
To evaluate English proficiency in a job interview, score the candidate's spoken answers on four separate criteria: grammar, vocabulary, communication effectiveness and conversational flow. Rate each one from 1 (Poor) to 5 (Excellent) against written descriptions of what each score means. Then set the bar each role really needs and check the evidence before you decide.
Language proficiency is how well a person can use a language to get real tasks done, such as explaining a problem, answering a follow-up question or calming an upset customer. A scoring rubric is a table of criteria and score levels that tells every reviewer what a 1, a 3 or a 5 looks like. Without one, two interviewers can hear the same answer and give very different ratings. With one, every candidate is judged against the same standard.
This guide gives you a ready-to-use 5-point rubric, a worked scoring example, a comparison with the CEFR and IELTS frameworks, and a simple process for reviewing scores fairly.
Key takeaways
- Score speaking on four criteria (grammar, vocabulary, communication effectiveness and conversational flow), each from 1 (Poor) to 5 (Excellent), instead of one gut-feel rating.
- Write what each score means for the role. Ask for a high level only where the job needs it: EEOC guidance says an English fluency requirement must be needed to do the job well.
- The four criteria line up closely with the CEFR's qualitative aspects of spoken language and the IELTS speaking criteria, but there is no official conversion between these scales.
- In InterviewFlowAI, the AI rates these four criteria during each interview. Scores appear on the Communication tab of the candidate report, and a person checks the recording and transcript before deciding.
The 4 Key Dimensions of Language Evaluation
A fair assessment splits language into four parts. This way you measure both correct language and how well the candidate uses it in a real conversation.
- Grammar & Syntax: Evaluates the structural correctness of the candidate's sentences.
- Vocabulary & Expression: Measures the range, precision, and professional appropriateness of the words chosen.
- Communication Effectiveness: Assesses how clearly and impactfully the candidate delivers their message.
- Conversational Flow: Looks at the natural cadence, relevance, and coherence of the dialogue.
The 5-Point Scoring Rubric
Each dimension uses the same 1 to 5 scale, from "Poor" (1) to "Excellent" (5). Use the descriptions below to train your interviewers or to set up an AI screening tool. In InterviewFlowAI, these are the four criteria the AI scores. Owners can rewrite what each score means for a role, or use AI Tune to draft it.
1. Grammar & Syntax
Correct sentences are the base of clear communication. This criterion checks how well the candidate builds them.
- 5 (Excellent): Exceptional grammatical control with sophisticated sentence structures.
- 4 (Good): Strong grammatical accuracy with minor errors that don't impede communication.
- 3 (Average): Generally correct grammar with occasional errors that don't affect meaning.
- 2 (Below Avg): Noticeable grammatical issues that may impact communication effectiveness.
- 1 (Poor): Frequent grammatical errors that significantly hinder communication.
2. Vocabulary & Expression
This criterion asks whether the candidate has the words they need to do the job well.
- 5 (Excellent): Sophisticated, precise vocabulary with excellent professional terminology.
- 4 (Good): Strong vocabulary range with appropriate professional language.
- 3 (Average): Adequate vocabulary for professional communication.
- 2 (Below Avg): Limited vocabulary that occasionally affects clarity.
- 1 (Poor): Very basic vocabulary that significantly limits communication.
3. Communication Effectiveness
Knowing the words is only half the job. This criterion checks whether the candidate can get their point across to the people they will work with.
- 5 (Excellent): Exceptionally clear and compelling communication for professional settings.
- 4 (Good): Clear, well-structured communication with strong professional impact.
- 3 (Average): Generally effective communication that meets professional standards.
- 2 (Below Avg): Communication challenges that may impact professional effectiveness.
- 1 (Poor): Significant communication barriers that would hinder workplace performance.
4. Conversational Flow
This criterion covers the live side of speaking. It looks at how well the candidate responds in a real-time conversation.
- 5 (Excellent): Natural, engaging conversational flow with excellent response relevance.
- 4 (Good): Smooth conversational flow with well-connected ideas.
- 3 (Average): Generally coherent flow with minor hesitations that don't impede understanding.
- 2 (Below Avg): Noticeable flow issues that affect effectiveness.
