Abstract

Work by Sudo (2013) and Gyuris (2017) has shown that the bias of polar interrogatives varies both cross-linguistically as well as − in languages that possess more than one interrogative structure − across clause types. We take the “bias profileˮ of an individual polar interrogative type to be a non-empty choice from the power sets of evidential bias options,(℘({+ev,−ev,%ev})−{∅}), and epistemic bias options, (℘({+ep,−ep,%ep})−{∅}), for each of its expressive instantiations as positive polar question (PPQ), and negative polar questions with inside (IN-NPQ) and outside negation (ON-NPQ) in the sense of Ladd (1981). By simple extrapolation we predict the existence of (7×7)3 = 117649 such bias profiles. Our study explores the “spaceˮ of bias profiles in a way that eclectically mixes general principles used as top-down heuristics, principles related to the discussion of bias in the literature, and some bottom-up observation-based theorizing where our small sample of biasprofiles from English, Japanese, and Hungarian points in the direction of “interestingˮ (partial) generalizations. Throughout we calculate the “effectivenessˮ of our principles in the sense of stating the extent to which they reduce the space of bias profiles numerically. Our discussion then focuses on salient counterexamples to and plausible motivations for or against the most“effectiveˮ principles. Particular attention is paid to complementary choices of evidential and epistemic biases, which, encoded via the principle of Static Complementarity (together with Convexity) leads to a reduction of the space of bias profiles to just (4×2)3 = 512 permissible types.