ASSESSMENT · EVERY DISCIPLINE

Every student speaking — oral evidence you can actually grade

You ask a question out loud. The student answers out loud. You need that to happen for a whole class, in a form you can grade, without standing next to each kid. Six ways to do it with what we already run — nothing to buy, no new login for students.

Worked here in Mandarin 4, because that's where the question came from. The structure is the same for a theology seminar, a lab explanation, or a student defending a proof. Read it as a menu. Pick one or two. Nobody does six.

Students recording video or audio? They can record with Studio Capture, Screencast-o-Matic, Clipchamp, PowerPoint or Sound Recorder, and they turn it in through Studio so you keep your own copy. Step-by-step pictures and directions you can paste into your assignment: Every way to record → · One way to turn it in →

The whole page in one line: Options 1, 2 and 5 are presentational or interpretive — cheap, repeatable, and gameable. Options 3, 4 and 6 are interpersonal — expensive and impossible to script. Most courses collect only the first kind and then wonder why the grade doesn't match the student. You need some of both, and you do not need much of the second.

Not a language teacher? The translation is exact. A recorded explanation of a solution is presentational. A one-on-one where you ask “but why did you divide there?” is interpersonal — and it is the only one of the two that finds out whether the student understood or memorized. The three modes, from ACTFL.

START HERE

Why these six differ — it isn't the technology

All six use a microphone and Canvas. What separates them is who the student is talking to, and whether anyone can talk back. World languages has had a name for this distinction for thirty years, and it's the most useful thing on this page even if you don't teach a language.

Presentational

One-way, to an audience that isn't there. Rehearsable by design. A recorded answer, a summary, an artist's statement.

You get: their best, most-prepared work.
You don't get: what they can do without warning.

Interpretive

They take something in — your question, a text, a recording — and show they understood it. No negotiation; the source can't clarify.

You get: comprehension, cleanly separated from production.
You don't get: spontaneity.

Interpersonal

Two-way and unrehearsed. Meaning gets negotiated in real time — someone asks again, someone repairs, someone follows up.

You get: the truest picture, and the only unfakeable one.
You don't get: it cheaply. It costs live time.

The six, at a glance

They differ in one practical way that matters more than any feature: whether the time cost lands on your evening or on a class period. Options 1, 2, 3, 5 and 6 cost you grading time at home. Option 4 costs class minutes and no grading at all.

OPTION 5 · FOUNDATIONS

The Plain Assignment

Questions in the description, one recording back. Ten minutes to set up.

Costs: your evening
Mode: presentational
Do it to: prove the mics work

OPTION 1 · ENHANCE

The Weekly Discussion

One question a week. Everyone records. They hear each other.

Costs: ~5 min/week
Mode: presentational, edging interpersonal
Do it to: build the habit

OPTION 2 · DEEPEN

The Timed Prompt Video

Your voice asks five questions on a countdown. They answer in real time.

Costs: one build, reused all year
Mode: interpretive + presentational
Do it to: grade a checkpoint

OPTION 3 · INNOVATE

The Paired Conversation

Two students, three unscripted minutes, one recording between them.

Costs: gives time back
Mode: interpersonal
Do it to: hear real talk

OPTION 4 · DEEPEN

The Live One-on-One

Four minutes, one on one, in Teams. Recorded. Nothing to script.

Costs: class time, zero grading
Mode: interpersonal
Do it to: know what's real

OPTION 6 · DEEPEN

The Defense

They defend their own finished paper out loud. Six standing questions, one sitting.

Costs: 15 min to set up, then grading
Mode: interpersonal
Do it to: find out whose thinking it is

If you want a recommendation: run Option 5 this week to prove thirty microphones work, then Option 1 weekly, Option 2 at each unit checkpoint, and Option 4 twice a semester for the grade that counts. Option 3 goes wherever you want students talking to each other instead of to you. Any one of these is more oral evidence than most courses collect all year.

This is not a world languages page

World languages worked this out first because they had to — you cannot assess speaking on paper. Everyone else has the same problem and has been quietly accepting written proxies for it. Six rooms, six real moments. Find the one that sounds like yours; the option numbers above tell you how to run it.

