Talk:Standard German phonology

Voiced allophone of /ʃ/
I just remembered that a linguist friend once mentioned a quite mysterious phenomenon: in certain words, such as wuscheln and kuscheln, the /ʃ/ is consistently realised as voiced [ʒ] by many people, although it does not seem to contrast with voiceless /ʃ/ anywhere. This phenomenon (for which I do not have any explanation, etymological or otherwise) was completely alien to me – I could at best vaguely recall encountering this pronunciation; I think it is limited to the traditionally Low- and Central-German-speaking regions and unknown in the south. Is anyone else familiar with it? Oh, it just occurred me that I had already left a comment on this phenomenon here. --Florian Blaschke (talk) 21:52, 26 April 2020 (UTC)
 * This parallels instead of  in Fussel (which you already mentioned in the archived discussion on de.WP) and Dussel. They all have the fricative between stressed  and, which makes me wonder if Muschel and tuscheln also affected. Love —LiliCharlie (talk) 22:48, 26 April 2020 (UTC)
 * Two things I will consider tomorrow: 1. Upper and Central German dialects don't consistently contrast intervocalic and . They tend to be voiceless in Upper German, and voiced/lenis in Central German; niesen and (ge-/nutz-)nießen are often pronounced the same. 2. The Dutch vowel system is similar to the Standard German one, but one difference is that it lacks the  contrast. Does this also apply to certain Low German dialects? Are Fussel and Fusel pronounced the same by some speakers? Love —LiliCharlie (talk) 23:40, 26 April 2020 (UTC)
 * Another instance of de:Binnendeutsche Konsonantenschwächung? --mach &#x1f648;&#x1f649;&#x1f64a; 05:09, 27 April 2020 (UTC)
 * As for Dussel and Fussel, we have to consider source dialects which have a intervocalic /s~z/-contrast after short vowels, i.e. which have not fully neutralized the length contrast as in Nase (< MHG nase) and Blase (< MHG blāse). For dialect borrowings with short vowels + /z/, our defective SG German spelling gives us two choices: use single s to indicate voicing, or double ss to indicate shortness of the preceding vowel. The latter is the "standard" choice, which is also applied to loanwords from Yiddish, e.g. Massel, Schlamassel, vermasseln (< מזל‎ mazl).
 * A similar word is verbaslen, which is also pronounced with a short /a/ by some speakers who distinguish intervocalic /s~z/ (at least I do). –Austronesier (talk) 08:10, 27 April 2020 (UTC)
 * That's a really good point; it confirms my suspicion that the phenomenon originates in traditional dialects and was transferred from there to Standard German, which does not have a comparable contrast (by the way, what regional variety do you speak?). So, which has /z/, the cognate of Nase or that of Blase? --Florian Blaschke (talk) 15:04, 27 April 2020 (UTC)
 * (flip-flopping between northern-ish colloquial German and a moderate Rhine-Main regiolect, occasionally drifting into traditional South Hessian depending on interlocutor) After another has come to my mind (quasseln), I guess it's a dialect phenomenon in an environment preceding [l̩]. Either it is sporadic voicing of [s] in [Vsl̩] → [Vzl̩], or sporadic shortening of the long vowel in [V:zl̩] → [Vzl̩]. Based on Dusel (cognate with Dussel) and verbaslen, I opt for the latter, but that's just a shot in the dark. In Massel, it is retention of original [Vzl̩]. (Btw I have [ʃ] in kuscheln, but [ʒ] in nuscheln. When code-switching to dialect, all become [ʒ̊]). –Austronesier (talk) 18:50, 27 April 2020 (UTC)
 * As for Dussel and Fussel, we have to consider source dialects which have a intervocalic /s~z/-contrast after short vowels, i.e. which have not fully neutralized the length contrast as in Nase (< MHG nase) and Blase (< MHG blāse). – As a speaker of a dialect that strongly keeps such length contrasts, I do not believe this is a comprehensive explanation. The dialect has lenis S in both «Nase» and «blaase», but fortis S in words like «hasse» or «Straasse». In my eyes, the odd thing about standard German S is not so much a deficiency in the alphabet, but rather the very retention of the fortis-lenis contrast. In the other standard German fricatives, the exact same fortis-lenis contrast has been lost completely so pairs like «schnaufen» and «Haufen» make perfect rhymes. I suspect that the distinction between fortis and lenis S is not as widespread as the orthography and the prescriptive pronunciation guides would have us believe. Instead, I suspect that most speakers pronounce the SS in «Schlamassel» exactly like the SS in «Schlüssel», be it with a fortis in the south, or with a lenis in the center and the north. --mach &#x1f648;&#x1f649;&#x1f64a; 20:09, 27 April 2020 (UTC)
