...as being compatible and well-correlated, in terms of the existence of relevant outward behavioral patterns and external similarities in all ways but those directly attributable to distinct functional use, with paired functional preferences.
Personally, something that I like very much about the three-letter model of referring to types (eg EIE, as opposed to ENFj) is that it leaves out the rational/irrational (j/p) indicator. I personally think that it would be altogether better if people stopped thinking of "people" as being irrational or rational, and started thinking of the functions in this manner instead.
I have come to think that it is hilarious that people describe others and themselves in terms of "temperaments" that correlate to outward behaviors. Personally, I don't think the temperaments really apply to people in any way other than their function usage. It seems to have become some kind of mythology among people in this forum that "IPs are lazy," "EPs are spontaneous," "IJs are rigid," and "EJs are active," and these traits are used as litmus tests for the supposed temperaments of people.
In the reality of Socionics theory, temperaments refer to the nature of functions. Now, in case you have forgotten, functions are not governing factors of external behavioral traits, as in Meyers-Briggs and related theories, but rather methods of information metabolism. They work on such a "smaller" scale, perceptually speaking, than the nuances of a person's external behavior, that it truly is ridiculous to assume that the two are directly related in any kind of meaningful, recognizable trend.
These constructs in no way necessitate or dictate any kind of outward behavioral patterns. I will agree that correlations can, in fact, be made, but the idea that there are underlying governing tendencies that can be used as "models" into which to fit people as a basis for typing seems absurd to me. However what I REALLY find preposterous is that people reject the idea of another person being a type because he/she does not "appear" to have a certain temperament; using it on a discriminatory basis (in terms of negating a typing; not the "mean" kind of discrimination) seems highly fallacious. It is potable that there are some vague behavioral tendencies that transfer (and would have to be dependent on subtype as well, IMO) from function usage to outward behaviors, but the idea that a person's behavioral "temperament" is in some way governed or has a strong correlation to their Socionics temperament seems, to me, to be a futile and baseless notion.
I will say that, with time, it is possible for a Socionist to develop a personal system of indicators by the use of which they are, perhaps, able to correlate vague notions of what being a "rational dominant" might look like, or how an "extravert" typically behaves. However, this takes time and the practiced typing of, I daresay, thousands of people, and, in the end, is really nothing more than a feeling, and can't really be used by anyone but the typer himself as "real" evidence for a typing; it's not the kind of thing you can explain any more than by saying "This is my impression." The usage of "traits" as concrete indicators, a sort of "checklist" for what a person of a given temperament should or shouldn't be like, is absolutely laughable, and should be stopped immediately by the use of any who wish to consider themselves credible typers.