Hi. Your Icelandic IPA additions contain sufficiently many errors that it would be preferable not to add them. If you assemble a list of IPAs I could review them for you, but reviewing them after they've been inserted into articles is too time consuming. – Thjarkur (talk) 13:11, 17 May 2021 (UTC)

@Þjarkur: I can fix IPAs. And I have been trying to reach out to other editors in the past for help and review, but I've had so few answers. Could you tell me some of the patterns of errors in particular I've been inserting? I really would like to know, and have my edits reviewed in general. - Gilgamesh (talk) 13:22, 17 May 2021 (UTC)

@Þjarkur: I looked at some of your recent contributions to see what IPA you've corrected, and I've been noticing some patterns.

• I do mark final -n(n) as voiceless, but I was told that final sonorants are devoiced like this at a pause. Are you saying this only affects -r and -l? I'm sort of presuming that if it doesn't affect -n, then it probably doesn't affect -m, either.
• Devoiced r in -rs. My reference for this was A Short Overview of the Icelandic Sound System Pronunciation Variants and Phonetic Transcription by Eiríkur Rögnvaldsson (2020). I did specifically cover the pattern that r was devoiced before s...but didn't say that was only within a morpheme. I applied the devoicing across morpheme boundaries, like in Skeiðar·ár·sandur, etc., and apparently that was wrong?
• Lómagnúpur. Yeah, I do remember fumbling a bit on that. I tried to google the name's etymology, look up the individual morphemes in dictionaries. After all this, I thought it was divisible as Ló·mag·núpur, but I guess I was wrong and it's Lóma·gnúpur. Morpheme boundaries can be a little tricky to discern when you're not a native speaker, which is why I've specifically opted not to add IPA for some terms I couldn't figure out in that regard. I did, however, fix a previous error on my own, in Stuðlagil, where I had previously divided the word as Stuð·lagil instead of Stuðla·gil.
• In Uxatindar, etc. The [xs] vs. [ks] issue also affects other Wikipedia articles concerning Icelandic. Apparently [xs] is...traditional? But Eiríkur Rögnvaldsson (2020) says few people under the age of 60 use it, and most people use [ks]. Since the Icelandic phonology article (which I've complained in Talk:Icelandic phonology is kind of a mess) only mentions [xs], I was going under the assumption that that is traditional IPA. But I couldn't ignore what I learnt about [ks], so I included both pronunciations.
• In Eyja- og Miklaholtshreppur, etc. I was, again, going by Eiríkur Rögnvaldsson (2020) which indicates that the t in lts (within a morpheme, I'm assuming) may be silent. And I previously found inconsistencies in this regard in IPA already present in particles, like Surtsey (someone else's IPA with [r̥ts]) and Surtshellir (someone else's IPA with [r̥s]). Going by the reference, I replaced these examples with [r̥(t)s], and used that going forward.
• In Okjökull (and I suspect also Langjökull), etc. None of the references suggested that palatalization of k and g did not occur across morpheme boundaries. I previously specifically tried to ask people for help on Lang·jökull, but, as has often been the problem, I couldn't find people knowledgeable in it. I then turned to audio clips for guidance. I don't know if [kˌj] is formal and [cˌ] is casual or whatever, so I...used my best judgment based on the materials available to me.
• In Arnavatnsheiði, that was just an inconsistency. The article is titled Arnavatnsheiði, but the lede calls it Arnarvatnsheiði, and I didn't notice that difference until right now. Iceland-related articles in general probably need a lot more review than they've been getting. But I did get vatns right as [vass] (again, thanks to Eiríkur Rögnvaldsson (2020)), as a less educated non-native guess might have been [vahtn̥s].
• In Dalvíkurbyggð and other -byggð articles. I read that final [ð] becomes [θ] in pause, and that final [ɣ] is [x] in pause, but I would have never would have guessed that final [ɣð] would become [ɣθ] instead of [xθ].
• In Svalbarðsstrandarhreppur, that just seems...unpredictable. I divided the word among morphemes (Sval·barðs·strandar·hreppur), and gave sval a long vowel because it ended with only one consonant. What factor made it short? Was it the b at the start of the next morpheme creating a cross-morpheme consonant cluster lb? Or is svalbarðs itself counted as a single morpheme? Or is sval a special morpheme with different vowel rules? All very good questions. Not always easy to find answers for. And, again, Iceland-related articles seem to suffer overall from a lack of attention and review.
• In Grjúpán. Looks like another morpheme boundary confusion. Grjúp·án, not Grjú·pán.
• In Iceland itself. You may have to forgive this confusion. I keep seeing the IPA for Ísland indicated inconsistently in lots of different places. [ˈistlant]? [ˈiːsˌlant]? [ˈiːstˌlant]? It now seems apparent that Ísland is to be treated as a unitary morpheme rather than divided into Ís·land.
• In Ófærufoss. Again, morpheme boundary confusion. But this time I made the mistake of looking at how the name is spelt in interwikis for a hint. In retrospect, if it actually had been Óf·æru·foss instead of Ó·færu·foss, the first f might actually have been [(v)] instead of [v] (going by Eiríkur Rögnvaldsson (2020)).
• In Barðsneshorn, etc. The issue of IPA for ðs can be a bit...confusing. In the audio clips I listen to for words like that, it keeps sounding like they're saying [θs].
• In Egil's Saga, etc. Yeah...that particular confusion came from Wiktionary, of all places. Not specifically the name Eigill, but so many other terms with that diphthonging reflex. (Wiktionary also has its own issues with consistency.) Eiríkur Rögnvaldsson (2020) says the lack of diphthonging is mostly regional (Southern Iceland), but Wiktionary's Icelandic rhymes pages more vaguely say it's some speakers. So whenever I saw any combination of a stressed vowel before gi within the same morpheme, I did both variations.
• In Bárðar saga Snæfellsáss. That again was the cross-morpheme -r-s- confusion. I already knew that devoicing was only pause-final.
• In Hoffellsjökull. Hoffell being treated as a single morpheme (instead of two as Hof·fell) was, again, hard to predict. I did notice that when you corrected the pronunciation, you used single [f] instead of double [ff]. That's the first suggestion I've seen since Eiríkur Rögnvaldsson (2020) that double consonants within a morpheme may be losing their contrast. All the other linguistics references I consulted didn't even cover that, and generally (along with Icelandic phonology article) treated double consonants as truly geminated. Is it always only possibly [f], or can it also be [ff]? - Gilgamesh (talk) 14:48, 17 May 2021 (UTC)