- 1 (Poor): Poor conversational flow that significantly impacts performance.
How the Rubric Compares With CEFR and IELTS
The Common European Framework of Reference for Languages (CEFR) is the Council of Europe's six-level scale for language ability, from A1 (basic user) to C2 (proficient user). Its Table 3 rates spoken performance on five aspects: range, accuracy, fluency, interaction and coherence. The IELTS Speaking test uses four equally weighted criteria, each scored in bands from 0 to 9.
The four rubric criteria cover much of the same ground:
| Rubric criterion | Closest CEFR aspect (Table 3) | Closest IELTS Speaking criterion |
|---|---|---|
| Grammar & Syntax | Accuracy | Grammatical range and accuracy |
| Vocabulary & Expression | Range | Lexical resource |
| Communication Effectiveness | Coherence | Fluency and coherence (the coherence part) |
| Conversational Flow | Fluency and interaction | Fluency and coherence (the fluency part) |
| No separate criterion | Phonology (added in the 2020 Companion Volume) | Pronunciation |
Two differences matter. First, there is no official conversion from a 1–5 rubric score to a CEFR level or an IELTS band, so don't treat a 4 as "B2". Second, the rubric has no separate pronunciation or accent score. That fits hiring, where the question is whether people can understand the candidate, not whether they sound like a native speaker.
If your company already writes job requirements in CEFR terms, you can borrow its wording. For example, the CEFR's B2 accuracy descriptor says the speaker "does not make errors which cause misunderstanding". That line would work well as the description of a 4 in Grammar & Syntax for a customer-facing role.
Worked Example: Scoring One Answer
Here's how a reviewer would apply the rubric to one answer. The role is customer support, and the question is "Tell me about a time you calmed down an upset customer." The answer below is illustrative, not taken from a real candidate.
"Last year a customer called because his order was late for the third time. He was very angry. First I listened and I didn't interrupt him. Then I said sorry and checked the tracking while he was on the line. The parcel was stuck in warehouse, so I asked them to send it again with express delivery. I also gave him a discount code. He thanked me, and he ordered again next month."
| Criterion | Score | Why |
|---|---|---|
| Grammar & Syntax | 4 (Good) | Mostly accurate. Two small slips ("stuck in warehouse", "next month") don't block the meaning. |
| Vocabulary & Expression | 3 (Average) | Clear, job-relevant words ("tracking", "express delivery", "discount code"), but a narrow range and repeated "and I". |
| Communication Effectiveness | 4 (Good) | A clear story in order: the problem, what they did and the result. |
| Conversational Flow | 4 (Good) | Stays on the question, and each idea leads to the next. |
Now compare those scores with the bar for the role. Say the support team agreed in advance that Communication Effectiveness and Conversational Flow need at least a 4, and Grammar & Syntax at least a 3. This candidate meets the bar. A technical writing role might need a 4 in Vocabulary & Expression, so the same answer would fall short there.
One answer is only a sample. In InterviewFlowAI, the AI rates the whole interview, so a reviewer should look at several answers before trusting any single score.
How to Set the Bar for Each Role
The right level depends on the job, and so does what's legal. In the U.S., the EEOC's guidance on national origin discrimination says an English fluency requirement is allowed only if it's needed to do the job well. It also warns that one uniform requirement across very different jobs, or a higher level than a job needs, may break Title VII.
The EEOC gives an example. A person may speak English well enough to work as a research assistant but not yet well enough to be a senior scientific writer. The same candidate can pass one bar and miss another, and that's fine as long as each bar fits its job.
Use these steps to set your bar:
- List the speaking tasks. Write down what the person will actually do in the language: customer calls, team meetings, client demos or support chats.
- Pick the criteria that matter most. InterviewFlowAI's own guidance gives an example: a support role might weigh conversational flow, while a technical writing role might weigh vocabulary.
- Set a minimum score for each criterion. Agree on it before interviews start, and write it down so every reviewer uses the same rule.
- Judge clarity, not accent. The EEOC says a merely noticeable accent is not the same as one that gets in the way of the spoken communication a job needs. Write your descriptions around being understood.
