Top 10 English proficiency test questions for potential candidates

Good English proficiency test questions for candidates measure the English the job actually uses: reading a client brief without missing the deadline buried in paragraph three, writing a reply that does not need a follow-up, following a fast phone call. Ten question types cover that ground. Each one below comes with an example you can lift and a note on what a strong answer really tells you.

Most hiring teams get this wrong in the same way. They ask for "fluent English" on the job post, then test grammar trivia that no one uses at work, and end up ranking candidates on how well they spot a misplaced comma. That is a problem for hiring quality, and it is a legal problem too. The US Equal Employment Opportunity Commission's 2016 guidance on national origin discrimination says employers "are advised to assess the level of fluency required for a job on a case-by-case basis", and warns that applying one uniform fluency bar across very different roles can breach Title VII. So "fluent" is not a standard. A level tied to a task is.

TL;DR

• Test the four skills the role uses (reading, writing, listening, speaking), not all four by default. A warehouse supervisor and an account manager need different things.
• Write every question from a real work task. If the candidate would never do it on the job, it does not belong in the test.
• Set the bar as a CEFR level (A1 to C2) with a score cutoff, not as the word "fluent". Even governments do it this way.
• Score with a rubric agreed before anyone sits the test. Unstructured judgement is where bias and inconsistency get in.
• Keep an accommodation route open and document why the level you picked matches the job. That is both fairer and more defensible.

Example English proficiency test questions and answers

Use four or five of these, chosen for the role, rather than all ten. A 30-minute test that mirrors the job beats a 90-minute one that mirrors a language exam. Each type below gives you what it measures, a question you can lift, and the signal in a good answer.

Grammar and sentence structure questions

Grammar questions are worth asking only where written output goes to a customer or into a record. Keep them in context instead of testing rules in the abstract.

Example question: "Correct the error in this line from a customer reply: She don't have access to the invoice yet."

A strong answer fixes the verb agreement ("She does not have") without rewriting the whole sentence. Candidates who rebuild the line from scratch are often hiding uncertainty about what was actually wrong.

Vocabulary test questions and word usage

A vocabulary test is the question type teams most often get wrong, because it is the easiest to make useless. "Choose the synonym for ephemeral" tells you about exam prep, not about work. Test the words the role uses.

Example question: "A client writes: Can you ballpark the cost? What are they asking for, and write a one-line reply."

Look for a candidate who reads "ballpark" as a rough estimate and answers with a range plus a caveat about what would change it. That is vocabulary plus judgement in one answer.

Reading comprehension questions

Give candidates a real document, not a passage about dolphins. A shortened client email, a policy paragraph, or a support ticket works well.

Example question: "Read this 150-word client email and answer: what is the deadline, what is the client actually unhappy about, and what do they want you to do first?"

The useful signal is whether they separate the stated request from the underlying complaint. Plenty of candidates with strong grammar miss that the client is annoyed about something they never named directly.

Listening comprehension questions

Listening is the skill most often skipped and most often the real bottleneck, especially for phone and video roles. Use audio or video prompts with natural pace and some crosstalk, not a slow studio recording.

Example question: "Listen to this 60-second call recording between a customer and an agent, then summarise the customer's problem in two sentences and name the next step the agent promised."

Scoring is simple: did they catch the commitment? Candidates who summarise the mood but miss the promised action will create follow-up work for everyone around them.

Written communication questions

This is the highest-value question type for most office roles, and the hardest to fake. Give a messy input and ask for a clean output.

Example question: "A delivery slipped by a week because a supplier missed a part. Write the email to the client. Keep it under 120 words, explain what happened, and propose a new date."

Read for structure before style: does the bad news come early, is there a concrete next step, is the tone level rather than either defensive or grovelling? Word limits matter here. Writing short under pressure is a skill in itself.

Spoken communication questions

Ask for speech when the role involves live conversation. A recorded one-way answer is enough for most screening; save live conversation for the interview stage.

Example question: "In 90 seconds, explain to a non-technical customer why their order is delayed and what you are doing about it."

Listen for plain words, a clear sequence, and no filler. Accent is not the thing being measured, and treating it as a score is a fast route to a discrimination claim.

Idioms and workplace expressions

Idiom questions are a fairness trap if you are not careful, because idioms are regional. Test only the expressions your own team genuinely uses in writing.

Example question: "A colleague writes: Let's circle back on this next week. What are they proposing, and is any action needed from you today?"

The right answer is that nothing is needed today and the topic returns next week. If your team does not write like this, drop this question type rather than keeping it for decoration.

Business email etiquette questions

Tone is a separate skill from grammar. A sentence can be correct and still cost you a client.

Example question: "Rewrite this reply so it stays firm but does not read as rude: No, we already told you that was not included."

Good answers keep the no, add the reason, and offer a route forward. Candidates who turn the no into a maybe have solved the tone problem by creating a worse one. If written etiquette is central to the role, pair this with an email etiquette assessment rather than relying on one question.

Inference and critical reading questions

Inference separates candidates who can process English from candidates who can act on it.

Example question: "A supplier writes: We may be able to meet that date if the shipment clears customs early. Can you promise the date to your client? Explain in one sentence."

The answer is no, and the reason is the conditional. Candidates who read "may be able to" as a yes will pass on commitments the business cannot keep.

Cross-cultural communication questions

For distributed teams, the risk is rarely vocabulary. It is a message that reads as blunt in one country and warm in another.

Example question: "You need an answer today from a colleague in another country who has not replied to two emails. Write the third message."

Strong answers name the deadline and the reason, offer a fast channel, and avoid both passive hinting and public escalation. There is no single correct version, which is exactly why this one needs a rubric and a second reviewer.

What should an English assessment in recruitment measure for each candidate?