 * I suspect I've completely misunderstood your point about Nase and Blase – did you simply mean that both have /z/ in those dialects, so that in them /z/ exists both after long and short vowels?
 * Oh yes, I totally forgot nuscheln and quasseln! You're right, it's striking that this phenomenon I'm talking about happens exclusively before (syllabic) /l/ – at least I can't think of a counterexample. Maybe mach's and your initial thoughts were misguided and the phenomenon has a more simple explanation like that: a phonetically rather strange (why only before syllabic /l/?) but regular sound law. I'm not sure about the alternative to interpret it as a shortening – it does seem to work better for /z/, and makes more sense phonetically, but as far as I can see it doesn't explain the [ʒ]s like in nuscheln. Or maybe they're a separate phenomenon again?
 * In a stereotypical Hessian accent, it sounds to me as if there is no voiceless/fortis intervocalic /s/ at all, all (?) intervocalic fricatives being voiced/lenis (as in the clichéd old saw about the oh so criminal Hessians ...), but I have to admit I have no clue what genuine Hessian dialects and Hessian-coloured regional Standard German are really like. --Florian Blaschke (talk) 22:48, 27 April 2020 (UTC)
 * So you also have lenis S in Schlüssel, but fortis in Schlüsse? I wouldn’t know the explanation for such a phenomenon. Very interesting, though (even if clearly unencyclopedic). Speculation: What if it were a remnant of former Konsonantenschwächung? When the distinction between lenis and fortis S was reintroduced, it was only done so in favourable environments: between vowels, but not between vowels and /l/ or finally. Such a partial reintroduction would not be unprecedented. I know of dialects that have Straass, but Ströössli – a remnant from former regular Strooss/Ströössli. --mach &#x1f648;&#x1f649;&#x1f64a; 05:13, 28 April 2020 (UTC)
 * In areas like Hesse and Upper Saxony, we have uniform lenis [z̥] in medial position. (SG speakers in my area percieve this as [z], which results in mock-dialect pronunciations like ['hɛzɪʃ] hessisch.) In northern colloquial speech, and also in relatively 'new' dialects like Berlinerisch or Missingsch, there is a clear distinction between Dussel and Kessel, and generally a genuine voicing contrast. Your "speculation" could thus mean that code-switchers from dialects with general lenis S "assigned" dialect words which lack a target in SG to lenis S [z] when switching to SG: dialect/regiolect [dʊz̥l̩] ≙ SG [dʊzl̩] because Dussel is not SG, but dialect/regiolect [kɛz̥l̩] ≙ SG [kɛsl̩], because Kessel exists in SG. Eventually, these dialect words entered into regional and supraregional speech which has a /s~z/-contrast. A possible scenario, but then, why does this only happen before [l̩]? –Austronesier (talk) 09:27, 28 April 2020 (UTC)
 * Ok, here's my final OR (🙊) on this: Dussel, Fussel, verbas(s)eln originate from a northern lect in which a long vowel in the environment [V:zl̩] is pronounced sufficiently short as to be percieved as a short vowel when borrowed into colloquial SG, especially if there is no equvalent target in SG. So /du:zəl/ [duˑzl̩] was "adjusted" to [dʊzl̩] rather than to [du:zl̩]. So no need to posit retention of the Middle High German / Middle Low German length contrast before $⟨s⟩$, which would be odd indeed in an area where vowels in open syllables were generally lengthened. –Austronesier (talk) 10:08, 28 April 2020 (UTC)
 * Just for clarification: my own variety of SG is very much southern and therefore my own realisation of Schlüssel cannot have /z/ even in theory, as my own variety lacks this phoneme entirely. That's why I found the phenomenon so amazing and curious. As for Ströössli, I'd like to mention that this form is curious and must itself be analogical given that MHG must have had stræzlîn (compare Bavarian Straßl with – secondarily shortened – front /a/), so you'd expect Sträässli with ää (or whatever MHG æ turns into in the respective dialect) as the diminutive of Strooss.
 * I see! Makes sense. So perhaps the voicing was secondarily extended from these examples to non-standard lexemes of comparable structure with /ʃ/? --Florian Blaschke (talk) 01:45, 29 April 2020 (UTC)
 * In Dutch merged with  (mostly in favor of the latter, except before nasals), not with . It's  that merged with . So, by extension, the question should probably be whether  can be merged with  in the Low German speaking area, and I've heard them nearly merged in some speakers (though whether that's common is another question) who use a very open  that approaches . But maybe they're kept distinct as  vs. . When it comes to the – merger, I'm not sure whether there's an actual merger in any area, but  tends to be shortened in e.g. Zug . Kbb2 (ex. Mr KEBAB) (talk) 09:04, 27 April 2020 (UTC)

