perfect English, but he keeps falling back into German” generated the code “code-switching”. The qualitative data analysis software ATLAS.ti supported this coding process.
In the subsequent step of analysis, we brought together different first-order codes, which linked together through higher-order categories. We teased out these categories and examined the theoretical relationships between them using the constant comparative method. We compared different parts each interview, different inter- views within each MNT, statements of interviewees on different hierarchical levels, in different teams and corporations. During this comparative process, connections between previously fragmented codes emerged. We merged similar codes into superordinate cat- egories or linked them to form a new category. With respect to language-based emotions, for example, we collapsed the codes “lan- guage-based inhibition”, “feeling of insecurity” and “fear of face loss” into the higher-order category “anxiety”. Regarding leadership strategies, we merged the codes “code-switching”, “speaking time” and “redundant communication” into the category “reducing the impact of language barriers”, consolidated the codes “humor” and “common goals” into the category “redirecting attention” and brought the codes “appreciation” and “meta-communication” together into the category “positive appraisal of linguistic diversity”. Leadership outcomes were condensed in a similar fashion. In the final step of analysis, we arranged these categories into a set of core findings. During the entire data analysis process we cycled back and forth between our data, codes and categories in an iterative fashion until no new categories emerged and saturation was reached. We also constantly recorded ideas about codes and their relationships in the form of theoretical memos.
Upon finalization of the data collection phase, we scrutinized the validity of our findings and conclusions by sharing an eight-page summary of findings and practical recommendations with all interviewees and by conducting a feedback round with selected infor- mants. Doing so yielded additional details that sharpened our conclusions and ensured that we captured the impact of language on MNT members' emotions and the moderating role of MNT leaders as they pervaded informants' experience.
Results
To answer our first research question, we will explore MNT members' negative emotional reactions to language barriers. Based on our analysis through constant comparison, we will classify these negative emotions into two discrete categories and thereby provide a systematic problem description. To address our second and main research question, we will show possible solutions by outlining a variety of measures MNT leaders have at their disposal to counter-act language-induced negative emotions. Our content analysis has shown that we can subsume these measures under three general strategies. To highlight the relevance of these findings, we will also outline the positive outcomes of successful emotion management through MNT leaders.
Defining the problem: two categories of negative emotional reactions to language barriers
Our critical incident investigations revealed that language barriers gave rise to substantial negative feelings, particularly at the be- ginning of a team's cooperation. During our iterative coding process, we discovered that we can subsume the various problems under two discrete categories. We labeled these two categories self-directed anxiety and other-directed resentment.
Self-directed anxiety triggered by language barriers
In terms of self-directed emotions, interviewees who judged their proficiency in the shared language as unsatisfactory, associated communication in a foreign language with “insecurity”,