Wikipedia Help NA‑class

This page is within the scope of the Wikipedia Help Project, a collaborative effort to improve Wikipedia's help documentation for readers and contributors. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks. To browse help related resources see the Help Menu or Help Directory. Or ask for help on your talk page and a volunteer will visit you there.Wikipedia HelpWikipedia:Help ProjectTemplate:Wikipedia Help ProjectHelp articles NA This page does not require a rating on the project's quality scale.

Linguistics: Phonetics NA‑class

This page is within the scope of WikiProject Linguistics, a collaborative effort to improve the coverage of linguistics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.LinguisticsWikipedia:WikiProject LinguisticsTemplate:WikiProject LinguisticsLinguistics articles NA This page does not require a rating on Wikipedia's content assessment scale. This page is supported by Phonetics Task Force.

There is a move discussion in progress on Help talk:IPA which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 16:16, 15 July 2017 (UTC)Reply[reply]

Hello. I've just begun fixing that article, and I've encountered a massive number of transcriptions with incorrect vowel length. Could someone go and recheck them? I'm sure that I missed some. Also, I can see that there are some (presumably) non-standard notations of consonant length, e.g. [ˈsvauŋːkʏr] instead of [ˈsvauŋkʏr]. That should also be fixed.

Morgunblaðið also needs a transcription check. Isn't it [ˈmɔrkʏnˌplaːðɪð], with a long [aː]? Mr KEBAB (talk) 16:58, 1 April 2018 (UTC)Reply[reply]

I'm looking at a book about the history of Icelandic and it seems to say (on p. 13) that the consonant following a stressed phonetically short vowel is or may be lengthened: "With the quantity shift every stressed syllable was long but the degree of stress could be amplified by drawing out the pronunciation of a long vowel or the enunciation of the consonant following a short vowel." This is consistent with the transcription you mention, [ˈsvauŋːkʏr]. Of course it's a separate question whether transcriptions on Wikipedia should show it. Some transcriptions on Wiktionary do, though they use slashes when they should use brackets. Icelandic phonology doesn't mention it. — Eru·tuon 23:16, 1 April 2018 (UTC)Reply[reply]

@Erutuon: So do you think that we should improve this guide and/or Icelandic phonology based on that information? Or is it there already and I just missed it?

Yeah, using slashes instead of brackets in Icelandic transcriptions on Wiktionary is a very prevalent mistake. Maybe someone should write a script to just replace the slashes with brackets, I don't know. But there are instances where using slashes to transcribe words with long consonants is correct. Compare man /man/, [maːn] and mann /mann/, [mann] (or /manː/, [manː], same thing really). This pair is actually used to prove the non-phonemicity of vowel length by the way. Mr KEBAB (talk) 13:31, 2 April 2018 (UTC)Reply[reply]

I've rechecked both and the information isn't there. So there's some room for improvement. Mr KEBAB (talk) 14:09, 2 April 2018 (UTC)Reply[reply]

@Mr KEBAB: If a source can be found, I suppose this phenomenon should be mentioned in Icelandic phonology. (The history book mentions the phenomenon too briefly to serve as a source.) Certainly gemination of a lone consonant should be transcribed, but I'm not sure about gemination of the first consonant in a cluster. It is transcribed in Swedish (see the "Stress, tone and syllabification" table). If it's quite salient, certainly. — Eru·tuon 21:38, 2 April 2018 (UTC)Reply[reply]

The Icelandic phonology article was already using [hp ht hc tk] for these sequences before I started any serious editing of it. But Help:IPA/Icelandic and most IPA in articles (including what I've added just to fit existing convention) have been using [ʰp ʰt ʰc ʰk] instead. I think it may be a better idea to use [hp ht hc tk] in all cases, and to update Help:IPA/Icelandic and article pronunciations to reflect this notation. Why?

• As noted in the phonology article, the preaspiration in [hp ht hc hk] is significantly longer than the postaspiration in [pʰ tʰ cʰ kʰ], making a full-quality [h] more appropriate for preaspiration.
• Preaspirated stops are associated with geminate consonant sequences pp tt kkj kk anyway. (They also occur in the post-vocalic sequences pl pn tl tn kl kn within a morpheme, as [hpl hpn htl htn hkl hkn].) [ʰp ʰt ʰc ʰk] have the misleading superficial appearance of being single-length consonants, which looks strange when compared to the IPA for the other geminate consonant sequences bb dd ggj gg, which can be represented as [pp tt cc kk] or as [pː tː cː kː] and in either case are still represented as longer than a single consonant.

- Gilgamesh (talk) 16:12, 7 May 2021 (UTC)Reply[reply]

I'm getting the distinct preliminary impression that this proposal may be by far the least controversial of my proposals here. - Gilgamesh (talk) 21:25, 8 May 2021 (UTC)Reply[reply]

I already posted this earlier at Talk:Icelandic phonology#Problematic_IPA_for_diphthongs,_especially_long_diphthongs, but realized it's at least as relevant here as it is there. Basically, the existing problem is the superficially ambiguous appearance of long diphthongs [aiː auː eiː œiː ouː]. Pronunciation guides for words in other unfamiliar languages help people who may have generic knowledge of IPA, this kind of notation can throw off the reader, as a gemination symbol ⟨ː⟩ tends to imply that the immediately preceding phonetic unit (and not more than one before it) is geminated, so a reader might look at [aiː auː eiː œiː ouː] and think it implies [a.iː a.uː e.iː œ.iː o.uː]. An alternative transcription, [aːi aːu eːi œːi oːu], isn't really any better because then the reader might think the following vowel symbol represents a separate phonetic unit and that this implies [aː.i aː.u eː.i œː.i oː.u]. Another alternative, [aːi̯ aːu̯ eːi̯ œːi̯ oːu̯], is problematic in that it presumes the geminated component falls entirely on the onset of the diphthong, when the phonology articles and the PDF references in general seem to make no such specific assumption—just that the diphthong as a unit is longer; additionally, the use of [i̯] is problematic in that someone with generic knowledge of IPA and diacritics might assume that those sequences are equivalent to [aːj aːw eːj œːj oːw], as IPA by default associated certain cardinal approximants as being equivalent to certain asyllabic vowels [j=i̯ ɥ=y̯ ɰ=ɯ̯ w=u̯], when such an assumption also cannot necessarily be made in regards to [j] in Icelandic, which means even the short versions [ai̯ au̯ ei̯ œi̯ ou̯] have the same problem. However, using a tie as in [a͡i a͡u e͡i œ͡i o͡u] makes it crystal clear that each of these pairings of vowels are one phonetic unit, so that the long diphthongs [a͡iː a͡uː e͡iː œ͡iː o͡uː] are immediately clearer at a glance. - Gilgamesh (talk) 10:15, 8 May 2021 (UTC)Reply[reply]

