COALA-1M Base Corpus

Overview The COALA-1M Base Corpus is a curated corpus of Latin texts specifically designed for the computational study of lexical semantics and semantic change in Latin. The corpus was developed within the COALA project and consists of a selection of Latin texts covering different periods of the history of the Latin language. The selection is intended to provide a balanced and diachronically structured textual basis for investigating the development of lexical meanings and patterns of semantic change through computational and corpus-based methods. This Zenodo release represents the initial publication of the COALA-1M Base Corpus. At this stage, the texts are provided in their original, unannotated form. Subsequent releases will incorporate additional linguistic and semantic annotation as the corpus development progresses. Contents of this release This release contains the following files: 1. COALA_1M_BC_metadata.csv A CSV file containing the metadata for all texts included in the corpus. The metadata provide detailed information about the individual texts and their authors, including bibliographical, chronological, geographical, generic, and corpus-related information. Where applicable, the metadata also provide links to external resources and identifiers, allowing the texts to be connected to established digital resources and authority files. The meaning and provenance of each metadata field are documented in detail in the Metadata Guidelines included in this release (see below). 2. COALA_1M_BC_metadata_guidelines.docx A document providing the metadata guidelines for the COALA-1M Base Corpus. The guidelines describe each metadata field included in the CSV file, specifying its meaning, the type of information it contains, and, where relevant, the conventions and external resources used to populate it. The document should therefore be consulted when interpreting or reusing the corpus metadata. 3. XML-BaseCorpus.zip The archive containing the Base Corpus texts as XML-TEI encoded files, organized by time periods. Text file names correspond to their ID in the metadata CSV sheet (with file extension (.xml) appended to them. Changelog 1.0.1 [updated] COALA_1M_BC_metadata.csv, fixed character encoding issues. 1.0 [added] XML-BaseCorpus.zip, the archive containing the Base Corpus texts as XML files [updated] COALA_1M_BC_metadata.csv, updated with new estimated word counts based on released XML and adjusted titles reflecting final cuts in texts. Corpus design The COALA Base Corpus was designed with a particular focus on diachronic lexical semantics. The selection of texts aims to provide evidence for the study of lexical items across different periods of Latin and literary genres (Farina & McGillivray, 2026) and to support the investigation of semantic stability, semantic variation, and semantic change. The corpus therefore combines: · diachronic coverage, through the inclusion of texts from different periods of Latin; · textual and authorial diversity, covering different authors, works, and genres; · rich textual metadata, documenting the chronological, geographical, bibliographical, and generic characteristics of the texts; · machine-readable textual data, provided in XML format; · a structure designed to support the subsequent addition of linguistic and semantic annotation. The corpus contains approximately 1 million words and is intended to serve as a base corpus for computational investigations of Latin lexical semantics. Data and annotation status The texts contained in this release are not yet semantically annotated. This release should therefore be understood as a publication of the corpus base layer, rather than as a fully annotated dataset. The current release establishes the composition and metadata of the COALA Base Corpus and provides a stable version of the underlying textual material. Future releases may include additional layers of linguistic and semantic annotation and may also introduce revisions to the corpus composition or metadata. Users are encouraged to cite the specific version of the corpus used in their research. Versioning This is the (complete) initial release of the COALA Base Corpus. Because the corpus is under active development, future versions may contain: · additional or revised texts; · corrections to the textual data; · updated or expanded metadata; · additional annotation layers; · methodological refinements to the corpus selection. When using the corpus in research, please refer to the specific version and DOI of the Zenodo release. Citation If you use the COALA Base Corpus in your research, please cite this Zenodo release: Farina, A., Liétard, B. & McGillivray, B. (2026) 10.5281/zenodo.22693836 Author Contribution AF: Conceptualisation; Data Curation; Formal Analysis; Investigation; Methodology. BL: Data Curation BMcG: Conceptualisation; Funding Acquisition; Methodology; Supervision. Acknowledgements We would like to thank the following scholars for their valuable advice and suggestions during the creation and development of the corpus: Filippo Briguglio, Krzysztof Nowak, Johann Ramminger. Their expertise and feedback contributed to the design of the corpus selection and to the methodological decisions underlying this resource. We are also grateful to Camillo Carlo Pellizzari di San Girolamo for his advice on Wikidata. Funding statement This work is supported by the UKRI under the Horizon Europe Guarantee (grant number UKRI947) for the project COALA (Computational Corpus Annotation for Quantitative Analysis of Latin Lexical Semantics) successfully evaluated by the ERC. Licence CC-BY SA 4.0. Cuzzolin, P., & Haverling, G. V. M. (2009). Syntax, sociolinguistics, and literary genres. In P. Baldi & P. Cuzzolin (Eds), New perspectives on historical Latin syntax (pp. 19–63). Mouton de Gruyter. Farina, A., & McGillivray, B. (2026). COALA-1M: Introducing a New Corpus for Manual Semantic Annotation in Latin. CLARIN Annual Conference 2026. 29 September – 1 October 2026, Brighton (United Kingdom).