A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
Speech coherence depends on matching information structure to listeners' cognitive processing capacity and background knowledge. Listeners cannot re-read or pause; coherence must account for working memory limits, the need for explicit connection-making, and the role of prior knowledge in understanding.
Design speeches on the same topic for different audience knowledge levels and test them with actual listeners; ask them to recall main points and to identify where they lost coherence. Compare your understanding of a complex spoken explanation to how you process written explanation of the same topic.
From discourse coherence you know that spoken language needs to flow — ideas should connect, transitions should signal direction, and the listener should never be left wondering how one statement relates to the previous one. From working memory research you know that listeners can hold only a small amount of unprocessed information in mind at any moment. Cognitive coherence is what results when you design speech that satisfies both constraints simultaneously: the discourse logic is sound, and the demands placed on working memory never exceed what listeners can actually process in real time.
The key difference between spoken and written language is that listeners cannot rewind. A reader who encounters a dense sentence can slow down, re-read, and hold the opening clause in memory while parsing the subordinate structures that follow. A listener who falls behind cannot pause the speaker. This means syntactic complexity has a direct processing cost in speech that it lacks on the page. A sentence that reads clearly may be impossible to follow when spoken aloud — not because it's grammatically wrong but because the time between a subject and its verb is too long, or because an embedding structure requires holding too many clauses open simultaneously.
Cognitive coherence requires several design choices that written prose doesn't always need. Given-new sequencing — starting a sentence with familiar information before introducing new information — lets listeners anchor the new content onto an existing mental representation rather than floating it without context. Explicit connectives ("and so," "but here's the catch," "this means that") do the logical linking work out loud, rather than leaving the inference implicit as writing can sometimes do. Topical repetition — returning to a keyword or phrase — refreshes the listener's mental topic anchor, compensating for the fact that they can't glance back at the beginning of a paragraph.
Audience knowledge matters enormously to cognitive coherence, in a way that differs from written communication. A speaker addressing novices must build each concept explicitly before using it; jargon that experts process automatically becomes an opaque roadblock for non-experts. The speaker must maintain a mental model of what the audience already knows and sequence information to match that model. When a speaker pitches content at the wrong knowledge level — either assuming too much or explaining too little — coherence collapses not because the logic is broken but because the audience lacks the prerequisites to connect the links the speaker is drawing. Designing for cognitive coherence means designing for a specific listener's mind, not just for a generic ideal of clarity.