During WWI an estimated of 650,000 colonial soldiers, recruited by the Triple-Entente, were sent to European battlefields. Yet, in most historiographies, the involvement of the non-white soldiers in the war has attracted scant comment. German propaganda campaigns alerted linguists and anthropologists to the presence of colonial soldiers in the POW camps who became the target of research. Resulting from this, the Lautarchiv (sound archive) in Berlin holds some 400 recordings of African prisoners of war (of altogether 1650 recordings with POWs), which were produced between 1915 and 1918 by the Königlkh Preussische Phonographische Kommission. Few of those recordings have so far been translated. In this paper I understand these recordings not as voices but as echoes – of accounts of the self, and of the war at times, using the concept of echo as a means to grapple with extraction, limitation, distance, and the distortion or outright effacement that is the result of mediation, the delay (or belatedness of hearing), and gaps in meaning and intelligibility. The restraints imposed on the speaker are a result of the linguists' will to extract grammar from semantics, so as to limit the potential distraction that a narrative of an ordeal (being hungry, wounded, homesick, betrayed, insulted) could entail. Conceptualizing the recorded voices and their translation as echoes, I seek to understand the status of voices that were recorded according to the logic of linguistic practice and the situation in the camps. This allows us to position these subaltern articulations in their mediated, distorted form as part of the colonial archive. My paper further engages with the questions that result from the precarious status of those recordings: do they represent subaltern positions of historiology of WWI? How can they be presented in an exhibition? This talk was recorded but cannot be shown for copyright reasons.