Help talk:IPA/Dutch

Rhotic

Hey user:Nardog. The Dutch rhotic has a number of possible pronunciations and the closest ones known to English speakers are the rolled R or the French R, and the bunched R in the coda which sounds close to the English language R, the approximant [ɹ]. --Esperfulmo (talk) 00:23, 13 December 2023 (UTC)

@Esperfulmo The article currently mentions 'rolled R', with a link to alveolar tap/flap, this seems weird, as the rolled R is defined as a trill. Are you sure the American [ɾ] is that different from the Dutch [ɾ]? Exarchus (talk) 10:31, 29 December 2023 (UTC)

Maybe you got confused by Wikipedia saying the trill is often reduced to a "flap". But there is no source saying it is a flap, as distinct from a tap, so I changed the phonology article to "tap". Exarchus (talk) 12:19, 29 December 2023 (UTC)

Well, the rolled R is either the trilled or the flapped R which are both from the variations of the phoneme for the Dutch.

Yes, very sure that the voiced alveolar flap (sounding like one trill) is very different from the voiced alveolar tap (sounding closer to the D sound/plosive). Yes, we see on Wikipedia both terms used interchangeably, but they don't sound the same. Even the user who made the audio file for both consonants failed to produce a correct tap sound. If you natively speak a language with one of them and another which has the other, you can definitely tell the difference. There was a suggestion to use a small capital D for the tap to distinguish it from the flap ɾ, but I don't know why both are transcribed with the same character. See the Americanist phonetic notation#Rhotics table. --Esperfulmo (talk) 12:24, 29 December 2023 (UTC)

The sound recording on wiktionary for 'city' (Audio (US)) sounds to me exactly like one would say Dutch 'sirrie' (except that this isn't a word). I see no reason why it can't be used as an approximation (and I'm not convinced it isn't exactly the same). Exarchus (talk) 14:00, 29 December 2023 (UTC)

Quoting from "The Sociophonetics and Phonology of Dutch r": "even those taps that are distinguishable in terms of their articulation from single-contact trills are auditorily/perceptually extremely similar to them." (p.170) Exarchus (talk) 14:29, 29 December 2023 (UTC)

From reading the mentioned paper, I can't conclude that Dutch 'taps' [ɾ] aren't real taps, but single-contact trills (which, as the paper indeed says, are articulatorily different). So I really think American intervocalic 't' is exactly the same thing. Exarchus (talk) 14:57, 29 December 2023 (UTC)

To quote extensively: "In conclusion, the predominance of taps in onsets makes any assumption that they are synchronically actually short (single-contact) trills less probable. If so, one would expect more variation with trills with two or three contacts, especially since the number of contacts are not under active control from the speaker. It is, in other words, not likely that there is an invariant trill target for all alveolar onset r, that is consequently not met due to articulatory or aerodynamic constraints. Nevertheless, since taps alternate freely with other manners of articulation in onsets, including trills (almost all speakers do variably use trills), it is demonstrably not the case that the tap is the single r-target either. This points towards an analysis in which both trills and taps, as well as fricatives and approximants, are available to all speakers who use alveolar variants in onsets." (p.176) Exarchus (talk) 15:56, 29 December 2023 (UTC)

OK, the paper mentions some difference between American [ɾ] and (in this case) Spanish [ɾ]:

"Taps can also be allophones of other, non-rhotic, sounds, such as, in American English, intervocalic /t/ in ladder, latter, city. There are phonetic differences between the two, apparent in x-ray footage (Ladefoged and Maddieson 1996): for an American English speaker producing a tap-/t/ there is anticipation during the preceding vowel in the form of retraction and raising of the tongue. The tongue is then moved forward to make the contact and returns to the floor of the mouth. For a Spanish speaker producing a tap-/r/, there is no such anticipation, but a quick upward and downward movement of only the tongue-tip." (p.171)

I still think using one as an approximation of the other is not far-fetched. Look at it the other way: how would you give an approximation for Dutch speakers of the American intervocalic 't'? The tap-r of course. Exarchus (talk) 17:43, 29 December 2023 (UTC)

Interesting quote from Articulatory Variation of the Alveolar Tap and Implications for Sound Change (Cathcart, 2012): "An acoustic and cineradiographic study by Monnot and Freeman (1972) shows that acoustically, English and Spanish alveolar taps are indistinguishable from each other."

I would not be surprised if someone showed that a perceived difference comes from the brain classifying the sounds as either /t/~/d/ or /r/... Exarchus (talk) 23:55, 29 December 2023 (UTC)

Assessing Rhotic Production by Bilingual Spanish Speakers (Cummings & Montrul, 2020): "The English flap and the Spanish tap are nearly identical, in that they are both produced with a rapid movement of the tongue tip or tongue blade against the alveolar ridge" Exarchus (talk) 10:01, 30 December 2023 (UTC)

What you may be thinking of is the situation in Arabic as I'm reading it on the Arabic phonology page: "The trill /r/ is sometimes reduced to a single vibration when single, but it remains potentially a trill, not a flap [ɾ]"

As I said, no indication this is the case in Dutch (or Spanish). Exarchus (talk) 13:33, 30 December 2023 (UTC)

So it would make sense to me to use [r̆] (r + breve) to indicate a single-vibration alveolar trill (what you seem to call 'flap'), and [ɾ] for the alveolar tap (often called flap).

And now I think I've finished rambling about taps/flaps ;-) Exarchus (talk) 15:39, 30 December 2023 (UTC)

I listened to wikt:city and it doesn't sound to me like a flap. For the case of the Egyptian pronunciation, the rhotic is in a free variation between a trill and a flap. The flap is quite more popular in the Nile Delta and some speakers from northern Egypt might exclusively use one variant. The flap is also used in Persian and other languages. They never sound quite close to the tap [D].

I know what you mean by recognizing sounds by one's brain that could be deceiving. Approximating the Dutch rhotic to the rolled R makes more sense, because:

• it alternates between an alveolar trill and a weaker variant
• the vast majority of people alternating between 2 variants have an alveolar trill versus an alveolar flap.

So which rhotics does your area use or what do you personally pronounce? There are those who use a uvular main variant. The Dutch case is very interesting! --Esperfulmo (talk) 17:30, 30 December 2023 (UTC)

An interesting example of a D like sound alternating with an R like sound in Hindustani. Listen to this cover of an Indian classic: [1], the title is spelled with a D: Thoda Resham Lagta Hai with an assumed /ɖ/ voiced retroflex plosive, even though it sounds like a [ɽ] voiced retroflex flap intervocalically. --Esperfulmo (talk) 18:14, 30 December 2023 (UTC)

- I know about this alternance between [ɖ] and [ɽ] in Indian languages and at first I found it strange as in my intuition, 'd' and 'r' have nothing to do with each other. But I figured out they really are articulatorily close.

- "I listened to wikt:city and it doesn't sound to me like a flap." In your terminology, it isn't a flap, but a tap and that's what the [ɾ] symbol is used for. And yes, it sounds like perfectly valid Dutch to me (when I think of it as an English word, I might be tempted to hear a /d/-like sound).

- The argument for approximating Dutch alveolar R as 'rolled R' is not that strong, as most of the time (in normal speech, not in sound fragments specifically pronounced for Wikipedia) it isn't a trill [r] and also not a reduced/short trill (what you call a flap), although people will intuitively think of it like that, but a tap. And that is what speakers of English know best anyway.

