This paper offers what is to our knowledge the first description and analysis of the encoding of perceptual reports in Arabic: we focus here on Modern Standard Arabic (MSA) and Maltese (MT) building on the analytic framework offered by Asudeh and Toivonen 2012 (AT). We show that a range of different syntactic constructions are used to encode perceptual reports with seem-class predicates while the core semantic observation developed in AT, concerning the distinction between types of PSOURCE (perceptual source), is found to hold robustly across the varieties of Arabic we have investigated. In the light of the data we outline, an important question for future work turns out to be that of distinguishing cases of (genuine) copy-raising from constructions with thematic subjects (for the verbs in question). While Maltese is ideologically and sociolinguistically a separate language, it shares many key aspects of its syntax with the Western vernaculars of Arabic, and is fruitfully considered as a dialect of Arabic for the purposes of cross-dialectal comparison.