Versions of this paper-not read from the present manuscript-were given from 1971 onward to colloquia at New York University, M.I.T., the University of California (Los Angeles), and elsewhere. The present version was written on the basis of a transcript of the M.I.T. version prepared by the editors of this volume. Donnellan himself heard the talk at U.C.L.A., and he has a forthcoming paper, “Speaker Reference, Descriptions and Anaphora,” that to a large extent appears to be a comment on considerations of the type mentioned here. (He does not, however, specifically refer to the present paper.) I decided not to alter the paper I gave in talks to take Donnellan's later views into account: largely I think the earlier version stands on its own, and the issues Donnellan raises in the later paper can be discussed elsewhere. Something should be said here, however, about the pronominalization phenomena mentioned on p. 270 below. In his forthcoming paper, Donnellan seems to think that these phenomena are incompatible with the suggestion that speaker's reference is a pragmatic notion. On the contrary, at the end of the present paper (and of the talk Donnellan heard), I emphasize these very phenomena and argue that they support this suggestion. See also footnote 31 below.