Does that mean that both parts of the long diphthongs are equally long, like [aːiː] etc.? We usually tend to drop the tie bars unless they're absolutely necessary. Sol505000 (talk) 17:32, 8 May 2021 (UTC)Reply[reply]

I think so, yes, more or less. But more importantly, I can make no assumptions in the relative length of the diphthongs' individual components, and neither can we. There is precedent for using tie bars for clarity, like for Polish where cz [t͡ʂ] and trz [tʂ] are not strictly homophones. In Icelandic, the spelling æ is the diphthong [a͡i], but the spelling aí (as in Aserbaísjan) is the two-vowel sequence [ai]. One (æ) may be overwhelmingly native while the other (aí) may be found chiefly in loanwords, but I am not assuming they are strictly homophones, either. - Gilgamesh (talk) 19:46, 8 May 2021 (UTC)Reply[reply]

Many (most? all?) Germanic languages require a long vowel to precede any vowel in a hiatus that is not separated by stress. This makes [aiː auː eiː œiː ouː] pretty clear as they are at least to some readers. With that being said, I don't really have an opinion on the tie bar. Sol505000 (talk) 20:46, 8 May 2021 (UTC)Reply[reply]

So if I'm reading you correctly, you mean [Vː.V] and [VˌV] are normal in Germanic languages, but not [V.V]? (What about when both short vowels are unstressed? Anyway...) There's still the problem that editors don't always remember to indicate long diphthongs the same way. Even I sometimes accidentally do ⟨aːi⟩ instead of ⟨aiː⟩ in my edits, because it goes against habit to use ⟨ː⟩ after any multi-symbol IPA that isn't tied. I've also noticed that I'm far from the only editor that does this, whether intentionally or by accident. Introducing inconsistent ⟨aiː⟩ vs. ⟨aːi⟩ breeds ambiguity. But if we consistently tie our diphthongs, not only does ⟨a͡iː⟩ notation look clearer, but we are less likely to accidentally use ⟨aː͡i⟩ notation because it immediately looks more awkward.

Anyway, I'd like to arrive at a multi-user consensus, but...it doesn't seem all that easy to reach a confident consensus with only two users, especially if one of those users admittedly has no opinion on a relevant area. :) I'd prefer we all understand our thoughts on this matter, so that we don't end up alarming third, fourth or fifth editors who sat out this discussion altogether. - Gilgamesh (talk) 21:23, 8 May 2021 (UTC)Reply[reply]

Perhaps not always in unstressed syllables: Stadion is [ˈstɑː.dɪ.ʊn] (with a short /ɪ/) in both Norwegian and Swedish, [ˈstaː.di.ɔn] in Dutch (with a short tense /i/ which, I imagine was long /iː/ in the past) and [ˈʃtaː.djɔn ~ ˈʃtaː.diː.ɔn] (typically transcribed as [ˈʃtaːdi̯ɔn] to indicate the varisyllabic nature of the close front unrounded vowel, or /ˈʃtaːdiːɔn/ in phonemic transcription) in German. Anyway, it's clearly a loanword. I'd have to do some digging before I gave you a more definitive answer. (English stadium [ˈsteɪ.djəm ~ ˈsteɪ.diː.əm] has been variously analyzed as /ˈsteɪdɪ.əm/ and /ˈsteɪdiːəm/ [or with the stupid non-phonemic schwee ⟨i⟩ symbol before the schwa]). Hiatuses are quite common in southern varieties of Standard German (and maybe also in the regional dialects spoken in the area) due to the much less extensive use of the glottal stop in the area. This could also be true of Western Standard German (where mit uns is commonly pronounced [mɪ.dʊns] instead of the prescriptive standard [mɪt.ʔʊns]), though I'm not sure about that.

Per Wikipedia:Manual of Style/Pronunciation#Other languages, we should use one type of transcription consistently.

If we really have to use the tie bars, IMO it's better to put them underneath the diphthong, like this: [a͜i a͜u e͜i œ͜i o͜u a͜iː a͜uː e͜iː œ͜iː o͜uː]. Some of our readers should be familiar with this usage from the 6th edition of Duden's Das Aussprachewörterbuch, where the German diphthongs are written with ⟨a͜i a͜u ɔ͜y⟩. Sol505000 (talk) 07:43, 9 May 2021 (UTC)Reply[reply]

TLDR but oppose any use of tie bars for diphthongs. We should stick to the recommendations of the IPA. Nardog (talk) 09:33, 9 May 2021 (UTC)Reply[reply]

@Nardog: "The recommendations of the IPA". Could you be more specific? - Gilgamesh (talk) 09:43, 9 May 2021 (UTC)Reply[reply]

I don't have a preference as long as it's IPA. The tie bar is not. Nardog (talk) 10:13, 9 May 2021 (UTC)Reply[reply]

@Nardog: What are you talking about? The tie (typography) is part of the International Phonetic Alphabet. What gave you the impression it's not a permitted IPA diacritic? And there are already other IPA transcriptions on Wikipedia that use it (as I mentioned above already), like for Polish IPA to differentiate /t͡ʂ/ (cz) from /tʂ/ (trz) which are not strictly homophones. - Gilgamesh (talk) 15:26, 9 May 2021 (UTC)Reply[reply]

