Phase I began with the knowledge that no spoken Spanish language data would be available for analysis; the original proposal stated that data already available at NMSU would be used at the program startup. However, after a meeting with Carol Van Ess-Dykema, the principal investigators became aware that future data collected by Pitt would be of a certain format, namely, setting up a business meeting. This had already been started for collecting English language data by the Janus project. Consequently, it was decided not to use the data mentioned in the proposal, and to collect new data using the Janus protocol.
An initial problem with this task was the fact that it was not known exactly how the Janus protocol was set up. We had an idea of what it was like from discussions with DoD contacts, and as time was of the essence decided to approximate the Janus protocol with one of our own. At that point data collection and transcription was begun.
For the Artwork data collection, speakers were told that they were to set up a meeting for the near future. They were given a schedule for a specific day, detailing hour by hour what commitments already existed. A coinciding half-hour time slot was open on the schedules for both speakers; other than that the period between 8:00 a.m. and 5:00 p.m. was ``booked up''. The first dialog was recorded and transcribed into English, and analyses of that text were begun. At the same time additional recordings and transcriptions of dialogs were carried out. Initially, the principal investigators thought that a plan analysis would be of interest, as the speakers evidently brought some type of plan- meeting structure with them to the task. Some plan-structure analyses were applied to the data, and a detailed plan-structure schema based on the initial dialog was developed.
The initial analyses resulted in the conclusion that more knowledge was needed about the structure of the Spanish dialog. For example, an analysis of plan structure presupposes some knowledge about Spanish discourse structure, which was not available. A bibliographic search resulted in finding no immediately pertinent research in this area. As a result, it was decided to analyze the dialogs from a discourse-oriented perspective, in order to establish a base for further analyses. Hence, a major task was to identify discourse functions in the dialogs, and the difficulties encountered in translating Spanish to English.
At about this point in time, it was recognized that the
translations could not be objectively analyzed, as they were the
work of one of the principal investigators. It was decided that a
professional translator who was not participating in the project
would be hired to do the remaining translations. This resulted in
ten translated dialogs.