Skip to main content

Tools for working with word frequencies from various corpora.

Project description

Author: Rob Speer

## Installation

wordfreq requires Python 3 and depends on a few other Python modules (msgpack-python, langcodes, and ftfy). You can install it and its dependencies in the usual way, either by getting it from pip:

pip3 install wordfreq

or by getting the repository and running its

python3 install

Japanese and Chinese have additional external dependencies so that they can be tokenized correctly.

To be able to look up word frequencies in Japanese, you need to additionally install mecab-python3, which itself depends on libmecab-dev and its dictionary. These commands will install them on Ubuntu:

sudo apt-get install mecab-ipadic-utf8 libmecab-dev pip3 install mecab-python3

To be able to look up word frequencies in Chinese, you need Jieba, a pure-Python Chinese tokenizer:

pip3 install jieba

These dependencies can also be requested as options when installing wordfreq. For example:

pip3 install wordfreq[mecab,jieba]

## Usage

wordfreq provides access to estimates of the frequency with which a word is used, in 18 languages (see Supported languages below). It loads efficiently-packed data structures that contain all words that appear at least once per million words.

The most useful function is:

word_frequency(word, lang, wordlist=’combined’, minimum=0.0)

This function looks up a word’s frequency in the given language, returning its frequency as a decimal between 0 and 1. In these examples, we’ll multiply the frequencies by a million (1e6) to get more readable numbers:

>>> from wordfreq import word_frequency
>>> word_frequency('cafe', 'en') * 1e6
>>> word_frequency('café', 'en') * 1e6
>>> word_frequency('cafe', 'fr') * 1e6
>>> word_frequency('café', 'fr') * 1e6

The parameters are:

  • word: a Unicode string containing the word to look up. Ideally the word is a single token according to our tokenizer, but if not, there is still hope – see Tokenization below.
  • lang: the BCP 47 or ISO 639 code of the language to use, such as ‘en’.
  • wordlist: which set of word frequencies to use. Current options are ‘combined’, which combines up to five different sources, and ‘twitter’, which returns frequencies observed on Twitter alone.
  • minimum: If the word is not in the list or has a frequency lower than minimum, return minimum instead. In some applications, you’ll want to set minimum=1e-6 to avoid a discontinuity where the list ends, because a frequency of 1e-6 (1 per million) is the threshold for being included in the list at all.

Other functions:

tokenize(text, lang) splits text in the given language into words, in the same way that the words in wordfreq’s data were counted in the first place. See Tokenization. Tokenizing Japanese requires the optional dependency mecab-python3 to be installed.

top_n_list(lang, n, wordlist=’combined’) returns the most common n words in the list, in descending frequency order.

>>> from wordfreq import top_n_list
>>> top_n_list('en', 10)
['the', 'of', 'to', 'in', 'and', 'a', 'i', 'you', 'is', 'it']
>>> top_n_list('es', 10)
['de', 'la', 'que', 'el', 'en', 'y', 'a', 'no', 'los', 'es']

iter_wordlist(lang, wordlist=’combined’) iterates through all the words in a wordlist, in descending frequency order.

get_frequency_dict(lang, wordlist=’combined’) returns all the frequencies in a wordlist as a dictionary, for cases where you’ll want to look up a lot of words and don’t need the wrapper that word_frequency provides.

supported_languages(wordlist=’combined’) returns a dictionary whose keys are language codes, and whose values are the data file that will be loaded to provide the requested wordlist in each language.

random_words(lang=’en’, wordlist=’combined’, nwords=5, bits_per_word=12) returns a selection of random words, separated by spaces. bits_per_word=n will select each random word from 2^n words.

If you happen to want an easy way to get [a memorable, xkcd-style password][xkcd936] with 60 bits of entropy, this function will almost do the job. In this case, you should actually run the similar function random_ascii_words, limiting the selection to words that can be typed in ASCII.


## Sources and supported languages

We compiled word frequencies from seven different sources, providing us examples of word usage on different topics at different levels of formality. The sources (and the abbreviations we’ll use for them) are:

  • LeedsIC: The Leeds Internet Corpus
  • SUBTLEX: The SUBTLEX word frequency lists
  • OpenSub: Data derived from OpenSubtitles but not from SUBTLEX
  • Twitter: Messages sampled from Twitter’s public stream
  • Wpedia: The full text of Wikipedia in 2015
  • Other: We get additional English frequencies from Google Books Syntactic Ngrams 2013, and Chinese frequencies from the frequency dictionary that comes with the Jieba tokenizer.

