I would say we SHOULD try to explain language evolution independently of genetics. But what do you mean by EVOLUTION? Ontogenetic or phylogenetic? If ontogenetic we should study Vygotski carefully. If phylogenetic we should be aware that there is no evolution to a higher point. All languages we know are at the same hight. So we have only change. Difficult to say why.

Klaus,

My question was about the referenced book by D.Dor and E.Jablonka

Remy,

excuse me, I did't see your reference. And I didn't read that book. But I stick to what I said about phylogenetic evolution of language to a higher point: There is no scientific proof of something like that.

It seems to me that you could have a look at work in Systemic Linguistics (Halliday, Martin, Matthiessen etc.). Language has co-evolved with the evolution of the brain. Eco social context: increasing complexity of all the physical and social processes shape and define the conditions of human life. The evolution of the meaning potential of the human species involved the human body as signifier and the ecosystem as signified.

https://www.youtube.com/watch?v=nC-blhaIUCk

I know this does not answer your question directly but Halliday does recognize concepts that you seem to be interested in such as phylogeneesis, ontogenesis and logogenesis.

Keith, when s.b. says "Language has co-evolved with the evolution of the brain" isn't it pure guessing?

@Keith and Klaus

Re co-evolution of language and brain:  For a book-length analysis of the issues, see Deacon, Terrence W. (1997) The Symbolic Species: The Co-evolution of Language and the Brain, New York: W. W. Norton.  As evidence of his credentials, Deacon earned a PhD from Harvard in 1984 with a dissertation on "Connections of the inferior periarcuate cortex in Macaca fascicularis: A comparative and experimental neuroanatomical investigation of language circuitry and its evolution. 379 pp."

The issues are controversial, but the evidence is more than anecdotal.  For a more recent book that discusses related issues, see Lieberman, Philip (2013) The Unpredictable Species: What Makes Humans Unique, Princeton: Princeton University Press.  Philip's son wrote a related book from a different perspective:  Lieberman, Daniel E. (2011) The Evolution of the Human Head, Cambridge, MA: Harvard University Press. 

For an overview with references, see the slides http://www.jfsowa.com/talks/goal2.pdf .

And thanks, Keith, for the pointer to Halliday's YouTube lecture.  He is an exact contemporary with Chomsky, but I believe his contributions to the study of language are more insightful and more important for future research.

I fully agree with John Sowa.

Remy, of course, languages are rooted in social environments, but that doesn't mean genetics is unimportant.  IMO many theoretical discussions on speech & language origins have become irrelevant since we know that Pleistocene Homo dispersed along African & Eurasian coasts (e.g. S.Munro 2010 Molluscs as Ecological Indicators in Palaeo-anthropological Contexts, PhD thesis Austr.Nat.Univ.Canberra) where they not only beach-combed & waded bipedally for littoral foods, but also parttime dived, which can explain human voluntary breathing control.

As Darwin & many others noticed, music (rhythm, variable & loud tones) was already present in gibbon duetting. Gibbons & our ancestors split early- or mid-Miocene. Much later (early-Pleistocene?) Homo spread intercontinentally along the coasts as far as Algeria, the Rift, Georgia & Java: Aïn-Hanech, Turkana, Dmanisi & Mojokerto, all 1.8 Ma, are coastal or connected to the coasts.  AFAWCS, the combination of gibbon-like song (Miocene apes) + adaptations for seafood collection (pre-sapiens Homo) provided all necessary elements for human speech:

- Voluntary breathing control evolved when Pleistocene Homo (

Given that grammar is typically considered not to be subject to selection, I have attached an excerpt from my 2015 OUP book (Evolutionary Syntax) in which I address this question by exploring a hypothetical scenario in which historical change and genetic selection can go hand in hand. It is a response to questions posed by reviewers, and is not meant to prove that historical change must be accompanied by genetic change, but only to suggest that nobody can guarantee that it cannot be. 

Marc,

Of course, but my question was set with regard to Dor & Jablonka's book. Basically, assuming that languages capabilities are set by genetics (see for instance Deacon), why not see language as a symbolic tool freed from genetics, with grammars reflecting the needs of communication and symbolic representation of human as social beings.

Ljiljana,

But then, what's the need of the assumption ?

