk] (without diacritic), which according to both Basbøll and Grønnum is a bit lower than the vowel in ven (narrow [ven]). Using [e̝] for the second to the highest front vowel will be straightforward, and as you say above, it will indicate that it's not the regular ⟨e⟩ –Austronesier (talk) 16:06, 7 April 2020 (UTC)

Done (in the summary I meant introduce, not "restore"). Nardog (talk) 00:37, 9 April 2020 (UTC)

@Austronesier: I'm having second thoughts about using ⟨ɛ⟩ for [ɛ̝, ɛ̝ː~ɛ̝j]. All sources except Grønnum transcribe them with the same symbols they use for what we now transcribe as [e(ː)]. Wouldn't ⟨e⟩ be simpler? Nardog (talk) 11:39, 9 April 2020 (UTC)

@Nardog: I'm a bit undecided here. At least Basbøll also employs [ɛ̝, ɛ̝ː~ɛ̝j] in Appendix A1 for normalized ⟨ɛ⟩ after R (grene, frisk), whereas normalized ⟨ɛ⟩ in other contexts represents narrow [e] (fælde, mæle). The SDU has the same Dania vowel «æ» in fælde, mæle and frisk, and also as first part of the diphthong in «æiͅ» in grene. The IPA key in the SDU equates Dania «æ» with IPA [e̞]. Brink et al. otherwise have very keen ears (they e.g. descirbe the vowel in trække as an extrashort diphthong [tʰʁɑ͜ækə], where Basbøll and Grønnum have narrow [tʰʁakə]), so I think we could also follow the SDU here. –Austronesier (talk) 12:47, 9 April 2020 (UTC)

@Austronesier: By "follow the SDU" you mean to use ⟨e⟩, I assume? That's what I'm inclined for as well, given ⟨ɛ̝⟩ and ⟨e̞⟩ are pretty much equivalent since the IPA does not provide a way to qualify "how much" raised or lowered. I wouldn't be surprised if Basbøll and Grønnum were using ⟨ɛ̝⟩ over ⟨e̞⟩ simply because ⟨ɛ⟩ is the established notation for this phone in the "normalized" IPA. A more immediate benefit, of course, is that now ⟨ɛ⟩ and ⟨æ⟩ in reference works can always be converted to ⟨e⟩ in our transcription, whether they're adjacent to /r/ or not. Nardog (talk) 01:06, 10 April 2020 (UTC)

[ɑ, ɑː]

An IPv6 user has repeatedly changed them to ⟨a, aː⟩ with no discussion. The discussions through which we arrived at the current set of symbols are documented here and at Talk:Danish phonology. I would also add that I find ⟨ɑ, ɑː⟩ superior, given ⟨a, aː⟩ are the symbols used for what we now transcribe with ⟨æ, æː⟩ in Danish literature. Nardog (talk) 10:21, 16 April 2020 (UTC)

I support to revert the inarticulate IP drive-by edits and to maintain the current consensus. Without edit summary, these edits are disruptive—regardless of their underlying rationale which we don't know due to our lack of a crystal ball. –Austronesier (talk) 12:15, 16 April 2020 (UTC)

If you aren't already, be sure to link to the relevant discussions in your edit summaries to give users (anonymous or not) to clearly see that your edits have consensus. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:34, 16 April 2020 (UTC)

I already did. Nardog (talk) 15:45, 16 April 2020 (UTC)

kat = [kæt]?

I do not believe the [æ] is correct! Just listen here: https://forvo.com/word/kat/#da - does not sound like [æ] at all. And I listened to many Danish sources, this always sounds just like a [ɛ], more or less the same as in German "hätte". — Preceding unsigned comment added by 84.58.193.219 (talk) 14:50, 29 October 2020 (UTC)

@IP: There is considerable height variation for the sound we transcribe as [æ] here (check e.g. https://forvo.com/search/mand/da/). Our transcription follows the descriptions by Basbøll and Grønnum, who actually use the symbol [a̝]. The vowel becomes increasingly raised with younger speakers, and [ɛ] is absolutely common now. This once highly stigmatized phenomenon (called det flade a – "the flat a") is described here: https://core.ac.uk/download/pdf/76379979.pdf.

