Why Only Us: Language and Evolution

by Robert C. Berwick, Noam Chomsky

Noam Chomsky and Robert Berwick draw on recent developments in linguistic theory to offer an evolutionary account of language and our remarkable, species-specific ability to acquire it.“A loosely connected collection of four essays that will fascinate anyone interested in the extraordinary phenomenon of language.”—New York Review of BooksWe are born crying, but those cries signal the first stirring of language. Within a year or so, infants master the sound system of their language; a few years after that, they are engaging in conversations. This remarkable, species-specific ability to acquire any human language—“the language faculty”—raises important biological questions about language, including how it has evolved. This book by two distinguished scholars—a computer scientist and a linguist—addresses the enduring question of the evolution of language.Robert Berwick and Noam Chomsky explain that until recently the evolutionary question could not be properly posed, because we did not have a clear idea of how to define “language” and therefore what it was that had evolved. But since the Minimalist Program, developed by Chomsky and others, we know the key ingredients of language and can put together an account of the evolution of human language and what distinguishes us from all other animals.Berwick and Chomsky discuss the biolinguistic perspective on language, which views language as a particular object of the biological world; the computational efficiency of language as a system of thought and understanding; the tension between Darwin's idea of gradual change and our contemporary understanding about evolutionary change and language; and evidence from nonhuman animals, in particular vocal learning in songbirds.

Genres: Linguistics, Nonfiction, Science, Language, Philosophy, Biology, History

247 pages, Kindle Edition

Published January 15, 2016

Reviews

[William Adams · Rating 2 out of 5]

The core idea in his monograph is that humans understand language hierarchically, an ability Chomsky and co-author Robert Berwick call “Merge.” Two mental objects can be merged into one, and that new, compound object can be processed linguistically as if it were a single object.

Explained in a review in The Economist (March 26th, 2016), a cat wearing a “hat” can become a “cat in the hat,” a noun phrase that functions grammatically as a single mental object. It can be merged with “the” to become “the cat in the hat,” and with other elements to become “The cat in the hat on the mat.” Whole sentences can in this way become single mental objects, as in “The cat in the hat on the mat came back.” Such compound mental objects can then be merged with other sentences to produce complex systems of thought.

Thus, language and thought are joined by this single ability, called “Merge,” which allows humans to think hierarchically. The gene for this hierarchical ability allows advanced thought, which confers evolutionary advantage so is conserved over generations.

Leaving aside the presumptive genetics, for which there is little or no evidence, the hypothesis reduces to, “Humans think hierarchically because language is processed hierarchically.” But we must allow that the causal arrow could run the other direction. Perhaps language is (or can be) processed hierarchically because we innately think and perceive hierarchically (e.g., due to Gestalt formation), not the reverse.

Another fundamental problem with the Merge hypothesis is that “cat” and “hat” are not free-standing mental objects, nor are any linguistic terms. Linguists, and most philosophers, labor under the illusion that there are such billiard balls of experience, each with attached linguistic label.

Rather, I think that such mental objects are generated post-experience, from social interaction. . Language is derivative of social interaction (as Wittgenstein convincingly argued). Words never stand as self-existent mental objects with their own reality the way numbers supposedly do.

The so-called Merge gene is thus not necessary. Humans do not have complex thought because they process language hierarchically. We process language hierarchically because our social interactions are fundamentally holistic from the beginning. Only later, with education, do we learn to analyze our complex, emotional, and prelinguistic social relationships into elements like words and conceptualized mental objects.

It's a thought-provoking read but should be absorbed with plenty of critical thinking to get past reflex biological scientism.

[Tatjana Volkova · Rating 5 out of 5]

В оригинале книга называется гораздо точнее: "Why Only Us" - "Почему только мы". Всё дело в том, что язык присущ только человеку. Обезьяну невозможно научить языку: если вы слышали об успешных экспериментах с шимпанзе, то успешность эта ложная: обезьяна лишь перебирала комбинации, но сказать, что она именно изучила язык - категорически нельзя. Даже такую простую вещь, как смысл слова "яблоко", в тех экспериментах она не усвоила.