And yes, the 'R' I use is alveolar (not uvular or 'bunched'), so I am very familiar with it. And I have done some 'experiments' with pronouncing R's and come to the conclusion that the 'short' alveolar R in Dutch is not a 'long' R reduced to one contact, as I find it difficult to consistently produce just one trill (and not two, or zero). So I think that by now I know perfectly what you mean with flap vs. tap, and the short alveolar realisation of R in Dutch is clearly a tap (when I pronounce it, I can't prolong it, if I try really hard, I have to change something inside my mouth, and then I get a trill). And I'm pretty sure that Spanish /ɾ/ is as well.

- Whether Persian uses a flap or a tap, I'm not sure, do you have first hand knowledge of this? I can imagine it could be Arabic influence. But then in this paper it is said that the most prominent rhotic variant in Persian is a fricative (about which Wikipedia doesn't say anything). So for the moment I'm confused about this. Exarchus (talk) 20:11, 30 December 2023 (UTC)

If you want to see frequencies of alveolar trill vs. alveolar tap in Dutch, you can look at 'The sociophonetics and phonology of Dutch r' (if you can find it) and then for example starting at page 87 for Antwerp: 14% voiced alveolar trill in word onset vs. 65.7 % voiced alveolar tap (+ some other varieties), intervocally 5.7 % trill vs. 68.6 % tap and at syllable coda it is apparently most often 'voiceless alveolar trill/tap with frication', something I hadn't heard about before and which I am no doubt constantly using. Exarchus (talk) 20:38, 30 December 2023 (UTC)

I suppose some of the (short) trills might be classified as 'flaps' (in your terminology), but that is not what the tap is. Exarchus (talk) 21:28, 30 December 2023 (UTC)

Thanks for sharing all of the sources <3

From the first look, the paper mentioned 3 Persian allophones and used the term tap. Never mind the Persian rhotic for now.

Regarding Spanish, it has a trilled version spelled with double R, e.g. caro vs. carro which doesn't sound like a long flap.

I actually don't consider myself using different linguistic terminology. The Americanist phonetic notation has two symbols for each of the consonants. A single trill still sounds stronger than the flap, and Egyptians using either the flap or the trill could normally geminate it. E.g. [ˈmɑr.rɑ, mɑɾ.ɾɑ] "once" vs. [ˈmɑ.rɑ, mɑ.ɾɑ] "woman" (vulgar)(مَرَّة vs. مَرَا). There are even those who use an approximant and geminate it, but an approximant is considered a lisp in Egypt. In Afroasiatic languages, gemination is common and phonemic.

Perhaps, the tap version can't be geminated or some languages can't geminate their rhotic. An interesting example is the word for "football" in Modern Hebrew was translated to כדור-רגל kadur regel, but evolved to כדורגל kaduregel because people pronounce it with a normally lengthed rhotic. Classical Hebrew accepted gemination, except for some consonants, like the rhotic. --Esperfulmo (talk) 02:26, 31 December 2023 (UTC)

Does the Spanish [r] (for example here) sound to you like an Arabic [r] (long trill) to you? Maybe there's a subtle difference there.

If you can find a description somewhere what Americanist phonetic notation actually means with 'tap' vs. 'flap', that would be useful. The Arabic phonology page also says: "the pronunciation of this single trill is between a trill [r] and a flap [ɾ]", where 'flap' is no doubt intended as 'tap'.

I can guarantee you that the tap [ɾ] in Dutch (and American English) is physically impossible to prolong/geminate. You'd have to start all over again and it sounds comical when I try it. As the Sociophonetics paper says (p.172), a short "schwa-like vocoid" gets inserted when [ɾ] is not preceded by a vowel, so repeating this tap sounds like [əɾəɾəɾəɾ]. Exarchus (talk) 09:36, 31 December 2023 (UTC)

This paper talks about Arabic 'taps', but I suppose you would call them 'flaps', because at one point it mentions "single long taps", which is impossible. So there seems to be total confusion about this in the literature. Exarchus (talk) 09:57, 31 December 2023 (UTC)

My personal guess is that the 'tap' will be more common cross-linguistically than the 'flap', because for producing a 'one-contact rhotic', the tap is ideal. Would be interesting to know for sure, and I'm really curious what the situation in Hindustani is. How does this sound to you? Sounds like a tap to me. Exarchus (talk) 10:08, 31 December 2023 (UTC)

The Urdu 'Audio (PK)' file for the same word seems to have a different R. Is this what you'd call a flap? Exarchus (talk) 10:25, 31 December 2023 (UTC)

The Urdu audio was a trill and the Hindi was a flap. Despite how the Urdu example was notated as a flap. --Esperfulmo (talk) 02:02, 1 January 2024 (UTC)

If you call the Hindi example a flap, then I'm confused and not sure anymore if there's a real difference between what you call a tap and a flap... Exarchus (talk) 15:36, 3 January 2024 (UTC)

In places where you might think there could be a geminated tap in Dutch, like the derivation 'doorrijden', I think that the first 'r' will in practice rather be this voiceless tap/trill with frication, and the second 'r' a voiced tap (in fast speech, this will rather be just one tap, or maybe an approximant; a trill would be emphatic I'd say). When I insist on making two voiced taps, there is a clear pause and it sounds like two words. Exarchus (talk) 13:34, 31 December 2023 (UTC)

I've started to notice something: languages that commonly have geminated consonants and that use the alveolar trill as main rhotic (Arabic, Italian, Finnish) do not seem to use the alveolar tap as a weakened form (it can be an allophone of /d/ in Finnish). As geminated consonants are normally just long versions of the short consonants, using a pair 'alveolar tap / alveolar trill' would be an exception to that rule, as the articulation is (as Sebregts calls it) "completely different". Dutch and Spanish don't generally have geminated consonants (don't let Dutch spelling confuse you there), so the brain doesn't have to have strict rules for producing short vs. long consonants.

At least that's my hypothesis. Exarchus (talk) 14:53, 31 December 2023 (UTC)

Notice that in Finnish, /d/ (which has a tap allophone) does not have a geminated form in native words. This might be relevant for my hypothesis. Exarchus (talk) 15:19, 31 December 2023 (UTC)

One thing about the Spanish trill: studies have shown that many or most speakers do not actually use the 'official' trill. This paper says: 'Hammond (1999:147) concludes that the “typical” trill is “absent in the normal discourse of the vast majority of native speakers.”'

That's why I linked to the 'rincón' page, where the speaker has a 'real' Spanish trill. Exarchus (talk) 21:04, 31 December 2023 (UTC)

The Spanish example sounded like an ordinary trill. --Esperfulmo (talk) 02:00, 1 January 2024 (UTC)

About Arabic, I was very specific on mentioning only Egyptian (Arabic), the spoken dialect. Most literature about "Arabic" studies one region of the Arabic speaking world and generalizes it from coast to coast (e.g. Watson researching Yemeni dialects), ignoring the great variation in local traditions and not distinguishing between reading Literary Arabic and spoken dialects. That's really another topic. In all Arabic dialects including Literary Arabic, all consonants accept gemination, even the [h] and the [ʔ]. I know it sounds very weird, but it's true.

If I'd distinguish the flap from the tap, I'd say the flap has a slight vibration whereas the tap has a sound similar to a weak plosive with an even lighter vibration than the flap.

