To expand on this: the mapping between images and other cues to a precise meaning is often actually pretty poor. To correctly navigate using images you often have to have prior knowledge, with the exception of the most downright obvious images or visual cues possible. On the other hand, text can have essentially arbitrary precision (although past a certain point it becomes difficult to parse) - and is thus actually often superior for first-use (or infrequent use) scenarios. The ideal is to have the best communication possible, and while I'm not sure as to the extent that Uber reaches that goal, text being unreadable and the app being unnavigable because you changed the language is, in my view, more on you than on the app developers.