For those who came in late, vowel hiatus is a common term for the occurrence of consecutive vowel sounds each of which serves as the nucleus of a syllable. For example, in the word “chaotic” we see consecutive syllabic [eɪ̯ɑ̩]. In the word “reality” we see consecutive syllabic [ia̩]. In the word “aortic” we see consecutive syllabic [eɔ̩ɚ̯].

Many languages limit or forbid vowel hiatus. Strategies for this are called “vowel hiatus resolution.” Among them are …

  • inserting a consonant between two consecutive vowel sounds (created by compounding or affixation, for example)

  • turning one of the two onsecutive vowel sounds into a glide and so creating a single syllable diphthong

  • turning the two vowels into one vowel that shares the features of both

Vowel hiatus resolution is discussed with considerably more detail and rigor at this link: https://www.ling.upenn.edu/~gene/courses/530/readings/Casali2011.pdf

Be that as it may, I have yet to find a description of the kinds of rules that prevent the formation of words comprising only four or more consecutive vowels. From what little I’ve heard, such words have never been attested in any natural language.

I’m trying to describe a conlang that allows the occurrence of two consecutive syllabic vowels, but not more. In natural language, and in descriptions of them, what kinds of rules constrain the number of consecutive syllabic vowel sounds, what are these rules called, and how are they written or described?

  • 2
    How about Ancient Greek Αἰαία, "(the island of) Aiaia"? I suppose that one will come down to how you analyze diphthongs. A bit of googling brings up a mountain named "Mauao" in Māori: four vowels, all in hiatus.
    – Draconis
    Feb 26 at 3:44
  • 2
    The point being, I think your premise (that languages necessarily have constraints against four vowels in a row) is false. Do you consider that a valid answer to the question?
    – Draconis
    Feb 26 at 3:46
  • 2
    Allowing for compounds, quite a lot of languages have no trouble with four or more consecutive vowels. The prime example (albeit constructed and semantically rather, erm, odd) is Finnish hääyöaieuutiset ‘news about wedding night intentions’, phonetically [ˈhæːyø̞ɑie̞uːt̪is̠e̞t̪]. Danish has quite long sequences of phonetic vowels as well, though underlyingly there are arguably consonants involved, e.g., bjergere ‘rescuers’, morphophonemically |bjɛrvərə|, phonetically [ˈp͉ɪ̯æɐ̯ʊ.ɒ̽.ɒ̽]. Feb 26 at 11:32
  • @Draconis: Yes, I do consider that a valid answer to my question. Obviously, I was wrong about the existence of the constraints I mentioned in my question. This new (for me) information will enable me to describe my conlang better. I can simply write that sequences of more than two consecutive syllabic vowels are "unattested" in my conlang--with tongue in cheek of course. Feb 26 at 13:31
  • Do you allow acronyms pronounced with hiatus?
    – Anixx
    Mar 3 at 22:15

2 Answers 2


From what little I’ve heard, [words comprising only four or more consecutive vowels] have never been attested in any natural language.

Such words certainly exist somewhere in some languages, although no language has such a form for the majority of its words. In Tahitian, there is a word oaoa meaning "narrow".

There doesn't necessarily have to be a rule in order for a certain form to be rare for words. It's possible that regular sound changes only rarely lead to a situation where a word is left with four or more vowel sounds and no consonant sounds.

In Latin, sequences of two vowels in hiatus are fairly common, sequences of three or more rare, but there is no absolute prohibition on having more than two vowels in hiatus: thus, Gāius has such a sequence (/gaː.i.us/).


There is no principled limit on the number of consecutive vowels possible in human language, see for example Kamba [nénéké.é.é.éetɛ̂] 'I have cried for it'; Hyman 1982 has a similar 6-vowel sequence in Gokana.

It is not unusual to find languages that allow two consecutive non-identical vowels – the language allows diphthongs. While the general rule is that any vowel defines its own syllable, languages may "interrupt" that parsing by organizing certain vowel sequences into a diphthong, thus you can get [goahti, giehta] in N. Saami, but not *[goehti] etc – the allowed vowel sequences are limited. Diphthongs are one source of vowel sequences, and are generally limited to two. Of course, there is also a labeling problem in distinguishing diphthongs from vowel-glide sequences.

Triphthongs exist, and are even more restricted. This is actually related to the onset-limit generalization noted decades ago by Clements, that onset and coda restrictions can be reduced to overlapping two-member filters, so that you can have [sp and [pr, therefore you can have [spr, but not *[sbr. Given that tautosyllabic vowel sequences are subject to overlapping two-vowel limits (for example [oa is okay and [ai is, therefore [oai might be, but *[aoi isn't), you intrinsically predict few triphthongs and virtually no quadriphthongs.

Furthermore, limits on the weight of syllables work against syllables with 4 vowels; well-justified analyses of syllables as having three moras are pretty close to non-existent. So all told, there is a predicted limit on the number of vowels within a syllable. However, there is no predicted limit on the number of consecutive syllables composed of just a vowel. Which is how Kamba and Gokana become relevant.

  • For analyses of syllables having three morae: Akkadian has a regular distinction between light (short vowel, no coda), heavy (long vowel or coda), and ultraheavy (long vowel and coda) syllables. For example, the stress can only fall on the ultima if it's ultraheavy. (Some dialects also have ultralong vowels resulting from contraction that can make a syllable ultraheavy on their own.)
    – Draconis
    Feb 26 at 17:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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