[ Pobierz całość w formacie PDF ]
.2.3 Triangulation: An overused term?Triangulation as a central methodological concept comes high on the list ofkey features of good research designs (Cohen and Manion, 1994: 233).Theway the term is conceptualized by scholars is however epistemologicallyvaried.Denzin s (1970: 472) early work indicated that there is more than onetype of triangulation:Data triangulation (the application of more than one sampling method for datacollection)Investigator triangulation (the involvement of more than one researcher)Theoretical triangulation (the use of more than one theoretical stance)Methodological triangulation (the use of more than one methodology).Data triangulation and methodological triangulation are arguably the mostcommon operationalizations of the term; the former refers to data gatheringmethods, while the latter is broader and refers to the use of more that onemethodology in a research design.Denzin also drew an interesting distinctionbetween inter-method and intra-method triangulation the former referringto the use of facets of the same method and the latter referring to the use oftwo (often contrasting) methods (see Schryer, 1993, for an example).Triangulation is often one of the key reasons for undertaking mixedmethods research.The question, however, is what triangulation means in thiscontext, as the use of the term is not consistent among researchers.Accordingto the typology of mixed methods designs suggested by Greene et al.(1989) but also more recently by (Bryman, 2006) the term stands for convergenceof findings and corroboration of research results.According to this view, theexpectation is that different datasets or different methodologies will lead tosimilar results and hence allow for confident interpretation (e.g.Lyons, 2000:280) of the findings and strengthen the researcher s conclusions.As such theterm is also widely associated with the concept of credibility of research find-ings.A problem associated with this approach is the assumption that data col-lected using different methods can necessarily be compared and/or contrastedin order to answer the same set of research questions.This view assumesthat there is one single objective reality or truth not only a problematic Combining Methods in Linguistic Research 35assumption (as seen in various chapters of this book), but also, as argued byHarden and Thomas (2005) one that ignores that data from different sourcesoften reveal conflicting realities.At the same time it is important to stress that triangulation (as definedabove) is not the only purpose of mixed methods research.In their early work,Greene et al.(1989) suggested an influential typology of mixed methodsdesigns including four mixed methods purposes (apart from triangulation):initiation aiming at discovering meaningful contradictions and the paradox ;complementarity aiming at shedding light on different aspects of the samephenomenon; development aiming at using findings elicited by the use ofone method for the design of the second or subsequent one; and expansion aiming at broadening the scope and objective of the research (see Tashakkoriand Teddlie, 2003, for further discussions of the model, and Bryman, 2006).Bryman (2006) further showed that a large number of scholars undertakemixed methods research in order to further elaborate their findings.I focus on triangulation here as it is the term most commonly used, andalso often used in a generic way to refer to all purposes of mixed methodsresearch.As Tashakkori and Teddlie argue, over the years triangulation hasbecome a veritable magical word (2003: 674), with the concept being criti-cized for being so broad that it is debatable whether it has any analytical value.Triangulation is so commonly associated with mixed methods research thatTashakkori and Teddlie encourage mixed methodologists to refrain fromusing it unless they specify how it was specifically defined in their researchcontext (2003: 674).Having said that, we need not question the value of triangulation per se butwe need to differentiate between the technical term and the concept of mixedmethods designs as a whole.Even though neither is a panacea for any researchdesign, when applied in relation to a robust conceptual framework (seeCreswell et al., 2003) triangulation (in either sense) does lead to a better under-standing of complex research questions and environments.For example,Dörnyei (2007) suggests that a better understanding of phenomena can emergefrom triangulated findings (whether convergent or divergent), and also reportson the value of mixed methods designs for classroom research where chal-lenges (such as the diversity of student/teacher body) may be addressedthrough versatile designs (I return to the issue of versatility in relation to mixedmethods later in this chapter).A final point about triangulation emerges from Bryman s analysis (2006)of 232 articles in the social sciences, suggesting that it is often an outcome of 36 Research Methods in Linguisticsmixed methods research despite the fact that the desire to triangulate was notthe original motivation for opting for this type of research.As put by Holmesand Meyerhoff (2003: 12), researchers fruitfully combine aspects of differentmethodologies to answer the questions that arise in the course of theirresearch , and often they are not concerned with the surrounding epistemo-logical debates (or they take what Teddlie and Tashakkori (2003) describeas the a-paradigmatic stance).In other words, researchers undertake mixedmethods research in order to answer their specific research questions withoutpositioning themselves to either qualitative, quantitative or mixed methodsparadigms (Harden and Thomas, 2005).Bryman (2006) further usefully dis-tinguishes between rationale (where explicitly stated) and practice: in 27% ofall articles he analysed, the researchers did not explicitly state the purpose forundertaking mixed methods research, and out of the 80 articles that applied atriangulation design, only 19 set this as an explicit rationale.Interestingly sur-veys (quantitative) and interviews (qualitative) seem to be the most dominantmethods used by researchers.Whether explicitly mentioned or not, it remains the case that, multilayereddesigns are often preferred to one-dimensional ones for eliciting rich findings(e.g.Northey, 1990).To further illustrate this, I now turn to studies that haveused a wide range of methodologies in the field of workplace discourse.2.4 Applying mixed methodologies inresearch on workplace discourse4Given the multifaceted nature of research on discourse, it has been argued thatcollecting data from different sources in an iterative way is an appropriate wayto address research questions in this area (Beaufort, 2000).While discoursestudies are often seen as by nature qualitative, being largely based on naturallyoccurring real-life data, recent work (e.g.Holmes and Marra, 2002) has shownhow quantitative and qualitative paradigms can be combined for a betterunderstanding of the interactants norms and practices in discourse [ Pobierz caÅ‚ość w formacie PDF ]