It was a very interesting discussion about rhotics.

Let's wrap it up so as no one says we're getting out of the topic of improving the content of the IPA page :) --Esperfulmo (talk) 02:20, 1 January 2024 (UTC)

wiktionary page

@Esperfulmo I came across the Dutch pronunciation page on wiktionary and I think it's full of inconsistencies/inaccuracies, this one is better. Would you want to update the wiktionary page (or just copy/paste this one there)?

One thing that might be improved here is that there's a difference in quality (in standard speech) between /ɑ/ and /aː/, so if you use 'father' for both, this is not so accurate. /aː/ is more like the General American example here (but long), and /ɑ/ more like the RP pronunciation example (but short). By the way, the sound example for Dutch 'bad' sounds to me like a Brabantian [a] pronunciation, so not the 'official' [ɑ]. (edit: I'm not so sure anymore it'd be Brabantian, but it does sound Belgian to me. Anyway, the difference in vowel quality between 'bad' and 'aap' in the recordings seems clear to me.) Exarchus (talk) 17:19, 28 December 2023 (UTC)

Well, I changed the approximation for /aː/ to "British lad, but long", I obviously mean modern British and not the older RP. Exarchus (talk) 21:39, 28 December 2023 (UTC)

Hey Exarchus. The table has approximations that will never be precise. It's for users who can't read IPA. About Wiktionary guides, I personally don't think there should be a page there for each language, since they are mostly copied from Wikipedia and a simple redirect would be sufficient. I normally don't rely of the guides there unless they have some peculiarities to follow there instead of what's common here, like Arabic transliterations. --Esperfulmo (talk) 12:28, 29 December 2023 (UTC)

Assimilation taken too far?

Hi everyone, I came across the phonetic transcription in the article about the Afsluitdijk, which is transcribed as [ˈɑfslœydɛik]. I see where this comes from, but the way in which assimilation in the <-td-> sequence is rendered here strikes me as excessive. The final [-t] of ‘sluit’, which becomes [-d] through regressive assimilation, does not disappear without a trace, certainly not in reasonably careful speech. If there was a (hypothetical) Afslui-dijk, it would not be pronounced exactly the same as the (existing) Afsluitdijk. In reality, I think that the sound is most likely to appear as an unreleased stop: [ˈɑfslœyd̚dɛɪk] (I know that this symbol is not used in the current guidelines), so the articulatory organs move to the position for articulating [d], but there is no audible burst. Within the existing guidelines, I would probably transcribe the word as [ˈɑfslœyd.dɛik], with the full stop indicating a syllable break. It would certainly not be wrong to pronounce the word this way. What do others think? Isoglosse (talk) 07:45, 30 October 2024 (UTC)

I have checked the archives of the talk pages about any discussion on this topic, but came up empty. I have recently stumbled across another example where I think that the assimilation that is transcribed assumes a speech style that does not match the audience (i.e., non-native speakers who seek guidance on pronunciation and are unlikely to pronounce names in an allegro style).

The name Marius Job Cohen is transcribed as [ˈmaːrijə ˈɕɔp koːˈɦɛn]. As I said before, I see where this comes from. A fluent speaker of Dutch might say that in rapid speech. But [ˈmaːrijəɕ ˈɕɔp koːˈɦɛn] (with the fricative repeated at the end of the first part) or [ˈmaːrijəɕ ˈjɔp koːˈɦɛn] (with the assimilation only transcribed in the first part) are equally correct and maybe better transcriptions. Assuming a careful speech style, [ˈmaːrijəs ˈjɔp koːˈɦɛn] (with no assimilation across syllable boundaries) is also possible. In my view, all transcriptions are valid – but which one is best?

I am not sure if the current transcription style, which omits symbols and thereby suggests a complete merger of the sounds involved in assimilation, serves encyclopedic users best. In analogy to the Afsluitdijk example above, I think that I would suggest to transcribe the name as [ˈmaːrijəɕ ˈjɔp koːˈɦɛn], which indicates the phonetic change, but leaves all parts of the name intact. If transcription guidelines would be clarified in this respect, I don’t think that all existing transcriptions would have to be changed, but that this could be implemented going forward.

As there have been no replies to my previous post, I would like to page some recent editors of the guidelines: @Sol505000 @IvanScrooge98, @Nardog, @Exarchus, @Esperfulmo. Happy to hear your take on that. Isoglosse (talk) 13:56, 22 November 2024 (UTC)

I don’t have strong feelings on either option, to be honest. And I’m not an expert on Dutch phonology either, so I can’t have an opinion as to what solutions best describe the usual realizations. What I care most about is consistency, whatever we decide to do. I’ve recently gone through all the transcriptions pointing here to fix old conventions that hadn’t been updated. ~ IvanScrooge98 (talk) 14:08, 22 November 2024 (UTC)

Thank you! I quickly browsed through Geert Booij’s The Phonology of Dutch (1999) this afternoon and found a couple of remarks that seemed relevant to me. In the chapter about phonological rules, he says:

“Phonological rules like Progressive Assimilation, which apply obligatorily within prosodic words and compounds, become optional in larger domains: for instance, in phonological phrases.” (p. 59; boldface added by me)

So for instance, devoicing /tz/ in ‘zoutzuur’ to [ts] or /tɣ/ to [tx] in ‘straatgoot’ is obligatory (because these form a prosodic word), but it can be optional in names like Gerrit Zalm, with /t|z/, or Albert Gerard Koenders, with /t|ɣ/ (where | indicates a word boundary).

In the introduction to connected speech phenomena, he says:

“In connected speech, words are subject to rules which are often optional, in that they are dependent on style of speech and speech rate. […] The use of these processes is often characteristic of less monitored, casual speech.” (p. 125; boldface added by me)

And in the subchapter about nasal assimilation (7.2.2):

“It should be remembered, however, that such observations are based on intuitions only, and require experimental underpinnings. Moreover, as pointed out by Nolan (1992), assimilations may be only partial.” (p. 147; boldface added by me)

So, what I am trying to say is: The way Dutch is currently transcribed treats optional, partial processes as obligatory, complete processes. I would suggest to err much more on the side of caution here and only reflect obligatory phonological processes in transcriptions, especially given the fact that we do not aim to transcribe rapid, connected speech. Isoglosse (talk) 17:13, 22 November 2024 (UTC)

I wonder whether there are experimental studies about the pronunciation of something like 'Afsluitdijk'. I'm inclined to think that those who regularly use the word really say [ˈɑfslœydɛik] (and consistently so), while those who aren't really familiar with it would tend to say [ˈɑfslœyd̚dɛɪk]. Exarchus (talk) 17:32, 22 November 2024 (UTC)

Transcribing Marius Job Cohen as [ˈmaːrijə ˈɕɔp koːˈɦɛn] seems a bit too far to me. It could suggest that the pronunciation of 'Marius' is [ˈmaːrijə] and that of 'Job' is [ˈɕɔp]. Not showing assimilations between words might be better. Exarchus (talk) 17:52, 22 November 2024 (UTC)

We might also introduce the ‿ symbol to the guide. I had considered proposing that because I agree that transcriptions like those are confusing. ~ IvanScrooge98 (talk) 18:02, 22 November 2024 (UTC)

That would be an idea, but is it necessary to show the assimilation when it isn't obligatory?

