In this paper we study one type of Spanish que (the equivalent to the complementizer "that") that can be shown to belong into the crosslinguistically restricted list of evidentials. In particular, we will claim that it encodes the (most basic) marks of non- first-hand or indirect (reported) evidence. Our point of departure is certain (apparently) independent clauses of Spanish headed by an overt complementizer (que). Some tests will be presented that support the idea that that one type of que introducing a well specified subset of root sentences shares most of the properties that have been claimed to characterize reportative evidentials in languages such as Quechua (Faller 2002, 2006). As for the properties of reportative que, it will be further shown that it does not encode any features related to epistemic modality (reliability or (im)probability) and we will propose that it is better analyzed as an illocutionary operator, affecting the illocutionary force (in line with Faller 2002 among others) and not as an epistemic modal (Izvorsky 1997 among others). In order to determine the nature of this reportative element and its origin, we contrast it with an old Spanish form, dizque, which exists nowadays in certain modern American varieties. This particle also has the properties of an evidential but behaves as an epistemic modal. In the last sections, we will propose that both evidential particles (que and dizque) are the result of a process of grammaticalization (i.e. 'upward reanalysis', or categorial change, of functional material, in the sense of Roberts and Roussou 2003) of the complex structure headed by a communication verb, dicen que "they say that". We will tentatively describe such process and introduce a hypothesis as to the nature and role of the parameter involved in the claimed reanalysis. [ABSTRACT FROM AUTHOR]