I'm not sure, but Assi could be a word that has short V++something else than. (Wessi and Ossi clearly have .) Love —LiliCharlie (talk) 10:16, 28 April 2020 (UTC)
 * Yes, that's because it's a shortening from Asozialer, which is phonemically /ˈaːzoːtsjaːlər/, so it's a bit of a special case. But you're right that it's unclear why it shouldn't automatically turn into /ˈaːziː/ or perhaps /ˈasiː/, given that SG rule that bans /z/ after short vowels, and given that my variety of SG lacks /z/, it struck me as very odd when I learned that "northerners" pronounce Assi as /ˈaziː/, despite the spelling. --Florian Blaschke (talk) 01:45, 29 April 2020 (UTC)
 * I have misunderstood your observation about the V_l̩ environment. I thought you were making an observation about your own speech. For the record, I have avoided ⟨z⟩ – I believe that truly voiced /z/ is only found in languages like Polish or French, whereas obstruent voice is accidental in virtually all varieties of German – except in /v/ which rather behaves as a continuant in German, and except in dialects like (if I am not mistaken) Styrian with its Slovene substratum. I do not understand why you’d expect the diminutive of Strooss to be Sträässli. The umlaut of o is ö, so Ströössli is completely regular, just like Sohn – Söhnlein etc.
 * Regarding Asozialer, I would guess that only the stressed syllable has length /azɔʹtsjaːlər/, so when you cut off the syllables, short /a/ is not surprising. In this region, you can hear /ʹaz̥o/ used as an adjective in youth speech, but then again, short vowels before lenis consonants are very common in our dialects.
 * In my dialect, most of the SCH words you have mentioned are not used or feel like recent loans (such as /ˈkxʊʒ̊lə/ or /ˈkxʊʃlə/ from standard German kuscheln). An exception might be /ˈnʊʒ̊lə/ (nuscheln), which has lenis /ʒ̊/ and would contrast with fortis /ʃ/ in words such as /ˈb̥ʏʃlə/ (büscheln). Generally speaking, non-initial lenis /ʒ̊/ appears to be rare in the native core vocabulary, but it certainly occurs in a few words, both after short vowel as in /ˈaʒ̊i/ and /ˈpeʒ̊ə/ (traditional dialect forms of the name Ernst and Peter), /ʹv̥aʒ̊o/ (Fascho); or after long vowel as in /ˈnʊːʒ̊ə/ ‘to rummage’, /ˈloːʒ̊ə/ (Loge). --mach &#x1f648;&#x1f649;&#x1f64a; 06:47, 29 April 2020 (UTC)
 * I think that's an immaterial point – our article does notate the phoneme as /z/, and at least intervocalically, /z/ is definitely voiced in "northern" varieties. As for Strooss, it's from MHG strâze, and its umlauted form had the same vowel as in kæse, which I know is Chääs in many Swiss German dialects, so Sträässli is absolutely expected as the original (!) form that was analogically (!) replaced by Ströössli. Rather, öö is old only in words like Bröötli, as in MHG, long œ is the umlauted counterpart of ô as in brôt. Sohn did not even have an o-like vowel in MHG – it was sun. (It's recorded as [z̥ũː] in conservative rural Central Bavarian, but has generally been replaced by [b̥ʊɐ] – hasn't it been replaced in Swiss German dialects as well?) Evidently MHG â ([aː] or [ɑː]) was rounded in many (?) Swiss German dialects and merged with the reflex of MHG ô (originally [ɔː]).
 * I'm pretty sure that asozial, and also Asozialer, has initial stress in northern SG varieties at least, and your phonological transcription looks off to me; I still stand by mine. Your dialect is Swiss German, right? SG generally disallows /z/ after short vowels, at least in the north; not sure about Swiss SG. --Florian Blaschke (talk) 08:39, 29 April 2020 (UTC)
 * I looked up asymmetrisch in Mackensen (2005) and Krech et al. (2009), and both pronouncing dictionaries give first and a variant pronunciation with stressed penultimate second. The strange thing here seems to be the shortness of the vowel in the initial stressed open syllable. Love —LiliCharlie (talk) 09:34, 29 April 2020 (UTC)
 * I know we use IPA [z], and I am not against it since it is widely used. I still believe voice is only an accidental feature of German obstruents, not a distinctive feature. Distinctive features are the ones that are essential to a phoneme. If they are not present, the phoneme will no longer be recognizable. Accidental features are the ones that may or may not occur, e.g. depending on the environment where a sound occurs. Even in northernmost German, the lenis obstruents are only accidentally voiced, especially between vowels. This is different from languages like French or Polish that have truly voiced obstruents. Jokingly, one might say that the IPA is to blame because they sanctified French phonetics by elevating voice to the one and only distinctive feature for twofold obstruent series – unlike e.g. older German transcription system that used broader definitions for twofold obstruent series.
 * With regard to umlaut, my point is that umlauts pairs like a/ä or o/ö are regular, while pairs like a/ö or o/ä are irregular – whether or not o is original or shifted from a. When Straass is shifted to Strooss, then unshifted Sträässli would be irregular – just as irregular as retaining shifted Ströössli when Strooss is shifted back to Straass. You seem to expect o/ä – do you know any variety that has it (e.g. Strooss and Sträässli at the same time)? That would be very interesting. And yes, Swiss German dialects have kept the original U in Sohn, /ˈz̥ʊːn/ in my dialect (with monosyllabic lengthening, which is more widespread than open-syllable lengthening in the southwest and less common in the north and – big surprise – far less well-known).
 * Slightly more on-topic, I do not think the combination of stressed short vowel + single consonant letter in (some pronunciations of) words like asozial or asymmetrisch is surprising. The rules for double consonant letters only fully apply to typical native words with monosyllabic stems. Polysyllabic words do not follow these rules, cf. other examples like (some pronunciations of) Telefon, Roboter, etc. --mach &#x1f648;&#x1f649;&#x1f64a; 11:13, 29 April 2020 (UTC)
 * Maybe it's because voicing can be easily measured and defined phonetically (vibration of the vocal chords during the production of a phone), while fortis/lenis is a highly abstract property that does not correlate with any concrete phonetic feature (it's apparently not the same as consonant length, either, for example). However, I don't think it matters; intervocalically, /z/ is at least partially voiced in the varieties I'm interested in here. Whether that voicing is accidental is of no concern to me. Moreover, dismissing the voicing as the relevant feature would only shift the problem: why are there lenis counterparts of /s/ and /ʃ/ in certain lexemes where they are not expected, and not there in other lexemes in the same kind of environment?