Когда обезьяна слышит речь, она слышит только шум. Когда же человеческий ребенок слышит речь, он выхватывает из неё слова, повинуясь какому-то внутреннему механизму. В голове маленького японца щёлкает рубильник: я родился в стране, где говорят на правоориентированном языке. То же происходит в голове маленького англичанина: щёлк, у нас тут левоориентированный язык. В нашей голове изначально заложены какие-то законы, алгоритмы, и суть их мы до сих пор не знаем, несмотря на весь прогресс нейрокартографии и экспериментов на мозге.

Книга сложная, в ней вам придется лавировать между генетикой, нейрофизиологией, лингвистикой, математикой и информатикой. Но в качестве награды вы получите такое Йозеф-Кнехтовское, касталийское чувство, которое у меня лично всегда возникает при чтении интервью и статей Ноама Хомского.

Пение птицы - это конечный автомат. Язык человека - это иерархическая синтаксическая структура. Любой человек может строить бесконечно длинные предложения при помощи порождающей грамматики. То есть, выходит, любому человеку изначально свойственно понятие рекурсии. Или вот ещё красивый пример: вопросительное слово проносится через всё предложение, как квантор в математике.

Язык - не средство коммуникации, а инструмент мышления. Есть мнение, что именно появлением языка мы и обязаны появлению абстракции, логики, умозаключений у человека. И появляется главный вопрос, вынесенный в заглавие книги - ПОЧЕМУ? Почему человеку присущ язык? Это загадка и для авторов книги. Они, в общем, и сами не знают ответа на свой вопрос, и честно это признают. Поэтому язык рассматривается как некоторый феномен, как некоторое странное и непостижимое явление природы. Только долгим, упорным изучением механизмов, алгоритмов, вычислительных систем мозга мы сможем когда-то приблизиться к решению этой загадки.

[David Bahry · Rating 3 out of 5]

A popular view of the evolution of the human natural language faculty, or “language instinct,” is that it is a complex adaptation to communicate hierarchical propositional structures over a serial channel, evolved by gradualistic natural selection for that function (Pinker and Bloom, 1990; Pinker, 1994). The authors of Why Only Us: Language and Evolution disagree with some or more of that evolutionary view, though as we’ll see there are tangles in interpreting what parts they actually disagree with and how strongly (Berwick and Chomsky, 2016).

The book, a collection of four essays, appears to be aimed at professional linguists, with occasional reference to technical details of such things as context-free grammar, multiple context-free grammar, and finite-state machines, with little hand-holding (cf. Piattelli-Palmarini, 2002). A lay reader with some nontechnical background—say, who has read Steven Pinker’s The Language Instinct (1994) and Richard Dawkins’ The Blind Watchmaker (1986)—will be able to follow some of the discussion but not all. If you’re fine with that, read it.

To summarize Berwick and Chomsky’s positions (their own summary is on pp. 110–111):

First: they distinguish the broad faculty of language from what they call the narrow faculty of language, seeing the broad faculty of language as comprising both the narrow faculty, and its interfaces with externalization mechanisms, e.g. speech including phonology, and with internalization, e.g. thought (cf. Hauser et al., 2002).

Second: they believe the narrow language faculty evolved as a tool of thought, not for communication (which I assume they consider to be either a spandrel [a side-effect of an adaptation; Gould and Lewontin, 1979] or an exaptation [a spandrel or adaptation that has since been refined by natural selection for a new function, but is still best not thought of as “existing for” that new function; Gould and Vrba, 1982]).

Third: they believe only humans have the narrow language faculty (I'm not sure whether this is a definition, or a positive claim), though songbirds for instance have convergently evolved aspects of externalization (cf. Berwick et al., 2011); and that the narrow faculty of language is human-universal, with the differences between e.g. Basque and Spanish only superficial.

