A delicate branching network of illuminated filaments and glass nodes.

Imagine reporting the same event in three ways: you saw rain falling, you noticed wet pavement and inferred rain, or a neighbor told you it rained. English can spell out those differences with phrases such as “I saw,” “it looks as if,” and “I was told.” In some languages, grammatical markers can make information source part of the sentence itself. Linguists call this category evidentiality. It helps explain what a sentence signals about how its speaker came to know something, though it does not reduce every language to one neat set of labels. [1] [2]

Source of information is not the same as certainty

A speaker can be highly confident in a report and uncertain about something personally observed. Evidentiality concerns the kind of access a speaker presents: sensory experience, inference, or information received from another person are common distinctions. Epistemic modality, by contrast, concerns a speaker’s evaluation of a claim, including its likelihood or certainty. These meanings can interact in actual sentences, but they are not interchangeable categories. [1]

English often expresses source with separate words or phrases: “I heard the road is closed,” “the road appears to be closed,” or simply “the road is closed.” The first wording attributes information to a report; the second suggests an inference. A language may grammaticalize some such distinctions with an affix, clitic, particle, or a form in its tense system. The World Atlas of Language Structures documents several structural strategies and also distinguishes languages with grammatical evidentials from those without them. Its typological sample is a comparison tool, not a complete census of every language. [2]

One proposition, different evidence

A useful way to see the idea is to hold the event constant and change the evidence. In descriptions of Cuzco Quechua, the sentence translated “It is raining” can be marked to convey direct evidence, a conjecture, or a report from another person. The forms often written =mi, =chá, and =si illustrate those distinctions in this variety. They do not mean that all Quechua varieties use an identical system, or that each marker maps perfectly onto a single English word. [3]

That last caution matters. A scholarly analysis of evidential marking in Quechua varieties describes overlapping meanings involving information source, inference, stance, and what is foregrounded in a conversation. It also reports that evidential marking is not obligatory in Cuzco Quechua in every sentence. Grammar descriptions are accounts of particular speech communities and data; they should not be turned into claims about what every speaker must say or believe. [3]

How languages package the distinction

A marker need not sit in the same place across languages. The WALS chapter documents evidentials expressed as verbal affixes or clitics, separate particles, and distinctions incorporated into tense or other grammatical forms. A familiar translation may therefore hide how the original sentence is built. Two forms translated with the same English past tense can differ in whether a speaker presents an event as directly known or indirectly learned. Conversely, a language without a dedicated grammatical evidential can still express source through vocabulary or a longer construction. [2]

For a translator or learner, the practical question is not “Which English word equals this marker?” but “What contrast is this form making here?” Check the language and variety, the sentence around the marker, and the source’s own examples. If a reference labels a form “direct,” treat that as a starting description, not proof that it always means eyewitness vision. Research on evidential systems has found that forms can carry additional discourse functions and resist one-to-one glosses. [3]

A short reading exercise

Try this with an English sentence: “The meeting has ended.” Write three versions that keep the event but change how the speaker knows it: “I was in the room when it ended”; “I can see the chairs are stacked, so it seems to have ended”; and “Mina told me it was over.” Then ask what changed. The event stays constant, while the route by which the speaker learned it becomes explicit. This is an invented exercise, not a translation of a Quechua sentence.

Now consult a grammar or learner resource for a language you study. Find one example sentence with an evidential marker and note the source’s label, the marker’s position, and whether the description says it is optional or required in that construction. Keep the example attached to its named language variety. If the resource gives only a broad English gloss, avoid inferring a sharper distinction than it documents. This habit is useful beyond linguistics: it keeps our explanations faithful to the evidence instead of turning a convenient label into a universal rule.

References & further reading

  1. SIL Global: Evidentiality ↗
  2. World Atlas of Language Structures: Coding of Evidentiality ↗
  3. Kalt, Acquisition, Loss and Innovation in Chuquisaca Quechua ↗

Our editorial approach.