 * The pair Strooss : Sträässli is indeed irregular and unpredictable from a phonetic and synchronic point of view – not necessarily from a structural one, though. If a dialect has [æː] as the umlauted counterpart of [oː] in certain cases, where it can be shown that the [oː] goes back to an older [aː], the problem has an explanation, if only a diachronic one. However, for dialectologists, such cases are quotidian. Just like the analogical leveling Strooss : Ströössli (patterned after models like Schooss : Schöössli, Flooss : Flöössli, Nooss : Nöössli), which makes the relationship regular and predictable again. However, from a point of view informed by MHG, like mine, it is Strooss : Sträässli that is regular and predictable and Ströössli is the surprising form – even if its explanation is immediately obvious! ;-)
 * I'm not sure which pronunciations you mean here. Do you mean /ˈteːləfoːn/ and /ˈroːbɔtər/? Oh, you probably mean /ˈtɛləfoːn/ and /ˈrɔbɔtər/, but that phonotactic structure is by no means irregular. Voiced/lenis stops are perfectly allowed in SG after short vowels – just think of /ˈrɔbə/. It's voiced/lenis fricatives – specifically /z/ – which I intuitively thought were not. But presumably I was simply misguided and there is no such phonotactic restriction in SG. And as for spelling – that's not what I was concerned with, but to me Assi definitely looks like it should be pronounced /ˈasiː/. Personally, I'd spell /ˈaziː/ Asi, but I concede that first, the SG spelling system is defective here and there is no obvious way to spell /z/ and /ʒ/ after short vowels, and second, that lenis /z/ (like lenis sibilants in general) after short vowels does strike me, with my native southeastern SG variety, as odd. Indeed, /ˈazʏmeːtrɪʃ/ irritates me ever so slightly too – I pronounce it with fortis /s/. --Florian Blaschke (talk) 19:38, 29 April 2020 (UTC)
 * Sure, phonetic voice is well-defined, but so are aspiration, length and force. A phonetic alphabet that claims to be international should not take sides by preferring one of them. And mind you that preferring voice does not reduce ambiguity. We have ambiguity as it is. In practice, the supposedly voiced signs are used for sounds that may be voiceless, e.g. the German lenis obstruents.
 * Your diachronic musings about the umlauts are interesting, but is there any evidence to support them? The evidence I know suggests all dialects that have shifted a to o have also shifted its umlaut ä to ö. Do you know of any dialect that has shifted a to o while keeping its umlaut ä, thus supporting your theory? Of course, it would have to be dialect that keeps rounded front vowels, because otherwise, if a dialect without rounded front vowels has Strooss/Streessli, there is no way of telling whether Streessli is a continuation of Sträässli or unrounded Ströössli.
 * With regard to lenis vs. fortis S or SCH after short vowels, I still suspect that a majority of German speakers does not distinguish lenis vs. fortis S or SCH. I fear most data on German phonetics is tainted by the prescriptive pronunciation manuals, being based on professional speakers who have been trained to follow the prescriptive pronunciation, on small groups of university students in speech laboratory settings, or on people who introspect so hard they convince themselves to perceive the prescriptive pronunciation. So I suspect a majority of German speakers does not have a fortis–lenis distinction in words like reissen/Kessel(/büscheln?) vs. reisen/Dussel/kuscheln – except for speakers who strictly follow the prescriptive pronunciation and except for speakers of dialects that have consistent fortis–lenis oppositions (including the Hauffen–schnaufen distinction I mentioned). Consequently, double ⟨ss⟩ in spelling would behave very much like the doubling of any other consonant, not indicating a different consonant pronunciation, but a checked vowel. Of course, I’d be happy to renounce my suspicion if there is convincing data, that is, data from near-spontaneous speech by laypeople, not from professional speakers or a few students. --mach &#x1f648;&#x1f649;&#x1f64a; 07:41, 1 May 2020 (UTC)
 * Concerning umlaut, I agree with . Umlaut is a fully prodictive synchronic process in modern German, even more so in Alemannic lects which have extended umlaut to word classes which historically had no umlaut. In such cases, paradigms tend to keep regular, which also includes regularzing the outcome of umlaut after rounding/raising of a. I am not aware of a dialect which has o umlauted to e/ä unless it displays general loss of rounding of ö and ü. And the Straass/Ströössli case is a nice example where we can witness several strata of vowel changes due to dialect mixing and/or interference of the standard language.
 * As for the distiction between fortis and lenis S (and SCH), I don't understand what is "prescriptive" about /fʊzəl/ or /ʃʊzəl/. These words belong to the colloquial lexicon, are seldom heard in official settings, and the expected "prescriptive" realization would be a spelling pronunciation like /fʊsəl/ or /ʃʊsəl/. Yet native speakers of northern colloquial German say [fʊzl̩] with a sibilant quite distinct from the one in [kɛsl̩] not because they aspire to comform to a prescriptive ideal, but that's what they here from their parents and peers. In most northern areas, people have shifted from local dialects to SG (in the shape of northern colloquial German) during the last 150 years, and they equated their lenis /z/ [z~z̥] with SG /z/ and their (marginal) fortis /s/ with SG /s/. And with the higher prestige of the northern pronunciation (standard and colloquial) and its overrepresentation in German media, the distinction starts to creep into the spoken language of areas which traditionally had merged medial /s/ (< MHG z) and /z/ (< MHG s), thus demerging reisen and reißen. FWIW, 's observiation about lenis S and SCH was not made under laboratory conditions I presume, but in informal interaction.
 * Btw, I distinguish between Assi /azi:/ (< Asoziale(r)) and Assi /asi:/ (< Assistent) when speaking with non-Hessians, but conflate them when speaking with local peers. My daughter would say /azi:/, but she doesn't use it–they call them /azlak/ nowadays. –Austronesier (talk) 09:54, 1 May 2020 (UTC)