- Tune the rubric wording. In InterviewFlowAI, open the interviewer, click Language, edit the descriptions under Scoring Rubric (or click AI Tune), then click Save Settings. Only Owners can change the rubric.
Language is usually one part of a wider screen. For a technical role, read how InterviewFlowAI evaluates skills in a conversational interview. For an engineering example, see how AI interviews screen backend developers.
How Scores Are Reviewed in InterviewFlowAI
A language score is a starting point for a decision, not the decision itself. Here's how a reviewer works through it:
- Open the candidate report. The Communication tab shows the four language scores, each out of 5.
- Check the evidence. Play the recording at 0.75× to 2× speed, search the transcript for a word, or click a transcript line to jump to that moment.
- Compare candidates. Ask FlowAI can summarize one candidate or compare up to 10 side by side.
- Decide. An Owner accepts or declines the candidate. The score informs that choice but doesn't make it.
- Recalibrate. If scores look too strict or too lenient for the role, an Owner edits the rubric descriptions so the scores reflect what the role needs.
Automate Your Evaluations with InterviewFlowAI
A fixed scale takes the guesswork out of screening, but scoring every interview by hand takes time. InterviewFlowAI runs first-round AI interviews and scores language on the same four criteria every time. For a wider view of the category, see our guide to how AI interview software works, what it costs and how to choose.
Plans start at $59 a month. On monthly Growth and Scale plans, an interview costs about $0.99 ($0.83 billed yearly), and an AI interview of up to 10 minutes uses one credit. AI resume screening is included on every plan. Proctoring for cheating detection is included at no extra cost.
Beyond language scoring, the platform covers the rest of the first-round screen:
- Custom AI interviewers: Video interviews on every plan, and phone interviews on Growth and above, where the AI calls candidates from your own connected Twilio number. Both use realistic AI voices, and video uses avatars.
- Integrations: Native integrations with Greenhouse, Ashby, Teamtailor, Workable, Recruit CRM and Zapier on Growth and above, plus webhooks and API integrations on Scale.
- Ask FlowAI: Ask questions about interview data, summarize a candidate, and compare up to 10 candidates side by side.
- Sourcing and outreach: Search for candidates with Talent Search (Growth and above) and send automated outreach sequences (Scale) from the same platform.
- Tailored experiences: Customize interview stages and application pages, and use skills-based or custom questions.
Clear rubrics plus AI scoring give your team the same evidence for every candidate, so decisions get faster and more consistent.
Methodology: This guide is based on InterviewFlowAI's product documentation and the public frameworks listed below: the Council of Europe's CEFR tables, the IELTS Speaking criteria and the EEOC's guidance, each checked on 7 November 2026. The worked example is illustrative and does not come from a real candidate.
Sources
- InterviewFlowAI Knowledge Base, Language proficiency, checked 7 November 2026
- InterviewFlowAI Knowledge Base, Candidate report, checked 7 November 2026
- InterviewFlowAI Knowledge Base, Plans, credits, and billing, checked 7 November 2026
- InterviewFlowAI Knowledge Base, Integrations, checked 7 November 2026
- InterviewFlowAI Knowledge Base, Ask FlowAI, checked 7 November 2026
- InterviewFlowAI Knowledge Base, Talent search, checked 7 November 2026
- Council of Europe, Qualitative aspects of spoken language use, Table 3 (CEFR 3.3), checked 7 November 2026
- Council of Europe, The CEFR Levels, checked 7 November 2026
- Council of Europe, CEFR Companion Volume (2020), Appendix 3: Qualitative features of spoken language (expanded with phonology), checked 7 November 2026
- IELTS, IELTS scoring in detail, checked 7 November 2026
- IELTS, IELTS Speaking key assessment criteria, checked 7 November 2026
- U.S. Equal Employment Opportunity Commission, EEOC Enforcement Guidance on National Origin Discrimination, checked 7 November 2026
Mukul Munjal
Founder & CEO, InterviewFlowAI
I'm Mukul Munjal, founder and CEO of InterviewFlowAI. I have 7 years of experience in engineering, 3 of them building voice agents, and I lead the team that builds our AI interviewer. Before publishing, I check competitor prices and features against each vendor's own pages.


