Question: Is there a matrix or table showing how audibly similar the different sounds from the IPA are? I'm looking for a scalar value that somehow measures the oral distance between two phonemes.

Example: The following sounds are "close"

ɑ (father) ɛ (dress)

while the following sounds are "distant"

b (buy) n (can)

Nomenclature: I'm putting quotes around "close" and "distant" since I don't know the proper term to quantify the oral similarity between two phonemes. Please edit or correct any mistakes I've made.

Motivation: I'm actually looking for this similarity matrix with a reduced set of the IPA, namely those sounds coming from the ARPABET.

  • You probably want a metric based on a featural characterisation of the sounds. – Colin Fine Sep 23 '15 at 14:56
  • 2
    "Similar" is absolutely subjective, in fact any two objects in the Universe have somethings in common and are similar in something, e.g. b and n are also 'close' because both are voiced and both are consonants. You need to define what you mean by 'similar'. – Yellow Sky Sep 23 '15 at 14:56
  • @YellowSky I wish I could, but as I noted in my answer I lack the terminology. In a very ambiguous way, "similar" would be the average ranking provided by thousands of listeners. There will be a lot of variance, but there should be some features that are universal (I think?). Perhaps something mentioned by Colin, the "featural characterisation" would suffice. – Hooked Sep 23 '15 at 15:05
  • 2
    For consonants, the similarity is the hamming distance (# of bits different) of the feature matrix. For vowels, it's the distance in the vowel trapezium (plus some features like rounding). – Mitch Sep 23 '15 at 15:10
  • 3
    There are no speakers of IPA so the question is meaningless. One could ask specifically about confusability in English; or Spanish. There are multiple IPA transcriptions of English, so this will not tell you much about IPA. – user6726 Sep 24 '15 at 16:11

As described (empirical similarity from many listeners), I think you want confusability, e.g. a ‘confusion matrix'. Here’s an example, from Munson et al. (2003): https://doi.org/10.1121/1.1536630

Full reference:
Munson, B., Donaldson, G. S., Allen, S. L., Collison, E. A., & Nelson, D. A. (2003). Patterns of phoneme perception errors by listeners with cochlear implants as a function of overall speech perception ability. The Journal of the Acoustical Society of America, 113(2), 925-935.


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.