Abstract

Legacy language materials: field notes, transcriptions, and audio recordings created by earlier generations of researchers, occupy a privileged position in the study and revitalization of endangered languages. When a language is no longer transmitted to children, when remaining speakers are elderly and few, and when contemporary speech is visibly marked by contact-induced change and domain restriction, materials collected decades earlier inevitably come to be perceived as capturing something more "complete," more "authentic," more "true." This perception is shared, though for different reasons, by linguists seeking diachronic baselines and by community members seeking an uncompromised model for revitalization.

This paper argues that the purity attributed to legacy materials is an ideological construction that obscures the cumulative mediations inherent in any archival record. Drawing on the Tomsk Field Archive of Siberian Indigenous Languages - a collection of over 180 volumes of handwritten field notes covering eight languages across four genetic groupings, assembled over four decades of systematic documentation beginning in 1947, I trace five categories of mediation, what I term "degrees of separation," between the archived record and the communicative events it purports to represent. These include the interpretive filtering inherent in real-time field transcription, the systematic smoothing introduced through secondary editing and "cleaning," the analytical choices embedded in notation normalization, the structural transformations imposed by digital rendering, and the epistemic flattening that occurs when legacy and contemporary data are integrated into undifferentiated multimedia databases.

Using Irvine and Gal's (2000) framework of language ideology, I show how two semiotic processes sustain the perception of archival purity. Iconization collapses the distinction between the archive and the language it documents, such that the collection comes to stand for the language itself. Erasure renders the successive mediations invisible, stripping legacy materials of their production history and presenting them as transparent windows onto a past linguistic reality. Together, these processes generate a paradox: the most heavily mediated version of the data, the one furthest removed from original speech events, carries the greatest institutional authority.

The paper concludes by drawing out implications in three domains: source-critical research practice, including a proposal for systematic "mediation metadata"; meaningful Indigenous data governance under the CARE Principles, with attention to the specific constraints of the Russian institutional context; and a broader reconsideration of archival purity as an ideological phenomenon that shapes not only how legacy materials are used but how "language" itself is conceptualized in documentary and revitalization work.