Look at the IPA chart. The ties are for affricates and double articulation. Using them for diphthongs is non-IPA. Nardog (talk) 15:55, 9 May 2021 (UTC)Reply[reply]

@Nardog: That's an interesting take on it. I can't say I would have concluded that tie bars cannot be used for diphthongs just because the chart has no examples using vowels, especially as I have otherwise seen the tie bar used for diphthongs for years. I have never once assumed that the tie bar use for one consonant transitioning to another as parts of an affricate unit, was specifically not also usable with vowels as part of a diphthong unit, nor have I ever heard of this kind of objection before now. The International Phonetic Alphabet article also does not say anything on this use of the tie bar being proscribed or unofficial. Is there a reliable source specifically defining this tie bar usage for diphthongs as non-IPA? - Gilgamesh (talk) 21:02, 9 May 2021 (UTC)Reply[reply]

Don't you realize that's a sort of argument from ignorance? I don't understand why the burden is on me to find a source that says it's not part of the IPA. Regardless, the fact of the matter is the IPA defines the tie bars as indicating an affricate or double articulation only, as not only seen on the IPA chart but also exemplified in the IPA Handbook, p. 22. The definition also appears on pp. 173, 182. The IPA has always defined the tie bars as for affricates from the beginning, and for double articulation as well since 1927, and never for diphthongs. I've never seen an official IPA publication use it for vowels, or I certainly would have included it in History of the International Phonetic Alphabet.

I wouldn't say I've never seen them used for diphthongs, but it's quite rare, and never by anyone closely associated or familiar with the development of the IPA, such as Daniel Jones, David Abercrombie, Peter Ladefoged, or their pupils. I can't recall ever seeing an Illustration of the IPA included in the Handbook or JIPA use them for vowels, and if there was it would be quite an anomaly. I'm curious where you've seen them "for years".

Ladefoged & Johnson (2015), the classic textbook, indeed only mention the tie bars in context of affricates or double articulation (including clicks), as does Laver (1994), who dubs the over-tie "a linker". Turns out, a good deal of textbooks and reference works don't even mention them outside the reprinted IPA charts (Abercrombie 1967, Bussmann, Trauth & Kazzazi 1996, Catford 1977, 2001, Collins & Mees 2013, Crystal 2008, Davenport & Hannahs 2010, Lodge 2009, Trask 1996, Zsiga 2013). Rogers (2000) does, but only in regard to double articulation. The only textbook referencing the use of tie bars for vowels I could find was Ball & Rahilly (1999), who mention it on one occasion and never use or discuss it again.

The IPA wouldn't proscribe it given adaptability is one of the alphabet's strengths (and failure to delineate what's official IPA and what's not is a problem that plagues that article). Authors are free to choose what to do in their transcriptions. But we as an encyclopedia that aims for NPOV and has a worldwide readership cannot count on readers being familiar with nonstandard conventions.

I would read ⟨i͡u⟩ as a simultaneous [i] and [u], i.e. something like [y] or [ʉ], and ⟨i͡uː⟩ as a prolonged version of it, and assume there must be a good justification for such an unusual notation, not least because virtually no other key does such a thing and not least because most of our keys follow the standard IPA even when there exist perhaps more popular language-specific conventions (e.g. Danish and Mandarin). In fact that's what I'd always assumed about Help:IPA/Marshallese, to which you are basically the sole contributor, until seeing this thread. I would genuinely be like, "Wow, what a peculiar language!" each time I stumbled on an article with a Marshallese transcription. And now you're telling me they're just regular old diphthongs? Good lord. I'm not saying it's you, it's definitely me, but bear in mind that's the kind of confusion you would likely invite should you use ties for diphthongs. Nardog (talk) 16:51, 10 May 2021 (UTC)Reply[reply]

Marshallese vowels are...complicated. It's a vertical vowel system. The vowel allophones necessarily involve smooth transition from one realization to another. In reality there are only four vowel phonemes, and we've struggled to find a clear way of transcribing them. But I suppose you're right that I can't prove that this tie usage for diphthongs was ever allowed in the IPA. It had never occurred to me that it wouldn't. I'd be interested in asking you for thoughts on improving Marshallese IPA. Austronesier, Erutuon and I were sporadically working on an overhaul of the system to something more readable (there seems to be no completely conventional IPA system, and even Byron W. Bender's works try to avoid IPA altogether), then covid happened and everyone's lives got turned upside down and all this stuff could no longer be my focus for a while. Before the disruption, my working ideas for Marshallese vowels do away with all the diphthong allophones, and instead transcribe the mid-point vowels of each allophone. For example, for the close vowel phoneme, instead of doing nine allophones [i iɯ iu ɯi ɯ ɯu ui uɯ u], we can instead use one of [i ɨ ɯ u͑ u] (yes, the [iu ui] diphthong allophones were always implicitly [iɯu uɯi] or [iɨɯu͑u uu͑ɯɨi]). The diphthongs themselves were never important, only the secondary articulations of the neighboring consonants, so the mid-point vowel is as good a symbol as any, and in many places actually more graphically similar to the orthography, like [ɨ] for i that neighbors both a palatalized full consonant and a velarized full consonant. At least for vowels between two full consonants; vowels neighboring semiconsonants (the approximant phonemes) is where things can get a little stranger, especially considering that the /j/ and /w/ semiconsonants can sometimes surface as real approximants as needed, but /ɰ/ is considered always a phonetic zero outside how it affects vowels. If and when I resume work on Marshallese article improvement either at Wikipedia or Wiktionary (where I also maintain a rather sophisticated pronunciation module written in Lua), I can invite you to offer your input in general style matters, even if you don't study Marshallese specifically. But back to the topic of IPA in general and Icelandic specifically...