I think that covers most of the issues you had with my IPA. If there any others, then please tell me. The general problem here is that, whenever there's an abrupt renewed international interest in a place, one of the first things people wonder is "how is that pronounced?" And when it comes to Iceland topics...that can be really complicated. Too many questions, not enough editors, and not enough knowledgeable people to consult. At some point even an imperfect editing effort can be better than no editing at all (WP:BOLD), though I do recognize that too many well-meaning haphazard lone edits can cumulatively deteriorate an article's quality, which especially seems true for the article Icelandic phonology. That article in particular seems to have been the source of a lot of misconceptions that other editors propagated to other articles, including the notion that all sonorants are voiceless after voiceless consonants in all positions (e.g. *[ˈeiːjaˌfjatl̥aˌjœːkʏtl̥]), though I did try to fix that myself when I learnt it wasn't strictly true. - Gilgamesh (talk) 14:37, 17 May 2021 (UTC)

@Þjarkur: I've been browsing through Special:WhatLinksHere/Template:IPA-is to see what IPA pronunciations I can repair as per these notes. I just reached Stöðvarfjörður, where I can't say with enough certainty that that pronunciation is correct. Is it correct? - Gilgamesh (talk) 21:00, 17 May 2021 (UTC)

Also, Seltjarnarnes. I only ask because I originally couldn't find a clear dictionary entry for sel, so I don't know if elongating its vowel is appropriate in this context. - Gilgamesh (talk) 21:29, 17 May 2021 (UTC)

I'm also not entirely certain the IPA for Þaralátursfjörður is correct. - Gilgamesh (talk) 22:37, 17 May 2021 (UTC)

