Definitions of reported speech are neither clear nor unambiguous, even in the case of the most classic forms, i.e. direct and indirect speech forms which have been recognized for a long time by the vulgate. In this contribution, we aim to show that these definifitions come to the fore in three different approaches: the first one would be associated with Charles Bally’s enunciative studies (1912, 1914) and Gérard Genette’s narratological study (1972), the second one appears in the approaches of “enunciative heterogeneity” by Jacqueline Authier-Revuz (1992, 1993), and the third one is presented in the works of Jacques Bres and Bertrand Verine on the dialogical principles in speech (2002). Our hypothesis is that ambiguities linked with the different theories of reported speech come from a heterogeneous (non stabilized) description of the status of the attributive speech in reported speech (also often called quoting speech or introductory syntagm) and may be the cause of divergences in the analyses of speech facts.