Yes, I've seen sporadic use of tied diphthongs here and there for years, sometimes on Wikipedia, sometimes off. No one before ever suggested to me that it was non-IPA usage. If what you're saying is true and it was never legitimate IPA usage, then it's something I must have picked up through example osmosis in all the years I've been studying and editing about linguistics, which would be at least a good two decades. What you told me here was...strange. "It couldn't actually be the truth." But I can't prove you are wrong. So, for the time being, I'll have to think of a different approach of this. I just know that ⟨VVː⟩ looks really awkward and my mind consistently speedbumps upon reading it. If it really is the case that ⟨VV̯⟩ is preferred for clearly-transcribed diphthongs, we can work around that, even if it ends up being the case that we should use something like ⟨VːV̯⟩ for long diphthongs. ⟨VV̯ː⟩ would still look wrong, as my mind reads it as geminating only a semivowel rather than a whole diphthong. - Gilgamesh (talk) 20:05, 10 May 2021 (UTC)Reply[reply]

But [aiː] is [Vː] like [aː], not [VVː]. The underlying phonemic makeup is just /V/ as length isn't contrastive in Icelandic. Sol505000 (talk) 22:33, 10 May 2021 (UTC)Reply[reply]

Not according to A Short Overview of the Icelandic Sound System, Pronunciation Variants and Phonetic Transcription (2020) by Eiríkur Rögnvaldsson. So far I can find no other references that agree with him, but he asserts that vowel length is becoming contrastive because consonant length is already being lost. (A similar process happened between Middle English and Early Modern English when the language went from a phase where consonant length was phonemic while vowel length was not, to vowel length suddenly becoming phonemic because consonant length lost its contrast.) I wouldn't immediately start transcribing all geminated consonants as single like he does, but if he's even half right, that means vowel length is no longer trivial. - Gilgamesh (talk) 05:49, 11 May 2021 (UTC)Reply[reply]

Fair enough. In that case those are /Vː/. Sol505000 (talk) 06:15, 11 May 2021 (UTC)Reply[reply]

Proposal withdrawn. No fundamental agreement in favor of using tie bars can be reached. - Gilgamesh (talk) 05:59, 11 May 2021 (UTC)Reply[reply]

@Nardog: @Sol505000: I made a new proposal further down, replacing this one. - Gilgamesh (talk) 10:08, 11 May 2021 (UTC)Reply[reply]

The use of mid-open vowel symbols [ɛ ɔ] for monophthongs and mid-close symbols [e o] for the onset of diphthongs [ei ou] has no basis in fact, as shown in Flego-Berkson (2020) p8, where indeed it's entirely common for the onset height of diphthongs to be more open [ɛ͡i ɔ͡u] than for the genuinely mid short monophthongs [e̞ o̞], and for long monophthongs to actually be opening diphthongs [e͡ɛː o͡ɔː] (or [ɪ͡ɛː ʊ͡ɔː]), all within the realm of free variation in a language not known for having major concurrent differences in dialect. This IPA guide already uses a common onset symbol [œ] for the vowels with mid central rounded onsets [œ œi]. (I have a separate proposal for the central vowel symbols, but I won't focus on that here.) So why not just the generic mid front unrounded symbol [e] for all front unrounded mid onsets [e eː e͡i e͡iː], and the generic mid back rounded symbol [o] for all back rounded onsets [o oː o͡u o͡uː]? - Gilgamesh (talk) 10:41, 8 May 2021 (UTC)Reply[reply]

We should absolutely switch over to [ɛi ɔu], yes. Sol505000 (talk) 17:49, 8 May 2021 (UTC)Reply[reply]

I was proposing using the more cardinal symbols [e o] for all such onsets, including for monophthongs. It seems more common for languages without meaningful mid-open and mid-close distinctions (such as in Finnish IPA) to use those symbols for the true mid vowels unless there is a compelling reason (such as in Ukrainian IPA) that the close-mid positions also associated with those symbols overlap with the allophonic space of neighboring phonemes. But for Icelandic, both mid-open and mid-close articulations appear to be free variations of common mid phonemes. As such, I would recommend [e o] replace [ɛ ɔ] entirely. - Gilgamesh (talk) 19:53, 8 May 2021 (UTC)Reply[reply]

I'm ambivalent about whether the symbols should be ⟨ɛ ɔ ɛi ɔu⟩ or ⟨e o ei ou⟩, except for the fact that an actual short cardinal [e] would probably be unambiguously heard as /ɪ/. If we switch over to ⟨e o ei ou⟩ IMO we should also transcribe the mid central rounded vowel with ⟨ø⟩. I'd rather go with ⟨ɛ ɔ ɛi ɔu⟩. Sol505000 (talk) 20:13, 8 May 2021 (UTC)Reply[reply]

Well, what I'm saying is that it would be reasonable to use ⟨ɛ⟩ if [e] were to risk overlapping with ⟨ɪ⟩, as I noted is a concern with Ukrainian IPA. But this doesn't appear to be an issue with Icelandic where mid vowels and near-close vowels are clearly separated, with no merging of /ɛ ɪ/. Icelandic does have certain contact mutations of vowels before certain consonants, but [ɛ ɪ] become [e͡i i] respectively before gi, ng and nk, with no actual risk of /ɛ ɪ/ themselves merging. To be clear, outside of the aforementioned mutation reflexes, the phonemes /ɛ e͡i ɪ i/ are all distinct, and their short and long allophone pairs [e̞~ɪ͡ɛ ɛ͡i~ɛ͡iː ɪ~ɪː i~iː] also remain distinct. This is not that strange, as most Germanic languages (and especially North Germanic languages) have large vowel inventories anyway. Icelandic is relatively unusual for a modern North Germanic language in having no phonemic distinction between mid-close and mid-open vowels. So, all things being equal, I'd rather use ⟨e o⟩ than ⟨ɛ ɔ⟩.

