So I was asked: Is there a reason why a language might have a more complex conjugation system?
Short version: Each language seems to have the same level of complexity as any other, but some manage to keep it in places where learners don't have to deal with it early on. Sometimes the complexity ends up in the conjugation system, where you can't get around it, and sometimes the complexity ends up in the sentence structure, where you can fake your way around it for a while and still do okay.
Basically, some aspects of language change can reduce superficial complexity by increasing underlying complexity. An example of a single instance of that happening in French is responsible for the consonants at the end of French words that are silent or not, depending on context. That is, a superficial layer of language (how many sounds are pronounced in a typical word) got simplified, but the expense was in producing greating underlying complexity -- people had to start thinking about which sounds in which contexts would be silent or not.
Now, normally that kind of complexity gets wiped away ("regularized") before another of the same kind of change can take effect. But if it just so happens that the initial complexity doesn't get smoothed out, then the complexity sort of piles up.
Several layers of that kind of thing happening is what produces the kind of complexity that you find in Navajo.
Here's a totally contrived example to illustrate things:
Imagine you have a language where all verb roots are single syllables, and you conjugate them by adding two prefixes: first a single-syllable prefix for the subject, and then a single-syllable prefix for tense. Suppose it looks like this, at what we'll call Era 1:yo-si-ha = I walked vo-si-ha = you walked yo-ma-ha = I will walk vo-ma-ha = you will walk
The exception in the command form, which is just the root plus a suffix "ko":ha-ko! = Walk!
And suppose that negation is a preceding word "me": </pre> me yo-si-ha = I didn't walk me vo-si-ha = you didn't walked me yo-ma-ha = I won't walk me vo-ma-ha = you won't walk me ha-ko! = don't walk! </pre>
Now, suppose that you always know where the stress in a word is: it's always on the first syllable of a word; so the above is pronounced like this:yósiha = I walked vósiha = you walked yómaha = I will walk vómaha = you will walk me yósiha = I did not walk me vósiha = you did not walk me yómaha = I will not walk me vómaha = you will not walk háko! = walk! me háko! = don't walk!
The language at this stage is very straightforward -- in this aspect at least; who knows what horrors lie elsewhere! Languages that are "simple" in one aspect usually have nasty surprises in another aspect.
Now hop in a time machine. Suppose that the h's in those unstressed syllables tend to get dropped -- but you can't drop them between identical vowels, or else it "sounds weird". So now you end up with:yósia = I walked vósia = you walked yómaa = I will walk vómaa = you will walk
Meanwhile, people stop thinking of "me" as a word by itself, and start treating it as a prefix that tries to attach to the next word in as short a form as possible. But where you can sort of say "myo", you definitely can't say "mvo" or "mho" as a single syllable, sort of like how "don't" is one syllable, but "doesn't" is two.
So now, in Era 2, the language looks like this:yósia = I walked vósia = you walked yómaa = I will walk vómaa = you will walk myósia = I did not walk mevósia = you did not walk myómaa = I will not walk mevómaa = you will not walk háko! = walk! meháko! = don't walk!
Notice that we used to know where the stress was in a word (always at the beginning), but now it's a bit messy.
Now, hop in the time machine again. That "aa" at the end of words starts to get pronounced as stressed, because if you don't stress it, it sounds like just "a" instead of "aa". So now you have: At the same time, unstressed a's at word-end start to sound more and more like "uh" and then people start dropping them altogether. And along the way a dialect appears where the tongue-twister "myo" gets turned into just "mu". So now, in Era 3, you have:yósi = I walked vósi = you walked yomá = I will walk vomá = you will walk músi = I did not walk mevósi = you did not walk mumá = I will not walk mevomá = you will not walk háko! = walk! meháko! = don't walk!
This is the amount of language change that could happen in as little as two or three centuries. Each little step (which is more than a single person normally perceives in their lifetime) "makes sense" by itself, and is the sort of thing that happens in language change all the tame. But as the changes pile up, things got very messy.
Remember that in Era 1, "I won't walk" was "me yomaha", and each syllable there had discrete meaning: "me" was not, "yo" was I, "ma" was will, and "ha" was walk. But look at the Era 3 form: "mumá". The best you can explain this is to say that the "m" means "not" and the "u" means "I" and the "ma" means "will" and an "invisible a" means "walk".
Here's the important fact: Children can pick up Era 3 language about as easily as children used to pick up Era 1. But adults have a hell of a hard time learning a language like Era 3.
Now, normally when a language gets this messy, people start "regularizing" it -- maybe in this case they declare "ha" (walk) to be irregular, and start replacing it with a new word, "po" (suppose it used to mean just "step" or something) that doesn't have that "irregularity", and maybe it starts looking like this, in Era 4-hypothetical:yóspo = I walked vóspo = you walked yompó = I will walk vompó = you will walk múspo = I did not walk mevóspo = you did not walk mumpó = I will not walk mevompó = you will not walk hápo! = walk! mehápo! = don't walk!
And suddenly things look a lot more learnable again. That sort of thing tends to happen, but it's not a force of nature or anything, so maybe Era 3 speech turns not into Era 4-hypothetical speech, but into something even nastier -- from the point of view of adults who're trying to learn it. From the perspective of adults who already speak it, or of little children who are learning it, there's no real difference between how "hard" any of these forms are -- they all seem perfectly natural.
It so happens that a lot of Native American languages are like Era 3 talk, or "worse"; and European languages tend to avoid being like Era 3 talk, sort of, most of the time -- and instead have messy syntaxes.
So if you learn a European language, you can learn to say "I walked" pretty easily, and only later do you find trouble when you find out that it needs some special complexity when you want to say "He wishes I would walk" or whatever. But the point is that you can sort of fake your way around that kind of thing -- if you say "He wishes I walk", it's still intelligible.
But in a language with complex word structure (instead of complex sentence structure), you can't fake it -- the big mess in the contrived Era 3 isn't the kind of thing you can sort of fudge. These mostly accidental differences between Native American and European languages are why Native American languages like Navajo are generally quite hard to learn except as a young child.