Hi Remy,

I think my previous comment addresses your question to some extent (and the short excerpt I attached), suggesting that not even grammar can be exempt or freed from genetics. 

Remy,

Do you mean why one needs the assumption that genetics played a role in language evolution? I would say that we do not need such an assumption, nor the assumption that genetics did not play any role in language evolution. What we need instead are very specific hypotheses addressing these questions, which can be tested, and such testing has already begun. 

Ljiljana & Remy, IMO everything in biology is subject to selection, no doubt language too is subject to selection, but as Ljiljana says, it's not easy to see genetic changes on a very small scale, and it's of course difficult to disentangle e.g. tonal skills in speaking from musical skills in singing.

Ljiljana, your view is very interesting, I've never considered this in so much detail. FWIW, some thoughts. In the earliest 2-word sentences, there was probably no sharp distinction between V & N-words (although I read once that in our brain, nouns are found in the posterior=sensitive part of the brain, and verbs in the anterior=motor part). Chimps in sign language have a lot of repetitions and no fixed word order AFAIK, e.g. "apple give me give" or so, but we can expect that what the speaker notices first usu.comes first in speech, and that the agent comes before the result (cf SV), although in the first intransitive "sentences" VS is perhaps about equally likely, e.g. "fall man". As soon as all this becomes formalised, we may expect more rigid word orders, usu.SV I'd think, although possibly early languages were often ergative, e.g. VE "fall man".

Marc,

"everything in biology is subject to selection", indeed, but while the ability to speak clearly belongs to biology, that's not necessarily true for grammar.

Ljiljana,

There is no symmetry between "assuming that genetics played a role in language evolution" and  "assuming that genetics didn't play a role in language evolution".

Marc,

Yes, I think that is true, that it is not possible to reconstruct either a SV or a VS rigid order for this two-word stage, and that it is best to leave the order free at this point, as all sorts of pragmatic and semantic influences may have played a role in the choice, as you say. The advantage of postulating such a simple (and incredibly vague) two-word stage for syntax is exactly that proceeding to a transitive stage (SOV, SVO, etc.) would have provided a clear and significant communicative benefit. My argument in the book is that true displacement (including talking about strange and non-existent things) is facilitated by complex syntax. One-word (e.g. apple) and two-word (apple eat) expressions are bound to the context (here-and-now), and are typically interpreted in such a way that they make pragmatic sense. But with this limited syntax, it is very hard to express something like "The apple ate the boy." One needs complex syntax to express bizarre things like this; as well as novel and entertaining things. Needless to say, a transitive construction (which requires layered syntax) is also less vague, and allows faster processing of information. 

What contributes to the complexity of syntax is that once it acquires transitive structures, the intransitive (absolutive-like) structures may still remain in use, in e.g. compounds and noun phrases, as has been reported. Ergative languages often have "split-ergativity" where transitive nom-acc patterns are used with e.g. animates (or noun phrases higher on the animacy hierarchy), with which ambiguity is more likely to arise, but absolutive patterns are used with  inanimates. I have also argued that so-called "middle" constructions in e.g. Serbian are absolutive based, and in fact reflect some properties of this two-word proto-syntax, with incredible vagueness. (There are some nice minimally contrasting pairs of examples, one involving a middle, and another a transitive structure, whose processing can be tested by fMRI experiments). But just the very existence of such middle constructions across languages (straddling the boundary between what is transitive and intransitive, what is passive and active) provides support for a gradualist, step-by-step evolution of syntax. Also, the postulation of this intransitive two-word stage can explain the vast variation across languages in the expression of transitivity. 

Remy,

Again, I think that the question regarding whether syntax/grammar evolved gradually, and whether this is genetically determined, has to be a matter of empirical investigation. But one first needs to reconstruct syntax down to the stages that make evolutionary sense, and where each new stage brings with it some concrete and tangible communicative benefits. One needs to explore some specific and falsifiable hypotheses. 

Ljiljana,

Again, that was not my question. My question was: why do we need this assumption, not how to check it.

Remy,

I probably do not understand what you are asking, then. What I tried to say is that we do not need it as an assumption, but rather as a research question, something that should be investigated. Does that make sense?

