On March 11, 2011, a magnitude 9.0–9.1 earthquake (東日本大震災) struck off the coast of Touhoku, Japan. The earthquake triggered powerful tsunami waves, which together with the earthquake itself wiped out much of the region's infrastructure. The disaster caused nearly 16,000 deaths, over 6,000 injuries, approximately 2,500 missing persons, and hundreds of thousands of displaced residents, ranking it as the most powerful earthquake on record in Japan and the most devastating disaster to strike Japan since World War II. The region continues to recover and rebuild, including from the nuclear crises that resulted from the accident at the damaged Fukushima Daiichi Nuclear Power Plant (福島第一原子力発電所事故).
The Japan Disasters Archive (JDA), a project from the Reischauer Institute of Japanese Studies at Harvard University, seeks to archive websites, news articles, video, images, and other media related to the March 11 disaster and its aftereffects in the time since. The archive links out to materials in partner projects, allowing users to contribute new materials as time goes on. Through the JDA interface, users can search for materials by relevant keywords, item date, language, and media type, making it an excellent resource for studying the ongoing development of Japan's response to the March 11 disaster.
There are many different ways to capture the experience of a disaster. In this series of visualizations, one subset of media from the JDA — a collection of testimonials about individuals' experiences of the disaster — is mapped and annotated. The testimonials portrayed here are responses to the question, その時、何をしていましたか? What were you doing at the time the disaster struck?
ハーバード大学のライシャワー日本研究所の「日本災害アーカイブ」(JDA) は、東日本大震災に関するウェブサイト、記事、動画、画像などの資料を保持するように活躍するアーカイブである。JDAを使ってユーザーはキーワード、言語、貢献日付などのフィルターで資料が検索できます。
Loading Visualization, Please Wait...
Roll over the visualization above to read individual testimonials
Narrative Arcs in Descriptive Language
In the visualization above, the specific language used in each testimonial to describe experiences of the disaster is annotated and analyzed. Each arc of circles represents a single testimonial; each circle represents a single word in that testimonial. The sizes of the circles are proportional to the relative frequencies of each word. Furthermore, the distance of each arc from the center corresponds to the relative geographic distance of the testimonial from the earthquake's epicenter. Roll over the circles to see the words, and click on individual circles to find their occurrences in other testimonials. The highest-frequency words are listed to the right of the visualization, including their counts across the testimonials. Click these words to highlight their instances across the testimonials.
Collective Memory and Geographic Displacement
The language used to describe collective memory of the 3.11.2011 disaster is geographically distributed in relation to the earthquake's epicenter. The visualization below illustrates these trends; while words such as 地震 and 揺れ are distributed widely among testimonials throughout the country, some words such as 放送 and 様子 are distributed further away from the epicenter.
Loading Visualization, Please Wait...
Loading Visualization, Please Wait...
Word Co-occurrence
Throughout the testimonials, certain words are used together, resulting in clusters of language that illustrate common ways of describing individuals' experiences of the disasters. In the visualization below, these word co-occurrences are visualized in a matrix. The top 50 most frequently used words are plotted on the vertical and horizontal axes, and each square represents the frequency with which each pair of words occurs together across all the testimonials. Each square is colored in proportion to the frequency of the word pair; brighter colors indicate pairs of words that appear together in the testimonials with high frequency, and darker colors indicate low-frequency pairs. When you hover over an individual square, you can see the word pair highlighted and the number of documents in which that pair occurs.
This visualization reveals interesting trends in the data. For example, the words 「地震」 and 「揺れ」 appear in high frequency together as well as with nearly all other high-frequency tokens. Other high-frequency pairings are predictable, such as 「駅」 and 「電車」. And yet there are other pairs of words that are intriguing in how frequently they occur. For example, 「津波」 and 「映像」 appear together in 12 documents, as do 「テレビ」 and 「被害」 . Such pairings provide an illustration of common experiences across all testimonials, regardless of proximity to the earthquake's epicenter.
Loading Visualization, Please Wait...
Data analysis and creation of this visualization were done by Steven Geofrey using D3.js. Tokenization of the testimonials portrayed here was performed using the Yahoo! Japan Morphological Analysis (日本語形態素解析) API. The source code for this project is available on GitHub.