ManyDictionaries

The goal of the ManyDictionaries project is to create machine-readable versions of every dictionary ever published. Dictionaries preserve important cultural heritage, and play a critical role in language revitalization efforts. Converting dictionaries to machine-readable formats unlocks their full potential, and allows them to be used for developing language technology (e.g. updated dictionaries that include images and audio) and large-scale scientific studies of languages around the world.

Recent technological developments mean that it is now feasible to create machine-readable versions of large collections of dictionaries. Leaving aside specialist and technical dictionaries, we estimate that the languages of the world are documented in between 50,000 and 100,000 published dictionaries. We aim to digitize as many of these dictionaries as possible and to make the results as accessible as possible to scholars and community members.

We respect the intellectual property and the hard work of the scholars who created these dictionaries, and our project is fully compliant with copyright law. For dictionaries that are out of copyright, machine-readable versions will be released on a publicly accessible platform. Copyrighted dictionaries will be processed in the USA and machine-readable versions will be stored on a secure server that is made accessible to scholars as permitted by copyright law.

There are other important efforts to develop collections of dictionaries, including

• ELEXIS
• PanLex
• Dictionaria
• PARADISEC
• https://reflex.cnrs.fr/

(add more here)

Our consortium includes researchers associated with some of these projects and our hope is to collaborate and coordinate with all of them. A coordinated approach means, for example, that researchers in different parts of the world don't need to spend time and resources in scanning and processing the same dictionaries.

Although we focus on dictionaries, grammars and corpora are also important, and machine-readable versions of these resources are tremendously valuable. Ultimately we hope to see ManyGrammars and ManyCorpora projects that match our aim of being as comprehensive as possible.