The aim of this article is to propose a taxonomy of speakers from a socio-pragmatic perspective by taking an original approach to the study of single-turn political discourse, that is, political speeches, rather than debates, interviews or press conferences. This limitation on the scope of the study stems from the fact that the categorisation advanced is not concerned with turn-taking, but concentrates on the speaker’s use of other voices in his/her representation of reality. Thus, a clear distinction is made between (...) the speaker and the sayer, namely the original speaker whose words are reported, or rather creatively reconstructed, by the current speaker ‘here and now’. The taxomony comprises a number of categories employed strategically by the current speaker in the service of different objectives. The speaker categories are inherently limited due to a single principle of categorisation: the form of the report. The singular speaker is the default type which enables the speaker to present himself/herself as an individual as well as his/her attitudes, values and beliefs, while the collective speaker enhances the speaker’s belonging to a group and allows him/her to speak on behalf of its members. The sayer categories, by contrast, are more varied due to three independent principles of categorisation: the form of the report, the genuineness of the sayer’s utterance and the number of accounts embedded in the narrating event. The approach taken to the analysis of the narrative passages that involve the proposed speaker types owes much to Chilton’s Discourse Space Theory and is concerned with conceptualisation of the speaker’s representation of events from chosen perspective. The corpus of speeches selected to investigate and illustrate individual categories consists of over 80 political speeches delivered by three Democratic American Presidents: John Kennedy, William Clinton and Barack Obama. (shrink)