A (contingent) content–parthood analysis of indirect speech reports

This article presents a semantic analysis of indirect speech reports. The analysis aims to explain a combination of two phenomena. First, there are true utterances of sentences of the form α said that φ which are used to report an utterance u of a sentence wherein φ's content is not u's content. This implies that in uttering a single sentence, one can say several things. Second, when the complements of these reports (and indeed, these reports themselves) are placed in conjunctions, the conjunctions are typically infelicitous. I argue that this combination of phenomena can be explained if speech reports report (perhaps contingent) parts of the contents of the sentences reported