@Þjarkur: All right, I did a largely thorough sweep through many different articles' IPA. But one thing you need to understand is that edits like these aren't just going to stop, whether they're by me or someone else. They will be well-meaning, but not always accurate. I ended up correcting all sorts of other people's edits that were far worse. It's true that Iceland topics often don't have enough review by other editors, but that still can't be a reason not to try to edit them. Again, WP:BOLD. There's no replacement for discussion, review and verification involving multiple users. I welcome these, and have searched them out where I could, but this relies on others responding and participating. And, unfortunately, experts on Icelandic linguistics seem few and far between, so there've been so many questions that have simply gone unanswered for a long time. Now, a few of my IPA entries may have had mistakes, but most of them are accurate, no? I'm doing something other editors haven't seen fit to do on this scale before now, and I'm trying to be as informed as I can going forward. But I need review. All these articles need review. Icelandic phonology needs review (and probably an entire rewrite). Icelandic orthography needs review. Help:IPA/Icelandic needs review. The various talk pages are an ideal places to do this. I have opened various discussions at Talk:Icelandic phonology and at Help talk:IPA/Icelandic to address some things, and for the most part they haven't gotten much response, and those that have haven't adequately addressed all the points of concern. But with all due respect, talk pages are not places to ask editors not to edit just because there may not be enough time for other editors to review their edits for possible errors. Edits happen on public wikis regardless of whether there's someone else to review them. - Gilgamesh (talk) 23:46, 17 May 2021 (UTC)

@Þjarkur: Some more questions where I'm not entirely certain.

• Bifröst (town) and Bifröst University. Since ​Bifröst seems to have originally been a mythological term, it's obviously not a new term. Is it to be treated as a unitary term, as [ˈpɪvrœst]? Or is it split up, like Bif·röst [ˈpɪːvˌrœst]?
• Maríuhöfn. Right now it's [ˈmaːriʏˌhœpn̥]. Should it be [ˈmaːriːʏˌhœpn̥]?

I will let you know if I think of any other uncertain edits. - Gilgamesh (talk) 13:14, 19 May 2021 (UTC)

@Þjarkur: Requesting review for Icelandic chicken. In particular, the IPA given for Haughænsni. - Gilgamesh (talk) 18:11, 20 May 2021 (UTC)

Gunnars saga Keldugnúpsfífls. I think I got the pronunciation right, but it should be reviewed. - Gilgamesh (talk) 00:52, 21 May 2021 (UTC)

@Thjarkur: Should I have been pinging only this name all this time? As you can see, I can be a bit of a restless binge editor, but I do care about the accuracy of my edits, which is why initially even when you didn't respond further, I browsed your own user contributions to glean for specific evidence of my own mistakes, and used those to try to correct other mistakes I'd made. But it would help to know when I'm on the right track, and when I'm on the wrong track. So many questions, not enough answers, from anyone it seems. This is a super-fascinating language, and I've already collected several linguistics PDFs on it to aid me, but it can be so hard to find interactive human experts to ask followup questions. Icelandic names and terms only have so much usefulness outside a purely textual context if a person doesn't know how to pronounce them. - Gilgamesh (talk) 17:19, 22 May 2021 (UTC)

Need review of IPA for Álagablettur. - Gilgamesh (talk) 20:10, 22 May 2021 (UTC)

Hæ hæ, either pinging method works but I had just left town for a few days. The issues you listed above are indeed the main issues. It is not possible to know how a term is split or whether adjacent terms influence vowel length without knowing the language. Additionally, consonant length differences are either so miniscule or non-existant that consonant length is not denoted in the standard broad transcription. "r" vs "rr" is sometimes perceptible, but "sólin" and "sólinn" are the same.

I disagree that error-prone edits are better than no edits, especially at a massive scale. There is no one reviewing these articles except occasionally me. An editor's talk page is the correct location to let them know they're out of their depth. Your recent edits still contain too many errors and I would honestly recommend reverting your additions. WP:BOLD is not an excuse for making massive changes on a topic one isn't sufficiently familiar with.

– Thjarkur (talk) 20:43, 23 May 2021 (UTC)

Problem is, what you propose would doom so many terms to the arcane realm of "esoterically unpronounceable." There's been such a recent surge of international interest in Iceland-related topics that it's really no longer acceptable for the vast majority of terms to have no educated attempts whatsoever to pronounce them. And when people don't know how, they'll try to find out how. Which initially means consulting the (still highly problematic) Icelandic linguistics articles, and beyond that, published papers on the subject. (I myself have collected these materials for both guidance and for use as references.[1][2][3][4][5]) And, for principled and especially dedicated editors, going further still to eagerly learn from their mistakes, which in itself necessarily requires the freedom to make mistakes in the first place. These articles need review. They need experts. And if no experts step up, we search them out, because we're curious, we crave knowledge, we want to know these things.

