An investigation of asymmetries of rhythm in speech production
Abstract
The human experience is filled with rhythmicity. From coordinated motor movements to memory recollection, humans use rhythmic patterns to accomplish a multitude of activities. In addition to internally regulated rhythmicity such as locomotion, humans are capable of synchronization to outside rhythmic stimuli as well. They use rhythm, both internal and external, in order to fine-tune common tasks of movement, perception, and memory. Rhythm is an integral part of language, manifesting as linguistic stress. Most of the world’s languages have a binary stress pattern, with an alternation of strong and weak beats. But there are a handful of languages that have been documented with a ternary stress pattern, where there is a lapse of two weak beats between strong beats. Linguists have tackled the question of how to theoretically represent ternary rhythm, but no one has answered the question why such a stark asymmetry between binary and ternary stress patterns exists in languages. This dissertation addresses that question by investigating how speech production is influenced by binary and ternary rhythmic patterns. Four production experiments used tongue twisters to look at error rate in binary and ternary meter. The first experiment presented 12 syllable long tongue twisters that subjects read as quickly as possible. Participants produced significantly more errors on ternary meter tongue twisters than on binary (p = 0.0003). In the second experiment, binary and ternary tongue twisters were presented aurally, with the subject repeating the stimulus for 5 seconds. Participants again produced significantly more errors in ternary meter than in binary meter (p = 0.021). In the third experiment, stress position was altered to first syllable, last syllable, or no stress. Ternary meter produced more errors than binary (p = 0.038). In the fourth experiment, binary and ternary rhythmic patterns were combined with binary and ternary syllable sequences to make matched or mismatched stimuli. In this instance, subjects produced fewer errors in ternary than in binary meter (p = 0.054). The results of three of the four experiments point to a binary rhythm preference in motor planning and control for speech production.