Ljiljana,

It does, and my question was: why should it be investigated ? Are the social factors, as presented by Dor & Jablonka not sufficient to explain language development (and genetics redundant) once the biological capabilities solidly set at biological level ? 

Remy, IMO languages would also evolve without genetics, and I'd think perhaps inherent (non-genetic) selection is the major factor in how a language changes (the best (= shortest, easiest to pronounce, most striking etc.) expressions, grammatical rules, sound changes etc. are selected for), but no doubt people with the best language skills will also be selected for, e.g. it's to be expected that languages gradually become more precise, shorter, easier to understand etc. partly thanks to our better hearing & sound pronouncing skills.

Marc and all the others,

you wrote "perhaps inherent (non-genetic) selection is the major factor in how a language changes (the best (= shortest, easiest to pronounce, most striking etc.) expressions, grammatical rules, sound changes etc. are selected for)"

but how did all the complicated grammar systems arise? They don't make much sense. In German for example we are always forced to use one of three genera (masculine, feminine, neutral), even when we speak about a tea-spoon - absolutely senseless. Even the systems of tenses don't make much sense. To say "yesterday" is much easier and makes more sense.

Klaus,

Just a couple of observations I have on this topic. Yes, many languages, including Slavic, have a gender system, which is a subtype of classifier systems. What is important about these systems is not so much which category each noun is assigned to (the assignments are largely arbitrary), as you notice, but that these nouns can now more easily agree with a verb, or with an adjective, or with other categories in the sentence with which they have a syntactic relationship. This can reduce the amount of ambiguity, and in general, as is the case with the other grammatical devices, can contribute to the speed of processing.

But having said that, my second observation is that not every single little phenomenon in grammar should be subjected to evolutionary scrutiny. Just like we would not want to say: “I bet you cannot explain why I have a mole on my back, and therefore there was no natural selection or evolution,” we also do not want to say “I bet you cannot explain why “book” is neuter in German, and feminine in Serbian, and thus there was no evolution of grammar.” 

@Klaus,

The fundamental fact that governs language evolution is that each generation of babies must learn a language without any formal training of any kind.  That means that anything that babies find hard to learn, remember, or use will be lost or simplified.

That implies a strong tendency toward simplification and regularization.  But children also need to express themselves in a way that other people (especially their parents) can understand  That forces them to learn and use patterns that are familiar to other people.

But there are two kinds of evolution:  the genetic evolution over the past 6 million years from apes to Homo sapiens, and the cultural evolution that is still taking place today.  Linguists, psycholinguists, and social linguists have written many volumes on these issues.

"How did all the complicated grammar systems arise?" Very very gradually, in small steps, over a long period, see e.g. MA Nowak & DC Krakauer 1999 "The evolution of language" PNAS 96:8082-33. It began with 1 sound = 1 word, but soon, when there became too many words, it became difficult to discern all "words", and combinations of 2 sounds = 1 word became necessary. In about the same way, combinations of words became necessary later: 2-word sentences (e.g. see Ljiljana's work). Fixed word order followed (but could later became superfluous, e.g. in well-declined languages such as Russian & Latin). For the evolution of grammar, see e.g. SC Dik & K Hengeveld 1997 "The theory of functional grammar" M de Gruyter NY. Transitive verbs had to discern between actor & patient (man woman kill), e.g. Spanish (vs other Romance languages) began using the preposition "a" to discern between "matò el hombre(S)" vs "matò al hombre(O)". In SOV languages, adverbs became postpositions became agglutinated endings became declination. Etc. In a comparable way, we see the evolution of declination in Spanish futurum "matar has" to mararàs" etc. That "boek/Buch" is neuter in Dutch/German has its roots in PIE, but in Dutch (as in English & Afrikaans) the distinction between neuter & non-neuter words has became largely superfluous, and Dutch dialects can have "de(n) boek" instead of "het boek". What has become superfluous tends to disappear, e.g. "-er" is pronounced "-e" in English, and "-en" is pronounced "-e" in Dutch.

