Cognitive psychologists and ergonomists have proposed various methods for the analysis of individual verbal protocols, but much less for dialogues in collective work settings. Many professional activities, however, are carried out by people working together through verbal interaction. From a perspective of cognitive ergonomics, we have developed principles for the analysis of collaborative design, amongst which the COMET method. In the framework of the MOSAIC project, researchers from cognitive ergonomics and linguistics have compared analysis methods adopted and results obtained. One of the directions has been to confront the methodological approaches adopted by linguistic interaction analysis and cognitive ergonomics to the analysis of the MOSAIC corpus through a twofold analysis of two identical segments of this corpus. The results of this confrontation concern two levels: the data-analysis method (the objects of analysis and the levels of structuring the corpus) and the interpretations adopted by each discipline (especially, the status attributed to a pre-existing theoretical framework). Interestingly, the methodological questions have been not only of a comparative nature: they have influenced also each individual analysis. The cognitive ergonomics' analysis indeed has been extended in two directions compared to previous analyses: consideration of interaction's multi-modality and introduction of a finer way to code the design solutions. The present chapter presents the conclusions of the confrontation between the two approaches.