PHOIBLE is a database of phonological inventories and distinctive features, encompassing more than 3000 inventories (and growing). Here are links to various parts of the project:

• The web-browsable published form of PHOIBLE is called “PHOIBLE Online” and is located at https://phoible.org. The data published there corresponds with the most recent release of this repository. The data on https://phoible.org is also available as a CLDF StructureDataset.

• This repository is https://github.com/phoible/dev. Data in machine-readable form (csv and RData) is available here. It contains the cleaned and aggregated data. It is not guaranteed to exactly match what is published at PHOIBLE Online, due to the occasional discovery and correction of errors and the addition of new languages. For this reason, it is recommended that you make use of the most recent release in your own analyses, rather than working from the tip of the master branch.

  Because this is the development repository, it also contains scripts for aggregating the raw data into a single file, testing the consistency and integrity of the data, etc. That is where we make sure the proper IPA symbols are used, or that every unique phoneme has a unique representation in the distinctive feature system. Most users will never need to interact with these scripts, and can either browse the online version, or work with either the CLDF StructureDataset or the csv or RData files in the data directory.

• The documentation is at http://phoible.github.io. This includes notational conventions, departures from official IPA usage, citation information, etc. If you have a question about PHOIBLE, chances are you should seek answers there first.

All of these sites should be considered works-in-progress. The “published” form of the data at https://phoible.org should be considered the most stable.