Thoughts on working memory (for self record)

man, I think I picked the hard topic to look at...
If I were to give the definition of "working memory", I would end up burying myself in the literatures of working memory, but still couldn't give out a clear model of definition for it without confusing myself to death. This is how I have been feeling whenever I found any study talking about this topic.
I gradually learned (still not very skillfuly) to first identify which camp they are coming from:

1. Second language acuqistion camp: these people are langauge educators, applied linguists, and people who like to borrow psychological concepts and studies. The implication of WM for them is the predictor for learning a second/foreign language. Intuitively everybody would think that if you can remember a sequence of foreign sounds and be able to repeat promptly, that means you have some kinda "talent" for learning languages.
But knowing this and to prove it doesn't solve my question. What's working meomry in a neurobiological sense?

2. Psychology camp: they are the people who first brought up the idea of working memory from the concept of short-term memory, including Baddaley himself. Since they made up the term, their studies are being quoted by others all the time, but only at the introduction. After that they just go on whatever directions they like to talk about this term.
3. Neurolinguistic people: looks like these people like fMRI very much. A lot of them believe in UG and language area it seems like. The attempt is to find out the location of working memory, which is prefronal cortex, along with some other regions like temporal gyrus, parietal areas being found in those imaging studies.
Sometimes these localization studies or ROI ("region of interest") look cool but couldn't explain what's going on underneath the cortex. When parietal cortex and cingulate both light up, what does it tell us about the brain?

4. Neuroscientist: Fuster, Goldman-Rakic et al...
These are the people whom I think have cool ideas about working memory. Fuster has this "cognit" neuronetwork explaning cognition in the cortex. He discovered the memory cells when monkeys doing delayed response task. Same kinda cells are found in different regions of the brain when doing other tasks such as tonal, visual-spatial, color, etc. Goldman-Rakic found cells firing at certain "memory field" when monkeys doing delayed response tasks through single cell recordings. Another way to understand WM and the relationship with prefrontal area might be through pathological studies, like autism or schizophrenia. For example, "autism" (though this term needs clear definitions too!) subjects tend to have working memory deficits, same as schizophrenia patients, WM integrity can even predict whether they can reintegrate socially or replase. Some studies connect these deficits to dopamine projection dysfunction in prefrontal cortex, which I am still trying to figure out. I got the feeling this might the direction I want to go just to avoid the endless debate on what working memory does and what it actually is.