IELTS Speaking Band Descriptors: What Grammatical Range Actually Costs You

A repeated mock result tells a very specific story: fluency and coherence have reached 7, lexical resource has reached 7, and grammatical range and accuracy stubbornly sits at 6. Nothing sounds broken. Nobody stops you to struggle. Yet the same column comes back with the same number, three mocks in a row.

The reason is usually not knowledge. It is that the grammar you use in the test is narrower than the grammar you know, and the ielts speaking band descriptors reward what the examiner can hear, not what you could produce if you were asked to write it. This guide reads the grammatical range and accuracy row the way a listener applies it, then works through six annotated answer excerpts, five structures worth rehearsing, the error patterns that hold a performance down, and a self-check you can run on your own recording tonight.

What the criterion measures, and what candidates assume

The Speaking test is a face-to-face interview that lasts 11-14 minutes across three parts, and it is recorded. The official test description states the criterion plainly: grammatical range and accuracy assesses the range of grammar you use and how accurately and appropriately you use it.

Two nouns, then. Range is counted across everything you said while answering real questions, so it is a property of the performance rather than of one lucky sentence. Accuracy is judged inside the structures you actually produced, which is why errors in a complex sentence cost more than the same slip in a short one: the listener has more to hold at once.

What candidates assume instead is that this is a rules column. They revise conditionals on paper, then play safe in the room with short, controlled sentences, and the recording sounds clean and small. The descriptor for band 6 already names the problem, because it asks for a mix of short and complex sentence forms and a variety of structures used with limited flexibility. A performance built from two sentence shapes cannot be flexible no matter how accurate it is.

There is one more rule that shapes everything below. The public descriptors carry two notes: a candidate must fully fit the positive features of the descriptor at a particular level, and a candidate is rated on average performance across all parts of the test. So the examiner is not awarding points for isolated good moments; the question is whether the whole performance fits a row.

Band 6, 7 and 8 in the descriptor's own words

The wording matters more than any summary of it, because the difference between the rows is small and lives in specific phrases. All three rows below are quoted from the official Speaking Band Descriptors.

Band	What the descriptor says about grammatical range and accuracy	What the examiner can hear
6	Produces a mix of short and complex sentence forms and a variety of structures with limited flexibility. Though errors frequently occur in complex structures, these rarely impede communication.	Two sentence shapes are present, the complex one is rare and repeated, and meaning survives the errors.
7	A range of structures flexibly used. Error-free sentences are frequent. Both simple and complex sentences are used effectively despite some errors. A few basic errors persist.	Several structures appear where they fit the question, most sentences arrive clean, and the errors that remain do not repeat in a pattern.
8	Wide range of structures, flexibly used. The majority of sentences are error free. Occasional inappropriacies and non-systematic errors occur. A few basic errors may persist.	The structure changes with the task, error-free sentences outnumber the rest by a wide margin, and what is left looks random rather than learned.

Read the three rows together and the practical test becomes clear. Band 6 and band 7 both tolerate errors, so accuracy alone does not separate them: band 7 requires that error-free sentences are frequent and that both sentence types are used effectively. Band 8 shifts the same words again, from frequent to the majority and from some errors to non-systematic errors. Nothing in the rows rewards a longer answer, a rarer tense, or a more impressive topic. All three reward structures that fit the question and land cleanly.

Six answers, one grammar move apart

Every excerpt below is a constructed teaching example written for this guide, not a real candidate, a real test question or a recorded performance. The question stays the same across the pair, and only one grammar move changes, so you can hear what that single move costs or earns. Each example is roughly the length of a Part 1 or Part 3 answer.

One: a relative clause that carries a reason. Question: whether you prefer studying at home. Band 6 version: I prefer studying at home. It is quiet and I can eat when I want. Band 7 version: I prefer studying at home, mainly because I control the noise, which matters when I am reading something dense. The move is the relative clause attached to a whole idea rather than to a noun. which matters when I am reading something dense links a cause to a consequence instead of adding a fourth short sentence.

Two: a second conditional in a Part 3 discussion. Question: whether cities should fund public libraries. Band 6 version: If the government spends money on libraries, it is good for students. It helps them study. Band 7 version: If a city cut that funding tomorrow, students who rely on the reading rooms would lose the only quiet place they can use. The move is the unreal conditional, which lets you discuss a hypothetical policy instead of repeating a general opinion. The risk is the verb form in the result clause; keep the modal and the bare infinitive together.

Three: a passive with a modal for impersonal statements. Question: how exam results should be used. Band 6 version: Teachers should check the results. They can see the problems. Band 7 version: Results should be checked against the tasks they were meant to measure, otherwise the wrong skill gets rewarded. The move is the passive with a modal, which lets you make a recommendation without naming a person. It is the structure that turns a repeated people should into something an examiner hears as range.

Four: a participle clause to sequence two events. Question: a time you had to solve a problem quickly. Band 6 version: I arrived late. Then I saw the queue and I called my manager. Band 7 version: Having missed the first train, I called my manager before I even reached the queue. The move is the perfect participle clause, which compresses two events into one sentence and puts them in order. It only works when both clauses share the same subject, so check who is doing what before you start the sentence.

Five: a comparative with a subordinate clause. Question: whether online lessons are as effective as classroom lessons. Band 6 version: Online lessons are good. But classroom lessons are better because of the teacher. Band 7 version: Online lessons are more efficient than they look, partly because the quiet students contribute more than they would in a room. The move is the comparative plus a clause of reason, which turns a blunt verdict into a measured comparison. This is the structure most useful in Part 3, where most questions are really comparisons in disguise.