And it's not just lack of pronunciation guides that appears to have plagued Iceland-related articles. It's also a general shortage of references, a tendency to editorialize in article bodies, and an osmosis of errors propagated by editors based on what's already elsewhere on the wiki. I recognize I'm not immune to this trend myself, but I have put great effort into fixing errors where I see them, improving problematic language, and flagging articles and statements for improper tone, lack of citation, or dubiousness, and, most importantly, learning and improving myself as a learner, researcher and editor as I go. The issue of pronunciation is far more nebulous still, because up until now there seems to have been very little effort by anyone to address this issue on a Wikipedia-wide basis outside a very small selection of terms. My general goal is to leave these articles in a relatively better condition than when I found them, and that's what I've been doing. "Relatively" being the key term, as it implies an improvement that itself may still be flawed but is better than what came before.

Now, if there's a problem, then please, by all means, help out by correcting articles, answering questions, sharing notes, because other editors will tend to involve themselves no matter what, and we do need additional participation by editors who have a keener knowledge about the topics, but more willing editing manpower is also still welcome. Wikipedia is an open wiki that anyone can edit precisely because its predecessor project, Nupedia, could only be developed at an unacceptably slow pace and tiny scope when all edits had to be reviewed or approved by experts. The Wikipedia process allows for more possible error to creep up in articles (not to mention the occasional outright vandalism), but at the tradeoff of greater speed, participation, and ultimately also review of existing articles. Writing or editing a random article doesn't automatically make it a good article, but the underlying freedom to edit it is still necessary. It's the review and collaborative improvement process that can turn random articles into good articles. In the meantime, article edits cannot be prevented or discouraged just because there could be errors and not be enough editors with enough time to review them all. Honest efforts to improve articles, even when they are imperfect, should be welcomed and encouraged, with review and correction given by others as needed until a random article becomes an above-average-quality article, to hopefully a good article. - Gilgamesh (talk) 21:58, 23 May 2021 (UTC)

All right, so...reviewing some of your linguistics points.

So final unstressed (at least unstressed) -inn is always [-ɪn], not [-ɪnn]? I'll keep that especially in mind going forward, and fix existing IPAs with [-ɪnn] when I reencounter them.

But as for doubled consonants themselves... There've been a lot of inconsistent details about that. Icelandic phonology (already acknowledged to be problematic) uses doubled consonant IPA [CC] only, and it was like that even before I got there. Wiktionary entries for Icelandic terms use [Cː] notation, so pretty much the same deal. As for the external references, Eiríkur Rögnvaldsson (2003) and (2020) use ungeminated consonant IPA, but Gussmann (2011) and Flego–Berkson (2020) use geminated consonant IPA, and the Flego–Berkson in particular present acoustic data on the audible durations of short and long vowels and single and double consonants. It seems there are different approaches to notating doubled consonants, and the Icelandic phonology article, Wiktionary, Gussmann and Flego–Berkson all preferring to show geminated consonant IPA, and Help:IPA/Icelandic in particular has noted (not my edits) that vowel length is not phonemic, though I can appreciate how that can change if consonant length distinction becomes lost. Now, it's possible there could be a lot of errors all over these places, and that you may be entirely correct, but it still remains that the only third party reference I've been able to find that insists on single consonant IPA comes from just one person, Eiríkur Rögnvaldsson. Additional third party citations and verification might be helpful, otherwise it contradicts other references and (for what they're worth) previously established practices on these wikis. And, yes, assuming you are fluent in Icelandic, I know it may sound strange for a non-speaker to request additional verification. But quite simply, the readily readable references are contradicting each other on whether consonant length is to be considered contrastive. When there's such a conflict of references, I'm inclined to use the notation that preserves rather than eliminates contrasts.

