On the meaning of intonational contours: a view from scalar inference

Abstract

This paper investigates the meaning of intonational contours by experimentally testing how they affect the likelihood of scalar inference (SI) calculation. Our main test case is the rise-fall-rise contour (RFR) which, based on prior theoretical work, is predicted to either increase or decrease the likelihood of SI. We conducted two experiments using an inference task: one where participants first produce a target sentence with their choice of contour and one where participants listen to a pre-recorded target sentence with a particular contour. The experiments converged in showing that the RFR increases SI rate relative to a neutral fall. Additionally, production data revealed the frequent use of another contour that resembles the Contradiction Contour, which we label Concession Contour. This contour also led to an increase in SI rate, although to a lesser extent than the RFR. In addition to informing the theoretical literature on RFR, our results also highlight the methodological importance of controlling for intonation in the study of SI.