Multimodal Output Management - Man-Machine Dialogue: Design and Challenges - page 160

The alignment takes on an important part in this context, especially after
works on prosodic as well as lexical alignment [BRE 96] and in general along
all the dimensions of language [PIC 04]. Recently, Branigan et al. [BRA 10]
emphasized not only how much the alignment was involved in MMD, but also
to what point it is different from alignment in human dialogue. Alignment with
a machine is seen as a strategy for communication to work. Thus, alignment is
seen as one of the many major challenges for MMD.
9.4. Conclusion
In addition to vocal returns, the system can manifest reactions and answers
through visual returns or even gestures if it can materialize itself with an
avatar displayed on screen. The task can also involve the presentation of
information to the user. The generation of all these messages at the system's
output involves many choices on its part: what information should be
highlighted? What part of the information should be rendered visually? What
part should be rendered orally? This chapter has discussed the necessary
parameters for such choices and the automatic generation techniques of
multimodal and linguistic messages. We also find here the linguistic
preoccupations mentioned in input processing as well as the cognitive aspect
preoccupations such as taking into account human factors to generate adapted
messages.