Portuguese MeterFree
EN

Methodology

How it is calculated.

Every rule the result follows, in plain words.

Updated 21. 11. 2026

What it is

The Portuguese Meter estimates your level of Portuguese on the six CEFR levels (A1 to C2) from four skills: grammar, vocabulary, reading and listening. It is an orientation estimate for your own use, not an exam.

What it measures

Grammar

Ser and estar, past tenses, subjunctive

articles and contractions (no, da, pelo), comparisons, future and conditional, future subjunctive, gender and agreement, if-clauses, linking words, object pronouns, past subjunctive, perfect and imperfect past, personal infinitive, por and para, prepositions, present subjunctive, present tense, relative clauses, ser and estar, tenho feito and compound tenses, the passive, verb phrases (acabar de, voltar a…)

Vocabulary

Everyday words, false friends, idioms

advanced vocabulary, everyday words, false friends, idioms, word partnerships (fazer/dar/ter…), work and study

Reading

From a notice to an opinion piece

reading between the lines, tone and attitude, understanding details, understanding the main idea

Listening

Numbers, requests, tone of voice

numbers and times, reading between the lines, tone and attitude, understanding details, understanding the main idea

How the test runs

You start by saying where you think you are. This only decides the first questions; it has a small weight and the test moves away from it quickly if your answers say otherwise.

The test then asks 30 questions (fewer if a skill has run out of questions). After each answer it updates its estimate and picks the next question that tells it the most about you, keeping a balance of skills and areas: grammar 10, vocabulary 8, reading 6, listening 6. Without an Portuguese voice on the device, listening is skipped and the other skills get more questions.

Listening texts are read by your browser's voice; you can play each one at most twice.

“I don't know” counts as not known. It is more honest than guessing, and it makes the estimate more accurate.

It measures international Portuguese: the texts follow the 1990 Spelling Agreement, no answer depends on words, spellings or constructions that differ between Brazil and Portugal, and the listening uses the voices your device has, in turn (Brazil, Portugal and others; you can choose one in the settings).

How the level is calculated

Every question has a difficulty on one scale. The chance of a right answer depends on how far your level is from that difficulty, with an allowance for lucky guesses (one in the number of options). This is a standard model of item response theory (three parameters, with the slope fixed at 1.4).

After each answer your level is estimated as the average of all levels that fit your answers, weighted by how well they fit (the EAP method). The levels are cut at fixed points of the scale: A1 below −2, A2 from −2, B1 from −1, B2 from 0, C1 from 1, C2 from 2.

The range you see covers about 80 % of the probability. “Close to” means your estimate lies in the outer quarter of the level, next to the named one.

Skills and areas

The level of each skill is estimated from that skill's questions only, pulled a little towards your overall level because there are fewer of them. Treat it as a direction, not an exact level.

The strongest area and the biggest gap compare how many questions in an area you got right with how many someone at your level would get right. An area is shown only with enough questions: at least 3 for the strongest area, at least 2 for a gap.

First-language traps

Some questions are built around mistakes typical for speakers of a particular first language (for example false friends). They are shown only to people who chose that first language, and the detailed analysis shows which of them you fell into.

The questions

The current bank has 300 questions. By level and skill:

LevelGrammarVocabularyReadingListeningTotal
A1875626
A2221981059
B1202081260
B22520101065
C1241711860
C21178430
Total110905050300

Limits

Version

Content version 0.1.0: the questions, their difficulty, the scoring and the texts. When the first number changes, results from before and after are not directly comparable.

All sources