But I will agree that the symbol height of the central rounded vowel ought to agree with the other mid-vowels, so if ⟨ɵ⟩ were not used, but ⟨e o⟩ were used, then ⟨ø⟩ would be more logical than ⟨œ⟩. - Gilgamesh (talk) 21:05, 8 May 2021 (UTC)Reply[reply]

Proposal withdrawn and reorganized. - Gilgamesh (talk) 20:01, 19 June 2021 (UTC)Reply[reply]

As noted in both Flego-Berkson (2020) and in Eiríkur Rögnvaldsson (2020), the onsets of the vowels ö au u are not front rounded [œ ʏ], but actually central rounded vowels, with Eiríkur in particular describing [œ] as having a position closer to that of a true schwa [ə]. Eiríkur also notes that the symbols [œ ʏ] are a matter of tradition, which is why he uses them. However, it's understood that front vowels [ai ɛ ei ɪ i] and the consonant [j] immediately after velar consonants [k kʰ ɣ] trigger those consonants becoming palatal [c cʰ j], but that the "front rounded" vowels [œ œi ʏ] do not trigger this change, with at least one other user in Talk:Icelandic phonology highlighting this confusion. But since these are actually central vowels and not front vowels at all, then using actual central vowel symbols might make this clearer at a glance.

In harmony with my other proposal using common mid vowel symbols for vowels with mid vowel onsets, the most obvious cardinal symbol to replace [œ] would be [ɵ], as it was historically adopted as the symbol for a rounded schwa. If my proposal for common mid vowel symbols is rejected, then it can also be noted that since 1993 there are now two separate cardinal symbols for rounded central vowels: [ɵ] for close-mid, and [ɞ] for open-mid, in harmony with other close-mid symbols [e o] and open-mid symbols [ɛ ɔ]. I wouldn't actually recommend going with separate [ɵ ɞ] for the same reason my mid-vowel symbol proposal exists: It's an artificial distinction without basis in fact.

The matter of IPA for the near-close rounded central vowel (represented orthographically by u and currently indicated with the more frontish symbol [ʏ]) is a little more complicated because there is currently no canonically-adopted IPA symbol for that specific vowel position. All possible symbols would require the use of diacritics [ʏ̈ ʉ̞ ʊ̈ ɵ̝]. However, the unmarked symbol [ʉ] seems to harmonize best with the unmarked symbol [ɵ]. There is an unofficial phonetic symbol [ᵿ] traditionally used in English phonological discussion and by the Oxford English Dictionary, and it has a canonical encoding in Unicode, but it has not been adopted by the IPA, so it would seem irresponsible to suggest its use here.

With the symbols I've suggested, [œ ʏ] could be replaced with [ɵ ʉ] respectively.

The monophthong a (currently indicated [a]) is also a central (unrounded) vowel, equivalent to [ɐ̞] or [ä]. A similar confusion also exists where the diphthong [ai] usually palatalizes a preceding velar consonant whereas the monophthong [a] does not, but I may address that in a separate proposal. - Gilgamesh (talk) 11:26, 8 May 2021 (UTC)Reply[reply]

In our transcriptions of Swedish, ⟨a⟩ stands for a central vowel and, AFAIK, it also doesn't palatalize a preceding velar. Perhaps we could switch over [ai] to [æi] if there are any sources that write it as such.

As for the close and mid central rounded vowels... I'd rather keep transcribing them ⟨ʏ œ⟩. ⟨ʏ⟩ is very commonly used for a close-mid central rounded vowel (so only a slightly lower vowel than in Icelandic) in transcriptions of Dutch, so our readers are probably already familiar with this usage. As far as ⟨ɵ⟩ is concerned, it's less distinct from ⟨ɔ⟩ and ⟨c⟩ than ⟨œ⟩. For that reason alone I'd keep ⟨œ⟩. Also, try approaching it from this angle: in canonical IPA, ⟨ʏ œ a⟩ stand for front vowels. In transcriptions of Icelandic, they all stand for central vowels, except in the diphthong [ai] which is a separate phoneme. Sol505000 (talk) 18:03, 8 May 2021 (UTC)Reply[reply]

I did have ideas for a possible proposal for the open vowels in Icelandic IPA, but I wanted to separate that from this proposal for the time being. And I'm already aware of Dutch IPA using front rounded symbols for central rounded articulations. But what Dutch doesn't seem to have is a contact palatalization of preceding velar consonants. Icelandic does have this, and it specifically involves front vowels but not central vowels. And I don't really see a graphical confusion between ⟨ɵ⟩ and ⟨c⟩ or ⟨ɔ⟩, any more than ⟨ɵ⟩ with ⟨e⟩ or ⟨ə⟩. Swedish IPA uses both ⟨ɵ⟩ and ⟨ɔ⟩ quite happily without graphical confusion.

The tentative idea I had for a proposal on open vowels (which I'm not proposing at this exact moment—I'm still thinking it over) is indeed to have separate onset symbols for palatalizing and non-palatalizing open vowels. My initial idea was [a] for a, [æ͡i] for palatalizing æ (with its notably front onset), [ɑ͡u] for á (with its notably back onset), and [a͡i] for the rarer non-palatalizing diphthong in words like gæi [ˈka͡ijɪ] and skagi [ˈska͡ijɪ]. An [æ a ɑ] front-central-back distinction (like in Bavarian IPA). But, at least in the same spirit of not using separate symbols for non-contrastive [ɛ e] and [ɔ o], the central and back symbols [a ɑ] could be collapsed into one symbol [a], as [ɑ] only occurs as part of [ɑ͡u] anyway. But then again, [æ] also only occurs as part of [æ͡i], so...hmm... (You can probably see why I hadn't finished thinking this over yet.) Another approach could be to use a common symbol [a] for all three open onsets, and just indicate palatalization of consonants or lack thereof where it occurs. But the problem with that is that the difference between [ca͡i] and [ka͡i] appears superficially opaque if you only use one symbol. But it's also true that palatalizing [æ͡i] and non-palatalizing [a͡i] only contrast in contact with a preceding velar consonant. So, in summary, the different possible options I'm thinking of:

• [æ͡i a a͡i ɑ͡u]
• [æ͡i a a͡i a͡u]
• [(c)a͡i a (k)a͡i a͡u] (most similar to current IPA)

My gut would go with [æ͡i a a͡i ɑ͡u], but I haven't reached a level of confidence to actually propose any of these, which is why I've currently left it out of my central rounded vowels proposal. - Gilgamesh (talk) 20:24, 8 May 2021 (UTC)Reply[reply]

Proposal withdrawn and reorganized. - Gilgamesh (talk) 20:01, 19 June 2021 (UTC)Reply[reply]

I previously proposed and withdrew a different proposal involving tie bars, but it became clear that no agreement could be reached in favor of using tie bars on diphthongs. But the issue of clarity of existing ⟨VVː⟩ notation (particularly how my mind always speedbumps when I read it) compels me to still suggest some form of clarification. I also previously suggested that we can't assume one component of a diphthong has more emphasis than the other, and I still can't see how we can assume they do. But lacking a better option for diphthong clarity, I propose the notation [ai̯] for short diphthongs and [aːi̯] for long monophthongs. I do not favor this option (the tie bars still looked clearest to me), but this is still clearer than bare [aiː] notation. - Gilgamesh (talk) 06:12, 11 May 2021 (UTC)Reply[reply]

It seems the otherwise unmarked ⟨VVː⟩ diphthong notation also appears to be conventional for long diphthongs in Faroese IPA, while Old English IPA uses ⟨VːV̯⟩ for its long diphthongs. Of course, the distinction is that both Icelandic and Faroese are living languages, while Old English is an ancient language whose phonological representation is largely academic. I'm honestly not entirely sure which option is the wisest course.

But it also seems, overall, that if the tie bar is excluded, then IPA seems hobbled in its ability to elegantly represent smoothly-transitioning diphthongs. I had considered the tie bar the most logical option not just for Icelandic diphthongs but most for diphthongs in most languages because (1) I'd already thought it was accepted usage through all the examples I'd already seen (example osmosis) and before recently I'd never heard it suggested otherwise, and (2) There seemed to be an intuitively straightforward analogy between tied diphthongs and tied affricates, at least as a hindsight justification for usage I had already long occasionally seen in practice. It is true that some uses of the tie bar imply dual articulation for other languages, like [k͡p]. But the tie bar in affricate notation, like [t͡ʃ], is not strictly a simultaneous articulation of the symbols, but a unitary sequence of one sound followed by the other in such a way that they can't be mistaken as belonging to separate syllables. After all, if you see [t͡ʃ], you don't read it the same as if it were [ʃ͡t], because it's not, strictly speaking, a simultaneous articulation of two different sounds in different parts of the mouth, but [t] seamlessly followed by [ʃ]. When an affricate is geminated, the notation I've seen for Faroese appears to be [t͡ʃː] (though it also seems worth mentioning that a similar sound can be represented in a language like Italian as [ttʃ]). Likewise, since two vowels in general cannot be genuinely dually articulated because it involves differences in the shape of the mouth overall, it logically followed that notation like [a͡i] can only reasonably refer to diphthongs. After all (and this example is not particular to any language), why use [ɛ͡e] to bizarrely notate a monophthong equidistant between [ɛ] and [e], when you could just use [ɛ̝] or [e̞]? A tie bar for diphthongs was a principle that never seemed to require any special logic leap, but rather seemed like a natural inference of IPA's allowances. I have attempted to address @Nardog:'s reservations about tied diphthongs, and I don't necessarily expect I can persuade him personally on this matter, but it still seems difficult for me to endorse the contradictory logic that a tie bar can be used for [k͡p] as a simultaneous articulation on one hand and for [t͡ʃ] as a smooth unitarily-articulated consonant sequence (an affricate) on the other hand and yet not also for [a͡i] as a smooth unitarily-articulated vowel sequence (a diphthong). I still cannot prove that tied diphthongs are officially approved IPA use, but short of a specific official reference discouraging them, and considering that tied diphthongs have already occasionally been in use since before my time, it never seemed necessary to require approval for what appeared to be a simple logical inference. To be confused by such a thing would be like...someone being in any way confused when someone refers to an audibly lengthened vowel as "geminated," when in truth that is exactly what happens when any sound (whether it be a consonant or vowel) is articulated for an audibly longer-than-normal duration. Consensus questions like these need not necessarily be a matter of disagreement according to individual taste in IPA notation, but a matter of what a hypothetical third party's conventional logical process is likely to infer from a notation, and whether a notation can be considered graphically clearer than its alternatives. And, with all due respect, I believe this is an important concern that deserves to be considered in full and not treated with a "tl;dr" and reflexive dismissal. (I also have an autism spectrum-related communications quirk, and as such, succinctness is not necessarily my greatest strength.) - Gilgamesh (talk) 07:39, 19 May 2021 (UTC)Reply[reply]

Proposal withdrawn. - Gilgamesh (talk) 20:01, 19 June 2021 (UTC)Reply[reply]

I may have made a mistake in making so many individual proposals and treating them like they could be independently discussed. But the truth is, all my proposals are linked. So here is one unified proposal with a pithier list:

• ⟨á⟩ [au] → [a͡u]
• ⟨au⟩ [œi] → [ɵ͡i]
• ⟨e⟩ [ɛ] → [e]
• ⟨ei⟩, ⟨ey⟩ [ei] → [e͡i]
• ⟨o⟩ [ɔ] → [o]
• ⟨ó⟩ [ou] → [o͡u]
• ⟨u⟩ [ʏ] → [ʉ]
• ⟨æ⟩ [ai] → [a͡i]
• ⟨ö⟩ [œ] → [ɵ]

Examples:

