An Empirical Study of i18n Collateral Changes and Bugs in GUIs of Android apps

Mobile markets allow developers to easily distribute mobile apps worldwide and collect complaints and feature requests in the form of user reviews and star ratings. Therefore, internationalization (i18n) of apps is a highly desired feature, which is currently supported in mobile platforms by using resources flies with strings that can be internationalized manually. This manual translation can be a time consuming and error-prone task when the app is targeted for different languages and the amount of strings to be internationalized is large. Moreover, the lack of consideration of the impact of internationalized, strings can drive to collateral (i.e., unexpected) changes and bugs in the GUI layout of apps. In this paper, we present an empirical study on how i18n can impact the GUIs of Android apps. In particular, we investigated the changes, bugs and bad practices related to GUIs when strings of a given default language (i.e., English in this case) are translated to 7 different languages. To this, we created a source- codeless approach, ITDroid, for automatically (i) translating strings, and (ii) detecting bad practices and collateral changes introduced in the GUIs of Android apps after translation. ITDroid was used on a set of 31 Android apps and their translated versions. Then, we manually validated the i18n changes that introduced bugs into the GUIs of the translated apps. Based on these results, we present a taxonomy of i18n changes and bugs found along in the apps as well as implications of our findings for practitioners and researchers. Online appendix: https://thesoftwaredesignlab.github.io/ITDroid/.