Now when it comes to knowing how to divide compound words into morphemes, I'm gradually picking this up especially for geographical terms with recurring vocabulary (á, blá, borg, breiða, brú, bær, dalur, ey, fjall, fjörður, foss, gjá, göng, hlíð, hraun, hreppur, hús, hver, höfn, jökull, kirkja, land, nes, rauð, reykja, vatn, vegur, vík, etc.) and their inflections (nominative, genitive, singular, plural, gender, definite articles, other recurring suffixes, etc.). Most of these haven't been difficult to figure out, as there have been a lot of repeating patterns, and most of these morphemes have full declension tables at their Wiktionary entries and other online Icelandic dictionaries. If I'm really uncertain where terms are divided up, I may ask for help, as I've already started doing since you initially contacted me. - Gilgamesh (talk) 03:13, 24 May 2021 (UTC)

@Þjarkur: I went to revisit some more of my own IPA edits, and I got to Gunnars saga Keldugnúpsfífls, saw the [knups], and thought, "oh, that must have been one of my brainfarts from earlier," and changed it back to [knufs]. But afterwards I had a hunch, checked the edit history, and saw that you were the one who had changed what actually been my own earlier [knufs] to [knups] for gnúps. Now, I can understand [xs] having become [ks] in recent generations, as this was specifically noted by Eiríkur Rögnvaldsson (2020) as a younger trend. (Previously all existing Wikipedia examples had used [xs] before I started adding [ks] as an alternative based on Eiríkur Rögnvaldsson (2020), which was itself before you started changing them all to [ks].) But the preexisting Icelandic orthography article has [scɪfs] for skips, and as a rule p becoming [f] before s, k or t. And that same Eiríkur Rögnvaldsson (2020) reference doesn't specifically address any similar case for [fs] vs. [ps]. So you're saying that the p being [f] before s rule is...not a rule at all? It's false? If it's actually false, then the Icelandic linguistics articles may be in even worse shape than I thought. - Gilgamesh (talk) 12:34, 24 May 2021 (UTC)

@Þjarkur: I see you edited Help:IPA/Icelandic. But you haven't addressed the concerns about conflicting references regarding geminate consonants. We still need additional reliable third-party references, beyond just Eiríkur Rögnvaldsson. You said IPA transcription of geminate consonants are not usually done, but we still have a whole body of Wiktionary entries (none of which I wrote) indicating geminate consonants.

This may also overlap with a concern I've raised at Talk:Icelandic phonology. Specific levels of phonological representation. You're insisting the usage in Template:IPA-is ought to be phonemic, but it uses [square brackets] which are phonetic usage. Use of /slashes/ is phonic/phonemic, and use of ⫽double slashes⫽ is morphophonemic. Either the template needs to be changed to a different level of representation, or phonetic IPA needs to be allowed.

I've also raised proposals in Help talk:IPA/Icelandic about using different notation conventions to improve readability, such as clearer long diphthongs, the use of only one height of mid vowel symbols, and especially replacing the symbols for the central rounded vowels to actual central rounded vowel symbols. All in all, I proposed making the following IPA substitutions:

