Digital technologies have been used for humanity studies since the 1940s. After decades of technological development, nowadays scholars have much more advanced tools for multi-aspect research of historical literary documents: geo-modelling and geo-analysis. (Definition of literary documents include not only literature works but also maps, government materials, etc.) GIS softwares as well as advanced data-processing tools have enabled researchers to build cartographic platforms which assembly and link gigantic amount of literary datum. Afterwards interesting results can be obtained. I will introduce three interesting projects of such kind from across the globe.
I. Taiwan Baotu Digitization
First let’s take a look at the project of digitization of Taiwan Baotu (old hand-drawn maps made in the 1900s). As mentioned before maps are also a form of literary document. And the very nature of artifact maps make them the most foundational element when building up digital platforms for historical geography studies. Researchers from various universities and institutes of Taiwan utilize an online map service platform, OpenStreetMap, to digitize the hand-drawn Taiwan Baotu and built their own interactive cartographic platform in order to show what actual life was like in Taiwan 100 years ago.
As a matter of fact, these Baotu had already been online before the study. But as it’s a multi-functional map with different kinds of legends and boundary lines and without longitude and latitude representations, its abundant information is relatively hard for modern scholars to utilize for study of geographical and social changes, or to share their results. To tackle with that, an interactive layer with several data sheets was build on top of OpenStreetMap with the help of QGIS. The final platform makes it possible for users to do online mark, to attach external information, to find the place through tapping the key word, to easier search for the places of the same function, etc. By utilizing to the maximum degree all sorts of information on the old Baotu, the whole platform is built just like a Google Map for 1900s Taiwan.
Old maps are the easiest to be digitized into cartographic platforms, but mapping of old stories and old emotions are not so intuitive. I will introduce them in the 2nd and 3rd part. These two projects have brought about interesting questions as well as interesting results.
II. ElfYelp: Cartographic Platform of Danish Folklores
Personally I think this is the most professional and most meaningful project among the three.
‘ElfYelp’ is the geo-analysis platform developed by UCLA specifically for studying ancient folklore of Denmark. Datum from 20431 legends are taken into account while compiling the cartographic platform in order to make possible future deeper researches on the relationship and inner patterns between storytelling and places. 772 labels are used for the stories to mark. Concentration of labels are calculated on the map based on where the stories were originated:
One thing particular about this project lie in that varies professional data-processing methods have been employed, including data mining and spatial search. Modified LGTA (Latent Geographical Topic Analysis) employs Gaussian spacial distributions to make sure that there’s a probability for each topic on any given polygonal region. Different from the perceptual circumstance, stories here are modeled like waves highest concentrated on one point and ‘decay’ with regard to distance (interpolating is also used to adjust), which corresponds to the real circumstance happened in history that stories and legends are spread over a certain area because of the travelling of men. The calculation is normalized by population.
The three major modes for using this platforms are as follows: 1. One gives the keyword or tag to find places with high concentration of the keyword or tag. In this way researchers may look for the pattern behind the high concentration (with the aid of historical maps): whether population density, landscape features or political/economical situations have to do with it. And they may further look for the reasons. 2. One selects a random area of land on the map, and keywords could be shown by order of ranking of concentration. Thus researchers could have a closer look at specific places. 3. One can also give two different keywords and compare the high-concentration areas in order to determine whether there’s a correspondence between the two. For example, in a preliminary research places with the most ‘witch’ stories are found to be closest to Catholic monasteries.
In all, not many results have been drawn from this platform, but the large database, clever data-processing methods and practical operational modes will certainly enable further exploration of how ancient stories could reflect their social, economical and natural conditions.
III. Maps of London by Emotions in Fictions
With the help of platforms like Named Entity Recognizer and Mechanical Turk, a crowd-sourcing experiment has been conducted by Stanford University in order to depict the emotion map of London based on 5000 English-language fictions of the 18th & 19th century. Unlike the other two projects, their main focus was not to develop their own geo-data platform but rather to utilize several existing platforms to obtain interesting results. They divided the city of London into several polygon neighborhoods and tagged several discrete places in circles in order to define the boundaries of target geographic locations and have a clear distinction on the maps. (This is different from the last project in that they are not following any strict criteria of dividing the map into small, equal regions, due to fact that fictions pay unbalanced attention to different areas.)
No geo-analytical database is generated during the process but three different categories of maps are obtained in the project. Each category consists of four maps featuring the four half-centuries between 1700 and 1900. The first category illustrates in different colors the likelihood for the place (neighborhood or discrete site) to be mentioned in fictions. The datum are collected by researchers as this is an objective issue. On the contrary, second and third categories of maps deal with, respectively, the possibility of the certain place to be the setting of fictions, and, positiveness of emotion linked to the place in the passages. These two issues are comparatively subjective and their datum are acquired through internet crowd-sourcing (according to survey results given by 20 readers). This crowd-sourcing method of data acquisition is also a unique way of conducting literary history studies.
By checking these maps, their research highlight is their abundant conclusions: 1. ‘stillness’: degree of literary attention of different places in London over these four half-centuries remain largely unchanged. 2. After crosschecking the detailed information of sites and neighborhoods on the map, it is revealed that there’s an authentic correlation between places and literary emotions. Positiveness of emotion is influenced by four factors: location of places, function of places, age of places and social class of places. 3. The linked emotions are stronger if the specific place is setting of the fiction.
IV. Brief Conclusion
Geo-related modelling and spatial analysis are tools of the 21st century which bring literary history researchers in front of computer screens. After the process of data digitization, cartographic model construction and characteristic analysis, interesting results could be obtained without having to bury the head into old books and documents.
Huang, Jheng-Peng, et al. “Old Traces, New Links: Representation of Taiwan Baotu in OpenStreetMap.”
Broadwell, Peter, et al. “ElfYelp: Geolocated Topic Models for Pattern Discovery in a Large Folklore Corpus”
Heuser, Ryan, et al. “Mapping the Emotions of London in Fiction, 1700-1900: A Crowdsourcing Experiment”