IELTS Speaking Band Descriptors Explained: What Examiners Are Actually Listening For

Most candidates prepare for IELTS Speaking without ever reading the document the examiner is actually scoring them against. They watch YouTube videos about “tips and tricks,” memorise a few impressive phrases, and walk into the test hoping fluency and a bit of vocabulary will carry them to a 7. Then the result comes back at 6.0 and they have no idea why, because nobody ever showed them what a 6 sounds like next to a 7 in the examiner’s own terms.

The IELTS Speaking test is marked against four public, named criteria: Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation. Each one has its own band descriptor table, and each band level is defined by specific, observable features of language — not vague impressions of how “good” you sounded. Examiners are trained to listen for these features and mark against them in real time. That means your score is not a mystery. It is the output of a checklist, and the checklist is public.

This article goes through all four criteria in plain English, explains what the examiner is actually listening for at each band, and — most usefully — pins down exactly what separates a Band 6 answer from a Band 7 answer on each one. Band 6 to 7 is where most serious candidates get stuck, and it is stuck for identifiable, fixable reasons, not some abstract lack of “level.”

If you take one thing from this article, take this: stop preparing for “the test” as a single blurry event. Prepare for four separate, nameable skill sets, because that is how it is being scored.

Book a free IELTS Speaking assessment call →

Fluency and Coherence

Fluency and Coherence measures two related things: how continuously you can speak, and how logically your ideas connect together. It is not about speaking fast. A slow, deliberate answer that flows logically can score higher than a fast answer full of dead stops and disconnected ideas.

The “fluency” half is about your willingness and ability to speak at length without long, unnatural pauses — pauses where you’ve simply run out of things to say or lost your thread. The “coherence” half is about whether your ideas are sequenced sensibly and linked together in a way a listener can follow without effort, using connectors that sound like real spoken English rather than something lifted from an essay.

That last point trips up a lot of candidates. Words like “moreover,” “furthermore,” and “in addition to this” are written-style linkers. Dropping them into spoken answers doesn’t sound advanced — it sounds rehearsed and slightly odd, because nobody talks like that. What examiners are listening for instead is natural spoken discourse markers: “well,” “actually,” “to be honest,” “I mean,” “the thing is,” “so yeah.” These are the connectors real fluent speakers use to organise thought in real time, and their natural, appropriate use is part of what coherence actually measures.

Band 6 is described as someone who is willing to speak at length but whose speech shows some hesitation, repetition, or self-correction, sometimes over-uses certain connectors, and occasionally loses coherence when they lose their thread. It is functional, understandable, generally coherent speech — but with visible seams.

Band 7 speaks at length without noticeable effort or loss of coherence, and any hesitation is language-related — meaning you’re pausing because you’re searching for the right word or grammar structure, not because you’ve lost the idea itself. Band 7 also uses a range of connectives and discourse markers with some flexibility, rather than leaning on the same two or three repeatedly.

Here is what actually moves you from 6 to 7:

Fill silence with real stalling language, not dead air. A dead pause — total silence while you think — reads as a fluency breakdown even if it’s only two seconds. Fluent speakers never go silent; they stall out loud. “That’s a good question, let me think…” keeps the speech continuous while your brain catches up. This is a trainable habit, not a talent.

Structure your Part 2 answer like a story, not a list of facts. Weak Part 2 answers are just four disconnected facts about the topic. A Band 7 narrative has shape: setup (when/where/who), detail (what happened, specifics), a turning point or feeling (what changed, how it felt), and reflection (what you think about it now). That shape is what makes two minutes of speech feel coherent rather than like reading down a checklist.

Plan the end of your sentence before you start it. Most self-correction happens because candidates start speaking before they’ve decided how the sentence finishes. Deciding the grammatical shape of a sentence a half-second before you say it eliminates most mid-sentence correction, and self-correction is explicitly one of the features that pulls an answer down toward Band 6.

