0

I am looking for similarity metrics based on IPA symbols in English. In other words, given two phonemes A and B (given in IPA format), I want to know how similar they are based on some metric, M. For example, M(ɒ, oʊ) would yield a higher score than M(æ,ɔɪ).

postnubilaphoebus
  • 101
  • 1
    Use the IPA charts themselves (with different dimensions added for features like roundness and ejectiveness), then just use the taxicab metric. Though IPA charts, strictly speaking, don't have phonemes on them. You'll hafta use different charts for each language anyway. – John Lawler Dec 03 '22 at 18:39
  • There isn't a unique similarity metric. People from California will tell you that /ɑ/ and /ɔ/ sound nearly exactly alike, because they have the cot-caught merger. People from the U.K. will say that they sound quite different. – Peter Shor Dec 03 '22 at 19:43
  • @JohnLawler Where can I find a list of features like roundness and ejectiveness? Sorry, not a linguist. – postnubilaphoebus Dec 03 '22 at 20:12
  • @PeterShor Then I will perhaps have to use spoken language and use a frequency domain representation of phonemes by the same speaker with a standard accent or something. – postnubilaphoebus Dec 03 '22 at 20:13
  • Most of the extra dimensions are only one or two items deep, and if you're using phonetic or phonemic symbols, all numbers are integers. Any raised or sub- letter is an additional dimension: [kʰ] is 1 up in the "aspirated" dimension; [kʷ] in the "rounded" dimension, [kʸ] in the "palatalized" dimension. You don't hafta be a linguist to use the chart; phonetics is a natural science. – John Lawler Dec 03 '22 at 21:48
  • [linguistics.se] might be a better place to ask this. – Barmar Dec 09 '22 at 21:23

0 Answers0