A language proficiency assessment for employees works when it does three things: it measures speaking, listening, reading, and writing separately against a CEFR scale, it is administered the same way for every person in the same role, and it is repeated on a fixed schedule. Self-ratings alone cannot meet the first condition, and a placement quiz given once cannot meet the third.
This guide covers the measurement itself: which format to use, what a workplace language assessment should score, how to turn results into training groups, when to retest, and how to keep the whole process fair and private. It sits between two neighboring guides. Our CEFR levels explained article defines what A1 through C2 mean at work. Our guide to building a language training program covers the organizational and role-level needs assessment that decides who gets tested in the first place. Start there if you have not yet decided which teams are in scope.
Why test employees' language skills instead of asking them?
Because people are unreliable judges of their own proficiency, and the error runs in the wrong direction for you. Weaker speakers tend to rate themselves high, and stronger speakers tend to rate themselves low. A 2019 study of 98 primary care providers, published in Medical Care, compared clinicians' self-rated non-English proficiency with a validated oral interview and found only a moderate correlation of 0.51 between the two. The authors concluded that clinicians in the middle bands needed formal testing before being cleared to work with patients in that language. Middle bands are exactly where most of your workforce sits.
Employers know the gap exists. In the ETS TOEIC Global English Skills Report published in March 2026, a survey of 1,300 HR directors across 17 countries, 68 percent of companies admitted they lack a reliable method for assessing English during recruitment, and 55 percent said they plan to test all new hires systematically. The problem is not new. ACTFL's survey of 1,200 US employers found that one in three language-dependent employers reports a language skills gap, and one in four has lost business because of it.
Testing does not close the gap by itself. It tells you where the gap is, how wide it is, and whether the training you buy afterward actually moves it. Without a baseline you have no way to show leadership that progress happened.
What should a workplace language assessment measure?
Four skills, scored separately, and weighted by what the job requires. A single overall score hides the pattern that matters most in business settings: employees who read and write at B2 but speak at A2. That profile is common among engineers, analysts, and anyone who learned the language in school. It is invisible on a written test.
Anchor every score to a CEFR level. The framework is the shared reference for most commercial tests, so a B1 on one vendor's speaking test can be compared with a B1 on another's, at least approximately. It is also the scale that lets you set a target by role and then measure distance to it.
| Role type | Skills to weight most | Assessment task that reveals it |
|---|---|---|
| Customer-facing (retail, hospitality, support) | Speaking and listening | Role-play a complaint or a question with a live examiner or a recorded prompt |
| Plant floor and field crews | Listening, then speaking | Follow a spoken safety instruction and repeat it back; read a warning label |
| Sales and account management | Speaking, then writing | Present a product for two minutes; write a follow-up email to a prospect |
| Technical and back office | Reading and writing | Summarize a procedure document; draft an internal status update |
| Managers of multilingual teams | All four, speaking first | Deliver feedback on a work scenario; answer questions from a direct report |
Two details separate a useful workplace language assessment from a school exam. First, the tasks should look like the job. Reading a warning label tells you more about a forklift operator than a paragraph about vacations does. Second, the test should be long enough to be reliable and short enough that people will finish it. Forty-five to sixty minutes for a four-skill assessment is a reasonable ceiling for working adults.
Self-rating, computer test, or live interview: which format fits?
Most companies end up using two of the three. The self-rating grid screens who needs a real test. The computer-adaptive test sets the baseline for everyone. The live interview goes to the roles where a speaking error costs money or safety.
| Format | What it measures well | Weakness | Typical cost per person |
|---|---|---|---|
| CEFR self-assessment grid | Rough triage; lets employees opt into the right band | Overrating at low levels, underrating at high levels; not defensible on its own | Free |
| Computer-adaptive test (recorded speaking, auto-scored listening and reading) | Consistent baseline across sites; fast turnaround; scales to hundreds | Speaking is scored from recordings, so interaction and repair skills are not observed | Roughly 55 to 65 dollars for a computer-delivered oral test |
| Live oral proficiency interview | Real interaction: clarification, turn-taking, handling an unexpected question | Scheduling; examiner cost; harder to standardize across many raters | Roughly 110 to 140 dollars per interview |
The cost figures come from the published federal price list for ACTFL assessments, where a computer-delivered oral interview runs about 57 to 62 dollars and a live examiner interview about 109 to 137 dollars. Other vendors price differently, but the ratio holds: a human examiner costs roughly twice a recorded one. For a cohort of 100 employees, that is the difference between about 6,000 and about 12,000 dollars for the speaking component alone, which is why a tiered approach usually wins.
Proctoring is the other decision. Unproctored online tests are fine for placement, where the employee has no incentive to cheat and every incentive to land in the right group. They are not fine for hiring decisions, promotions, or anything that goes in a personnel file. For those, use a proctored session or a live examiner.
How do you place employees into training groups?
