1:00-1:30PM: John Franklin - Hunting for Helen: A Lost Storytelling Engine in Greek Epic Poetry.

This paper will present and analyze a collection of storytelling motifs and variants drawn from ancient Greek epic poetry. Sources include key passages of Homer’s Odyssey, and various later authors who preserve plot summaries and story fragments that derive from several lost epic poems. The material derives from, I argue, an early narrative subsystem of the genre that has survived in various vestigial forms. I shall attempt to reduce the variants to a single narrative ‘engine’ that allowed oral formulaic singers to generative wandering adventures, akin to those of Odysseus, but set in the Eastern Mediterranean; it centered especially on the possession of Helen of Sparta/Troy, either an effort by her husband Menelaos to recapture her immediately after her abduction by the Trojan prince Paris; or wandering travels in the same area with Menelaos after being recaptured from Troy.

Abstract:
One of the fundamental goals of artificial intelligence is to build computers that understand language at a human level. Recent progress towards this goal has been fueled by deep learning, which represents words, sentences, and even documents with learned vectors of real-valued numbers. However, creative language—the sort found in novels, film, and comics—poses an immense challenge for such models because it contains a wide range of linguistic phenomena, from phrasal and sentential syntactic complexity to high-level discourse structures such as narrative and character arcs. In this talk, I discuss our recent work on applying deep learning to creative language understanding, as well as exploring the challenges that must be solved before further progress can be made. I begin with an general overview of deep learning before presenting model architectures for two tasks involving creative language understanding: 1) modeling dynamic relationships between fictional characters in novels, and 2) predicting dialogue and artwork from comic book panels. For both tasks, our models only achieve a surface-level understanding, limited by a lack of world knowledge, an inability to perform commonsense reasoning, and a reliance on huge amounts of data. I conclude by proposing ideas on how to push these models to produce deeper insights from creative language that might be of use to humanities researchers.

Bio:
Mohit Iyyer is an assistant professor in computer science at the University of Massachusetts, Amherst. His research focuses on designing deep neural networks for both traditional natural language processing tasks (e.g., question answering, sentiment analysis) and new problems that involve understanding creative language (e.g., modeling fictional narratives and characters). He received his PhD in computer science from the University of Maryland, College Park in 2017 and spent the past year as a researcher at the Allen Institute for Artificial Intelligence.

To make informed decisions about health, environmental, and financial issues, people often need to understand complex data. Narratives have been shown to be more persuasive in communicating such data than mere statements of facts. However, narratives without facts can be deceiving. I will describe extant research on best ways to enrich narratives with factual information using numerical and graphical formats, analogies, social context, and moral values.

In this talk I will give you a cognitive scientist's tour of cultural evolution. Stories, songs, arts, and technologies all survive by passing from person to person - and without a cognitive system to perceive, process, and produce these artifacts, culture would cease to be. One of the most intriguing aspects of culture is how it toggles between two distinct forms as it evolves: a private form in individuals' minds, and a public form that is out there for everyone to see. The interplay between these two forms of culture gives rise to complex dynamics over time, which I will illustrate with two neat examples: a mathematical model of idea-artifact coevolution, and some real-world cultural evolution data from a large-scale art project called Picbreeder.

11:00-12:00pm: Fritz Breithaupt - Provost Professor, Germanic Studies & Cognitive Science, Indiana University, Bloomington

Narrative and Emotion: Insights from the Telephone Game

When people retell stories, what guides their retelling? Most previous research on story retelling and story comprehension has focused on information accuracy as the key measure of stability in transmission. This paper suggests that there is a second, affective, dimension that provides stability for retellings, namely affects, emotions, and moral intuition. In a series of large-sample studies with multiple iterations of retellings, i.e. the telephone game, we found evidence that people are quite accurate in preserving all degrees of emotions and affects – even when the event that produced the emotion in the original story is dropped or changed. Thus, we propose that the preservation of emotion, affect, and intuition is an implicit goal of retelling. The work is based on studies by the Experimental Humanities Lab (www.experimentalhumanities.com).

1:30-2:30PM: Jim Bagrow - Information flow on networks

Modern society depends on the flow of information over online social networks, and popular social platforms now generate significant volumes of behavioral and communication data. However, it remains unclear what fundamental limits may exist when using these data to predict the activities and interests of individuals. In this talk, I will review the research on this problem, from observational studies to experiments and big data analyses. In my research group, we are developing mathematical models and applying tools from information theory to study the flow of information on networks. In recent work, we used these tools to estimate the predictive information content of the writings of Twitter users and to what extent that information flows between users. Distinct temporal and social effects are visible in this information flow, and these estimates provide a fundamental bound on the predictive accuracy achievable with these data. Due to the social flow of information, we estimate that approximately 95% of the potential predictive accuracy attainable for an individual is available within the social ties of that individual only, without requiring the individual's data. This work has implications for online privacy: individuals outside an online social platform may still be predictable with the data possessed by the platform's owner, due to the flow of information from social ties on that platform.

10:30 - 11:00am: Tina Escaja - RoboPoems: Poetry-inflected robots

RoboPoems: Poetry Inflected Robots.

Robopoem@s, in its Quadrupeds version, are based on five insect-like robots whose legs and bodies are engraved with the seven parts of a poem written from the point of view of the robot. Their interaction with sensors and voice activation allows for an exchange with humans and with each other, emphasizing the existential issues addressed in the poem. Binary constructs such as creator/creature are questioned by these creatures purposely chosen from open-source models resembling insects and spiders, thus emphasizing anxiety and removal from humans while underlying the already problematic relation between humans and technology. Technology and humanity are reframed by these wandering robopoem@s, which ultimately recounts existential assumptions through poetry and robotics.

1:30-2:30PM: Timothy Tangherlini - Folklore, Cultural Analytics, and the Science of Storytelling

Since the inception of the field, folkloristics--or the study of folklore--has been concerned with the study of stories. In this context, folklore can be understood as cultural expressive forms that circulate on and across social networks. In this short presentation, we explore two interrelated problems: the classification of stories and the analysis of story structure. In regard to the first problem, we explore how stories can be classified to align with a research problem, rather than the age-old problem of aligning research problems with existing classifiers. Based on work on flexible classifiers, we show how new questions emerge when a collection of stories is classified in this manner. In regard to the second problem, we show how a simple actant-relationship model can help us discover the generative narrative framework for stories within a tradition group. Such an approach allows us to develop a multi-scale understanding of the interactions between an underlying tradition, the domains of knowledge for that group, and the stories that people tell within and across those domains. Focusing primarily on legends--believable stories told as true--we consider the classification of stories of witches, ghosts and house elves in nineteenth century rural Denmark before moving on to discussions of rumors, the rise of the anti-vaccination movement in the United States, and the challenge of "fake news", Pizzagate, and the totalizing conspiracy theory of Q-Anon.

2:30-3:00PM: Mary Louise Kete - The Twin Sciences of Story: What does the study of literature tell us about the affordances of story?

There are at least two different kinds of sciences of story: What can be known about stories? What can be known through story? One is the science or systematic analysis of narrative, in particular, verbal narratives. What makes them different from other verbal forms? What are the component elements and how do they relate to one another? Story, in this sense, has been the object of science since at least Aristotle. The other considers narrative as a way to know (in the sense of the Latin verb, “scire,” at the root of our noun, science) and shifts attention to the work that narratives can do for readers, writers, tellers, buyers, givers and sellers of stories within particular historical and material conditions. Each of these sciences provides distinct ways to understand the limits and possibilities of narrative as both a “fact of the environment and a fact of behavior.”