Proficiency Levels

Proficiency levels

What the standards check, where they check it, and how Fluenough should describe its own A1, A2 and B1 stages. Research for the owner, 2026-10-23. This is a reference, not a decision. The open questions are at the end.

Sourcing. The publishers' own sites (Council of Europe, JLPT, TOPIK, Dakshina Bharat Hindi Prachar Sabha, Government of India) could not be reached when this was written. So most facts below are marked:

• (primary) read from the standard's own text or page;
• (secondary) read from Wikipedia or a search summary, to check before quoting in the app or the store;
• (inference) our reading of the sources, not a source's claim.

Check the secondary facts against the publisher before any store copy uses them.

1. The standards

CEFR (Council of Europe)

Item	What the sources say
Owner	Council of Europe. Framework 2001; Companion Volume 2020 (secondary).
What is checked	Can-do descriptors in reception, production, interaction and mediation, described as "language as social action" (secondary). Not word counts.
Levels	A1, A2, B1, B2, C1, C2. The 2020 Companion Volume adds Pre-A1 (secondary).
Certification	No central or official CEFR exam (secondary, Wikipedia). Exams are related to the levels, not certified by them.
Alignment rule	The Manual for Relating Language Examinations to the CEFR (2009) sets out the process a provider follows to show where an exam sits. Claims of a link need that evidence, and the standard-setting stage has been criticised for weak support (secondary, search summary).
Can "A1" be claimed without a test?	Inference: no. The levels are descriptors, so a course can say it aims at A1 descriptors (a design target). It cannot say it is A1 without evidence against the descriptors.
Official mapping to other standards	Disputed. The table of CEFR against ACTFL is one-directional and other mappings differ (secondary).

Sources: CEFR, Wikipedia (secondary); Companion Volume, rm.coe.int/16809ea0d4 (primary, not opened); Manual, rm.coe.int/1680667a2d (primary, not opened).

JLPT (Japan Foundation and JEES)

Item	What the sources say
Owner	Japan Foundation and Japan Educational Exchanges and Services (secondary).
What is checked	Language Knowledge (vocabulary, including kanji; grammar), Reading, Listening. No speaking. No writing. (secondary)
Levels	N5 to N1, five levels since the 2010 revision. N3 was added then (secondary).
Certification	Official score report. A pass needs the overall minimum and a minimum in each section (secondary). Overall minimums: N1 100/180, N2 90/180, N3 95/180, N4 90/180, N5 80/180. Each section needs at least 31.67 percent.
Vocabulary list	None published for the current test. The JLPT discourages studying from lists (secondary). The 1984 to 2009 format had a question-writer specification.
Approximate counts	Only for the old four-level format, unofficial: level 4 about 100 kanji and 800 words; level 3 about 300 and 1,500; level 2 about 1,000 and 6,000; level 1 about 2,000 and 10,000 (secondary). These do not map exactly onto N5 to N1.
Official CEFR mapping	Japan Foundation comparison table: N1 ≈ C1, N2 ≈ B2, N3 ≈ B1, N4 ≈ A2, N5 ≈ A1 (secondary). From December 2025, score reports carry a CEFR indication "for reference" (secondary).

Sources: JLPT, Wikipedia (secondary); JLPT official site, jlpt.jp (primary, not reached).

HSK (Chinese Testing International, for the Ministry of Education)

Item	What the sources say
Owner	Chinese Testing International (汉考国际) runs HSK. The Ministry of Education announced the 2021 word list (secondary).
What is checked	HSK 2.0: vocabulary and characters, with listening and reading tests (writing in higher levels, unverified). HSK 3.0: vocabulary, characters, grammar points, listening, reading and writing (secondary).
Levels 2.0	Levels 1 to 6, cumulative: words 150 / 300 / 600 / 1,200 / 2,500 / 5,000; characters 174 / 348 / 618 / 1,064 / 1,685 / 2,663 (secondary).
Levels 3.0	Nine levels in three stages: Beginner 1 to 3, Intermediate 4 to 6, Advanced 7 to 9, the last three sharing one exam. The 2025 syllabus, from November 2025, gives cumulative words 300 / 500 / 1,000 / 2,000 / 3,600 / 5,400 / 11,000 (levels 1 to 6, then 7 to 9) (secondary). The 2021 proposal gave different figures (11,092 words, about 3,000 characters). Sources disagree; check chinesetest.cn.
Certification	Official HSK certificate.
Official CEFR mapping	No official equivalence. Hanban said in 2010 that the six levels matched the six CEFR levels. German and French teacher associations reject this (for example HSK 6 as B2 or C1, not C2) (secondary). Inference: the disputes mean no mapping is settled.

Sources: HSK, Wikipedia (secondary); HSK 3.0 counts, search summaries (secondary, not checked against the official syllabus).

