The scale in one paragraph
The ICAO Language Proficiency Rating Scale, published in ICAO Annex 1 and explained in Doc 9835 (icao.int could not be fetched this week), rates a pilot or controller on six skills: pronunciation, structure, vocabulary, fluency, comprehension and interactions. Each skill is scored from 1, Pre-elementary, to 6, Expert. Level 4, Operational, is the minimum for international operations. Your overall level is the lowest of the six scores; the ELPAC site states the rule as "there is no compensatory marking" and the TEA notes call it "a mandatory ICAO requirement". The scale is about effect on the listener, not about error counts: the lines ask how often a feature of your English interferes with "ease of understanding" or with meaning. The test measures speaking and listening only, in plain English in an aviation context; the TEA page says it is "not a test of aviation phraseology" and "not a test of operational knowledge". Below, each skill gets its own section with the full Level 4 line quoted from the scale as reproduced on Mayflower College's ICAO Level 4 page (read 21 October 2026), the Level 3 line in our plain-English paraphrase, a worked example written by TalkDrill, and the habits that most often cost the mark. Further down is a real, published TEA result with the examiner's comment on each skill.
- Six skills, each rated 1 to 6. Overall level = lowest skill (ELPAC: "no compensatory marking").
- Level 4 is the operational minimum per ICAO Annex 1 as widely published.
- Speaking and listening only. No reading, no writing, no phraseology test.
- Level 4 lines below are quoted; Level 3, 5 and 6 lines are TalkDrill's paraphrase of the scale.
- Every worked example on this page was written by TalkDrill. None is a real candidate. The one real result shown is Mayflower's own published example.
The six levels and what the names mean
| Level | Name | In plain English (TalkDrill) | Licence |
|---|---|---|---|
| 1 | Pre-elementary | Isolated words and memorised phrases; cannot hold an exchange. | No |
| 2 | Elementary | Simple memorised exchanges on routine matters; breaks down as soon as something is unexpected. | No |
| 3 | Pre-operational | Can talk about work on a good day, but errors, pauses and misunderstandings frequently get in the way. | No |
| 4 | Operational | English that works: an accent and some errors, but the message gets through and the speaker checks when unsure. | Minimum; re-evaluated at least every 3 years (Annex 1, widely published) |
| 5 | Extended | English that handles the unexpected with ease; errors rare and harmless; paraphrases without hesitating. | Yes; at least every 6 years |
| 6 | Expert | The language is simply a tool; the listener forgets to notice it. | Yes; not re-evaluated |
The gap that matters is between 3 and 4, because it is the gap between holding the operational level and not. The levels and validity page covers the gap between 4, 5 and 6 and the re-evaluation intervals. This page concentrates on 3 against 4 for each skill.
Pronunciation
What it measures: the sounds, word stress, rhythm and intonation of your speech, judged only by how often they make the listener work. Level 4 (ICAO wording): "Pronunciation, stress, rhythm, and intonation are influenced by the first language or regional variation but only sometimes interfere with ease of understanding." Level 3 (TalkDrill paraphrase): influenced by the first language and frequently interfering with ease of understanding.
The word doing the work is sometimes against frequently. An accent is assumed at every level up to 6, and the holistic descriptors ask only for "a dialect or accent which is intelligible to the aeronautical community". The rater is not asking whether you sound British or American; the rater is marking the moments when a sound or a stress pattern costs the listener the meaning. On the radio those moments cluster around numbers, call signs and place names, because those are the words with no context to rescue them.
Worked example (written by TalkDrill)
Examiner: "What is your fuel endurance?" The candidate means one hour fifteen minutes. Candidate A says "one hour FIF-teen", stress on the first syllable and the final n swallowed, so the examiner hears "fifty" and has to ask again. Candidate B says "one hour fif-TEEN", stress on the second syllable and a clear final n. Both have an Indian accent. Only the first interfered. If it happens again and again across the test, it moves towards the "frequently" of our Level 3 paraphrase; an occasional slip that the candidate repairs fits the "sometimes" of Level 4.
