12 min read

CEFR Speaking Rubric Apps for Teachers: Build, Test, Moderate

Use CEFR speaking descriptors responsibly with task alignment, observable rubrics, app evidence, rater training, moderation, fairness, and appeals.

A CEFR-speaking app is credible only when the task, descriptors, scoring evidence, and decision align. CEFR is a framework of language activities and can-do descriptors, not a universal algorithm that turns any recording into an official level.

Use Council of Europe materials as the primary source. Commercial alignment claims require separate evaluation.

Key takeaways

  • Start with the communication activity, not a level label.
  • Select relevant CEFR descriptors rather than copying every category.
  • Write observable task evidence and test it on samples.
  • Moderate human ratings and audit automated outputs.
  • Distinguish classroom estimates from certified exam results.
  • Explain privacy, limitations, and review rights to learners.

In this guide: CEFR map | task | rubric | apps | moderation | fairness | reporting | template | pilot | FAQ

What CEFR speaking covers

The Council of Europe CEFR Companion Volume expands descriptors for communicative language activities and strategies, including oral production, oral interaction, mediation, online interaction, and phonological control. The framework also describes linguistic, sociolinguistic, and pragmatic competence.

The commonly cited qualitative table describes:

  • Range.
  • Accuracy.
  • Fluency.
  • Interaction.
  • Coherence.

These dimensions are useful, but a classroom task may also need task completion, intelligibility, evidence use, or professional appropriateness.

Start with the speaking task

Define:

  1. Mode: production, interaction, or mediation.
  2. Audience and purpose.
  3. Information available.
  4. Preparation and time.
  5. Expected outcome.
  6. Target level or progression.
  7. Evidence collected.

Example interaction task:

Learner resolves a fictional booking problem by clarifying details, proposing options, responding to one changed condition, and confirming the final action.

Relevant evidence includes turn-taking, clarification, response, repair, intelligibility, and task completion. A read-aloud score would not cover the construct.

Build the rubric

Select descriptors

Search the official descriptor bank for the activity and level. Preserve the can-do focus.

Translate into observable evidence

Vague:

"Good B1 interaction."

Observable:

"Responds to straightforward follow-up questions, asks for clarification when needed, and completes the familiar transaction with occasional support."

Define adjacent bands

Explain the difference between levels through performance, not adjectives.

Add task completion

CEFR language evidence does not automatically state whether the learner achieved the classroom outcome.

Test with samples

Ask raters to score independently and mark ambiguous wording.

Evaluate app evidence

An app claiming CEFR alignment should answer:

  • Which CEFR version and descriptors are used?
  • Which speaking modes and task types are sampled?
  • How long and varied is the evidence?
  • Does the score cover interaction or only monologue?
  • How was mapping established and validated?
  • Which learner groups were included?
  • How do microphones, noise, accent, age, and disability affect results?
  • Can a teacher inspect the performance and rationale?
  • Is human review available?
  • Is the result formative, institutional, or officially recognised by anyone?

Do not infer answers from a CEFR logo.

Train and moderate raters

Use:

  • Rubric orientation.
  • Anchor performances at several levels.
  • Independent trial rating.
  • Discussion of major differences.
  • Notes on boundary cases.
  • Double-scoring sample.
  • Periodic drift checks.
  • Written decision log.

Moderation question:

"Which exact performance evidence supports this band, and which descriptor rules out the adjacent band?"

Human judgment becomes more defensible when reasoning is visible.

Test fairness and technical conditions

Pilot across:

  • Proficiency range.
  • Expected accents.
  • Devices and microphones.
  • Quiet and normal classroom noise.
  • Different speaking rates.
  • Relevant accessibility needs.
  • Familiar and unfamiliar topics.
  • Repeated attempts.

Check whether one group receives systematic transcription or pronunciation penalties unrelated to the target construct. Investigate instead of averaging.

Use speaking assessment apps for teachers and pronunciation-feedback tools for testing workflows.

Report levels without overclaiming

Good classroom report:

"On this prepared B1-level presentation task, the performance met the class rubric for organisation and intelligibility and approached the interaction descriptor during questions. This is a formative classroom estimate, not an official CEFR certificate."

Weak report:

"The student is permanently B1."

State:

  • Task and date.
  • Preparation conditions.
  • Descriptor source.
  • Criteria and evidence.
  • Formative or summative purpose.
  • Role of automation.
  • Limitations.
  • Next can-do target.
  • Review route.

Do not combine incompatible tasks into one unexplained level

A prepared presentation and an unprepared conversation create different evidence. The first may show organisation, range, and sustained production; the second may show listening, turn-taking, clarification, and repair. If both matter, report the task results separately before making any broader classroom judgment.

For example, a learner may organise a strong prepared B2-style presentation while needing substantial support in spontaneous interaction. Averaging two proprietary numbers into one level can hide that instructional need. Use multiple tasks only when the assessment blueprint explains why each one is included, how it is weighted, and which descriptor evidence it supplies. Tell learners whether a result refers to one performance, one mode, a course decision, or a broader estimate. This makes the next learning target more useful than a single label without context.