Six: a past perfect that keeps a story in order. Question: a memorable shopping trip. Band 6 version: I went to the shop. The shop closed. So I went home. Band 7 version: By the time I got there, the shop had already closed, so I walked home with an empty bag. The move is the past perfect after a time marker, which makes the sequence of a Part 2 story unambiguous without repeating and then.

Six excerpts, six moves, one pattern: in each pair the band 7 version does not use harder vocabulary or a rarer tense. It uses one structure that matches the job the question is asking for.

Five structures worth rehearsing

Rehearse structures as tools with a job, not as decorations. Each line below names the structure, gives a shape you can reuse, and states where it earns its place.

• Relative clause for a reason: which matters because ... after a full clause. Use it in Part 1 and Part 3 whenever a short answer would otherwise stop at and it is good.
• Second conditional for a hypothetical: If ... were ..., ... would .... Use it in Part 3 when the question asks what should happen, rather than what does happen.
• Passive with a modal for recommendations: should be checked, can be measured. Use it when you want to discuss a process without inventing a person to blame.
• Participle clause for sequence: Having ... , I .... Use it in Part 2 to compress the first step of a story; check that one subject does both actions.
• Past perfect after a time marker: By the time ... , ... had .... Use it once in a story to set the order, then return to the simple past.

One rule covers all five. Choose the structure because the question needs that relationship, then say it in the simplest version of the structure that still shows the relationship. A relative clause with a clear referent is worth more than a long one that loses its subject halfway.

Errors that hold a performance at band 6

Some errors are compatible with band 7, because the descriptor allows a few basic errors to persist. Others press the performance down, and they do it by attacking the positive features of the row rather than by being frequent. The table separates the two.

Error pattern	What the examiner hears	Why it holds the score down	The fix
A complex clause started and abandoned mid-sentence	Repairs, restarts and I mean at the point where the structure became difficult	Flexibility is judged on completed structures; abandoned ones show range that is not yet controlled	Prepare the ending of the clause before you start it, or say the short version and add the detail in a second sentence
The same relative clause on every answer	... which is very good returning across answers on different topics	Range is counted across the performance, so one repeated shape reads as a single structure	Attach the clause to a different job each time: a reason, a contrast, a result
Tense drift inside one story	The past, then the present, then the past again in the same account	The listener loses the order of events, and the error hides inside the sentence that carries the content	Fix one time frame for the story and mark it once with By the time or After that
Agreement lost inside a long clause	A singular verb after a plural head noun, several words later	A basic error that recurs looks systematic, and the row reserves only a few persistent ones	Keep the subject and verb close, and start a new sentence rather than adding another clause
A question word order carried into reported speech	I do not know where is it repeated in Part 3	The structure is used often in discussion, so the same basic error becomes a pattern	Learn one frame, I am not sure whether ..., and keep the statement order after it
A memorised chunk used where it does not fit	According to me or a formal connective in a casual Part 1 answer	Flexibility includes appropriateness, and a chunk that ignores the register sounds recited	Keep the ready-made phrase, but change the sentence around it so it answers this question

The pattern across the table is that frequency is not the villain; repetition is. A candidate can make six small slips and still be heard at band 7 if the structures underneath them are varied and complete. A candidate who makes the same mistake every time a complex structure appears is showing the examiner a limit in range, which is exactly what the band 6 row describes.

A self-check you can run on your own recording

This is a private practice routine, not an official scoring method, and it uses only the wording the descriptors already give you.

1. Record three answers on three different topics: one Part 1 question, one Part 2 long turn, and one Part 3 discussion question. Keep each to about 60 seconds so the review stays fast.
2. Type out only the sentences that contain more than one clause. Ignore the short ones; they are not where the score is decided.
3. Next to each long sentence, write the name of the structure you used: relative clause, conditional, passive, participle, past perfect, comparative. If two sentences get the same label, that is your range limit showing.
4. Mark each of those sentences again for accuracy. Put a dot for clean, a cross for an error, and a question mark for anything you had to repair while speaking.
5. Count: at least three different structure names, and most of the long sentences clean, is the private target that matches the band 7 wording about range used flexibly and frequent error-free sentences.
6. Choose one structure that never appeared, and rehearse it in the next session on three unrelated questions. Repeat the recording and compare the labels, not the confidence.

Frequently asked questions

What is the difference between grammatical range and accuracy in IELTS Speaking?

Range is the variety of structures you produce across a performance, and accuracy is how many errors those structures contain. The band descriptors put them in one column because the examiner hears both at once, but they behave differently in practice: a safe answer full of short sentences can be accurate and still score low on range, while an ambitious answer can show real range and lose marks because the complex sentences do not land. Separating the two when you review a recording is what makes the column actionable.

Can I reach band 7 in IELTS Speaking with simple sentences?

Not on their own. The band 7 row requires a range of structures used flexibly, error-free sentences that are frequent, and both simple and complex sentences used effectively, so a performance built only from short forms cannot satisfy it. Simple sentences still have a job, because they carry detail and keep the answer moving, but they have to sit next to complex structures that fit the question. Because a few basic errors are allowed at band 7, the aim is not error-free speech but structures that are varied, appropriate and complete.

How do I practise complex sentences without sounding memorised?

Attach each structure to a purpose rather than to a topic, then rehearse it on three or four unrelated questions. A conditional rehearsed only on a prepared hobby script will appear in the test as a script, and the descriptor judges whether the structure fits the question. Practise out loud and record, because the useful signal is whether the structure arrives while you are still thinking about the content. Two or three structures used deliberately across a whole practice session will do more for range than one long, difficult sentence copied from a sample answer.