Research continuum

Backstage at the lab

Reading beyond time and across disciplines

One scientist working on the archives said, “A great earthquake historically occurs once every hundred to several hundreds of years in a given region. To understand earthquakes and to make use of lessons learned for future disaster prevention, one must read old literary works on past disasters.” Modern scientific research goes beyond the boundaries of traditional scientific research domains. Collaborations between scientists and humanities scholars have become more common. Digital technologies have accelerated the process of deciphering old literary works. Here we interview scientists who have gone beyond their usual fields to inquire about the impact that reading old archives has had on their research.

Seismologists and old literary works

One day in early Autumn 2015, a group was struggling with an old drawing shown on a screen under the high ceiling of the Science Seminar House on the University’s North Campus. They were sitting in a U-shape, and reading brush letters written on the drawing. Because the age and gender of the group were diverse, upon first glance, it was hard to tell their purpose. They were slowly deciphering the texts, “In the eve…ning, there was…a mountainous tsu…nami…?”

The drawing vividly depicted people in a panic following a large volcanic eruption of Mt. Asama, which occurred in the third year of Tenmei, or 1783. The drawing contained explanatory texts written in brush letters, and a lecturer was teaching the group how to read the letters.

This is a snapshot taken during a camp workshop of the Old Earthquake Study Group led by Professor Ichiro Nakanishi of the Seismology Lab. The study group meets weekly. In addition, the group holds an annual camp workshop aimed at intensively reading old literary works that describe past disasters. The group is comprised mostly of students in the sciences and researchers in disaster science, but also includes non-researchers and librarians.

One of the group members, Yasuyuki Kano, an assistant professor of the Disaster Prevention Research Institute (DPRI), explained, “I love measuring – measuring just anything. I have conducted research on changes in water level data collected by monitoring underground water, and earthquake-induced expansion and contraction of the Earth’s surface.”

His response begs the question, why is Kano, who is an experimental scientist, a member of a study group reading old literary works?

The answer turned out to be quite simple. “Numerical data on earthquakes have been collected since the Meiji period, but a great earthquake occurs only once every couple of hundred years. Because data has only been collected for the last 150 years or so, knowledge about earthquakes is very limited. To improve our understanding, we must read about old earthquakes recorded in old writings.”

However, old literary works pose a great challenge to modern scientific researchers. Drawings and literature describing old earthquakes are written in the sosho or the gyosho style, both of which are a style of kuzushiji, or deformed, cursive characters. It is therefore necessary to spend substantial time and effort to read them fluently. This reading and rewriting into modern script is called honkoku, or transcription We then asked Dr. Kano, isn’t it more efficient to ask students or researchers in Japanese literature or history who are supposed to be familiar with honkoku, to read old literary works rather than for scientists to tackle them?

He responded with an example. “For instance, suppose you find an account of a massive wave in the sea in the archives. If you are a disaster prevention science researcher, you can tell by reading the account if the massive wave is a tsunami caused by an earthquake or a large wave caused by a typhoon.”

Research on earthquakes in the distant past is not a new concept. The trailblazers of Japan’s seismology, who experienced the 1923 Great Kanto earthquake firsthand, belonged to a generation that could read old literary works. In fact, they also conducted research on records of old quakes. In recent years, the extensive research done by Tatsuo Usami, a professor emeritus of the University of Tokyo, has advanced honkoku and the organization of earthquake accounts in old writings, which had been stored in a number of places throughout Japan. “One of the tasks of future research on old earthquakes is to read more archives to extract information, and organize this into a searchable form for future generations,” said Kano. Kyoto University, which has a large collection of literature about earthquakes, is essentially an unlimited room for study.

However, the goal of earthquake research is not merely to read old literary works; it is also to organize and analyze the data obtained through honkoku, and publicize the findings in a usable form. This will allow mitigation strategies against earthquakes to be prepared in the near and the distant future. When Dr. Kano was trying to find an efficient reading method, he found SMART-GS.

