The Gold Standard of Anthropology | Membership Categorisation Analysis, Category-Bound Activities and Inference

1. Begin with the action, not a demographic label

Back to contents

Start from what participants are doing: blaming, requesting, explaining, joking, recruiting help or defending a decision. Then ask whether a category is named, described or made inferentially available in that sequence. A participant may say “as her supervisor” to authorise advice, while another may resist that footing. Age, gender or occupation in a transcript header is not automatically relevant to the action. The analytic burden is to show how the category enters the interaction and what it helps participants accomplish.

2. Separate categories, collections and devices

Back to contents

A membership category is one possible label for a person; categories can be heard as belonging to collections such as family, workplace or school. A membership categorisation device is a locally usable way of organising people from such a collection. The same person can be categorised as parent, customer, engineer or neighbour. Analysts should not choose one “real” identity. They should show which device the participants make relevant and how alternative devices are invited, blocked or replaced.

3. Treat category-bound activities as defeasible inferences

Back to contents

Some activities are conventionally heard as expectable for a category: teachers mark work, doctors diagnose and hosts welcome guests. These category-bound activities support recognition, but they are not universal laws. A statement such as “she checked the dosage twice” may invite hearing the person as a clinician only in a suitable context. Record the words, sequence and shared setting that license the inference, and retain rival hearings rather than converting cultural expectation into a factual attribute.

4. Use economy and consistency cautiously

Back to contents

Participants can infer a collection from a small number of category references, and once a device is established they may hear later references consistently through it. These classic observations are analytic resources, not mechanical coding rules. A later turn can switch device deliberately, and a single description can support several collections. Test the proposed device against subsequent uptake, repair, laughter, agreement or resistance. If participants revise the categorisation, the analysis must follow that revision.

5. Join category analysis to sequential analysis

Back to contents

Categories gain meaning through where and how they occur. A category in an accusation works differently from the same word in self-deprecation or a form-filling answer. Examine the prior turn, the category formulation, the response and the next action. Pronouns, descriptions, reported speech, lists and extreme-case formulations can all make membership relevant without a bare label. Sequential evidence prevents the analyst from treating isolated nouns as a complete social taxonomy.

6. Worked example: “the new volunteer”

Back to contents

In an invented community-kitchen meeting, Mina says, “The new volunteer left the freezer open again.” Raj replies, “You mean Tariq? He was covering two stations because Sam went home.” Mina then says, “I’m not blaming him; I’m saying a coordinator should have checked.” “New volunteer” first supports an inference of inexperience and accountability. Raj replaces the category with a personal name and situational workload, resisting that inference. Mina shifts responsibility to “coordinator.” The finding is not that volunteers are careless; it is that participants negotiate which category should organise responsibility for this event.

7. Build collections without stripping away context

Back to contents

A robust study gathers comparable category practices using an explicit search rule and enough surrounding talk to recover the action. Code the category expression, proposed collection, category-bound predicate, sequence position, speaker footing and recipient response. Keep cases that fail, are repaired or receive no uptake. Counts can describe a collection, but they cannot substitute for showing why cases are comparable or how a formulation performed its work.

8. Changed-context transfer check: an online moderation thread

Back to contents

Move from a spoken meeting to a forum where a moderator writes, “As a first-time poster, please read the rules.” The same method asks what category is invoked, which obligation is bound to it and how the recipient responds. The context adds profile badges, algorithmic labels, asynchronous replies and absent audiences. A platform-assigned “new member” tag may supply the category before anyone chooses it. Transfer the inference discipline, then analyse how the medium distributes categorising rights and visibility.

9. Know the limits and common misconceptions

Back to contents

Membership categorisation analysis is not a licence to infer hidden prejudice from any category word, and it is not a list of cultural stereotypes. Participants can exploit, parody or reject conventional associations. Silence does not prove acceptance, while an analyst’s plausible interpretation is not automatically a members’ method. Categories are locally occasioned but can connect to durable institutions and inequalities. The task is to demonstrate that connection rather than choosing between “only local” and “entirely structural” accounts.

10. Audit transcripts, searches and analytic decisions

Back to contents

Retain source recordings, transcription conventions, search terms, inclusion rules, candidate cases and excluded near-matches. Link every claim to the sequence that supports it, including prosody, gaze or on-screen context where available. Record alternative categorisations considered during analysis and show how recipient uptake strengthened or weakened each reading. Protect identities without replacing category expressions that are themselves analytically important; explain any necessary redaction.

11. Frequently asked questions

Back to contents

Must a category be spoken explicitly? No; descriptions and category-bound activities can make it inferentially available, but the evidence must be shown. Are stereotypes the same as category-bound activities? No; the latter are participants’ defeasible practical associations, not analyst-endorsed truths. Can MCA use counts? Yes, after cases are sequentially established. Is every identity relevant at once? No. Does a category explain behaviour? It can organise an account, but it does not prove motive or essence.

12. The gold standard keeps categorisation participant-grounded

Back to contents

Strong membership categorisation analysis shows how people assemble social identity in the course of action. It distinguishes categories from collections, treats bound activities as defeasible, follows sequential uptake and preserves rejected or shifting devices. The result is neither a demographic inventory nor free association. It is an inspectable account of how participants make categories consequential, and of where the evidence stops.