Fourth: they subscribe to the “minimalist program” in linguistics, proposed by Chomsky in the 1990s (cf. Chomsky, 1995a; Boeckx, 2006). They view the narrow language faculty as comprising only recursion, via the operation Merge, plus a mental lexicon of word-like linguistic atoms on which Merge can operate. (Merge, if it exists, takes two objects and combines them without further modification into a hierarchically-structured object—e.g. turns {read} and {books} into {read, books}.) They also view puzzling aspects of sentences’ surface structure as due to principles of efficient computation, rather than more arbitrary reasons. A prominent example is deletion of the repeated word in wh-movement: John is eating an apple transforms into Guess what John is eating, not into Guess what John is eating what; Berwick and Chomsky argue that “[t]his property has always seemed paradoxical, a kind of imperfection of language. … But it falls within the [Strong Minimalist Thesis], automatically. … With all but one of the occurrences of what suppressed, the computational burden is greatly eased …” (pp. 72–74).

Fifth: they believe Merge is simple and its evolutionary emergence rapid (at least relative to evolutionary time scales). It is less clear to me how long they think it took the externalization interface to evolve, and to what degree they think its interface was already a language-ready pre-adaptation, vs. required extensive adaptation by natural selection. Near the end of the book, they give interesting, admittedly-speculative, neuroanatomic evidence that “slight rewiring” of the brain as a basis for the evolution of language may not be unrealistic—no longer only general speculation, but a hypothesis about a specific piece of neurological wiring: a connected loop between language-related areas in adults, which is incompletely myelinated in infants and not fully connected in other primates. Right or not, it illustrates a lesson of evolutionary developmental biology: some apparent macromutations “are only macromutations if we look, naïvely, at the finished product, the adult. If we look at the processes of embryological development they turn out to be micromutations, in the sense that only a small change in the embryonic instructions had a large apparent effect in the adult” (Dawkins, 1986). Without detailed knowledge of development and neurology, it might be cheap to just assume such simplifications are available, but with such knowledge, we can better judge which mutations really are easily “available” to simple mutations via existing developmental processes.*

Sixth: they believe that human natural language evolved only relatively recently—before the out-of-Africa migrations ~60,000 years ago, but since the appearance of anatomically modern humans in Africa about ~200,000 years ago, their most confident guess being ~80,000 years ago, the time of the first clear evidence of symbolic behaviour, at Blombos Cave in South Africa. The authors doubt that Neanderthals, who diverged from us a few hundred thousand years prior, had language. It is frustrating that they seem to waffle on how confident they are in this doubt: at times the supposed recent evolution of human language is treated as though a tentative hypothesis; at other times it is treated as a “fact” constraining the set of plausible hypotheses for how language evolved. Also frustrating is that they seem to waffle on how constraining a recent origin would be for evolutionary scenarios anyway. Sometimes they imply that a recent origin would mean language has to be simple and its evolution not gradualistic; but near the end they suggest that the time frame they posit (200,000-80,000=120,000 years) is long enough that even the most famous complex adaptation, human-style eyeballs, could comfortably evolve. That being so, it is unclear why they treat their recent-origin hypothesis as in tension with the language-as-complex-adaptation hypothesis, rather than treating these hypotheses as independent.

Readers interested in the more technical details of linguistic theory referred to in the book (I didn’t always understand them) can probably consult any recent textbook on syntax, but would do well to read one that highlights the minimalist program (e.g. Boeckx, 2006). Readers interested in some of the math behind evolutionary theory can start with Martin Nowak’s Evolutionary Dynamics (2006), which Berwick and Chomsky cite in addition to the denser (Gillespie, 2004) and (Rice, 2004); or perhaps Otto and Day’s A Biologist’s Guide to Mathematical Modeling in Ecology and Evolution (2007). A frustration in reading the book was the occasional aside on Chomsky’s gripes with other linguists and philosophers of language, uncited and little-elaborated, such as his war against referentialism; this seems to allude to views discussed more fully in e.g. (Chomsky, 1995b).

Given that Pinker (say) believes the evolution of the human faculty of language including word order (Man bites dog vs. Dog bites man) etc. was gradualistic, while Chomsky (say) believes the evolution of Merge was non-gradualistic: do they even disagree, or at least, is their disagreement even major, apart from what definitions and distinctions they find useful for research? I’m not sure.

