As the need for gender-inclusive language has become a highly debated topic over the years, gendered biases in speech are unfortunately often picked up and propagated by modern language models trained on large amounts of text. While remedial efforts are underway, grammatically gendered languages such as German pose some unique challenges in generating gender-inclusive language for corrective model training or fine-tuning. We assembled GIL-GALaD, a corpus of German gender-inclusive language from different sources such as social media, news articles, public speeches and academic publications. Our corpus includes the most common types of modifications of generic masculine forms of nouns and spans 30 years (1993-2023), containing over 800,000 instances of gender-inclusive language. Tools for corpus usage and extension are to be included in the release. During corpus assembly, we were also able to gain some insights into which types of gender-inclusive language were used in practice throughout the years and across different domains.

This paper describes our approach and results for the SemEval 2024 task of identifying the token index in a mixed text where a switch from human authorship to machine-generated text occurs. We explore two BiLSTMs, one over sentence feature vectors to predict the index of the sentence containing such a change and another over character embeddings of the text. As sentence features, we compute token count, mean token length, standard deviation of token length, counts for punctuation and space characters, various readability scores, word frequency class and word part-of-speech class counts for each sentence. class counts. The evaluation is performed on mean absolute error (MAE) between predicted and actual boundary word index. While our competition results were notably below the baseline, there may still be useful aspects to our approach.