Typing every word the narrator says feels responsible and often blocks learning. Pause to encode claims and relationships — then resume — so notes become a schema, not a transcript.
Video lessons create a special note-taking trap. The audio keeps moving. Students try to keep up. Working memory splits between listening, reading on-screen text, and typing. John Sweller’s cognitive load theory predicts what happens next: extraneous load rises, generative processing falls, and you end with a document that looks complete and a mind that does not.
Richard Mayer’s multimedia principles sharpen the advice. When words and pictures are already competing, adding a stenography task is rarely the brave move. The brave move is segmenting: pause, produce a short generative capture, continue.
Notes during video should be encoding tools — not closed-captioning you maintain by hand.
Transcript vs generative notes
| Note style | Cognitive job | Later value |
|---|---|---|
| Verbatim typing while playing | Split attention | Re-read fluency; weak schema |
| Bullet dump of slide text | Copying | Recognition of the slide |
| Pause → claim + why + link | Organize / integrate | Teach-back friendly |
| Sketch of mechanism + 1 probe | Dual coding + retrieval setup | High |
Encoding quality over completeness
A half-page note you can teach from beats five pages you can only recognize. Completeness is a stenographer’s KPI. Understanding is yours.
The pause–encode–resume protocol
- Play until a major claim, definition, or diagram lands
- Pause (Mayer’s segmenting in student hands)
- Write one claim in your words
- Add why it matters or how it links to a prior idea
- Optional: one probe you might forget
- Resume
If you cannot write the claim without looking, rewatch that segment only, then encode. Do not restart the whole lecture to soothe anxiety.
What to capture (and what to ignore)
Capture:
- Causal chains and definitions you will need to produce
- Distinctions between near-twins
- Diagrams redrawn simply
- Misconceptions the lesson warns about
- Timestamp of anything still muddy
Ignore (usually):
- Jokes, transitions, and repeated slogans
- Decorative visuals
- Full sentences that merely duplicate the slide
Seductive details feel sticky and often steal the schema’s seat.
Live class constraint
If you cannot pause, use micro-captures: one claim every few minutes in the margin. After class, convert claims into probes before any full rewatch.
Dual channels without drowning
When a diagram is the point, stop typing prose and trace or redraw. Paivio’s dual coding idea helps only if the nonverbal trace is yours. Looking at a figure while transcribing narration often means neither channel gets generative work.
After the lesson, close the video and redraw from memory. That is the note’s exam.
Why verbatim typing fails multimedia lessons
Kenneth Kiewra’s note-taking research distinguishes functions of notes: encoding during the event, and an external store for later. Video students often maximize the store (pages of captions) while sabotaging encoding (split attention). Mayer’s redundancy and coherence principles warn that on-screen text plus narration already load verbal channels; adding your own stenography is a third verbal stream fighting for the same scarce capacity.
Laptop versus longhand debates (Mueller and Oppenheimer’s work is often cited here) matter less than what you produce. Verbatim laptop transcripts can be worse because speed enables mindless capture. Slow generative bullets — claim, why, link — force organization even on a keyboard.
Selection is not the whole job
Fiorella and Mayer’s generative learning model needs select → organize → integrate. Continuous typing mostly selects. Pausing to paraphrase and link is where organization and integration happen.
Timestamp as a metacognitive bookmark
A timestamp next to a muddy claim is a plan, not clutter. It lets you jump for targeted repair later — the only rewatch that usually pays — without restarting the whole player. Pair each timestamp with a probe you will attempt closed-book first. If the probe succeeds tomorrow, you may never need the clip.
Students who skip timestamps either rewatch everything (fluency tax) or avoid rewatch entirely (gap remains). The bookmark turns confusion into a scheduled act.
Seductive details and the “interesting” trap
Harp and Mayer’s work on seductive details shows that interesting but irrelevant asides can hurt retention of the main model. Video lessons are full of them: anecdotes, animations, jokes. Capture the causal spine. Leave the glitter. If a detail does not help you teach the claim, it is entertainment — enjoyable, optional for notes.
After the video: notes are not the finish line
Same day:
- Teach-back from notes alone
- Turn weak points into retrieval cues
- Schedule spaced returns
Later:
- Interleaved questions
- Papers that remix the ideas
Notes that never leave the page are storage theater.
Pair with smart revision notes
Capture during the lesson can stay messy. A short revision pass — organize, cut, map — converts capture into a durable artifact. Mixing those jobs mid-video is how transcripts happen.
The split-attention principle in student hands
Mayer and Chandler’s split-attention findings (and related CLT work) show that forcing learners to integrate spatially or temporally separated sources raises extraneous load. Video already splits eyes and ears across slides, captions, and voice. Your notes should not become a fourth competing stream. Pause so note-taking is sequential to listening — not parallel combat.
When you must note live, use a fixed template with tiny slots (claim / why / ?) so handwriting or typing stays short. Templates reduce decision load mid-stream.
Notes beside the lesson
Sukrat Lecture’s side workspace keeps notes, guide, and chat beside the slide — so you can pause, encode, and ask without a tab odyssey. Product page: /products/features/lecture
FAQ
Should I take notes on paper or screen?
Whichever you will use for teach-back and later retrieval setup. Medium matters less than the pause–encode habit.
What about provided slides?
Annotate relationships and questions; don’t retype the deck.
How many notes per 20-minute lesson?
Often less than you think — 8–15 high-quality lines plus one sketch beats a novella.
What if the video has no pause-friendly chapters?
Invent seams at claim boundaries anyway. Count major claims, not minutes. A forced pause every few claims beats continuous typing through an unchaptered file.
Should I note examples word-for-word?
Capture the structure of the example and invent a cousin later. Verbatim examples train recognition of that story, not transfer.
Related reading
- Checkpoints Beat Passive Watching
- Segmenting Principle
- Generative Learning During Lectures
- Encode, Don’t Transcribe
- Revision Notes vs Capture Notes
- From Lecture to Spaced Review
- Dual Coding
Keep going across Sukrat
Pause is part of note-taking
If the video never stops, your notes are often a race you lose on purpose. Segment. Encode the claim. Link it. Resume.
Aim for a schema you can teach — not a transcript you can highlight. Then move the weak points into spaced retrieval and transfer practice.
The point of notes is still in your head when the player is closed.