Program Language
The language of the Twitter user interface is the language that the user chooses to interact with and not necessarily the language that they choose to tweet in. When comparing user interface language with whether location service are enabled or not we find 123 different languages, many of which are in single of double figures, therefore we present only the 20 most frequently occurring user interface choices in Table 5 below. There is a statistically significant association between user interface language and whether location services are enabled both when taking only the top 20 (x 2 = 83, 122df, p<0.001) and all languages (x 2 = 82, 19df, p<0.001) although the latter is undermined by 48.8% of cells having an expected count of less than 5, hence the need to be selective.
8%), closely accompanied by individuals who work together into the Chinese (24.8%), Korean (26.8%) and Italian language (twenty seven.5%). Men and women probably to enable brand new settings use the Portuguese user interface (57.0%) accompanied by Indonesian (55.6%), Language (51.2%) and you may Turkish (47.9%). One may speculate as to the reasons these differences take place in relation so you’re able to cultural and you may political contexts, nevertheless the variations in preference are clear and you can obvious.
The same analysis of the top 20 countries for users who do and do not geotag shows the same top 20 countries (Table 6) and, as above, there is a significant association between the behaviour and language of interface (x 2 = 23, 19df, p<0.001). However, although Russian-language user interface users were the least likely to enable location settings they by no means have the lowest geotagging rate (2.5%). It is Korean interface users that are the least likely to actually geotag their content (0.3%) followed closely by Japanese (0.8%), Arabic (0.9%) and German (1.3%). Those who use the Turkish interface are the most likely to use geotagging (8.8%) then Indonesian (6.3%), Portuguese (5.7%) and Thai (5.2%).
Along with speculation more than these distinctions exists, Tables 5 and you can 6 demonstrate that there can be a user software vocabulary impression from inside the play that molds behavior in both if or not venue services are let and you may if or not a user spends geotagging. Program words is not a good proxy getting venue thus these can not be dubbed as country level outcomes, but perhaps there are cultural differences in perceptions on Fb play with and privacy whereby software language will act as an excellent proxy.
Associate Tweet Vocabulary
The language of individual tweets can be derived using the Language Detection Library for Java . 66 languages were identified in the dataset and the language of the last tweet of 1,681,075 users could not be identified (5.6%). There is a statistically significant association between these 67 languages and whether location services are enabled (x 2 = 1050644.2, 65df, p<0.001) but, as with user interface language, we present the 20 most frequently occurring languages below in Table 7 (x 2 = 1041865.3, 19df, p<0.001).
As when looking at user interface code, profiles just who tweeted during the Russian had been at least browsing have area qualities enabled (18.2%) followed closely by Ukrainian (twenty two.4%), Korean (twenty-eight.9%) and Arabic (29.5%) tweeters. Users composing when you look at the Portuguese was the most appropriate to possess location properties permitted (58.5%) closely trailed by Indonesian (55.8%), the Austronesian code out of Tagalog (the official label getting Filipino-54.2%) and you may Thai (51.8%).
We present a similar analysis of the top 20 languages for in Table 8 (using ‘Dataset2′) for users who did and did not use geotagging. Note that the 19 of the top 20 most frequent languages are the same as in Table 7 with Ukrainian being replaced at 20 th position by Slovenian. The tweet language could not be identified for 1,503,269 users (6.3%) and the association is significant when only including the top 20 most frequent 
languages (x 2 = 26, 19df, p<0.001). As with user interface language in Table 6, the least likely groups to use geotagging are those who tweet in Korean (0.4%), followed by Japanese (0.8%), Arabic (0.9%), Russian and German (both 2.0%). Again, mirroring the results in Table 6, Turkish tweeters are the most likely to geotag (8.3%), then Indonesian (7.0%), Portuguese (5.9%) and Thai (5.6%).