Abstract

The semantics of belief reports has recently received a great deal of attention. Speech reports have largely been left behind in this discussion. Here I extend a familiar recent account of attitude reports, the Russellian theory, to the special case of speech reports. I then consider how it compares to Davidson’s paratactic theory with respect to a few examples that raise special problems about speech reports. Neither theory accounts for everything we want to say about these cases. I suggest that the problem lies in an assumption common to both theories, that in reporting what others say, we aim to represent what was said exactly as the original speaker represented it, in so far as this is possible.