Skip to main content

Builds a multisource English lexicon

Project description

🗽 CityLex: a free multisource English lexical database

PyPI version Supported Python versions CircleCI

CityLex is an English lexical database intended to replace or enhance databases like CELEX. It combines data from up to seven unique sources, including frequency norms, morphological analyses, and pronunciations. Since these have varying license conditions (some are proprietary, others restrict redistribution), we do not provide the database as is. Rather the user must generate a personal copy by executing a Python script, enabling whatever sources they wish to use.

Building your own CityLex

To install CityLex execute

pip install citylex

To see the available data sources and options, execute citylex --help.

To generate the lexicon, execute citylex with at least one source enabled using command-line flags. As most of the data is downloaded from outline sources, an internet connection is normally required. The process takes roughly four minutes with all sources enabled; much of the time is spent downloading large files.

To generate a lexicon with all the sources that don't require manual downloads, execute

citylex --cmudict \
        --elp \
        --subtlex_uk \
        --subtlex_us \
        --udlexicons \
        --unimorph \
        --wikipron_uk \

File formats

Two files are produced. The first, by default citylex.tsv, is a standard wide-format "tab separated values" (TSV) file, of the sort that can be read into Excel or R. Some fields (particularly pronunciations and morphological analyses) can have multiple entries per wordform. In this case, they are separated using the ^ character.

Advanced users may wish to make use of the second file, by default citylex.textproto, a text-format protocol buffer which provides a better representation of the repeated fields. To parse this file in Python, use the following snippet:

import citylex

lexicon = citylex.read_textproto("citylex.textproto")

This will parse the text-format data and populate lexicon. One can then iterate over lexicon.entry like a Python dictionary.

Non-redistributable data sources

Not all CityLex data can be obtained automatically from online sources. If you wish to enable CELEX features, follow the instructions below.

This proprietary resource must be obtained from the Linguistic Data Consortium as LDC96L14.tgz. The file should be decompressed using

tar -xzf LDC96L14.tgz

This will produce a directory named celex2. To enable CELEX2 features, use --celex and pass the local path of this directory as an argument to --celex_path.

For more information

For contributors

To regenerate you will need to install the Protocol Buffers C++ runtime for your platform, making sure the version number (e.g., the one returned by protoc --version matches that of protobuf in requirements.txt. Then, run protoc --python_out=. citylex.proto.


The CityLex codebase are distributed under the Apache 2.0 license. Please see License.txt for details.

All other data sources bear their original licenses chosen by their creators; see citylex --help for more information.


CityLex was created by Kyle Gorman.

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 citylex, version 0.1.8
Filename, size File type Python version Upload date Hashes
Filename, size citylex-0.1.8.tar.gz (10.6 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page