Letter to the editor : A note on classification experiments in acoustic phonetics

R.L.H.M. Smits, L.F.M. Bosch, ten

    Research output: Contribution to journalArticleAcademicpeer-review

    Abstract

    In recent years, several classification experiments have addressed the value of global spectral properties for the classification of stop consonants by humans or machines. The results of some of these studies seem to be at variance with each other. While Blumstein & Stevens (1979) and other investigators found invariance in global spectral properties, Suomi (1985) reports that the global spectral properties are highly dependent on the vowel context, and classification of stops based on these properties needs to operate in a context-dependent fashion. Through a theoretical exposition, as well as a comparison of past classification studies, it is shown that the results of classification experiments may strongly depend on the employed classification strategies. In particular it is shown that, when the study of Suomi (1985) is interpreted in accordance with some basic principles of the theory of classification, the general claim of the absence of invariance in the global spectral properties of stop consonants is not warranted on the basis of the reported results.
    Original languageEnglish
    Pages (from-to)477-485
    JournalJournal of Phonetics
    Volume23
    Issue number4
    DOIs
    Publication statusPublished - 1995

    Fingerprint

    Dive into the research topics of 'Letter to the editor : A note on classification experiments in acoustic phonetics'. Together they form a unique fingerprint.

    Cite this