Our students like this activity and created a significant amount of high-quality documents (only one presentation was clearly plagiarized).
Here are the word clouds results from two open feedback questions: “What do you like about using VoiceThread in the course?” and “What do you not like about using VoiceThread in the course?”. The figures render the students' answers as word clouds in which the frequency of a word is shown with the font size (common words like "a", "the" are removed).
According to the feedback, students highly value being able to communicate ("communication") in "English" with their "friends" (this includes their classmates and the "foreign" students). They find VoiceThread "interesting" and like the multimedia features ("voice", "pictures", "video"). The emphasis of on "make" and "share" indicates that the students like to perform these activities often associated with constructivism and enabled by Web 2.0.

When asked about what they did not like, most complaints were about the slow connection speed of VoiceThread, from the class-room but also from at home and work. While watching the threads was possible most of the time, commenting, especially with voice fre-quently did not succeed. One student stated that the registration was too complicated (even though it required only to select a user name, pass-word and to give an email address).

While VoiceThread had advantages, a very often too slow connection, disappearing comments, incorrect statistics and difficult group management made the learning unnecessarily complicated and sometimes frustrating. An alternative would be to build up a multimedia discussion tool from several Web 2.0 services, such as video sharing sites (YouTube or YouKu, the Chinese counterpart), photo sharing sites (e.g., Flickr), video blogs, etc., and to integrate them into a Personal Learning and Teaching Environment. That is what we are currently exploring in the context of the ROLE project.
The word clouds images were created with Tagxedo, a very helpful service.
We (Kerstin Borau, Daniel Jackson and Scott Grant) submitted a more detailed discussion of the study to ICWL. In case you are interested in more details, let me know and I will post some more.