Boldur Barbat
November 7th 2002, Montegrotto, Italy
The paper addressed the role of voice communication in interactions with machines

Boldur first clarified the meaning of some of the terms used in the title: e.g. ‘language’ implies natural, spoken language, while the plural ‘languages’ is used to describe all other languages. He noted that the whole issue of how we can best interact with machines had become more important as, socially, we were now in a large feedback loop through the advent of the WW and other ICTs which both make more communication possible and require it. Various developments leading to more communication were listed: including the rise of the WWW, broadband, the development new interfaces and increasing system intelligence. In addition, there were various reasons why this topic should be particularly salient to the COST269 action, which is why it had been considered as a candidate topic in the past.

The paper raised a number of issues concerning debates about how important it was to have interfaces that used spoken language, including some beliefs that they are a panacea and the major direction in which we should be moving. For example, it was problematic to see orality as more ‘natural’ and other languages as ‘artificial’, in that all languages are socially codified signs, including the body language that can itself be used within mediated communications.

The next sections list the ‘human-centred’ pros and cons of spoken language interfaces. The former include the importance of speech generally in human communication, various specific situations where it is essential or important and examples of where it has already proved useful in relation to ICTs. The list of disadvantages is even longer (see the paper), and includes that fact that spoken language creates distortions and is not well suited to human-machine interaction. This is followed by a list of current technological developments, and some evaluations of them, which have a bearing on the issue of spoken language interfaces (see the paper). On the one hand these include improvements in voice recognition, but on the other users hear human sounding voices they anticipate that they are interacting with a person and this sets their expectations too high.

Ultimately the aim of the paper is to be problem-posing rather than problem-solving, outlining the various considerations we need to take into account. Boldur proposed that this, and other such problem-posing papers from other workgroups, appear on our conference web-page to show our thoughts about issues and discussions as part of the process of preparing for Helsinki. The paper gave some other examples of topics that his colleagues might be likely to discuss at the conference.