How many consonants?

I was thinking about how many distinct types of consonants there are and came up with a back-of-the-envelope figure of 300. The IPA has about 130 different consonant symbols and then there are some other diacritics and length to factor in, so I guess about 300. Of course that’s assuming that the IPA symbols line up with the actual diversty of consonants.

Has anyone tried a more rigorous quantification?

10 thoughts on “How many consonants?

  1. Pavel Iosad

    No, 300 is far too few: after all !Xòõ alone manages to muster half that figure, and largely with the help of clicks, without utilizing the full range of possibilities in the pulmonic part.

    That must be a pretty trivial puzzle of simple combinatorics; I’d be surprised to learn no-one has done that yet.

  2. Michael Becker

    Suppose we found out that human languages draw from an inventory of exactly 792 consonants. How would one use this information? I suppose I could think of some practical uses… Did you have any theoretical point in mind?

  3. Mike Maxwell

    Doesn’t the answer depend on certain questions–like is consonant X in language A the same as consonant Y in language B? I suspect the answer to that may be unclear in some cases–like maybe the fortis vs. lenis consonants in Korean.

    Another issue, I would think, is whether [X] is a single consonant, or some sort of sequence. Decades ago, when I learned phonology in SIL, resolving this an any particular language was referred to as ‘resegmentation’, and typically involved decisions about affricates and geminates. Thus /ts/ in English is best analyzed as a consonant cluster, whereas in Tzeltal it (and the glottalized counterpart) functions as a single consonant. I haven’t seen this kind of issue addressed in generative phonology. But in the context of Pavel Iosad’s reply, there does seem to be some uncertainty about the number of consonants in !Xòõ, which revolves precisely around the question of whether certain consonants are better analyzed as clusters. There’s some discussion of this in the Wikipedia article for this language (http://en.wikipedia.org/wiki/!X%C3%B3%C3%B5_language), attributing a lower count to Ladefoged.

    As for practical uses, from time to time people have wanted to build a universal phone(me) recognizer, which could listen to a (good!) recording and come up with a transcription. There are obviously lots of feasibility issues, but one especially important one is the number of consonants and vowels that might have to be recognized (as well as other ‘things’ that need to be recognized, like tone, stress, etc.). Such counts would be particularly important because, I am told, the phonetic properties of a stop, for example, are quite different before different vowels, and also different before other consonants.

  4. Ed

    300? How did you arrive at this number?>

    Just a WAG.

    Suppose we found out that human languages draw from an inventory of exactly 792 consonants. How would one use this information? I suppose I could think of some practical uses… Did you have any theoretical point in mind?

    Mostly just idle thinking. But, this is exactly the kind of wow number that can be used to start off a discussion of linguistics for non-linguists. I find that most normal people just aren’t aware of the scope of diversity in natural language. That lack of awareness makes the kind of functional arguments for prescriptivism very attractive. You know, “well language is for communication and strict adherence to prescriptive rules enhance communication.” So there was some of that thinking in there.

Leave a Reply

Your email address will not be published.