The biggests problems in language evolution IMO were the earlier phases of speech, but since we realise that early-Pleistocene Homo (for a shorter or longer period) dispersed intercontinentally (cf coastal sites from Algeria to Java 1.8 Ma) along African & Eurasian coasts (beach-coming, wading bipedally & diving for waterside & shallow aquatic foods), these problems are largely solved IMO. Our far ancestors already had some sort of gibbon-like songs (duet of musical loud & variable sounds), and when they began eating seafood, they didn't have to bite & chew, but often had to suck & swallow their foods, which explains e.g. how consonants could evolve: the ability to close the mouth (at first for swallowing soft & slippery foods) at the lips, teeth, palate & uvula (labial, dental, palatal & uvular consonants). When they began diving for cray- & shellfish, they acquired voluntary control of their airways (breath-hold diving), allowing to emit sound at free will (i.e. not only emotionally any more, as in gibbon-like song). Aquatic foods are rich in brain-specific nutrients (DHA, iodine) whiih explains why H.erectus could evolve larger brains (possibly necessary for attaching arbitrary meanings to sounds?) than apes-australopiths.

The further steps are not insuperable IMO (Nowak, Ljiljana, Dik etc.).

Marc,

I agree with your outline and the claim that the further steps are not insuperable -- given a few million years..  There was an earlier pointer to the YouTube of Halliday's lecture, which addresses the social and cultural issues that affect the ongoing evolution.   That same YouTube page has videos  by various linguists and anthropologists.  Among them is a talk by Halliday's colleague Christian Matthiessen on a semiotic view of evolution: https://www.youtube.com/watch?v=U15qHWJcfT4 .

Another link on that page points to a 2010 lecture by Noam Chomsky, who is rehashing his views on the "poverty of the stimulus":  https://www.youtube.com/watch?v=urrNTVxuCxs .  Most of the others disagree with him, but Chomsky still has a large fan club.

The following slides present my outline of the issues:  http://www.jfsowa.com/talks/goal2.pdf .

Ljiljana,

I think the development of many languages went into a direction which we can’t understand as improvement. Between 50 and 90 % of grammar structures of many languages are either not sufficient or not necessary to make syntactic relationship clear and the sense of verbal expressions unambiguous.

John, you wrote

„That means that anything that babies find hard to learn, remember, or use will be lost or simplified. That implies a strong tendency toward simplification and regularization.“

I think this is more fantasy than reality.

What about the babies of the !Xóõ people. Their language is as much developed as ours, but they have to learn 164 phonemes (while other languages show that 10 phonemes can be sufficient, too). And they do it.

„cultural evolution that is still taking place today. Linguists, psycholinguists, and social linguists have written many volumes on these issues.“

I think it needs much fantasy to write such books.

Marc, you wrote

„It began with 1 sound = 1 word, but soon, when there became too many words, it became difficult to discern all "words", and combinations of 2 sounds = 1 word became necessary. In about the same way, combinations of words became necessary later: 2-word sentences“

That’s an interesting idea – but not more. There is no proof at all. Science - in my eyes – is something else.

Klaus,

I think that all (or at least nearly all) the contributors to this list agree with you on the following point:  Every modern natural language has sufficient internal resources (syntactic and semantic) to express anything expressible in any other NL -- provided that its speakers borrow or adapt whatever vocabulary may be necessary to express topics or points of view that they had not previously encountered.

I'm sure that Ljiljana, Marc, Halliday, Mathiessen, and the authors of the many writings you dismissed as "fantasy" would also agree.  We consider that point so obvious that there is no need to defend it.

But it's also true that scientific and cultural developments during the past three millennia have enriched modern languages with vocabulary that goes far beyond anything that our ancestors ever dreamed of.  The way those languages have accommodated the new developments is an important issue that Halliday, Matthiessen, and others have explored.

Before dismissing their work, I would urge you to listen to their lectures or read their publications -- and take them seriously.  You might come to different conclusions, but it's important to consider the questions they asked and analyzed.

Klaus,

If I can just respond to the few comments addressed to me - I have a class to teach pretty soon. 

I think that we need to keep reminding ourselves that evolution is not always improvement, and what “improvement” means can be relative to a very specific situation. To take one example, suppose, for the sake of argument, that our human species in the next 20 or so generations becomes physically weaker and less athletic due to our modern sedentary life. If great physical fitness is no longer necessary for survival, such ability can be masked (not readily visible to selection factors), and not selected for, or at least less selected for than in the past. It can thus be gradually diminished. I think for most people this would not be considered as improvement, but it would be evolution.

