Discreteness of phonology

[Harkiran B]

At the end of the Sound Categories video, there was a brief mention of the problem between the discreteness of phonology, and the fact that acoustics are continuous – I would think that the continuity of acoustics really becomes a problem when feature changes between sounds are continuous – so the hardest to discretize would be two consecutive vowels, then a consonant followed by a vowel, and then two consonants, etc. Would this be a correct understanding?

[Simon]

When you say “discretize” I think you perhaps mean “segment” – dividing the continuous speech stream into a sequence of units.

You are correct in thinking that a sequence of vowels is hard to segment (think about diphthongs) and that some consonants are relatively easy. This has implications for concatenative speech synthesis, in which we segment and then re-sequence recorded speech.

The concern of phonology is to place speech units into a finite set of discrete categories that can distinguish the words of a language (think about minimal pairs).

[Author]

Posts