When it comes to complexity and language, any complexity you add to the morphology is complexity you take away from the syntax, and vice-versa. For example, when you learn all the noun cases of Finnish, it buys you having to remember fewer constructions with adpositions—or fewer verb augmentations, if the language went that way.

Syntactically, Valyrian is usually (MODIFIER) NOMINATIVE-NOUN (MODIFIER) OTHER-CASE-NOUN* (ADVERB) VERB. It's quite simple. There's not a lot you have to remember, and things can move around a little bit, if it feels right. You don't have to remember a ton of auxiliaries with different applications and slightly different usages. For the most part the heavy hitters (the nouns and verbs themselves) take care of things rather nicely. This is what complexity within the words themselves buys you: simplicity elsewhere.

The reason you get this is because all languages are doing the same thing: describing human experience. And humans are the same language to language. The other small tidbit is that when creating a naturalistic language—and it doesn't matter what method you use—you are, unconsciously or not, aiming for the lowest common denominator in terms of grammatical complexity. You don't have to do that, but generally if you're trying to create a language for humans with no other goals, you do. With a language like Ithkuil, John was intentionally pushing away from what is standard in human languages, and so there are needless levels of complexity that push beyond the boundaries of ordinary human language.

Now, when I say "needless", this is what I mean.

In Turkish, if you want to say "The girl is reading a book", you say:

Kız kitap okuyor.

Turkish is a language with noun cases, but you only see the nominative here. Why? Because the girl is reading A book. When the object is indefinite in Turksih you don't need to use the accusative case—in fact, you shouldn't. If you wanted to say "The girl is reading the book", that's when the accusative case pops up:

Kız kitabı okuyor.

Okay, with this in mind, you've introduced—just in the nouns—four possibilities:

Nominative + indefinite

Nominative + definite

Accusative + indefinite

Accusative + definite

In a maximally complex language, all of this would be marked. In Turkish, only one of these is marked. (Well, maybe two, if you were to say Bir kız for nominative + indefinite. Turkish has an indefinite article that pops up sometimes.) Certainly there are languages where all of these have some sort of marking, but then those very same languages will have other situations where maximal marking is possible but not present.

Human languages all have this in common. There are areas in the language where more categories could be marked but are not. It doesn't matter what the language is. This is because humans have limits for how much junk they'll tolerate in the language they're using. It isn't long before something that could be inferred from context is inferred from context. It collapses every so often (i.e. too little is marked and so marking pops up), but the unconscious goal is for the language to have a balance between morphological and syntactic complexity and also explicitness and implicitness.

A language doesn't have to do this, though, and so conlangs can be more or less explicit/implicit. Can they work? Certainly, but they may be more than humans will comfortably tolerate, and so humans may not want to use them.

Take Láadan, for example. Had Láadan been created later it might have had a better shot at being used, but this was 1982 before conlangers had started getting together. Láadan primary flaw is that it's trying to be a deep philosophical experiment while also trying to be a language a lot of people speak. That was never going to work. Suzette Haden Elgin lamented that maybe women didn't want a language of their own to use, and so the experiment was doomed from the start. A simpler explanation is she saw an ocean and built a train to cross it.

In Láadan, every sentence begins with one of six speech act particles (copied from Wikipedia):

Bíi: Indicates a declarative sentence (usually optional)

Báa: ndicates a question

Bó: Indicates a command; very rare, except to small children

Bóo: Indicates a request; this is the usual imperative/"command" form

Bé: Indicates a promise

Bée: Indicates a warning

And then in addition to that, every sentence ends with one of the following (also copied from Wikipedia):

wa: Known to speaker because perceived by speaker, externally or internally

wi: Known to speaker because self-evident

we: Perceived by speaker in a dream

wáa: Assumed true by speaker because speaker trusts source

waá: Assumed false by speaker because speaker distrusts source; if evil intent by the source is also assumed, the form is waálh

wo: Imagined or invented by speaker, hypothetical

wóo: Used to indicate that the speaker states a total lack of knowledge as to the validity of the matter

This is too much! Evidential systems in language exist, but they are so much smaller than this, and usually the markers pull double duty—and there's often a null marker.

Again, though, it's about the goals! This is fine for a philosophical language. And if it was simply a philosophical language, then how many people "speak" it is irrelevant. For example, John Quijada doesn't lament that after twenty years there isn't a community of Ithkuil speakers—indeed, he's baffled whenever he hears of someone who wants to try to "speak" Ithkuil. It's not designed for that, and so the metric isn't a fair one. Based on the structure of Láadan, I'd argue the same: the number of speakers/users isn't a fair metric, and shouldn't have been a design goal. Because while a language like High Valyrian looks more complex, with its declension classes and conjugations, Láadan is more complex in that it exceeds the expectations of explicitness a human user expects from a language.

This is why conlangers shouldn't study Navajo. It is by far an outlier in terms of how much complexity is in the grammar alone, and studying it will make it sound like the native speakers use all this grammar. Which they don't, when it's implicit, which the textbook can't tell you (since nominalized verbs usually need more structural support than unambiguous nouns, it varies highly).

The biggest conlang I had inspired by Navajo, Hlūf, I had to make a billion features optional, because of the story:

"Apisawekumumehaeskelelewihē" is how the textbook will tell you to write "Alright (concessive), let's suppose she causes you to be hurt over and over again." This is how you'll speak if you're giving a speech in parliament.

"Api, asa kumumehaᴉhwē aoe 'kelel?" is how a native speaker would assemble the sentence using a local system of mutations, and thinking through the sentence as it's being said.

"Appi assa wıᴉ kūhru *gestures of repeated punching* moha, meha meha owë?" is how my protagonists will say it.

That's supposed to be a native English speaker in a foreign land, making a clumsy pidgin out of the dictionary terms - isolating lemmas, using extra pronouns, and second-language-errors like mixing up "kumu" with "kūhru," meaning "to make" like crafting an object, not like causing an emotion.

So there was a justified artlangy excuse to make the language "complicated" - the story requires non-linguist readers to tell the fluent from the clumsy speakers at a glance. Over the course of the story, the reader should hopefully remember a couple words, and the isolating pidgin will make it so eventually they can recognize a suspicious keyword. Even spoken aloud, this wouldn't work. But you can pick up the pattern when reading comic speech balloons "fluent speakers use long words and choppy speakers use short ones." They may not know what verb conjunct slots or oligosynthesis are, and neither did I when I was a kid, that's okay!

i really do

but i don't wanna draw like shit

cause i did a poll on my tumblr asking who wanted plant hrt

nine people voted yes

0 no

but i draw like shit and i dont wanna give those 9 people shit

(idk why im sending an ask about that i just feel like telling someone yk)