But I am inclined to think that the transcription [ˈɑfslœydɛik] is fine. Exarchus (talk) 18:33, 22 November 2024 (UTC)

Two points about this:

I think that Booij’s (or actually Nolan’s) observation about assimilation being partial is important. The fact that there is no audible stop at the end of the second syllable of ‘Afsluitdijk’ does not mean that the (elided) stop has not left its trace in that syllable (e.g., with respect to vowel quality or quantity). Not transcribing it at all seems to suggest to me that it was not there in the first place and that it must never be pronounced, not even in slow and careful speech. But this would be incorrect.

I am happy to agree that people who frequently talk about the Afsluitdijk are likely to say [ˈɑfslœydɛik] or something along these lines. But [ˈɑfslœyd.dɛik] is correct as well – and much clearer for non-native speakers who look for guidance. I would like to find a transcription convention that makes sure that the pronunciation we represent is common (in a careful speech style), but also gives enough information to readers. If they have any proficiency in Dutch, connected speech phenomena will happen anyway.

Two minor points I also wanted to mention: First, let’s not forget that ‘Afsluitdijk’ is not the same as the name of a person. I am much more willing to agree with [ˈɑfslœydɛik] than with [ˈmaːrijə ˈɕɔp]. Second, we try to make transcriptions that work for all varieties of Dutch. I am not sure if Dutch as spoken in Belgium and Suriname behaves exactly the same as Netherlandic Dutch when it comes to assimilation. Staying closer to a phonological transcription (similar to the way English is transcribed here) ensures that the transcriptions will apply to the language as a whole. Isoglosse (talk) 18:52, 22 November 2024 (UTC)

This paper says: "Dutch is reported to prohibit long consonants, including those derived by external sandhi". The next line does indicate that this apparently isn't always correct, but the example seems to be about external sandhi ('raaf fraai' vs. 'ra fraai').

So while you can no doubt say [ˈɑfslœyd̚dɛɪk] as a way to clarify how the word is composed (and this is useful for people who never heard about this 'Afsluitdijk'), I doubt it is natural Dutch. It's not because people think they are pronouncing it differently than they would say 'Afsluidijk', that there really is a phonetical difference, that's where you need an experimental study. Exarchus (talk) 19:26, 22 November 2024 (UTC)

Great find, thank you! There is a manuscript version of the paper that is freely accessible and a poster that nicely summarises the study.

The results say that ‘fake geminates’ (i.e., double /r/ that arises when a word ending in /r/ is adjacent to a word that begins with /r/) are not longer than a single /r/, but that the resulting sound carries characteristics of both the final /r/ (which is retroflex/bunched in some varieties of Dutch) and the initial /r/ (which is uvular or alveolar). Or in the words of the authors:

“These results suggest that Dutch degemination does not involve categorical segmental deletion, as previously suggested, but instead it has constraints on phonetic consonant duration that limit temporal gemination.” (my boldface)

Assuming that other consonant combinations behave similarly, we could conclude that [ˈmaːrijə ˈɕɔp] is an incorrect transcription for ‘Marius Job’ because it suggests that /s/ and /j/ merge into one single sound (that would be ‘categorical segmental deletion’). Based on these results, it seems more likely that there is a sound that starts like an /s/ and ends like a /j/ (with an /ɕ/-like transition phase). The transcription that renders this best may simply be [ˈmaːrijəs ˈjɔp].

With respect to the ‘Afsluitdijk’, I think it makes sense to remind ourselves of the audience. No person who is remotely fluent in Dutch would look up the pronunciation of that word. This is for people who don’t speak Dutch. Based on the current transcription, they might end up saying (clearly and carefully) [ˈɑf.slœy.dɛik], which is more likely to come across as incorrect than saying [ˈɑf.slœyd.dɛik]. Isoglosse (talk) 11:42, 23 November 2024 (UTC)

I think the paper gives some arguments for not bothering with giving (optional) assimilation between words.

But I don't think it is relevant for the case of 'Afsluitdijk' as this is a compound, where Booij suggests (in the quote you gave previously) that such rules are obligatory. Exarchus (talk) 12:12, 23 November 2024 (UTC)

Agree. I didn’t want to imply that the paper is relevant for the case of ‘Afsluitdijk’. (As you said, it is a compound rather than a phrase. And the paper is about fricatives rather than stop consonants.)

One thing to consider is that final consonants are also relevant for the distribution of vowels. When we completely omit a final consonant, this seems to imply things about the distribution that are probably incorrect. Certain vowels only occur in closed (or checked) syllables. In Dutch, /ɔ/ would be an example. /oː/ occurs in both open and closed syllables, so we have both /poː/ ‘po’ and /poːt/ ‘poot’, but only /pɔt/ ‘pot’, not */pɔ/. (Old news for anyone who speaks Dutch, of course.) But if we transcribe ‘potdicht’ as [pɔˈdɪxt], we suddenly have the ‘impossible’ syllable /pɔ/. From a phonological perspective, [pɔdˈdɪxt] would solve this problem, but only if it is not phonetically inaccurate.

So it would be really nice to also find research on the actual pronunciation of these stop combinations, but I haven’t been able to find anything so far. Isoglosse (talk) 15:09, 23 November 2024 (UTC)

I don't think there would be a problem in having /pɔtˈdɪxt/ resulting in [pɔˈdɪxt]. But /ɔ/ very much occurs in open syllables, as in 'potten'.

And I don't believe [pɔdˈdɪxt] can be correct. By the way, the Belgian pronunciation is rather [pɔˈtɪxt] I think. Exarchus (talk) 15:56, 23 November 2024 (UTC)

Cases like ‘potten’ are usually explained by ambisyllabicity, which means that the consonant is assumed to belong both to the first and the second syllable, so basically /ˈpɔt.tən/, which of course is phonetically rendered as [ˈpɔtən]. B-class vowels in the following quotation are /ɪ, ʏ, ɛ, ɔ, ɑ/:

“Intervocalic consonants following a B-class vowel always become ambisyllabic in Dutch, at least under the assumption that B-class vowels only occur in closed syllables.”

It’s a bit weird, I think, that Dutch linguistics has worried about syllables with B-class vowels in cases like ‘potten’, but does not seem to have considered or discussed cases like ‘potdicht’.

Interesting observation, by the way, relating to Belgian Dutch. We’re not on Wiktionary, luckily, so we don’t have to worry too much about ‘potdicht’. But it is important that we try to find a transcription style that works for any variety of Dutch. Isoglosse (talk) 16:40, 23 November 2024 (UTC)

"It’s a bit weird, I think, that Dutch linguistics has worried about syllables with B-class vowels in cases like ‘potten’, but does not seem to have considered or discussed cases like ‘potdicht’."

I think they are supposed to behave exactly the same. Exarchus (talk) 16:54, 23 November 2024 (UTC)

Hi Isoglosse. I think your last suggestion is meaningful and scientific. We shouldn't overdoe casual speech transcription. In fact, we don't do that in practice with other languages. --Esperfulmo (talk) 23:10, 22 November 2024 (UTC)

Hi, thank you for your reply! Good point about checking what the guidelines for other languages say (English may be relevant as a benchmark and maybe German as a related language).