Innovation spawned by the Faculty of Letters

Screenshot of SMART-GS showing an imported writing sample on an old earthquake (Courtesy of Yuta Hashimoto)

SMART-GS is a software program developed by Susumu Hayashi, a professor of the University’s Faculty of Letters, in order to transcribe the handwritten notes of German Mathematician David Hilbert and trace his philosophies. Dr. Hayashi imported Hilbert’s handwritten notes, which were made available by the University of Göttingen, into a computer as image files using SMART-GS, and added his interpretations or annotations to the original notes as text data. When an illegible letter is encountered using SMART-GS, it is exported as image data and compared to similar letters by placing these side-by-side and considering the context where the letter originated. Then a conjecture is drawn about the letter. A link of the letter image can be left as supporting data for the conjecture. Additionally, the transcription and annotation can be shared with other researchers, allowing collaborations across multiple disciplines.

Kano, who was searching the internet for useful tools to advance the work of the Old Earthquake Study Group, stumbled upon SMART-GS by accident. He later realized it had been developed at his same university. Kano and Hayashi, who work in completely different fields, did not have a chance to meet until early Spring 2014, when they were assigned to the same room as admission exam administrators. Their relationship has evolved since then. Hayashi and Yuta Hashimoto, a graduate student in Professor Hayashi’s research group, who had been involved in the development of SMART-GS, have become frequent visitors to the Old Earthquake Study Group, leading to rapid advances in their collaboration.

How is it that this kind of software was developed at the Faculty of Letters? Professor Hayashi, the developer, is presently an expert in historical sociology, but he majored in mathematics as an undergraduate and graduate student. He also studied computational science and wrote books on programming. Hashimoto, who has been familiar with computers since his childhood, graduated from the Faculty of Letters, and worked in the business sector as a system engineer for three years. Because both Professor Hayashi and Hashimoto were familiar with computer science, they were able to develop SMART-GS, which caters to the needs of researchers.

Kazu Nagai, another professor in the Faculty of Letters, used SMART-GS in a honkoku publication project involving the handwritten diaries of Yuzaburo Kuratomi, the chairman of the Privy Council of Japan, over a seven-year period during the transition from the Taisho to the Showa period. Actually, it was during this project that SMART-GS evolved to incorporate a work-sharing function based on the requests of researchers involved in the project.

In recent years, digital humanities have received increased attention in Japan and around the world. The humanities encompass disciplines dealing with human nature, including literature, history, linguistics, art, etc. Digital humanities fuse two seemingly remote fields: digital technologies and humanities. Digital humanities emphasize questioning humanities through innovation and sharing of research methods such as letter data analysis, creation of image databases, and applications of spatiotemporal data. Kyoto University is a pioneer of this field in Japan. In September 2015, a Japan Digital Humanities Conference was held at Kyoto University, including many researchers from overseas.

The initial objective of SMART-GS development was to read the handwritten notes of Hilbert independently. However, researchers in need of this tool prompted the developer to make it sharable among researchers, realizing more fruitful collaborations. The software has further evolved via the Internet into a tool that is free from the constraints of time or distance, allowing researchers to interpret literature in depth at an unprecedented speed. The software has also enlightened researchers because mundane reading for their research has evolved into collaborations that go beyond the boundaries of academic disciplines and national borders.

“People who want to read kuzushiji are everywhere,” said Hashimoto.

Reading beyond borders

Kuzushiji learning app, KuLA

In February 2016, a research project “Development of an International Educational Program for Japanese Historical Writings” (led by Yoichi Iikura at Osaka University), of which Hashimoto is also a member, announced the release of KuLA, a kuzushiji learning app for smartphones and tablet computers. This allows users to learn how to read kuzushiji in a quiz format, aiding not only Japanese students but also those who are studying Japanese history or Japanese classic literature overseas. Feedback from overseas researchers who want to read the original texts of old archives is reflected in its design, making it highly suitable for learning kuzushiji.

This application began attracting attention in social networking circles even during its development stage. Readers of this article might be somewhat surprised by the fact that those who want to read kuzushiji coincide with social networkers. In fact, there has been a kuzushiji boom among players of an online sword game, in which famous Japanese swords are personified. These gamers wanted to read and write kuzushiji.

