Linguistic Landscapes at Scale: Affordances and Limitations of a Mobile Data Collection Approach
David Joseph Wrisley
Abstract – 2016 LAUD Conference
Institut für Fremdsprachliche Philologien, Fach Anglistik
A pilot study of the linguistic landscape of Beirut began in the Fall of 2015. The data has been collected by mobile crowdsourcing: a group of fifteen Lebanese language students who use the resultant dataset in a course project. At the time of writing this abstract, the total dataset has reached 600 samples from the city of Beirut and its suburbs. After initial analysis, the data model will be refined and the project will continue, running until the Spring of 2017. The linguistic situation of the Lebanese Republic–and in particular the capital city Beirut–is complicated by a number of factors: (1) differing patterns of multilingual practice marked by class and communitarianism, (2) the orthographic possibilities of alternating between Arabic and Latin scripts and Oriental and Western numerals and (4) what I called „audience-specific multilingual non-equivalence.“ Can LL studies say something about the intended audience of different national communities of varied multilingual practices? Unlike notable LL studies that examine the imprint of immigrant communities in specific neighborhoods of diverse metropolitan environments (Blommaerts), this project describes „indigenous“ multilingualisms, without restricting the zones in which data is collected and without prescribing a level of granularity of the data. Instead of qualitative ethnography focusing on deep local context of a small contained area, the project resituates the notion of ethnography in a data-driven, geospatial frame. Geosemantics, and discourses in place (Scollon/Scollon) are replaced with the possibility of spatially-bound multilinguistic patterns.
Written language samples are captured by mobile application and the image samples are human tagged with metadata based on a boundary vocabulary of contextual and linguistic features specific to the Lebanese language situation. The platform employed permits live visualization of the aggregate of the data. It also allows access to a growing database of geotagged images of written samples of language, and their visualization of them on a map interface using queries based on their metadata. The scope of data achieved in a small amount of time produces interesting visual results, but ones that we must „read“ with care. We must create a continuum, including which kinds of language samples are not location-based, which ones might be location-specific, as well as those which are most susceptible to revealing spatial patterns. The overwhelming amount of bilingual samples collected were unsurprisingly English/Arabic. In the paper I will discuss some of the others patterns in LL that the study has found about languages in contact in the Beirut area (Arabic/French, Latin transliteration of Arabic, minority languages as Armenian and Tagalog) as well as refinement needed for the metadata.
Analysis in a data-driven LL study recalls debates in literary studies between distant and close reading, having both affordances and limitations. Larger sets of data reveal patterns that were not previously visible and abstract representations of them (like a map) provide new avenues for contextualizing such data. There is, however, a need for flexible visualization possibilities to keep all scales in mind. Interface design, for example, facilitates an explorative close-distant readings with the possibility to visualize language sample and its metadata in the information box.
Keywords: LL, mobile data collection, map visualization, spatial humanities, digital humanities, GIS, Arab world, network analysis