Words, in time

Back to Article
Back to Article

Words, in time

Hang on for a minute...we're trying to find some more stories you might like.


Email This Story






In Sept. 2014, SLU English professor Dr. Jonathan Sawday was discussing, with his graduate students, the essay “Literary History as a Challenge to Literary Theory” by Hans Robert Jauss. In this essay, Jauss puts forth the idea of the “horizon of expectations” that a reader has as he or she engages with a text. At the March 23 English department-sponsored Ong Symposium on the Digital Humanities – an event named after the world-renowned SLU English professor and linguist Fr. Walter Ong, S.J. – Sawday discussed this classroom experience and pointed to its importance in initiating the research that he and three of his students (Geoff Brewer, Lauren Kersey and Seth Strickland) presented at the symposium.

The horizon of expectations is, Sawday said, “The reader’s sense as they move through a text and encounter something new or unexpected, more often at a structural level, which forces them to adjust their preconceived notions … which, in turn, help them to make sense of what they’re reading.”

Jauss’ proposition planted an idea with Sawday and his students about whether such “structural levels” of literature could be identified – in a very precise, algorithmic way.

“While we were reading and discussing Jauss, we also talked a little about the work that is going on in the U.S. and in Europe about developments in author attribution, the ways in which texts, or portions of texts, are attributed as a work of Author Y as opposed to Author Z using computational linguistics,” Sawday said.

“Could we, we wondered, train a computer to spot anachronistic texts – texts that are in the wrong place at the wrong time? In other words, could we teach a computer to detect what we termed ‘time signals’ beyond the more familiar and very successful efforts to identify genres and authors using computer programs?”

Going off of the successes of previous researchers – who have created ways for computers to recognize certain aspects of texts – Sawday and his students, with the help of SLU computer science students, developed such a technique to try and have computers place texts into certain time periods.

Sawday and his team decided to use the early modern English literary time epochs outlined by the Victorian literary critic George Saintsbury. Saintsbury’s epochs are the 15th century, 1500-1588; Early Elizabethan, 1590-1599; Jacobean, 1600-1624; Carolinian, 1625-1639; and Augustan, or, “the rest,” as Sawday put it.

The computer program used a number of factors—including word use and type—to try and place literary works within these preconceived time periods. The process was not without challenges.

“Despite all the emphasis we’ve placed on the change of linguistic and literary systems over time, and despite how defamiliarizing it often feels to read documents from earlier periods, a large number of words in the English language do not change within a span of two hundred years,” Kersey stressed. “The highest frequency words, in particular, the words that constitute the core of our language, are the least likely to change.”

Strickland pointed to additional challenges the students faced in conducting this type of research. He referenced the three laws of robotics, defined by the author Isaac Asimov, as an example of how computational research—however exciting and groundbreaking—still has its limitations. Asimov’s laws—which articulate that a robot can never harm a human, must follow the orders of humans (unless such orders run up against the first law) and are required to ensure their existence, without violating the first two laws – can be applied to the computers used in the student’s research.

If a SLU student were to create a robot, Strickland said, the student would program these laws into the robot. But the idea of “doing no harm” varies among students; it might be defined as not allowing Strickland to binge-watch “House of Cards” on his laptop, something that could be seen as detrimental to his cognitive health. Similarly, there are limitations to the literary time-period research. Because Strickland and his colleagues are the ones who defined the time epochs for the computer, “we … cannot know what we miss,” Strickland stressed. In other words, definition matters; using different criteria would change the scope of the research project.

In the end, though, the project – which is entitled “Digital Queries into Early Modern Periodization Schemes” – has given Sawday and the students a chance to examine the connections between technology and literature.

“Our project is an exercise in artificial reading,” Sawday said. “But that does not mean that it is, in some way, anti-humanist or even antihuman. As Walter Ong said in probably his most famous work, Orality and Literacy, … ‘technologies are artificial, but artificiality is natural to humans.’”