Rhythmic classification of languages based on voice timing

Speech rhythm classes can be distinguished acoustically and perceptually from the variability of consonantal durations and the relative durations of vocalic intervals. The present research investigated whether this distinction can be made robustly, simply on the basis of voice timing, by measuring the durational characteristics of voiced and voiceless intervals in fluent speech. We show that voice patterns — in terms of vocal fold vibration — provide an effective basis for classification and that they can be automatically processed for large datasets. The possible implications that this finding can have on the ability of infants to distinguish between languages of different rhythmic classes are discussed.