Lexical Resource

Lexical Resource is the criterion candidates misunderstand most. They think it means using “big words.” That is not what it measures. It measures how flexibly, accurately, and naturally you use vocabulary to talk about a wide range of topics, including topics you didn’t prepare for.

Three things matter more than word rarity: collocation, paraphrase, and appropriacy.

Collocation means words that naturally pair together in English — “make a decision,” “make an effort,” “have a conversation,” “hold a meeting.” These pairings aren’t logical, they’re fixed by convention — and getting them wrong is one of the fastest ways to reveal that vocabulary is memorised rather than internalised.

Paraphrase is the ability to say the same idea in different words — essential because you will frequently not know or momentarily forget the exact word you want. It also shows up in how you open your answer: repeating the examiner’s question word-for-word back at them is a Band 5-6 habit. Rephrasing the question in your own words at the start of your answer is a small, visible signal of lexical control.

Band 6 uses vocabulary with enough flexibility to discuss familiar and unfamiliar topics, but the range is limited for less common topics, and there are some inappropriate word choices along with occasional paraphrasing.

Band 7 shows flexible vocabulary to discuss a variety of topics, uses some less common and idiomatic vocabulary, shows awareness of style and collocation with some inappropriate choices, and paraphrases effectively.

What moves you from 6 to 7, specifically:

Natural collocations under time pressure. Not rare words — correct pairings, produced fast, without stopping to check them. This is built by noticing and drilling collocations in the topics you actually get asked about, not by memorising vocabulary lists of “advanced” isolated words.

One genuine idiomatic expression per answer, used correctly. Not five. Not forced in where it doesn’t fit. Examiners can tell instantly when a candidate has bolted on a memorised idiom that doesn’t match the context — it actually damages the score because it signals a scripted answer rather than natural range.

Paraphrase the question, don’t recycle it. Before answering, briefly restate the question idea in your own words. It buys you a second to think, and it directly demonstrates the paraphrasing skill the descriptor names outright.

Grammatical Range and Accuracy

This is the criterion most candidates fear because they assume it means “no mistakes.” It doesn’t. It measures two separate things combined into one score: range (how many different grammatical structures you use) and accuracy (how correctly you use them). Crucially, the descriptors care more about range than about being error-free. A grammatically perfect answer made entirely of simple sentences scores lower than a more ambitious answer with a complex structure or two and a minor error, because the second candidate has demonstrated more of the language.

Band 6 uses a mix of simple and complex sentence forms, but with limited flexibility — meaning the complex structures tend to be a small, repeated set rather than a genuine range, and errors are frequent enough that they sometimes cause some difficulty for the listener, though basic meaning still gets through.

Band 7 produces a range of complex structures with some flexibility, and while errors still occur, they rarely impede communication. The key shift from 6 to 7 isn’t fewer errors so much as more structures attempted, more successfully.

Two structures the descriptors are specifically listening for at this gate:

Hypothetical and conditional language. “If I had more time, I would learn another instrument.” Conditionals — especially second and third conditional, and modal-based hypotheticals like “I would have,” “I might have” — are one of the clearest, most checkable signals of range that examiners are trained to notice. Most Band 6 candidates simply avoid this structure altogether because it’s harder to produce under pressure, which is exactly why using it correctly and naturally is disproportionately valuable.

Comparatives and superlatives beyond “-er” and “-est.” “Better,” “bigger,” “the biggest” are Band 5-6 comparative forms — correct, but limited. Band 7 range includes constructions like “far more expensive than,” “nowhere near as convenient as,” “by far the most useful,” “increasingly popular.” These aren’t obscure grammar — they’re common, natural ways fluent speakers actually compare things — but most candidates default to the simplest form under exam stress because it’s the first thing that comes to mind.

The practical takeaway: accuracy matters, but it is not the ceiling on your score the way most candidates assume. A candidate who attempts conditionals and varied comparatives, and makes the occasional small error doing it, will consistently outscore a candidate who plays it safe with only simple present and past tense sentences, however clean those sentences are.

