Since the 1980s, there have been a substantial number of models of the rhetorical structure of the various sections of the research article (RA) genre in English, which have also been used to compare it with that of comparable sections of the RA in other languages with applied purposes in mind. Basic text units in these academic discourse models are the rhetorical move and step. These are considered as socio-psychological constructs with specific functional-semantic values which are commonly recognized by members of the research community as typical of each section in the RA genre. However, the process of annotation giving rise to many of these move analysis models has often been subjected to criticism mainly because of the fact that it has been applied by a single researcher, raising questions about its reliability. Although some studies have attempted to overcome these shortcomings by using independent coders, very few offer the percentage agreement figure and even fewer, the Kappa coefficient, as measures of inter-coder reliability. Also, very few studies offer clear definitions of moves and steps that allow other researchers to easily replicate their studies. In the present paper, we have used insights from the methodological principles used in the qualitative analysis literature to offer clear definitions of moves and steps and have analyzed and annotated the rhetorical structure of a pilot sample of 15 pairs of RAS from the EXEMPRAES (Exemplary Empirical Research Articles in English and Spanish) corpora representing a wide range of disciplinary areas. Our paper discusses the usefulness of such calculations as the average Cohen’s Kappa coefficient and the average agreement percentage in the process of revising the coding scheme itself, as well as demonstrates the reliability of the coding done by at least two independent coders in the standard final test.
All rights reserved