This dissertation aims to contribute to the field of discourse analysis by focusing on three discourse marker candidates, namely buoys, PALM-UP and the sign SAME, in French Belgian Sign Language (LSFB) and Catalan Sign Language (LSC). The first issue in the study of discourse markers is their identification, which has been based on three criteria: to be syntactically optional, to be non-truth-conditional and to constrain the inferential mechanisms of interpretation processes. PALM-UP and SAME may function as discourse markers, whereas from the category of buoys only list buoys have discourse-marking functions. The analysis of the discourse-marking tokens reveals intra and inter-linguistic differences at the level of their distribution per genre, per signer, their functions and their position in discourse. In the corpus analysed in this research, storytelling seems to be the genre that presents more differences in the use of discourse markers as compared to the other genres in the two sign languages. This finding underlines the need of using different types of productions in order to study how discourse is structured. There does not seem to be any sociolinguistic pattern that defines the use of discourse markers: it purely depends on the signer's preferences as it has been reported in the literature on spoken languages. PALM-UP is the most frequent and polysemous discourse marker, followed by SAME. From the three discourse markers under study, list buoys are the less frequent in spontaneous discourse and the less polysemous. The three discourse markers share most functions with their counterparts in the two SLs, but there are other functions that seem to be language-specific, particularly in the use of SAME. The position in discourse is established taking as reference the delimitation of turns and the adaptation of the Basic Discourse Units Segmentation Model (which combines an independent syntactic and prosodic segmentation) to the signed modality. Some discourse markers present a correlation between their function and their position that refines their description as in spoken languages. This study reveals that discourse markers have common properties in the two modalities (spoken and signed).