Understanding meaning does not guarantee inspecting form
When reading for meaning, attention naturally goes to the message. That is normally a good thing. But if the learning goal is to notice how a condition, hedge or clause is built, the useful structure can pass by without becoming an object of inspection. A carefully placed cue can create a brief invitation to look again.
The key word is carefully. The purpose is not to force the learner to consciously analyse every token. It is to make one relevant feature easier to locate while the complete sentence remains visible.
Textual enhancement has evidence, but the effect is not magical
A 2025 meta-analysis of textual enhancement studies reported a small-to-medium aggregate effect on L2 development across the included research. The result supports taking visual salience seriously, while also showing why simplistic claims are risky: effects varied, and making something visually obvious is not the same as guaranteeing durable learning.
For Metkagram, the responsible interpretation is a design rationale. Inline Marks may help direct attention to reviewed structure. Whether the particular Mark–Frame workflow improves a defined learning outcome compared with a suitable alternative remains a question for direct experiments.
A cue should lead somewhere
Highlighting on its own can become a dead end. The learner notices a coloured span, says “yes, I see it,” and moves on. Metkagram links the cue to a Frame, then to examples, a Move, Contrasts and retrieval. The visual moment is useful because it opens a path from perception to reuse.
This also changes how we judge annotation quality. A Mark is not successful because it looks elegant. It is successful as interface design when the learner can tell what it points to and use that observation to understand a reusable structure.
Less annotation can be more informative
An over-annotated sentence creates a new decoding task: first learn the label system, then recover the language hiding underneath it. Metkagram therefore treats annotation as selective support. Some examples need several cues; others need only one; future languages may start with Frames and no annotation at all.
A good practical question is: what do we want the learner to notice on this pass? If the answer is “everything,” the instructional target is probably not yet clear enough.