• Brennisteinsfjöll [ˈprɛnnɪˌsteinsˌfjœtl̥] → [ˈprennɪˌste͡insˌfjɵtl̥]
• Eldgjá [ˈɛltˌcauː] → [ˈeltˌca͡uː]
• Fœreyjar [ˈfaiːrˌeiːjar̥] → [ˈfa͡iːrˌe͡iːjar̥]
• Holuhraun [ˈhɔːlʏˌr̥œiːn] → [ˈhoːlʉˌr̥ɵ͡iːn]
• Jökulhlaup [ˈjœːkʏl̥ˌl̥œiːp] → [ˈjɵːkʉl̥ˌl̥ɵ͡iːp]
• Kirkjubæjarklaustur [ˈcʰɪr̥cʏˌpaiːjarˌkʰlœistʏr̥] → [ˈcʰɪr̥cʉˌpa͡iːjarˌkʰlɵ͡istʉr̥]
• Marskálkseyjar [ˈmar̥skaul̥ksˌeiːjar̥] → [ˈmar̥ska͡ul̥ksˌe͡iːjar̥]
• Neskaupstaður [ˈnɛsˌkʰœipˌstaːðʏr̥] → [ˈnesˌkʰɵ͡ipˌstaːðʉr̥]
• Sauðárkrókur [ˈsœiːðˌaurˌkʰrouːkʏr̥] → [ˈsɵ͡iːðˌa͡urˌkʰro͡uːkʉr̥]
• Sogið [ˈsɔijɪθ] → [ˈso͡ijɪθ]
• Söguöld [ˈsœːɣʏˌœlt] → [ˈsɵːɣʉˌɵlt]
• Tálknafjörður [ˈtʰaul̥knaˌfjœrðʏr̥] → [ˈtʰa͡ul̥knaˌfjɵrðʉr̥]
• Þjórsá [ˈθjour̥sˌauː] → [ˈθjo͡ur̥sˌa͡uː]
• Þorbjörn [ˈθɔrˌpjœ(r)tn̥] → [ˈθorˌpjɵ(r)tn̥]

Now, my more in-depth rationales. These details are considerably less pithy.

Rationale for ⟨au⟩ [œi] → [ɵ͡i], ⟨u⟩ [ʏ] → [ʉ] and ⟨ö⟩ [œ] → [ɵ]:

• According to Flego, Stefon; Berkson, Kelly (July 2020), "A Phonetic Illustration of the Sound System of Icelandic", page 8, these are central vowels, not front vowels. (In the case of ⟨au⟩, the diphthong's onset is still central and largely coincides with ⟨ö⟩.) This is corroborated at least for ⟨ö⟩ by Eiríkur Rögnvaldsson (2020), "A Short Overview of the Icelandic Sound System, Pronunciation Variants and Phonetic Transcription", page 5, which describes it as being close to true schwa.
• The front unrounded vowels ⟨e⟩, ⟨ei⟩, ⟨ey⟩, ⟨i⟩, ⟨í⟩, ⟨y⟩, ⟨ý⟩, ⟨æ⟩ and the consonant ⟨j⟩ regularly trigger palatalization of preceding velar consonants ⟨g⟩ and ⟨k⟩, but the central vowels do not.
• The onset of the diphthong ⟨au⟩ [ɵ͡i] coincides with the (short) realization of the vowel ⟨ö⟩ [ɵ]. This appears to have been historically true as well, with Old Norse ⟨au⟩ having been */ɔ͡u/ when historic ⟨ǫ⟩ was */ɔ/. Why the differences in spelling then? Modern Icelandic orthography was standardized largely on the far more recent standard normalization of Old Norse as an otherwise dead literary language. This standard normalization prescribed the spelling ⟨au⟩ where *⟨ǫu⟩ may have always been more phonologically accurate. (A modern Icelandic spelling of ⟨öu⟩ might have certainly been less confusing, but things are what they are.)
• The use of central IPA symbols for these vowels is not only more accurate, but reduces confusion associated with the principle that front vowels trigger palatalization. Since these central vowels are not truly front vowels and in any event also do not trigger palatalization, they shouldn't be notated with front vowel symbols.
• In the case of ⟨u⟩ [ʏ] → [ʉ], the vowel is actually a near-close vowel [ʉ̞]. There are no dedicated official IPA symbols for central near-close vowels. The symbol ⟨ᵿ⟩ would seem like a more natural choice, and is used in some English dictionaries as an archiphoneme for the free variation of /ʊ~ə/, but the symbol has never been officially adopted into the IPA.

Rationale for ⟨e⟩ [ɛ] → [e], ⟨o⟩ [ɔ] → [o] and ⟨ö⟩ [œ] → [ɵ]:

• The use of open-mid IPA symbols for monophthongs ⟨e⟩, ⟨o⟩ and ⟨ö⟩ has always been artificial and not based in reality, especially when compared to close-mid symbols for the equivalent onsets of diphthongs ⟨ei⟩, ⟨ó⟩ and ⟨au⟩. If anything, the monophthong allophones are usually less open than the onsets of their equivalent diphthongs, as the diphthong onsets can commonly be as low as [ɛ͡i, ɔ͡u, ɞ͡i], and the monophthong long allophone offset can be as high as [e͡ɛ, o͡ɔ, ɵ͡ɞ] or even [ɪ͡ɛː, ʊ͡ɔː, ʉ̞͡ɞː]. But in all these cases, the vowel height of the mid vowel is subject to a certain free variation which seems beyond that of even a [phonetic] IPA transcription, though I could easily see this changing in the future with regard to the monophthong long allophones. The simplest allophones, the monophthong short allophones, are true-mid vowels [e̞, o̞, ɵ̞], and since they do not truly contrast against separate close-mid or open-mid vowels, and can allophonically drift into both close-mid and open-mid positions, it seems least complicated to use the unmarked symbols [e, o, ɵ] for all mid monophthongs and equivalent diphthong onsets.

