For a teaching material I needed a good example of vocalic mutation of the root, aka Umlaut, and I got stuck at the fact that, while the Umlaut is often postulated for some reconstructed languages, there seem to be few, if any, good examples of this phenomenon in some contemporary living languages.

By Umlaut/vocalic mutation I mean a morpho-phonological mechanism such that the vowel of the root (or stem) is assimilated to the vowel of the ending. Note that I want the ending vowel to be preserved, therefore such examples as the Old English nom.sg. fōt ~ nom.pl. fēt are not accepted, because the ending that provoked the vowel mutation is never attested in the written sources.

Curisouly enough, the opposite phenomenon, i.e. the vocalic harmony (when it's the ending vowel to change accordingly to the vowel of the root), is well attested cross-linguistically, for example in Turkish. Not sure what general conclusion can be drawn here, but still...

  • 2
    The difficulty of finding cases where umlaut alternations have been phonemicized but the trigger is still present implies something important about umlaut, i.e. that it tends to get phonologized only when the trigger is lost. I don't know if it's a suitable approach for your specific class, but if I were teaching umlaut, I might stress that very point.
    – TKR
    Sep 8, 2017 at 19:46
  • Yes, this seems to be a logical conclusion. I wonder why it's so. Why is it different in this respect from the progressive assimilation? Sep 8, 2017 at 19:56
  • 1
    By progressive assimilation do you mean vowel harmony? That is rather different in that the direction of assimilation is root-to-suffix rather than suffix-to-root, so the trigger (the root) is not likely to ever be lost. A closer parallel might be a case where a prefix triggers umlaut in the first vowel of a root while being lost itself; I don't know if that's been described.
    – TKR
    Sep 8, 2017 at 20:33
  • @TKR I realize I've made a stupid rhetorical question in my preceding comments. You're right about the root being always preserved. Sep 8, 2017 at 20:38
  • There's suffix to root assimilation in the French mid vowels (search for articles about French vowel harmony by Zsuzsanna Fagyal), for example j'aidais /ʒɛdɛ/ vs vous aidez /vuzede/. Of course, what's really happening is feature spread from the stressed vowel, and it's unlikely the trigger can be lost without serious changes in the prosody of the language. Furthermore, it can affect polysyllabic monomorphemic words as well, so it's not really what you're looking for but I felt it was worth a mention Sep 10, 2017 at 19:18

3 Answers 3


As I understand it, the essential character which you're seeking is that it is regressive suffix-to-root assimilation (not progressive, and not bidirectional), and the trigger has to actually be there. This would leave out cases that are regressive root-to-prefix as well (various dialects of Arabic have such rounding "harmonies").

Various Romance languages ("Italian" and "Spanish") have suffix-triggered raisings, a.k.a. metaphony, which I believe are triggered only by suffixes. Jose Hualde has written on that (not that I can remember a specific paper). There is also regressive non-iterative harmony in North Saami which is triggered by a suffix and applies within the root, though only to epenthetic vowels connected with "Q3" (both rounding and fronting).

It seems that Bengali and Assamese have such a non-iterative suffix-triggered harmony, but more digging into word structure is necessary (one putative example might exemplify harmony within a suffix – or, a different analysis of the suffix).

For illustrative purposes, metaphony might be the best bet. There may be other examples, depending on whether you want to broaden the characterization. For example Icelandic u-umlaut is arguable iterative. Likewise Uyghur umlaut.

  • 1
    Examples of metaphony in the Italian dialects are well known, but are also quite similar to those of Umlaut: the trigger is never there anymore. What we observe is the diachronic outcome of a preceding metaphony. E.g. in the dialect of Naples a metaphonic vowel change characterises the gender of the adjectives: f.sg. rossə ~ m.sg. ruossə (from *rossa ~ *rossu). Notice that the endings are not distinguished anymore, in the present day language. Sep 8, 2017 at 17:34
  • 1
    Languages such as so-called "Italian" and "Spanish"
    – Draconis
    Sep 8, 2017 at 17:37
  • @user6726 I could have generalised too much. In some dialects the final vowel is effectively still preserved. Sep 8, 2017 at 17:42
  • Versions described by Kaze and Hualde don't apparently have that problem. I don't know, though, whether all instances of umlaut in these languages are phonetically transparent.
    – user6726
    Sep 8, 2017 at 17:43
  • I don't know if this adds / helps , Bengali and Assamese , tend to make inherent Hindi schwas to rounded ones Sep 29, 2017 at 14:16

In German, umlaut is admittedly no longer productive, but it is still very much in evidence in words like Mann > Männer, Kuh > Kühe, and many more.

  • 1
    But I am looking for a productive, regular and predictable system. Because, otherwise, we cannot even speak about a morphological phenomenon: we just have some words that show a diachronic contextual change... Nothing similar to vocalic harmony which is perfectly productive and regular. Sep 8, 2017 at 17:38
  • 2
    @ArtemijKeidan: I think umlaut in German is definitely a morphological phenomenon--it's just fairly unproductive. Singular and plural forms of nouns are morphologically related synchronically, not just diachronically. Sep 8, 2017 at 18:10
  • @sumelic I am not sure we can speak about morphology when there is no regular pattern observable. What we have is just a bunch of nouns with mutated vowel in the plural. The selection of the mutated plural is unpredictable. Now, make a comparison with Turkish plural marker -lar/-ler: all nouns select one or the other on the grounds of a well-defined pattern. Sep 8, 2017 at 18:16
  • 1
    @ArtemijKeidan. Turkish has lots of exceptions with regard to vowel harmony, mainly, but not only, in loan words. I doubt whether anything in language is"perfectly productive and regular".
    – fdb
    Sep 8, 2017 at 18:18
  • 2
    What about umlauting diminutives like -chen or -lein?
    – jlawler
    Sep 16, 2017 at 23:17

Maybe Standard High German counts in here:

In High German, the plural of Fuß "foot" is Füße with double marking: both umlaut and the ending -e /ə/ occur. Note that the dialectal basis of the so-called "Lutheran e" is rather restricted, it is absent in the North, the West, and the South of the German linguistic area and only occurs in the Central and Central-East area.

In the other dialects, the process is completed, e.g. Bavarian Fuß/Fieß or Ripuarian Fooß/Fööß show the same stage of development as English foot/feet.

There are more German words following this pattern, e.g. Kuh "cow", Zug "train", Trog "trough".

Another pattern is given by Fass "barrel" with plural Fässer (both umlaut and -er occur).

  • Is this different from my answer?
    – fdb
    Sep 29, 2017 at 14:08
  • 1
    @fdb Yes, because of the emphasis on double plural marking and reference to German dialects whre the process is already completed. (I will expand a little on this). Your answer emphasises the non-productivity of German Umlaut (that is debatable, see e.g. jlawlers comment) Sep 29, 2017 at 14:14
  • Umlaut and suffixing are part of the same parcel. The whole idea of Umlaut is that the vowel in the stem is assimilated to a front vowel in the following syllable.
    – fdb
    Sep 29, 2017 at 14:17
  • @fdb Yes, I agree. And later on economy leads to the weakening deletion of the suffix. BTW, the vowel in the following syllable does not need to be a front vowel, for Old Norse there are also a- and u-umlauts Sep 29, 2017 at 14:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.