Classroom rubric template

Category Emerging Developing Secure Extending
Task completion Outcome mostly absent Partial outcome Outcome achieved Outcome achieved and adapted
Intelligibility Frequent listener effort Some repeated effort Key meaning generally clear Meaning clear under complexity
Range Very limited forms Sufficient for simple parts Sufficient for task Flexible and precise for task
Accuracy Errors often obscure Errors sometimes interfere Control supports meaning Consistent control under demand
Fluency Breakdown blocks task Uneven but continues Maintains usable flow Adapts pace and reformulates
Interaction Limited response Responds with support Initiates, responds, clarifies Builds and manages interaction
Coherence Ideas isolated Some useful linking Organised relevant message Flexible organisation and emphasis
Repair Abandons problem Repairs with support Clarifies or rephrases Anticipates and resolves ambiguity

Replace band wording with level- and task-specific descriptors. Do not label this table official CEFR certification.

Seven-day pilot

Day Action Evidence
1 Define construct and task Alignment statement
2 Select official descriptors Source record
3 Draft observable bands Rater interpretation notes
4 Score anchor samples Agreement and ambiguity
5 Test app and audio conditions Error patterns
6 Moderate and revise Decision log
7 Explain to learners Transparency and review route

Where SpeakShark fits

SpeakShark can provide formative conversational evidence and feedback. It does not issue an official CEFR certificate and should not be used for consequential level decisions without validation and human review.

Use AI speaking assignments, best AI tools for teachers, and large-class speaking activities.

Start a fictional formative task, compare evidence with your CEFR-informed rubric, then moderate the sample with another teacher.

How we researched this guide

We used current Council of Europe CEFR descriptor, Companion Volume, self-assessment, and qualitative spoken-language resources. The classroom template is our adaptation and is explicitly not an official certification instrument. Commercial alignment claims require separate evidence.

Sources

  1. Council of Europe CEFR descriptors, reviewed October 21, 2026 for official can-do descriptor resources.
  2. CEFR Companion Volume, reviewed for production, interaction, mediation, phonology, and competence scales.
  3. CEFR qualitative spoken-language table, reviewed for range, accuracy, fluency, interaction, and coherence.
  4. CEFR and European Language Portfolio, reviewed for activities, domains, levels, and can-do self-assessment context.

FAQ

Is there an official CEFR speaking rubric app?
The Council of Europe publishes CEFR descriptors and reference materials, not one mandatory commercial grading app. Products may claim CEFR alignment, but teachers should inspect the task, mapping method, evidence, validation, and level interpretation. A vendor label does not make an automated score an official CEFR result.
What CEFR categories can teachers use for speaking?
The CEFR provides descriptors across spoken production, spoken interaction, strategies, mediation, and communicative competence. Its qualitative spoken-language table includes range, accuracy, fluency, interaction, and coherence. Teachers should select only dimensions that match the task and add task completion or intelligibility where needed.
How do teachers turn CEFR descriptors into a classroom rubric?
Choose the target activity and level, identify relevant can-do descriptors, rewrite them as observable task evidence without changing the intended construct, define several performance bands, test them against sample performances, train raters, and revise ambiguity. Cite the source and avoid claiming formal certification.
Can an app assign a CEFR speaking level automatically?
Some products report CEFR-aligned outputs, but consequential use requires evidence about task coverage, scoring model, reliability, validity, learner population, audio conditions, and human review. A short monologue cannot represent every CEFR mode, especially interaction and mediation, simply because software returns a level.
How should teachers moderate CEFR speaking scores?
Use anchor performances, independent rating, discussion of major differences, periodic drift checks, and a written decision log. Sample across tasks, levels, raters, and learner groups. Keep the original performance available under approved retention rules and provide a review route for contested consequential scores.
What should teachers tell students about CEFR app scores?
Explain what task was completed, which descriptors were used, whether the result is formative or official, how automation contributed, known limitations, how recordings are handled, and how to request review. Do not present a practice-app estimate as a certified exam result or permanent statement about ability.

The Council of Europe publishes the framework and descriptors, not one mandatory commercial app. Vendor alignment claims require evidence.

Select relevant production, interaction, mediation, strategy, range, accuracy, fluency, interaction, coherence, and phonological descriptors for the task.

Choose activity and level, select descriptors, write observable evidence, define bands, test samples, train raters, and cite the source.

It can output a claim, but consequential use needs task coverage, validation, reliability, learner-population evidence, technical testing, and human review.

Use anchors, independent rating, discussion, drift checks, sample double-scoring, decision logs, approved retention, and a review route.

Explain task, descriptors, purpose, automation, limitations, data handling, review rights, and whether the result is formative or officially recognised.

SpeakShark details

We are a speaking improvement tool. SpeakShark is not an exam preparation provider and is not affiliated with any employer, school, test provider, or any exam board. For official equipment, monitoring, identification, score, and test-day rules, go to the exam body, test provider, administrator, or invitation. Use SpeakShark to make your spoken English stronger, and use official material to learn the test.

Curious how it works? Explore SpeakShark's features or see plans and pricing.

Keep reading

English Classroom Instructions for Teachers: Clear Phrase Guide

Give clear classroom instructions in English with attention, outcome, steps, grouping, timing, checking, transitions, correction, and closing phrases.

English Pronunciation Feedback Tools for Teachers: 2026 Guide

Compare pronunciation feedback tools for sounds, stress, rhythm, intelligibility, recording, classroom use, teacher review, privacy, and grading.

Speaking Assessment Apps for Teachers: 2026 Evaluation Guide

Evaluate speaking assessment tools for tasks, rubrics, recordings, pronunciation, interaction, teacher review, moderation, privacy, and appeals.