• Replace [ʰp ʰt ʰc ʰk] with [hp ht hc hk] because the preaspiration is much more pronounced than the postaspiration of [pʰ tʰ cʰ kʰ]. This change was carried out. In truth there was no comment to this proposal (it was there for a while but other users kept skipping it and replying to other threads instead), but the [hp ht hc hk] usage was already being used at Icelandic phonology and overwhelmingly in Wiktionary entries.
• Use clear tied diphthongs [a͡i] instead of [ai], especially for the clarity of long diphthongs [a͡iː] because [aiː] is visually ambiguous to users with only general IPA knowledge, as it looks like [a] followed by [iː] rather than a unitary phonete. Another user objected to any use off the tie bar for diphthongs, and ultimately consensus could not be reached. The proposal was withdrawn. The user who objected most strongly to tied diphthongs also admitted that he didn't read the rest of the proposal text anyway.
• Use common mid vowel symbols without specifying vowel height. Current practice as open-mid symbols for monophthongs [ɛ, œ, ɔ] and close-mid symbols for the onset of diphthongs [ei, ou] except strangely for [œi]. I proposed replacing every instance of [ɛ] with [e], and every instance of [ɔ] with [o]. Depending on whether my central vowel proposal proposal is adopted, [œ] would be replaced with [ɵ] (using central rounded vowel symbols) or [ø] (using front rounded vowel symbols). And this applies to the phonic and morphophonemic representation. A relatively more narrowly phonetic representation with [square brackets] could allow for realizations where they actually exist, such as [ɪɛ] instead of /eː/. Not enough users could form opinions on this for consensus to be established. It's worth noting that where Wikipedia has been using [œi] for au, Wiktionary entries have been using [øy]. I would propose [ɵi] instead for this diphthong.
• Use central rounded vowel symbols for the central rounded vowel phonemes. This would involve replacing every instance of [ʏ] with [ʉ] (the IPA has never formally adopted [ᵿ]), and every instance of [œ] with either [ɵ] or [ɞ] (depending on whatever consensus can be reached in regards to my mid vowel symbol proposal). This is necessitated by the confusing rule that velar consonants k and g become palatal consonants after front vowels, but not after [œ, ʏ] because they are central vowels rather than front vowels (and several of the external references corroborate this). Someone with basic IPA knowledge learns that [œ, ʏ] are front vowels, and at least one other user in Talk:Icelandic phonology raised this discrepency in regards to Icelandic. In truth, many of the central vowel symbols have only been fully adopted by the IPA in recent decades, so I can understand why it may not have become convention for Icelandic IPA in earlier decades. Not enough users could form opinions on this for consensus to be established.
• Clearly mark semivowels of diphthongs so that long diphthongs are easier to read. [ai̯, aːi̯] instead of [ai, aiː]. I proposed this after withdrawing the proposal for tied diphthongs. In truth, I still don't favor this proposal for the same reason I prefer tied diphthongs. No one responded.

As it is, it has been difficult to locate and corral other editors knowledgeable enough in this particular language topic to participate in discussions and consensus-building. I think their recurring absence has contributed to too many (I'm assuming otherwise well-meaning) lone editors spreading misconceptions among the articles by using Wikipedia as its own reference (a circular reference). For example, the assumption that [m n ɲ ŋ r l] can only be voiceless [m̥ n̥ ɲ̊ ŋ̊ r̥ l̥] in proximity to other voiceless consonants or stops, leading to the spread of erroneous notation like [tl̥a] for lla. The body text description in Icelandic phonology still describes that as the case. I have not edited that description as parts of it feel more like an (otherwise well-worded) essay someone jotted out and there are no citations or footnotes for most of the claims. Most of that article probably needs to be completely rewritten from the ground up (and possibly also the Icelandic orthography article), and that particular task would probably be biting off far more than I can chew. - Gilgamesh (talk) 23:14, 24 May 2021 (UTC)

I should also note that your recent edits to Help:IPA/Icelandic have made that page self-contradictory now. The page already noted (not my edits): "Vowels are usually long if they are stressed and followed by no more than one consonant, double consonants counting as more than one. Vowel length is not phonemic." With some kind of contrast existing between [VCː] and [VːC], it goes to reason that either vowel length can be non-phonemic or consonant length can be non-phonemic, but not both at the same time. Ultimately the wiki's articles on this topic need to not contradict each other, but again, this comes down to conflicting references on issues such as the contrastiveness of consonant length, and the fact that all the third-party published references insisting consonant length is non-phonemic appear to come from a single source: Eiríkur Rögnvaldsson. If he were the only source or no one contradicted him on this point then this wouldn't be the issue it is, but he's contradicted by multiple other references. We need additional references and verification, otherwise we can't just assume as a matter of fact that consonant length must be treated as completely non-phonemic. - Gilgamesh (talk) 00:44, 25 May 2021 (UTC)

@Thjarkur: Need review of IPA at Árbær. In particular, the IPA for Ártúnsholt, Norðlingaholt and Selás.

• From what I could gather, Ártúnsholt is split up as Ár·túns·holt, but I could be wrong.
• The IPA for Norðlingaholt is uncertain with the norðlinga part. Eiríkur Rögnvaldsson (2020) says rðl is pronounced [rtl], which I rendered [(r)tl] for the same reasons rl is rendered [(r)tl]. I know norðlinga probably wouldn't have that IPA if it were split into norð·linga, except that -lingur is a suffix fused to norð rather than a separate word, right?
• Selás is split up as Sel·ás, right? I also kept the vowel in sel- short because in this usage it behaves like a prefix...right?

That's all for the time being. - Gilgamesh (talk) 22:49, 29 May 2021 (UTC)