Based on the analysis of 267 tokens derived from editorial columns primarily drawn from two Persian newspapers, following on earlier studies by Chafe, Jahani, Lazard, Dahl, Adel, and Dafouze, and inspired by a series of Hyland’s studies on metadiscourse signals, this study has aimed at investigating evidential markers in these columns. In order to come to grips with the types of evidentials, first, we classify them into two major types – inferential and reportative; the reportative evidentials are further classified into (...) four types. The reportative classification is based, in the first place, on whether the source of information comes from an individual or from a government body, hence institutional. The further classifications are based on their identifiability/specificity. Results show that inferential evidentials comprise about 15% of all the tokens. Among the four reportative types, those whose source is individual and identified/specified and those that are institutional and unidentified/unspecified, coded as TYPE 1 and TYPE 4, respectively, have the highest frequency. The results overall show that Persian editorials in these two papers feature a high frequency of attribution of information to identified sources when the source is individual, but to unidentified sources when the source is institutional. The results also support other authors who claim that the imperfective aspect marker mi-, which is frequent in Persian, is a marker worthy of consideration in evidentiality. (shrink)