Rhythm, tempo, and F0 in language discrimination

Abstract

Existing literature on language rhythm suggests that languages can be discriminated in AAX “oddball” experiments only if they belong to different rhythm classes. Since exceptions to this pattern have been noted (e.g., the discrimination of Polish from English though both are said to be stress-timed), a series of five AAX experiments was run to test whether successful discrimination depends on factors other than rhythm class. English utterances were used as the base (AA) with Polish, Greek, Korean, Spanish, and Danish as the target languages. The stimuli were converted to sasasa while either retaining their original F0 or with flat F0; further, either the original tempo of each stimulus was retained or the stimuli (within each experiment) were manipulated to all have the same tempo (2 * 2 overall design, tempo * F0). The results show that tempo played the major role in discrimination with F0 cues being used when tempo differences between the two languages were eliminated. Rhythm class, on the other hand, played no significant part in discrimination suggesting that previous experiments may have interpreted as an effect of rhythm class the fact that tempo is often (though not always) slower in languages classified as stress-timed.