 * Sorry for not being clear. What I suspect to be prescriptive influence is not /fʊzel, ʃʊzəl/, but the existence of a fortis–lenis contrast in S and SCH. My null hypothesis is that most speakers have just a single S sound in their normal everyday speech (whether or not they are Hessian) – the same one in words like Schussel, Asi, Hase, reisen and in words like Kessel, Assi, Strasse, reissen. At the same time, I think most speakers honestly believe that they differentiate two kinds of S sounds, not because that is what they hear from their parents and peers, but because that is what they observe in writing and what they have learnt in school and what they hear from professional speakers and what they produce themselves in careful, slow speech. Therefore, it is only natural that the distinction will be found when elicited, whether by introspection or by discussion.
 * It is not important whether the non-differentiated S sound is transcribed with /s/ or /z/ (I would prefer a lenis sign that does not imply voicedness). But I would expect that its northern pronunciation sounds unusually voiced to southern ears, especially after short vowels, where voicing is blocked in the south. In ensuing introspection or discussion, the distinction will be found naturally.
 * I am aware that my believe in this null hypothesis is by itself problematic when I use it to dismiss evidence to the contrary. But I hope/expect I am sufficiently open-minded to accept evidence to the contrary if the methodology is sound. I am setting a high standard though. Most phonetic data is gathered in laboratory settings with a ridiculously tiny sample sizes, and there is a tradition of pure introspection. Of course, this is only the standard I am setting for my personal conviction. There are no such restrictions on Wikipedia. --mach &#x1f648;&#x1f649;&#x1f64a; 14:39, 1 May 2020 (UTC)
 * It should be easy to show pictures of ein Hase vs. eine Straße, etc. to preschoolers who can't yet read or write, and ask them what they see. I imagine many northerners will not produce exact rhymes. Love —LiliCharlie (talk) 15:44, 1 May 2020 (UTC)
 * Sorry for the off-topic discussion, but I don't know of any evidence that phones can directly change into other phones dependent on whether there is a related form or not. If the [æː] in "Middle High Alemannic" (k)chæse turns into [ɛː] or whatever, or stays the same, so should the [æː] in stræzlîn – at least initially. Once old â [aː] has been rounded to [oː] it is trivial to analogically replace the now irregular diminutive with a regular diminutive with the synchronically expected umlaut [øː], but [æː] won't turn into [øː] directly – unless [xæːs] also turns into [xøːs]. And I can predict that it won't, and that there will most likely be evidence that shows this (if the evidence is rich enough), even if you might have to go back a few centuries to find something like strôz(e) besides stræzlî in the same dialect. That's how sound change and analogy are known to work. It's an empirically based insight.
 * As for the question of whether a contrast between fortis /s/ and lenis /z/ as well as fortis /ʃ/ and lenis /ʒ/ exists in SG at all (or, for that matter, in dialects), that's an interesting problem that's also directly relevant for the article overleaf. I'm really curious what the literature has to say about this, and if any doubt about this contrast is voiced at all. --Florian Blaschke (talk) 22:01, 1 May 2020 (UTC)
 * From what you are saying, your “empirically based insight” (that â can shift to ô while its umlaut æ remains unshifted) not only lacks supporting evidence, but cannot be falsified, since you will dismiss all contrary evidence (like SDS I 89) as mere analogy or as not far enough back in time. Continuing this discussion seems pointless.
 * Phonetic research with prescholars preschoolers is an interesting idea. Apparently, some researchers do this, e.g. the Felicitas Kleber’s group, though I have not found such analyses with S or for northern German. --mach &#x1f648;&#x1f649;&#x1f64a; 12:53, 2 May 2020 (UTC)
 * Very few prescholars are also preschoolers, I'm afraid. Love —LiliCharlie (talk) 13:05, 2 May 2020 (UTC)
 * Your contrary hypothesis, that MHG æ can shift to [øː] directly, even when it usually doesn't in the same dialect – curiously exactly in those cases when it's not supported by a related form –, without any analogical reformation having interfered at any point, is no less unfalsifiable. On the other hand, mine is supported by empirical and logical arguments, and thus the standard way to account for this phenomenon: regular sound change, plus proportional analogy, is widely attested, irregular or "morphologically sensitive" sound change like postulated by you is not. Lost evidence is a regular occurrence, and I doubt you've even looked for the kind of evidence I've inferred likely exists (i. e. an older form of the dialect in question still featuring strôz(e) besides stræzlî). --Florian Blaschke (talk) 15:30, 21 May 2020 (UTC)
 * Of course I have not looked for your “lost evidence”. Why should I? I have produced evidence for my point of view: numerous dialects that have _Strooss/Ströössli_ (cf. SDS I 89). You have not produced any evidence for your point of view (speculation does not count), and now you want me to do that job for you? Seriously? --mach &#x1f648;&#x1f649;&#x1f64a; 15:47, 21 May 2020 (UTC)
 * The fact that numerous dialects have _Strooss/Ströössli_ now does not say anything about the historical intermediate stage postulated by me on principled grounds (with good reason, not arbitrarily as you seem to think) – there is no solid evidence for sound change working the way you imagine it, conditioned by morphological alternations rather than phonetical environments. Since you are postulating an unattested and widely refuted phenomenon, the burden of proof is on you. --Florian Blaschke (talk) 08:37, 22 May 2020 (UTC)
 * Relevant citations: Blevins, Juliette. 2004. Evolutionary Phonology: The emergence of sound patterns. Cambridge: Cambridge University Press.
 * Hill, Eugen. 2009. “Die Präferenztheorie in der historischen Phonologie aus junggrammatischer Perspektive”. Zeitschrift für Sprachwissenschaft 28. 231–263.
 * Blevins 2009, a case study. --Florian Blaschke (talk) 08:50, 22 May 2020 (UTC)
 * I would say that in native words in standard pronunciation /z/ is prohibited after short vowels. I don't think this is a set rule of German phonology however, this combination just happens to not occur in native words as a result of diachronic sound changes, it's not actually prohibited. Germans don't struggle with pronouncing /z/ after a short vowel. Thus I see no problem with it occuring in slang / regional / nonstandard words or pronunciations, or in loanwords. "Asi" is both a slang term and derived from a loanword. Tajoshu (talk) 19:00, 26 June 2024 (UTC)