So our transcription is sound, but maybe representative of what is becoming an "old-fashioned" pronuncation. But since there is no comprehensive description of Danish vowel sounds in the speech of the contemporary young and mid generation, we stick to the Basbøll/Grønnum transcription for consistency. –Austronesier (talk) 16:52, 30 October 2020 (UTC)

Unvoiced fricative, græde

1. Danish phonology says "/l, j, r/ are voiceless [l̥, ç~ɕ, χ] after /p, t, k/", so shouldn't græde be [ˈkχ-] instead of [ˈkʁ-]?
2. Secondly, it's a poor example, because younger speakers, such as myself, pronounce the vowel as more like an [ɑ]. If you compare the DDO audio for kat and græde, it's obviously not the same vowel.
   Assuming this page is meant to help editors without formal linguistic training to transcribe words, it needs to use examples that are shared among age groups.__Gamren (talk) 21:20, 13 November 2020 (UTC)

@Gamren:

1. "/l, j, r/ are voiceless" holds for the position after the phonemes /p, t, k/ i.e. phonetic [pʰ, tsʰ, kʰ]. They remain voiced following phonemic /b, d, g/, even though the latter are phonetically [p, t, k].
2. This is admittedly problematic, since græde is pronounced by most younger speakers as [ˈkʁɑːðə] (or better: [ˈkʁɑːð̩]), thus homophonous to grade. So actually, [æː] is non-existent for this group of speakers. For older speakers, græde and grade are still distinct, and the first has a vowel that is more or less the long version of the kat-vowel. That's how the græde-vowel is described by Grønnum and Basbøll, who belong to that group of speakers. What we can do is to add a note saying that this vowel is extremely rare anyway, and is realized as [ɑː] by younger speakers. This shift is well-documented in Grønnum (2005). –Austronesier (talk) 08:29, 14 November 2020 (UTC)

Add The DDO lies halfway, by still distinguishing græde and grade, but using a vowel for the former that is clearly more open than the kat-vowel. The vowel in kat is flat out flad in the DDO sound sample. Quite interestingly, the DDO transcribes kat and ret with the same vowel, although in the sound samples, the latter clearly is more open, and actually corresponds to the vowel in græde. –Austronesier (talk) 08:46, 14 November 2020 (UTC)

Table format

@Nardog: mine may not have been an improvement stricto sensu, but do you not agree that listing the large amount of vowels of Danish the way I did may be a lot more immediate to the reader? Also, grouping syllabic consonants in the same list as pure vowels seems to me a little off and a bit misleading in narrow phonetic terms. 〜イヴァンスクルージ九十八［IvanScrooge98］（会話） 23:40, 17 May 2021 (UTC)

I don't. Your version of the key is gratuitously wider, with the "N/A" parts making it even more pointless. I for one find the alternation of short and long vowels easy on the eye. In the vast majority of cases we use "vowels" and "consonants" in the phonological sense, or otherwise we would be counting approximants among vowels. The syllabic consonants in Danish are weak vowels whichever way you look at it. Also, if you're making a separate section for syllabic consonants then it should either include [ɐ] (from a phonological perspective) or exclude [ð̩] (from a phonetic one). Nardog (talk) 01:24, 18 May 2021 (UTC)

I have compared the tables in mobile view on my phone. The long version requires more scrolling, but little swiping. What I like better about the long version is that I can immediately see the consonant part of the table; in the wide version, a first-time reader cannot tell whether they will have to scroll or swipe to find the consonants. So I opt for the older version here.

The unstressed syllabic segments are tricky. It appears counterintuitive at a first glance to have [ə] and [n̩] in the same block, but a division into two parts creates a cutoff problem, especially with respect to where to put [ð̩] then. No strong opinion here. –Austronesier (talk) 12:22, 18 May 2021 (UTC)

[e] for 'ven' and 'mæle' needs to be changed back to [ɛ]

I see that Nardog made this change relatively unilaterally in April 2020 and later tried to present it as consensus in getting me to follow it.

