Download CEFR readability datasets

As part of our pilot project for the European Language Grid, EDIA has developed several datasets that can be used for training AI models on CEFR readability classification. These datasets consist of texts from various sources, labelled on CEFR readability level.

Please fill in the form below to get access to the datasets. The datasets are available for non-commercial, academic purposes (CC-BY-NC) only.

Name

First Name

Last Name

Academic organisation

Email

LinkedIn Profile

Optional - To help us verify your academic affiliation.

http://

Which datasets are you interested in?

Dutch dataset (1.200 texts labelled on CEFR readability level)

German dataset (500 texts labelled on CEFR readability level)

English (1.200 texts labelled on CEFR readability level)

What would you like to use the dataset(s) for?

Please describe your research project.

Citing

When citing these resources in your research, please use:

Breuker, M. (2023). CEFR Labelling and Assessment Services. In: Rehm, G. (eds) European Language Grid. Cognitive Technologies. Springer, Cham. https://doi.org/10.1007/978-3-031-17258-8_16

Readability API

Based on the datasets, we have created several CEFR text classification models which can be used through our Readability API. For more information see our developer documentation.