Tibetan Dialect Style Transformation Dataset

This dataset is primarily derived from contemporary novels and electronic book texts collected from the Tibetan Qiongmai Literature website. It is systematically constructed using a bidirectional mapping strategy between written and spoken Tibetan texts, with a focus on research on stylistic transformation across Tibetan dialects. During the data collection stage, relevant reference materials, including Eight Major Tibetan Operas, Investigation and Research on Tibetan Dialects, and A Grammatical Study of the Khampa Tibetan Dialect: The Grammar of the Derge Dialect, were consulted. Original textual data were obtained through a combination of web crawling techniques, manual screening, and automated processing. Invalid, duplicated, and evidently anomalous texts were removed to ensure data standardization and consistency.In the data processing stage, all texts were subjected to unified encoding and format normalization. Based on predefined annotation guidelines, annotations were applied for sentence identifiers, written-text content, spoken-text content, and dialect type labels. All samples underwent at least one round of manual verification, with a focus on textual completeness, annotation consistency, and format correctness, in order to reduce potential annotation errors and improve data reliability.The dataset is released as a compressed package containing a single .tsv structured text file, comprising a total of 35,050 parallel sentence pairs of Tibetan spoken and written texts. The data file is organized in a structured text format, and the definitions of fields and their corresponding relationships are described in this dataset description and in the associated data paper. A small number of samples contained missing or incomplete information in their original sources and were marked or removed during data preparation. This dataset provides basic data support for research on stylistic transformation in Tibetan dialects.