This document discusses how to model text data in a database. It explains that while XML is useful for representing logical structure, it does not capture complex relationships within text like figures of speech that crosscut the structure. Relational databases provide a better model as they can represent these network-like relationships. The document presents an example of a database that stored words as individual records linked to information about figures, grammar, and other annotations to generate a rich interactive text edition and discover patterns not visible to simple readers. Modeling text as structured data in a database allows for new insights into the cultural patterns conveyed by the text.