Ong, J. H. & Burnham, D. .
The MARCS Institute, University of Western Sydney
Given that different musical scales divide the octave differently, listeners must learn what musical pitch categories exist in their native musical system. For example, the Western major scale divides the octave into seven intervals whereas the Javanese Slendro scale divides the octave into five intervals. It is unknown what mechanism underlies category formation for such scales, though it is hypothesised that musical pitch categories are acquired using the same mechanism as phonetic categories (Patel, 2008). This study examines whether musical pitch categories can be acquired based on the distributional structure of the input alone (distributional learning). Non-musicians were randomly assigned to one of two distributions of a continuum spanning a novel musical pitch minimal pair: Unimodal, which should induce formation of a single category, or Bimodal, which should induce formation of two categories. Their performance was assessed before and after training using a discrimination task. The results indicated that the Bimodal group, but not the Unimodal group, could reliably discriminate the musical pitch minimal pair and generalise discrimination to other similar but unfamiliar minimal pairs. This suggests that musical pitch categories can be acquired based on frequency information alone, similar to the manner in which phonetic categories are acquired.