Our interest in human–computer interaction started with our participation in the TT2
project (“Trans–Type-2”, 2002–2005—http://www.tt2.atosorigin.es), funded by the
European Union (EU) and coordinated by Atos Origin, which dealt with the development
of statistical-based technologies for computer assisted translation.
Several years earlier, we had coordinated one of the first EU-funded projects
on spoken machine translation (EuTrans, 1996–2000—http://prhlt.iti.es/w/eutrans)
and, by the time TT2 started, we had already been working for years in machine
translation (MT) in general. So we knew very well which was one of the major bottlenecks
for the adoption of the MT technology available at that time by professional
translation agencies: Many professional translators preferred to type by themselves
all the text from scratch, rather than trying to take advantage of the (few) correct
words of a MT-produced text, while fixing the (many) translation errors and sloppy
sentences. Clearly, by post-editing the error-prone text produced by a MT system,
these professionals felt they were not in command of the translation process; instead,
they saw themselves just as dumb assistants of a foolish system which was
producing flaky results that they had to figure out how to amend (the state of affairs
about post-editing has improved over the years but the feeling of lack of control
persists).
In TT2 we learnt quite a few facts about the central role of human feedback in the
development of assistive technologies and how this feedback can lead to great human/
machine performance improvements if it is adequately taken into account in the
mathematical formulation under which systems are developed. We also understood
very well that, in these technologies, the traditional, accuracy-based performance
criteria is not sufficiently adequate and performance has to be mainly assessed in
terms of estimated human–machine interaction effort. In one word, assistive technology
has to be developed in such a way that the human user feels in command of
the system, rather than the other way around, and human-interaction effort reduction
must be the fundamental driving force behind system design. In TT2 we also
started to realize that multimodal processing is somehow implicitly present in all
interactive systems and that this can be advantageously exploited to improve overall
system performance and usability.