That our languages are not perfect does not mean that they did not evolve. Our eyesight and hearing are not perfect either, and allow for so much “ambiguity” (if you want), but they did evolve. Evolution does not guarantee a perfect outcome, or a good outcome, for that matter. It just keeps tweaking and tinkering with us.

As for your claim that 50% to 90% of grammar structures are either not sufficient or not necessary, I would have to know what specifically you have in mind here, and how these percentages were arrived at. If this includes the discussion of gender or classifier systems, which assign all nouns into classes, then again if you look at each noun it may seem unnecessary (and bizarre) to assign it to an arbitrary gender class. But, as I pointed out above, this is just a part (or symptom) of a larger phenomenon, agreement, which is prevalent across languages. If this larger phenomenon, agreement, has a communicative benefit overall, in reducing ambiguity (e.g. since subjects agree with the verb, and objects typically do not, that can lead to a reduction in ambiguity), and increasing the speed of processing, then we cannot claim that gender systems are dysfunctional.

But at the same time, it is quite possible that some features of language (or biological phenomena in general) are, or become dysfunctional over time, but that by itself is not an argument against evolution.

Most importantly, when we discuss these issues, it is important to break them down into specific claims, rather than very general claims, which are typically unfalsifiable, such as Grammar is dysfunctional, or Syntax is optimal. I think the most important aspect of doing science (as opposed to “fantasy”) is to make specific and falsifiable claims (hypotheses), even if many of them will be proven wrong. Even when they are proven wrong, scientific progress is made. 

And, as John pointed out, it is important to keep an open mind. 

"It began with 1 sound = 1 word, but soon, when there became too many words, it became difficult to discern all "words", and combinations of 2 sounds = 1 word became necessary. In about the same way, combinations of words became necessary later: 2-word sentences“ That’s an interesting idea – but not more. There is no proof at all. Science - in my eyes – is something else.

Just a short answer, later more perhaps: Klaus, have you read Nowak & Krakauer 1999 "The evolution of language" PNAS 96:8082-33?

Marc, you asked me:

"Klaus, have you read Nowak & Krakauer 1999 "The evolution of language" PNAS 96:8082-33?"

No.

Ljiljana, you wrote:

"As for your claim that 50% to 90% of grammar structures are either not sufficient or not necessary, I would have to know what specifically you have in mind here, and how these percentages were arrived at."

That’s my own estimate. What I have in mind is – for example –  morphology which as a whole is dispensable, because there are languages, too, which have only syntax (isolating languages).

Ljiljana and Klaus,

Re grammar structures that are not sufficient or not necessary:  It's hard to find an exact number.  But it's true that all NL grammars are (a) redundant and (b) ambiguous.   For simple sentences, grammar alone is almost sufficient, and semantics and pragmatics alone are almost sufficient.   Both properties together, (a) and (b), are essential for language learning, fault tolerance, flexibility, and efficiency (i.e., high-speed communication in a noisy environment).

Re isolating languages:  No languages are completely isolating.  Chinese is usually considered isolating because native Chinese speakers who have never studied linguistics say so.  But Chinese linguists point out that there are well-defined words that consist of one to four syllables, and there are particles that have well-defined positions with respect to other  morphemes.

Furthermore, it's a spelling  convention whether you write Lebensversicherungsgesellschaftsangestellter or life insurance company employee -- German puts S between the components and English replaces S with a blank.

For high-speed transmission, most languages omit or contract syntactic features that are easy to infer.  In standard English, for example, 'did not' => 'didn't; 'will not' => 'won't'.  In Pittsburgh, the native dialect allows contractions such as 'why don't you' => 'whyncha'.

As for inflections, linguists hypothesize and find abundant evidence for their evolution as contractions.  In Japanese, for example, nouns are followed by distinct particles that mark case relations.   When Japanese is written in Latin letters, a spelling convention puts a blank between the noun and the particle.  Linguists hypothesize an early version of IndoEuropean with subject pronouns that followed verbs,  By contraction and sound changes, those particles became part of the so-called word.

Biologists who study evolution agree with Ljiljana that the process leads to many ad hoc, non-optimal solutions.  For example, the vertebrate eye has a blind spot in the middle of the visual field.  But the octopus eye, which is functionally very similar, does not have that blind spot.  Both vertebrates and molluscs evolved from bilateral worms that had simple eye-spots, but evolution led to different placements of the optical nerves.

