Dataset of Inflecional Construct Form noun classes in Kibiri [prm]

I am currently finishing my PhD at Université Sorbonne Nouvelle, and LACITO (Laboratoire de Langues et Civilisations à Tradition Orale, under the supervision of Prof. Sebastian Fedden, for the grammar of Kibiri. Kibiri is a severely endangered Papuan language spoken in Gulf Province, Papua New Guinea. This dataset contains 265 Kibiri nouns annotated for their absolute (ABST) and construct (CSTR) forms, together with: the Kibiri orthographic transcription, their assignment to inflectional construct classes, syllabic structure, final vowel (clusters), and semantic classes. The dataset was compiled during fieldwork and corpus annotation and forms the empirical basis for the analysis presented in the article "Absolute and Construct Forms in Kibiri: Evaluating Creissels’ Typology in a Papuan Context" (under review). Each entry includes the absolute form, construct form, class membership, phonological properties (such as final vowel cluster and syllable structure), and semantic classification where available. The dataset was used to conduct a quantitative exploration of noun class assignment using entropy and information gain measures in order to evaluate the relative contribution of phonological and semantic predictors. The results indicate that phonological structure provides limited predictive power, while semantic categories show little conditioning, suggesting that class membership in Kibiri is largely lexically specified. The dataset is provided in CSV format to facilitate reuse and reproducibility. Researchers are free to reuse the data for non-commercial purposes provided appropriate attribution is given to the author and the dataset DOI is cited.