I recently had a discussion with a friend about whether a "cave man-style" language was likely to have ever existed. You know, the stereotypical "Fire bad! Need hunt, go tree-place now!" sort of thing. I guess for this purpose we can define it as a language with very few (if any) structural rules, verb conjugations, prefixes/suffixes, variable word forms, complex sentence structures, or grammatical rules. A language where each word has a single form, and you communicate thoughts just by putting words next to each other without structure or organization. "Need food. Rabbit run fast. Hungry."

On the one hand, it seems natural that the first phase of language development would be very simple, even awkward, before linguistic rules were solidified. (This was my friend's argument.) The idea is that without any education or enforced consistency the new language would remain merely functional, complex only to the point necessary for communicating information, and would only grow beyond that with the advent of "linguistics" as a field of study, which might not happen for thousands of years.

On the other hand, (this was my argument), I'd imagine that as soon as language begins to develop and you start making up words for actions, objects and places, it probably wouldn't be very long before "tree-place" becomes "the local forest," and "lion bad" becomes "lions are bad" or even "when you see a lion it is bad." I tend to think sentence structure arises naturally from the organization of the speaker's thoughts, and while nitty-gritty grammar rules (like "who" vs "whom") might not develop for a long time, things like verb tenses, possessive nouns, and prepositions would develop quite quickly. (Btw, I know some languages still don't use some of these things, but the overarching idea of sentence structure and grammatical complexity is more what I'm talking about, rather than verb tenses specifically, etc.)

So my question is this: do we have any evidence of the very, very first phases of linguistic invention? Maybe some feral group of children who developed their own language, or a remote population uninfluenced by local tongues? Is there any way to know how quickly complex sentence structure developed? Or have those first few steps been lost to history?

  • 6
    'without structure or organization. "Need food. Rabbit run fast. Hungry."' It looks like you're already assuming SVO. There are other things here which to you might look like lack of structure, like droppable subject, lack of subject marking on the verb, zero copula etc, but lots of modern languages have these features.
    – dainichi
    Commented Dec 18, 2013 at 7:08
  • We can only theorize. I expect so (that languages with only uninflected nouns existed), but there's no set of circumstances nowadays to check (no 'cave-men' around). The nominal cave men, the Cro-Magnon (~20K-30K years ago), most certainly had as modern a language system as people nowadays. If there were such a thing as a time machine, we'd probably have to check much longer ago, say ~200-300 thousand years ago.
    – Mitch
    Commented Dec 21, 2013 at 16:32
  • I'm curious to know how natives of languages with diverse structures and features imagine a "cave-man-style" language sounds like.
    – bli
    Commented Dec 21, 2013 at 20:43
  • @bli It seems almost impossible for us to conceive of a communicatively efficient language completely devoid of any inflection/structure. Such a language would be 'weakly compositional', in the sense of Pagin and Westerstahl (people.su.se/~ppagin/papers/pwcompass1e.pdf), i.e. the meaning of a sentence would be a function of its parts, but with no heed to how they combine. "John hit Mary", "Mary hit John", "John Mary hit", etc etc. would all mean the same thing. Seems to me that ordering effects would emerge pretty damn quickly just for the sake of communicative efficiency.
    – P Elliott
    Commented Dec 22, 2013 at 13:25
  • This is all the more interesting given that ape's have sign language.
    – vectory
    Commented Apr 5, 2019 at 22:53

5 Answers 5


This is essentially the lexical protolanguage hypothesis of language evolution, favored by Derek Bickerton, Ray Jackendoff, and others. You can find a nice discussion of this in Tecumseh Fitch's 2010 book The Evolution of Language.


The early versions of Nicaraguan Sign Language were more or less like that - these deaf-mute people would sign sentences like "I go house Pedro", "this road not good", "I want food now" and so on. As it evolved, newer speakers have added complexity onto it: agreement between verb and object, markers for time and mood, spatial references, and so on.

  • Interesting. Any idea how long that process took?
    – Nerrolken
    Commented Dec 19, 2013 at 17:42
  • 1
    I guess it was fairly rapidly. Don't know the exact time frame but shouldn't be that difficult to find out: There've been plenty of studies comparing the 1st generation of signalers of the language with younger generations, contrasting not just the terseness/complexities of their language but also their reasoning and spatial abilities. After all, this language only came into being in the late 70's, early 80's.
    – Joe Pineda
    Commented Dec 19, 2013 at 18:42
  • Be warned, though, that the house-specific signs these people were using before being gathered all together were based on the common gestures typical in Nicaragua, and the order and usage patterns were inevitably influenced by the children's relatives' native Spanish language. So even though NSL grammar has a lot of differences from that of Spanish, it did receive a strong influence from it during its formation phase for it wasn't developed completely "ex-nihilo".
    – Joe Pineda
    Commented Dec 20, 2013 at 14:03
  • It was very quick. As soon as there was a population of deaf children under the critical age, seeing the older children haltingly communicate in a manual pidgin, they developed a manual creole. Idioma de signos nicaraguense.
    – Colin Fine
    Commented Dec 20, 2013 at 22:54

My answer is a little late but maybe someone will still read it...

