At the moment I am reading Guy Deutscher's "The Unfolding of Language", in which he hypothesises that modern human language began as sequences of individual words (e.g. "girl run climb tree" or "do work give master") without anything like what we might consider as "grammatical structures" or function words that do not refer to content. Over time, humans gradually develop function words and grammatical structures (e.g. in "do work give master", the word "give" eventually develops into the preposition "for"(just as the word 给 in Chinese is both used as a verb "give" and a preposition "for"). I am happy to accept this hypothesis (language had to come from somewhere).

I have also come to accept what seems to be effectively a rule in Linguistics in that all languages are equally as "good" as one another in terms of expressing ideas (since any argument about a particular language being "better" or "worse" is going to be based on a subjective and arbitrary idea of what is "good"). Languages do not get better or worse over time; they just change. We could say therefore that despite all the change that has taken place, speakers of PIE could just as effectively express ideas (the ideas that they needed to express, at least) as English or Latin or Greek or Chinese speakers.

However, there seems to me to be a huge contradiction between the two ideas that:

1: modern languages had to evolve from something that was not as sophisticated and therefore as effective at expressing ideas as itself (e.g. "girl run climb tree" is not as sophisticated as "the girl runs to the tree and climbs it" for example")


2: all languages over time and space are equally as good as each other in terms of expressing ideas

At what point did the evolution of "grammar" stop and all languages throughout time became as "good" as each other?

I am also not sure how all this relates to the idea of innateness. The earliest human speech proposed by Deutscher obviously did not have any sort of recursion in it. If we (for argument's sake) accept that language is innate, did this evolution in our DNA take place at the same time as our language was evolving, which then allowed languages to use recursion? Did this evolution of our "language faculty" then stop (which allows us now to claim that every attested human language is equally expressive as all of the others)?

Thank you very much for your help!

  • 4
    2 - Languages are equally expressive if we average across all ideas, but not equally expressive at expressing each idea. Clearly English is better at expressing thoughts on containerised cloud infra than Hittite is. Sep 22, 2018 at 11:13
  • 1
    There are just trade-offs that lead to natural limits, it's hard to make them more expressive, you can increase the lexicon or grammatical paradigms but people may not actually learn them so it's not functionally part of their language. Manifested expressiveness is as much a function of culture and knowledge as of language. Sep 22, 2018 at 11:15
  • 1
    Ultimately we don't know the origin of the languages. We certainly have no evidence for more than 10 thousand years ago or so.
    – curiousdannii
    Sep 22, 2018 at 13:03
  • 1
    Related question: linguistics.stackexchange.com/questions/11128/… Sep 22, 2018 at 19:19

2 Answers 2


I take the following to be the central question: “At what point did the evolution of "grammar" stop and all languages throughout time became as "good" as each other?”. The question contains a few important premises, one of which is that the evolution of grammar has stopped. In biology, the term “evolve” is traditionally used to refer to changes in heritable characteristics. Some linguists reject the notion that there is a heritable cognitive faculty of language (as distinct from general human cognitive abilities), and therefore it is not a foregone conclusion that “grammar”, specifically, has evolved in the biological sense. I don’t know of anyone who denies that there has been evolution of general cognitive abilities in hominids, so we might safely substitute “cognitive evolution” for “grammatical evolution”.

However, linguists tend to use the term “evolve” in a broader sense, usually simply meaning “change”. Using this broader sense of “evolve”, we can say that Modern English evolved from Old English, or American English evolved from various extant dialects of British English spoken 400 or so years ago. But that is not a change in a heritable characteristic of the species. If you compare a hypothetical and speculative “cognitive characteristics of hominini some millions of years ago” to modern human cognitive characteristics, there will surely be some heritable evolution. The question is, what is it and who did it affect? It is not know whether there have been any heritable changes to the human language faculty, during the course of human evolution. (By comparison, we know from the fossil record that there have been numerous biological changes in “humans”, whether that be homo erectus, homo sapiens or various extinct intermediaries such as h.s. idaltu).

