Choose a language
English
Opens with you, I and be, because it's built from what people say in films and series, not what they write.
Spanish
Ser and estar both make the top ten, so the difference between them is the first thing it teaches.
French
Être and avoir come before every article: the two verbs to learn before anything else.
German
Sein and haben sit in the top four, and du outranks nicht, a sign of how much of it is film dialogue.
Italian
Essere and avere sit in the top three, with non close behind: the core of nearly every sentence you'll hear.
Portuguese
Ser and estar both make the top eight, along with ir, the verb for going and for the near future.
Russian
Я and не open the list, and every word carries a romanization, so you can read it before you know Cyrillic.
Japanese
Particles fill the top: の, を, に and は come first, each with its reading, drawn from news and web text rather than subtitles.
Mandarin Chinese
Every word comes with pinyin, and the list opens with 我们 and 什么, two words you'll need in your first conversation.
What every list contains
- Rank and frequency. The word's position and its raw count in the source corpus. Each word is listed once, under its dictionary form, with the counts of all its inflected forms added up and the most frequent forms shown beside it.
- IPA. A phonemic or phonetic transcription, so you can pronounce a word you have only read.
- Type. Noun, verb, adjective, particle and so on, with a filter on every list to show only the types you want. For a pronoun or article form, the dictionary form it belongs to.
- Meaning. A short English gloss of the word's primary sense.
- Example sentence. A real sentence using the word, with an English translation.
- Readings. Pinyin for Chinese, kana and rōmaji for Japanese, a romanization for Russian.
- Tags. Gender for nouns, aspect for Russian verbs, and register (colloquial, slang, derogatory, vulgar), each one a filter on the page and a column in the downloads. Vulgar words stay hidden until you ask for them.
How the lists are built
The ranking comes from FrequencyWords (OpenSubtitles 2018), a word-count of film and television subtitles, the closest large corpus there is to everyday spoken language. Japanese uses the Leipzig Corpora Collection instead, because subtitle tokenizers split Japanese into fragments rather than words.
Each candidate word is then checked against Wiktionary: tokens that are not words of the language (names, typos, foreign words, tokenizer fragments) are dropped, and the survivors take their pronunciation, part of speech, meaning and example from the dictionary entry. Where Wiktionary has no usage example, one is drawn from Tatoeba, the open collection of translated sentences.
Inflected forms ("est", "está", "was") are counted under their dictionary form ("être", "estar", "be"), so each word takes one place in the ranking, with the forms met most often listed beside it. Pronoun and article forms (me, him, la) keep entries of their own, as do comparatives (better, mieux) and female equivalents, since a learner learns each of them as a word.
Using the lists in your classroom or blog
Every list is free to copy, print, embed and adapt. The frequency data and dictionary content are open-licensed (the exact licence is stated on each list's page); the lists themselves are released under the same terms. If you reuse one, link back to its page so your readers can get the latest version and the downloads.