Workflows for multilingual corpus building

As we announced on social media and on DARIAH’s main blog, our workshop in Hamburg at the end of August was particularly fruitful. During this three-day event on Creating, Managing and Archiving Textual Corpora in Under-resourced Languages, we created three main workflows:

Building textual corpora for under-resourced Languages

Managing textual corpora in under-resourced languages

Archiving textual corpora

And a specific one mainly for corpus building with audio files:

Preparing minority or endangered language corpora for annotation in SIL FLEx

All of these workflows have been published on the Social Sciences & Humanities (SSH) Open Marketplace. They describe the main steps to follow in order to build a corpus, taking into consideration the issues that arise when dealing with under-represented languages. The last workflow is an important piece that facilitates the incorporation of audio files into one’s corpus. Audio files can represent an important part of the data available for a given topic in an under-represented language, in particular when it comes to endangered ones. Since the lack of data can prevent scholar to use many digital tools that require a certain quantity of data, the inclusion of all files, including audio, can make a difference and save a whole project.

With the success of our workshop and its outcomes, we hope to continue the work initiated here in promoting FAIR and CARE in the production and preservation of multilingual digital data. There are indeed many steps to digitise raw data and analyse them automatically or semi-automatically, and only with the kind of initiative that we have launched with institutional support that is not profit-driven can we hope to diversify the digital data available and counter the monolingual trend that has dominated so far.

---

---