FluencyScreen
← All articles

How to assess a candidate's spoken English

If a role involves talking to people, you need to assess candidate English the way the job does — out loud, under time pressure, in real conversation.

Resumes tell you someone studied English. They don't tell you whether they can hold a customer call. A certificate, a self-rating of "fluent," or a clean cover letter all describe the candidate at their most prepared. The job rarely happens at their most prepared.

So the question is narrow and practical: can this person speak well enough to do the work? This is how to answer it without a 30-minute panel interview for every applicant.

Why resumes and written tests miss speaking

Written English and spoken English are different skills, and people are rarely equal at both. A candidate can edit a paragraph for ten minutes, run it through a tool, and produce something polished. None of that survives a live call, where there is no time to revise and no second draft.

Multiple-choice grammar tests have the same gap. They measure whether someone recognizes a correct sentence, not whether they can build one while a customer is waiting. You can score full marks on a written quiz and still freeze when asked to explain a refund policy out loud.

  • A resume shows the best-case version, written and reviewed.
  • A written test measures recognition, not production under pressure.
  • A self-reported level ("C1," "fluent") is a guess, not a measurement.
  • Speaking is the part of the job you most need to verify — and the part these miss.

What to listen for: the six skills

"Good English" is too vague to score consistently. Break it into parts so two people listening to the same answer reach the same conclusion. To test spoken English in a way that holds up, listen for six things:

  • Grammar — are sentences structured well enough to be understood, even with small errors?
  • Vocabulary — is the range wide enough for the role, or does it run out on anything specific?
  • Fluency — does speech flow, or stall in long pauses and restarts?
  • Coherence — do answers stay organized and on point across a few sentences?
  • Comprehension — did they actually understand the question, or answer a different one?
  • Task completion — when asked to do something, like explain or summarize, did they get it done?

Note what is not on this list: accent. Accent is not proficiency, and screening for it is both unfair and a hiring mistake — plenty of clear, capable speakers have strong accents. Judge whether you can follow the meaning, not where the person learned to speak.

Manual vs. automated assessment

You can assess candidate English by hand: schedule a call, ask a few questions, and grade against a rubric. It works, and a structured human conversation is hard to beat for senior or unusual roles. The problem is volume. Live calls don't scale, scheduling eats days, and the early-round assessor is rarely your most consistent grader.

Automated screening trades some of that nuance for speed and consistency. The candidate opens a one-time link, talks to an AI voice interviewer for ten to fifteen minutes, and answers around ten questions — no account, no scheduling. The transcript is scored against the CEFR scale on the six skills above, and you get a level, per-skill scores, strengths and weaknesses, a recommendation, and the recording.

The honest framing: use automation for the first pass, to decide who is worth a human conversation. It removes the people who clearly can't do the speaking part and lets you spend live time on the ones who can. You can see a sample report to judge whether the output is detailed enough for your decisions.

A simple rubric you can use

Whether you grade by hand or read an automated report, anchor it to CEFR so "good enough" means the same thing every time. The scale runs A1 to C2: A1 and A2 are basic, B1 and B2 are independent, and C1 and C2 are proficient. Most hiring bars sit in the independent band.

  • A1–A2 (basic): can handle simple, predictable exchanges. Usually below the bar for customer-facing or collaborative work.
  • B1 (independent, lower): can manage routine tasks but strains on anything unexpected. Fine for some internal roles.
  • B2 (independent, upper): can hold a fluent work conversation without strain. A reasonable bar for most customer and team-facing jobs.
  • C1–C2 (proficient): comfortable with nuance, persuasion, and complex topics. Worth it for senior, sales, or high-stakes roles.

Pick the band the role actually needs and write it into the job — "B2 or above for support," for example — rather than leaving it to each interviewer's gut. Then the assessment isn't a vibe; it's a level you can compare across every candidate.

One caution: the candidate doesn't see their own score, so the report is a hiring tool, not a coaching one. If you want to set the level requirement per role and run the first pass automatically, the pricing page lays out the screening volume each plan covers.