Cross-Linguistic Dataset of Different Types of Lexical Semantic Change

Although semantic change types such as broadening, narrowing, amelioration, pejoration, metaphor, and metonymy are well established, their relative prevalence remains understudied. To address this gap, we present a novel dataset based on DatSemShift, a database documenting semantic change across 2,248 languages from 20 language families. We manually annotated 234 semantic shift patterns (i.e., general patterns of change observed in multiple languages) from this database involving both completed semantic changes and polysemies, comprising 2,782 individual realizations (instances of change in different languages) from 792 languages. The dataset includes part-of-speech information, semantic change classifications, confidence scores, and annotation rationales. We also release annotation guidelines to facilitate future extensions of the dataset or creation of new datasets.