Skip to main content

Easy-to-use word translations for 3,564 language pairs

Project description

image image image image


Easy-to-use word translations for 3,564 language pairs.

This is the official code accompanying our LREC 2020 paper.


  • A large collection of freely & publicly available bilingual lexicons for 3,564 language pairs across 62 unique languages.
  • Easy-to-use Python interface for accessing top-k word translations and for building a new bilingual lexicon from a custom parallel corpus.
  • Constructed using a simple approach that yields bilingual lexicons with high coverage and competitive translation quality.


First, install the package using pip:

pip install word2word


git clone
python install

Then, in Python, download the model and retrieve top-5 word translations of any given word to the desired language:

from word2word import Word2word
en2fr = Word2word("en", "fr")
# out: ['pomme', 'pommes', 'pommier', 'tartes', 'fleurs']


Supported Languages

We provide top-k word-to-word translations across all available pairs from OpenSubtitles2018. This amounts to a total of 3,564 language pairs across 62 unique languages.

The full list is provided here.


Our approach computes top-k word translations based on the co-occurrence statistics between cross-lingual word pairs in a parallel corpus. We additionally introduce a correction term that controls for any confounding effect coming from other source words within the same sentence. The resulting method is an efficient and scalable approach that allows us to construct large bilingual dictionaries from any given parallel corpus.

For more details, see the Methodology section of our paper.

Building a Bilingual Lexicon on a Custom Parallel Corpus

The word2word package also provides interface for building a custom bilingual lexicon using a different parallel corpus. Here, we show an example of building one from the Medline English-French dataset:

from word2word import Word2word

# custom parallel data: data/pubmed.en-fr.en, data/
my_en2fr = Word2word.make("en", "fr", "data/pubmed.en-fr")
# ...building...
# out: ['mitochondriale', 'mitochondriales', 'mitochondrial', 
#       'cytopathies', 'mitochondriaux']

When built from source, the bilingual lexicon can also be constructed from the command line as follows:

python --lang1 en --lang2 fr --datapref data/pubmed.en-fr

In both cases, the custom lexicon (saved to datapref/ by default) can be re-loaded in Python:

from word2word import Word2word
my_en2fr = Word2word.load("en", "fr", "data/pubmed.en-fr")
# Loaded word2word custom bilingual lexicon from data/pubmed.en-fr/en-fr.pkl


In both the Python interface and the command line interface, make uses multiprocessing with 16 CPUs by default. The number of CPU workers can be adjusted by setting num_workers=N (Python) or --num_workers N (command line).


If you use word2word for research, please cite our paper:

 author = {Yo Joong Choe and Kyubyong Park and Dongwoo Kim},
 title = {word2word: A Collection of Bilingual Lexicons for 3,564 Language Pairs},
 booktitle = {Proceedings of the 12th International Conference on Language Resources and Evaluation (LREC 2020)},
 year = {2020}

All of our pre-computed bilingual lexicons were constructed from the publicly available OpenSubtitles2018 dataset:

    title = "{O}pen{S}ubtitles2018: Statistical Rescoring of Sentence Alignments in Large, Noisy Parallel Corpora",
    author = {Lison, Pierre  and
      Tiedemann, J{\"o}rg  and
      Kouylekov, Milen},
    booktitle = "Proceedings of the Eleventh International Conference on Language Resources and Evaluation ({LREC} 2018)",
    month = may,
    year = "2018",
    address = "Miyazaki, Japan",
    publisher = "European Language Resources Association (ELRA)",
    url = "",


Kyubyong Park, Dongwoo Kim, and YJ Choe

Project details

Download files

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

Source Distribution

word2word-1.0.0.tar.gz (21.6 kB view hashes)

Uploaded source

Built Distribution

word2word-1.0.0-py3-none-any.whl (32.0 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page