Abstract
In this paper, I consider the slope of two successive tones and I argue that tone sandhi should be factored in “slope correspondence” (i.e. preservation of the steepness relation between two successive tones). Empirical evidence comes from the asymmetrical behaviors of the two rising tones in Hangzhou Chinese, in which in sandhi initial position, the sharp rising tone 14 is retained whereas the smooth rising tone 24 is completely flattened (i.e. “Preservation of the Marked”). I shall show that previously proposed tonal faithfulness and markedness constraints in the OT literature both fail to adequately explain the phenomenon in question, while the slope correspondence constraints provides a unied account for the Hangzhou Chinese data.