There are at least two notable differences between English and Japanese attitude attributions. A first difference is that Japanese has more complementizer options and different verbs select a different set of complementizers (Kuno, 1973). For example, the factive verb koukaisuru (‘regret’) takes the complement type koto, but it is incompatible with to, whereas the verb of saying iu (‘say’) is compatible with to, but not with koto. A second difference, which I focus on in this work, is that the Japanese knowledge and belief verbs require the aspect marker -te-i- to form attitude attributions, whose general semantic contribution is to create either a progressive or perfect statement. I discuss the role of -te-i- in knowledge attributions in Japanese together with a thesis in epistemology.

The thesis I discuss in this paper is epistemic contextualism, according to which the propositions expressed by the sentences of the form S knows that p vary in an epistemologically distinctive way depending on the contexts of use (DeRose, 1992; Lewis, 1996, among others). One major difficulty facing epistemic contextualism is that we seem to have no good linguistic reason to believe that the verb know is a context-sensitive expression (Stanley, 2004, 2005). I argue that a knowledge attribution in Japanese is a species of modal expressions, such as must, which are widely held to be context-sensitive. Thus, epistemic contextualism holds with respect to Japanese. The modal account of Japanese knowledge attributions is based on Nishiyama’s (2006) modal analysis of the aspect marker -te-i-, which in turn stems from Portner’s (1998) modal analysis of the progressive.