Send message to Authors

Abstract

A call has been made in recent years to de-colonize machine learning and AI by putting them into use for emancipation of marginalized voices instead of profit. Replying to this call, our project implements an NLP analysis of over 200 travelogues describing the life of ethno-social communities on the territory of present-day Kazakhstan between the 1760s and 1910s. Travelogue literature, written mostly by Russian administrative and military officials and European travellers, constitutes one of the richest sources of knowledge about people of Central Asia at the time. Subject to critical analysis, this literature - manifesting hybridity of a “contact zone” - becomes indispensable for recovering the voices of indigenous population. Our study answers the question which factors determined to what extent the local voices were represented or reduced to the passive objects of imperial gaze. A team of undergraduate coders manually processed the travelogues from Turkestanskii sbornik, selecting over 3000 passages covering boundary-setting features, symbols and markers of local identities (e.g. rites, material culture, self-names, etc.). In addition, those passages were labeled according to the presence of indigenous voices. Then, the shares of local voice, top keywords, and bi-gram probabilities were calculated centuries, decades, and authors in R. We find that indigenous presence was filtered through the imperial gaze with weak local voices as the norm across all decades, indicating that Central Asian people were described almost entirely through an external lens. Yet, local voice was suppressed not uniformly or incidentally, but depended on the author’s identity, depth of local embeddedness and purpose of writing. The local voice markedly receded in the travelogues by the end of the nineteenth century signifying the final Russian conquest of Central Asia. In other words, similarly to Africa and Latin America, political colonial expansion invigorated imperial gaze. Ethnographers, culturally embedded travellers, and orientalists included the higher shares of moderate and strong local voices, while writers with military and administrative affiliations proved to be the most exclusive. Thus, we add a more systematic nuance to the conclusions of those scholars who have found diversity or even “the anti-conquest” in imperial travel writing.