Exploring TEI structures to find distinctive features of text types
Berlin-Brandenburg Academy of Sciences and Humanities, Germany
Speakers deal with text types (e.g. newspaper, letter, leaflet) successfully every day: They are able to apply the proper text type in a given context, for a certain communicative purpose, according to specific social constraints. However, extensive linguistic discussions on the factors that substantially constitute text types have not come to an end, yet.
Among the key distinctive features of text types the textual structuring has been regularly counted in though its importance compared to other factors could not be finally resolved. Today, with large TEI corpora at hand carrying information on (logical and layout) text structures, it becomes possible to automatically evaluate the relevance of textual structuring for the differentiation of text types. In addition, TEI structures can be included in the recognition of other features whose distribution depends on certain text structures. Hence, next to other criteria it seems straightforward to take a closer look at TEI structuring for the extraction of distinctive features for text types.
The current paper presents an approach to identify distinctive features of devotional text types. Three examples, where TEI structuring is considered, are discussed, namely (1) intertextuality as indicated by bibliographic references, (2) repetition of words and phrases in certain structural contexts, and (3) the level text structuredness in general. The features evaluated here were mentioned in previous (predominantly not corpus-based) studies on distinctive features of devotional literature and of text types in general.
The study is based on three 17th century corpora: manuals of devotion (4,057,497 tokens), funeral sermons (6,910,357 tokens), and a reference corpus of diverse text types (21,862,811 tokens). Texts are taken from the Deutsches Textarchiv corpus and are all tagged according to the TEI format DTABf. It will be shown that TEI tagging can help to safely extract these features and to achieve their more sophisticated interpretation.
Reconceiving TEI models of theatrical performance text with reference to promptbooks
1University of Waterloo, Canada; 2University of British Columbia; 3Medicine Hat College; 4SUNY Oswego; 5University of Victoria, Canada
This paper explores and suggests a revision to the conceptual model of “dramatic text” currently evidenced in TEI discourse. Our critique arises out of the challenges our research team has encountered developing an encoding protocol for promptbooks in the collection of the Canadian Stratford Shakespeare Festival Archives. We have argued elsewhere that promptbooks—by which we mean the promptbooks defined by Charles H. Shattuck (1965) as the “book[s] actually used by prompters or stage managers in conducting performances” (1965, 5)—are ontologically distinct from other forms of theatrical performance texts in the sense that “a promptbook records a series of utterances to which intended performance events are mapped in a relative temporal sequence” (Roberts-Smith et al. forthcoming 2019).
In this paper, we work through the implications of that claim for the TEI Guidelines, and suggest a reconceptualization of TEI’s discursive model of theatrical performance texts based on: 1. long-acknowledged limitations of the existing Guidelines to address the complex ontologies of a “dramatic work” (Mylonas and Lavaigno 1995); 2. a summary of more recent theoretical work on the ontologies of dramatic and theatrical/performance texts (including frameworks proposed by Osborne 1996; Clopper 2001; Erne 2003; Schafer 2006; Kidnie 2009; Worthen 2005, 2010, 2011; Werstine 2012; Roberts-Smith et al. 2013; Griffin 2018); and 3. a survey of existing TEI Guidelines for Performance Texts (TEI 7), TEI listServ discussions (with a focus on the 2001 debate among Bauman, Sperberg-McQueen, and others), and other TEI extensions, including the Music Encoding Initiative (MEI).
Our central argument is that the tension between event and text that Mylonas and Lavaigno see as problematic in encoding “dramatic works” (1995) applies neither to performance texts traditionally understood as such nor to promptbooks. A more functional conceptualization in TEI would accommodate dramatic, theatrical/performance, and prompt texts as three distinct ontologies.
What is a Line? Encoding and Counting Lines in Early Modern Dramatic Texts
1University of Victoria, Canada; 2University of British Columbia, Canada; 3University of Leeds, UK; 4University of Nevada at Reno, USA
What we consider to be a line—whether we locate its beginning <lb/>, describe it as a topographic line <line>, or a line of verse <l>— is both deeply significant and highly unstable. Lines numbers a key critical metric in Early Modern Drama. The 1623 first folio of Shakespeare’s plays, has even been subjected to a copyrighted canonical numbering system (Through Line Numbers). Not just a convenient unit for citation, lines are the dominant measure of play and role length. Major critical arguments depend on whether and how much a character speaks in verse or prose; at the same time, editorial relineation is often based on the very arguments that in turn depend on lineation. And yet, as in modern typesetting and digital interfaces, early modern prose lines break where the page or column width demands. Characters do not always speak in full lines. Part lines may be prose or verse. Multiple short utterances may be set on one typographical line. Complicating the mise-en-page still further is the fact that prose was often set as verse and vice versa, to fit text to page space. Stable, canonical lines are a common output of a scholarly edition, yet all editorial theory since McGann has emphasized the instability of such textual features. We discuss the tension between citable and fluid texts, outline how we use the @ed and @edRef attributes in our lineation, and introduce our prototype whereby various lineation systems are interoperable. The LEMDO prototype allows for canonical line identifiers, but does not foreclose other possibilities; in fact, LEMDO proliferates @xml:ids so that projects within the LEMDO ecosystem can make project-level decisions about what constitutes a line. Finally, we gesture towards future reconciliation of lineation systems through linked-open data.