TOPIK (National Institute for International Education, Korea)

Item	What the sources say
Owner	NIIED, a branch of the Ministry of Education, has run TOPIK since 2011 (secondary). The vocabulary grading source is not confirmed here: check the National Institute of Korean Language.
What is checked	TOPIK I: listening and reading, 100 points each. TOPIK II: listening, writing and reading, 100 points each. No speaking (secondary).
Levels	TOPIK I: levels 1 and 2. TOPIK II: levels 3 to 6 (secondary, since the July 2014 reform).
Pass marks	TOPIK I: over 80 (level 1), over 140 (level 2). TOPIK II: over 120 (level 3), over 150 (4), over 190 (5), over 230 (6), out of 300 (secondary).
Certification	Official TOPIK certificate.
Official CEFR mapping	None stated in the source read (secondary).

Sources: TOPIK, Wikipedia (secondary); TOPIK official site, topik.go.kr (primary, not reached).

ACTFL (United States)

Not researched in depth. The public ACTFL levels are Novice, Intermediate, Advanced and Superior (inference from prior knowledge, not checked in this session). ACTFL scores speaking, writing, listening and reading by rated performance, and its mapping to CEFR is disputed (secondary, CEFR article). Check actfl.org before any use.

2. Indian languages

Item	What exists	What does not
Hindi, state exams	Dakshina Bharat Hindi Prachar Sabha (Chennai) runs: Parichaya, Prathamic, Madhyama, Rashtrabhasha, Praveshika, Visharadh (two parts), Praveen (two parts) (secondary). A Government of India FAQ says Praveen is equal to a Hindi BA (secondary, search).	No stated skills per exam, no CEFR link, no per-level word count (secondary).
Bengali, Telugu, Kannada	CIIL (Central Institute of Indian Languages) has a "Tests of Language Proficiency" series for these languages, built to a Class X (secondary school) standard (bookseller listings and a search summary, not verified on CIIL's site).	No national level scale found, and no CEFR link found.
Three-language formula	The 1968 formula; NEP 2020 removed prescribed languages, so the states choose. "Two of the three languages should be native to India" (secondary).	The formula sets no proficiency levels and no tests (secondary).
CEFR mapping for any Indian standard	None official. A commercial hiring test (TestGorilla) claims Hindi C1. A blog's Kannada claim conflates two frameworks. Both are unreliable (search).

Conclusion (inference): no Indian standard for Bengali, Telugu or Kannada reaches a level scale that can be cited, and none maps to CEFR. Hindi has the clearest ladder, the DBHPS exams, but no published word or skill list in the sources read.

Sources: DBHPS, Wikipedia (secondary); Three-language formula, Wikipedia (secondary); CIIL, Wikipedia (secondary; no tests named there); Government of India FAQ on Hindi examinations (primary, not reached; search summary only).

3. Constructed languages: Toki Pona

• No proficiency standard. No test, and no CEFR-style level (secondary).
• Word lists from the language's own books: pu (2014) has about 120 main words plus a few synonyms (counts vary, 120 to 125). The 2021 dictionary gives 137 "essential" words: the original 120 plus 16 ku suli (major) words. It also lists 45 ku lili (minor) words. The sources' arithmetic does not fully agree (secondary).
• Inference: any Toki Pona level has to be counted in words, from these lists. A CEFR label has no grounding at all for this course.

Source: Toki Pona, Wikipedia (secondary).

4. How apps describe CEFR

• Duolingo, courses (secondary, Duolingo's blog as found by search). Its courses were rebuilt "so the material follows the CEFR". A1 and A2 content is covered, with about 800 words introduced at each level. B1 was "under development" when the post was written. The stated goal is that all courses "meet the CEFR standards". Duolingo does not say its courses are certified.
• Duolingo English Test (secondary). Its CEFR mapping came from a concordance study, validated externally, and the method was reviewed by the Chartered Institute of Linguists (CIOL). This is the claim Duolingo backs with a study. Its courses use weaker words.

Source: Duolingo courses blog (secondary, search); DET CEFR alignment, CIOL-hosted PDF (secondary, search).

Lesson (inference). Words like "follows" or "built around" are what a course can honestly say. "Aligned" or "certified" need a study behind them.

5. What Fluenough checks today

Area	What the app does	Source in repo
Stage labels	A1, A2, B1 on paths. Bronze, silver, gold badges on the Level tile.	lib/features/stats/level_badge.dart
Word counts per stage	Validator targets A1 700, A2 900, B1 1,200 words. It warns outside half to 1.5 times the target. A path may set its own sizes, counting roots.	tools/validate_decks.py (line 290, line 6104); #615
Grammar topics	A count of topics covered, shown as "12 of 30 grammar topics".	wiki/Deck-Format.md (The B1 plan)
Readings	Passages drilled with questions. Not graded. Readings are based on ISO 15919.	wiki/Readings-ISO-15919.md; ADR-0019
Listening and speaking	Practised in drills. Not graded.	wiki/Speaking-and-Listening.md; ADR-0014, ADR-0034

