A geolinguistic approach for comprehending local influence in OpenStreetMap

OpenStreetMap (OSM) thrives on allowing anyone in the world to contribute features to a free online geographic database, thereby allowing international mixes of contributors to create the map in any given place. Using South America as a test area, I explore the geography of OSM contributors by applying automated language identification to the free-form comments that contributors make when saving their work. By cross-referencing these languages with selfreported hometowns from user profiles, I evaluate the effectiveness of language detection as a method for inferring the percentage of local contributors versus the percentage of "armchair mappers" from elsewhere. I show that most English-speaking contributors to the South American OSM are from outside the continent (rather than multilingual locals). The percentage of English usage is higher in poor areas and rural areas, suggesting that residents of these places exercise less control over their map contents. Finally, I demonstrate that some features related to daily needs of health, education, and transportation are mapped with higher priority by contributors who speak the local language. These findings give researchers and organizations a deeper understanding of the OSM contributor base and potential shortcomings that might affect the data's fitness for use in any given place.