English · Option 6 YOU ASK “What did you almost argue instead, and why didn’t you?” YOU HEAR “I was going to say the parties are about loneliness. But I couldn’t find anywhere he’s actually alone in a party scene — he’s always watching. So I changed it to him being a spectator at his own party.” WHAT THE PAPER DIDN’T TELL YOU The essay was a B. That answer is an A. She tested a claim against the text and abandoned it on evidence — and the finished paper had nowhere to show the one thing you most wanted to see.	Theology · Option 4 YOU ASK “State your position on the reading. Now give me the strongest objection to it — and answer that one.” YOU HEAR “I said suffering can have meaning. The strongest objection is… a child with cancer. And I don’t have an answer for that one. I think the meaning is what people do around them, not the suffering itself. But that’s not the same thing.” WHAT THE PAPER DIDN’T TELL YOU He is doing theology in real time and he knows exactly where his own argument thins out. A written reflection closes that gap with a tidy sentence. Reflection you can hear is rarer than reflection you can grade.	Social Studies · Option 3 YOU ASK “Sixty seconds arguing your side. Then sixty seconds giving the best version of the other side — the version its smartest defender would give.” YOU HEAR Fluent for sixty seconds on their own side. Then, on the other: “They… they just wanted to keep their power, I guess.” Stops at thirty. WHAT THE PAPER DIDN’T TELL YOU That gap is the whole assessment. A student who cannot voice the opposing case has picked a team, not understood a dispute — and on paper a one-sided essay and a well-understood argument look identical.
Mathematics · Option 4 YOU ASK “Walk me through it. You had x² = 5x and you divided both sides by x. Why was that allowed?” YOU HEAR “Because you can do the same thing to both sides. So x equals 5.” WHAT THE PAPER DIDN’T TELL YOU x = 0 is also a solution, and dividing by x threw it away. The written work said x = 5 and earned the mark. Forty seconds of talking showed a procedure learned without the rule underneath it — and that she will lose a root every time a problem is built to catch exactly this. You can teach it Tuesday.	Science · Option 5 YOU ASK “Before the lab: predict what happens, and why. After the lab: you were wrong. Account for it.” YOU HEAR “I said it would get warmer because dissolving releases energy. It got colder. So… breaking apart the salt takes more energy than the water gives back when it surrounds the ions. I had the direction backwards.” WHAT THE PAPER DIDN’T TELL YOU The lab report said results differed from the hypothesis due to experimental error. The recording is a time-stamped record of a mind changing. No lab report has ever captured that, because there is no box for it.	Fine Arts & CS · Option 4 YOU ASK “Explain what this function does — and why you wrote it this way instead of the obvious way.” YOU HEAR “It… loops through and checks each one.” Pause. “I’d have to look at it again.” WHAT THE PAPER DIDN’T TELL YOU The code ran and passed every test. Ninety seconds told you what no test suite can: whether the person submitting it could have written it. The same move works on an artist’s statement, spoken about the piece hanging in front of you.

Option 1 · The Weekly Discussion ENHANCE

A Canvas Discussion where the replies are recordings instead of typing. You post one question; every student records an answer; then they listen to each other. This is what Flipgrid used to do — Microsoft retired it and folded it into Teams, so don't go looking.

YOUR SIDE — ABOUT 5 MINUTES A WEEK

1. New Discussion. Record your question right in the post.

2. Check “Users must post before seeing replies.” This is the setting that makes it work — without it, the last student just repeats the first.

3. Post Monday, due Friday. Grade complete / not complete.

4. Leave a recorded reply on five or six — not thirty, not every week. Rotate.

STUDENT SIDE — ABOUT 3 MINUTES

1. Open it. They can't see anyone's answer yet.

2. Reply, record, speak 60–90 seconds, post.

3. The thread opens. They respond to one classmate — with a question, not “good job.”

Grade it light. This is practice, not assessment. Students can re-record all they want and you'll never know — that's fine, because volume is the point. The moment it becomes a weekly graded performance, they stop talking and start writing scripts.

Option 2 · The Timed Prompt Video DEEPEN · THE ONE THAT MATCHES A REAL ORAL EXAM

You build one short video that asks five questions with a visible countdown between them. Students play it on headphones and record themselves answering in real time. The countdown is the whole design. They can't pause and compose, so what you hear is what they can actually produce under pressure.

YOUR SIDE — ONE AFTERNOON, THEN REUSED ALL YEAR

1. Write three sets of five questions — three, not fifteen. Rotate them across the semester.

2. Build each in PowerPoint: one slide per question, a countdown animation, a chime. Record your voice over it, export to video. Swapping Set B for Set A later is a text edit, not a re-record.

3. Upload to Studio, embed in an assignment, attach the rubric.

4. Use Assign To to give different thirds different sets — or simply don't say which set it is until the window opens. That alone does most of the work.