Habits that often cost the mark, by first language: v and w merged (Hindi, Punjabi, Gujarati), th said as t or d (many Indian languages), retroflex t and d that change the sound of "tower" and "door" (Telugu, Tamil, Kannada, Malayalam), p and b merged (Arabic), final consonants dropped so "eight" and "A" or "five" and "fie" blur (Vietnamese, Tagalog, Bahasa), and word stress placed by the rules of the first language (Turkish, Malayalam). Each has a short fix; the mistakes by first language page gives the drill for yours.
Structure
What it measures: grammar and sentence patterns, again by effect. Level 4 (ICAO wording): "Basic grammatical structures and sentence patterns are used creatively and are usually well controlled. Errors may occur, particularly in unusual or unexpected circumstances, but rarely interfere with meaning." Level 3 (TalkDrill paraphrase): basic structures associated with predictable situations are not always well controlled, and errors frequently interfere with meaning.
"Used creatively" is the phrase candidates miss. It means you build sentences for the situation in front of you instead of reproducing a memorised one. A pilot who can say "We diverted because the weather closed in" but cannot re-arrange it to "If the weather closes in we will divert" is at the edge of a 3, because the second sentence is the one a non-routine situation needs. The second half of the line is the reassurance: errors are expected "in unusual or unexpected circumstances", and they only count against you when they interfere with meaning.
Worked example (written by TalkDrill)
Examiner: "Tell me about a time you had to change your plan in flight." Level 3: "Weather is bad. We are diverting. Captain decide." The listener cannot tell whether this happened or is happening, or who decided what. Level 4: "The weather at destination went below minima, so the captain decided to divert to the alternate and I informed ATC." Tense is clear, the sequence is clear, and one slip ("at destination" without "the") costs nothing.
Habits that often cost the mark: the progressive for states and completed events ("we are having a problem since ten minutes"), dropped articles that remove the subject of the sentence, "will" and "would" confused in conditionals, no past-tense marker because the first language marks time with a separate word (Vietnamese, Bahasa, Tagalog), and the verb pushed to the end of the sentence (Turkish, Hindi). None of these matters until it changes the meaning; the drill is to narrate one real flight a day in the past tense, aloud.
Vocabulary
What it measures: the range and accuracy of your words on work-related topics, and your ability to paraphrase when the exact word is missing. Level 4 (ICAO wording): "Vocabulary range and accuracy are usually sufficient to communicate effectively on common, concrete, and work-related topics. Can often paraphrase successfully when lacking vocabulary in unusual or unexpected circumstances." Level 3 (TalkDrill paraphrase): range and accuracy often sufficient on common topics but limited, and the speaker often cannot paraphrase.
Paraphrase is the whole difference. A Level 4 speaker who does not know "chocks" says "the blocks they put in front of the wheels" and the exchange continues. A Level 3 speaker stops, searches, and the exchange dies. The rater is listening for that recovery as much as for the word itself.
Worked example (written by TalkDrill)
Picture: a baggage loader has struck the fuselage. Level 3: "The... the vehicle... it is... damage. The plane is damage." Level 4: "A baggage vehicle, the one with the belt, has hit the side of the aircraft below the cargo door. I can see a dent and the ground crew are checking it." The candidate did not know "belt loader" and did not need it.
Habits that often cost the mark: freezing instead of describing; local expressions a listener from another country will not know ("prepone", "do the needful"); "ma'am" and "sir" used as fillers; and technical words known in writing but never said aloud, so the stress is guessed. The drill is to describe three objects in the cockpit or on the apron every day without naming them.
Fluency
What it measures: tempo and flow, and in particular what happens to them when you move from rehearsed speech to spontaneous speech. Level 4 (ICAO wording): "Produces stretches of language at an appropriate tempo. There may be occasional loss of fluency on transition from rehearsed or formulaic speech to spontaneous interaction, but this does not prevent effective communication. Can make limited use of discourse markers or connectors. Fillers are not distracting." Level 3 (TalkDrill paraphrase): stretches of language are produced but phrasing and pausing are often inappropriate, and hesitations or slowness may prevent effective communication.
