A scenario that is not on the syllabus is a demo. Students are ruthless prioritizers: optional practice loses to graded everything, every week, no matter how good the practice is. This chapter is about writing role play into the course structure so it actually happens — and doing it in the order that builds trust instead of burning it.
The stakes ladder
The oldest advice in the role-play literature is also the most commonly violated in AI rollouts. Harbour and Connick, via the NIU guide: if you plan to use role play as a graded exercise, introduce small, non-graded role plays early and often first. Students meeting an unfamiliar format and a grade at the same time experience only the grade — and the format takes the blame for the anxiety.
Operationalized, the ladder has three rungs:
Rung 1: Ungraded, weeks 1–3. One short, low-difficulty scenario, framed explicitly as format familiarization. The goal is that every student has met the persona, heard their own transcript, and seen the feedback format before anything counts. Rung 1 is also your calibration window: transcripts from these weeks tell you whether the difficulty knobs from Chapter 11 are set right, while the cost of being wrong is zero.
Rung 2: Completion-graded, weeks 3–8. Sessions count for participation or completion — did you run it, did you attach the reflection — but the rubric score does not affect the grade. This is where the habit forms and where formative feedback does its work. Scores are visible to students (they want them), but the message is explicit: this window is for getting better, not for performing.
Rung 3: Score-graded, from mid-term. The best-attempt pattern from Chapter 12: unlimited private practice, submit the session that clears the bar, attach the comparative reflection memo. By now the format is familiar, the difficulty is calibrated, and the grade measures skill rather than novelty-shock. For courses running an AI viva (Chapter 9), this is when it lands — after weeks of the same examiner being available as a practice partner.
The ladder is one semester compressed. Programs that skip to rung 3 in week two generate the complaints that get pilots cancelled; the same assessment in week ten, after eight weeks of rungs 1 and 2, generates none. The difference is not the assessment. It is the ramp.
Assignment patterns
Four patterns cover nearly every course design:
Pre-class prep. The scenario runs before the session that depends on it: negotiate with the AI counterpart before the negotiation workshop, take the practice history before the standardized-patient day, spar before the Model UN (Chapter 7). The class then starts from attempt three instead of attempt zero, and expensive synchronous time — the actor session, the seminar — is spent performing rather than fumbling. This is the single highest-yield pattern, because it upgrades what the institution already pays most for.
The gated module. Chapter 14’s embedded step, formalized: reading, then a five-minute conversation applying it, then the quiz unlocks. The conversation is where the reading either becomes usable or reveals it didn’t land — a comprehension check that cannot be pattern-matched from the text.
Post-class remediation. Feedback loops that end with a target: students who miss a criterion on the graded attempt get a targeted drill on exactly that criterion, retry, and resubmit. The rubric’s per-criterion structure (Chapter 12) makes the prescription automatic: low score on open-ended questioning routes to the questioning drill, not to “practice more.”
The recurring ritual. For skill-heavy courses, a standing weekly slot: every Monday, ten minutes, this week’s scenario. Small, boring, and cumulative — the everboarding logic that sales teams use, applied to a semester. Rituals survive busy weeks; one-off assignments don’t.
Live where the coursework lives
Friction kills habits, and the biggest friction is a separate destination. If students manage their academic lives in Canvas, Moodle, or Blackboard, the role play should be a page in the module, not a link to elsewhere: embedded in the assignment, launched in place, no second login. Completion and scores should flow to where the instructor already looks for them.
The same logic governs communication: the scenario appears in the same weekly announcement, the same module checklist, the same rhythm as every other deliverable. The more the role play is administratively indistinguishable from a problem set, the faster it stops being an event and becomes coursework.
On Tough Tongue: scenarios embed via iframe in Canvas, Moodle, Blackboard, or any LMS that allows iframe embedding, so the session runs inside the module page. Session logs with timestamps, duration, and rubric scores export for the gradebook, and the API supports programmatic assignment and completion tracking for programs that automate the gated-module pattern.
Exercise
Take your current syllabus and place exactly three touchpoints on it: one rung-1 ungraded scenario in the first three weeks, one rung-2 completion-graded pattern (choose from the four above) in the middle stretch, and one rung-3 score-graded best-attempt submission after mid-term. Write each as a real syllabus line — title, week, deliverable, weight. If you cannot find room, remove one reading response; you will not miss it, and your students will not either.