5. Grade by scrubbing. Same question order in every file, so you can jump to question 4 across all thirty.

STUDENT SIDE — ABOUT 6 MINUTES, ONE SITTING

1. Headphones in. Start a recording of themselves.

2. Play your prompt video in another tab.

3. Before each answer they say aloud: “Set B, question three.” That one sentence means your voice never has to be in their file, and you always know what they're answering.

4. Stop, submit. No headphones? Play the prompt on their phone, record on the laptop.

Do not have students play your video on speakers hoping the mic catches both voices. Browsers cancel echo by default — the feature that stops you hearing yourself on a Teams call — and it will subtract your voice out of their recording. Headphones, or a second device.

And nothing stops a student re-recording. Canvas can't lock that down and we won't pretend it can. What stops scripting is that they don't know which five questions are coming, plus a rubric weighted on flow and follow-through — a read-aloud paragraph sounds nothing like a person thinking, and you will hear it in four seconds.

Option 3 · The Paired Conversation INNOVATE · THE ONLY UNSCRIPTABLE ONE

Two students talk to each other for three minutes and submit one recording between them. Neither knows what the other will say, so there is nothing to prepare. It is also the mode you're actually assessing when you say “respond to questions” — and it halves your grading pile.

YOUR SIDE — ABOUT 20 MINUTES TO SET UP

1. Make it a Canvas Group assignment with pairs. One upload covers both and both get the grade — otherwise you're sorting thirty files for fifteen conversations.

2. Give each pair a task card, not a topic: two roles and something to settle. “One of you is planning the weekend; one of you has no money.” A topic produces two monologues.

3. Require at least three questions each. That single rule stops one strong student carrying a quiet one.

4. You assign the partners. Left alone they pair by friendship.

STUDENT SIDE — ABOUT 10 MINUTES

1. Meet your partner — in class, or a scheduled Teams call.

2. Read the card. Two minutes to think. No writing it out.

3. Record three minutes. One of you uploads. Done for both.

What breaks it: pairing outside class hours. If they have to coordinate schedules alone, a third won't. Give them the last fifteen minutes of a period, or assign partners and require a Teams call — Teams at least supplies a calendar invite and a record button. Absences break pairs; keep a trio in your pocket.

Option 4 · The Live One-on-One DEEPEN · HIGHEST VALIDITY

Four minutes, one on one, in a recorded Teams meeting. The highest-validity oral assessment available to you, requiring no technology you don't already have. It's also the only one where you can follow up — “oh really? say more about that” — which is where you find out what a student actually has.

YOUR SIDE — TWO PERIODS, TWICE A SEMESTER

1. Post a sign-up. Four-minute slots, back to back.

2. The rest of the class works on something silent and independent while you go one at a time.

3. Open with the same three questions for everyone, then follow wherever it goes.

4. Score on the rubric while they talk. You are done grading when the last student walks out.

5. Keep the recording. Not for grading — it's what you show a parent who questions the grade.

STUDENT SIDE — FOUR MINUTES

1. Sign up for a slot.

2. Show up and talk. Nothing to install, nothing to upload, nothing to go wrong.

3. Tell them in advance it's recorded and why. It settles nerves and it's the right thing to do.

The trade is class time for grading time, and it's a good trade. Thirty students at four minutes is two periods. Twice a semester is four periods for the only two grades in the course that cannot be faked, produced by someone else, or generated. Most of us spend more than four periods on review days.

Option 5 · The Plain Assignment FOUNDATIONS · START HERE THIS WEEK

The ten-minute version. Questions typed or recorded in the description, one recording back. It has the weakest defense against scripting of anything here — and that is not what it's for. It exists so you find out this week whether thirty microphones work, before you spend an afternoon building Option 2.

YOUR SIDE — 10 MINUTES

1. New Assignment. Type or record five questions in the description.

2. Submission type: Media Recording for short audio, or File Upload if you want them going through Studio.

3. Grade in SpeedGrader, and leave your feedback as a recorded comment — twenty seconds of speaking beats a typed paragraph, and in a language course it's one more minute of the target language in their ear.

STUDENT SIDE — 5 MINUTES

1. Open the assignment, read the questions.

2. Hit record, answer all five, submit.

Option 6 · The Defense DEEPEN · ENGLISH, THEOLOGY, SOCIAL STUDIES — AND THE AI ANSWER

The student turns in the paper. Then, in one sitting, they defend it out loud for five minutes — thesis in their own words, why they cut the source they cut, where the argument is weakest, explain the paragraph on page three without reading it. The questions come from their own text, so there is nothing to prepare that isn't just… having written it.

