Résumé:
The present dissertation aimed at investigating master two students’ of English problems in the
use metadiscourse markers in the general introductions of their dissertations and gauging the
potential causes of those problems. To achieve the first aim, a corpus analysis of fifteen master
dissertation randomly chosen on the basis of convenience have been identified, codified, and
analysed following Hyland’s (2005) model, which assigns metadiscourse features to two
metadiscourse functions ,interactive and interactional. To achieve the second aim, a teachers’
interview was designed and implemented to five supervisors randomly chosen among those who
supervised the dissertations in question. The corpus analysis was based on Hyland’s (2005)
model which assigns discourse markers to two major metafunctions: interactional and interactive
revealed significant results. As far as interactive metadiscourse markers are concerned, the
corpus analysis has revealed that master students have demonstrated an adequate use of
transitions, frame markers and endophoric markers, but failed to demonstrate the required level
with regard to two markers, namely evidentials and code glosses. As for the interactional
metadiscourse metafunction, the analysis of the corpus under study has revealed that master two
students have failed in demonstrating an adequate level with regard to all the defining features of
this metafunction,namelyhedges,boosters,attitudemarkers,engagementmarkers,and self-mention.
It is worth noting here that engagement markers have been totally absent in the data. As far as
the interview is concerned, the analysis of the results has shown that the adequate use of
discourse markers does not constitute a focal point of the feedback they offer to their supervisees
due, at least partly, to lack of time and students’ fragility at the level of grammar and writing.
The most interesting finding generated by the analysis of the interview is the supervisors
‘admittance that they are responsible for inhibiting supervisees from using hedges, boosters,
attitude markers and self-mentions. This finding suggests that the supervisors in question do not
adhere to a constructivist view of academic writing which totally rejects the principle suggesting
that discourse is but a mere linguistic representation of a set of impersonal and universal truths.