The Gold Standard of Anthropology | Interactional Repair, Other-Initiation and Social Accountability

1. Begin with a repair sequence, not an assumed mistake

Back to contents

Identify the trouble source, the repair initiation, the repair solution and the return to the ongoing activity. A trouble source is whatever participants treat as problematic; it need not be objectively false or grammatically defective. A researcher may hear a name correctly while the recipient displays uncertainty about which person is meant. Start from observable conduct—delay, repetition, a questioning expression, a candidate understanding—and show how the next turns make the problem relevant.

2. Separate who initiates repair from who completes it

Back to contents

Repair organisation distinguishes initiation from completion. A speaker can halt and replace a word before anyone responds. A recipient can initiate repair and give the original speaker an opportunity to clarify. Direct other-correction is possible, but many ordinary sequences favour self-repair by the speaker of the trouble source. These are sequential opportunities, not personality types. Count the pathway only after locating where initiation and solution occur.

3. Use position to understand what repair is doing

Back to contents

Other-initiation commonly appears in the next turn, but later repair positions matter. Third-position repair can occur when a speaker discovers from the recipient’s response that an earlier turn was misunderstood. Fourth-position repair can expose a problem one step later still. Position changes the evidence available to participants. A delayed correction after consequences have unfolded is not analytically equivalent to an immediate “who?” or “which one?”.

4. Distinguish open from restricted repair initiators

Back to contents

An open form such as “huh?” indicates trouble without locating it precisely. A restricted form can repeat a word, offer a candidate hearing or ask a targeted question such as “the east gate?”. Greater specificity can speed repair, but it also proposes what the problem is. Compare the form with the response: repetition may confirm hearing trouble, whereas replacement or explanation may reveal reference or understanding trouble.

5. Read accountability in the design of the initiation

Back to contents

Repair initiations can make one party’s conduct answerable. “What do you mean?” may attribute obscurity to the prior speaker; “Did you say Tuesday?” can present the difficulty as hearing; a candidate understanding can display effort and reduce blame. In classrooms, clinics or interviews, institutional roles shape who may challenge whom and how directly. Accountability should be demonstrated through wording, timing and uptake, not inferred from politeness stereotypes.

6. Worked example: repairing a departure time

Back to contents

In an invented bus-planning exchange, A says, “It leaves at seven fifteen from Pier Two.” B repeats, “Seven fifteen?” with rising intonation. A responds, “Sorry—seventeen fifteen, five fifteen in the afternoon,” and B says, “Right, after the museum,” before returning to the itinerary. B’s repeat targets the time without accusing A of error. A treats the problem as a mistaken production, supplies two equivalent formats and takes responsibility with “sorry.” The resumption confirms a locally sufficient repair. The analyst should not claim that B misheard: the recording and A’s correction support the narrower conclusion that the time became repairable and was replaced.

7. Build collections and retain deviant cases

Back to contents

A single vivid excerpt can illustrate a practice but cannot establish its distribution. Assemble comparable cases with explicit inclusion rules, recording the trouble source, initiation form, position, response and outcome. Inspect cases where repair fails, is resisted or changes the activity. Frequencies can support a pattern, yet rare practices may be socially consequential. Collection building and close sequential analysis should correct each other.

8. Changed-context transfer check: a clinical handover

Back to contents

Move from casual planning to a nurse–doctor handover. The sequence still contains trouble source, initiation and repair, but stakes and authority change the available designs. “Which dose?” may be a neutral safety check, while silence or a softened candidate can reflect role asymmetry. Documentation systems and alarms can also initiate practical repair. Transfer the sequential method, then add institutional responsibilities, risk and evidence from the work setting.

9. Know the limits and common misconceptions

Back to contents

Repair is not a direct window into confusion, memory or intent. An open-class initiator does not prove total non-understanding, and absence of repair does not prove mutual comprehension. Transcription choices can hide gaze, overlap or embodied repair. Cross-linguistic similarities do not erase language-specific grammar and social relations. Researchers should not turn a preference for self-repair into a moral rule that direct correction is universally rude.

10. Audit the collection from recording to claim

Back to contents

Retain source recordings securely, transcript conventions, timing decisions, case-selection rules, exclusions and links between every excerpt and its media. Mark uncertain hearing rather than repairing the transcript silently. Compare candidate collections across speakers, activities and languages where the claim requires it. Publish enough sequence before and after the repair for readers to inspect the proposed trouble and resumption.

11. Frequently asked questions

Back to contents

Is every repetition a repair initiation? No; repetition can agree, challenge, display surprise or continue an activity. Does “huh?” mean the speaker was unclear? Not necessarily; it locates a problem without assigning a cause. Can repair be non-verbal? Yes, gaze, gesture and object handling can initiate or accomplish repair. Does self-repair always happen first? It is often structurally favoured, not guaranteed. Can counts replace transcript analysis? No; categories depend on sequential evidence.

12. The gold standard makes repair socially inspectable

Back to contents

Strong repair analysis shows how participants themselves make trouble visible and resolve it. It separates initiation from completion, respects sequential position, compares open and restricted designs and treats accountability as something built in the exchange. The result is more than a catalogue of errors. It explains how people protect progress, knowledge and relationships while restoring enough shared understanding to continue.