Text dating refers to the computer-assisted dating of texts. It involves examining whether linguistic features can be used to draw conclusions about a text’s date of composition or publication. This approach is based on recurring patterns in word usage, phrasing, and stylistic conventions. 

In this project, I applied text dating to a sample of French novels. The texts were first prepared for machine analysis and then modeled using bag-of-words and TF-IDF methods. Based on this, various machine learning models were employed to attempt to date the texts based on their linguistic features and to provide the most accurate estimate possible using regression.

What was particularly important here was not only whether a model could predict a plausible date. Equally central was the interpretability of the results: Which words, phrases, or stylistic patterns actually contribute to the chronological classification? This makes text dating interesting not only as a predictive method but also as an analytical tool in the digital humanities.

The project therefore combines technical modeling with literary interpretation. It demonstrates how machine-based methods can be used to identify diachronic patterns in larger text collections, while also making it clear that such results must always be interpreted within the context of genre, authorship, corpus composition, and metadata.