RuCoLA: Russian Corpus of Linguistic Acceptability

TL;DR

RuCoLA is a new high-quality Russian linguistic acceptability dataset designed to evaluate and improve language models' grammatical and semantic understanding in Russian, including in out-of-domain text generation scenarios.

Contribution

This paper introduces RuCoLA, the first comprehensive Russian acceptability corpus with in-domain and out-of-domain data, and provides baseline experiments highlighting current language models' limitations.

Findings

Language models lag behind humans in acceptability detection.

Models struggle with morphological and semantic errors.

RuCoLA enables benchmarking of Russian language model competence.

Abstract

Linguistic acceptability (LA) attracts the attention of the research community due to its many uses, such as testing the grammatical knowledge of language models and filtering implausible texts with acceptability classifiers. However, the application scope of LA in languages other than English is limited due to the lack of high-quality resources. To this end, we introduce the Russian Corpus of Linguistic Acceptability (RuCoLA), built from the ground up under the well-established binary LA approach. RuCoLA consists of $9.8$k in-domain sentences from linguistic publications and $3.6$k out-of-domain sentences produced by generative models. The out-of-domain set is created to facilitate the practical use of acceptability for improving language generation. Our paper describes the data collection protocol and presents a fine-grained analysis of acceptability classification experiments with a range of baseline approaches. In particular, we demonstrate that the most widely used language models still fall behind humans by a large margin, especially when detecting morphological and semantic errors. We release RuCoLA, the code of experiments, and a public leaderboard (rucola-benchmark.com) to assess the linguistic competence of language models for Russian.