The phonetic basis for tonal melody mapping

Tonal melody mapping (for example, one tone per syllable, in left-to-right fashion) has been proposed as a common way of analyzing lexical tonal patterns for languages like Mende, Etung, and Kikuyu (Leben 1971, 1973, 1978; Edmondson and Bendor-Samuel 1966; Clements 1984). I argue that the mapping analyses of lexical tones in these cases are unwarranted and the surface tonal patterns resulted from such 'mapping' are in fact due to constraints on the distribution of contour tones. These constraints are phonetically grounded, and are based on the principles that contours are preferentially limited to syllables with longer duration. These include, but are not limited to, the final syllables in a prosodic domain (Oller 1973, Klatt 1975) and syllables in shorter words (Lehiste 1972, Lindblom and Rapp 1973).