Assimilation may not be equally strong in all languages, but much of it is just unavoidable coarticulation. I don’t think that John Bolton’s name would be transcribed /ˈdʒɒm ˈboʊltən/ according to the English guidelines. (There is no transcription, of course, because it’s a common English name.) Or Jan Böhmermann’s name is [jan ˈbøːmɐˌman] rather than [jam ˈbøː-], although a certain degree of assimilation/coarticulation is present in English and German as well. Changing/clarifying the Dutch guidelines would make sure that they fall in line with the guidelines of related languages. Isoglosse (talk) 11:51, 23 November 2024 (UTC)

Let me briefly summarise and suggest next steps:

• There seems to be agreement on the fact that transcriptions must not reflect optional phonological processes in phrases (e.g., names). I would suggest to stop transcribing assimilation in names altogether and add a paragraph about this in the guidelines.
  • Here are some examples of how transcriptions would change:
    • Nasals
      • Jan Blokhuijsen: [ˈjɑm ˈblɔkˌɦœyzə(n)] → [ˈjɑn ˈblɔkˌɦœyzə(n)]
      • Henk Krol: [ˈɦɛŋ ˈkrɔl] → [ˈɦɛŋk ˈkrɔl]
      • Martin van Marum: [ˈmɑrtɪɱ vɑ ˈmaːrʏm] → [ˈmɑrtɪn vɑn ˈmaːrʏm]
      • Arnon Yasha Yves Grünberg: [ˈɑrnɔ ˈɲɑɕaː ˈiːf ˈxrʏmbɛr(ə)x] → [ˈɑrnɔn ˈjɑɕaː ˈiːf ˈxrʏmbɛr(ə)x]
    • Stops and fricatives
      • Hans de Koning: [ˈɦɑnz də ˈkoːnɪŋ] → [ˈɦɑns də ˈkoːnɪŋ]
      • Frits Zernike: [ˈfrɪt ˈsɛrnikə] → [ˈfrɪts ˈzɛrnikə]
      • Coenraad Jacob Temminck: [ˈkunraː ˈtɕaːkɔp ˈtɛmɪŋk] → [ˈkunraːt ˈjaːkɔp ˈtɛmɪŋk]
      • Marius Job Cohen: [ˈmaːrijə ˈɕɔp koːˈɦɛn] → [ˈmaːrijəs ˈjɔp koːˈɦɛn]
  • I would be willing to keep assimilation of van [vɑn] to preceding voiceless sounds, e.g. Chris van der Klaauw: [ˈkrɪs fɑn dər ˈklʌu]. This would also greatly reduce the number of transcriptions that have to be changed. (I would, however, advise against footnotes that say “In isolation, van is pronounced [vɑn].” I don’t think anybody needs that.)
  • We could consider focusing on names first and leave sentences like Ik zou je het liefste in een doosje willen doen out of this.
• There is less agreement on how to transcribe word-internal assimilation, especially when it leads to potential gemination (e.g., Afsluitdijk). I would suggest to leave this topic alone and keep the transcriptions as they are (until there is more evidence).

Would you agree with that, @Exarchus, @Esperfulmo, @IvanScrooge98 and others? Isoglosse (talk) 17:00, 28 November 2024 (UTC)

I’m not convinced we should treat voicing/devoicing in two different ways as you seem to be suggesting, and I’m more inclined to keep all instances of assimilation ([ˈɦɑnz də ˈkoːnɪŋ] etc.). I would also tend to keep all internal assimilations, keeping in mind there’s normally never a pause mid-word; optional geminations resulting from that could be marked as [ˈɑfslœy(d)dɛik] (after all, /ən/ is already marked as [ə(n)], [ə(ɱ)] and so on). I’m fine with dropping non-voice assimilations across word boundary. ~ IvanScrooge98 (talk) 17:22, 28 November 2024 (UTC)

A problem with [ˈɦɑnz də ˈkoːnɪŋ] is still that it can suggest that 'Hans' is pronounced [ɦɑnz], so you might have to use the ‿ symbol, but what if people simply want to know how the first name is pronounced?

I'm trying to find out how this is handled for other languages (Polish?) but they mostly don't have IPA. Exarchus (talk) 18:11, 28 November 2024 (UTC)

Let’s also not forget that this type of voice assimilation is not even obligatory within words. For ‘bewijsbaar’ (provable), a Dutch pronunciation dictionary gives two transcriptions, namely [bəˈʋɛisbaːr] and [bəˈʋɛizbaːr] (I converted the transcription to our style). The same holds for most other relevant lemmas. (They were not completely consistent with that, but it seems more like an editorial oversight. The dictionary was published 46 years ago, long before Ctrl+F made our lives easier.) I would argue that the same is true in [ˈɦɑnz də ˈkoːnɪŋ]: Assimilation is possible, but optional. Isoglosse (talk) 18:26, 28 November 2024 (UTC)

An example with /sd/ would be ‘bisdom’ (bishopric), where both [ˈbɪsdɔm] and [ˈbɪzdɔm] are given. So I don’t think that we have to exclude that case from the list of assimilations that could be dropped. Isoglosse (talk) 18:30, 28 November 2024 (UTC)

Ok, I guess you mentioned that before but I got lost in the long discussion on Dutch phonology haha. Then I suppose we should only keep the assimilations that do happen as a rule, if there are any.

If we want to follow our practice with other languages like Exarchus suggests, I think we should look at fellow Germanic languages, for which we mostly don’t mark assimilations as they’re generally optional. (On the other hand, in Polish, like most other Slavic languages, voice assimilation is regular so we normally go along with it in our IPAs.) ~ IvanScrooge98 (talk) 19:24, 28 November 2024 (UTC)

One case where I'm not sure what's best is a surname like 'van Marum'. Sometimes the 'van' in these names is joined with the rest: both 'Van Mechelen' en 'Vanmechelen' exist. It seems weird to give different transcriptions based on spelling. So I think a case can be made to transcribe ˈmɑrtɪn vɑ‿ˈmaːrʏm Exarchus (talk) 21:02, 28 November 2024 (UTC)

That sounds ok, but I'm inclined to think that you might want to keep [fɑn] because that simply is the most common pronunciation of 'van' in the Netherlands. Exarchus (talk) 17:24, 28 November 2024 (UTC)

Yeah, you may have a point there that I am somewhat influenced by Netherlandic Dutch 😉

But don’t get me wrong: I don’t want to keep [fɑn], but I could live with it in existing transcriptions for now (and would see changing these transcriptions as low priority). But in the interest of consistency, any new transcriptions should use [vɑn]. Anyway, thanks for confirming that you would generally agree with these changes. Isoglosse (talk) 18:15, 28 November 2024 (UTC)

[ˈfrɪt ˈsɛrnikə] is the best bad example, since it misleads the reader that the first name is "Frit". I'd say such pronunciations are way too casual to even consider in an encyclopedia! IPA transcription in practice is transcribing quite careful pronunciations, not lazy utterances. I already saw similar bad transcriptions for the Hebrew phonology which I corrected and cited. I also want to ask, what is the best transcription for Tiësto, accordingly? I saw it notated as if it were spelled Kesto and now it is notated as if it were Tjesto. Why do we have the diaeresis Ë, then? Shouldn't we also correct this? It's like Tuesday transcribed /ˈtʃuːzˌdeɪ/ when it's just an optional pronunciation. Esperfulmo (talk) 01:09, 29 November 2024 (UTC)

