Analysis of PanSp vs. ArgSp users.

(A) & (B) Differential distributions of PanSp vs. ArgSp tweets over CABA+ and CABA extents, respectively. Blue (red) indicates where PanSp (ArgSp) tweets are more prominent. The graph in (A) does not show a strong distinction between the city and the surrounding areas. The most convincing feature is arguably the strip of blue parallel and close to the shore. This cluster of blue bins is similar in shape to the pattern of tourist attractions presented in Fig 10(C), suggesting that tourism may play a role in determining the use of PanSp. The graph in (B), however, does not show a distinctive pattern, showing that the choice of dialect is not strongly determined by location within the city. It should be noted, however, that the overall count of ArgSp tweets outweighs that of PanSp tweets, marking the former as the more standard. (C) & (D) Differential distributions of tweets by users associated with PanSp tweets vs tweets by users associated with ArgSp tweets from (A) and (B), respectively. As in the English vs Spanish case, clear differences exist between the tweet-defined relative distribution and the user-defined one, highlighting the distinction between the geosocial contexts and the social groups associated with the two dialects. In particular, the graph in (C) shows a clear preference by PanSp users for the surrounding areas of the city versus the city itself. The graph in (D), however, shows that within the city there is no strong spatial pattern in the users of PanSp. It should be noted that tweets containing PanSp or ArgSp tokens are a small subset of the total tweet corpus. As a result, the inclusion of all tweets by the users increases the counts considerably. (I) & (F) Normalized angle histograms of the data in (B) and (D). As in the English vs Spanish case, the tweet-defined counts show a much weaker correlation than the user-defined counts indicating that despite the lack of a strong spatial pattern, a geosocial context influencing the choice of dialect does seem to exist. CABA+ Statistics (ArgSp vs. PanSp Users): NT = 594,312, NR = 345,841, SPCC = 0.03, KSS = 0.02, GMI = -0.05. CABA Statistics (ArgSp vs. PanSp Users): NT = 332,455, NR = 192,621, SPCC = 0.02, KSS = 0.04, GMI = 0.009. CABA+ Statistics (ArgSp vs. PanSp Tweets): NT = 18,731, NR = 5,607, SPCC = 0.10, KSS = 0.05, GMI = -0.004. CABA Statistics (ArgSp vs. PanSp Tweets): NT = 10,113, NR = 3,324, SPCC = 0.06, KSS = 0.05, GMI = 0.02. Base map and data from OpenStreetMap and OpenStreetMap Foundation under the Open Database License (Fig 12A–12D).