bohm2 said:
It's Chomsky's emphasis on just recursion alone that seems hard to swallow, I think. But I'm pretty sympathetic to Chomsky's argument against adaptionalism and his emphasis on physical constraints as argued here:
The Boeckx article is a good read. Nice on the philosophical history. Amusing in the conceit that the minimalist program may discover general cognitive primitives - making Chomskian linguistics again the heart of the cogsci revolution!
The point is that either/or approaches here are always going to distort the science. It is quite clear that adaptationalism can't be "everything". But then neither can rationalism. Instead, these are the two extremes that then allow a complexity to arise between them (the usual systems point of view).
So that is my problem with Chomsky - the extent to which he tries to make it all about rationalism (which leads to his programmatic rejection of E-language, mutterings about hopeful monsters, vague utterances about "other factors", etc). And equally it is why I would have a problem with SR behaviourism, or Dawkins/Dennett style neo-Darwinism. People are always splitting into opposing camps (representing the thesis and antithesis) and so failing to model the synthesis - the actual way extremes get mixed to form the resulting system.
So language evolution (and the saltatory mental change it created in H.sap, as Boeckx agrees) is going to be a mix of rationalist and empiricist factors - formal and material causes. This is just what systems are.
The question then is what are the forms that constrained language evolution? And have the Chomskyites identified the right targets for research?
Generally Chomsky's big thing has been hierarchical organisation (as the deep "cognitive" structure behind the "superficial" surface structure of the linear sentence).
But I keep pointing out that the hierarchy is not special to language. It is in fact the most general kind of organisational form. Hierarchical causality underpins the whole of reality in the systems view. And it certainly is the general form when it comes to brain architecture. So of course hierarchy is a big and foundational aspect of language, but it is not specific to language. It was not what was new when language first evolved!
Evo-devo suggests other targets for the significant change perhaps. At least, evo-devo as Boeckx understands it - the kind that focuses on self-organised criticality and edge of chaos type dynamical phenomenon. So maybe the essential rational form is something like Kauffman's autocatalytic sets?
Well yes, this kind of "Turing constraint" is important to the evo-devo approach. Creative self-organisation and dissipative structure are exactly the kind of potentials that life harnesses. This level of chemico-physical form is the raw material of biotic systems. But Boeckx is completely missing the more important realisation of theoretical biology - the still higher laws of form described by semiotics, modelling relations theory and the epistemic cut. All that Santa Fe brand complexity stuff, and Prigogine dissipative structure stuff, is a theory of nativist form for the physico-chemical realm of nature. Life/mind depends on a still more highly specified level of form.
The key, as I've said plenty of times, is the ability to construct constraints. Life/mind has semiotic mechanisms like genes and words (and in a lesser way, membranes, pores, axons, spikes) that can harness the Santa Fe/Prigogine/Turing style dynamical emergence. So if we are talking about embracing rational laws of form as a rightful part of science here, then that is what semiotics is about. It identifies the thing that is actually new and so marks a sharp evo-devo transition between the non-living and the living, the mindless and the mindful.
Semiotics is not the only candidate for the forms that define bios of course. The Bayesian paradigm has also emerged as something that seems very central to understanding life, and more especially mind. It captures the anticipatory aspect of any semiosis, and also roots the semiosis in something concretely measurable - information/entropy.
So there is a natural convergence going on there at the moment. Anticipation and constraint go together because a reduction in surprisal is dichotomously also a maximisation of behavioural variety. That is, the better you get at predicting the world, the more tightly defined become its surprises, and so in turn the more precisely you can learn the something new it takes to refine your future predictions.
Anyway, this illustrates - Boeckx's point - that none of us are naked adaptationalists here. It is basic to the systems' point of view that we must also identify the laws of form. But the system's POV also stresses that the laws of form are themselves a hierarchy. There are some very general laws (like hierarchical organisation), and then the more specified ones that identify the essential transitions, such as the kind that produce phase transitions in condensed matter physics, or separate life from non-life.
So what of language evolution? What, if anything, are the essential novelties that led to that saltatory change? What are its specific rational forms?
Recursion is not special to speech. But a strong separation between what Chomsky calls E- and I-language is. So it is in fact the relationship between the two that should be the focus of attention. It is the semiotic interaction that arises between the hierarchical (holistic) organisation and the serial (digitally-constained) expression that is the key. This is the justification for vocal tract/phonology arguments about language evolution. The first thing that had to happen was exactly this kind of serial constraint on hierarchical output.
But there is more. The step from phonology to semantics is the tricky one. And it contains the further question of whether this step was achieved mainly by genetics or by cultural learning. Although in fact - from a "laws of form" perspective - this does not really matter so much. The answer is that it is going to be a bit of both most likely. And either way, it is the organising form that matters, not the material, the medium, in which the change became encoded - biological or cultural evolution. If you are indeed embracing rationalism, then the nature vs nurture dichotomy has even less force because the action lies now somewhere more "Platonic".
Now there are two aspects to the semantics of speech - the way words, mere noises, encode meanings. And in good systems fashion, one is local, one global. Or one lexical, one syntactical.
So you have the lexical. Words, through associative learning, come to stand for particular meanings - they are symbols that reliably constrain our state of thought (or more properly, given a Bayesian perspective, constrain our state of expectation). A word like "cat" or "blue" exerts a top-down boundedness on our current thinking. And then words can be freely combined to construct even more constrained states of thought - as of course the mimimalist program recognises with its focus on the operation of merge.
So part of UG is this "universal lexicon", this ability to construct bottom-up the states of mental constraint, Bayesian expectation, that normally, in animals, only exists as top-down hierarchical organisation.