Wait. The pronunciation was formerly noted as [ˈcɛstoː], when this help page used the ⟨c⟩ symbol for assimilated /tj/ (I went through most of these old transcripions myself in order to have them match the help); that was never a transcription of /ˈkɛstoː/. The dieresis does not mark a phonetic dieresis but it’s used to denote that each letter in ⟨ie⟩ represents a separate phoneme (as /jɛ/) instead of what would be usual in Dutch, i.e. /i/. ~ IvanScrooge98 (talk) 01:19, 29 November 2024 (UTC)

I'm personally not inclined to say [tɕ-] here but simply [ˈtjɛsto]. See new topic for length of final vowel. Exarchus (talk) 10:42, 29 November 2024 (UTC)

We must be consistent. If other instances of word-internal /tj/ are transcribed as [tɕ], then this one should be too, all the more since both phonemes belong to the same syllable; now, if this is actually /tiˈɛstoː/ and not /ˈtjɛstoː/, things are different. ~ IvanScrooge98 (talk) 11:24, 29 November 2024 (UTC)

Maybe it can be analysed as a kind of non-native diphthong /iɛ/. It isn't written 'Tjesto'. Exarchus (talk) 11:37, 29 November 2024 (UTC)

Tiësto himself definitely says [ˈtɕɛstɵu] or maybe [ˈt̠ʲɛstɵu], with a Czech/Slovak-style alveolo-palatal stop with just a hint of voiceless frication. It is not [j] unless you're going for a really broad transcription. The underlying phonemes are thus /ˈtjɛstoː/. (Also, tuesday is pronounced with an initial /tʃ/ by the vast majority of Brits nowadays. It's not optional, it's THE standard way of saying the word.) Sol505000 (talk) 13:00, 29 November 2024 (UTC)

If that's how he pronounces it, then I'm not going to argue. Exarchus (talk) 13:30, 29 November 2024 (UTC)

We should transcribe the most common assimilations and degeminations. 'van Marum' is, to the best of my knowledge, pronounced [vɑˈmaːrʏm]. It doesn't violate the Dutch phonotactics because we're not claiming that the underlying (phonemic) structure of 'van' is /vɑ/, with a final lax vowel. Don't mistake phonetics with phonology. Of course 'van' is /vɑn/ phonemically, nobody in their right mind would deny that. But I'm not sure why potdicht couldn't be /ˈpɔdɪxt/ (I hear an initial stress on File:Nl-potdicht.ogg) the same way that English cupboard is /ˈkʌbə(r)d/. Sol505000 (talk) 12:39, 29 November 2024 (UTC)

About the specific point of 'van Marum', I'm inclined to agree, as I said above, because it's simply one (sur)name. But for assimilation between names, Booij indicates that this is optional, see quote above: "it can be optional in names like Gerrit Zalm, with /t|z/". [I now see this wasn't actually a quote from Booij, but he does say it's "optional in larger domains" (larger than prosodic words and compounds)] Exarchus (talk) 13:35, 29 November 2024 (UTC)

Here is the actual quote from Booij (already cited above): “Phonological rules like Progressive Assimilation, which apply obligatorily within prosodic words and compounds, become optional in larger domains”.

Do you have any additional references (or quotes from Booij) that contradict this description, @Sol505000?

I don’t say that these assimilations are not encountered in rapid speech, but they are, well, optional. In the context of an encyclopedia that is mostly used by non-native speakers of Dutch, I think that a different transcription style would serve users much better.

But I’m just repeating what has been discussed at length above. So please provide any references or data beyond ‘to the best of my knowledge’ or we will proceed to make the changes that most editors involved in this discussion have agreed with. Isoglosse (talk) 18:54, 29 November 2024 (UTC)

And doesn't 'van Marum' behave just like a compound? The normal stress pattern is /vɑnˈmaːrʏm/, with only one syllable stressed. The space is there only to match the orthographical form. And spellings such as Vanderhaeghe for 'van der Hage' and Vandendriessche for 'van den Driese' only strengthen this perception of mine. Sol505000 (talk) 19:00, 29 November 2024 (UTC)

I agree with you that ‘van Marum’ is different from ‘Marius Job’. But let me reiterate two points:

First, assimilation also happens in related languages, but it is never transcribed in Wikipedia – not in English, not in German. The article on Martin Van Buren clearly shows that difference: Phonetically, /nb/ is likely to be [mb] in both languages, but it is only transcribed as such in Dutch. Why? I don’t think that there is any factual reason for that.

Second, research on Dutch /r/ has shown that “degemination does not involve categorical segmental deletion”. These are the best data that we have on what happens phonetically when two sounds meet. The current transcriptions fail to reflect these data by merging the two segments into one. I would argue that this is wrong. Not transcribing assimilation may suggest slightly slower and more formal speech, but it is correct. So we would move from a transcription that is so extreme with respect to assimilation that it could be described as wrong to a transcription that is correct, albeit a tad more formal.

Finally, we have to find transcription style that is both useful and workable. There is agreement among most people involved here that the current style is not ideal. The new style should make it easier for non-native users to get a grasp of the pronunciation, but it should not be overly complicated for transcribers. So, while I agree with you, as I said above, that ‘van Marum’ is different from ‘Marius Job’, I think that differentiating in such detail would go too far.

Would it help if the sentence that I would suggest that we add to the guidelines said: “Assimilation is possible and common in casual speech, but remains optional in larger phrases. Due to its optionality and for clarity, it is not indicated between different parts of a name (link to some names with updated transcriptions). It can be indicated in the transcription of phrases (link to a phrase like Ik zou je het liefste in een doosje willen doen). Within words, it is obligatory and must therefore be transcribed (link to a single-word lemma).”? That way, it should be clear that assimilation is nothing outlandish, but that we refrain from indicating it in transcriptions to help readers. I would appreciate your approval of that. Isoglosse (talk) 20:34, 29 November 2024 (UTC)

"while I agree with you, as I said above, that ‘van Marum’ is different from ‘Marius Job’, I think that differentiating in such detail would go too far."

But then you would need to give different transcriptions for "Van Mechelen" vs. "Vanmechelen"... Exarchus (talk) 17:54, 1 December 2024 (UTC)

And we should drop the /x-ɣ/ distinction for Northern SD. They're not two phonemes but one, realized mostly as a voiceless uvular [χ]. This way we would get rid of absurd notes such as that "in isolation, Geert is pronounced [ɣeːrt]" - right, if you're a Belgian, or speak with a soft G which isn't standard in the Netherlands. Standard Netherlandic pronunciations begin with an [χ], followed by a monophthongal [ɪː] or a centering [ɪə] and whatever /r/ you use, as long as it's not another voiceless uvular fricative: [χɪːrt]. Sol505000 (talk) 13:31, 29 November 2024 (UTC)

Then you'd need to have different transcriptions depending on which country the person comes from. That might be an idea, but I'm not sure everyone would agree with that. Exarchus (talk) 13:38, 29 November 2024 (UTC)

Let’s please not open this can of worms. That’s a whole different topic. Isoglosse (talk) 18:55, 29 November 2024 (UTC)