I think you have misunderstood my point of view, and I am sorry I have not understood your point of view well enough to point this out earlier. I am not claiming that there is a morphologically conditioned sound change from ÄÄ to ÖÖ. Instead, the question is whether diminutive umlaut or sound change takes precedence.

In my point of view, sound change takes precedence over diminutive umlaut. The diminutive has the umlaut of whatever vowel the base word happens to have, derived in a productive synchronic process. When the base word is affected by sound change, the diminutive will have the umlaut of the affected vowel.

In your point of view, diminutive umlaut takes precedence over sound change. The diminutive keeps the umlaut of the base word’s original vowel. When the base word is affected by sound change, there no longer is a synchronic umlaut relation between the diminutive vowel and the base word’s affected vowel. In a secondary step, the umlaut relation could be restored by analogy, so the outcome would be the same as in my point of view.

If the outcome is the same, then how can the two cases be distinguished? Occam’s razor favours my point of view since it does not require a secondary step for explaining Strooss/Ströössli. Also, the diminutive umlaut certainly is synchronically productive, cf. umlaut diminutives of anglicisms like Jöbli, Compüterli, or Glöönli. For the two-step point of view to be convincing, there needs to be evidence for the intermediate stage where the diminutive no longer is a synchronic umlaut of the base vowel. Such a stage would be very surprising and highly unusual. There would be two different kinds of diminutives for words with OO. Some would have regular diminutives with synchronic umlaut ÖÖ (words like Stooss/Stöössli), while others would have irregular diminutives with the base word’s original vowel’s umlaut ÄÄ (words like Strooss/Sträässli). Such irregular diminutives are hard to accept unless there is evidence. The literature you have cited is just arguing against morphologically conditioned sound changes, which is something I have never postulated. --mach &#x1f648;&#x1f649;&#x1f64a; 13:05, 22 May 2020 (UTC)


