IELTS speaking is graded on four criteria: Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation. Each is worth 25% of your overall speaking band. Most candidates have heard these terms but have only a vague understanding of what each one rewards and penalizes in practice. A clearer understanding changes which aspects of your speech deserve the most attention.
Get an IELTS-style score, instant feedback, and a clearer next attempt.
Fluency and Coherence: what moves this score
Fluency is not speed — it is the absence of strain. The examiner is asking: does this person speak at a natural pace without evident searching, blocking, or self-interruption? Long silences, repeated false starts, and frequent 'um...um...um' sequences all penalize this criterion. Short pauses for natural breath or thought do not. The key is whether the overall impression of the speech is effortless or labored.
Coherence is the organizational half: does the response make logical sense and proceed in a direction the listener can follow? A candidate who gives a well-organized 60-second Part 2 talk scores higher on Coherence than one who speaks fluently for 90 seconds but jumps between points without signaling. Both halves matter equally, and most candidates are stronger on one than the other — knowing which is your weaker half tells you exactly what to target in practice.
Lexical Resource: vocabulary beyond knowing words
Lexical Resource rewards three things: range (variety of vocabulary), precision (using the right word for the exact meaning), and flexibility (paraphrasing when you can't recall a specific word). A candidate who uses only high-frequency, basic vocabulary throughout the exam caps at around band 6, even if every word is used correctly. A candidate who attempts more sophisticated vocabulary but makes occasional errors can score band 7 or higher, because the descriptor rewards 'attempts to use less common vocabulary' even when it 'produces some inaccuracies.'
The most practical implication: using simpler vocabulary correctly is not rewarded as highly as using more ambitious vocabulary with occasional errors. This seems counterintuitive, but the descriptor is explicitly designed to encourage lexical risk-taking. If you know a more precise word, use it — an imprecise attempt is scored higher than a precise simple word, because it demonstrates range. Candidates who play it safe by defaulting to simple vocabulary they are certain about are limiting their Lexical Resource ceiling.
Grammatical Range and Accuracy: what actually counts as complex
Grammatical Range rewards variety in sentence structure — the examiner is noting whether you use only simple sentences or mix in compound and complex ones. Subordinate clauses ('although,' 'even though,' 'which means that'), relative clauses ('the city where I grew up,' 'something that has always interested me'), conditionals ('if this continues,' 'had I known earlier'), and passive voice ('this has been largely driven by') all demonstrate range.
Accuracy rewards correct use of these structures. The examiner is not counting every error — they are noting error frequency and whether errors impede communication. Minor agreement errors ('people is') are noted but do not heavily penalize. Systematic errors that appear repeatedly ('I have went,' 'she don't') signal a persistent fossilized pattern and penalize more significantly. Errors on complex structures score better than errors on simple ones, for the same reason as Lexical Resource: attempting a complex structure that fails partially is worth more than succeeding consistently on simple ones.
Practice this topic now
See your score first, fix one weak pattern, and retry the same topic with clearer fluency and stronger structure.
Pronunciation: the most misunderstood criterion
Pronunciation is not assessed on accent — the band descriptors explicitly do not mention accent. What is assessed: intelligibility (can the listener understand you without effort?), word stress accuracy (is the emphasis placed on the correct syllable?), sentence-level stress and rhythm (does the natural emphasis of sentences fall where a native speaker would place it?), and features of connected speech (linking, reduction, intonation).
The single highest-value pronunciation improvement for most non-native speakers is sentence stress — placing stronger emphasis on content words (nouns, main verbs, adjectives) and lighter emphasis on function words (the, a, of, to, that). Native English listeners process speech partly by tracking stressed syllables. When stress patterns are heavily non-native, listening effort increases and the examiner's Pronunciation score is affected. This is practisable: record yourself, listen for where you place stress, and compare it to a native speaker saying the same sentence.