What the blanks actually are
Three of the four Reading parts end with a second short text full of blanks, and those blanks are worth 13 of the 38 Reading marks — about a third of the section. People revise "reading comprehension" for weeks and never practise the thing carrying a third of the marks. The arithmetic:
| Part | Questions | Blanks |
|---|---|---|
| 1 — Correspondence | 11 | 5 |
| 2 — Apply a Diagram | 8 | 3 |
| 3 — Information | 9 | 0 |
| 4 — Viewpoints | 10 | 5 |
| Total | 38 | 13 |
Note Part 3: no second text at all. The part people brace for is the only one that is what they prepared for.
The failure: answering by ear
The blanks look like a grammar exercise and are not one. In a well-built blank every option fits the sentence and all are grammatical. Only one is consistent with what the main text said.
So picking the one that "sounds right" means choosing at random among the survivors of a test that eliminated nobody. That is answering by ear, and it is expensive because it feels confident. The fix: for each blank, find the line in the main text that rules options out. If nothing is ruled out, you have not found the line yet.
Three worked blanks
(original AlmiWorld practice items)
Part 1 — the friendly reply
From the main text, an email between neighbours:
I've booked the community hall kitchen for the Saturday, but only from noon onwards — the seniors' yoga group has it until then, and they finish right on twelve.
From the response you complete:
Count me in. I'll head over ______ so we can get the food started straight away.
- at nine — normal for a barbecue, but the kitchen is not free then. Ruled out by the main text, not by the sentence.
- at noon — correct. The earliest the kitchen is free, which "straight away" needs.
- the night before — fits the sentence, sounds helpful, contradicts "the Saturday".
- after the yoga class finishes at one — the skimmer's trap: it quotes the email's vocabulary back at you and gets the time wrong. They finish at twelve.
Part 2 — the email about a diagram
The main text is not prose but a diagram with point-form text and visuals:
| Session | Days | Time | Member fee | Non-member fee |
|---|---|---|---|---|
| Adult Lengths | Mon / Wed | 6:30–7:30 am | $48 | $72 |
| Aquafit | Tue / Thu | 7:00–8:00 pm | $60 | $90 |
| Family Swim | Sat | 1:00–3:00 pm | Free | $5 per visit |
From the email, written to someone who is not a centre member and cannot do mornings:
Aquafit looks like your best option, then. The session will cost you ______ in total.
- $60 — true of the diagram, false of the person. Aquafit does cost $60 — for members. He is not one. This catches almost everyone, because you can point at it in the table.
- $90 — correct. Non-member column, Aquafit row.
- $72 — a real number, but for Adult Lengths, a morning session.
- nothing — Family Swim is the free one, for members only.
The right cell is where the person's constraints intersect — not the one you find fastest.
Part 4 — the reader's comment
From the main text, an article on turning two lanes of Marchmont Road into a bike lane:
Councillor Renée Aubert argues that reallocating the road space would cut collisions at the three Marchmont intersections. Gerald Mwangi, who owns a hardware shop on the stretch, counters that removing kerbside parking would push his customers towards out-of-town retail parks.
From the comment, by a reader who has just said Aubert deserves more attention:
Her argument that ______ addresses the concern that actually keeps people off bicycles.
- reallocating road space would reduce collisions — correct. Aubert's argument, and about safety, which "keeps people off bicycles" points to.
- removing kerbside parking would drive customers to retail parks — grammatical, accurate, in the article. It is Mwangi's argument. Choosing it makes the commenter praise Aubert for her opponent's point.
- the consultation period was too short — could appear in any article about any proposal. Nobody here said it.
- cyclists should be licensed before gaining road space — a real opinion, but it opposes Aubert while the sentence praises her.
What the blanks are actually testing
Factual consistency. The blank names a time, a price, a place, a quantity. The main text fixes it, and every other option is a real fact from elsewhere, or a near-miss — noon versus one, $60 versus $90.
Logical connectors. The blank is so, however, because, even though. The wrong option reverses the relationship between two clauses that are both true. "The kitchen is free from noon, so I'll come at nine" is grammatical and absurd. Check which way the arrow points.
Stance consistency. Part 4's speciality. The comment holds a position and the blank must stay on that side — the right reason, attributed to the right person. An option that quietly switches you to the opposing viewpoint is the standard distractor.
Register: the second text inherits its part
- Part 1 — "conversational, social, friendly, familiar".
- Part 2 — "informative, helpful, not personal".
- Part 4 — "formal, complex: similar to a news article".
A word that is right in Part 1 can be wrong in Part 4. "That's a bit of a nightmare" belongs in a reply to a neighbour, nowhere near a council proposal — and "I remain unconvinced by the councillor's reasoning" is right in Part 4 and strange between friends. When two options survive on meaning, register is the tiebreak.
The shape that works
- Read the main text once for meaning before you touch the second text.
- Read the whole sentence around the blank — most are decided by the clause that follows.
- Ask which kind it is — fact, connector, or stance. That tells you what to look for.
- Eliminate one option by pointing at a line in the main text before choosing anything.
- If two survive, apply register.
- Answer every blank. There are no points deducted for wrong answers.
What actually costs you the mark
- Choosing the option that sounds most natural. They all do — that is the design.
- Treating the blanks as grammar. Grammar eliminates nothing here.
- Taking the fastest number to find in a Part 2 diagram instead of the one matching the person.
- Letting a Part 4 comment agree with the wrong opinion-holder because the option was really said.
- Overspending on the first half. The blanks sit at the end of the part, the test "will automatically move forward", and "you can't go back to a previous part."
- Leaving anything empty. Nothing is deducted, so a guess beats silence.
Try one
[Try Reading Part 1 →] — one full task, marked, no card required.