SloERRANT is a Slovene adaptation of ERRANT (ERRor ANnotation Toolkit, accessible at https://github.com/chrisjbryant/errant), designed for automatic extraction and classification of grammatical errors in Slovene text. It takes pairs of original and corrected sentences as input and identifies and classifies the edits between them. The tool introduces Slovene-specific edit merging and classification rules, distinguishing errors at four linguistic levels: orthography, morphology, vocabulary, and syntax. It also supports optional morphological grouping. It produces annotations in the M2 format, and includes a scoring component for comparing and evaluating grammatical error correction (GEC) systems against reference corrections. SloERRANT can process text using Slovene spaCy models or pre-annotated CoNLL-U input. For installation, usage instructions, and further documentation, see the README in the attached file.
