The Gold Standard of Anthropology | Evidentials, Reported Knowledge and Source-of-Information Claims

1. Start with the source claim, not an English label

Back to contents

Evidentiality concerns how a speaker presents access to information: direct perception, inference, report, assumption or another locally organised source. Some languages grammatically require a choice; others rely more on verbs, adverbs, particles or constructions. Begin with the forms people actually use and the contrasts speakers recognise. Translating every marker as apparently, reportedly or I heard can erase distinctions in sensory channel, access, responsibility or social authority. Record the original expression, its position in the turn and the surrounding talk before assigning an analytic category.

2. Separate evidentiality from epistemic certainty

Back to contents

A source marker and a confidence judgement are related but not identical. A speaker may report hearsay with high confidence because the source is trusted, or describe something personally seen while expressing doubt about its interpretation. Do not automatically rank direct evidence above reported or inferred evidence. Ask what the form conventionally marks, what stance the speaker takes and how recipients respond. This distinction keeps an evidential analysis from collapsing into a simple strong-versus-weak certainty scale.

3. Map grammatical obligation and optional choice

Back to contents

Where an evidential category is obligatory, speakers may have to mark source even when it is not the conversational focus. Where marking is optional, choosing a source phrase can itself be socially noticeable. Describe the paradigm, the environments where forms occur and the alternatives that were available. Absence of an optional marker is not proof that source was irrelevant; obligatory marking is not proof that every use carries the same interactional emphasis. Grammar provides resources and constraints, while particular encounters give them social force.

4. Track whose knowledge is being represented

Back to contents

A reported claim contains at least a current speaker, a source and a proposition, and it may also contain a quoted or animated voice. Keep these roles separate. The current speaker can endorse, distance from or merely transmit another person’s words. Passive expressions such as it is said may background the source, while named attribution can assign responsibility. Follow pronouns, quotatives, reporting verbs and shifts in footing to identify who is presented as knowing, saying and answerable for the claim.

5. Read source marking inside epistemic relationships

Back to contents

Rights to know are uneven. A patient may have primary access to pain, an elder to a lineage history, a technician to an instrument reading and a witness to an event. An evidential form can respect those territories, challenge them or claim independent access. Analyse who can legitimately assert the information, who must cite a source and who is expected to accept or verify it. The same grammatical marker may sound careful in one relationship and evasive in another.

6. Worked example: a reported water warning

Back to contents

At a village meeting, a resident says that the upstream water is unsafe using a reported-evidence particle. One analyst codes the statement as low confidence. The next turns show something else: the resident names a monitoring officer, others recognise that officer’s role and the chair immediately changes the collection plan. The source marker does not weaken the claim; it locates institutional responsibility. When a later speaker repeats the warning without attribution, the chair asks who measured it. The evidential analysis succeeds by linking form, source, uptake and action rather than assigning a universal confidence score.

7. Follow evidentials across a sequence

Back to contents

The meaning of a source claim often becomes visible after it is challenged, repeated or reformulated. A recipient may ask Did you see it?, treat hearsay as sufficient, request a named source or supply a stronger basis. Code the first claim together with repair, agreement, disagreement and later retellings. Sequential evidence can show whether participants treated a marker as distancing, accountable reporting, inference or shared knowledge. A sentence isolated from its response can conceal the very social work the form performs.

8. Changed-context transfer check: a school incident

Back to contents

Move from a community meeting to a student reporting that a classmate damaged equipment. I saw it, she told me and apparently may distribute responsibility differently, but school roles also matter. A teacher may require firsthand access for discipline while accepting reported information to begin an inquiry. The transferable method identifies source, entitlement, uptake and consequence. It does not convert grammatical evidence into legal proof or assume that a child’s reported source is inherently unreliable.

9. Know the limits and common misconceptions

Back to contents

Evidentials do not mechanically reveal truthfulness, certainty or deception. Direct perception can be mistaken, inference can be excellent and hearsay can come from a highly accountable institution. English glosses may hide richer categories, and elicited sentences may not reproduce conversational use. Researchers can also overgeneralise from one genre or speaker. Treat typological labels as comparative tools, not replacements for local grammar and ethnography, and state when the evidence supports only a narrow claim about a particular interaction.

10. Audit transcription, translation and coding

Back to contents

Archive recordings where consent permits, time-aligned transcripts, original forms, morpheme-by-morpheme glosses, free translations and the coding guide. Retain who spoke, participation roles, preceding source talk, recipient response and later repair. Double-code ambiguous tokens, document disagreements and test whether results depend on treating a form as evidential, epistemic or quotative. Compare interviews, naturally occurring interaction and relevant grammatical descriptions so that translation choices remain inspectable.

11. Frequently asked questions

Back to contents

Is every report an evidential? Not necessarily; the analytic boundary depends on the language and framework. Does direct evidence mean true? No. Can a form mark both source and certainty? Yes, but the two dimensions should still be distinguished analytically. Why study recipient responses? They reveal how participants understood the source claim in practice. Can English express evidentiality? It has many lexical and constructional resources even though source marking is not obligatory in the same way as in some languages.

12. The gold standard keeps source, stance and authority apart

Back to contents

A strong analysis begins with linguistic form, identifies the source relationship and then follows how the claim is received. It distinguishes information access from confidence, reported voice from current commitment and grammatical category from social entitlement. The result is not a universal ladder of evidence. It is a documented account of how people make knowledge traceable, assign responsibility and negotiate who may speak for what in a particular language and setting.