That is a genuine evolutionary novelty. It is not completely novel because, as said, genes also do the semiotic trick of constructing constraints. But it does seem a purer, less restricted, form of semiosis, or symbolism. And it may have its own completely novel aspects. Well, that is the kind of research question that can be posed once it is accepted that we are indeed exploring the laws of form, and their hierarchical complexification.
But then there is the global aspect of semantics - which the minimalist progam would appear to hope to cover with the operation of move. And this is the way that there is organising meaning at the level of sentences. Or as Chomskyites would prefer to see it, I-language.
To cut a long story short, my view is that the saltatory step here was the evolution of subject-verb-object sentence structure. Because what this did was encode the notion of efficient cause. It created the crisp mental habit of reducing reality to statements of cause and effect logic - tales of who did what to whom.
The animal mind is holistic in modelling reality. It does of course pick out efficient cause - that is what SR Behaviourism was all about. But it only does this in a contextual or situational fashion. So efficient cause remains entangled with an accompanying set of material, formal and final causes so far as the animal is concerned. A symbol like the ringing of a bell only has a meaning in a particular context (which includes final causes like whether you happen to be hungry or not, as well as other aspects of context like a history of reward in a similar experimental set-up).
So - again as Boeckx recognises - the new trick was to be able to break out of the animal mode of thought by being able to construct generalisations. Generalised models of efficient causality.
The question then arises whether this Rubicon step was a matter of biological or cultural evolution? The key thing is that this is a rational principle (what could be more rational than going back to Aristotle's foundational analysis of causality?). So what matters here is to identify it as the general crucial novelty in the human story. But then the question of whether the step was genetic or memetic is a valid subsidiary research question of interest to the paleolinguist.
So does the brain seem innately wired for subject~object distinctions? Or is the formal idea of efficient cause - the Newtonian idea that there is always a pusher and a pushee - just extremely learnable because the brain was preadapted to making long-range phonal connections?
I think the current evidence suggests that the brain is genetically more general purpose, and the habit of forming subject-verb-object structure sentences was a pretty sudden cultural invention. It crystalised a way of viewing and remembering the world that was so powerful that the small group of H.saps who developed it, took off and never looked back. Though as a meme, no reason why it would not have spread through social contact and migration.
So to sum up, rationalism is as essential to the full evolutionary view taken by systems science and theoretical biology as empiricism. The holistic view of causes demands that we seek the laws of form - the universal constraints of reality - as much as its fundamental materials, the stuff out of which complexly ordered realities get constructed.
There is then going to be a hierarchy of the laws of form. There is an emergent story in which simplicity develops into complexity. And novelties in form, in the nature of constraints, will mark the major observed transitions.
As we agree, the animal/human transition is a major one. So we should expect novel constraints to be one half of the explanation (the other half being the material/effective causes beloved of neo-Darwinism).
Chomsky certainly puts forward candidates for these novel forms. First UG, then principles and parameters, now the minimalist program. But Chomsky has always been hamstrung by his rationalist prejudices. He wants it to be the whole story (which become Platonism). And it has caused him to miss the aspects of formal organisation which are in fact the critical ones when it comes to life/mind. ie: the semiosis, the serial constraint, the epistemic cut, the Bayesian prediction.
But from an evolutionary perspective, we would expect to see a phonology-first emergence of a formal organisation (the serial constraint on hierarchical output) because otherwise how else could the necessary biological pre-adaptations be explained? The evolution of the vocal tract demands a theory, it can't be treated as a hopeful monster, and even a spandrel is pretty limp (the whole spandrel concept is weak, like all Gould's proto-evo-devo work really - again Boeckx is on the mark there).
Then the jump from phonal machinery to semantic/semiotic mechanism is a short one, in terms of further material change at least. But quickly revolutionary in terms of formal change of course. Suddenly H.sap had the machinery to construct words and sentences. To locally constrain states of thought to exact meanings, and to globally organise states of thought so that they articulated statements of efficient cause.
Suddenly you had a rational being (oh the irony!). Humans with the mental habit of viewing absolutely everything through the universal prism of cause and effect logic. And likewise, finding it hard not to reduce everything to just this one notion of cause.
The reason that there are not many systems scientists about is that the very tool of human thought - our SOV-based language - works against thinking in any other more holistic fashion.
And even the rationalists - those claiming to be investigating the laws of form - are still thinking in terms of efficient cause. Chomskyian UG is typical in talking about how hierarchical structure gets constructed from the bottom-up (when the brain itself is decomposing vague intentions into crisp hierarchical states of organisation, working from the top-down - again, the Bayesian view where global expectations constrain the information processing).
The very terms that Chomsky chooses - like I-language - betrays this basic misorientation. He sees the deep structure, ie: the global constraints of hierarchical form - as "inside". But to be more global - to be contextual and situational - the I-language has to be in fact "outside", larger in scale. It is a small but significant terminological confusion. It already sets a field on the wrong path.
Boeckx also gets things back to front with his hopeful claim that the minimalist program will prove to be foundational to the evo-devo rationalist project generally. The paleolinguistic question is certainly right at the heart of things - it is where material complexity, in the form of the human mind, is indeed the most complex. And a specific theory of the formal novelties is going to be required.
But it is hardly then foundational. Quite the reverse. Systems science, condensed matter physics, dissipative structure theory and other modelling discourses are going to provide the more general laws of form here. Then paleolinguistics has to pick up the story of the further specific laws that emerged at the crucial saltatory transition of grammatically-structured speech.