On Wednesday, October 19, at The Catholic University of America, our new Dean of the School of Arts and Sciences, Dr. Aaron Dominguez, gave a talk titled “Science & Math in the Humanities” to the Department of Library and Information Science. Dean Dominguez was the Associate Dean for Research and Global Engagement and a Full Professor of Physics and Astronomy at the University of Nebraska — Lincoln (UNL). Dr. Dominguez, whose area of research is experimental high energy physics, has a strong history of research and grant activity.
What connections can we made in the liberal arts? Specifically, can we connect creative mathematical and physical thinking with creative thinking in the humanities? If so, what is to be gained by this? He provided two examples: text analysis and network analysis.
Dean Dominguez started from his background in particle physics and how he thinks as a physicist.
Dean Dominguez has worked at the CERN Large Hadron Collider in Switzerland for many years. As physicists, they had to distinguish and separate the various types of particles in order to extract meaning from the noise. He noted that this strategy/methodology is used in text analysis and other digital humanities areas.
How does this apply to digital humanities? Dean Dominguez used the humorous video, ‘Kurt Vonnegut on the Shapes of Stories.’ to illustrate that stories and narratives have structure that can be quantified and graphed:
Professor Matthew Jockers has tried to do just that—see if there are hidden structures in large body of texts. Dean Dominguez has been friends with Jockers since they were colleagues at the University of Nebraska-Lincoln. Dean Dominguez mentioned that most of what followed in his lecture is from Jocker’s work which can be found here:
Jocker’s blog: http://www.matthewjockers.net
Access to R package (“Syuzhet”) for text analysis: https://github.com/mjockers/syuzhet
I highly recommend Jocker’s book for folks interested in learning text analysis (and R!). I use the book in my digital humanities course that I teach in the department of library and information science at CUA.
Opinion and sentiment analysis is a hard problem. You will need to go through a body of text and ‘tag’ each word as positive, negative, or neutral. Lexicons of positive and negative words have been developed over the years and a researcher interested in doing this type of analysis should seek them out as it would be a real time saver. Dean Dominguez uses the following example:
You can try this method with whole novels. Jockers used Portrait of the Artist as a Young Man as a test case:
The plot looks indistinguishable from the ‘noise’ so we use a low pass filter (similar method outlined from the CERN slides above):
Try a Discrete Cosine Transformation as a Low Pass (another way to do a Fourier transformation):
The plot begins to stand out from the noise.
Dean Dominguez asks: But What about Human Readers?:
- How well does this simple approach correspond to real human readers interpretation sentiment?
- In order to debug your computer code or understand your physics equipment, it must be calibrated. I.e. it must reproduce the known correct answer.
- Pay students to read novels and score overall sentiment of each sentence is negative, neutral, positive.
- Compare with computer-generated response.
“Proof of Principle is demonstrated,” says Dominguez:
- “The fairly simple algorithm seems to capture the basic plot arc of the novel
- Without filtering, it is lost in the high-frequency noise
- Calibration signal exists (human coders)
- Jockers has a 40,000 corpus of novels which he is running this on. Looking for archetypal plot arcs is doable
- Similar technique could be used for related analysis on video and audio; suspense; threat”
How can we encourage such connections between scholars in the digital humanities?
While Dean Dominguez met Jockers personally from their work University of Nebraska, Lincoln, he noted that there are many such potential collaborations across disciplines.
The second example deals with network analysis. The question that needs to be asked: “How can we recognize existing and potential possible collaborations?” As a starting point, Dean Dominguez used the College of Arts and Sciences at the University of Nebraska, Lincoln, to see what types of relationships existed between the departments, the scholars, and the administration.
The College has 18 departments in the humanities, social sciences, and sciences. Dominguez wanted to know:
- “What does our externally funded research portfolio and engagement look like?
- Can we gain with further interdisciplinary research activities?
- What role do the existing interdisciplinary research centers play?
- What do the current research collaborations look like in terms of departments and individual PIs?
- What potential collaborations seem to be missing?”
To answer these questions, he turned to network analysis using Gephi. After outlining the methodology (i.e. what are nodes, edges, closeness centrality, etc., see below)……………….
The results on the next couple of photographs show relationships between departments. While some departments are linked ‘naturally,’ other departments are linked by way of another department. Are there opportunities available for collaboration between departments that are not directly linked but seem to have some commonalities?
The photo below shows the relationships between faculty. Notice the circular set of separate dots in the center of the photograph: these denote humanities faculty members. The lone humanities scholar is not a cliche!
Dean Dominguez hopes to apply this network analysis to the departments and faculty at Catholic University to see what underlying relationships exist.
In summary, Dean Dominguez demonstrated with two specific examples of how creative thinking can take place between the sciences and the humanities and that hidden relationships can be discovered. CUA is the perfect place to do this given the liberal arts mandate of the University and the inherent interdisciplinariness of many of CUA’s programs. Dean Dominguez noted that since we are a Catholic school, ‘there is an orthogonal dimension as well: we ask what the meaning of things is, what is true, what is the right way to do it to help people.’
40 people showed up for the colloquium which suggests an increasing interest in DH research at CUA.