A Semantic Study of Citations in Literature Reviews of Information Systems Research Articles Through a Move- and Paradigm-Specific Approach


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date2 Aug 2021


Literature review (LR) is an integral part of research writing. Its significance lies in its dual role in contextualizing one’s own research in a related strand of scholarship while at the same time developing a theoretical framework or a methodological focus for the current study and identifying important omissions or gaps in the knowledge base that it seeks to address. However, writing an LR is a very demanding task for novice writers (Belcher, 1994; Feak & Swales, 2009; Flowerdew, 2000) as it requires them to make many decisions on the appropriate sorts of knowledge to cite and the right amounts to include. Those who misunderstand the purposes of citing even often over-cite and cite indiscriminately, resulting in what Belcher (1994) describes as bibliomaniac citations and what Ridley (2012) refers to as a lack of authorial control in their own writing. This situation clearly suggests that there is a strong need to examine citations in texts produced by experienced writers such as those that have been published so as to generate some useful learning models for novice writers to follow. 

Unfortunately, there has been very little work done in that regard. Prior citation studies have seldom taken a section-specific approach to investigate what writers cite or how their ideas are cited in specific parts of RAs by relating them to the rhetorical purposes of the parts of the texts. Therefore, the study presented in the current thesis is an attempt to fill the research void by exploring what sorts of knowledge (semantic matters) tended to be cited in LRs of RAs of three different research paradigms in the IS discipline (N = 90), that is, the positivist, pragmatist, and interpretivist paradigms, which are also known as behavioral science research ‘BSR’ (N = 30), design science research ‘DSR’ (N = 30), and interpretive research ‘IR’ (N = 30).

Assuming that the types of ideas that writers tended to cite were shaped and constrained by the move-steps of their writing and the research paradigms of their studies, the current study adopted a move- and paradigm-based approach to analyze citation content in the LRs of the IS RAs. One part of the study was a move analysis conducted through a move-based approach grounded in Swales’ (1990) well-attested CARS model to identify source ideas that occurred in specific moves and their respective steps of each of the three LR corpora (i.e., BSR, DSR, & IR). Following that a citation analysis was performed using a paradigm-based approach to investigate the types of ideas cited in the different move-steps of the LRs from the three subfields of the IS discipline (i.e., BSR, DSR, and IR). To facilitate this investigation, Kwan and Chan’s (2014) semantic typology of citations was drawn upon. The typology assumes that in research writing three types of epistemic entities tend to be cited, namely theoretical, methodological and research-related matters. Using this typology, the types of ideas that the writers tended to cite in the different move-steps of their LRs were classified and compared to identify possible paradigmatic variations between the three paradigm-specific LR corpora.

Findings showed that while all three types of cited ideas proposed in Kwan and Chan’s typology are present in the LR texts, there exists a notable portion of cited ideas that involve non-epistemically-framed matters describing primarily real-world phenomena. It was also found that the four types of citations are distributed unevenly across the LR texts and that the writers cite specific ideas in each of the categories. For example, when citing theoretical literature, the writers tend to refer to theoretical terminologies, definitions, and propositions in their LRs. When citing research literature, they also tend to make reference to research aims/focuses, acts, and findings. 

Findings also demonstrated significant paradigmatic variations in some of the cited ideas between the three LR corpora under study. For example when citing theoretical matters, the BSR LRs tend to refer to generic or proper names of constructs of behavioral or cognitive phenomena in comparison with the DSR LRs where those of design concepts and design artifacts are often cited and the IR LRs where those of sensitizing concepts of social phenomena are often cited. Concerning research citations, it was found that research aims/focuses tend to be cited in the BSR LRs that indicate causal or associative relationships to constructs of behavioral or cognitive phenomena, which are absent in the research citations found in the DSR and IR LRs. With regard to methodological citations, it was observed that only the DSR LRs cited methodological matters and these matters tend to make reference to design-oriented methodologies for artifact construction, as opposed to empirical methodologies as alluded to in Kwan and Chan’s (cited above) semantic typology of citations adopted in the present study. All these differences can be attributed primarily to the distinct nature of each of the three research paradigms on which the LR texts being produced are based. 

In short, the current study showed that the move-steps of a text and the research paradigm of a study has a significant bearing on the types of ideas that writers tend to cite in their own writing. Such paradigm-dependent and move-step-specific use of citations reveals one tacit nature of citations that has significant implications for citation instruction, which Linell (1998a, 1998b) describes as re-contextualization. That is, citations in LRs involve re-contextualizing source ideas in such a way that masks the original contexts of writing. Although this phenomenon is common in the LR texts examined in the present study, it has rarely been discussed in existing writing manuals. Based on the above findings, this thesis draws pedagogical implications concerning how to teach novice writers to use and integrate citations for reconstruction of prior knowledge in LRs.