No script on screen

Claim: words on screen do not repeat the voice: show key terms and the evidence, and say the rest.

What it asks

  • Put on the slide the key terms from what you say, not the sentences themselves.
  • Let the figure or table carry the evidence, and say the explanation aloud.
  • Keep the words short: about 20 to 40 words per slide, 80 at most (Beamer 2026, convention).
  • Do not read the slide to the room.

Why

Spoken plus written text equals written text alone. It beats spoken text alone only for low prior knowledge, system-paced and picture-free material. Key terms taken from the narration did better than verbatim text (Adesope 2012, 57 studies).

Identical on-screen text and narration hurt recall and transfer. A small discrepancy between text and voice did best. Learners preferred the identical text and learned less from it (Yue 2013, two experiments).

Talks already work this way: slides and speech run in parallel with different wording, and text slides are rephrased aloud (Rowley-Jolivet 2002 and 2012, secondary).

Limits

  • The edge over spoken text alone holds for low prior knowledge, system-paced and picture-free material. "Never put text on the slide" overreaches (Adesope 2012).
  • Most source lessons are short narrated computer lessons. No talk was studied in a live room.
  • Accessibility guides ask the speaker to say every slide word aloud (Rector 2019; ACM 2026). This runs against the redundancy result. The two have not been tested together, and no study of blind or low-vision listeners was found.
  • A deck read alone is self-paced. The held evidence was gathered under system pacing, so it does not bind that deck (inference, untested).
  • The word counts are conventions. Each comes from practice, not from a test.