Literature is not just read: often, it is also performed. The two main modes of performance are readings, often of poetry, and plays that were written to be performed.

The first kind of performance applies a broad range of strategies for giving a voice to the words, something that is easy to decode by listening (e.g. affectionate/subdued), but the more subtle details of pitch and speed may be lost, especially if there is a large material of recordings. However, there is software that can analyze the features of performances, and help map how readers interpret poetry.

When it comes to plays, there are numerous factors that must be taken into account, including the very uneven documentation of performances, which varies with technological opportunities and recording styles, which may be more or less dynamic.

Therefore, film and media studies overlap, and, for example, it could be useful to document the number of cuts that were made. Another approach, also touched on under Character, is to focus on the information that the texts reveal, notably, the complexity of actors, who may be studied through networks. In the study of both vocal performances and staging, it is not enough to rely on data, but they may provide an indispensable context for understanding and situating a particular performance.



As drama relies just as much on visual elements as on the distribution of words in spoken dialogues, investigating the use of stage properties (props) in a play can be an approach to analysing the performance of it. One computational approach could be to measure the frequency of props used within a single play, the oeuvre of a playwright or even an entire period of dramatic writing. This can be done by taking the total amount of lines in a play and dividing that with the number of props. You can also try to categorise the props appearing in a play. 

See the chapter “Stage Properties: Bed, Blood, and Beyond” from Hugh Craig and Brett Greately-Hirsch’s Style, Computers, and Early Modern Drama for a discussion of quantitative approaches to stage props. You can for instance get inspired by the description of how Douglas Bruster graphed the frequency of properties in Shakespeare’s plays across time, resulting in a V-shaped curve revealing a gradual decrease in prop-use after the histories of the early 1590s which were characterised by a high frequency of props, followed by a significant increase beginning about 1605. Moreover, Bruster found that Shakespeare’s tragedies often had the greatest number of props, histories the second greatest, and comedies the least and concluded that the genre of a play affects both the number and the kinds of props appearing on stage. It might not always be straightforward to find the number of props in a play, but you will often find it located in the appendices to a play or kept track of in some scholarly overviews, such as Martin Wiggin’s British Drama 1533-1642: A Catalogue. This measurement can be used to gain an insight into generic differences, peculiarities of different authors, historical developments or differences between national scenes. 

A different approach is to use visualization tools to map the locations of performances of a play or several. You can get inspired by IbsenStage, which is a database of all known performances of Ibsen's plays worldwide. It is a quantitative research tool that contains data on more than 20,000 productions of Ibsen's plays and provides an accessible online resource for researching the performances. IbsenStage is continuously updated by the Centre for Ibsen studies at the University of Oslo, and continues to grow as historical and contemporary data are entered into the database.


As digital humanities have traditionally approached literature in the written format, analysing audio recordings of text readings opens new, exciting perspectives into literary performances. We might expect that literary texts are read out in an expressive way, different from every-day speech, but without computational methods it is difficult to prove. A quantitative analysis can therefore help to find out if there are any specific characteristics of auditory fiction. MacArthur (2016) analyzed a sample of performances from 100 poets and was able to detect differences between poet readings and conversational speech. 

Following her methods, conduct your own investigation of audible literature with the open-source tools Gentle and Drift. Try for example comparing poetry slam recordings with classical poetry or audio books with radio drama. Start from an exploratory visual investigation of pitch graphs and think how it changes your perception of the auditory features. Having extracted features such as pitch, pause length and frequency, you can then statistically analyse the data. 

As to drama, performance is a core part of it, yet computationally underresearched. The project Dramavis by Frank Fischer has generated character networks of dramatic texts, all the code available on GitHub. Moreover, the interactive graphs by Xanthos et al. (2016) are available to explore online, and fork from GitHub. The network dynamics can reveal aspects of the performance, presence on stage and plot evolution even without seeing the play.


Scripts and sites


Translation »