John,

are you sure that Chinese linguists are not influenced by indo-european linguists?

Even if Chinese (Mandarin) or - better - Vietnamese (Annamese) are not fully isolating languages, I assume, that it is possible to construct a fully isolating language.

@ Remy

To answer your question: Given your linguistic premises are genetics necessary to explain language evolution ?

No. I don't think so. While connections can be made between genetics and language in some way or another, I still believe that language is fundamentally a cultural activity that is passed on socially from person to person, generation to generation -- not a biological one. Of course, biology, chemistry, philosophy, psychology, linguistics, etc., -- are all connected in some way or another. BUT chemistry is not necessary for explaining modern dance or culinary traditions. Biology can't explain Literature. And so on. Trying to "explain" language in genetic terms would also be extremely reductionist, and therefore not necessary. I think that we need to acknowledge how fundamentally different language and genetics are from each other in order to even begin talking about their "co-evolution".

I think the search for the origin of language is also unnecessary -- whether you talk about its origin in the brain or its origin in the "first speech that ever happened between two people". We shouldn't look for the "causes" of language. We should look at how language conditions our relationship to each other and the world, as a cultural phenomenon. By doing that, we can have a better understanding of the relationship between culture and biology: for example, mating rituals or birthing laws can affect reproduction patterns; discoveries made in biology will require us to modify language so we can talk about them; brain damage will impair speech, etc.

I believe that language is highly complex and always changing. No word, phrase or text is ever repeated twice. Any relative stability in language is a cultural convention, not a biological condition or requirement. Language also follows a different kind of time than genetics. The change from Middle to Modern English, for example, does not correspond in any way to genetic evolution. And why is it that almost anyone can learn any language on earth, regardless of their genetic make-up?

Again, I do think we can make connections between genetics and language, but it would be at the cost of reducing languages to something that they aren't and we would be ignoring their complexity and diversity, and therefore not explaining anything really.

Are we genetically predisposed to learning and speaking languages? Probably. BUT languages always have to be learned and practiced culturally. Languages always precede us: we are born into a world where languages are already spoken between us. It is passed down socially, not genetically.

That's why I think it's not necessary.  Nevertheless,  I am still fascinated by the relationship between genetics and language.

IMO you should, Klaus, Nowak's paper is short, not difficult (what can't always be said of his other (biological) publications), and easy to find (PNAS).

Dik on functional grammar is IMO a splendid book (2 tomes), and a pleasure to read, although it takes a lot of time.

John, thanks for the interesting links & your excellent comments.

I don't think we all here very much disagree. Of course, how languages today evolve is mostly on their own (cultural) rules, different from genetic evolution (which is slower), but OTOH there's still biological selection on our speaking & hearing equipment, which also influences how languages change. The distinction between nature & nurture is sometimes impossible to make, but in any case, languages are still becoming more & more efficient AFAICS.

I'm happy to see some convergence in this thread.

I'd also like to comment on Klaus's two questions:

Re Chinese linguists:  Most linguists have some understanding of the structure of multiple languages (although some have criticized some of Chomsky's students for doing little or no research on languages other than English).

Re possibility of purely isolating languages:  Many artificial languages for computer systems have little or no morphology.  But there's a question about how to represent numbers.  If you consider '376' to be a world-like symbol with an internal morphology, then the string of Chinese characters that represent that same value would have to be considered a single word-like symbol.

Re John's and Klaus' thread concerning isolating languages...

In fact, David Gil has argued in various publications that Riau Indonesian is such a language, almost purely isolating. At the same time, he reports massive vagueness in this language, such that something like "chicken eat" can mean almost anything that has to do with chicken and with eating, including, but not limited to: "The chicken is eating. Somebody is eating the chicken. The chicken that has been eaten." Below are two relevant publications of his. The latter is in the book which just came out, and which addresses the question of whether all languages must be equally complex or not:

Gil, David (2005). ‘Isolating-monocategorial-associational language,’ in

H. Cohen and C. Lefebvre (eds.), Handbook of Categorizaton in

Cognitive Science. Amsterdam: Elsevier, 347–379.