I believe language started just with sound some higher mammals use - sounds to show pleasure, fear, danger, aggression, and so on.

Humans eventually distinguished loads more such sounds and according emotions, and probably added more types of proto-words - like proto-verbs. Emotions for pleasure got more refined and could mean 'let's eat' or 'let's have fun'. Necessity to replicate such fine nuances to not be considered stupid, miss out on vital clues, or to be able to manipulate others to ones own advantage improved our language learning and vocal abilities, so ever more distinctions could be made.

Eventually, we applied sounds to everything - the noun was born. We also applied them to our fellow humans - probably first with things like emotions noted about another one. 'Let's eat' sounds for someone who likes to eat a lot, aggressive sounds for the alpha male, and so on. Probably also 'run away' sounds for the ones not known for their bravery, or such.

The next step was to combine the two options humanity evolved - emotions/verbs and names/nouns. As they had the same origins and sounded similar, early humans probably quickly made complicated constructions to distinguish one from the other. And there was some evolutionary pressure to understand those things. Suddenly, it was possible for early human 1 to tell early human 2 and 3 to come along for hunting and tell early human 4 and 5 to stay at home to watch out for danger. Even if the grammar was limited to 2 words and some helper sounds to indicate which was the noun and which was the verb. It is unlikely that there was a specific order in the words (verb-noun or noun-verb) - that requires intellectual discipline, which early humans did not have.

A running evolution made sure humans understood the pretty quickly pretty complex proto-grammar. Sounds could be added to indicate something was or something will be, for instance.

Eventually, humans learned to distinguish between subject and object. But still without any special order, and therefore with complicated ways to indicate which is which. Adverb and adjectives followed a few hundred thousand years later. At which time we probably already had recognisable words, recognisable grammar, and so on - complicated through lots of helper sounds to indicate which is which, probably even some hints of what would later turn into cases and such.

The fact that people's increased linguistic capacity allowed them to add more and more otherwise meaningless helper sounds eventually led to words independent of emotions or things. Which allowed articles, prepositions, pronouns, conjunctions, and so on, and with that indirect objects and such. Whereby every new addition made the grammar more complex. Here and there, simplifications were made and standards established which made grammar simpler again, yet still effective.

Language increased to the level of universal concrete languages, and then to abstract languages - a process which is still ongoing, with some languages being more useful for abstract and still understandable content (like English), and others not so much.

There was also the invention of modern grammar, like a fixed verb, subject, object and indirect object position, to avoid complicated cases and other awkward ways to express things. Languages became structured through standardisation and then 'clear' through simplification. Also a process which languages are at a different level in. Chinese is probably among the most advanced in that regard, because their writing system discouraged prefixes, suffixes, sound changes and other more primitive grammar constructions, so that the Chinese were forced to create actual words/signs for every meaning, thus creating an extremely isolating language.


You may be interested in pidgin/creole formation. I have to imagine that early phases of pidgin development have very simple grammars.

  • 2
    By definition pidgins have simple grammars. The development of creoles is thought to possibly tell us something about the 'default' settings for the human language faculty, but it's not something we can be certain about. Commented Dec 18, 2013 at 4:47

Any isolating language would have this type of structure: no difference between parts of speech, no conjugation, no cases, no declentions.

The structures like 'fire bad', or noun + adjective, are typical for Russian.

These structures are also typical for Arabic.

Since even animals have a certain type of language, we have no sufficient data on the existance of a protolanguage of human beings.

  • 1
    Haven't ever heard of constructions like 'fire bad' in Russian or Semitic (I can talk about Hebrew only). The only resemblance is that in both cases there is normally no present tense form for 'to be', but there is still agreement in gender and number: 'Огонь - плох.'
    – alephreish
    Commented Dec 20, 2013 at 19:03
  • I studied Mandarin for a few quarters years ago. It doesn't grammatically mark for some features that English marks, and does grammatically mark for some features that English does not mark. I see no reason to assume that isolating languages are more primitive (i.e. significantly similar to the first languages) because a) isolating languages, like Mandarin, are just as godawful complicated as languages with more bound morphemes, and b) no one knows what the first languages were like. Commented Dec 25, 2013 at 21:06
  • One can say also Огонь - плохо with an adverbal predicate.
    – Manjusri
    Commented Dec 26, 2013 at 10:05
  • @James Grossmann: what are the grammatical features unmarked in English but not in Mandarin? Are they marked isolatingly, or by flection, like in English?
    – Manjusri
    Commented Dec 26, 2013 at 10:08
  • Almost all Chinese grammatical categories are marked with particles rather than inflections. Noun class, which English lacks, is is marked by a system of classifiers, each one a free morpheme. Mandarin also has a system of final particles that grammatically express, for example, reduced forcefulness in a question or command, convey a friendly warning, and convey aspectual information. Mandarin also has particles that mark focus. See ttps://en.wikipedia.org/wiki/Mandarin_grammar#Other_cases for more details. Commented Dec 27, 2013 at 10:08

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.