Are there phoneme sequences ("pairs") that have not been found in any natural language?

I imagine there are some number of sequences that are physically impossible, but also some that are physically possible for humans to produce but which, either due to chance or to some more interesting reason, have not appeared in any known natural language.

  • 2
    By "phoneme pairs" or "compound phonemes" do you mean consecutive phonemes, simultaneous phonemes, or something else? Dec 3, 2018 at 18:16
  • @MarkBeadles: I mean consecutive phonemes. E.g., /m/-/p/ is a common pair, but is there any language in which /p/-/m/ appears? (Granted, some pairs might not be physically possible; that would be of interest too.)
    – feetwet
    Dec 3, 2018 at 21:20
  • 2
    [pm] is possible (and found even in English, as in one pronunciation of happen), in the form of a nasal release.
    – Nardog
    Dec 3, 2018 at 22:21
  • 1
    The answer to your question must vary depending on whether you are actually asking in terms of phonemes and not phones. If a succession of the main allophones of two phonemes is "physically impossible", a language might employ an allophone instead of avoiding the combinaiton altogether.
    – Nardog
    Dec 3, 2018 at 22:26
  • 1
    English has your /p/-/m/ sequence across word boundaries eg 'top man'. But I suspect you're asking about occurrence within words, right? But even there I could offer English 'topmost'. So what I think you're really interested in is occurrences within a single syllable, is that correct? If so perhaps you could add that to the question? BTW Arrernte (Pama-Nyungan, Australia) has /pm/ syllable-initially. Dec 4, 2018 at 20:43

1 Answer 1


First, a word about what a phoneme is: a phoneme is "a sound" which a language uses as one of its primitive elements for constructing utterances. For instance /p/ in "paper, spit". It turns out that there are physical differences between the three kinds of p in those words, and those physical differences are excluded when coming up with the list of phonemes of a language. In other words, a phoneme is an abstraction from actual speech, omitting certain kinds of differences. Also note that this is about pronunciation and not spelling, so "rough, fish, phylum" all contain the phoneme /f/. Some examples of phonemes in English are /p, b, t/. Some physically-different sounds that exist in English which are not phonemes include [k kʰ kʲ kʲʰ], that is, these are physical variants of a single phoneme, /k/. (They are called "allophones"). Some examples of phonemes in Hindi are /p pʰ b bʰ/. English physically has p and , but there is a rule telling you when you use one versus the other, so the difference is not primitive. English simply does not have . A list of phonemes is always "in language X".

A "pair of phonemes" would then be "any two phonemes in the language". For example /f, s/ or /m, æ/. The notion of "pair of phonemes" is not a useful notion in language, since it is arbitrary. However, there are some kinds of "pairs of phonemes" that could be useful, for example the three pairs /d, n/, /b, m/, /g, ŋ/ exemplify English sounds which differ in a single, unified way (presence of absence of nasal airflow).

If by "pair of phonemes" you mean "any two-member subset of the set of phonemes in language X", then it follows that "all pairs of phonemes exist". Since /bʰ/ is not a phoneme in English, /p, bʰ/ is not a pair of phonemes in English – we fail on the "phonemes in language X" criterion. This definition of "pair of phonemes" is constructive – you can always construct all of the two-member subsets of any set containing more than 1 element. You may want to know that all languages have at least 2 phonemes.

So, "No, by definition".

OTOH, if the question is about sequences of phonemes in a specific order, then the trick is to find really low-frequency (of occurrence) sounds, and scanning the grammars of languages that have such sounds. Clicks are low frequency of occurrence. In no language with clicks do you ever find two consecutive clicks. That's probably a sample problem arising from the extreme rareness of clicks.

  • 6
    I think they're asking about pairs of consecutive phonemes, which don't always exist: English never has */hŋ/, for example, even though it has both /h/ and /ŋ/. In other words, this is a phonotactics question.
    – Draconis
    Dec 3, 2018 at 17:48
  • @Draconis is correct about the intent of the original question. This answer is helpful background, and provides one example: "no language [contains] two consecutive clicks."
    – feetwet
    Dec 3, 2018 at 21:38
  • There are no sequences of clicks because they only occur syllable initially. So I guess a way to identify non-occurring sequences is to look at phonotactics (as @Draconis said) to identify those phonemes restricted to onset or coda — they cannot participate in some sequences. I don't think it's necessarily about rarity, eg /ŋ/ and /h/ are not rare in English but we can't have the sequence /h/ /ŋ/ (but /ŋ/ /h/ is ok). Dec 6, 2018 at 22:31
  • Nothing is necessarily the case here: the odds are highest if you pick low-frequency phonemes then exploit sequence gaps in the languages that have them. Sequences with ŋ or h will not be in that list.
    – user6726
    Dec 6, 2018 at 23:49

Your Answer

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

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