From the room, last lesson
Six sentences produced during the cross-examination. Every one of them was heard, understood, and left alone.
That is the problem this lesson is about.
Six sentences. Two are grammar; four are strength. Repair both kinds.
1. She mustn’t have known about the meeting.
2. He must be avoiding me — he hasn’t replied since Thursday. — four days is not elimination
3. They must been waiting outside for an hour.
4. Maybe she must be tired.
5. It might not have been posted — the box was empty on Monday. — that settles it
6. He must have forgotten on purpose. — a conclusion about a motive, from nothing
The finding
Errors that sound like opinions
Say I have went and somebody will correct you, or at least visibly notice. Say she must have forgotten when you mean she might have forgotten, and nothing happens at all. The sentence is well formed. It is idiomatic. It is exactly what a native speaker would say — if a native speaker believed it.
What has actually happened is that you made an accusation, and the room received one, and the conversation moved on carrying a small dent you will never hear about. Nobody corrects an opinion. They just adjust their picture of you slightly, and of the person you were talking about.
Two things make this systematic rather than occasional. The first is that Portuguese dever — deve estar cansada, deve ter esquecido — sits lower on the scale than English must. It is closer to "probably". So the obvious translation lands one full step above where you meant to be, and it lands there every time, invisibly.
The second is that Portuguese carries much of this in adverbs — talvez, certamente, com certeza — while English fixes the strength in the modal itself. That is why maybe she must be tired appears: the adverb is doing the work the speaker expects it to do, and the modal has been left to look after itself.
The consequence is unusual for a grammar point. This is not accuracy for a test. It is accuracy about other people — how much you are claiming to know about someone from what you can see of them. Which is also, in the end, what this unit is about.
Three questions.
1. Why does nobody correct a modal that is one step too strong?
2. What is the effect of translating deve as must?
3. Where does English fix the strength of a claim?
Three habits, and the two traps
What to do instead
| Habit | What you do | Why it works |
|---|---|---|
| 1 · Ask it of yourself | Before must leaves your mouth: what rules out the alternatives? | If you cannot name the thing that eliminates, you do not have a must. You have a feeling about likelihood, and English has a different word for that. |
| 2 · Default down one | Genuinely unsure between must and might? Say might. | Underclaiming costs you the appearance of confidence. Overclaiming costs you the benefit of the doubt you were extending to somebody. The prices are not equal. |
| 3 · Attach the evidence | Say the because in the same breath as the claim. | A claim with its evidence attached checks itself. You hear the gap as you say it, which is the only correction that arrives in time. |
The two traps, one last time
| Trap | What comes out | What it should be |
|---|---|---|
| mustn’t for negative deduction | She can’t be at home. | |
| mustn’t in the past | They can’t have seen it. | |
| Adverb doing the modal’s job | She might be tired. | |
| have lost in the /əv/ | He must have been waiting. |
Eight pieces of evidence. Match each to the strongest claim it will actually carry — no stronger.
Four questions.
1. You cannot name what rules out the alternatives. What do you say?
2. Which costs more: sounding less certain than you are, or more certain than you are?
3. What is the point of saying the because out loud?
4. Which evidence can buy a must?
Downgrading, on purpose
Two to four words per gap. Ask the challenge question at every gap before you write.
He hasn’t answered since Thursday, so he the message — or he busy, which is what I would assume about anybody else. His badge shows no entry on Friday, so he in the building that day. His calendar had the meeting in it, so he about it at some point. He looked away when I mentioned the report, which anything at all, although it that he already knew.
Eight sentences. Seven need repair; one is already right, and it is the one most people change.
1. She mustn’t have got my email.
2. Maybe he must be stuck in traffic.
3. They must been talking about it before I came in.
4. He must be angry with me — he was quiet at lunch.
5. She can’t be in the building — her badge hasn’t been used today. — correct as written
6. It might not have been him — he was in another country.
7. He must to have known about it.
8. She must have ignored it deliberately. — you have no evidence about intention
The recorded minute
Task card · Speaking + analysis · Pairs · 10 minutes
Speak for sixty seconds speculating about one of: the person who lived in your flat before you, the owner of the desk in Lesson 1, or whoever returned the library book.
Sixty seconds of talk, then four minutes of grading, then swap.
- Speaker: sixty seconds, no stopping. Say it badly rather than stopping to choose a modal.
- Listener: write every claim, with its modal, exactly as said.
- Together, put the challenge question to each claim. Downgrade every one that cannot answer it.
- Count the downgrades. That number is what you carry into the written piece.
One thing to hold on to while you do this. The subject of every claim here is a person, and the evidence is whatever they happened to leave behind. That is a very small window, and the difference between must and might is the difference between deciding who somebody is and wondering.
The downgrade count
Two questions.
1. Most of your downgrades were claims about…
2. What will you do differently in the next sixty seconds of speculation?
Accuracy about people
Two questions. Twenty seconds.
1. Why is this the one grammar point nobody corrects for you?
2. What kind of evidence buys the top of the scale?
Next. Two hundred and fifty words about a stranger — written slowly enough that every modal can be chosen, and defended.