Using “evolution” in the broader sense “has changed”, without the genetic implications, grammars are always changing, and the assumtion that evolution has stopped is false. Using the term to refer to genetic change, the premise that there has been evolution within the time-span of “humans” is unsupported. It is beyond doubt that there has been cognitive evolution in hominids, but there is no evidence that the basis of language has genetically evolved since the emergence of “humans” as distinct from other hominini. The development of the language faculty in its current form may indeed be the defining criterion by which “humans” can be distinguished from other animals. We need to examine the premise that “modern languages had to evolve from something that was not as sophisticated and therefore (not) as effective at expressing ideas”. In the biological sense, this is probably true in that before there was human language, there were pre-human cognitive abilities which did not include the conceptual faculty of modern (or pre-paleolithic) humans. Human cognition evolved to the point that open-ended systems of concepts like “girl”, “boy”, “person”, “ape”, “tree”, “leaf” could be formed and manifested via distinctive sounds – chimpanzees still don’t do this. The conceptual faculty also developed so that propositions could be formed. When and how this happened is pretty much anyone’s guess.

It is not unreasonable to think that the first human languages had the general capacity to form a hierarchy concepts and to express propositions, and that there was little by way of learned structural baggage. We know that the order of words in an utterance is not genetically predetermined, and we know that morphologically complexity is not genetically mandated. We could therefore assume that the first human language had no polymorphemic words, and word order was free, so maybe Proto-Human sentences were like “girl tree climb” ~ “climb girl tree” ~ “tree climb girl” and in princple the messages might have been confusing. The adoption of a conventional rule “put the actor before the verb” and “put the theme after the verb” could have emerged and been learned by subsequent generations, without the need for any biological evolution.

As for changes in sophistication of language (genetic or otherwise), there is no objective measure of “sophistication”, indeed I don’t even understand what it means for a language to be “unsophisticated” as opposed to being “sophisticated”. There is also no efficiency metric for the expression of ideas. In Logoori, you can express very complicated ideas in a single word, for example rwávataakakʊ́ʊ́ndaamiɲɪrá which we require well over a dozen words to express in English – ‘when they (humans) did not make you curse for me, four days ago or further in the past’. Logoori is so sophisticated and efficient that you can say yaakadééka to mean ‘he cooked, this is just general information’ versus – with a change in the tone of the word – you can say yáákadééka ‘we have been waiting for him to cook, and he has finally finished the cooking’. Unfortunately, the language requires verbose story-telling methods to express the difference between ‘He bought a car’ and ‘He bought the car’.

The claim that earliest human speech did not have recursion is literally question-begging.


The problem with Deutscher’s theory is that it posits the exact opposite of what we can observe in real languages across time. If we look at the long-term development from Latin to Romance; from Sanskrit and Old Iranian to modern Indo-Aryan and Iranian; from Ancient to Modern Greek; from Old Aramaic to Modern Aramaic; from Classical Arabic to modern Arabic dialects; even from Old English to modern English – what we observe everywhere is the shedding of inflections, the abandonment of complex morphological structures, and the movement towards an “isolating” typology.

An “isolating” language like Chinese is not more and not less capable of expressing complex ideas than highly inflected languages like Sanskrit or Greek. What does make a difference is whether any given language has the technical vocabulary needed to express such ideas. You cannot talk about mathematics without having a copious and clearly deliminated battery of mathematical concepts. These can be borrowed from other languages (as mainly happens in English), or newly formed from native elements (as in Greek and Chinese).

  • 1
    I disagree with Deutscher as well, but there is no evidence that languages are all moving towards being isolating. Dixon proposed that languages go through a cycle of being fusional, analytic, then agglutinative. Maybe these languages will gain inflection later on?
    – Toby Mak
    Sep 7, 2019 at 9:34

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.