Gil, David (2014). ‘Sign languages, Creoles, and the development of

predication,’ in Frederick J. Newmeyer and Laurel B. Preston

(eds.), Measuring Linguistic Complexity. Oxford: Oxford

University Press.

Ljiljana,

Thanks for the references to Gil's work.  After a bit of googling, I found many of his writings scattered around the WWW.  I like the idea that we should inject "a little bit of Riau Indonesian into our theories of Universal Grammar."  Source: http://westernlinguistics.ca/afla/meetings/afla4/abstracts/gil_little_green_linguists_lana_in_the_riau_archipelago.pdf

But the qualifier 'almost' in front of 'isolating' is significant.  Gil's colleagues have related his observations to pidgins, creoles, and the sign languages learned by apes.  There is a continuum.  People who have little knowledge of each other's native language can communicate with almost no grammar. But when the subject matter becomes more complex, they need to invent some conventions to clarify the relationships.  When they start doing arithmetic, they need conventions for representing numbers.

As another example, two chess players can discuss a game very precisely, but only if they agree on certain conventions.  With one set of conventions, they say "I'll move my knight to the third rank on the King Bishop file."  Then they abbreviate that statement to N-KB3.  With different conventions, they label the 64 squares from a1 to h8 and say "I'll move the knight from g1 to f3."

I believe that's the way grammar develops.  People can get along with minimal conventions for simple situations.  But when they feel the need to express complex relationships, they establish conventions.  The commonalities and differences among languages result from commonalities and differences in the ways people think about the world, communicate with other people, and agree on useful conventions.

In general, I believe that Halliday and his colleagues are closer to heart of the matter than Chomsky and his UG fans.  Following is a review of one of their books: http://www.jfsowa.com/pubs/halliday.pdf .

I don't think that I disagree with you, John. And David Gil probably does not either. The main postulate of my 2015 OUP book is that this two-word (small clause, absolutive-like) proto-grammar is the foundation for all languages and constructions, the foundation upon which languages can develop complexities in various directions. And many of these grammatical devices do in fact lead to disambiguation and to more precision. My argument is also that one can arrive at this proto-stage not just impressionistically, but by following a precise reconstruction method, in this case based on Chomsky's (1995) Minimalist theory. 

Thanks again for the links, John. Yes, I think many of us agree that Chomsky should best be forgotten asap. I haven't read Halliday cs, but I guess their functional approach of how syntax evolved isn't really different from SC Dik 1997 "The theory of functional grammar" Mouton de Gruyter NY.

IMO we now have all the essential steps in how speech arose: music (gibbon-like duets), voluntary breathing (diving for littoral foods

Just to add  a footnote to this discussion...

In  my view, various approaches (functionalist, formalist) often provide really useful tools for studying different aspects of language, including language evolution. Instead of trying to shoot down this approach or another completely (which both sides like to do), I found it very useful to consider the tools of both types of approaches in trying to contribute to the answer to the question of how syntax evolved. 

If I understand correctly that seems to push syntax towards physics as opposed to the social behavior direction suggested by Dor & Jablonka.. 

While there isn't much doubt about the role of biological evolution in the emergence of linguistic capabilities, new questions arise when, as Dor & Jablonka suggest, a distinction is made between communication and symbolic representation, the former shared with primates, the latter specific to humans. Then, if natural evolution did mark the distinction, it should also be observable in linguistic constructs. And that could be the inflexion point where linguistics would get their autonomy from genetics.

Remy,

The distinction between symbols and communication by all species is part of the much broader field of semiotics.  Human languages are the most complex semiotic systems, but there's a continuum of  sign processes from bacteria to humans (and perhaps beyond).  For a discussion of zoosemiotics, see the article on the "Umwelt" by John Deely: http://manoftheword.files.wordpress.com/2013/01/umwelt-deely.pdf .

For a summary of Peirce's classification of sign types and their relationship to perception, language, and reasoning, see pages 3 to 9 of http://www.jfsowa.com/pubs/rolelog.pdf .

There's much more to be said about all these issues, but Peirce's categories help to clarify many complex issues.  For a summary of how they relate to the philosophies of Whitehead, Wittgenstein, and others, see http://www.jfsowa.com/pubs/signproc.htm .