TOEFL Build a Sentence 2026: The 5-Step Method, 8 Grammar Traps, and 12 Practice Examples
Build a Sentence is the first task you see in the redesigned TOEFL Writing section, and the easiest place on the whole test to leak band points without realizing it. You drag five to seven word chunks into a grammatical sentence, you have about 45 seconds per item, and ten items together pile into your Writing band. The catch: scoring is binary. One wrong placement and that item scores zero, no partial credit, no second look. This guide gives you the 5-step decoding method, the eight grammar traps ETS hides in the chunk pool, a 7-minute pacing plan, and twelve fully worked practice examples that mirror the real item difficulty curve.
1. What the Build a Sentence task actually tests
The task is the opening segment of the Writing section in the January 2026 TOEFL iBT redesign. You see a short context sentence (sometimes a question), then a row of five to seven clickable word or phrase chunks. Your job is to drag those chunks into a single grammatical English sentence that fits the context.
Roughly 10 items run for about 7 minutes total, which works out to around 45 seconds per sentence. A subset of items sneak in one extra chunk that does not belong — a decoy. The decoy almost always looks plausible at first glance: a near-synonym, an alternate verb tense, or a phrase that fits one slot but creates a grammatical clash elsewhere. Reading every chunk before placing the first one is the single most effective habit you can build.
For the wider context of how this task fits in the new test, see our complete TOEFL iBT 2026 format changes guide.
2. How the binary 1/0 scoring works
Build a Sentence does not use the holistic 1.0 to 6.0 rubric the rest of the writing section uses. Each item is scored 1 point for a perfectly correct sentence and 0 points for any error — a missing word, a wrong order, a misplaced modifier, anything. Those 10 raw points feed into the overall Writing band on the 1-6 band scale alongside the Email and Academic Discussion tasks.
| Raw score (out of 10) | Approximate band contribution | What it signals |
|---|---|---|
| 9-10 | Pushes Writing band toward 5.5-6.0 | Solid grammar foundation. Errors only on traps you have not seen before. |
| 7-8 | Supports a Writing band of 4.5-5.5 | Good control of common patterns; weakness on inversion, conditionals, or relative clauses. |
| 5-6 | Caps Writing band at around 4.0-4.5 | Pacing is fine but pattern recognition is shallow; missing decoys regularly. |
| 3-4 | Caps Writing band at 3.0-3.5 | Foundational grammar gaps. Typical when writers can produce English but cannot edit it under time pressure. |
| 0-2 | Caps Writing band at 1.0-2.5 | Word order itself is unstable. Drill basics before attempting the full task. |
Two points to keep in mind. First, the gap between 7/10 and 9/10 is mostly trap recognition, which is trainable inside two weeks. Second, because the rest of the Writing section depends on holistic 1-6 scoring, a strong Build a Sentence performance gives you a buffer when your Email or Academic Discussion response slips a half-band on test day.
3. The 5-step decoding method
Strong test takers do not assemble these sentences left to right. They decode the structure first, then snap the chunks into place. Use these five steps in order on every item:
-
1
Read every chunk first. Do not start placing. A 5-second scan of all chunks reveals the verb, the subject, and any decoy. Without this scan you will commit to a wrong skeleton and waste 30 seconds rebuilding.
-
2
Lock the verb. The conjugated main verb anchors the sentence. Find it first. If you see two candidates, one is in a subordinate clause or one is a participle. Decide which is the head verb before placing anything.
-
3
Match subject to verb. Check number agreement. "The students" pairs with "have" or "are," not "has" or "is." If a chunk forces a singular agreement, it tells you the subject. ETS uses this to mark sentences quickly.
-
4
Place modifiers last. Adverbs, prepositional phrases, and relative clauses go where they modify. Do not snap them in by feel — ask which word they describe and place them adjacent. Floating modifiers are the single biggest source of zero-point items.
-
5
Read it aloud silently. Once chunks are placed, mouth the sentence in your head once. If anything sounds clunky, suspect a misplaced modifier or a missed decoy. A sentence that "sounds wrong" is almost always wrong.
