The article defines a new referential problem of ethnographic description: the verbalization of the “silent” dimension of the social. As a documentary procedure, description has been devalued by more advanced recording techniques that set a naturalistic standard concerning the reification of qualitative “data.” I discuss this standard from the perspective of the sociology of knowledge and replace it by a challenge unknown to all empirical procedures relying on primary verbalizations of informants. Descriptions have to solve the problems of the voiceless, (...) the silent, the unspeakable, the pre-linguistic, and the indescribable. Ethnography puts something into words, which did not exist in language before. To respond to this task, descriptions have to turn away from the logic of recording and develop into a theory-oriented research practice. (shrink)