How we estimate English vocabulary difficulty

We analyze the English subtitles available in our catalog. Movie and TV dictionaries are generated from these files. For TV shows, the number of covered episodes is displayed; results are not a claim about episodes for which we have no subtitles.

Subtitle data and coverage

Unique vocabulary and repeated words

The vocabulary bar counts unique lemmas and phrasal verbs once each. The occurrence-weighted measure instead counts eligible word tokens: a repeated word contributes each time it appears. Phrasal verbs are excluded from this second measure so that component words are not counted twice. Neither measure estimates the percentage of dialogue you personally understand.

How CEFR estimates work

A1 through C1 labels come from the lexical classifications stored in our dictionaries. They describe vocabulary, not an official CEFR certification of a film or a learner. The A2, B1, and B2 study suggestions use the share above that level plus rare vocabulary: over 20% suggests substantial preparation, over 10% suggests reviewing recurring words, and lower values suggest trying a short scene. These are transparent study heuristics, not calibrated listening tests.

Names, noise, and unclassified vocabulary

Proper names and entries marked as noise are excluded from occurrence-weighted calculations. Words without a reliable classification remain unclassified and are displayed separately. Rare means the dictionary classifies an item as uncommon; it does not automatically mean the word is above C1. Subtitle mistakes, informal language, and spelling variants can affect classifications.

Examples, definitions, and comparisons

Preview terms are chosen from recurring B1, B2, C1, and rare entries with at least three occurrences. Examples are short excerpts from available subtitles. English meanings use the AvocadoCinema study glossary or existing dictionary definitions with source links. A general dictionary meaning may not match the exact scene. Comparisons favor shared genres and similar advanced-vocabulary shares, not identical storylines or speaking styles.

What these estimates cannot tell you

Accents, speed, overlapping speech, background sound, songs, context, and prior familiarity with the story affect listening. Songs or repeated lines can raise frequency counts. Review a short scene to judge suitability. TV totals pool available episodes and can cover much more vocabulary than one film. Data dates identify a generated snapshot, not the last time we watched or manually reviewed a title.