In this talk, I compare two types of epistemic verbs with an infinitival complement that appear to be quite similar at the surface level in English and Japanese, and attempt to elucidate some important similarities and differences between the two languages, taking into account recent developments of principles-and-parameters/minimalist approach in generative grammar.

English has a well-known syntactic contrast between "Raising" (seem, appear, etc.) and "Exceptional Case Marking (ECM)" (believe, consider, etc) constructions, though the verbs involved in them can be classified together as "epistemic verbs" (verbs of thinking, feeling, perceiving, etc.) in semantic terms.

(1)	a. Raising: Mary seems to me [t to be smart]
	b. ECM: I believe [Mary to be smart]

There have been various analyses proposed for each of these constructions in generative grammar , but one simple way of characterizing the contrast in (1) is to say that ECM verbs like "believe" are the transitive counterparts of Raising verbs like seem, or conversely, the latter are the unaccusative counterparts of the former.

The Japanese also has a similar contrast in omo-e-ru 'seem' vs. omo(w)-u 'think':

(2)	a. Hanako-ga Taroo-ni kawai-ku omo-e-ru 'Hanako seems to Taro to be adorable'
	b. Taroo-ga Hanako-o kawai-ku omo(w)-u 'Taro thinks (of) Hanako to be adorable'

The first question I would like to address in this talk is: Is it descriptively adequate to capture the contrast in (2) in the same way as that in (1). I show that the answer is negative, despite of their apparent similarities. Specifically, I demonstrate that (2a) differs from (1a) in that the matrix subject position is occupied by the dative phrase and the embedded subject receives nominative Case in situ; in other words, the word order in (1a) is derived by scrambling the nominative phrase out of the embedded subject position to the sentence-initial position, rather than by raising it to the matrix subject position.

Given this descriptive observation, a question naturally arises as to what makes this difference between the two language. I attempt to provide an answer to the question by focusing on two basic lexical differences between them; one is the categorial status of the to-phrase in English and ni-phrase in Japanese, and the other, the presence/absence of an Agr(eement) element in I(nfl).

If I have time, I also include in my talk some discussion on the syntactic function of the Japanese "spontaneous" ("zihatsu") morpheme ("e" in "omo-e-ru") with reference to voice alternation in general.