HandsOn AI

Measure Japanese speaking proficiency,
simply and accurately.

A CEFR-aligned Japanese speaking proficiency test — the speaking counterpart to JLPT. Results come back as fast as the same day, and candidates take it on a phone or PC in about 20 minutes.

Contact Us
Admin panel candidate list (Japanese interface): level distribution at the top, and each candidate’s CEFR level and status below
7,500+
test takers to date
20 languages
for result reports
Same day
fastest turnaround for results
Admin panel
Test distribution and results in one place

A shared yardstick for evaluating speaking

First page of the result report (Japanese): overall CEFR level, T-score and group rank, and the Can-do list
Feedback on each indicator (Japanese): range, accuracy and fluency, each with the current state and points to improve, quoting the candidate’s actual speech
An excerpt of the feedback on individual indicators (Japanese)

Assess speaking with CEFR levels and Can-Do statements

  • A CEFR level for speaking proficiency

    An overall level from A1 to C1+, with achievement shown for each question.

  • Can-do statements that make ability concrete

    See what the candidate can do in specific situations — such as stating a destination at a station information desk — marked Achieved, Almost or Not yet.

  • Five indicators that show what to work on

    Written feedback on range, accuracy, fluency, interaction and coherence explains the current state and points to improve.

Task achievement comes first. The assessment considers both whether the candidate achieved the task and how they communicated.

T-score and rank show relative standing

Two candidates at A2.2 can still be far apart. The report pairs a T-score — the candidate’s position in the distribution of all test takers — with a rank within the same test group, giving you a basis for selection, class placement and scholarships.

Overall result card (Japanese): CEFR level A2.2, T-score 50.7 (top 47% of all test takers), group rank 105 of 118, and a level bar

T-score

Position among all test takers

The mean is 50. It shows where a candidate sits within the same CEFR level and the top percentage they fall into, and it is comparable across classes and schools.

Group rank

Rank within the same test group

Calculated per test group — a class, a hiring round — with ties sharing a rank. Sort by it in the admin panel to pull out top performers or candidates who need follow-up.

The T-score and rank appear in both the candidate report and the admin panel.

Reports translated into 20 languages

Result reports can be delivered in the test taker’s own language, including the Can-do list and the feedback on each indicator (current state and points to improve). Candidates understand their own challenges correctly, and reports are easy to share with Sending Organizations and families.

Report in Burmese: the level and Can-do list in the candidate’s language
Burmese

Academic supervision

Designed under the supervision of our academic advisor, for accurate CEFR and Can-do judgments.

Uichi Kamiyoshi

Musashino University, Faculty of Global Studies, Department of Japanese Language Communication, Professor

Uichi Kamiyoshi

Background

Graduated from Tokyo Gakugei University, Faculty of Education. Completed doctoral coursework at Osaka University, Graduate School of Language and Culture. After serving as an elementary school teacher in Kitakyushu, Senior Japanese Language Specialist and Chief Consultant at the Association for Overseas Technical Cooperation and Sustainable Partnerships (AOTS), and specially appointed lecturer at Nagasaki University of Foreign Studies, he joined Musashino University as Associate Professor in April 2016, and has held his current position since April 2023. His specialties include language education policy and community Japanese language education.

Test Structure

Listen, explain and respond. Four task formats assess speaking in situations close to real conversation.

About 20 minutesFully online

  1. PART 1

    Speech

    Share ideas on a given topic.

  2. PART 2

    Visual speech

    Describe the information in a graph or diagram.

  3. PART 3

    Listening & speech

    Listen, answer comprehension questions, and give a short speech.

  4. PART 4

    Role-play

    Respond to automated voice prompts in a given situation and role.

Test screen on a smartphone: listening instructions and tabs for speech, visual speech, listening and role-play

Review test recordings

Review recordings in the admin panel to check how candidates answered, including for entrance exams.

Use Cases

For every stage of the workforce pipeline

Japanese Language Schools

KAI Japanese Language School
KAI Japanese Language School
JILC
JILCCase study (Japanese)
  • Placement test
  • Term-end test for the conversation course
  • Level report at completion

Recruitment Agencies & Sending Organizations

Dive
DiveCase study
ASIA to JAPAN
ASIA to JAPANCase study (Japanese)
  • Candidate screening
  • Proof of level before and after placement

Training Companies & Hiring Companies

Originator
Originator
  • Quantitative measurement before and after training
  • Setting the right training goals

Rank, T-score and test history,
in one admin panel

Distribute tests, track completion and manage results in one place. No collecting papers, scoring or transcribing — results accumulate as test history automatically.

