Some general pointers (not sure about the literature that is out there) and hopefully people can build on some of the things I have listed here:
If you are talking about a usability test then you should be focusing on goals and tasks first and understand the success rate. I suspect that people's emotions would be affected by how usable something is, so you want to make sure that you understand the effect and impact.
For gathering emotional reaction, I think you are looking at more qualitative rather than quantitative metrics, so you might want to first build up a list of words that describe the product and the experiences that people would have.
The behaviour that users display after interaction with form elements can be explicit or implicitly. For the explicit behaviour you'll have to try and capture these either by raw data (e.g. video or eye/mouse tracking) or accumulation of information (heat map), and for the implicit behaviour you might have to design a survey or questionnaire (or interview them directly) to extract these.
You are correct in pointing out that what people say and do can be different, but it needs to be addressed in the context of what you are asking them to do. You may even find that in your case what people say and do match quite well, so you shouldn't make that assumption first until you look at the data, but make sure that you do capture both what they say and do. Generally if what users say and do vary, it can be attributed to their expectations of not just the application, but also their own assumptions about what their behaviour might be.
I am looking for some help here...