Rationale for ⟨á⟩ [au] → [a͡u], ⟨au⟩ [œi] → [ɵ͡i], ⟨ei⟩, ⟨ey⟩ [ei] → [e͡i], ⟨ó⟩ [ou] → [o͡u] and ⟨æ⟩ [ai] → [a͡i]:

• All Icelandic vowels, diphthongs included, have short and long allophones. The traditional IPA notation for diphthong long allophones, ⟨aiː⟩, is ambiguous because ⟨ː⟩ modifies the immediately preceding IPA symbol, so to the readers with only generic IPA knowledge, the entire sequence can easily be misread as ⟨a.iː⟩. There is also a common tendency for such sequences to be misnotated as ⟨aːi⟩ with the intended meaning of ⟨aːi̯⟩. I previously proposed ⟨ai̯, aːi̯⟩ notation as an alternative, but the non-syllabic diacritic involved is easily overlooked especially if it precedes ⟨j⟩ or ⟨ˌ⟩. By contrast, a consistent tie bar for ⟨a͡i, a͡iː⟩ makes it clear that "this diphthong is a phonological unit." There has been some noted dispute over whether it is valid to use a tie bar for diphthongs. Nardog argues that the tie bar was only ever intended for consonants, because official IPA references only give examples for the tie bar using consonants. I argue that a diphthong is the vowel equivalent of an affricate consonant, where one sound is closely followed by another sound as part of a combined unitary sound, and that this is a simple logical inference that never needed special approval. And, as a third user has also mentioned, tie bars for diphthongs have already seen occasional use in published linguistics materials for German.

Special mention for ⟨a⟩ [a], ⟨á⟩ [a͡u] and ⟨æ⟩ [a͡i]:

• The frontness of each the onsets of each of these vowels varies considerably, with [a͡i] actually being true front ⟨æ⟩ [æ̞͡i], ⟨á⟩ [a͡u] actually being true back [ɑ͡u], and ⟨a⟩ [a] actually being true central [ɐ̞]. The sequence [a͡ijɪ] can also arise from the spelling ⟨agi⟩, but its onset is true central [ɐ̞͡ijɪ]. To add to the inconsistency, some loanwords with ⟨æ⟩ actually have [ɐ̞͡i] instead of [æ̞͡i]. But besides the fact that an [æ̞] onset palatalizes a preceding velar consonant whereas [ɐ̞] and [ɑ] do not, none of these onsets actually meaningfully contrast, and the symbol ⟨a⟩ is selected as the free variation symbol for all three possible onsets, with any palatalization reflected only in the symbol used for the preceding consonant where applicable. I have not seen fit to propose this status quo be changed.

Special mention for geminated consonants:

• Icelandic has the strange situation where doubled consonants (using ⟨dd⟩ as an example) have geminated morphophonemes ⫽dd⫽ and noticeably geminated surface phonetes [tt], but, if Eiríkur (2020) is correct, non-geminated intermediate phonemes /t/. It would seem that Icelandic is at a point of transitioning from being a language where vowel length is non-phonemic, to becoming a language where it is instead consonant length that is non-phonemic; a similar transition happened to most other Germanic languages in the Middle Ages, including during Middle English, with Icelandic only now appearing to do the same. So perhaps it is true that [t] and [tt] may not meaningfully contrast even if they still noticeably differ, which I believe can be characterized as a moribund contrast—something still audibly productive in speech but no longer necessarily perceived by its own speakers, and may very well vanish altogether without a trace in future generations' speech. But for the time being, considering that gemination still exists at the orthographic, morphophonemic and surface phonetic level, I would be inclined to still notate geminated consonants, at least in a [phonetic] transcription (which this template uses), for the sake of an added clarity that is not actually incorrect and does no real harm.

This all forms the basis of my combined repackaged proposal, with both pithy and detailed segments. Any part of it can be discussed, with suggestions for modification, etc. I just really want to do this by an organized and engaged consensus process if at all possible. - Gilgamesh (talk) 06:13, 19 June 2021 (UTC)Reply[reply]

This seems well argued. -- Rei (talk) 11:54, 28 November 2023 (UTC)Reply[reply]

In this article https://www.researchgate.net/publication/282778942_Phonetic_Transcription_Guideline_Icelandic "au" is represented as [œy] which also corresponds to my understanding of the IPA and the Icelandic language. Hugstar (talk) 15:46, 28 September 2023 (UTC)Reply[reply]

Done. Árnason (2011: 58) says the rounded quality in the second component of /œi/ is "spread from the [œ]-component as a phonetic effect due to co-articulation" but nonetheless uses [œy] in phonetic transcriptions. Nardog (talk) 05:25, 30 September 2023 (UTC)Reply[reply]

The page currently lists 'boot' for 'U', and 'food' for 'Ú', but "boot" and "food" are both the same sound, a close back rounded vowel, at least in the English I know. 'Ú' is indeed pronounced like "food" (and "boot"), but "U" is pronounced like the I in "pig", but said with rounded lips like 'Ö'. Example here - "Kaffið er þunnt".

'U' is only pronounced like "boot" when it comes before 'NG' or 'NK' - see here. -- Rei (talk) 12:57, 21 November 2023 (UTC)Reply[reply]

ED: I see that someone further down proposed (ʉ) / close central rounded vowel. This sounds closer to accurate to me. Isn't it actually ⟨ʉ̞⟩ / near-close central rounded vowel? That sounds most correct to me, more in the direction of [⟨ø⟩ / Close-mid_front_rounded_vowel, aka Ö]. Other Icelandic speakers (or ideally scholars), what's your take?

All I can say with absolute certainty is, the way it currently stands, "boot" and "food" have the same vowel sound to me, and that vowel sound is "Ú". -- Rei (talk) 14:41, 21 November 2023 (UTC)Reply[reply]

I went ahead and made the changes, since there was no input from anyone else. I also noticed what the problem was: the example of ungur fell afoul of the ung/unk rule, which makes it pronounced like 'ú'. But that's an exception, not the general case. I replaced the example with 'full', to stand in contrast with 'fúll'.-- Rei (talk) 11:40, 28 November 2023 (UTC)Reply[reply]