 * The more regular and productive a morphohonological process is (such as diminuitive umlaut), the less it will tolerate irregularites. So it is quite possible that paradigmatical levelling can occur at the same time as the sound shift that triggers it, without an intermediate "messy" stage. I'll try to look for attested cases, I vaguely remember to once have read about it. Btw, just wondering: is aa/öö the regular umlaut pattern in lects that have Straass/Ströössli? –Austronesier (talk) 15:42, 22 May 2020 (UTC)
 * I do not know Zurich German sufficiently to tell whether all AA umlaut to ÖÖ. Maybe we can ask ? --mach &#x1f648;&#x1f649;&#x1f64a; 16:46, 22 May 2020 (UTC)
 * I didn't read the entire discussion here ... But Ströössli is called «Systemumlaut» in traditional dialectology (that means: analogical, e.g. intraparadigmatic). Straass/Ströössli only appears in Zurich German and in no other Swiss German dialect because Zurich German had once, like all other northern Swiss German dialects, «Verdumpfung» from MHG. /aː/ to /ɔː/ – thence the plural, the diminutive and other derivations per /ɶː/, e.g. /rɑːt/ (so today, but earlier: /rɔːt/) «Rat», /rœːt/ «Räte», /rœːtɪg/ «rätig», /rœːtlə/ «räteln», or /ʃtrɑːsː/ (so today, but earlier: /ʃtrɔːsː/) «Straße», /ʃtrœːsːlɪ/ «Sträßlein». Later, in the 16th century? and on unknown grounds, there took place «Rücksenkung», e.g. /ɔː/ was lowered back to /ɑː/ in most parts of the canton of Zurich (plus Rapperswil), whilst in the Zurich Upperland /ɔː/ was closed to /oː/. But the plurals, the diminutives and the other derivations were NOT touched by this younger sound change, thus today /rɑːt/ «Rat», /rœːt/ «Räte», /rœːtɪg/ «rätig», /rœːtlə/ «räteln», and /ʃtrɑːsː/ «Straße», /ʃtrœːsːlɪ/ «Sträßlein». The same in Zurich Upperland: now /ʃtroːsː/, but still /ʃtrœsːsːlɪ/ etc. By the way: Concerning to Systemumlaut there is some fluctuation, e.g. Verdaacht > verdèèchtig and verdö̀ö̀chtig.
 * But BEYOND of the cases of «Systemumlaut», the Umlaut of MHG. /aː/ ist regulary lautgesetzlich /ɛː/, e.g. /ʃʋɛːr/ «schwer» < OHG. swâri, or /rɛːtsəl/ «Rätsel» (in the latter case the connection to Rat, raten wasn't so clear, thus no Systemumlaut, and it seems to be a borrowing from the literary language; but: /rɑːtə/ «raten», /rœːtlə/ «räteln»). And in cases of SECONDARY lengthening, there are /ɑː/ > /æː/ (like the historic Secondary Umlaut), /ɑː/ > /ɛː/, and /ɑː/ > /e/ (like the historic Primary Umlaut), e.g. /ɑːrm/ «Arm», /æːrm/ «Arme», /æːrmlɪ/ «Ärmlein»; /ɑːrm/ «arm», /ɛːrmər/ «ärmer»; /bɑːd/ «Bad», /beder/ «Bäder», /bedlɪ/ «Bädlein». --Freigut (talk) 09:12, 23 May 2020 (UTC)
 * Having read the entire discussion (which is very interesting), Straass/Ströössli only appears in Zurich German and in no other Swiss German dialect because Zurich German had once, like all other northern Swiss German dialects, «Verdumpfung» from MHG. /aː/ to /ɔː/ is pretty much what I suspected (minus the geographic part) and I was about to suggest it. I wouldn't do it as eloquently, so thank you for explaining it. Kbb2 (ex. Mr KEBAB) (talk) 10:54, 23 May 2020 (UTC)

Velar nasal as /nɡ/
How much support is there for this analysis? It might be reasonable historically or morphophonologically but it sounds like a fringe view synchronically. If it is, we might as well not mention it at all. Nardog (talk) 23:30, 19 November 2020 (UTC)
 * Do you know any recent Fundstelle for this topic beyond the UTB source which I added last year? Nardog is still curious about it, and so am I, but keyword searches only lead me to the textbook witticism about the complementary distribution of and, or to quite old sources like this one, which mentions Isačenko (1970) as trigger of the debate. –Austronesier (talk) 07:32, 2 March 2021 (UTC)


 * No I don't. I remember I liked the brief presentation of ten categories of arguments in Lass (1984:216–220) which is based on Dressler (1981). All old stuff. Love —LiliCharlie (talk) 09:13, 2 March 2021 (UTC)
 * The theory does predict correctly the occurrences of [ŋ] vs [ŋg] based on the phonological environment, however I see one point that clearly speaks against it, and that is [ŋ] in final position. In this theory, final [ŋ] would represent a word-final occurence of /ng/, contrasted with /nk/, which seems highly unlikely to me given that for every other obstruent, there is no such fortis-lenis distinction in final position due to final obstruent devoicing. The theory would only make sense to me for those speakers that pronounce final  as [ŋk(ʰ)] and thus don't have this distinction. Tajoshu (talk) 19:07, 26 June 2024 (UTC)

Diphthongs
Where can I find further information on the difference between phonemic and phonetic diphthongs? The difference is clear, but I would like to know where these terms come from. —  GPodkolzin   Talk 14:19, 9 July 2021 (UTC)

Voiced sibilants
Related to the discussion further above, I'm surprised to find that there is no mention in the article of the fact that in some (southern) varieties of Standard German, voiced sibilants are consistently devoiced or even merge with their voiceless counterparts, at least in some environments (such as word-initially). As far as I'm aware, this is the case in Austrian Standard German and Swiss Standard German, for instance, and also in my own regional Bavarian-coloured variety of Standard German. Even in the north, /ʒ/ and /d͡ʒ/ are not always realised as such, and may not even be accepted as phonemes at all. --Florian Blaschke (talk) 17:40, 17 March 2023 (UTC)
 * There is a note to that regard: “The obstruents /b, d, ɡ, z, ʒ, dʒ/ are voiceless lenis [b̥, d̥, ɡ̊, z̥, ʒ̊, d̥ʒ̊] in southern varieties, and they contrast with voiceless fortis [p, t, k, s, ʃ, tʃ].” I am not entirely convinced of the latter part of that note. Devoiced [ʒ̊, d̥ʒ̊] merges with [ʃ, tʃ] – in stark contrast with devoiced [b̥, d̥, ɡ̊, z̥] which remain distinct from [p, t, k, s]. I am amending that note accordingly. --mach &#x1f648;&#x1f649;&#x1f64a; 19:04, 17 March 2023 (UTC)
 * Thank you! I used search strings like /z/, /z or z/, and completely missed that paragraph while skimming too. At least the paragraph is now improved (though it could use a reference). I just realised there's a whole section on the subject! --Florian Blaschke (talk) 23:29, 17 March 2023 (UTC)

