Can stress-timing and syllable-timing be perceived?

Abstract

The listeners’ ability to classify languages as stress- or syllable-timed was indirectly tested in two experiments. In the first, American, Greek, and Korean listeners rated low-pass filtered utterances of English, German, Greek, Italian, Korean, and Spanish for similarity to a series of non-speech trochees, using a 7-point scale. In a second experiment (with English listeners), low-pass filtering was replaced by “sasasa” (in which consonantal intervals are replaced by [s] and vocalic intervals by [a]); stimuli either retained their original F0 or had flat F0. Stress-timed English and German were expected to be rated more similar to the trochee series, since their rhythm is based on the alternation of stressed and unstressed syllables and is thus akin to trochees; low ratings were expected for syllable-timed Spanish and Italian, and intermediate ratings for (unclassified) Korean and Greek. Ratings did not support rhythm class distinctions: English was rated least trochee-like by all listener groups, and the other five languages were rated similarly to each other. Further, intonation played a more important role in classification than previously expected. Overall, the results cast doubt on the impressionistic basis of rhythm classes and more generally on the idea that speech rhythm can be equated with timing.