Readers interested in the debate may find valuable a previous exchange between Pinker and colleagues defending the complex-adaptation-for-communication hypothesis, and Chomsky and colleagues defending their framing of the broad/narrow distinction plus the simple-recent-origin hypothesis for the narrow faculty (Hauser et al., 2002; Pinker and Jackendoff, 2005; Fitch et al., 2005; Jackendoff and Pinker, 2005). Other readers of the book will be interested in its bibliography, which is quite valuable aside from your position on that debate, including recent work on the neurology and genetics of language, and on linguistic analysis of birdsong. Finally, for any wanting to know the recent positions of Noam Chomsky, inventor of modern linguistics and whose positions are surely worth knowing, and wanting to see what biolinguistics research taking those positions seriously might look like, the book will be invaluable and may inspire just such future research.

The book is interesting, and will be valuable for biolinguistics researchers, whether or not they agree with its arguments and conclusions. Its greatest weakness is its framing, which seems to take overly-seriously (e.g. p. 3) Alfred Russell Wallace’s worry that our smarts are too smart and our language too rich to have been of use to our stone age ancestors, who lived simple lives, thereby presenting a puzzle to evolutionary research. To me that seems patronizing, and more importantly, false. Anybody who’s ever sat around the campfire with friends, rivals, and their crush knows that there was always plenty to talk about.

References

Barkow, J. H., Cosmides, L., and Tooby, J. (1992). The Adapted Mind: Evolutionary Psychology and the Generation of Culture. Oxford University Press.

Berwick, R. C. and Chomsky, N. (2016). Why Only Us: Language and Evolution. The MIT Press.

Berwick, R. C. et al. (2011). Songs to syntax: the linguistics of birdsong. Trends in Cognitive Sciences 15, pp. 113–121.

Boeckx, C. (2006). Linguistic Minimalism: Origins, Concepts, Methods, and Aims. Oxford University Press.

Carroll, S. B. (2005). Endless Forms Most Beautiful: The New Science of Evo Devo. W. W. Norton & Company.

Chomsky, N. (1995a). The Minimalist Program. The MIT Press.

Chomsky, N. (1995b). Language and nature. Mind 104, pp. 1–61.

Davies, J. A. (2014). Life Unfolding: How the Human Body Creates Itself. Oxford University Press.

Dawkins, R. (1986). The Blind Watchmaker. Longman.

Fitch, W. T., Hauser, M. D., and Chomsky, N. (2005). The evolution of the language faculty: Clarifications and implications. Cognition 97, pp. 179–210.

Gillespie, J. H. (2004). Population Genetics: A Concise Guide (2nd ed.). The Johns Hopkins University Press.

Gould, S. J. (1985). A short way to corn. In: The Flamingo’s Smile. W. W. Norton & Company.

Gould, S. J. and Lewontin, R. C. (1979). The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist programme. Proceedings of the Royal Society of London. Series B, Biological Sciences 205, pp. 581–598.

Gould, S. J. and Vrba, E. S. (1982). Exaptation—a missing term in the science of form. Paleobiology 8, pp. 4–15.

Hauser, M. D., Chomsky, N., and Fitch, W. T. (2002). The faculty of language: what is it, who has it, and how did it evolve? Science 298, pp. 1569–1579.

Jackendoff, R. and Pinker, S. (2005). The nature of the language faculty and its implications for the evolution of language (Reply to Fitch, Hauser, and Chomsky). Cognition 97, pp. 211–225.

Lewontin, R. C. (1998). The evolution of cognition: questions we will never answer. In: Scarborough, D. and Sternberg, S. (eds.) 1988. An Invitation to Cognitive Science. Vol. 4. Methods, Models, and Conceptual Issues. The MIT Press

Nowak, M. A. (2006). Evolutionary Dynamics: Exploring the Equations of Life. Harvard University Press.

Otto, S. P. and Day, T. (2007). A Biologist’s Guide to Mathematical Modeling in Ecology and Evolution. Princeton University Press.

