SpeechSpectrum

The same speech, transcribed under different conventions.

Research Study

Participating in our Prolific study? Enter your ID, complete a short demographics survey, then read and evaluate a transcript.

Enter Study

Internal Testing

Run the full participant flow (consent → demographics → content acknowledgement → study) with an auto-generated TEST Prolific ID.

Test Random Condition Test Demo Condition

Study Design

Counterbalancing & Randomization

Participant enters Random assignment to 1 of 5 conditions (between-subjects) A: Verbatim B: Non-Verbatim C: Enhanced D: Bullet Points E: SpeechSpectrum Random assignment to 1 of 4 Latin-square orders (within-subjects) Order A Davis Bates Lorenze Feliciano Order B Bates Lorenze Feliciano Davis Order C Lorenze Feliciano Davis Bates Order D Feliciano Davis Bates Lorenze Complete

Condition assignment (between-subjects): Each participant is randomly assigned to one of 5 conditions — A: Verbatim, B: Non-Verbatim, C: Enhanced, D: Bullet Points, or E: SpeechSpectrum. The condition determines how transcripts are presented throughout the entire session.

Conversation order (within-subjects, 4×4 Latin square): All participants read all 4 parole hearings — Davis, Bates, Lorenze, and Feliciano — but the order is one of 4 fixed Latin-square permutations, randomly assigned. With 36 participants per condition and 4 orders, this gives 9 participants per (condition × order) cell.

Primary test: 5 (condition) × 4 (transcript) mixed ANOVA on comprehension accuracy, with the within-between interaction as the focal effect. 8 questions per conversation × 4 conversations = 32 items per participant.

Question order: Multiple-choice questions are reshuffled for each transcript. Free-response questions always appear at the end. Question order is logged per transcript.

What gets logged: Condition assignment, transcript order, question order, per-question timing, per-transcript timing, total session time, style switch events with timing (demo condition), highlighted text per question with style attribution, all survey responses, and demographic data.

Transcripts

View each transcript in a single style, or use the SpeechSpectrum demo to switch between all four.