An English language proficiency test measures some mix of four skills, reading, writing, listening and speaking, against a defined level. In hiring, it should measure only the skills the role uses, at the level the role needs, scored the same way for every candidate. Anything beyond that is testing for its own sake.

That ordering is the point of the Testlify Competency-to-Evidence Matrix, which maps every role to the competencies that matter, then connects each competency to measurable evidence through assessments, simulations, interviews, references and structured feedback. Applied here it means one thing: do not start with a test. Start with the role, name the English tasks it contains, and only then pick question types.

Two things fall out of that table. First, no role needs all four skills at the same level, so a single blended score hides the thing you care about. Report by skill. Second, writing matters in more roles than speaking does, which is the reverse of how most interview processes behave.

What is the most effective way to test English proficiency for recruitment?

Define the level before you write a question, test only job-relevant skills, score against a rubric fixed in advance, and keep the whole thing short enough that candidates finish it. Validation comes last and matters most: check the test actually separates people who do the job well from people who do not.

1. Write down the English tasks in the role. Pull them from the job description, not from memory. If a task is not in the job description, it should not be in the test, and the EEOC guidance above is the reason to put it in writing.
2. Pick a level, not an adjective. The Council of Europe's Common European Framework of Reference for Languages (CEFR) runs A1 through C2 and is the common scale. The UK government states its own English requirement for citizenship and settlement as a qualification at B1, B2, C1 or C2 level, which is a useful reminder that a named level is the normal way to express this.
3. Choose four or five question types from the ten above. Weight them the way the job weights them. A typing and written-accuracy score, for instance, can be weighted deliberately: a default of 60% accuracy and 40% speed is adjustable when the role cares more about one than the other.
4. Set the cutoff before you see any scores. Pick the pass mark from the role analysis, write it down, and apply it to everyone. A certificate threshold with a floor of 40% is a configuration choice, so decide it on purpose rather than accepting a default.
5. Standardise delivery. Same questions or same question bank, same time limits, same instructions. A smaller question bank raises the chance that two candidates see overlapping questions, so there is a real trade-off between anti-cheating variety and comparability.
6. Score blind, with two readers on open answers. Written and spoken answers need a rubric and a second reviewer. Multiple-reviewer scoring is what stops one person's taste in prose becoming the hiring standard.
7. Validate it. The EEOC's technical assistance on employment tests and selection procedures is direct about the bar: "The test or selection procedure must be job-related and its results appropriate for the employer's purpose," and tests should be "properly validated for the positions and purposes for which they are used." Item-level statistics make this practical. A difficulty index and a discrimination index per question show which items are too easy, too hard, or failing to separate strong candidates from weak ones, and a quality-risk flag surfaces questions with very low accuracy or high skip rates. Cut those items and re-run.

Why bother with step seven at all? Because the evidence on selection methods keeps moving. A 2022 reanalysis in the Journal of Applied Psychology by Sackett and colleagues revised many long-accepted validity estimates downward after questioning how earlier studies corrected for range restriction. The exact numbers are still argued over in print. The pattern that survives every reanalysis is the one that matters here: structured, standardised evaluation predicts job performance better than an unstructured conversation does. A rubric-scored English task beats a recruiter's impression of a phone call, even when the recruiter is good.

Which roles actually need an English proficiency test?

Any role where English carries the work, which is narrower than most job posts imply. The clearest cases:

• Outsourcing and offshore delivery teams, where a written handover between time zones is the whole coordination mechanism and a vague update costs a day.
• Agency and marketing roles, where the output is English and register control is the job, not a side skill.
• Customer support and inside sales, where listening speed under pressure decides whether the call resolves.
• Client-facing operations in retail, real estate and professional services, where one badly worded message reaches a customer directly.
• High-volume hiring of any kind, where a consistent score is the only way to compare 200 applicants without 200 phone screens.

The test is the wrong tool wherever the job description cannot name an English task. If you cannot write the task down, do not test for it.

If language coverage beyond English matters, the same logic applies per language, and Testlify's language assessments cover several, with CEFR-style reporting available as a report type.

Where do English tests go wrong, and how do you keep them fair?

Four failure modes account for most of it, and all four are fixable at design time rather than after a complaint.

Testing above the job. A C1 bar on a role that needs B1 filters out capable people and narrows your pipeline for no gain. Pick the level from the task list.

Scoring accent or dialect. Comprehensibility is a fair thing to assess. Accent is not. Keep it out of the rubric and out of reviewer notes.

Test conditions that measure nerves. Some candidates speak well and freeze in a timed recording. Allow a stated number of attempts, say what the time limits are up front (a recording cap of five minutes per video answer, for example, set in advance), and give preparation time before speaking tasks.

No route to ask for an adjustment. Testlify supports a candidate-initiated accommodation request that asks whether the candidate is fluent in English and about conditions affecting memory or concentration, then routes it to an administrator who adjusts the session. Be precise about what that is: a human-reviewed request flow, not automatic extra time. Employers switch it on per workspace for candidates with accessibility needs or limited language proficiency.

Pro tip: run your own test on two people already doing the job well and two who struggled. If the test does not separate them, it will not separate your candidates either. This takes an afternoon and it is the cheapest validity evidence you will ever collect.

There is a design consequence to all of this in 2026, now that any candidate can draft a written answer with an AI tool. Spread the weight. A spoken answer and a listening task are harder to outsource than a paragraph of prose, and per-skill reporting (pronunciation insights, for instance, score accuracy, fluency and prosody separately) lets you see where the ability actually sits instead of trusting one blended number.

Ready to build this properly? Testlify's English proficiency assessments run from A1 to A2 basics through advanced levels, with per-skill reporting, rubric scoring and multiple reviewers. Book a demo and bring one job description with you, and the sensible test design for that role usually takes about 20 minutes to settle.

FAQs