The following 17 languages are well-supported, with reasonable tokenization and at least 3 different sources of word frequencies:

Language Code SUBTLEX OpenSub LeedsIC Twitter Wpedia Other ──────────────────┼───────────────────────────────────────────────────── Arabic ar │ - Yes Yes Yes Yes - German de │ Yes - Yes Yes[1] Yes - Greek el │ - Yes Yes Yes Yes - English en │ Yes Yes Yes Yes Yes Google Books Spanish es │ - Yes Yes Yes Yes - French fr │ - Yes Yes Yes Yes - Indonesian id │ - Yes - Yes Yes - Italian it │ - Yes Yes Yes Yes - Japanese ja │ - - Yes Yes Yes - Malay ms │ - Yes - Yes Yes - Dutch nl │ Yes Yes - Yes Yes - Polish pl │ - Yes - Yes Yes - Portuguese pt │ - Yes Yes Yes Yes - Russian ru │ - Yes Yes Yes Yes - Swedish sv │ - Yes - Yes Yes - Turkish tr │ - Yes - Yes Yes - Chinese zh │ Yes - Yes - - Jieba

Additionally, Korean is marginally supported. You can look up frequencies in it, but we have too few data sources for it so far:

Language Code SUBTLEX OpenSub LeedsIC Twitter Wpedia ──────────────────┼─────────────────────────────────────── Korean ko │ - - - Yes Yes

[1] We’ve counted the frequencies from tweets in German, such as they are, but you should be aware that German is not a frequently-used language on Twitter. Germans just don’t tweet that much.

## Tokenization

wordfreq uses the Python package regex, which is a more advanced implementation of regular expressions than the standard library, to separate text into tokens that can be counted consistently. regex produces tokens that follow the recommendations in [Unicode Annex #29, Text Segmentation][uax29].

There are language-specific exceptions:

  • In Arabic, it additionally normalizes ligatures and removes combining marks.
  • In Japanese, instead of using the regex library, it uses the external library mecab-python3. This is an optional dependency of wordfreq, and compiling it requires the libmecab-dev system package to be installed.
  • In Chinese, it uses the external Python library jieba, another optional dependency.


When wordfreq’s frequency lists are built in the first place, the words are tokenized according to this function.

Because tokenization in the real world is far from consistent, wordfreq will also try to deal gracefully when you query it with texts that actually break into multiple tokens:

>>> word_frequency('New York', 'en')
>>> word_frequency('北京地铁', 'zh')  # "Beijing Subway"

The word frequencies are combined with the half-harmonic-mean function in order to provide an estimate of what their combined frequency would be. In Chinese, where the word breaks must be inferred from the frequency of the resulting words, there is also a penalty to the word frequency for each word break that must be inferred.

This method of combining word frequencies implicitly assumes that you’re asking about words that frequently appear together. It’s not multiplying the frequencies, because that would assume they are statistically unrelated. So if you give it an uncommon combination of tokens, it will hugely over-estimate their frequency:

>>> word_frequency('owl-flavored', 'en')

## License

wordfreq is freely redistributable under the MIT license (see MIT-LICENSE.txt), and it includes data files that may be redistributed under a Creative Commons Attribution-ShareAlike 4.0 license (

wordfreq contains data extracted from Google Books Ngrams ( and Google Books Syntactic Ngrams ( The terms of use of this data are:

Ngram Viewer graphs and data may be freely used for any purpose, although acknowledgement of Google Books Ngram Viewer as the source, and inclusion of a link to, would be appreciated.

It also contains data derived from the following Creative Commons-licensed sources:

It contains data from various SUBTLEX word lists: SUBTLEX-US, SUBTLEX-UK, SUBTLEX-CH, SUBTLEX-DE, and SUBTLEX-NL, created by Marc Brysbaert et al. (see citations below) and available at

I (Rob Speer) have obtained permission by e-mail from Marc Brysbaert to distribute these wordlists in wordfreq, to be used for any purpose, not just for academic use, under these conditions:

  • Wordfreq and code derived from it must credit the SUBTLEX authors.
  • It must remain clear that SUBTLEX is freely available data.

These terms are similar to the Creative Commons Attribution-ShareAlike license.

Some additional data was collected by a custom application that watches the streaming Twitter API, in accordance with Twitter’s Developer Agreement & Policy. This software gives statistics about words that are commonly used on Twitter; it does not display or republish any Twitter content.

## Citations to work that wordfreq is built on

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for wordfreq, version 1.2
Filename, size File type Python version Upload date Hashes
Filename, size wordfreq-1.2.tar.gz (10.1 MB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page