: Obligatory possessive inflection (58A) and possessive classification (59A).
The "RoBERTa" designation suggests this data has been pre-processed or formatted for use with the (Robustly Optimized BERT Pretraining Approach) large language model, likely for tasks like cross-lingual transfer or testing a model's metalinguistic knowledge. Included Linguistic Features (Chapters 37–70)
: Testing if models like RoBERTa or XLM-RoBERTa have "learned" the typological rules of specific languages during pre-training. WALS roberta sets 37-70.zip
World languages with features and coordinates - Dataset Search
: Noun phrase conjunction (63A) versus verbal conjunction (64A). Verbal Categories (Chapters 65–70) : World languages with features and coordinates - Dataset
The features in this range are essential for understanding how different languages handle noun and verb structures. :
: Leveraging the broad cross-linguistic data in WALS to improve how models handle the hundreds of languages that lack large amounts of training text. : Using the WALS database features as labels
: Using the WALS database features as labels to see if a model's internal representations (embeddings) cluster according to known linguistic traits, such as whether a language uses definite articles.