Abstract The distinction between direct and indirect speech has long been known not to reflect the crosslinguistic diversity of speech reporting strategies. Yet prominent typological approaches remain firmly grounded in that traditional distinction and look to place language-specific strategies on a universal continuum , treating them as deviations from the “direct” and “indirect” ideals . We argue that despite their methodological attractiveness, continuum approaches do not provide a solid basis for crosslinguistic comparison. We aim to present an alternative by exploring the syntax of logophoric speech, which has been commonly treated in the literature as representative of “semi-direct” discourse. Based on data from two unrelated languages, Wan (Mande) and Ainu (isolate), we show that certain varieties of logophoric speech share a number of syntactic properties with direct speech, and none with indirect speech. Many of the properties of indirect speech that are traditionally described in terms of perspective follow from its syntactically subordinate status. Constructions involving direct and logophoric speech, on the other hand, belong to a separate, universal type of structure. Our findings suggest that the alleged direct/indirect continuum conflates two independent aspects of speech reporting: the syntactic configuration in which the report is integrated, and language-specific meaning of indexical elements.