Methodological contributions – Bridging qualitative and quantitative

103

Nordic universities maintain the importance of the local languages in teaching and academic dissemination, we do not find a moderating role of local language proficiency. This suggests that, in our sample, the activities which are carried out in English, such as the majority of international publication and networking, are prioritised in collaboration performance evaluations.

The findings and their enabling conditions demonstrate how the specificity of conditions contribute to the functioning of the mechanisms behind English as an organisational language. However, as causality is conjunctional in critical realist terms, the same outcomes of the empirical studies may be observed in other cases, but caused by other mechanisms (Rihoux and Ragin, 2009; Welch et al., 2011).

Similarly, the same mechanism and similar enabling conditions may produce other outcomes in other cases. For instance, post-merger organisations based outside the Nordic countries may also use English as an organisational language, but supported by a different ideology. Furthermore, in a sample of only Norwegian universities, despite also hosting an English-speaking scholarly environment, peers may emphasise local language proficiency more in performance evaluations, due to the strong role of Norwegian in local education (Ljosland, 2014). Thus, the findings of this thesis are relevant beyond the context of the empirical material, but need to be adapted and contextualised according to the specific linguistic and institutional conditions.

8.2 Methodological contributions – Bridging qualitative and quantitative methods

104

contributes to this movement by showing how an abductive case study has shifted both empirical focus and theoretical emphasis in response to emerging salient topics. While initially opting to study the relationship between language policy and language practices, I discovered that the still underway post-merger integration, coupled with the low degree of language policy codification and awareness, highlighted discursive legitimation as the most relevant theoretical concept to study.

Furthermore, the access conditions and COVID-19 pandemic restrictions required a flexible design which still maintains the conceptual focus on English as an organisational language. A rigid, more procedural study would have stuck to the policy/practice relationship, only to find little theoretical value in this finding alone (Piekkari and Welch, 2018). However, the other data sources, participant observation notes and corporate documents, revealed a multilingual organisation in the early stages of post-merger integration, where English plays a strong, but uneven and undefined role as the organisational language. Thus, I contextualised the weak policy/practice relationship and shifted the analytical focus to legitimation, given the strong presence of normalisation in the interviews from the Danish/German context.

The use of semi-structured interviews as the empirical basis for discourse analysis is a further methodological contribution of the present thesis. Recent discussions of methods in language-sensitive IB and management research have encouraged scholars to go beyond interviews as the main data point, in order to capture the interactive element of language in natural speech situations (Tietze, 2020). However, the value of semi-structured interviews in the study of ideological articulation is re-emphasised, since the method accesses individual-level legitimation, which is subsequently contextualised in the enabling factors of the organisation. The depth of data has allowed for a rich explanation of several legitimation categories, but their interpretation has relied on extensive physical and virtual stays in the organisation, in order to make sense of legitimation in the context

105

of the post-merger situation. Thus, the methodological value of ethnographic methods is also emphasised (Hammersley and Atkinson, 1995; Rees and Gatenby, 2014), as the importance of encountering organisational members outside interview situations, as well as the ability to access organisational information portals, provides rich contextual information about the organisation, which could not have been accessed if contact was limited to interviews.

The application and theorisation of the quantitative analysis in Paper 3 contributes to the critical realist literature of quantitative methodology (e.g.

Downward et al., 2002; Hurrell, 2014; Zachariadis et al., 2013). The questionnaire results in Paper 3 function as an independent basis for theorising mechanisms in predominantly Nordic academic contexts. The study is a rare statistical analysis of a moderating variable role interpreted through the Bourdieusian conceptualisation of linguistic capital (1977; 1991) and theorised as a mechanism in line with critical realism. Regression analysis is frequently dismissed by many interpretive qualitative scholars and some critical realist scholars for generalising based on falsely stabilised relationships in an artificial closed system (Sayer, 1992).

However, I argue that Paper 3 demonstrates that hierarchical regression analysis is a potent approach to compare the explanatory power of different independent variables and analyse statistical interaction, where one variable moderates the relationship between two other variables. Since the critical realist research agenda seeks to uncover how interaction between entities on the empirical level reveals structures in the real, the analysis of Paper 3 strengthens the case for hierarchical regression analysis in realist approaches.

A further methodological contribution of the present thesis is the role of critical realism in tying actor-level qualitative and quantitative data together in the theorisation of a single phenomenon – English as an organisational language. On the one hand, the critical realist ontological conceptualisation of an objective reality which can be theorised based on empirical-level indicators of mechanisms and their

106

enabling conditions (Fleetwood, 2005; Ackroyd, 2010) means that both qualitative and quantitative methods may provide data which reveals structures on the real layer. On the other hand, the ontological position frames a methodological discussion of how different methods reflect the empirical layer of reality. Critical realist scholars are no strangers to combining qualitative and quantitative methods, as well as the epistemological discussion of their implications (e.g. Downward et al., 2002; Hurrell, 2014; Zachariadis et al., 2013). Some scholars have argued that the current ontological position is well suited for consolidating the qualitative-quantitative divide in social sciences (e.g. Bryman, 1988). A crucial starting point in critical realist approaches to combining methods, is the overarching purpose of explaining social phenomenon through thick description. This agenda may have led to the domination of qualitative methods, and case studies in particular, due to their strength in outlining the conditions for generative mechanisms and the entities which form them. While some realist scholars have questioned the ability of quantitative methods to complement these strengths in a meaningful way (Sayer, 1992), there are critical realist voices in favour of pursuing both qualitative and quantitative approaches, in order to capture different types of relationships at different levels (Brown and Roberts, 2014; Hurrell, 2014). This does not mean that each approach is intended to cover the deficiencies of the other, as it comes with specific methodological strengths for the given research design and context. Thus, in critical realist designs, ideally, quantitative methods are not employed to offer generalisable findings in a larger sample, but rather to identify patterns of variable interaction within the given context.

Unlike most examples of critical realist mixed-methods research, the present thesis combines two separate empirical investigations on a theoretical level, each in their own setting, to frame how one single conceptual phenomenon functions in different contexts. Hurrell (2014) and Sayer’s (1992) distinction between intensive and extensive methods gives an outline of the strengths of each approach. As an

107

extensive approach, the hierarchical regression analysis in Paper 3 uncovers a pattern on a larger scale across several organisational contexts with a key common trait – a predominantly Nordic academic sector. The identified pattern involving the moderating role of English proficiency suggests a frequent relationship which, in turn, indicates the dominance of English as an organisational language. However, frequency is not conflated with causality here (Hurrell 2014). To uncover the micro-level operationalisation of beliefs about the role of English, the single-case qualitative study in Paper 2 was the most apt approach for intensive research of ideology and legitimation in one single organisation. In the interviews, individual-level articulation of ideology through legitimation outlines the discursive link between the legitimated role of English and the expectation that organisational members should and will align with this role. Thus, the intensive approach of Paper 2 illustrates the micro-level production and reproduction of English as an organisational language, while the extensive approach of Paper 3 demonstrates its implications by identifying a broader pattern in peer evaluations.