This is the most useful idea on this page for anyone who doesn't teach a language, and it is the most honest answer we have to “did a machine write this?” — because it stops asking that question and asks a better one: can this student think about this work?

YOUR SIDE — 15 MINUTES, THEN IT'S JUST GRADING

1. Write six standing questions that work on any paper in the unit. They never change, students know them in advance, and knowing them doesn't help unless you did the work. Examples below.

2. Record them once with a countdown, same as Option 2 — or ask them live if you'd rather (that's Option 4 with a paper in front of you).

3. Open a ten-minute window for a five-minute recording. One sitting. In class with headphones, in the hallway, at a desk in the back.

4. Weight it from day one. The defense is a stated share of the assignment grade for everyone, always — not something you bolt on when you get suspicious.

SIX QUESTIONS THAT ONLY THE AUTHOR CAN ANSWER

1. What's your thesis, in one sentence, without looking?

2. What did you almost argue instead, and why didn't you?

3. Which source did you cut, and what was wrong with it?

4. Where is this weakest? What would a smart person attack?

5. Explain your third paragraph to someone who hasn't read it.

6. What changed in your thinking between the first draft and this one?

Read this part before you use it. A defense is not a lie detector.

Nervous students defend badly. Multilingual students defend badly in their second language. Anxious, introverted, and slow-processing students defend badly. A weak defense is evidence about speaking, not evidence about authorship — and treating it as proof of cheating will land on exactly the students least able to push back. If something genuinely doesn't add up, that is a conversation and the academic integrity process, never a grade you assign in the moment.

Here's why it works anyway. You are not catching anyone. You are moving where the grade lives. If a real share of the grade comes from talking about the work, then having a machine produce the work stops paying — the shortcut doesn't reach the part that counts. That is a design change, not a surveillance change, and it's the reason this holds up better than any detector. Detectors are never proof. A student explaining their own thinking is just… the assignment.

Pairs naturally with Make the writing visible — that page makes the process observable; this one makes the understanding audible. Together they're most of an answer.

Grow down the pathway

Same microphone at every rung. What climbs is not the technology — it's how much of the thinking the task actually requires, and who owns it. A teacher can sit at Foundations here and at Innovate somewhere else; that's normal and it's the honest description of most of us.

CANVAS LEVEL I

Foundations

Every student has spoken once, and you have it. One assignment, one recording, submitted and gradeable. The point is that the evidence exists at all — most courses never collect it.

Here: Option 5.

CANVAS LEVEL II

Enhance

It's regular, and they hear each other. Same slot every week, light structure, post-before-you-see-replies. Speaking stops being an event and becomes the ordinary texture of the course.

Here: Option 1.

CANVAS LEVEL III

Deepen

The design of the task produces the evidence. The countdown, the undisclosed question set, the follow-up you couldn't have predicted, the defense of your own paper. You built conditions that make understanding visible instead of hoping to spot it.

Here: Options 2, 4 and 6.

CANVAS LEVEL IV

Innovate

Students own it. They interview each other; they write the question sets; they keep a semester-long audio portfolio and argue, on the rubric, that they've moved. You are no longer the only person in the room who can hear growth.

Here: Option 3, and what it grows into.

The placement rule, same as everywhere in this course: rungs are depth of intent, not which feature. Recording a video is not Innovate because it's video. Handing students the microphone and the judgment is. If you want help placing something, bring one assignment and we'll put it on the ladder together.

What you need to be able to do — and where to learn it

Nothing here needs a workshop. Each rung adds one or two skills to the one below it, and every one of them is a ten-minute thing you can learn the afternoon you need it.

Foundations Option 5	Create an assignment and set its submission type · get a browser to allow the microphone · open SpeedGrader · leave a recorded comment instead of typing one.	Canvas Studio — the hub Students create with Studio — the four submission routes, side by side
Enhance Option 1	Everything above, plus: create a discussion and find “users must post before seeing replies” · record inside the reply editor · grade complete/incomplete so a weekly task doesn't become a weekly grading job.	Studio · Enhance — student video discussion (this is our Flip replacement)
Deepen Options 2, 4, 6	Plus: record a narrated slide show in PowerPoint and export it as video · upload to Studio and embed it · build or import a Canvas rubric · use Assign To to send different versions to different students · record a Teams meeting. The rubric is the real skill here — the rest is buttons.	Record a slide show with narration Turn your presentation into a video Extended time & accommodations — every way to do it
Innovate Option 3 and beyond	Plus: build Canvas Groups and a group assignment · design a task card that forces two-way talk · hand students the rubric and make them use it on themselves. The hard part stopped being technical two rungs ago.	Build groups that actually work The structure does the work — why the task design, not the grouping, decides whether they talk

Whichever you pick, these four apply

Record one yourself first

Submit it as a student before you assign it. Nearly every “the recorder doesn't work” email is one of two permissions: the browser's microphone prompt for our Canvas site, or Settings → Privacy & security → Microphone in Windows 11. Walk one class through it together, once, and the tickets stop.

