Context Sampling Audit¶
Procedure — instantiates Contextual Selective Propagation
Samples real usage across relevant communities, channels, documents, and situations to reveal which meanings are actually active and where transfer risk lives.
Before you can govern how a meaning travels, you have to know what the meaning currently is — in every place it already lives, not just where you assume it lives. Context Sampling Audit is the front-end discovery procedure that pulls a structured sample of real usage — actual sentences, posts, slides, forms, and overheard uses — from each community and channel that touches the term, then reads back what sense each one is really carrying. Its defining move is that it works from observed instances rather than from opinion or a canonical definition: you don't ask people what a word means, you collect how they use it and let the divergence surface itself. The output is a grounded record of the focal term, the range of live senses, and a map of which community holds which — the raw material every downstream mechanism consumes.
Example¶
A museum's digital team notices the word "curate" is drifting inside their own organization and decides to audit it before it causes trouble. They set a sample frame — the curatorial department, the marketing team, the social-media queue, the education program, and the legal language in an active grant agreement — and pull a dozen real usages from each. The curators use "curate" narrowly: scholarly selection plus long-term physical care of objects. Marketing uses it broadly, as in "our curated gift shop." Social media has bleached it further — "a curated feed" means little more than "we chose some things." The grant document uses it operationally, tying "curatorial responsibility" to a legal duty of custodianship.
The audit doesn't rule on which sense is right. It produces a recorded profile: five active senses, ranked from the narrow custodial meaning to the near-empty promotional one, tagged to the communities that hold them, with the collision flagged — marketing's loose "curated shop" is one channel away from language that the grant treats as a binding custodial obligation. That single artifact turns a vague unease into a concrete transfer-risk the team can now decide what to do about.
How it works¶
The procedure is observational, and its steps are what set it apart from simply asking around:
- Define the sample frame. Enumerate the communities, channels, document types, and situations the term actually appears in — deliberately including low-status and informal ones, since that is where drift hides.
- Pull real instances, not reports. Collect concrete uses in the wild rather than definitions people offer on request; self-report smooths over exactly the divergence you are hunting.
- Sense-code each instance. Tag every instance with the sense it carries and the context it came from, sampling until new instances stop revealing new senses.
- Surface divergence and transfer risk. Lay the coded senses against their contexts and flag the pairs where a sense in one channel would be misread — or become unsafe — if it propagated into another.
Tuning parameters¶
- Sample-frame breadth — how many communities and channels you draw from. Wider frames catch more drift but cost more to collect and code; narrow frames are fast but blind to the fringe where meaning mutates first.
- Instances per context — how deep you sample before calling a context saturated. More instances stabilize the sense estimate; too few let one loud voice stand in for a whole community.
- Recency window — how far back usages may be drawn from. A tight window catches live meaning; a wide one captures legacy senses still in circulation but risks mixing eras.
- Coding granularity — how finely you split senses. Fine coding surfaces subtle narrowing/widening; coarse coding is quicker but can merge two senses that matter operationally.
- Observed-vs-elicited mix — how much you supplement wild instances with light elicitation. Some elicitation fills gaps in rare contexts, but the more you lean on it the more the audit inherits self-report bias.
When it helps, and when it misleads¶
Its strength is that it anchors the whole intervention in what people actually do with a term, so later boundary decisions rest on evidence rather than on the governing team's assumptions about "what everyone knows it means." It is the cheapest way to discover that an apparent consensus is really six communities using one word for six things.
Its central failure mode is sampling bias: a convenience sample — whatever usage was easiest to grab — quietly overweights the loudest or most accessible community and mistakes its sense for the sense.[n1] A tidy audit built on a skewed frame produces confident, wrong conclusions about where meaning lives, and it goes stale as usage moves on. The guarding discipline is to fix the sample frame before collecting, deliberately reach into low-visibility contexts, and date the audit so its shelf life is honest rather than assumed permanent.
How it implements the components¶
Context Sampling Audit fills the discovery-and-recording front of the archetype — the components that establish what is traveling and where it currently lives:
focal_signifier_or_practice_record— the audit names the focal term and records its range of active source senses drawn from real instances.meaning_shift_profile— coding instances by sense reveals the narrowing, widening, and bleaching already underway across the sampled contexts.context_and_community_map— tagging each sense to the community and channel that holds it produces the first map of who means what.
It does not trace the pathway between those contexts (Propagation Pathway Graph, propagation_pathway_trace) or set any transfer boundary rules (Channel-Specific Scope Note, selective_transfer_boundary_rules); the ongoing uptake feedback loop belongs to Uptake Sentinel Monitoring.
Related¶
- Instantiates: Contextual Selective Propagation — the audit supplies the grounded usage picture the rest of the intervention builds on.
- Sibling mechanisms: Sense-Boundary Comparison Table · Propagation Pathway Graph · Channel-Specific Scope Note · Bridge-Context Interview · Uptake Sentinel Monitoring · Meaning Version Snapshot · Semantic Quarantine Marker
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Samples real usage across relevant communities, channels, documents, and situations to reveal which meanings are actually active and where transfer risk lives, making its operative form a bounded evaluation of existing evidence or work that produces a finding or disposition.
Independent corroboration: The frozen evidence defines Context Sampling Audit as 'Samples real usage across relevant communities, channels, documents, and situations to reveal which meanings are actually active and where transfer risk lives', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Linguistics & Semiotics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Sociolinguistics cohered empirical sampling of actual usage across communities and situations to map live senses rather than assume a canonical meaning.
Related originating lineages:
- Ethnography & Qualitative Methods — Field-based collection preserves community, channel, and situational context for each use.
- Statistics & Experimental Design — A prespecified sampling frame guards against convenience samples overrepresenting visible groups.
Review resolution: Sociolinguistic usage sampling is primary, with ethnographic context preservation and statistical sampling frames constitutive; the exact semantic audit is specialized and synthesized.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] Convenience sampling draws cases because they are easy to reach rather than because they represent the population. In a usage audit it systematically over-samples the most visible or vocal community, so the sense that dominates the sample may not be the sense that dominates the field — the standard corrective is a pre-specified sampling frame that forces coverage of low-visibility contexts. ↩