This is the line that catches candidates who prepared well. The introduction and the experience questions are rehearsable, and a rehearsed answer can sound like a 5. Then the examiner asks something unplanned, and the tempo halves. The rater has been waiting for exactly that transition. Fluency at Level 4 does not mean fast; it means the tempo stays appropriate when the material is new. In the published TEA example below, the summary for a Level 4 candidate says "there is repeated loss of fluency when searching for vocabulary and structures" and that "the interruptions to her flow of speech hold her to Level 4".
Worked example (written by TalkDrill)
Examiner: "You mentioned a go-around. What did the cabin crew do?" Level 3: "The cabin crew... ah... they... ah... cabin crew is... they sit... ah..." for eight seconds before a sentence forms. Level 4: "They were already seated for landing, so they stayed strapped in, and after we were established in the climb the captain made a short PA and I called the purser to check the cabin." Two brief pauses, no filler longer than "so".
Habits that often cost the mark: a learned introduction delivered at speed followed by a cliff; "actually", "basically" and "like" in every clause; translating in the head before speaking; and silence instead of a holding phrase. The holding phrases that keep a 4 are short: "Let me think", "Right, so", "In that case". Learn five and use them.
Comprehension
What it measures: how accurately you understand spoken English on work-related topics, including when the speaker's accent is unfamiliar and when the situation takes an unexpected turn. Level 4 (ICAO wording): "Comprehension is mostly accurate on common, concrete, and work-related topics when the accent or variety used is sufficiently intelligible for an international community of users. When the speaker is confronted with a linguistic or situational complication or an unexpected turn of events, comprehension may be slower or require clarification strategies." Level 3 (TalkDrill paraphrase): often accurate on common topics when the accent is sufficiently intelligible, but the speaker may fail to understand a linguistic or situational complication or an unexpected turn of events.
The test design puts pressure on this skill in two ways. First, tests play recordings of speakers whose accents you do not hear at your home airport. TEA Part 2, per the TEA notes, is "a series of recordings of international speakers of English" in three sections: six short non-routine recordings after each of which you report the message and say whether a pilot or a controller was speaking, four longer recordings you take notes on, and three short recordings you question. Second, the Level 4 line allows you to be slower and to ask for clarification; the Level 3 line is for the candidate who does not realise a clarification is needed. TEA's notes are precise about the trade-off: recordings "will only be played once unless you ask for repetition", "you cannot hear recordings a third time", and "regular repetition will suggest that your Comprehension is slower and this may affect your Comprehension score". Asking once, for a specific item, is a strategy. Asking every time is a symptom.
Worked example (written by TalkDrill)
Recording, spoken quickly with an unfamiliar accent: "Due to a vehicle on the taxiway, hold position at Bravo three, expect further taxi in five minutes." Level 3 re-tell: "Taxi to Bravo three." Level 4 re-tell: "There is a vehicle on the taxiway, so we hold at Bravo three and wait about five minutes. I did not catch whether the vehicle was on our taxiway, so I would confirm that. It sounded like a controller speaking." The second candidate missed something and said so, and answered the "who was speaking" question TEA Part 2A asks.
Habits that often cost the mark: training only on one or two familiar accents; listening for key words instead of the whole message; and a reluctance to ask a senior speaker to repeat. The drill is ten minutes a day of recordings from pilots and controllers in accents you do not know, re-told aloud, with every uncertain item named.
Interactions
What it measures: how you manage the exchange: whether you respond promptly and usefully, whether you take the initiative, and whether you check, confirm and clarify when something is unclear. Level 4 (ICAO wording): "Responses are usually immediate, appropriate, and informative. Initiates and maintains exchanges even when dealing with an unexpected turn of events. Deals adequately with apparent misunderstandings by checking, confirming, or clarifying." Level 3 (TalkDrill paraphrase): responses sometimes immediate, appropriate and informative; can initiate and maintain exchanges with reasonable ease on familiar topics but generally not when there is an unexpected turn.
