Erik,
You and others have raised some great points here about competencies. Yes, we have worked hard with the NWS to get people thinking more in terms of job responsibilities versus traditional disciplinary divisions. Its a hard perspective to change, given that that is often how we are educated at school, but also how expertise is developed and recognized (we speak of radar experts, but how often do we speak of short-term forecasting experts?).
I have to admit a little concern sometimes that people can take competencies too far, and begin to feel that teaching and assessing the isolated sub-sub competencies will allow people to learn and demonstrate that they can fulfill the actual broader job responsibilities. That is why I like the approach that Ian describes, that he and Jeff are promoting, and that the CAeM group are applying in writing Aeronautical Forecaster job competencies. Starting with the highest level job responsibilities not only helps make sure that you are getting to all the important more detailed competencies, knowledge, and skills, but also reminds you that you also have to teach, provide practice in, and assess the larger competency as well. People have to be able to perform the larger task in context. This is also why I like ROMATSA's insistence on making on-the-job training a formal part of the training program, and also part of the assessment as well.
By the way, a really nice but rather long treatment of this whole-task vs. part-task issue is covered in a book by a professor from the Netherlands, van Merrienboer's "Training Complex Cognitive Skills." (I have to confess not having read the whole thing, but even the overview of the model provides a lot of good food for thought.)
Joe is the first author of that presentation and paper you mentioned. I'll see if I can find a copy to add to this session's resources as requested. Ian may have some good things to share also.