For the /x-ɣ/ issue, see [2]. They write that "Reactions of ABN speakers do not confirm the view of earlier writers (...) who state that word-initial or stressed syllable-initial ⟨g⟩ represents /ɣ/ rather than /x/. Consequently, goed gegaan 'went well' is /ˈxut xəˈxaːn/ and not /ˈɣut ɣəˈɣaːn/. (...) Since Dutch does not permit final lenis consonants, this leaves only unstressed medial position as a context for any potential occurrences of /ɣ/. Many Dutch native speakers claim to use /ɣ/ for non-stressed intervocalic ⟨g⟩ thus contrasting the medial consonants in pairs such as lachen /ˈlɑxə/ 'to laugh' and vlaggen /ˈvlɑɣə/ 'flags'. However, this is certainly not true of all ABN speakers (...) and it is our impression that few in fact maintain such a contrast with any degree of consitency." A few pages later, they say that "For those speakers who have /ɣ/ as a phoneme, it is typically realized as an articulation which is less energetic than /x/. It is velar rather than uvular, and lacks the scrapiness associated with /x/, but does not (at least in ABN) have vocal fold vibration. (...) In the South of the Netherlands and in Belgium, (...) /ɣ/ may be in regular opposition to /x/ and also occurs as a representation of word-initial ⟨g⟩. For such varieties of Dutch, /ɣ/ is generally a VOICED velar fricative." So even for those who contrast /ɣ/ with /x/ in the North, the actual phonetic outcome is... [x] and [χ]. [ˈvlɑxə(n)] vs. [ˈlɑχə(n)]. Surely this is absurdly narrow for the purpose of this guide? Sol505000 (talk) 11:01, 1 December 2024 (UTC)

currently missing: short [a], [e], [o], [ø]

Currently, nothing is said about the short versions of /aː/, /eː/, /oː/ and /øː/, which occur in unstressed syllables, unless in a word-final closed syllable. That's the rule given by Gussenhoven at p.344.
So this means that the transcription of Tiësto (mentioned in the above topic) should have final [-o]. The transcription at Gerda Kraan, viz. [ɣeːˈrɑrdaː ˈɣɛrdaː maːˈrijaː ˈkraːn], then also has 5 mistakes. Exarchus (talk) 10:34, 29 November 2024 (UTC)

That rule is not mentioned at Dutch phonology (at least, I couldn’t find it), so it might be a reason why it was overlooked. Or we might find there’s someone who considers this to be a feature of more casual and less careful speech, in which case we’d be contradicting the approach we are following in the discussion above. I’ll tag @Sol505000 again as they’ve been pretty active in adding/amending Dutch IPAs and they seem to know a lot, unlike me lol. ~ IvanScrooge98 (talk) 11:30, 29 November 2024 (UTC)

It's there: "The native tense vowels /eː, øː, oː, aː/ are long [eː, øː, oː, aː] in stressed syllables and short [e, ø, o, a] elsewhere."

This isn't entirely correct as it doesn't mention word-final closed syllables. Exarchus (talk) 11:34, 29 November 2024 (UTC)

Yes, and it doesn't account for the lengthening/lack of shortening before /r/. -baar has a long [aː] because of both. Sol505000 (talk) 13:21, 29 November 2024 (UTC)

Agreed on [a], but in the Netherlands /eː, øː, oː/ have long been pronounced as [ei, øy, ɵu] (notice the centralized starting point of the latter - it's not [ou]), not [eː, øː, oː], which is a regionally marked realization. This complicates Gussenhoven's rule. There may be a shortening, but if the diphthongal effect is still there then we probably shouldn't transcribe it ("it" meaning the shortening), at least not in articles tied more to the Netherlands than Belgium or other countries. This, among other things, is why I find outdated IPA transcriptions a little bit infuriating :P There should be more Geoff Lindseys in the field of linguistics.

Tiësto is pronounced [ˈtɕɛstɵu] in unmarked (NL) Standard Dutch - and apparently also in Breda, where Tiësto comes from (not surprising, considering that Brabant is where diphthongizing /eː, øː, oː/ has probably originated). He definitely says [ˈtɕɛstɵu]. A Belgian SD pronunciation is probably closer to [ˈtɕɛsto], with a whole different vowel (probably shorter, yes). [ˈtɕɛsto] would be old-fashioned or non-standard in the Netherlands and so would [ˈtɕɛstou], with a fully back starting point. /oː/ contrasts with /øː/ by the backness of the ending point, rather than anything else - another proof they're definitely NOT monophthongs in contemporary Northern speech. You could transcribe them as [ɵʉ] for eu vs. [ɵu] for oo and it would still yield the correct pronunciation (there's usually a rather marginal difference in backness, with eu starting front-central or central and oo back-central, which is too little to distinguish the two vowels alone). Sol505000 (talk) 12:28, 29 November 2024 (UTC)

I think that the way English – another pluricentric language – is transcribed in Wikipedia is a helpful example. The transcriptions indicate the maximum number of vowel distinctions that can be made and indicate all consonants that can be pronounced. Based on this, the actual pronunciation in different Anglophone parts of the world (which may merge some distinctions and drop some consonants) can be inferred.

I would argue that we should try to treat Dutch – as a pluricentric language – similarly. It is quite beautiful that long vowels are transcribed ‘phonemically’ and that /ɣ/ and /x/ are distinguished in transcriptions. This is particularly important due to the fact that not every variety of Netherlandic Dutch diphthongises long vowels or drops the distinction between /ɣ/ and /x/. Many people outside the Randstad do not follow the traditional ‘standard’ of Northern Dutch, yet do not consider themselves as dialect or non-standard speakers. It is an added benefit of the current transcription style that we do not have to draw sharp lines here. Isoglosse (talk) 20:53, 29 November 2024 (UTC)

Your point makes sense and I tried but failed to implement it to Arabic. Moving back to Dutch, I definitely agree with you on distinguishing the two fricatives, even though I normally hear one: [x].

For diphthongs, maybe let's just follow common traditions if they transcribe them as long vowels or diphthongs. I have just one problem with "long" vowels in final unstressed syllables. Does it make sense? Do modern languages actually elongate vowels in such positions? English is an example. Mentioning Arabic as another example, the final long vowels are just a "classical" pronunciation of the literary dialect, which is mostly not followed even in educated speech. --Esperfulmo (talk) 23:27, 30 November 2024 (UTC)

"one problem with "long" vowels in final unstressed syllables. Does it make sense?"

Not in final open syllables, that's the point of this topic. There should be a category like at Help:IPA/Standard German for 'Shortened vowels'. Exarchus (talk) 09:37, 1 December 2024 (UTC)

The way we transcribe English is not a good example because diaphonemic transcriptions need to be enclosed within double slashes, //ˌlaɪk ˈðɪs//. [ˈðɪs ʔɪz ˈnɑɾ ə ˌdaɪəfəˈnimɪk ˌtʃɹɛənˈskɹɪpʃən], but an allophonic one. It can be broad (e.g. [ˌtrænˈskrɪpʃən] for 'transcription'), but it's still allophonic. The guides we should use as a comparison are Help:IPA/German or Help:IPA/Portuguese - both pluricentric languages, with separate national standards. I see no reason not to treat Dutch this way either. The national standards of Dutch have been very well defined in the literature (at least the european ones) by Gussenhoven, Verhoeven, Collins & Mees and Booij (and more authors).

