Hi all, I’m a machine learning engineer at Duolingo, and I wanted to share a project I’ve been working on called CEFR Checker. It aligns text from many languages to a common language learning proficiency standard (the CEFR) using an ordinal regression model trained over multilingual word embeddings and corpus word frequency estimates. We use this tool when we create Duolingo Podcasts and interactive Stories to help adapt the material to the appropriate CEFR level for beginner, intermediate or advanced learners.

We’re sharing it publicly in case others find it useful as well.

The tool is completely free to use and you can check it out here.

Here’s a blog post I wrote about the project if you’d like more info.