Piattelli-Palmarini, M. (2002). Grammar: The barest essentials. Nature 416, p. 129.

Pinker, S. (1994). The Language Instinct. William Morrow and Company.

Pinker, S. and Bloom, P. (1990). Natural language and natural selection. Behavioral and Brain Sciences 13, pp. 707–784.

Pinker, S. and Jackendoff, R. (2005). The faculty of language: what’s special about it? Cognition 95, pp. 201–236.

Rice, S. H. (2004). Evolutionary Theory: Mathematical and Conceptual Foundations. Sinauer Associates.

*On developmental biology, see Jamie Davies’ (2014) Life Unfolding: How the Human Body Creates Itself. On evolutionary developmental biology, see Sean Carroll’s (2005) Endless Forms Most Beautiful. For a theoretical example of a seeming macromutation that (if the model is correct) is really a developmental micromutation, see Stephen Jay Gould’s summary of a model of the evolution of corn from teosinte (1985).

Chapter 1 of Why Only Us has a critical discussion of gradualist, micromutationist assumptions—common in evolutionary psychology including linguistics—that I think is interesting but incomplete. The popular view of evolution common to evo psych writers, including Pinker, Tooby and Cosmides, etc. (cf. Barkow et al., 1992), is basically the view depicted in e.g. Richard Dawkins’ (1986) The Blind Watchmaker, so let’s see what he has to say (pp. 230–236). “There are very good reasons for rejecting all such saltationist theories of evolution,” he writes, highlighting two: “The first of these points was put by the great statistician and biologist R. A. Fisher … The other general reason for not believing in true saltation is also a statistical one, and its force also depends quantitatively on how macro is the macromutation we are postulating. In this case it is concerned with the complexity involved in evolutionary changes …” (the first point is illustrated by Fisher’s thought experiment about the adjustments to a microscope’s fine-tuning which are more vs. less likely to result in improved focus, if the microscope is already pretty-well focused; the second point is illustrated by the extreme improbability that a tornado in a junkyard would by chance throw together an assembled Boeing 747). Although the discussion by Berwick and Chomsky of micro- vs. medium-sized mutational effects is interesting (pp. 32–37), they only discuss it in terms of the first of the two arguments, Fisher’s microscope model. It is however unclear to me how much Berwick and Chomsky actually disagree with the popular view: they agree that “‘hopeful monsters’ really do seem out of the question” (p. 33), while only defending a role for mutations of modest effect; this seems quite compatible with the opinions of mainstream evo psych. (Perhaps there is semantic ambiguity about how small “micro” is? After all, nobody thinks mutational effects are literally infinitesimal; there has only been finite time since life began, yet life evolved.)

A critique of the possibility of doing evolutionary psychology at all (given brains and behaviours don’t fossilize, etc.), cited by Berwick and Chomsky, is (Lewontin, 1998).

[Owlseyes]

https://www.theatlantic.com/science/a...

[Heather Browning · Rating 2 out of 5]

Most of this material came across as highly speculative, and the scientific backing was thin. An interesting theory - language for thought rather than communication - but would need a lot more detail and empirical backing to be convincing.

[Kateryna Bobrovnyk · Rating 4 out of 5]

Непідготовленим читачам з боку біології/генетики (як мені) буде складнувато, але все одно дуже цікаво.

[Petros · Rating 2 out of 5]

In this book, Berwick and Chomsky, explore what language is, how it works and how it may have arisen.


The authors make the following points:


1) The starting point for the authors is that language has three key parts: i) an “internal computational system” of human language syntax (consisting of word-like elements/concepts and the combinatorial operation “Merge”), ii) a sensorimotor system of externalization (communicating those thoughts to others, via speech or sign language), and iii) an interface to connect concepts to action (the “conceptional/intentional”, or “semantic/pragmatic”, interface).