It is said that more than one million writings were made by the end of the Edo period in Japan. The National Institute of Japanese Literature, which is a research institute of historical writings, heads a large database project “Construction of the International Collaborative Network on Japanese Classical Books.” The objectives of this project are to digitize historical writings found throughout Japan and publicize them over the internet. If historical writings written in kuzushiji become easily accessible, anyone can access the wisdom and intelligence of people who lived before the Edo period. Even a kuzushiji non-expert can appreciate this. This project in combination with the collaborative reading functions of applications, such as SMART-GS, should facilitate collaborations among strangers, which may help decipher even dauntingly difficult kuzushiji.

Furthermore, there are successful examples of cloud transcription taking place overseas. For example, the University of London in England has organized a project that transcribes a huge number of handwritten notes left by Bentham, a philosopher. To date more than 350,000 people have participated.

Kano has a strong desire to read old literary works about disasters in the Old Earthquake Study Group in collaboration with people living in areas affected by such disasters, and then apply this knowledge to actual disaster prevention strategies. If kuzushiji learning and cloud honkoku become more popular, the movement will elucidate records not only of past earthquakes but also of other kinds of disasters. This will greatly contribute toward disaster prevention research today.

Hashimoto also offers another application, a “Kin-digit Reader”, to facilitate reading of the “Kindai Digital Library” (for kindai, or modern period), in which digitized data of the National Diet Library since the Meiji period have been provided. (The Kindai Digital Library was integrated into the “National Diet Library Digital Collection” in May 2016.)

“There is another significant point in developing this kind of digital reading-support technology,” said Hashimoto, “the interpretation of data varies by time. For instance, ‘Keian no Ofuregaki’, literally the Keian regulations (Keian is the period corresponding to the reign of the emperor), which can be found in history textbooks, had been taught as regulations issued by the bakufu, or shogunate, to control farmers. However, a new interpretation, which has been shared among researchers of late, regards ‘Keian no Ofuregaki’ as local regulations originally issued by a local authority that were subsequently spread to a broader area. This new interpretation has been reflected in descriptions in recent history textbooks. If this kind of interpretation can be accumulated and saved, people 100 years from now will be able to study ‘our interpretations’ as historical data.”

Reading archives is one research method used around the world. Reading literature with fresh pairs of eyes should produce superior results. Moreover, reading with additional eyes from overseas will advance international collaborative research and generate new findings. A new world of reading facilitated by digital technologies and the internet is just around the corner, and readers will not always have to be full-time researchers. Anyone will be able participate in these projects.

Interdisciplinary research potential at Kyoto University

In recent years, the fusion of humanities and science or interdisciplinary studies has encouraged the development of new research fields that go beyond typical areas of inquiry.

Traditionally, interdisciplinary studies have involved researchers applying a useful result in a particular field to their original area of research. “I am simply not used to the idea that everything is divided into humanities or science. I have an impression that many of the researchers in Kyoto University are like me,” says Hashimoto.

One scientist in the Old Earthquake Study Group has increasingly devoted his time to the reading of kuzushiji. Takatoshi Sakazaki (of the Research Institute for Sustainable Humanosphere and a research fellow of the Japan Society for the Promotion of Science), an expert in atmospheric physics, has lately been studying typhoons using old literature.

“Numerical data analysis is what I usually do, but I originally liked history. It is pure enjoyment for me to be able to read old literary works. People in those days were very attentive. They wrote very detailed accounts of weather conditions during a war. If there is no honkoku version, I have to read them on my own.”

The approach of the Old Earthquake Study Group has no boundaries between humanities and science. Curious minds transcend traditionally established boundaries easily and effortlessly through the sheer joy of learning.

Yasuyuki Kano: Assistant Professor,
Research Center for Earthquake Prediction, DPRI, Kyoto University

Yuta Hashimoto: PhD candidate, Graduate School of Letters,
Kyoto University