Speech corpora constructed in field linguistics are usually characterised by their small volume. They mostly concern little described languages, for which data are scarce. Field work also has its particularities, which means that recordings are closer to “spontaneous” than to laboratory speech. These characteristics influence the methods used for constituting and managing these corpora. To illustrate these specific methods, we will present a program for archiving field linguistics documents, aiming to preserve, perpetuate, normalise and circulate these documents. In particular, we will examine past and present formalisms and tools for managing corpora of this type. To finish, we will evoke some problems in organisation and law concerning the circulation of these resources.