Distribute tests and send reminders

Choose a difficulty and issue test URLs. Invitation emails and re-test links are sent from the admin panel too

Completion status and level distribution

See at a glance who hasn’t finished, who is awaiting scoring, and how many candidates sit at each level — by school or class

Group rank, T-score and CEFR level in one list

Sort and filter to pull out top performers or candidates who need follow-up

Test history and progress

Line up a candidate’s past tests and see how far they have come since the first one

Admin panel candidate list (Japanese interface): level counts at the top, and columns for CEFR level, T-score, group rank, learning history and qualifications

Detect signs of misconduct,
keep assessments fair

The system detects signs of misconduct in online testing to help keep assessments fair. It records when a candidate switches to another screen during the test, and for tests that record the candidate’s face it also detects signs in the video. Detections are shown in the admin panel to help you review results.

Misconduct signals in the admin panel (Japanese interface): video analysis (one detection, no missing footage), one screen switch, and a timeline showing a hidden page and a freeze with links to the footage

Switching screens

Records when the candidate switches to another tab or window during the test

Multiple people in frame

Detects signs that someone other than the candidate appears in the video

Face not visible

Detects periods when no face can be confirmed in the video

Video stopped or blocked

Detects signs that the camera stopped or was covered

Automated detection is reference information for review. A detection alone does not establish misconduct.

Ready to use for grading
and level reporting

Unlike reading or grammar, speaking has never had a shared yardstick. Grades based on teacher impressions drift, and interview-style tests are a heavy burden at the end of every term.

Because the test returns CEFR-aligned levels and Can-do statements, it can be used as it is for grading a conversation course or reporting levels at completion — in line with Japan’s Reference Framework for Japanese Language Education.

As the term-end test for a conversation course

Everyone takes it at the end of term; task achievement and the five indicators back up the grade

For CEFR-aligned level reporting

CEFR levels and Can-do statements leave a record based on language activities

For calibrating between teachers

A shared yardstick makes discussions about scoring differences concrete

Measure, develop,
and confirm growth

Don’t stop at a single test: measure, practice, then measure again. Re-assessment uses the same criteria, so growth is recorded exactly as it happened.

1

Measure speaking ability

Japanese Speaking Assessment (1st time)

2

Practice

Repeated practice with AI Role-Play — a place to practice for candidates below the target level

3

Confirm the change

Japanese Speaking Assessment (2nd time)

AI Japanese Learning (AI Role-Play and Mini Tests) in step 2 also works as an achievement test for the situations and sentence patterns covered in class.

Frequently asked questions

Q.When are results available?
Scoring is done by AI, so results are available as fast as the day of the test, in the admin panel and in the candidate report.
Q.What do candidates need?
A smartphone or PC and a quiet place. No app is required — candidates take the test from the URL you distribute. It takes about 20 minutes.
Q.How is it different from JLPT?
JLPT measures language knowledge, reading and listening; speaking is outside its scope. This test focuses on speaking and rates it on the CEFR scale. Combined with JLPT results, you can see both knowledge and practical use.
Q.What is the difference between a proficiency test and an achievement test?
A proficiency test measures overall ability regardless of curriculum; an achievement test checks what was learned in a given course or textbook. This is a proficiency test. For per-lesson achievement, combine it with AI Role-Play and Mini Tests (AI Japanese Learning).
Q.Can it be used to grade a conversation course?
Yes. It returns task achievement (Can-do) and five indicators that can be presented as the basis for a grade. Regular testing each term also shows progress.
Q.Is it aligned with the CEFR and Japan’s reference framework?
Yes. Judgments follow the CEFR and Japan’s Reference Framework for Japanese Language Education. The level and Can-do statements for spoken interaction and production are kept as a record.
Q.Can it handle groups with mixed levels?
Yes. There are two difficulties, Beginner and Comprehensive, and Comprehensive rates every level from A1 to C1+. Many schools run it as a placement test for everyone first and use the results for class placement.
Q.Is the AI scoring reliable?
Five independent AI raters judge each item separately and the result is decided by majority. The measured agreement between any two raters is 96.5%, and a random sample is audited for quality every month.
Q.Can you confirm that the candidate took the test themselves?
Test videos are recorded and can be reviewed in the admin panel. The system also detects switching to another screen, multiple people in frame, no visible face, or a stopped video, and shows these in the admin panel. Automated detection is reference information for review; a detection alone does not establish misconduct.
Q.How is the T-score calculated?
It shows a candidate’s relative position in the score distribution of all test takers, with the mean set at 50. Use it to see where someone stands within the same CEFR level, or to compare across classes and schools. It appears alongside the group rank in both the report and the admin panel.
Q.Do candidates get the report in their own language?
Yes. Reports including the Can-do list and the feedback on each indicator can be delivered in 20 languages, including Vietnamese, Indonesian, Burmese, Nepali, Chinese and English. Administrators review results in Japanese, and the Japanese version is authoritative.
Q.How is it priced?
Pricing is based on the number of test takers. See the pricing page for details.

Let us help you find the best solution

Whether you need a consultation, a quote, or a demo — feel free to reach out.

Contact Us