Getting Started

My first project is to take some of the lessons from social network analysis and see what they yield when applied to literary networks. As my object of analysis, I've decided to use the dataset compiled in 『現代詩1920-1944ーモダニズム詩誌作品要覧』. It provides a record of every poem or article published in 166 different magazines (2,695 volumes in total) between 1920 and 1944, and includes 103,485 works by 4,171 different authors. Owing to the large amount of material and the necessary time it would take to do all of the data entry myself, I've decided to begin with a small subset of poets and trace the networks by which they are linked. It is also hoped that even with a small fraction of poets represented, I might be able to build the foundation of a morophological index to the ways that poets participated in the culture of modern poetry journals. Data from the above volume will also be combined with biographical data to see what kinds of correspondences emerge across different kinds of connections (education, birthplace, literary style).

To begin with, I've decided to work with the subset of poets who contributed to 「学校」(gakkō), a regional poetry magazine started by Kusano Shinpei in 1928. It was a short-lived magazine (printed using the mimeograph) whose contributors consisted of both local poets with ties to the Fukushima region and metropolitan poets already well established in poetry circles. My reasons for starting with this magazine have to do with its small size, its regional locus, and a familiarity with the material. If the initial results turn out to be promising, I will repeat what I'm doing here with a journal of similar size (perhaps a small-time metropolitan publication).

The first step is to devise an effective and efficient method for moving the data from its current form into something that I can manipulate as network data. This means creating a database in which I can store and easily access the relevant information. Until I have some time to teach myself about database design, however, I've chosen to work in Excel to create a simple spreadsheet listing names (with unique IDs), birth dates, birthplace, education, poetry journals contributed to (with unique IDs), and number of poems or essays contributed to each journal. Inputting this information for each of the 25 poets who contributed to Gakkō, while time consuming, will be the easy part.

Having input information for just a third of the poets at this point, a couple of questions and ideas have already come up as to what problems I might encounter at the analysis stage:

1. Does the short life-span of many of these magazines present a problem in terms of capturing social network diversity at the synchronic and diachronic levels? In other words, might we have a situation in which poets moved rapidly through different magazines, resulting in rather simple network formations at any one point in time?

2. This is related to the first point. Is there an effective way to display the life-spans of each magazine alongside the network data? We will need to create some sort of chronological chart to do this.

3. What do we do with those magazines that may have had only a couple of issues? Should we simply reject them as isolates?

4. There is inevitably going to be a certain amount of abstraction taking place here, and one wonders what we lose by giving equal value (i.e., one point) to every contribution. Does a short two-line poem equal the same as a long-verse narrative poem spanning several pages? And what about contributions that appear in the same issue? Should they have equal value?

 

First Steps

This blog has been created with two simple purposes in mind. First, to keep a record of my missteps and small successes as I pursue projects under the rubric of digital humanities 2.0. Second, to serve as a space to introduce and reflect on some of the ways that our appreciation of literature and literary history might benefit from the careful (and always tentative) application of quantitative methods. My specialization in modern Japanese literature will naturally skew these objectives toward certain kinds of content, but my hope is that the methodologies on display here will be of general interest to those in other fields as well.