Listening test for hiring

20 questions on recorded announcements, voicemails, conversations and discussions, in 12 to 16 minutes. Each recording plays up to twice, the test adapts after the first 8 questions, and the CEFR level is ready the moment the candidate finishes.

Listening · Part 2Recording 1 of 3
4 questions

You will hear two colleagues planning a training day for their team.

1. Why does the woman want to move the training day?

  • The trainer cannot come that week
  • Too many of the team are away then
  • Someone else has booked the room
  • She has not finished the materials

3 more questions on this recording below.

Play 1 of 20:48 / 1:17
Plays up to twice. The transcript is never shown.Sample question
Length
12 to 16 min
Questions
20
Plays per recording
Up to 2
From
$0.60 a test

What is a listening test?

A listening test for hiring measures how well a candidate understands a language when they hear it: a time and a place in a voicemail, the point of a conversation, and, at higher levels, what a speaker implies or how they feel about what they are saying. Evalingo's is 20 multiple-choice questions on recordings made for each CEFR level, taken online from a link, with a level from A1 to C2 the moment the candidate finishes. It runs in English and German today, with more languages on the way.

Listening is the skill a speaking test assumes and does not measure: a candidate can answer fluently and still miss what a fast or unfamiliar speaker says, and on a phone queue that is the failure customers notice first. Every question here has one right answer, so the result is a calculation, instant and identical every time.

How the test adapts

Everyone starts with the same 8 short clips. How many they get right decides which three longer recordings come next, so most of the test is spent at the candidate's own level rather than on recordings far too easy or far too hard for them.

Part 1 · everyone

8 short recordings, one question each

A2B1B2

Centred where the routing decision is made, so every question here helps decide what comes next.

Lower route

Part 1: 4 or fewer right

A1A2B1

Middle route

Part 1: 5 to 6 right

B1B2C1

Upper route

Part 1: 7 or more right

B2C1C2

Part 2 is 3 longer recordings with 4 questions each, one at each level of the route. Neighbouring routes overlap, so a candidate near a cut-off is still measured across their real level. All 20 answers count towards the result, whichever route they took.

Paced by the audio, not a clock
Each screen gives time to read the questions first, 10 seconds in Part 1 and 30 or 45 in Part 2, then plays the recording. The candidate can start it sooner with Play now, and nothing moves on until they press Next.
Up to 2 plays
Each recording plays twice at most and the second play is optional, as in most established listening exams. A play cut off by the page being hidden, such as by an incoming phone call, is given back.
Questions on screen, never the transcript
The questions stay visible while the audio runs. The transcript is never shown, so the test measures listening rather than reading.
A sound check first
Before the first recording the candidate checks their sound and, where the browser allows, chooses which speakers or headphones to use. Nothing here can be read instead of heard, so going on without sound comes with a warning.
A voice for each speaker
Conversations use a separate voice for each speaker, and the pace follows the level: slow and clear with pauses at A1, natural speed with the hesitations of real speech from C1.
Extra time when needed
An employer can grant 25%, 50% or 100% extra time as an accommodation. It lengthens the reading time before each recording, never the audio.

What the questions ask

Each question targets one thing a listener has to do with what they hear. Lower levels ask mostly for stated facts and the main point; higher levels ask what is implied, what a speaker thinks, and how an argument is built.

Detail
A specific fact stated in the recording: a time, a price, a place, a name.
Main point
What the recording is mainly about, or the overall message.
Purpose
Why the recording, or a part of it, was said.
Vocabulary
What a word or phrase means where it is used.
Reference
What a word such as "it" or "this" points back to.
Inference
What the recording implies without stating it.
Attitude
The speaker's opinion, feeling or tone.
Structure
How an argument is built, or what one part of it is doing.

What a recording sounds like at each level

Recordings are scripted to the CEFR Companion Volume listening scales. As the level rises, speech gets faster and less carefully articulated, conversations carry more disagreement and qualification, and more of the meaning is in tone and implication rather than stated outright.

