The Best Language Assessment Tools for Hiring in 2026

If you need to check whether candidates can actually speak and write a language, there are more options than there were five years ago, and they are less alike than their websites suggest. Some test four skills over an hour. Some test grammar with multiple choice questions. Some score a spoken answer in minutes.

We make one of these tools, so read this with that in mind. We have tried to describe every product here, including our own, by what it is genuinely good at and where it is the wrong choice. Everything below is taken from each vendor's own published material as of October 2026. Prices are left out deliberately: they change often, most are quoted per contract, and comparing them fairly would take more caveats than numbers.

What to decide before you compare anything

Three questions narrow the field faster than any feature table.

Which skills does the job actually use? A support agent on the phone needs speaking and listening. Someone answering tickets needs writing. A tool that tests reading and grammar will rank candidates, but not on the thing the job demands.

Do you need a level, or a pass mark? A CEFR level (A1 to C2) means the same thing to your recruiter, your hiring manager and your client. A proprietary score is only meaningful to people who know that scale.

How many languages, and does anyone on your team speak them? If the answer to the second part is no, you need a tool that scores without you, in every language you hire for, on the same scale.

The tools

Versant by Pearson

The longest-established name in automated English testing, and the one many large contact centres already use.

• Tests: a family of English tests, from a 17 minute speaking and listening test to a 60 minute professional English test, including a four-skill version.
• Scale: the Global Scale of English (10 to 90), which Pearson maps to CEFR.
• Best for: large organisations hiring for English at volume who want a long research record behind the score, and who need listening and reading measured too.
• Worth knowing: the focus is English. The tests are long by screening standards, and the GSE is a scale your hiring managers may need to learn.

EF SET (EF Education First)

A widely used English test from one of the largest language training companies.

• Tests: reading and listening in the standard version, with a four-skill version that adds speaking and writing.
• Scale: 0 to 100, aligned to CEFR, with a certificate for the test taker.
• Best for: benchmarking an existing workforce's English, especially alongside English training, where the certificate is something employees value.
• Worth knowing: English only. The best-known version measures receptive skills, so check which version you are buying if speaking matters for the role.

Emmersion (IXL Learning)

An AI speaking assessment, acquired by IXL Learning in 2022.

• Tests: speaking and writing, about 15 minutes, in a set of languages that includes English, Spanish, French, German, Italian, Japanese, Mandarin, Portuguese and Tagalog.
• Scale: its own 100-point TrueNorth scale, mapped to CEFR, ACTFL and others.
• Best for: teams who want automated speaking scores in several languages and already work with ACTFL as well as CEFR.
• Worth knowing: being part of a larger education company is a stability signal for some buyers, but check the current roadmap for the hiring product specifically.

TestGorilla

A general skills-testing platform with a large library, of which language tests are one part.

• Tests: CEFR-graded language tests in around 16 languages, mostly grammar, vocabulary, reading and listening in a multiple choice format, alongside hundreds of non-language tests.
• Scale: separate tests per CEFR level (a B1 test, a C1 test) with a score on each.
• Best for: teams who want one platform for cognitive, technical and language screening together, and for whom a language check is one step among several.
• Worth knowing: you pick a level's test in advance rather than getting a level back, and multiple choice does not tell you how someone sounds on a call.

Hallo

An AI language platform covering a very wide range of languages.

• Tests: speaking, writing, listening and reading, sold separately, in more languages than anyone else here (their site lists over 90), plus AI interviews and a tutoring product.
• Scale: CEFR, with feedback on fluency, vocabulary, grammar, pronunciation and coherence.
• Best for: organisations that need an uncommon language, or that want all four skills from one AI vendor.
• Worth knowing: the product range is broad, so be clear which product you are evaluating. Their main call to action is a pricing and demo request.

SmallTalk2Me

An AI English assessment focused on speaking and writing.

• Tests: speaking and writing in about 15 minutes, with a detailed report across many measured parameters, and custom branded tests.
• Scale: CEFR A1 to C2.
• Best for: teams hiring for English who want a self-serve tool with a very detailed speaking report.
• Worth knowing: English is the focus, so it fits less well if you hire in several languages.

Evalingo

Ours. An AI assessment of speaking and writing, built for screening candidates before an interview.

• Tests: speaking (about 7 minutes) and writing (about 15), each scored on five dimensions and each given its own CEFR level, in 13 languages on one scale. Questions are read aloud to the candidate in the language being tested, and the interface around them follows the candidate's own language.
• Scale: CEFR A1 to C2, with the ten dimension scores and the reasoning behind each level.
• Best for: hiring teams screening for languages nobody on the team speaks, BPOs screening agents at volume, and anyone who wants to start the same day without a sales call. From $0.60 an assessment on annual billing, ten free to try.
• Worth knowing: we do not score listening or reading as separate skills, we cover 13 languages rather than 90, and we are a young company without the decades of research behind Pearson's tests.

A note on Pipplet

Pipplet, owned by ETS, was a common choice for multilingual testing until ETS closed it on 20 December 2025. If you are replacing it, the main thing to check is that your new tool reports CEFR, so the level requirements already written into your job descriptions keep their meaning. We wrote a short guide to switching from Pipplet.

How to choose

If speaking is the job, test speaking. Support, sales, healthcare front desks and anything on the phone. Multiple choice grammar tests will rank candidates, but not on what the role demands. Versant, Emmersion, SmallTalk2Me, Hallo and Evalingo all score a spoken answer.

If you hire in several languages, insist on one scale. A B2 should mean the same in Spanish and German, or you cannot set one requirement for a multilingual team. Check how many of your languages each tool covers, then check they all report the same way. Our languages page shows how this works for us.

If you need all four skills, that narrows it quickly. Versant's four-skill test, EF SET's four-skill version and Hallo cover listening and reading as well as speaking and writing. We do not.

If you need English only, the field is widest. Every tool here does English. Choose on length, report detail and how it fits your workflow.

Try it on people you already know. Every one of these tools offers some way to test before you commit. Send it to five colleagues or recent hires whose level you are sure of. If the results match what you know, the tool is measuring the right thing. If they do not, no feature list makes up for it.

Frequently asked questions

What is the most accurate language assessment for hiring? There is no single answer, because accuracy depends on what is being measured. A test is accurate for hiring when it measures the skill the job uses, consistently, on a scale your team understands. The fastest way to judge is to run a tool on people whose level you already know.

Is CEFR better than a proprietary score? For hiring, usually. CEFR is the international standard, so a level means the same thing to a recruiter, a hiring manager, a client and the candidate. Proprietary scales can be more granular, but everyone reading the result has to learn them.

Can AI score speaking reliably? For screening, yes, provided every candidate gets the same structure, timings and rubric, and the result shows its reasoning so a person can check it. AI scoring removes the variation between human interviewers, which is often the bigger source of inconsistency.

How long should a language test for hiring take? Long enough to hear unscripted speech and read real writing, and short enough that candidates finish it. Fifteen to twenty-five minutes covers speaking and writing; tests over an hour cost you candidates who have other offers.