This paper proposes a new approach for analysing daily activities in a laboratory. The case study presented is an analysis of shop-talk around a microscope. In addition to the classical approaches, such as ethnomethodology and anthropology of science, I argue that a microsemiotic approach could be useful to better understand what is at stake. The semiotic approach I shall use here was proposed by a group of Belgian semioticians: Groupe μ. This semiotic approach leads to a constructivist point of view: (...) the meaning of a visual representation is progressively constructed and is very context-dependent. This semiotic approach is fruitful because it allows a very precise analysis of shop-talk recorded data, and gives a better account of the materiality of visual representations. (shrink)