How it works

Διαβάστε στα Ελληνικά

BRAVE (Brief Receptive Adaptive Vocabulary Evaluation) estimates your receptive vocabulary: the number of words you recognize and understand when reading. Counting them all directly would require checking every word in a large dictionary. Instead, we use Item Response Theory (IRT), a mathematical approach to designing and analyzing tests.

The test presents words of different difficulty levels and asks whether you know them. For example, “γάτα” (cat) is easier than “αποστάτης” (renegade). The model combines your answers with the difficulty of the test words to estimate your vocabulary size.

To keep the test short and as precise as possible, we use Computerized Adaptive Testing (CAT). After each answer, the estimate is updated. The next word is selected to be close to your estimated level and provide as much additional information as possible. The test ends when the required precision is reached. This means that beginners and people with large vocabularies see different words.

Reference dictionary

We use the Dictionary of Standard Modern Greek (Triantafyllidis Dictionary) to define what counts as a word. Its 45,000 headwords form the basis of our estimate. We count headwords rather than treating every inflected form as a separate word.

We chose this dictionary because it is an authoritative and widely used reference focused on standard Modern Greek. It is not a complete inventory of the language: its coverage of specialized terms, archaic words, and other less common vocabulary is limited.

Word frequencies

Word difficulty is related, among other things, to how often a word is used. Common words are generally known by more people than rare ones. We collected frequency data for each headword from the Hellenic National Corpus (HNC). These data help us express the vocabulary estimate as a number of known words.

Response checks

The result assumes that you mark words as known only if you know their meaning. We use two checks:

  • Made-up words. The test includes some invented words. If you mark more than one as known, a warning appears with your result.
  • Meaning checks. For some words you mark as known, we ask you to choose the correct meaning from four options. More than three incorrect answers also trigger a warning.

These checks do not change the vocabulary estimate itself. If either check flags a problem, the result may be unreliable. In that case, we show a warning and do not display a confidence interval.

The team