Group by speaking level first, then by job function. Speaking is the skill that most often lags, and a class where half the room is A2 and half is B2 in conversation serves neither half. Reading and writing gaps can be handled with differentiated homework inside a group, but a speaking gap forces a split.
Hold each group to a single CEFR level, or at most a half-level spread, and keep it small enough that everyone talks. Six to eight learners is a practical ceiling for a speaking-heavy class. Where the headcount at one level is too small to form a group, private or paired sessions cost more per hour but far less than an oversized mixed class that nobody finishes.
Placement has one more job: it sets the target. A customer service representative who tests at A2 with a role requirement of B1 has a defined distance to cover, and that distance can be planned in hours. LanguageCert, an awarding body that publishes guided learning hours per CEFR level, estimates about 180 hours of guided study for the B1 level and another 180 for B2. Treat those as planning figures, not promises, and match them against the weekly hours your schedule can actually protect.
Retest at three points: baseline, end of block, and on a fixed calendar
Retest before training starts, at the end of each training block, and on a fixed calendar for roles where proficiency is a job requirement. Each retest should use the same instrument as the baseline, or a documented equivalent, so that movement reflects the learner rather than the test.
The end-of-block retest is the one companies skip, and it is the one that proves the program worked. If your training runs in twelve-week blocks, retest speaking and the weakest secondary skill at week twelve. Advancing a full CEFR level takes longer than one block for most adults, so use a sub-level scale (A2.1, A2.2, and so on) or a scored band inside each level, otherwise you will report "no change" for six months and then a jump.
For standing requirements, borrow the convention testing bodies already use. Language Testing International, which administers ACTFL assessments, treats test results as valid for two years. The logic is sound: proficiency can decline as well as improve, so a certificate describes the person on the day they were tested. Two years is a sensible default for a bilingual pay premium or a certified-interpreter designation. Twelve months is better for safety-critical roles where the language is used under pressure and rarely practiced otherwise.
Whatever cadence you set, the results should feed the numbers you report to leadership. Our guide on how to measure language training ROI shows how proficiency gains sit alongside business metrics in that report, and our 90-day implementation plan shows where each retest lands in a rollout calendar.
Fairness, privacy, and Title VII: the rules for testing employees
Test the requirement, not the person. Under US law, a language proficiency requirement is a national origin issue, and the Equal Employment Opportunity Commission's guidance is direct: an English fluency requirement is permissible only if required for the effective performance of the position, assessed case by case. Applying one uniform standard across dissimilar jobs, or demanding more fluency than the job needs, can violate Title VII. The same logic protects you when you test for Spanish or French: tie the required level to documented job tasks, write it down before anyone is tested, and apply it to everyone in the role.
Four practices keep an assessment program on the right side of that line and keep employees willing to take part:
- Separate placement from evaluation. Tell employees, in writing, that placement results are used to build training groups and are not part of performance reviews. Then honor it.
- Limit who sees individual scores. The learner, the training coordinator, and the instructor need the level. The manager needs the target and the progress. Leadership needs aggregates, and no one at any level should receive a ranked list of names.
- Score accent as communication, not as identity. The EEOC guidance also addresses accent: it may count against a person only when it materially interferes with the job. Rubrics should score whether the message was understood, not how it sounded.
- Offer accommodations. Extra time, a quiet room, or a live examiner instead of a recorded one for someone with a hearing or processing difference. Ask before the test day, not after.
If you outsource the testing, put these rules in the contract. A vendor that cannot describe its rater training, its score confidentiality, and its accommodation process will struggle with all three when it matters. Our checklist for evaluating a language training vendor includes the assessment questions to ask.
How to test employees' language skills: the checklist
- Write the required CEFR level per role, tied to specific job tasks, before testing anyone.
- Choose the format per tier: self-rating grid to screen, computer-adaptive test for the baseline, live interview for high-stakes roles.
- Score speaking, listening, reading, and writing separately. Report the profile, not one number.
- Use job-like tasks and keep the session under an hour.
- Place by speaking level first, then by function. Cap speaking groups at eight.
- Set the retest calendar in the same document as the baseline: end of each block, plus a standing cadence for certified roles.
- Publish the confidentiality and accommodation rules to employees before the first test.
- Store baseline, target, and each retest together so progress and ROI can be reported from one record.
Done this way, assessment stops feeling like a hurdle and becomes the first concrete step of a program employees actually finish. It is also the step that makes every other part of corporate language training measurable.
Assess your team with Edlingo
Every Edlingo corporate program begins with a four-skill, CEFR-aligned placement assessment, delivered online with a live speaking component where the role calls for it. We set a target level per role with you, build groups by speaking level, retest at the end of each block on the same scale, and report progress against the baseline. Individual results stay with the learner and the coordinator. Leadership sees the cohort.
Request a Free Team Assessment →
Still defining which roles need which level? Start with CEFR levels explained. Budgeting the program that follows? See corporate language training cost, or contact us to talk through your situation.