How much of a functional load must a phone carry to be considered its own phoneme?

For example, my idiolect of English has a marginally distinctive glottal stop. However, it exists distinctively in only two words I can think of, both interjections:

  • /ʔju/ [ʔjɨw] eew, distinct from

    • /ju/ [jɨw] you/yew/ewe and

    • /hju/ [hj̊ɨw] hue/hew; and

  • /ˌɐˈʔɐ/ [ˌʔɐˈʔɐ] uh-uh, distinct from /ˌɐˈhɐ/ [ˌʔɐˈhɐ] uh-huh.

(This is independent of the common utterance-initial glottal stop insertion before vowels, and /ʔ/ as an allophone of /t/ in British English.) Is two minimal pairs enough to establish a phoneme?

  • The phonemic glottal stop is not necessary to maintain contrastiveness in either of these examples. The underlying syllables could just be vowel-initial in both cases, contrasting with h-initial or glide-initial. Sep 17, 2015 at 12:32

3 Answers 3


Interjections may not be very representative of the phonology of a language. For instance, "shhh" (being just a consonant) obviously doesn't obey the minimal word restrictions in English. Now, functional words like "they" and "that" can behave in weird ways too and often don't need to obey quite the same restrictions as lexical words. But functional words tend to have high token frequency, so their effect on a language's sound inventory is comparatively strong. In English, this means constant reinforcement of the distinction between /θ/ and /ð/, even while the voicing distinction is often lost in lexical words. Interjections, on the other hand, probably have low frequency.

So my guess would be that interjections don't often get used in minimal pairs, or when they do get used, it's with greater care and more caveats. I suppose you could look for other clues as to whether you have a phonemic glottal stop, such as cooking up a small perceptual experiment for yourself.

I don't know if you already checked out some of the literature on characterizing marginal contrastiveness. I've come across Goldsmith 1995 (his introduction to the Handbook of Phonological Theory) or Hall 2009 (her dissertation) for a more quantitative approach.


A 1996 paper published in Language by Ladefoged and Everett treats this question, dealing with sounds which are unusual in terms of their phonetics.


Even one minimal pair can be enough to establish a phoneme. But it might actually be an allophone of some other phoneme.

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.