In both his earlier and later writings, Frege claims that distinct sentences can express the same thought, and that there is a correspondence between the parts of a thought and the parts of the sentence expressing it. The joint assertion of these claims gives rise to a problem: how can there be a correspondence between the parts of one thought and the parts of distinct sentences? This paper discusses Michael Dummett's and Gregory Currie's interpretations of Frege's views on the analysis (...) of thoughts and proposes an alternative interpretation which answers the above problem and provides some insight into Frege's epistemology of thoughts and his theory of human communication. (shrink)