Language Enhanced Virtual Scenarios (LEVS)
Many simulation modalities struggle with the practicalities of learning communication skills. Sophisticated mannequins such as Laerdal’s SimMan may have an embedded speaker to simulate spoken voice but you can’t talk to him. For most such simulated communications, we have had to rely on Standardized Patients: actors who play the role of a patient most convincingly. There is great fidelity but the chief problem lies in cost and scalability, with a 1:1 ratio between SP and learner.
We are seeing the increasing use of chatbots in commercial customer-facing enterprise web sites. For a limited number of possible responses, these are now able to handle an impressive degree of accents and voiced phrases. However they do require many weeks to program the corpus of responses and logical pathways.
In response to this need, we modified the approach taken by many text-chat interfaces: the wee text box that you see when you ask to “chat” with an agent. This approach has been very successful in many ways.
Originally called Turk Talk, this method was based on the concept of the Mechanical Turk, which we have documented here. Centuries ago, a Byzantine emperor impressed the audiences of his court with an automaton that could play chess. The ruse was that the automaton had a human hidden inside: a human playing the role of computer.
Amazon has built upon this with their Mechanical Turk service so the phrase is well known and widely used.
Our TTalk interface has proven to be remarkably effective when integrated with OLab, our virtual scenario platform. We have described this in several places.