This is the skill that candidates under-prepare when they rehearse complete answers for an examiner: fully, politely, without questions. The rater wants something else. A short answer, then a question. A read-back when a number is given. "Confirm" when something does not fit. TEA builds a whole section around it: in Part 2C, per the notes, "you have 20 seconds to ask the speaker questions to find out more about the situation", and then the examiner asks what advice you would give. The holistic descriptors below say why: the test exists to check that you can "recognize and resolve misunderstandings".
Worked example (written by TalkDrill)
Role play. Examiner, as controller: "Climb flight level three one zero, and be advised there is reported moderate turbulence between two eight zero and three two zero." Level 3: "Climb flight level three one zero." Level 4: "Climbing flight level three one zero. Confirm the turbulence is between two eight zero and three two zero? In that case request to stop climb at two seven zero until clear of the area." The second candidate read back, checked, and took the initiative.
Habits that often cost the mark: one-word answers followed by silence; never asking a question in the whole test; politeness formulas instead of content; and no read-back because the test "is not phraseology". It is not, but reading back a number to check it is a clarification strategy, and that counts.
The holistic descriptors: what the whole test is for
Above the six-skill scale ICAO sets out holistic descriptors: statements of what a proficient speaker can do. The Mayflower College page lists five (quoted with their original spelling). Proficient speakers shall:
- "communicate effectively in voice-only (telephone/radiotelephone) and in face-to-face situations";
- "communicate on common, concrete and work-related topics with accuracy and clarity";
- "use appropriate communicative strategies to exchange messages and to recognize and resolve misunderstandings (e.g. to check, confirm, or clarify information) in a general or work-related context";
- "handle successfully and with relative ease the linguistic challenges presented by a complication or unexpected turn of events that occurs within the context of a routine work situation or communicative task with which they are otherwise familiar";
- "use a dialect or accent which is intelligible to the aeronautical community".
The fourth is the one candidates overlook. It does not ask you to know every word for every emergency. It asks you to handle "a complication or unexpected turn of events" inside a situation that is "otherwise familiar": a routine approach that suddenly is not. That is exactly what TEA Part 2 and ELPAC Paper 2 Task 1 are built to create.
The ELPAC site says its Paper 2 is rated against criteria "which derive from the ICAO rating scale and the holistic descriptors". Read together with the six lines, the descriptors explain why compliant tests contain a voice-only listening component, a face-to-face or live speaking component, and at least one non-routine situation. TEA's Part 2 is voice-only recordings and its Parts 1 and 3 are face to face; ELPAC's Paper 1 is listening and its Paper 2 is live speaking. They also explain why the TEA page insists the test is "not a test of aviation phraseology". Phraseology is what you use when everything is routine; the descriptors are about what you do when it is not. Our non-routine situations page is built around that switch.
How raters use the scale on the day, and a real TEA result
Formats differ, but the two we verified this week show two rating models.
| TEA (aviation-english.com and the TEA Notes for Candidates) | ELPAC (elpac.eurocontrol.int) | |
|---|---|---|
| Who rates | "A face-to-face, human interaction test with 1 examiner" | Paper 2: "two accredited assessors", an English Language Expert and an Operational Expert |
| What is rated | "Only your test performance"; examiners "cannot consider any aspect of previous experience or previous test scores" | "Each of the six ICAO criteria is rated independently" |
| Recording | "The test will be recorded for security and quality purposes" | On disagreement, "the recording of Paper 2 is reviewed by an independent Third Assessor"; that decision is final |
| Overall level | "The lowest of these scores (which is a mandatory ICAO requirement)" | "The lowest rating across the six criteria from both assessors" |
| What you receive | All six scores on the certificate, posted to your centre "within 10 working days" | The user organisation "issues the official report to you" |
In both models the rater is not totting up errors. The rater listens for the moments the listener had to work, places each skill on the line that describes those moments best, and reports the lowest.