4. The 8 grammar traps ETS hides in chunks
After analyzing dozens of recent Build a Sentence items, eight trap patterns dominate. Internalize the list and your accuracy on hard items jumps from roughly 60 percent to over 90 percent.
1. Subject-verb agreement across an interrupting phrase
A long noun phrase between subject and verb hides the real subject. "The collection of artifacts on the upper floors is..." not are. The chunks make both forms available; only one is correct.
2. Misplaced relative clause
"Who" / "which" / "that" must sit immediately after the noun they modify. A chunk like "which the professor edited" has to land next to a manuscript or a paper, not next to the year or the publisher.
3. Adverb placement (especially "only," "even," "almost")
"Only Anna spoke at the meeting" and "Anna only spoke at the meeting" mean different things. ETS sets up contexts where one position is correct and the other creates an absurd reading.
4. Inversion after negative adverbials
"Rarely does he arrive on time" requires inversion. "Rarely he does arrive on time" is wrong. If a chunk starts with rarely, seldom, not until, or never, expect inversion in the chunks that follow.
5. Conditional tense matching
"If she had known, she would have called" — both halves shift one tense back. Mixed tenses are a near-guaranteed wrong sentence. When you see if, find both verb chunks before placing either.
6. Gerund vs infinitive after verbs
"Avoid going" is correct; "avoid to go" is wrong. "Decide to go" is correct; "decide going" is wrong. Each verb has a fixed pattern. The decoy chunk usually offers the wrong form alongside the right one.
7. Article use with countable / uncountable nouns
"Information" takes no article and no plural; "data" is treated as plural in academic English; "research" is uncountable. A chunk like "an information" or "researches" is a deliberate decoy.
8. Comparative and parallel structure
"More effective than..." needs a parallel noun phrase or clause on both sides. "She prefers reading to write" is wrong; "She prefers reading to writing" is correct. Watch the form of the second item.
5. The 7-minute pacing plan
Ten items in 7 minutes works out to 42 seconds per item, but the items vary in difficulty. Use this allocation under the global timer:
- 1 First sweep: 4 minutes. Move quickly through items that look easy. Aim for 25 seconds each on the first 6 to 7 items. If an item triggers a trap-pattern flag, leave it for the second sweep.
- 2 Second sweep: 2 minutes. Tackle the harder items you flagged. Use the full 5-step method on each. Aim for accuracy, not speed.
- 3 Final sweep: 1 minute. Re-read each completed sentence in your head. Catch the one you almost certainly got wrong on the first sweep — there will be one.
Note: the test uses one global timer for the whole 10-item block, so you can spend more time on hard items as long as you protect the total. Once you click Next on an item, you cannot return.
6. Practice 1-6: Foundational patterns
Each example below shows the chunks as they would appear on screen, then the correct sentence with the working chunks in brackets, then a short explanation of the trap or pattern. Try each before reading the answer.
Cleaner reading: "The paintings on display in the gallery are until next month."
7. Practice 7-12: Trap-heavy items
The next six items mirror the difficulty of the harder half of the on-screen block. Each contains either a decoy chunk or a layered trap. The decoy chunks are highlighted in amber.
Decoy chunk: "postponing" (not used).
Decoy chunk: "to write" (not used).
Decoy chunk: "the audience reacted" (not used; would create non-inverted form).
Decoy chunk: "have reached" (treats "the board" as plural, which is wrong in standard American English).
8. Common mistakes that cost points
Placing chunks before reading all of them
The single biggest accuracy killer. You commit to a wrong skeleton based on the first three chunks, then have to rebuild when chunk five contradicts you. Always scan all chunks before placing.
Ignoring the decoy
If the item gives you 5 chunks for a sentence that could be made with 4, do not assume all 5 must be used. Some items deliberately have an extra. Build the most natural sentence first, then test whether removing or swapping a chunk improves it.
Trusting "sounds right" without checking the verb
English learners often tune their ear to common patterns and miss formal grammar requirements like inversion or article rules. The 5-step method forces you to check the verb explicitly, which catches these.
