We have minimal languages with few words, like Toki Pona, and I have made a rough language with about 4000 words which covers most of what you would need to say as base concepts (4000 base concepts), as an extremely rough estimate. Then you combine these things to get the millions of specific concepts, roughly speaking. But these languages, as far as I know, have "modern" grammars roughly speaking.

In doing a conlang, and thinking of languages like Vietnamese, Chinese, or even Hebrew, Arabic, and Swahili, you always want to make it speakable like a "modern natural language".

I have a line of imagination whereby I'm thinking that early in language evolution languages might not have been fully formed into "natural" languages like they are today. Instead they might have been, well, for lack of a better term, "primitive". By primitive I mean, they would have lacked subtle grammatical concepts and may not have even had "words" for highly abstract concepts. In thinking briefly about it, words like even "move", "make", "do", etc., these are all extremely abstract. Actions like "jump", or "sleep" or "climb" are easy to point to, as are comparing things to get adjectives like "red", or objects like "tree" and "leaf". So there is potentially a long gap in language development between realizing we need words for the abstract concepts to make it easier to talk about things, and when we just had things for the obvious concepts (things/features/actions which you can easily point to, for example).

In addition, I try and think how they could have even come to the idea of creating chains of sounds (like words), and it seems that would be a long evolution. Maybe they started realizing they could make individual sounds ("mm"), then two-sound chains ("ma"), then further 1-syllable sounds, then to two syllable sounds. Who knows. But in guessing, I would like to build a conlang which takes 1 syllable words and sequences them into a thought, without making it flow smoothly like a modern natural language.

Are there any languages like this (conlangs, creoles, etc.)? Any sources of inspiration I should take a look at?

What I'm imagining is something like:

Tree. Climb. You. Me.
Tree. Climb. You. Me.... Eat. Fruit... Take. Nap... Then. Walk. There (gesture point).

Well, really, "take" is an abstract concept. So it would be like "Nap" instead probably.

The order of the words seems like it would be more fluid rather than fixed (English is more fixed word order). You would talk slower because you need to let an imagination sink in for each word stated (in the listener), since all this is new.

  • Tree (point to tree, pause for a few seconds)
  • Climb (gesture climbing, pause for a few seconds)
  • You (point to you, pause for a few seconds)
  • Me (point to me, pause for a few seconds)
  • etc..

I don't have a better way of thinking about how a "primitive" language might look. I don't mean it needs to be choppy and stereotypically "caveman-like", but that's what it seems like I'm saying. But no, I am trying to say something where each word has magical meaning, which you have to ponder to understand, and use other associations and gestures to get the meaning of the speaker, before there are grammatical niceties like adpositions and such (assuming English in this example), or affixes.

Are there any conlangs like this which I could look at? If so, knowing a quick example of how it works would be helpful. If not, what about creoles or pidgins, how do they compare? If nothing comes to mind, how could you string together concepts without modern natural language grammatical niceties, in your imagination?

  • Based on natural human languages, if there's any general trait of "primitive" languages, it's probably that they're more complex. Simplification of cases, tenses, etc over time seems quite common. What you're describing seems more like a pidgin, which can indeed be quite simple, but as soon as people start learning them as a first language they instantly gain in complexity and sophistication.
    – curiousdannii
    Commented Mar 13, 2023 at 5:30
  • @curiousdannii if what you were saying is true, then it would be a snap of the fingers to teach chimpanzees or other animals how to speak. I feel like you are intentionally ignoring the fact that there is a huge gap learning-wise (evolutionarily too) between voice-sound-making and primitive word stringing together. Once you get to stringing words together, then yeah maybe it can jump easily to "modern natural language", but I'm asking about the first to second phase transition here, not the second to third (as I've outlined).
    – Lance
    Commented Mar 13, 2023 at 9:54
  • 2
    No, I was saying that our evidence of human languages shows that so-called "primitive" languages are just as complex as "modern" languages, if not more so. If anything I'd posit that perhaps their morphological complexity was simplified to compensate for a gradual increase in vocabulary/semantic complexity, though I have no actual evidence for that of course.
    – curiousdannii
    Commented Mar 13, 2023 at 11:24

2 Answers 2


I don't know of any attested languages that work this way, constructed or natural. Humans are very good at finding ways to communicate, and if you put children together with no exposure to outside language, they will invariably come up with something complex and sophisticated with a full-fledged grammar. Look at how pidgins turn into creoles*, for example, or the origin of Nicaraguan Sign Language (ISN)*. Indeed, some linguists (most famously Noam Chomsky) point to this as evidence that language is innate, and humans have a built-in "universal grammar" mechanism in our brains to make this work.

If you want to find a "primitive" language like this, you're going to have to look back tens of thousands of years at least, which means we're firmly in the realm of speculation. But you might find Ljiljana Progovac's work interesting; she analyzes certain "fossil" constructions that are remarkably consistent across living languages, looks at how they compare to early stages of infant language development (like the "two-word" stage), and speculates about how people might have once communicated using only these constructions.

Her basic thesis is that the first stage of communication was assigning names to things and actions ("tree", "run", "eat", etc), then stringing two of these together to express actions on objects ("I eat", "eat plant", etc; notably, she points out that most languages allow these sorts of two-word constructions where the noun is either the subject or the object of the verb, so a "turncoat" turns his coat but a "turntable" is a table that turns). Eventually, the next step was using one of these two-word constructions as a component in a two-word construction, introducing recursion to the system. She speculates that this would have been an evolutionary benefit since it would make someone better at insults, which could be used to resolve social disputes in a non-violent way (and thus not injuring other members of the group).

This is, again, very speculative. But it might be an interesting foundation for a conlang.

* These examples aren't entirely uncontroversial: there are a few different theories out there about how creoles form, and some linguists believe there was outside intervention involved in ISN. But the general consensus—and what we teach our introductory linguistics students—is what I've said here.


There is of course the Pleistocenese, somewhat plausible artlang meant for the Neanderthals. In many ways it breaks our preconceptions, it has no separate phonemes (word=morpheme=syllable≈phoneme), all the "words" are uninflected monosyllabic units, there are no regular word classes, no higher grammatical abstractions (the syntax is a patchwork of irregularities). The language is on one hand rather complicated for a modern human to learn, on the other hand quite incapable of conveying elaborate philosophical ideas (evenif you somehow added the vocabulary).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.