Early oral historians defended their sources by arguing that spoken accounts could be checked and were as reliable as documents (Vansina, 1985). A later generation changed the question. Portelli (1991) showed that the ways in which narrators misdate, merge and reshape events are themselves evidence of what those events meant to them. On this view an interview yields two things: information about the past, and a present-day account of a life, shaped by the encounter with the interviewer. Current theory treats the second as the distinctive contribution of the method (Abrams, 2016). For a declining craft the implication is that stories of decline should be recorded as interpretations, with attention to who tells them and to whom.

Halbwachs (1992) argued that individuals remember within frameworks supplied by the groups they belong to, so that a family, a caste or guild, or a neighbourhood sustains a shared version of its past. Connerton (1989) added that societies also remember through habitual bodily practice, in posture, gesture and routine, and that such memory is passed on without being put into words. A loom is a place where both kinds meet. The rhythm of the shuttle is bodily memory; the story of how the village came to weave a certain cloth is collective memory. A method confined to the second would miss the first.

The difficulty of describing skill is a classic problem. Polanyi (1966) summed it up in the remark that people know more than they can tell. Studies of apprenticeship have since described how such knowledge is acquired: newcomers begin with simple tasks at the edge of the work and move towards full participation, learning by watching and doing more than by instruction (Lave and Wenger, 1991; Marchand, 2008). Sennett (2008) stresses that skill is a trained responsiveness to materials and not the execution of a plan. Learning of this kind depends on years spent near the work, which is what is lost when children leave the loom shed for school and wage labour.

Mannheim (1952) proposed that people who meet the same historical changes at the same stage of life form a generation with a shared outlook. In a weaving village the idea has a concrete form. Those who learned as children when weaving was the only livelihood, those who trained but left, and those who have never woven stand in different relations to the craft and will narrate its decline differently. The method treats these positions as a structuring device for the whole project. Doing so also guards against a common bias of salvage projects, which is to interview only the oldest practitioners and so to record the craft solely as its elders wish it remembered.

The project begins with open meetings at which the team explains what it hopes to do and asks what residents would want recorded. Frisch (1990) called the principle shared authority: the interpretation of a life belongs to narrator and historian together. In practice this means a short written statement of aims, agreed with a weavers' cooperative or village council where one exists, and revised if the meetings show that the team's interests and the residents' diverge.

Consent is sought from each narrator before recording and confirmed afterwards, when the person knows what has been said. The agreement should state who holds rights in the recording, where copies will be kept, who may hear them and whether any passages are closed. Professional guidance treats these as continuing obligations and not a form signed once (Oral History Association, 2018). Designs and techniques may have commercial value to the families who own them, and narrators should be able to withhold them from public access.

A household list is drawn up with local help, noting for each adult the relation to weaving: active weaver, former weaver, never a weaver. It also notes roles that interviews with weavers alone would miss, among them the women and children who wind bobbins and prepare warps, dyers, loom carpenters, yarn dealers and the traders who commission cloth. Narrators are then chosen to cover the grid of generation and role. The aim is range, and no claim of statistical representativeness should be attached to the selection.

Each narrator is first invited to tell the story of his or her life with little interruption, and only afterwards asked about topics on a common guide: learning, the working day, relations with merchants, festivals and the ritual uses of cloth, the moment of leaving or staying. The life-story form lets narrators set their own emphasis and shows where weaving sits in a life (Thompson, 2000). Interviews are held in the narrator's first language, at home, and where possible by an interviewer from the region. Interviewers note silences and topics avoided, which accounts of working lives often contain and which carry meaning of their own (Abrams, 2016).

A second session takes place where the work is done. The narrator demonstrates a sequence of operations on video and talks through it, handles tools, old cloths, pattern books or photographs, and names parts and processes. Objects prompt recollection that questions do not, and demonstration records what cannot be said. Where a loom stands idle, a former weaver may be asked to sit at it again. Sessions that bring an older weaver together with a younger relative are especially useful, because the explanation is then pitched at a learner and the exchange between the two is itself a record of how transmission worked or failed. Terms for tools, motions and faults are gathered into a glossary as the sessions proceed.

Recordings are transcribed in the original language before translation, keeping dialect words and technical terms and marking pauses and laughter. A time-coded index by topic makes the audio and video searchable, which matters because the recording, with its tone and gesture, remains the primary source and the transcript a guide to it (Portelli, 1991).

Analysis proceeds by reading across the generational grid. Thematic analysis, in which transcripts are coded and the codes grouped into themes through repeated reading, offers a transparent procedure (Braun and Clarke, 2006); it should be supplemented by attention to narrative form, to how stories are told as well as what they contain. Draft interpretations are taken back to narrators and to a village meeting, and disagreements are recorded. Copies of all material remain in the village in a usable form, and a full set is deposited in a public archive under the agreed conditions.