Say what you like, but it IS both quite taxing to create more animated models as well as totally unnecessary. Would be more helpful, sure. I'm not arguing that it wouldn't. Only that there was no shortage of ways to KNOW where you were going and who you were speaking to, based on much more than just character models. Even in these days of much more limited features and far smaller production schedules, things made sense because the overall game design catered to the player's needs. (This in direct contrast to today's designs which excessively pander to the player well beyond simply catering to their needs.)
For example, a group of people in an open room:
|o o o o o
| o
|o o o o o
|
In the above formation, your eye is naturally drawn to the character in the middle. It didn't take unique symbols to accomplish this.
Now add the idea of objects, like furniture:
[ ]o o o o
[ ] o o o o
[ _____] o o
[ ] o o o o o
[ ]o o o o o o
[ .___. ] o o
[_] o [_]o o o
You're naturally drawn to the character centered between the objects, again without need for unique symbols. It could be a the bottom of the mass, in the middle, at the top, the sides, or anywhere. But the scenery itself directs you to where you need to go. When you add a social context to it, it has an even more powerful draw on the player. "Guy behind a desk? Must speak to him!" All of these are done very simply with visual design, not big signs directing you to where you need to go.
(I'd make much more dynamic shapes, except the forum formatting won't let me, so let's just make do with very simplistic, right-angle formations.)
Let's not forget, these designs are all IN ADDITION to character models (using a little bit of variation, not outright unique models) and descriptions. If you're used to newer games which have more money to throw around where all the goons you take out look the same so you know they're unimportant but you immediately recognize the important characters because they're the ONLY ones in the entire game that have their unique model, that just means you're conditioned to the current expectation. It doesn't make it necessary.
I continue to marvel at how much I need to "relearn" when playing older games (or games built on older principles) because of how modern games just throw conditioning at me all the time. Sometimes I notice it, like my complaint that all the NCR troopers in FONV or all the raiders in FO3 or all the hunters in TLU or etc etc all look and act and sound exactly the same. But much of the time it goes by quite unnoticed. Then, pull out an old classic, and there's huge traction because I've grown accustomed to being pandered to. I've forgotten all the skills that came so easily to me when I was younger. Because it really was EASY. We've just forgotten that.
And by the way, I meant that "a depiction should at least somewhat fit a description."
Now it makes sense. I'd still say that's a convenience, nothing more. If the description matches the look, great. If they all look the same but the descriptions are all I have to go by to set one apart from another, that's fine, too. Books got by on imagination, and games did too, once. They don't have to skirt by on nothing but player imagination, but they CAN.