Chasing perfection on item 1
Spending 90 seconds on the first item to feel confident burns time you need for harder items later. The block is scored on accuracy across 10 items, not on how clean your easy items felt.
Skipping the read-aloud check
A 3-second silent read at the end catches roughly one error in five for most students. That is one extra band point in expectation across the block.
9. A 14-day Build a Sentence study plan
Build a Sentence rewards repetitions over theory. Two weeks of focused practice moves most students from a 5/10 baseline to 8/10 or higher.
Days 1-3: Pattern audit
Work through 30 untimed practice items. After each, identify which of the 8 traps it tested. Build a personal list of which two or three traps you miss most.
Days 4-7: Focused drills
Do 10 items per day, but only on the patterns you missed in days 1-3. Speed does not matter yet; pattern accuracy does. Aim for 9/10 on focused sets.
Days 8-11: Timed full blocks
Run a full 10-item block under the 7-minute timer once daily. Score yourself. Review every miss and tag it to one of the 8 traps. Track your block scores.
Days 12-14: Test simulation
Put Build a Sentence inside a full Writing-section simulation back-to-back with Email and Academic Discussion. The goal is preserving accuracy under fatigue, not just under time.
You can run the timed blocks and the full Writing simulation inside our free TOEFL writing practice tests, which include the Build a Sentence interface, the 7-minute timer, and item-level scoring.
10. FAQ
How many Build a Sentence questions are on the TOEFL 2026?
Approximately 10 sentence-reordering items appear in roughly 7 minutes, the opening segment of the Writing section. Each item gives you 5 to 7 clickable word chunks plus, occasionally, one decoy chunk that should not be used. You have about 45 seconds per sentence on average, but the timer is for the whole block, so you can spend longer on hard ones.
How is the TOEFL Build a Sentence task scored?
Scoring is binary: 1 point for a perfectly correct sentence, 0 points for any error. There is no partial credit. The 10 items together feed into your overall Writing band on the 1.0 to 6.0 scale, which means a strong Build a Sentence performance can lift your writing band by 0.5 points or more even if your Email or Academic Discussion responses are uneven.
Are there decoy words in the Build a Sentence task?
Sometimes. A subset of items includes one extra chunk that does not belong in the correct sentence. The decoy is usually a near-synonym or a word that fits grammatically but creates a meaning that contradicts the rest of the sentence. Read every chunk before you place the first one so you can spot the decoy early.
Is there typing required on Build a Sentence?
No. The interface is click-and-drag. You select chunks in the desired order or drag them into a sentence frame. There is no spelling, no typing, and no autocorrect. That makes the task purely a test of grammar and word order, not vocabulary recall or keyboard speed.
Can I change my answer after placing the chunks?
Yes, within the time limit and within the same item. You can rearrange the chunks as many times as you like before submitting. However, once you move to the next sentence you cannot return to a previous item, so always confirm the chunk order reads aloud as a clean sentence before clicking Next.
What grammar should I review for the Build a Sentence task?
The most heavily tested patterns are subject-verb agreement, word order in subordinate clauses, placement of adverbs and prepositional phrases, conditionals, relative clauses, gerund versus infinitive after verbs, and inversion after negative adverbials. Working through 50 practice items is more useful than reading grammar rules in the abstract, since the task is about pattern recognition under time pressure.
Build a Sentence rewards trained eyes more than memorized rules. Two weeks of disciplined practice on the eight trap patterns is enough to push most test takers from average to near-perfect on this block, and that gain alone can lift the overall Writing band by half a point. Once you have it down, lock the gain in by stitching it together with the longer writing tasks: see our Write an Email template guide, our Academic Discussion guide, and the full TOEFL Writing Section 2026 overview.
Drill Build a Sentence with real 2026 items
Our free TOEFLMock writing tests include a 10-item Build a Sentence block with the official 7-minute timer and item-level scoring against the 1-6 rubric.
Start Free Writing PracticeTest preparation specialist and former classroom instructor. Designs full-length mock content aligned to the 2026 ETS redesign and writes study-plan, format, and score-requirement guides.