Most transcriptions, as one can see under § Comparison of transcription schemes, transcribe it as ⟨ɛ⟩. Regardless, it's clearly an ⟨ɛ⟩. If you listen to this performance, for example, the close-mid front unrounded vowel ⟨e⟩ in "det", two words prior, is clearly not the same vowel as med or brede, where it is the open-mid front unrounded vowel ⟨ɛ⟩. Wiktionary lists the pronunciation as mɛð, both phonemically and phonetically.

I only found one brief mention by Nardog of this particular change, at the very end of a discussion about something else. · • SUM1 • · (talk) 18:39, 16 January 2022 (UTC)

No, it's was not done unilaterally, but is based on consensus and the result of hard word in Talk:Danish phonology. These transcriptions are supported by multiple sources, cf. the narrow phonetic transcriptions in Basbøll (2005) and Grønnum (2005); you can also check the right column in the pronunciation-guide of the Danske Ordbog, where the vowel in mæle is described as [eː] in nøjagtig (= exact) IPA. The transcription as [ɛː] is merely conventional, but our readers deserve an accurate transcription. –Austronesier (talk) 20:04, 16 January 2022 (UTC)

PS: Brink et al. transcribe this vowel in Den Store Danske Udtaleordbog in their practical Dania-derived notation as ⟨æ⟩, but they describe the sound in the introduction as IPA [e̞]. For Brink et al., Basbøll and Grønnum (and also in our transcription), IPA [ɛ(ː)] is the sound heard in words like male (conventional: [mæːlə]). –Austronesier (talk) 20:36, 16 January 2022 (UTC)

To transcribe a contrast between [ɛː eː e̝ː] with ⟨æː ɛː eː⟩ is really misleading. The cardinal-like qualities of [ɛː] and [eː] are normally given the transcription ⟨æː ɛː⟩, which makes no sense, especially in the latter case. Why on earth would you choose something else than a simple Latin letter ⟨e⟩ for an [e]-like vowel (apart from ⟨ɪ⟩, whenever there are good phonological reasons to use it)? Not to mention that there is a number of other languages (French, Hindustani, German, Italian, Portuguese, Slovene, and even Swedish) where ⟨ɛ⟩ and ⟨e⟩ are used for cardinal-like (or nearly so) qualities, and the contrast between the two is often even described as an "open E" [ɛ] vs. "closed E" [e], mirroring the labels "open-mid" and "close-mid" used by the IPA. The difference between the Danish [eː] vs. [e̝ː] is very different to what we can find in those languages (most importantly, the Danish [eː] is basically the "closed E" of French, German etc., whereas the Danish [e̝ː] is just a little bit closer/higher than that), which are spoken by hundreds of millions of speakers. It is a very bad idea to redefine the symbols ⟨æː ɛː eː⟩ for a language like Danish, which is hardly understood (let alone spoken) by anyone in comparison with the aforementioned languages combined. As far as it is possible, there should be a continuity between IPA transcriptions of various languages - especially if you want to redefine symbols for a (considerably) less known language.

The cardinal-like vowels should be given the corresponding cardinal vowel symbols first, and only then should the rest get their own symbols - with diacritics, if that is needed (I don't think anyone has a problem with the nasal diacritic in IPA transcriptions of French?!). Now, I know that a vowel shift has taken place over the last 50 years or so, but then that should definitely be shown in transcription. It doesn't matter that some of the symbols would overlap (when you compare the reformed transcription with the traditional one), over time people would get used to the differences (and if they wouldn't, tough luck - but some standards should be maintained in IPA transcriptions. Nobody uses ⟨k⟩ to denote an alveolar trill). Sol505000 (talk) 13:57, 10 February 2022 (UTC)

