Different processing mechanism of garden-path ambiguity in music and language : an empirical approach.

Abstract

Ambiguity is a ubiquitous phenomenon in both music and language. In the past several decades, scholars have acquired much understanding of it by examining its subtype called “garden-path ambiguity” in language. This type of ambiguity brings about several interesting questions in music: How can we define the garden-path ambiguity in music? Can we observe the garden-path effect in music? Does the human parsing mechanism (or the human parser) retain all the plausible interpretations in its initial processing (parallel model), or does it keep only the best-so-far interpretation at a time (serial model) when coping with such ambiguity? Though roughly touched upon by Meyer (1956), Bernstein (1976), Jackendoff (1991) and Temperley (2001), these questions have never been systematically theorized and empirically investigated. This paper first proposes a theoretical framework for what constitutes the garden-path ambiguity in music. Then it reports two experiments addressing the question about the parsing mechanism. In Experiment 1, a type of garden-path musical stimuli called “chimeric melodies” were presented auditorily to the participants. Results indicate that the processing of such garden-path stimuli does not cost significantly more time relative to the unambiguous controls. In Experiment 2, another similar type of chimeric melodies were presented but this time only visually in scores. In this experiment, the eye-fixation duration at the disambiguation region was significantly longer than other parts of the stimuli. Taken together, while the garden-path effect is verified in music, the processing model is dependent on the presentation modality. When presented auditorily (Experiment 1), the processing model of the garden-path ambiguity in music is best explained by a parallel model, whereas when presented visually (Experiment 2), it is best explained by a serial model. Further neuro-scientific research such as EEG study is suggested for more modality-neutral comparison and more real-time information.