Why do we say "map-territory relation" rather than "territory-map relation"?

"Map-territory" suggests altering the abstracting to fit the map.
"Territory-map" suggests altering the map to fit the abstracting.

Can this formulation suggest:
"Map-territory" -> intensional orientation. [accommodation]
"Territory-map" -> extensional orientation. [assimilation]

Hmmm. "Curiouser and curiouser.", said Alice.

Language is delivered as a linear temporal sequence, and in English the word order sequence is important, and it determines the meaning, as "man bites dog" temporally reverses "dog bites man". In other languages, case indicators such as German 'der' and 'den', Russian word endings, etc., allow for reversed word order without the subsequent change of meaning, although even in these languages the temporal order suggests the primary focus of attention - analogous to the difference between a passive voice and an active voice statement in English. I suspect that if we had been saying territory-map relation for seven or more decades, it would roll of the tongue as easily as map-territory does today. How can we tell whether the comfortable ring of an oft-repeated phrase might have been just as comfortably experienced after similar repetitions in a different word order? Do we have any evidence or research into cadence preferences that have been adjusted or normalized to account for frequency of use? Both mother-son and father-daughter contain the temporal origination sequence. In English classes we are also taught the importance of parallelism in effective communication.

That the word "relation" does not carry with it any implicit order or hierarchy, does not warrant concluding that symmetry may be presumed by default; the default assumption is asymmetry unless explicitly stated otherwise. Other than the relations of "identity" and "non-identity", what is the preponderance of symmetrical and asymmetrical relations in our environment and language? Care to list many? Cousin? Bother? Parent? Child, Co-worker? Co-ed? son, daughter, cover, support, want, etc., etc.

As for leaving out, every formulation purportedly "about" something necessarily "leaves something" out. (You cannot say "all" about anything.) You may pick any "kind" you wish, and so may others.

But "symmetrical?", I'm surprised at you.

Because language is delivered sequentially, and because time is irreversible, we would "naturally" presume asymmetry. This would be especially the case because, 1) the map is not the territory, 2) the map covers not all the territory, and 3) the map reflects the map maker. Number 1) might be evaluated as symmetrical, but number 2), I should think, would not, and certainly number 3) would not. We can have a map as an abstraction from a territory, but we cannot have a "territory" as an abstraction from a map.

In mathematics a "relation" is defined as a set of ordered pairs (n-tuples). The key word here is "ordered". The word relation, I believe, always connotes an order of something "related to" something else. "Correlation", on the other hand, explicitly connotes a "two-way" connection - implying a symmetric relation. For a relation to be presumed symmetric, it must be explicitly so stated, consequently, "relation", unqualified, contains implicitly that it may or may not be symmetric.

Perhaps we should try out territory-map relation for a while; get the feel of it; see if it starts to make any difference in our thinking, reacting, etc.

Perhaps for some of you my "language is delivered as a linear temporal sequence" could have been better said as, "We speak one word at at time. We hear one word at at time." Any "putting together" takes place in the brain later, including the recollection of something previously heard. I recommend On Intelligence by Jeff Hawkins. Read particularly the pages indexed by the word 'prediction'. We are continually anticipating the next word.

In German both of these translate to only one English sentence:
Der Mann beißt den Hund.
Den Hund beißt der Mann.

"The man bites the dog."

The articles 'den' and 'der' indicate subject and object case, but in English it is word order that performs this function. Both English and German have passive as well as active sentence forms, and these are both different from the above.

In Russian the equivalent expressions translate into active and passive voice respectively, so the word order is important in identifying the primary focus of attention, whether it be the man or the dog.

Spoonerisms can fe bun. My somewhat colorful grandmother used to say "fart smeller".

Both stuttering and some forms of dyslexia have been shown by some research to involve sequential processing timing errors at lower levels.

By the time a spoonerism is uttered the linear sequence has been produced after the processing that reversed the syllables. The altered syllables have been uttered in sequence (and subsequently heard in the uttered sequence), one syllable at a time.

In hearing "The man, bites the dog" the timing sequence is critical to "hearing" the unspoken comma, or not, and thence getting or not getting the intended message, as dose the the context that allows one to even expect a poetic alteration of "normal" speech.

The congugation of an article in German is a spoken word; the comma is not.
"Den mann" tells the listener that the man is the object of a verb action, and to expect a verb and a subject actor.
"Der mann" tells the listener that the man is the subject actor, and to expect a verb.

In English, "The man" tells the listener the later, but a passive voice verb requires a retroactive "fix-up" of the expectation.

In both cases the "man" is first placed in the focus of our attention as the primary interest.

A smart feller knows that emphasis alters the focus.

"The man bit the dog." tells us it was not someone else doing the biting.
"The man bit the dog." tells us it was not some other action than biting.
"The man bit the dog." tells us it was not something else that was bitten.

Each case alters what our ordinary linear expectation begins with.

Jeff Hawkins book On Intelligence tells how we can do that with a slow organic "computer" that comes up with answers in under a hundred sequential steps, while high speed sequential digital computers can't in literally billions of steps. Highly recommended reading.