Grammatical Range and Accuracy: The IELTS Grammar Criterion Decoded

IELTS Speaking and Writing are each marked against four criteria, and Grammatical Range and Accuracy is one of them, worth exactly 25% of that sub-score. The other three are Fluency and Coherence, Lexical Resource, and Pronunciation in Speaking; Task Achievement or Task Response, and Coherence and Cohesion, in Writing. GRA sits alongside them as an equal partner — it is never scored in isolation, and a weak GRA band pulls your overall Speaking or Writing score down even if your ideas and vocabulary are excellent.

The criterion blends two separate judgements into one band. 'Range' asks how many different structures you attempt — simple sentences only, or a mix of complex sentences, conditionals, passives, and relative clauses. 'Accuracy' asks how many of those attempts are error-free. Examiners hold both in mind at once: a candidate who writes flawless simple sentences all essay long is capped at a mid-band, and so is a candidate who reaches for ambitious grammar but gets most of it wrong. The two dimensions must rise together.

These are the exact wordings examiners are trained against, taken from the public IELTS band descriptors for Speaking, alongside the closely mirrored descriptors for Writing. Read down a single band row and the same pattern repeats: range language ('a mix of', 'a range of', 'a wide range of', 'a full range of') paired with accuracy language ('rarely cause comprehension problems', 'frequently produces error-free sentences', 'majority of error-free sentences', 'apart from slips'). Learning to hear your own English in these terms is the single most useful diagnostic skill you can build.

Band 5 speech and writing is recognisable by errors inside ordinary sentences — subject-verb agreement, missing or wrong articles, tense slips — alongside complex sentences that almost always break down. The jump to Band 6 does not require dramatically more ambitious grammar; it requires the simple sentences to become reliably correct first, because the descriptor language changes from 'reasonable accuracy' with errors that 'may cause some comprehension problems' to errors that 'rarely cause comprehension problems'.

Alongside that reliability, Band 6 expects a genuine 'mix' of simple and complex forms — so add a handful of complex sentences you can build correctly under pressure: a first conditional, a defining relative clause, a because-clause. You do not need every complex sentence to be perfect at this stage, only enough of them to show the mix examiners are listening and reading for.

• She go to the office by bus every morning.(basic subject-verb agreement error blocks Band 6)
• She goes to the office by bus every morning.(simple sentence made reliably accurate)
• I think education important for a country future.(missing verb and article inside the complex clause)
• I think that education is important for a country's future, because it creates opportunities.(one accurate complex sentence is enough to show range)

This is the jump most candidates get stuck on, because it asks for something more specific than 'more grammar' — it asks for structures that were already being attempted at Band 6 to start coming out correct more often than not. The descriptor moves from 'may make frequent mistakes with complex structures' to 'frequently produces error-free sentences, though some grammatical mistakes persist'. In other words: keep attempting the same range, but raise the success rate.

Four structures do most of the work at this stage: relative clauses, first-to-third conditionals, the passive voice, and concession language (although, despite, even though). Alongside them, one accuracy habit matters as much as any new structure — eliminating comma splices and run-on sentences, where two independent clauses are joined with only a comma. A single uncorrected comma splice in a Task 2 essay is exactly the kind of 'persisting mistake' that keeps a script at Band 6.

• The prices are high, people still buy the product.(comma splice — two independent clauses need more than a comma)
• Although the prices are high, people still buy the product.(concession clause resolves the splice and adds range)
• The bridge, which was completed in 2019, cut the journey time in half.(defining relative clause used accurately)

Band 8 asks for structures beyond the Band 7 toolkit, deployed flexibly rather than as fixed formulas, with the majority of sentences entirely error-free. Three structures reliably signal this: mixed conditionals (combining an unreal past condition with a present result, or vice versa), participle clauses that compress two clauses into one, and 'distancing' passives that report a claim without naming who makes it — 'It is widely believed that…', 'It has been argued that…'.

Cleft sentences ('What really drives the change is…') and mild inversion ('Rarely does a policy achieve both goals at once') start to appear naturally rather than as rehearsed set-pieces at this level. The test is flexibility: can you use the structure in a fresh sentence you have not practised, not just recite a memorised template.

• Had the city invested in public transport earlier, congestion would not be such a problem today.(mixed conditional — unreal past cause, present result)
• It is widely believed that remote work has improved employee wellbeing.(distancing passive — no need to name who believes it)
• Faced with rising costs, many small businesses have started raising their prices.(participle clause compresses a because-clause)

Band 9 does not mean zero errors — it means the rare error that remains looks like a slip a native speaker might make under pressure, not a systematic gap in knowledge. A self-corrected false start ('I think that — sorry, what I mean is…') or a single dropped preposition in ninety seconds of continuous speech does not cost a Band 9 candidate anything, because the descriptor explicitly allows for 'slips characteristic of native speaker speech'.

What separates 8 from 9 in practice is less about learning new grammar and more about removing the last patterns of hesitation before complex structures — using inversion, cleft sentences, and mixed conditionals as unremarkable, everyday choices rather than a special register reserved for the exam. At this level, grammar stops being something you are visibly building and becomes something you are simply using.

• Not until the final report was published did the true scale of the problem become clear.(inversion used unremarkably, mid-conversation)