Kazakh Legal Morphological Relations Benchmark: Word2Vec and FastText Evaluation, Transfer, and Error Analysis

This reproducibility package supports the study “Distributional modelling of Kazakh morphological relations in a legal corpus: Retrieval, transfer, and error locality in Word2Vec and FastText.” It contains the corpus-derived morphological benchmark, analogy evaluation outputs, relation-vector analyses, legal-subdomain transfer results, robustness and dependence-aware statistical analyses, linguistic error diagnostics, publication-ready tables and figures, manifests, file inventories, and checksum information. The source legal documents and trained model binaries are not included. The package is intended to support transparent verification and reproduction of the reported analyses.