I think you'll find just as many (northern) speakers being dissatisfied with the current IPA transcriptions of Dutch misrepresenting their speech as if they spoke Belgian Dutch, which is actually a minority variant. More speakers speak (NL) Dutch than (BE) Dutch and the northern pronunciation is far more influential in L2 teaching and it's encoded in foreign brains as "how Dutch sounds like". I'm certainly dissatisfied with Nederland being transcribed [ˈneːdərlɑnt], rather than [ˈneidərlɑnt] (even the Dutch royal family uses [ei] here - or do they have a Low Saxon/Limburg accent and I am mistaken? Both varieties are clearly non-standard outside of the respective regions, by the way) or Gouda being transcribed [ˈɣʌudaː], rather than [ˈχʌuda]. There are two mistakes in the former transcription, not one. Similarly, [ˌɣrɑxtə(ŋ)ˈɣɔrdəl] is not a good guide to the pronunciation of Grachtengordel, with two voicing mistakes regarding to velars: [ˌχrɑχtə(ŋ)ˈχɔrdəl]. I'm not a big fan of coda [r] before consonants, that also seems non-standard or very old-fashioned (no matter whether you trill it or tap it). You mostly get a retroflex or uvular approximant here, or just a schwa offglide. But I can live with ⟨r⟩ in this position as well, given how ridiculously variable /r/ is in Dutch.

If you want to transcribe vowel shortening then we need to separate the dialects. ADO as in ADO Den Haag does not (normally) feature a short monophthong at the end *[ˈaːdo] but a closing diphthong: [ˈaːdɵu], exactly the same as the stressed one in zo [zɵu]. There may be a length difference, but not a QUALITY difference. Maybe it can get reduced to [ɵ] mid-phrase before a consonant, as in [ˌaːdɵ dɛn ˈɦaːχ] - that seems possible. But it's not how Gussenhoven describes it, and it would be too much detail anyway for this guide. Listen to how Louis van Gaal speaks Spanish: [3]. Both stressed and unstressed O's are turned to [ɵu], he can't suppress it very well (he does at the beginning). Same with the [e] -> [ei] change. Surely pacto is not /ˈpakˌto/ in Spanish? And if it is, how would van Gaal know it? I think that Gussenhoven's rule is just wrong for the mid /eː, øː, oː/, he should've used the correct symbols ⟨ei, øy, ɵu⟩ and checked the diphthongality of ee, eu, oo together with their length. Lesser stress generally results in a reduced length, that's nothing surprising. Analyzing ee, eu and oo as monophthongs in contemporary Northern speech is a terrible mistake (and so is transcribing ui as ⟨œy⟩, which makes it sound like eu - that's another story), and it's trying to make the reality conform to Booij's dogma ("meer is pronounced [mɪər], not [meir], and that's a proof that it's a phonemic monophthong"), rather than the other way around. American English without the Mary–marry–merry merger also doesn't (usually) have a phonetic [eɪ] before /r/, it's typically [e] or [eə]. Yet nobody has a problem with writing /ˈmeɪri/. Wells certainly transcribes it so. Also, the pre-r allophony should be transcribed explicitly also in Dutch. What's stopping our readers from interpreting Herengracht [ˈɦeːrə(ŋ)ˌɣrɑxt] (proper IPA: [ˈɦɪːrə(ŋ)ˌχrɑχt]) as Hee Run Gracht [ˈɦeirə(ŋ)ˌχrɑχt] if they remember that "/eː/ is a diphthong in the North"? I think that we should make our transcription as noob-friendly as reasonably possible.

More examples: Hengelo [ˈɦɛŋəlɵu], Enschede [ˈɛnsχədei], NAVO [ˈnaːvɵu]. I see no evidence of Gussenhoven's rule being applied to the mid vowels. Maybe it applies word-internally, or phrase-internally before a consonant, in the case of word-final vowels. Sol505000 (talk) 09:59, 1 December 2024 (UTC)

You no doubt have a point that having an NL vs. BE standard would have its advantages. But let me simply make the observation that as a speaker from the province of Antwerp, the supposed Belgian standard comes across as artificial when it comes to vowels: all of /ɑ/, /ɪ/, /ʏ/, /i/, /y/ (in those cases not before /r/), /ɛi/ and /œy/ are pronounced differently than what's supposed to be standard. And then there's the matter of /u/, which consists of two phonemes /u/ and /uː/ here. Not that I'd want to propose several 'southern Dutch' standards, just saying that the Belgian standard isn't very ideal, but so be it I guess. Exarchus (talk) 11:09, 1 December 2024 (UTC)

Btw, I find your use of [ɵu] rather peculiar. The Dutch phonology article says: "The starting point of [oʊ] is centralized back ([ö])". That's not the same as [ɵ] and in a not particularly narrow transcription, [ou] would be good enough. It has actually been suggested that [ɵ] would be the best symbol for /ʏ/, so using [ɵu] could suggest we are talking about */ʏu/. Exarchus (talk) 20:00, 1 December 2024 (UTC)

Belgian Standard Dutch is just older Northern Standard Dutch with a regional spin (soft G). I think that many speakers find that it doesn't represent their speech very well, that's normal (and not the case at all in the north, there the standard is an actual, real standard that many people use, or closely approximate). When it comes to additional phonemic distinctions, a minority of northern speakers contrast /ʊ/ with /ɔ/, probably transferring the regional Low Saxon/Limburgish distinction over to their standard Dutch. Yet, we don't transcribe it (not least because they don't agree which words have which vowels). The 'potdicht' audio posted above by me clearly has [ʊ]: [ˈpʊdɪχt], but I'm not sure whether the speaker would consider it to be a phoneme separate from [ɔ]. He has a Low Saxon background, for what it's worth. Also, I think that Booij considers this [ʊ] to be the same as a shortened /oː/, thus [ˈtɕɛstʊ]. Similarly, a shortened /eː/ would be [ɪ] and a shortened /øː/ would be [ʏ]. Do we need extra symbols for those? I doubt it.

Rounded vowels are perceived as more back than the unrounded ones. The actual difference between [ɵ] and [ö] is not big. I do use it to indicate a back-central starting point, yes, to avoid diacritics. I think Wells uses it very similarly in AoE for one Irish pronunciation of STRUT. When you measure the formants, [ɵ] is back-central in the vowel space, with F1 and F2 matching that of [ɤ] which is back-central as well. The difference is in F3 (thus rounding). I know that the formant space is not the same as the official IPA quadrilateral (on the former, rounded vowels turn out to be retracted and sometimes raised in relation to their unrounded counterparts, that's why you get central [ɵ ɞ] in the back-central position where [ɤ ʌ] are, similarly [y ø] are front-central in terms of F2 and approach [ɨ ɘ] in backness, differing mainly in F3), but you probably get my point that [ɵ] and [ö] are not that different. Same with [ʉ] vs. [ü], etc. I have both [ʉ] and [ü] as allophones of /u/ in my native language and can never tell which one I'm using without measuring the formants, the difference is just too small. The textbooks tell me that it can only be [ü], but I'm afraid not.

⟨ou⟩ may be a good enough transcription, I have no problem with it. To me, it's a distinct phone that you can hear in e.g. Maastricht, or in older Northern Standard Dutch (that's why Collins & Mees talk about 'potential diphthongs'). But ⟨ɵu⟩ might be overly narrow/confusing when you consider that ⟨ɵ⟩ is used by some writers for the short lax ⟨u⟩ (orthographic representation), which most sources write ⟨ʏ⟩ - which is a good choice (I find Northern German /ʏ/ to be virtually identical to this). Sol505000 (talk) 21:36, 1 December 2024 (UTC)