Pronunciation

This is the most misunderstood criterion of all four, and it’s worth stating plainly: Pronunciation is not about sounding native. The official band descriptors do not mention accent as a target at any band, including Band 9. What they describe, even at the top band, is a speaker whose speech is easy to understand throughout, where L1 (first-language) accent has minimal effect on intelligibility. In other words, you can keep your accent all the way to a Band 9, provided a listener can understand you without strain.

What the criterion actually measures is intelligibility: can the examiner follow what you’re saying, effortlessly, without having to guess. That is a function of specific, learnable features — not of erasing your accent, which is neither necessary nor realistic in the exam timeframe.

Band 6 speaks with some effective use of features like word and sentence stress, chunking, and intonation, but this isn’t sustained, and there may be lapses that mean the listener occasionally has to work to understand.

Band 7 shows all the positive features of Band 6 more consistently, though there may still be minor difficulties with individual sounds or the odd inconsistency in stress/intonation — but overall, the speaker is easy to follow throughout, and any L1 influence has minimal effect on that.

Two trainable features do most of the work here:

Sentence stress for meaning. In English, the word that carries the actual point of the sentence gets stressed, and that stress physically shifts depending on what you mean. Native and highly intelligible non-native speakers stress the content word carrying the important new information — not every word equally, which is what makes flat, monotone speech hard to follow even when every individual sound is pronounced “correctly.”

Chunking, or thought grouping. Fluent, intelligible speech isn’t produced word by word — it’s produced in small grammatical chunks with tiny pauses at natural boundaries: after a subject, before a new clause, around a prepositional phrase. Speaking without any chunking, or pausing randomly mid-phrase, is one of the most common reasons an otherwise clear speaker is hard to follow.

Neither of these requires changing your accent. Both are about the shape and rhythm of speech sitting on top of whatever accent you have.

Why “just practise speaking more” doesn’t work past Band 6

This is the advice almost every candidate gets, and it’s why so many plateau at 5.5-6.5 for months. “Just speak more English” treats fluency, vocabulary, grammar, and pronunciation as one vague, undifferentiated “level” that improves through general exposure — like a bar filling up the more you talk.

It doesn’t work that way past a certain point, because each of the four criteria above is a distinct, nameable skill with its own specific gap. General speaking practice will absolutely get you from a low level to a functional one. But once you’re already reasonably fluent — which most candidates stuck at Band 6 are — undirected practice just repeats the same errors at higher speed.

The candidates who move from 6 to 7 are, almost without exception, the ones who stopped practising “speaking” in general and started practising named skills: today I am specifically working on conditional structures; today I am specifically drilling sentence stress; today I am specifically eliminating dead pauses by replacing them with stalling phrases.

How to actually diagnose your own gaps

You cannot fix what you can’t name, and most candidates genuinely don’t know which of the four criteria is holding their score down. Two things fix that.

Record yourself answering real test questions, then listen back against the descriptors above, one criterion at a time. Don’t just listen for “how did that sound overall.” Listen once purely for pauses and self-correction. Listen again purely for collocation errors and repeated vocabulary. Listen again purely for which grammatical structures you actually attempted. Listen a final time purely for stress and chunking.

Get feedback that names the specific skill, not just a number. “You’re a 6” tells you nothing you can act on. “You’re avoiding conditionals and defaulting to simple comparatives” tells you exactly what to drill this week.

There are two practical ways to get that kind of specific, skill-level feedback. Actually Speak’s IELTS Speaking practice system is built around exactly this: structured practice against the four criteria with feedback tied to named, fixable skills rather than a vague overall impression. For candidates who want direct, personalised coaching on their own recorded answers, 1:1 coaching with Matthew Fabling at Smooth English Company works through this same descriptor-by-descriptor process, pinpointing which specific gate — fluency, lexis, grammar, or pronunciation — is capping the score and building a plan around it.

Either way, the principle is the same: stop training for an overall “level” and start training against the four things that are actually being marked.