How to Build an Adult ESL Speaking Rubric and Prompt Cards
ESL speaking prompts invent CEFR levels, fake fluency scores, and children's topics for adults who need to talk to a landlord. The useful card is a rubric you can defend plus prompt cards that match the level you actually named, with no made-up scores and no children's cartoon themes unless you asked for them.
The matching generator is the Adult ESL Speaking Rubric and Prompt Cards prompt. Browse related cards in the PromptDig library (Browse more prompts). When a filled run survives, share the version you actually use (Share a prompt).
Level from Inputs, not from a guessed exam
Inputs: class setting (community, workplace, college ESL), claimed CEFR or other level if you have it, goals (phone a clinic, job interview small talk, landlord repair), constraints (no homework printer, 10-minute stations, mixed literacy), forbidden (fake scores, children's topics if you said adult, medical advice, immigration legal advice). If CEFR is unnamed, write unknown and describe the level in plain teacher language: can hold a 1-minute turn with support, or cannot.
Do not assign a B2 because the topic is "professional." Do not print a 92 percent speaking score. The rubric is criteria, not a number the model made up.
Rubric rows should be observable: intelligibility, task completion, turn-taking, vocabulary on the topic, self-repair. Scale in words (emerging / developing / meets) or in points you specified. If you did not specify points, do not invent a 20-point analytic table that looks official.
Prompt cards adults can actually use
Each card: a situation, a role, a goal, two supports (sentence frames or a word bank from Inputs), a time box, a note on what success looks like using the rubric. Topics from Goals only. If the class needs clinic phone calls, do not add a vacation-planning card to look fun.
Frames are optional language, not a script to read. Adults who can already say the line should be allowed to drop the frame. The card should say so.
Literacy mix: if some students do not read comfortably, the card needs a teacher-read version. Do not assume everyone can scan a paragraph of instructions.
Sets: six to ten cards is enough for a rotation. More than that clones. Vary the goal (ask, clarify, refuse, confirm) not just the noun.
Not-advice, gaps, and what the rubric will not do
Clinic and housing topics can be practiced as language. They are not medical or legal advice. Add a one-line banner on those cards. De-identify any real names from a placement interview you paste.
Gaps: missing level, missing minutes per station, missing whether recording is allowed, missing first-language mix, missing whether you want pair or teacher-led. Thin Inputs get a three-row rubric, four cards, and a long gaps list.
This is not a placement test, not TOEFL, not a scored official CEFR exam. If you need a placement tool, say so and still refuse a fake band.
Fill the card, then run
Name the adult goals. If you do not have CEFR, say unknown. Ask for rubric plus cards, not a score report.
Goal: [Speaking rubric and adult prompt cards]
Spec / paste: [Setting, goals, time box]
Stack versions: [CEFR or unknown]
Constraints: [Literacy mix, no printer, station length]
Forbidden: [Fake scores, child topics, medical or legal advice]
Acceptance checks: [Observable rows, adult situations, not-advice banner]
When a run leaves CEFR unknown, writes landlord-repair cards with a not-advice line, and refuses a numeric fluency score, share that filled card. The next class will try to ask for a fake B2 anyway.
Print the cards at the size you can actually copy. If there is no printer, keep each card under a short teacher-read paragraph. Re-run when goals change from clinic to interview. Do not keep the clinic not-advice banner on a job-talk card just because it was in the last deck.