Let's say some hypothetical language had the phones [g],[k],[ʔ],[h]

and we determined /g/ to be the UR of the following allophones:

/g/: [g],[k],[ʔ]

But upon further examination, you may be able to argue for the glottal stop being the UR of another subset of allophones:

/ʔ/: [ʔ], [h]

So based of this definition, the segment ʔ is functioning both as an allophone of the phoneme /g/, and as its own phoneme /ʔ/.

Is this possible?

  • 1
    The question does not make much sense to me. Using ʔ for both a phone and a phoneme does not change the fact that they are two different things. So are you really asking if a phone can be an allophone of several phonemes? – dainichi Dec 16 '13 at 4:59
  • @dainichi That's pretty much it. But adding onto your last sentence: I was asking if some phone P "can be an allophone of several phonemes", including a phoneme that is the same as phone P. I think this is best illustrated by the plural /z/ ending in /kætz/->[kæts] and /dagz/->[dagz]. This shows that [s] is an allophone of /z/, even though [s] is also an allophone of /s/. What confused me is that if there exists a phoneme /s/ you could infer that the segment 's' provides meaningful contrast, and therefore would not make much sense to be an allophone of another phoneme. – RECURSIVE FARTS Dec 16 '13 at 6:07
  • "a phoneme that is the same as phone P". A phoneme is not a phone, you're comparing apples to oranges. "the segment 's'" What do you mean by segment here? A phone? A phoneme? A morpheme? – dainichi Dec 18 '13 at 7:31

Yes, this has been suggested decades ago. The representatives of the so called Moscow phonological school (in the Soviet Union, greatly influenced by Nikolai Trubetzkoy) argued that in such cases we deal with archiphonemes, /X/. One of the most famous examples is below:

(1) kosa (braid.NOM.SG.) - kos (GEN.PL)

(2) koza (goat.NOM.SG.)- koz (GEN.PL.), the latter pronounced as 'kos'.

In Russian, /s/ and /z/ are considered phonemes- there are a lot of minimal pairs, but in this example the opposition is neutralized. It's not quite clear, based on phonology only, whether we have an allophone [s] of the phoneme /s/ or /z/ in (2).

| improve this answer | |
  • Yes maybe that's what I'm looking for! This seems akin to the English plural ending X -> Xz / [+plural], where you may get alternations for words like [kaets] and [dagz] (lazy IPA). Am I correct in assuming so? – RECURSIVE FARTS Dec 15 '13 at 7:46

This would be tantamount to an archi-phoneme.


  • /g/: [g], [ʔ]
  • /k/ [k], [ʔ]

    and that [ʔ] in a given position, eg word final, corresponds to both [g] and [k] elsewhere then for that position you have an archi-phoneme that includes /g/ and /k/.

    ' Please note that it's the phone [ʔ] that is the allophone of two phonemes. A phoneme is not a sound but a class of sounds.
| improve this answer | |

If ʔ is an allophone of g, I don't see how you could argue that ʔ is also its own phoneme.

Wouldn't the model look like this instead?:

/g/ : [g] , [k] , [ʔ] , [h]

| improve this answer | |
  • Likewise you could argue: /ʔ/: [g] , [k], [ʔ], [h] But this is an arbitrary data set from some made up language I just used as an example. I frustratingly can't think of any examples in languages that I'm fluent in, however I've encountered this phoneme/allophone-sharing dilemma when working on actual data sets. – RECURSIVE FARTS Dec 15 '13 at 0:24
  • I don't think there's a problem with same phone being an allophone of two distinct phonemes, for one of which it is also the primary allophone. That's what the question is about, right @RECURSIVEFARTS? – Gaston Ümlaut Dec 15 '13 at 0:55
  • It's close. I'm wondering about certain allophones of a certain phoneme functioning as phonemes themselves. If you look at my example, the glottal stop is both an allophone of /g/ and its own phoneme. – RECURSIVE FARTS Dec 15 '13 at 1:05
  • I'm not sure if this is an example of what you're looking for, but perhaps something similar's going on in english when it's whispered? In sentence A: He's my big. The /b/ in "big" is pronounced as a [p]. I guess you could consider [p] to be both an allophone of /b/ as well as its own phoneme? Edit: I give up trying to edit this comment. – zheshishei Dec 15 '13 at 2:12

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.