Name a Plan B in the instructions

Every school laptop already has Sound Recorder. Record, save, upload. Put it in the assignment text with the folder spelled out — Documents → Sound Recordings — so a stuck student solves it at 10pm instead of emailing you. And leave file types unrestricted, or the .m4a it produces will bounce.

One rubric across all of them

Four rows, used for the weekly discussion, the checkpoint and the interview alike, so students are measured against one standard all year. Describe hesitation as a marker of level, never as a deduction — docking points for it penalizes anxiety, not thinking.

Build the accommodated version now

Anything timed needs a longer-countdown version — and Assign To delivers it quietly to just those students, with nobody else seeing a difference. Build it the same afternoon as the other three, not the week after the first student has to ask for it.

The reason this page exists right now

Speaking is the one form of evidence that cannot be delegated — not to a machine, not to a friend, not to last year's student. That was always true and it never mattered much, because written work was hard enough to fake that we didn't need the backup. It isn't anymore.

Which is why the move is not more detection. It's collecting one kind of evidence you were never collecting — and finding, as a side effect, that it was the more interesting kind all along. Nobody ever finished listening to thirty students explain their own thinking and wished they'd read thirty more paragraphs instead.

What we looked at and set aside

Studio video quizzes	They look like the obvious answer and they aren't — the embedded questions are multiple choice, true/false and multiple answer only. There is no way for a student to speak an answer inside one.
Recording inside a quiz	On paper this is exactly the right shape — question, answer, question, answer, one at a time. In practice the recorder inside a New Quizzes essay answer is inconsistent, and it stops working entirely when LockDown Browser is required. Don't build a graded assessment on a button that might not appear. Worth re-testing next year.
Starting in Classic Quizzes	Classic is steadier at that one trick, but New Quizzes native integration reached enforcement on 15 August 2026. Anything built in Classic now gets built twice.
Screen recording the prompt	Adds a tool to teach and a way to fail, and capturing computer audio is not a reliable option in Studio Capture. The “say the question number out loud” rule removes the need entirely.
Buying a speaking platform	Tools exist that enforce one-shot timed responses with no re-records, and they cost money and add a login for every student. Run Options 2, 4 and 6 for a semester first. If you can then name exactly what those didn't give you, that's a real case for a purchase — and a far better one than anyone could make today.

Where to read further

For the rubric — borrow, don't invent

Language teachers have spent forty years describing what speaking looks like at each level. The descriptors work for any subject once you swap the vocabulary — and they are free.

NCSSFL-ACTFL Can-Do Statements — what a student at each level can actually do, in student-facing language. The fastest route to a rubric students understand.

ACTFL Performance Descriptors — the teacher-facing version, and where the “hesitation is a level marker, not a deduction” principle comes from.

The three modes — the frame at the top of this page.

Already in this course

Canvas Studio — the mechanics behind every option here.

Students create with Studio — all four submission routes compared.

Build groups that actually work — who talks to whom, and why the structure decides it.

Extended time & accommodations — how to deliver the longer-countdown version.

Make the writing visible — the process-first companion to Option 6.

Tell us which one and we'll build it with you

Pick one or two. For whichever you choose, the materials come to you ready to use — your job is recording your own voice, not building anything.

The question sets
Three sets of five in your subject, each with a recommended countdown, plus the extended-time version.

The countdown deck
PowerPoint, timer and chime already built. Open it and record over it.

The rubric
Four rows, ready to import into Canvas, usable across every option on this page.

The student instructions
Question-number rule, headphone rule, Sound Recorder fallback. Paste straight into the assignment.

Or bring one assignment you already give and we'll work out what its spoken version would be. That conversation takes about fifteen minutes and it is usually the fastest way in.

THE RUBRIC IS BUILT — IT IS ALREADY IN THIS COURSE

Four rows — comprehensibility, task completion, flow, accuracy — at four levels, sitting in the course rubric library. Open it, use Copy To to drop it into your own course, then attach it to any speaking assignment. Every rating carries the rule that matters: hesitation is a level marker, not a deduction. Speaking — one standard all year →