Trismegistos Text Irregularities is a tool for the collection and analysis of examples of linguistic variation. It is based on the editorial corrections and regularizations to texts available in the Papyrological Navigator.
The examples were collected by a scrape of the DDbDP full text in HTML format (.html.) from the PN website (state of 4 January 2014) on the basis of Trismegistos numbers (texid), followed by a conversion to plain text form (.txt) and an import in Filemaker for data manipulation. All elements marked with an asterisk in the main text were matched to the corresponding entries in the critical apparatus and moved to a separate related database. The editorial regularizations, corrections, additions and omissions were separated from other markup in the apparatus. This makes it possible to search for the 'irregular' (in text) form and the 'regularized' (apparatus) form of all editorial linguistic annotations to texts in the PN.
To enable analysis of different types of phonological and morphological variation, the plain irregular and corrected forms (without brackets and accents) are compared and the difference is described by the following actions: addition, omission and interchange (A instead of B). All attestations of these types of errors can be found using the search interface.
As of January 2016, in-text omissions and additions, annotated by pointed brackets < > and braces { }, are included as well. At the same time, the tool was expanded to include ancient insertions, deletions and corrections, both those annotated in-text above the line \ / and by double square brackets ⟦ ⟧, and the ones mentioned in the apparatus (corr. ex). In April 2016, the type of error was further specified by including the context of error: the letter before and after the interchange, addition or omission in question.
The current Text Irregularities tool contains all modern corrections (in-text and in apparatus), the types of errors and the context of errors. Similar information for all ancient corrections will be made available in the near future.
Since the analysis of the type of error and context of error was mainly achieved by computer-assisted comparison, mistakes or wrong interpretations may occur and human quality checks are desirable. To date (25 April 2016), up to 100.737 of the 131.348 modern corrections (77%) have received a first check. The context of error and the ancient corrections still need to be evaluated. Please contact us with questions and comments and if you would like to contribute to this process.