The phoneme is identified as /m/, and the main allophone is a standard [m].

The places of articulation of [m] and [x] are very far away, and the manners of articulation are very different. They seem quite unrelated, but I come across a problem which [x] being an allophone of /m/ can solve.

Is there any linguistic pattern that is compatible with them being allophones? Is there any example in natural languages?


The specific context is Chinese and the pairing vowel is /o/. The syllable is [xo], which phonologically should be /mo/. Since the Middle Chinese onset /m/ has split into /m/ and /w/, is it possible that [xo] is derived from /wo/, with more friction?

  • Are you asking if it's common or if it's at all attested? The answer to the question in the title is a definitive "no"; the answer to the question in the body, I'm not sure about.
    – Draconis
    Commented Dec 7, 2022 at 5:19
  • There is no possible answer without some elementary statement of the reason for thinking that m has x as an allophone. Do you have any data to support the premise?
    – user6726
    Commented Dec 7, 2022 at 5:24
  • @Draconis Thanks for pointing it out. I mean if it's at all attested. I've changed the title accordingly.
    – lilysirius
    Commented Dec 7, 2022 at 6:13
  • I've adjusted the bracketing used in the question. Strictly speaking a phone cannot be an allophone of another phone, instead both would be allophones of a phoneme and that phoneme ought to be in slashes rather than square brackets. Alternatively square brackets could be kept if the question were rephrased as /m/ and /x/ being allophones of the same phoneme (with the identity of that phoneme left unspecified)
    – Tristan
    Commented Dec 7, 2022 at 9:42
  • 2
    Which form of Chinese is this? Mandarin has no [xo]; even counting /h/ as [x] (actual realisation varies), there’s only [xu̯o] and [xou̯]. Can you give the exact context? Commented Dec 8, 2022 at 10:02

2 Answers 2


The only thing I can think of that might remotely connect to this, is the reconstructed Old Chinese initial of 黑, which is /*m̥ˤək/ according to Baxter-Sagart. This led to alternations such as 黑/墨 ([x-], [m-]) which reflect the voiceless nasal in the initial. In the case of 墨, reconstructions generally posit that some other conditioning factor in the onset changes the correspondence.

I wish the original poster would have given more details (i.e. which characters the claimed alternation corresponds to, at the very least).

If this is the alternation in question, then this would be motivated not by direct regular correspondence of [x] and [m] but by other diachronic sound change rules conditioned by other parts of the environment.

  • 1
    Closer examples: 海 Middle Chinese xojX ← m̥ˤəʔ in Baxter-Sagart; 黑 xok ←m̥ˤək. Quite a number of pharyngealized ones, 況 xjwangH ← m̥aŋ-s as a non-pharyngealized example.
    – user6726
    Commented Dec 8, 2022 at 20:05

I'm not aware of any attestations of this.

It's possible for allophones to seem completely different from each other, such as (American) English [ɾ] and [ʔ]. But for these, we have very clear evidence of their allophony: they alternate in forms like "writer" versus "written".

If you have evidence of an alternation like this, then I'd say that's pretty conclusive evidence. Maybe they're allophones in the modern language, or maybe they developed along different paths from the same historical phoneme, and the alternation is a fossil.

If the only evidence is complementary distribution, though, I would not personally call them allophones. It's like "heng" in English: [h] only appears in the onset of a syllable in American English, and [ŋ] only in the coda, so you could theoretically call them both allophones of a hypothetical /ꜧ/. But in practice, they sound totally different, appear entirely unrelated, and never show any alternation between them (Arabic words with coda [h] aren't realized with [ŋ] by English-speakers for example).

So while you could consider them the same phoneme, there's usually* not much reason to; it doesn't make anything clearer or better elucidate the data. It's more common to just call them two separate phonemes with restricted distributions.

* Though you might take this into account when designing a writing system, for example; you could probably use the same glyph for both of them without introducing ambiguity, if you had a way to distinguish "hanger" from "hang her". Hangul does this, using the same jamo for a null onset and a coda [ŋ].

  • with the final comment about Hangul, is this quite analogous? I don't know about the development of Korean, but Mandarin null onsets largely derive from earlier ŋ onsets. It seems likely that, even if a similar development didn't occur in Korean, the evolution in Chinese might have affected the orthography
    – Tristan
    Commented Dec 9, 2022 at 11:38
  • @Tristan It's possible, but orthographies tend to be deliberately modified as the language changes (English notwithstanding)—my understanding is that there used to be separate jamo for /ŋ/ and /∅/, which were later merged because of the complementary distribution.
    – Draconis
    Commented Dec 9, 2022 at 16:43
  • I really don't think that trend holds. Hebrew, Tibetan, Thai, French, German etc have all remained fairly fixed with only minor spelling reforms (comparable to American English). If there used to be separate jamo that would rule it out though
    – Tristan
    Commented Dec 12, 2022 at 10:10

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.