It comes from a psychologist named Keirsey who made a bunch of observations after using the MBTI for many years and reworked it.
His book, Please Understand Me II, is the first one I ever read with regards to personality systems, so I'm a bit biased and show some favoritism to it.
Anyway, there are major things different about Keirsey's approach:
- He throws all this nonsense about Fi, Te, Ne, dominant, teritiary functions, etc. out the window. He boils it down to just E/I - N/S - T/F - P/J.
- He develops a model that includes
observable behavioral tendancies: e.g. whether one uses abstract or concrete communication, whether it's directive or informative, etc. This is a huge advantage in my opinion, because it isn't about some obscure "psychological dynamics" (like whether we use Ti or Te more) that we can't even prove exist, and is about something you can observe directly.
- He identifies major motivations and so on that he sees in the types. This, again, has an advantage because he comes up with information that treats the types holistically, and, say, approaches an ENFP as being more than just the sum of E + N + F + P.
He sees the biggest differences in NFs, NTs, SPs, and SJs in terms of these skills sets, motivations, intelligences, etc. So those are the four temperaments, and the individual types (e.g. ENFP) is just a specialization of their temperament (in this case, NF).
I enjoy his approach, honestly. I'd consider myself more in his camp than the classical MBTI camp. I think in terms of temperaments more than not, although IMO there's value in both approaches.
Personality Test - Keirsey Temperament Website