Minimal replication dataset for "Textual Overlap Rather Than Domain Alignment: A Comparative Study of Fine-Tuning Strategies for Specialised Machine Translation with Large Language Models"

This record provides the minimal dataset and code package necessary to replicate the reported findings of the study. The deposited materials support reproduction of the test-set summaries, item-level translation-quality metrics, BLEU pass-rate likelihood-ratio G^2 tests, paired t-tests, paired Cohen's d_z calculations, and figure/table source data reported in the manuscript. The full bilingual fine-tuning corpus is not redistributed in this record because parts of the underlying Chinese-English political materials derive from third-party officially published sources that may be subject to copyright restrictions. To support transparency and reproducibility within these constraints, this record provides corpus-source metadata, dataset documentation, test-set evaluation data, item-level metric data, statistical analysis files, training logs, and code sufficient to reproduce the reported findings.