Language Dispersion Patterns in Bruxelles Through the Lens of Twitter
A continuation of my previous experiment (see url below), this piece provides a comparative view of the top 10 (+ 1 native) commonly used languages in geolocated tweets collected from Bruxelles in 2013. The dataset originally consists of 2,199,905 geolocated tweets but has been filtered down to 1,796,980 (≈82%) based on the languages in this visualisation. The languages have been arranged in descending order by tweet quantity beginning with French, the native language. Overlaid ellipses indicate the median dispersion centres of the tweets in each mini-map. The size of each ellipse serves to describe the concentration of tweets for that language. Small ellipse suggest that a given language is confined to a certain region of Bruxelles while larger ellipses imply that the language is spread across the city.
There are several immediate discoveries to be made. First, Turkish tweets can be distinguished as having a distinctively different dispersion pattern from other languages. Next, the overall dispersion pattern across all languages (except English) in the months of July, August, September and October differ from those of the rest of the year. This difference may be a result of summer activities. Finally, tweets in Turkish, Indonesian and Tagalog are dispersed differently in October although tweets in Tagalog appear to have an erratic dispersion pattern in general.
This work has been funded by KU Leuven as and developed under the guidance of prof. Andrew Vande Moere. For more information related to our research, please visit: http://rxd.architectuur.kuleuven.be