zanotowane.pl doc.pisz.pl pdf.pisz.pl centka.pev.pl
.2.3 Triangulation: An overused term?Triangulation as a central methodological concept comes high on the list ofkey features of good research designs (Cohen and Manion, 1994: 233).Theway the term is conceptualized by scholars is however epistemologicallyvaried.Denzin s (1970: 472) early work indicated that there is more than onetype of triangulation:Data triangulation (the application of more than one sampling method for datacollection)Investigator triangulation (the involvement of more than one researcher)Theoretical triangulation (the use of more than one theoretical stance)Methodological triangulation (the use of more than one methodology).Data triangulation and methodological triangulation are arguably the mostcommon operationalizations of the term; the former refers to data gatheringmethods, while the latter is broader and refers to the use of more that onemethodology in a research design.Denzin also drew an interesting distinctionbetween inter-method and intra-method triangulation the former referringto the use of facets of the same method and the latter referring to the use oftwo (often contrasting) methods (see Schryer, 1993, for an example).Triangulation is often one of the key reasons for undertaking mixedmethods research.The question, however, is what triangulation means in thiscontext, as the use of the term is not consistent among researchers.Accordingto the typology of mixed methods designs suggested by Greene et al.(1989) but also more recently by (Bryman, 2006) the term stands for convergenceof findings and corroboration of research results.According to this view, theexpectation is that different datasets or different methodologies will lead tosimilar results and hence allow for confident interpretation (e.g.Lyons, 2000:280) of the findings and strengthen the researcher s conclusions.As such theterm is also widely associated with the concept of credibility of research find-ings.A problem associated with this approach is the assumption that data col-lected using different methods can necessarily be compared and/or contrastedin order to answer the same set of research questions.This view assumesthat there is one single objective reality or truth not only a problematic Combining Methods in Linguistic Research 35assumption (as seen in various chapters of this book), but also, as argued byHarden and Thomas (2005) one that ignores that data from different sourcesoften reveal conflicting realities.At the same time it is important to stress that triangulation (as definedabove) is not the only purpose of mixed methods research.In their early work,Greene et al.(1989) suggested an influential typology of mixed methodsdesigns including four mixed methods purposes (apart from triangulation):initiation aiming at discovering meaningful contradictions and the paradox ;complementarity aiming at shedding light on different aspects of the samephenomenon; development aiming at using findings elicited by the use ofone method for the design of the second or subsequent one; and expansion aiming at broadening the scope and objective of the research (see Tashakkoriand Teddlie, 2003, for further discussions of the model, and Bryman, 2006).Bryman (2006) further showed that a large number of scholars undertakemixed methods research in order to further elaborate their findings.I focus on triangulation here as it is the term most commonly used, andalso often used in a generic way to refer to all purposes of mixed methodsresearch.As Tashakkori and Teddlie argue, over the years triangulation hasbecome a veritable magical word (2003: 674), with the concept being criti-cized for being so broad that it is debatable whether it has any analytical value.Triangulation is so commonly associated with mixed methods research thatTashakkori and Teddlie encourage mixed methodologists to refrain fromusing it unless they specify how it was specifically defined in their researchcontext (2003: 674).Having said that, we need not question the value of triangulation per se butwe need to differentiate between the technical term and the concept of mixedmethods designs as a whole.Even though neither is a panacea for any researchdesign, when applied in relation to a robust conceptual framework (seeCreswell et al., 2003) triangulation (in either sense) does lead to a better under-standing of complex research questions and environments.For example,Dörnyei (2007) suggests that a better understanding of phenomena can emergefrom triangulated findings (whether convergent or divergent), and also reportson the value of mixed methods designs for classroom research where chal-lenges (such as the diversity of student/teacher body) may be addressedthrough versatile designs (I return to the issue of versatility in relation to mixedmethods later in this chapter).A final point about triangulation emerges from Bryman s analysis (2006)of 232 articles in the social sciences, suggesting that it is often an outcome of 36 Research Methods in Linguisticsmixed methods research despite the fact that the desire to triangulate was notthe original motivation for opting for this type of research.As put by Holmesand Meyerhoff (2003: 12), researchers fruitfully combine aspects of differentmethodologies to answer the questions that arise in the course of theirresearch , and often they are not concerned with the surrounding epistemo-logical debates (or they take what Teddlie and Tashakkori (2003) describeas the a-paradigmatic stance).In other words, researchers undertake mixedmethods research in order to answer their specific research questions withoutpositioning themselves to either qualitative, quantitative or mixed methodsparadigms (Harden and Thomas, 2005).Bryman (2006) further usefully dis-tinguishes between rationale (where explicitly stated) and practice: in 27% ofall articles he analysed, the researchers did not explicitly state the purpose forundertaking mixed methods research, and out of the 80 articles that applied atriangulation design, only 19 set this as an explicit rationale.Interestingly sur-veys (quantitative) and interviews (qualitative) seem to be the most dominantmethods used by researchers.Whether explicitly mentioned or not, it remains the case that, multilayereddesigns are often preferred to one-dimensional ones for eliciting rich findings(e.g.Northey, 1990).To further illustrate this, I now turn to studies that haveused a wide range of methodologies in the field of workplace discourse.2.4 Applying mixed methodologies inresearch on workplace discourse4Given the multifaceted nature of research on discourse, it has been argued thatcollecting data from different sources in an iterative way is an appropriate wayto address research questions in this area (Beaufort, 2000).While discoursestudies are often seen as by nature qualitative, being largely based on naturallyoccurring real-life data, recent work (e.g.Holmes and Marra, 2002) has shownhow quantitative and qualitative paradigms can be combined for a betterunderstanding of the interactants norms and practices in discourse [ Pobierz caÅ‚ość w formacie PDF ]