LevelWhat they hearDeliveryWords
A1voicemail, simple announcement, self-introductionVery slow, clear, standard accent, long pauses between sentences.40 to 55
A2announcement, voicemail, simple two-person exchangeSlow and clear, standard accent.70 to 95
B1conversation, service conversation, phone callClear standard speech at a moderate pace.120 to 145
B2planning discussion, interview, radio interviewNormal conversational speed, standard accents, natural features of speech such as contractions and discourse markers.170 to 200
C1talk excerpt, panel discussion, explanatory talkNatural speed with the hesitations, reformulations and asides of real speech.230 to 260
C2panel discussion, expert interview, opinion podcastFast natural speed, colloquial delivery, a wider range of accents.250 to 280

Who needs a listening test

Roles that commonly need it

  • call centre and voice support agents
  • dispatchers
  • front desk and reception staff
  • hospitality staff
  • care workers
  • warehouse and site staff

On a phone queue, listening is half of every call. An agent has to catch a name, a number and a complaint from a caller who may be fast, upset or on a bad line, and a speaking test only checks the half where the agent talks.

It matters on site too: spoken instructions on a warehouse floor, a shift handover and a safety briefing all assume the listener understood, and none of them gets repeated. Paired with a speaking test, it gives you both directions of a conversation, each with its own level.

What comes back

A CEFR level from A1 to C2 and a score from 0 to 100 on the same bands as Speaking and Writing, ready the moment the last screen is submitted.

The result shows right answers per part and per recording, and which route Part 1 picked. A sitting that cannot be told apart from guessing, or one moved through before the recordings could be heard, gets "Not assessable" instead of a level.

How the scoring works
Sample result

Listening · English

Score 63
CEFR level
B2
Part 1, short recordings
6 / 8
Route chosen
B1 to C1
Part 2, B1 recording
4 / 4
Part 2, B2 recording
3 / 4
Part 2, C1 recording
2 / 4
Total
15 / 20
Every answer opens on the admin page: the recording, the question, what was chosen and what was right.

Questions people ask

How long does the listening test take?
12 to 16 minutes. It depends mostly on whether the candidate plays each recording once or twice, since the test moves at the pace of the audio rather than against a clock.
How many times can candidates hear each recording?
Up to 2. The second play is optional, and the questions are on screen throughout, so a candidate can read them before the audio starts and check their answers on the second play.
How are the recordings made?
Each recording is scripted to its CEFR level and voiced with natural-sounding AI voices, one per speaker, at the pace the level calls for. Every clip is transcribed back and checked against its script, and a recording is remade whenever its script changes, so what the candidate hears always matches what they are scored against.
How is the listening test scored?
By calculation, not by AI. Every question has one keyed answer and a difficulty on a common scale, and the candidate's level is estimated from which questions they got right, so a hard question answered correctly counts for more than an easy one. The model allows for guessing, with a floor of one in 4, which stops lucky guesses at the bottom of the scale from reading as ability. The same answers always produce the same level.
Why does the listening test adapt to the candidate?
A fixed 20-question test spends most of its questions far above or below any one candidate, and those tell you very little. Routing on the first 8 answers puts the next 12 at the candidate's own range, where each answer says the most. Because the result is calculated from how hard each question is, candidates on different routes are still placed on one scale.
What stops a candidate guessing their way through?
A check compares the pattern of answers with pure guessing. A genuine low-level candidate gets the easy questions right and the hard ones wrong, while a guesser scatters, so a sitting that cannot be told apart from guessing comes back as "Not assessable" rather than as a level. So does a sitting where most screens were answered before the recording could have been heard.
Which languages is the listening test available in?
English and German today, and it is being extended to every language we assess. Each language gets its own bank of recordings and questions, written and reviewed to the CEFR level descriptions, and becomes available once that bank is complete. Speaking and writing are available in 13 languages now.
Can we use our own recordings and questions?
No. Reading and listening draw only from our bank, which is written to level specifications and reviewed question by question, so a level means the same for every account. Question libraries you write yourself apply to speaking and writing.
Can I see which questions a candidate got wrong?
Yes, on the result page: every recording with its questions, what the candidate chose and what was right. That detail stays in the dashboard. The API and webhooks report the level, the score and right answers per part, so the questions themselves are never copied into other systems.
Can the listening test be combined with other skills?
Reading and listening can be ordered together as one assessment of 26 to 32 minutes, and the overall level is the average of the two. Speaking and writing are ordered as a separate assessment of about 20 minutes, and every skill reports on the same CEFR scale, so the levels can be read side by side.
How much does the listening test cost?
From $0.60 a test on annual billing, with no setup fee and no per-seat licence. You can start free with ten assessments and no card.

Hear the test before your candidates do

Ten free assessments, no card. Take the listening test yourself, then send it to a few people whose level you already know.

No credit card required.