Fricative realisation of /j/
I'm surprised that [ʝ] is described as a common or even normal realisation of /j/ in several sources. In my personal idiolect, /j/ may feature slight frication when it is especially stressed, but otherwise it does not seem to be anything other than an approximant, and may even be lowered from a [i̯] to a half-vowel as open as [e̯]. I do not recall hearing [ʝ] often from other speakers either, even from more northern regions in Germany – except perhaps in a strong far northern accent influenced by Low German –, and in any case it does not seem to be a common realisation outside of Northern Germany and parts of the Rhineland, perhaps. I'm pretty sure that such a realisation would be perceptively salient enough to stand out to me, like it does in certain northern regiolects. That my impression is probably not completely mistaken is suggested by this discussion, where many participants, even other Wikipedians with linguistic and phonetic training, do not even seem to be passively familiar with this realisation, or judge it rare. Additional evidence is the fact that there is not even the slightest mention of it in the German version of this article, either (unless I've missed it), and it solidly classifies /j/ as an approximant. In any case it is odd that most of the sources do not give any indications towards a regional distribution of this realisation, and some even treat it as the norm. --Florian Blaschke (talk) 23:49, 17 March 2023 (UTC)

Problems with adding map
I tried to add this map to the article https://commons.wikimedia.org/wiki/File:Velar_fronting_after_ae.pdf, but it does not show up. Help is appreciated. The source is page 456 of https://langsci-press.org/catalog/book/331. The pdf is open access/CC-BY Jasy jatere (talk) 07:49, 21 March 2023 (UTC)

Final-obstruent devoicing is not phonemic?
AFAIU the conventional wisdom in phonology is that if Rad and Rat are pronounced identically, then they are represented with the same string of phonemes. I think you're mistaking morphophonemes for phonemes. Nardog (talk) 20:42, 27 July 2023 (UTC)
 * Yes, but not all authors recognize morphophonemic transcriptions as separate from the phonemic ones. I've only seen Basbøll and maybe some other Danish authors separate the two. For Dutch, Booij gives only phonemic transcription, and he insists on writing ik (cf. wij ), ik  (cf. wij ) and ik  (cf. wij ). The final  are the same as  in this position and nobody argues otherwise. Unless we want to separate the morphophonemic analysis from the phonemic one, I think that my approach is correct. Rad is stored in the brain together with the plural Räder  and so is Rat (plural: Räte ). And let's not even dive into Slavic declension and conjugation. Speakers are obviously aware of the underlying distinction, even if they are not able to produce it at will, or at least struggle to do so. Whenever a phonemic transcription is too confusing, a phonetic one can be swapped for it instead - unless we're discussing phonemes. I see no problem with that.
 * I'm not sure whether the list of homophones is that long, but it probably features more common words. Sol505000 (talk) 21:07, 27 July 2023 (UTC)
 * Wiese (1996:200) supports this. Sol505000 (talk) 21:19, 27 July 2023 (UTC)
 * In the conventional phonemic analysis of German, Auslautverhärtung is the neutralization of the phonemic voicing contrast in final and pre-consonantal position. Since the choice of phoneme from the voiced/unvoiced pair becomes arbitrary, many phoneticians have posited archiphonemes in this position (e.g. /T/ for the voice-neutral apical stop). But more commonly, the phoneme that corresponds to actual produced sound is chosen, thus simply /t/ instead of /T/. A level of abstraction that incorporates morphophonological processes is also not uncommon (cf. Wiese), but heavily relies on a particular theoretical approach. Treating final devoicing as phonetic only is clearly not NPOV. –Austronesier (talk) 21:40, 27 July 2023 (UTC)
 * Speakers are obviously aware of the underlying distinction, even if they are not able to produce it at will, or at least struggle to do so Speakers are probably aware mostly because they know that Rat and Rad are spelled differently - I've certainly had occasions where I've discovered I was saying two things as homophones that I always assumed were said differently because they are spelled differently. Many older stages of the language wrote final voiced stops as their devoiced counterparts: Nibelungen str. 2: si wart ein scœne wîp (=Weib); Str. 5: so was ir lant (=Land) genant, Str. 18 sît lebte diu vil guote vil manegen lieben tac (=Tag). Without knowledge of modern German spelling, the plural forms would probably appear to just randomly sometimes include a voiced or an unvoiced consonant.--Ermenrich (talk) 20:21, 31 July 2023 (UTC)


 * A related issue, also relevant to the section about the phonemic analysis of /ŋ/ as /nɡ/, final devoicing is even found for /ŋ/ ~ /nɡ/ in the north of Germany, yielding /ŋk/ ~ /nk/. In varieties with this feature, it is quite evident that [ŋ] is to be analysed as /nɡ/ phonemically, indeed. --Florian Blaschke (talk) 11:10, 10 October 2023 (UTC)

Sound changes and mergers section entirely unsourced
The entire section on sound changes and mergers is unsourced. Furthermore, the first section makes claims about a merger of the vowels ä and e when the vowel section claims that there is a dispute about whether there is actually any difference between the two.—-Ermenrich (talk) 13:01, 24 November 2023 (UTC)