An approach to complex texts in multiple documents

Abstract

Abstract This article describes an approach to the treatment of texts in complex large textual traditions. Editors are interested in the text as it appears line-by-line in each document, and in how the versions of the text differ from document to document. It is useful to define a text as the record of an act of communication, inscribed in a document: thus, the instance of the act of communication we identify as Geoffrey Chaucer’s Canterbury Tales, as it appears in the Hengwrt manuscript. In this view, every text has a dual aspect: it is both the words as they are inscribed in a particular document, and as they constitute an act of communication and its parts. This presents challenges for scholars who wish to record both aspects. In encoding implementations, these two aspects are commonly treated as ‘overlapping hierarchies’. However, the ‘overlapping hierarchy’ model does not deal with cases where text segments are not contiguous in either aspect and cannot overlap cleanly. To meet these cases, the Textual Communities project developed an architecture in which the two aspects are represented as distinct and independent hierarchies (trees), with text segments referenced to nodes on each tree. The linking of text segments to the two trees is managed by a JSON database, accessed through transcription and collation tools presented in a Web interface. Textual Communities does not implement the whole of this architecture in terms of validation, ingestion, and processing. Full exploration and implementation of the architecture here described are challenges for future scholars.