For example, the participants were not familiar with terms like application and icons. The pictograms and icons we referred to as ‘little pictures’, for example. We encouraged them to talk-outloud especially about any problems they encountered or parts they found unclear. We stressed that if they did not understand the application or parts of it was not their fault but the programmers. Once we felt that they were confident and understood the main features of the application we started the usability test, which focused on multi-sentence icons, message composition and reading, and specifically the karaoke feature. We started with the two different versions of the multi-sentence icons both of which were available from different icons on the same screen. We compared the pre-listen version we introduced in the first prototype with the pop-up based one introduced in the second prototype. Recall that in the pre-listen version users had to tap and hold on the icon to listen to all the sentences associated to it in a row. Then, to select the sentence number i, users had to tap on the icon i times. In the pop-up version users were required to tap and hold on the icon and a little pop-up appeared on the screen with all the associated sentences. To select a sentence, users simply had to tap on it. For both versions we asked them to longtap on the multi-sentence icon and queried whether they had an idea on how to append one of the offered sentences to the message editor. For the composition, we situated them in the following scenario: “Let’s suppose you received an SMS from Amisha a friend of yours. You can see you have a new message from Amisha in your INBOX [participants are in the thread screen]. Now, you can tap on this message to see why Amisha is sending you this SMS [participants navigate to the message details screen (see Figure 1)].” After they had navigated to the Conversation screen we made them listen to what Amisha had sent by tapping on the play button next to the text message “Cinema tonight?” that was on top of the list. To double check that they had understood the audio message we asked them to explain why Amisha had sent them an SMS. After their explanation, we asked them to reply that they were not free tonight with “Tonight, no.” To make it easier, we broke this task into two subtasks through which we walked the participants: (1) We asked them to reuse the word “tonight” from the previous message by first finding it in the previous message and to append it to the message editor. When necessary we reminded them to use a double tap on the word. (2) We asked them to find the icon “No” from the list of icons in the Quick sender screen and to append it to the editor. Before sending they had to review the composed message by tapping on the “play” button. For reading we tested what happened if we removed the karaoke function (the words currently played were highlighted in red). We had two versions of the application: one with karaoke support and another one in which the whole sentence was played out but with no visual feedback in the UI. We tested two sentences in French “When do you come back?“ (“Tu rentres quand?”) and “Cinema tonight?” (“Cine ce soir?”). First we asked the participants to play out the sentences and identify as many words as they could in the karaoke version. Then we asked them to repeat this with the same sentences in the version without the karaoke. They could listen to the message as many times as they wanted. At the end of each sub-task completed we provided congratulatory or encouraging feedback. For the usability test, the video recordings were our first source of data collection. We reviewed the recordings and for each performed task reviewed the kind of errors they made and on which screen it occurred. Due to the low number of participants we did not conduct any statistical tests, however. During the debrief interviews, we elicited if they found the application easy to use and whether it would be useful for them in their everyday lives.