/a/ and /aː/ missing from key

I notice that the symbols /a/ and /aː/ (the front open vowel in IPA) are present in Wiktionary's phonemic entries (for example in land, valke, a total of 1395 times in my data), but not present in the key here. I'm not a speaker of Danish, and was wondering if any anyone might help me figure this out and correct the key. My first guess was that /æ/ (IPA's near-open front) in the key might actually correspond to /a/ in the majority of phonemic entries, since there are only 39 instances of /æ/ in Wiktionary. In other words, most of Wiktionary's entries actually use Den Danske Ordbog's transcription /a/, which is not reflected in the key. My first check was the word "græde" in the key here for /æ/ which is given as [ˈgʁaːðə] in the Ordbog, and "række" is [ˈʁagə]; while in Wiktionary, we have "græde" /ɡræːðə/ but no phonemic transcription for "række" (just the narrow transcription [ˈʁaɡ̊ə]).

However, when I look up some more of the 39 Wiktionary /æ/ words, I get inconsistent results from the Ordbog, for instance "regering" [ʁεˈgeˀɐ̯eŋ], modsat [ˈmoðˌsad], lukaf [lɔgˈæˀ], anden [ˈanəð], all of which have /æ/ in Wiktionary. I wonder if these 39 /æ/ words are mostly incorrect, and the key's /æ/ simply corresponds to /a/ in the majority of entries (in which case this should be fixed)? Thank you for any clarifications. For reference, here are the 39 /æ/ words: ['a', "a'erne", "a'ernes", "a'ers", "a'et", "a'ets", "a's", 'aldring', 'andet', 'bejae', 'børnehave', 'fladfisk', 'grundlovsdag', 'græde', 'indebrændt', 'kvæk', 'lukaf', 'manden', 'mexicansk', 'modsat', 'momentan', 'napoli', 'nervøs', 'nervøsitet', 'nicaraguansk', 'nævn', 'nævne', 'opmærksomhed', 'overfladisk', 'regering', 'samtale', 'sandslot', 'sfære', 'sydafrikaner', 'sydafrikansk', 'testikel', 'ugedag', 'vandløs', 'velsmagende'] — Preceding unsigned comment added by Craig Baker (talk • contribs) 15:38, 31 March 2022 (UTC)

Contradiction in "kræft" between DDO and the article

The table claims that the sound "æ" in "kræft" is different from "æ" in "række". Moreover, according to the table the former is transcribed as "ɑ" while the latter as "a" in DDO. However, I looked up "kræft" and "række" in DDO to find that the pronunciation is introduced as [ˈkʁafd] and [ˈʁagə], both using "a" and not using "ɑ" for the former. Is it a mistake? I'm new to Danish so I may be missing something. Hydro I. Hue (talk) 18:27, 1 November 2022 (UTC)

Scroll down. Nardog (talk) 18:28, 1 November 2022 (UTC)

I'm afraid to say that I cannot find any clue below. Is something mentioned in the footnote or do I need to refer something in this talk page...? Hydro I. Hue (talk) 16:57, 2 November 2022 (UTC)

What do you see in the "Comparison of transcription schemes" section? Nardog (talk) 17:13, 2 November 2022 (UTC)

OK I'm sorry I didn't clarify enough. I am actually talking about the "Comparison of transcription schemes" table. In the second column is an examle word "kræft", and in the eighth, "række". The entry for "Den Danske Ordbog" is ɑ and a, but in actual DDO I see [ˈkʁafd] and [ˈʁagə]. Hydro I. Hue (talk) 06:10, 3 November 2022 (UTC)

Oh, my bad, I thought you were talking about the difference between our IPA and DDO. I don't remember DDO diverging from Basbøll in terms of vowel analysis, but I guess we overlooked it. DDO's guide to pronunciation does mention the potential homophony of kræft and kraft FWIW. Nardog (talk) 15:53, 3 November 2022 (UTC)

Thanks for fixing it up! Hydro I. Hue (talk) 21:09, 5 November 2022 (UTC)