摘要
Based on firsthand acoustic data, this paper aims to determine how many phonologically contrastive falling tones exist in tonal languages, and what kinds of distinctive features are needed to specify them. These goals are achieved by using a tonal model called the Multi-Register and Four-Level Model, which represents tones along four parameters: register, length, height, and contour. Having excluded a quasi-falling tone, this paper identifies seven Falling Tonotypes in the M Register: High, Low, MS-High, MS-Low, Deferred-High, Deferred-Low, and Slight Falling. Four of these also occur in the L Register with special voice qualities. In total, there are eleven Falling Tonotypes, which can be specified according to five distinctive features.