A real published result: TEA example, Level 4
Mayflower publishes an example Level 4 TEA candidate from its Plymouth centre, linked from the TEA page. It gives her six scores and a performance summary that explains each one. It is the clearest picture we found of the scale in use, so here it is, with the summary quoted against each skill (read 21 October 2026).
| Skill | Score | From the performance summary (quoted) |
|---|---|---|
| Pronunciation | 5 | "generally clear but deviations from clear and accurate vowel and consonant production can be noted in different parts of test" |
| Structure | 4 | "patterns of non-interfering errors of basic structures are common, even in work-related or predictable contexts, and errors which may interfere with meaning are present in more cognitively demanding situations" |
| Vocabulary | 4 | "repeated errors in word form and regular mis-use of words (e.g. mis-translations or mis-collocations) are in evidence, which prevent her from reaching Level 5" |
| Fluency | 4 | "repeated loss of fluency when searching for vocabulary and structures. The interruptions to her flow of speech hold her to Level 4." |
| Comprehension | 4 | "mostly accurate throughout, although she misses the nuance of some of the more complex items"; "many reports contain noticeable omissions of specific details" |
| Interactions | 5 | "able to manage the speaker/listener relationship effectively and offer informative and appropriate responses. Many responses are not, however, immediate or effortless." |
| Overall | 4 | The lowest of the six |
Two lessons sit in that table. The first is the lowest-score rule in action: two 5s did not lift the overall level. The second is the language raters use: "non-interfering errors", "repeated loss of fluency", "noticeable omissions of specific details". Notice too that interactions scored 5 even though "many responses are not, however, immediate or effortless": the summary explains why it was not a 6, not why it was not a 4. These are the lines of the scale, applied to one person. Read your own recordings with the same vocabulary. The tests compared page has the TEA and ELPAC formats side by side.
A self-check sheet you can use after every recording
Record yourself describing a picture for two minutes and answering three follow-up questions, then answer these twelve questions. Each pair maps to one skill. Two "no" answers in a pair suggest that skill is your lowest, and your level, until it moves.
| Skill | Check 1 | Check 2 |
|---|---|---|
| Pronunciation | Did every number and place name come out so that a stranger could write it down first time? | Did you stress "fifteen", "sixteen" and "seventeen" on the second syllable and the tens on the first? |
| Structure | Can a listener tell from your verbs what happened, what is happening and what will happen? | Did you build at least one "if ... then" or "because ... so" sentence that you had not rehearsed? |
| Vocabulary | When a word was missing, did you describe the thing instead of stopping? | Did you avoid local expressions that a listener from another country would not know? |
| Fluency | Did your tempo stay roughly the same after the first unplanned question? | Was every pause short, or filled with a holding phrase? |
| Comprehension | Did you re-tell the whole message, not only the key words? | When you missed something, did you ask once, for that item, instead of asking for the whole thing again? |
| Interactions | Did you ask at least two questions in the whole recording? | Did you read back or confirm every number you were given? |
Practice route
- Read the TEA example first. Ten minutes with the published Level 4 candidate and its performance summary teaches you what a rater writes down.
- Baseline today. One picture from the practice bank, two minutes, three follow-ups, then the self-check sheet above. Write the lowest skill at the top of your plan.
- Daily, ten minutes: a non-routine role-play in TalkDrill role-play practice, where the AI plays the controller or the captain and introduces something unexpected. Target: two questions and one read-back per exchange. This trains comprehension and interactions together.
- Daily, five minutes: numbers, call signs and place names aloud in read-aloud practice, focused on the sounds your first language merges.
- Weekly: a full recording scored with the sheet. Keep the recordings; the fourth one next to the first is the progress report that matters.
- Before the test: the 30-day roadmap arranges all of this into four weeks, with a full mock in week four. A human mock with an ICAO-trained rater is not on TalkDrill yet; until it is, the AI mock and this sheet are what we offer.
Every page in this module is linked from the aviation English hub; the official resources page lists the ICAO, EUROCONTROL, Mayflower and DGCA links with the date each was last checked.