Gap table

What the standards check	CEFR	JLPT	HSK	TOPIK	Fluenough
Vocabulary list or count	Not a count (descriptors)	No official list	Yes	Yes	Yes (word counts)
Grammar	Descriptors	Tested	Tested (3.0 counts points)	Inside reading and listening	Topics covered, not tested
Characters or kanji	No	Yes (kanji)	Yes	No	Script lessons only
Listening	Yes	Yes	Yes	Yes	Drills, not graded
Reading	Yes	Yes	Yes	Yes	Drills, not graded
Writing	Yes	No	Yes (writing tests)	Yes (TOPIK II)	Not assessed
Speaking	Yes	No	Not in sources read	No	Not assessed
Interaction and mediation	Yes	No	No	No	Not assessed
Certificate	No central one	Yes	Yes	Yes	No

Plain reading: the app covers vocabulary, grammar topics and practice in reading and listening. It assesses nothing. It has no speaking, writing or interaction check, and no certificate.

6. Options for truthful wording

Ranked by honesty and usefulness. Each has a main downside.

1. (b) Course stages, with what they measure. For example "Stage 1 · about 700 words · 12 grammar topics". Every number is counted by the app. Honest: high. Useful: medium, since learners lose the shared meaning of "B1". Downside: learners must learn new names. "Stage" still suggests a ladder, so a footnote is needed.
2. (d) Can-do self-checks per stage, using CEFR descriptors. Honest: high for a standards claim, if the descriptors are quoted exactly and marked self-rated. Useful: high. Downside: a lot of writing per course, and the descriptors' reuse terms must be checked first (not checked). Self-rating is not assessment.
3. (a) Keep A1, A2, B1 with a "CEFR-inspired vocabulary milestone" footnote. Honest: medium. The codes still point to CEFR, and footnotes get missed. Useful: high now, with no rename. Downside: the store will read "B1" as "B1 certified".
4. (c) Native labels per language (JLPT N-level, HSK band, TOPIK level). Honest: low to medium. The app does not test JLPT (it has no speaking or writing, and no list), so "N5" would mean "about N5 vocabulary, unofficial". Useful: high for Japanese, Chinese and Korean learners. Downside: HSK 2.0 and 3.0 disagree, every scale needs its own counts, and there is no equivalent for Bengali, Telugu, Kannada, Marathi or Toki Pona.

Recommendation: (b). Show the stage name and the counts the app actually measures, with a footnote that the stages are not an exam. Keep the YAML milestone values (A1, A2, B1) as internal codes so decks and the validator do not change; only display text changes. Add (d) later for one course, if the owner wants it. Downside: a rename costs UI work and loses a term learners know.

Wording

• App stage label: "Stage 1 · about 700 words" (in place of "A1").
• App footnote (on Progress): "Stages count the words and grammar topics in a course. They are not an exam and do not measure your level against CEFR, JLPT, HSK or TOPIK."
• Store listing: "Courses are built in three stages of words and grammar. There is no exam and no certificate."
• If native labels are chosen later (option c): "Roughly the vocabulary of JLPT N5 (unofficial comparison)."

Words to avoid

• "certified", "certificate" (unless one is issued), "official level", "exam-ready", "passes"
• "CEFR level", "A1 course", "B1 course" in store copy (use "stage")
• "aligned to CEFR" or "CEFR-aligned" (needs a study; say "follows" or "built around" if true)
• "fluent", "fluency" as a promise; "native level"; "JLPT N5 level", "HSK 3 level"
• "follows ISO 15919" for readings (say "based on ISO 15919", per the owner's rule)

7. Open questions for the owner

1. Rename the stages?
   • (A) Stage 1, 2, 3 with counts (recommended).
   • (B) Keep A1, A2, B1 with a "CEFR-inspired" footnote. Quick, but the store copy will still mislead.
   • (C) Different names per language. More work, and the names will not match across courses.
2. Show native labels (JLPT, HSK, TOPIK) for Japanese, Chinese, Korean?
   • No. Simplest and most honest.
   • Yes, marked unofficial, with the words-only comparison.
   • If yes: HSK 2.0 or 3.0? Their word lists differ.
3. Are the word targets (700, 900, 1,200) meant to match a standard?
   • No: keep them as course design numbers, and say so in the footnote.
   • Yes: they would have to match a published word list, and only some languages have one.
4. Add can-do self-checks per stage?
   • None for now.
   • English only, one course first (the B1 course).
   • Per language, after checking the CoE descriptor reuse terms (not yet checked).
5. Hindi and the DBHPS ladder. Name Hindustani stages after the DBHPS exams (Prathamic, Madhyama), or not? Their skills and word lists are not published in the sources read.
6. Toki Pona. Count by the pu and ku lists, or use stage names only?
7. Store wording. Approve the copy in section 6, or change it?

8. Before anything is published

• Open the primary pages: the Council of Europe's Companion Volume and Manual, jlpt.jp, topik.go.kr, chinesetest.cn (HSK 2025 syllabus), the DBHPS site, and the Government of India FAQ.
• Replace each secondary fact in this page with the primary wording, or drop it.
• Check the CoE descriptor reuse terms before any can-do work.