2) Songbirds and modern humans have independently evolved similar vocal learning abilities (the ability to learn distinctive, ordered sounds and, perhaps, even abstract sound category formation —infant ‘phoneme tuning’), with similar gene mechanisms and similar/homologous brain vocal cortex motor regions and projections to brainstem vocal motor neurons (this includes the way the FOXP2 gene is expressed and regulates relevant brain region connectivity). This points to the sensorimotor system of externalization being similar between modern humans and songbirds.


3) The FOXP2 protein is identical in Sapiens, Neanderthals and Denisovans, but has two amino acid differences in compared to chimps: therefore, it may have been target for recent natural selection in humans. Based on observations (in the famous “KE family”) and on experimental research (insertion of the mutated FOXP2 gene into mice), defects in the FOXP2 gene result in verbal dyspraxia (the individual knows what they want to say, but their brain has difficulty coordinating the muscle movements necessary to say those words), but also in a more general motor dyspraxia (deficits in learning rapid/sequential fine-motor skills). Therefore, FOXP2 may be part of the externalization process, but not of the “internal computation system”.


4) Between humans and chimpanzees, 99% of their proteins (i.e. the available macromolecules involved in the biochemistry of the organism) are identical. Between modern humans and Neanderthals, there are just 87 gene differences (differences in the proteins our genes code for - that’s around 0.1% difference), but some 3000 regulatory region differences (the gene regulatory system includes non-coding/junk DNA and non-DNA components that regulate gene activity). A gene mutation acts on many different systems, and is more likely to have deleterious effects, whereas a regulatory region mutation can be more narrowly targeted (therefore a regulatory region change should be possible in much less time than a gene change). This means that regulatory changes, rather than changes to protein-coding genes, might be where the evolution action predominantly lies (there is a fairly conserved genome —some even pose a universal animal genome— but different ways to utilize it). This would be a case of exaptation (putting previously available biological machinery to different uses), allowing for quick changes consistent with a non-gradualist view of evolution.


5) Technological progress in archaic humans was not gradual: Mode 1 Oldowan tools are found 2.5 million ya (or earlier) and were then maintained unchanged for perhaps a million years until the innovation of Mode 2 Acheulean hand axes (which lasted another million or so years). It is only after the appearance of the first modern humans that we see the beginning of the first rapid changes in both tools and the appearance of unambiguously symbolic artifacts (the geometric engravings in Blombos cave date to 80,000ya). This pattern of stasis followed by innovative jumps points to non-gradualist evolutionary changes in cognition. Having said that, brain size increased fairly gradually, reaching its current level perhaps 100,000ya.


6) There is no evidence for symbolic thought in archaic human species (no art, and very long periods of no technological progress), but all modern humans have identical capacities for language acquisition. Therefore, the authors conclude, language must have arisen sometime after the evolution of anatomically modern humans (circa 200,000 years ago) and before Homo Sapiens’ exit from Africa (circa 60,000ya), possibly around 80,000ya.


7) The most elementary property of human language capacity is that it enables us to generate (construct and interpret) a “discreet infinity of hierarchically structured expressions”:
• discreet = made up of a specific number of discreet words
• infinity = we can create an infinite number of different sentences
• hierarchically structured = expressions enveloped within others, in a hierarchical structure (i.e. syntax)
This means that the “generative procedure” of language is based on items (word-like entities that constitute the “atoms of computation”) plus computational properties.


8) The simplest assumption is that the “generative procedure” of internal computation (i.e. syntax) emerged suddenly as the result of a minor mutation. That means the “generative procedure” should be very simple.


9) Syntax can be reduced to one simple procedure: merging two items (words) into one, which can then be used as a new item (a new “subphrase”, to be merged with other items). The authors call this operation “Merge”, and it can be repeated to form an “infinity of hierarchically structured expressions”.

So the inner, computational, syntax for all languages is “Merge”, but then each language has its own externalization syntax: some languages have a subject-verb-object and others have a subject-object-verb order, some languages have inflections others don’t, etc. The deeper layer is part of linguistic thought, the superficial level is part of linguistic externalization.


10) And now, here is an important argument: the inner form of syntax is optimized for computational efficiency (this is the “Strong Minimalist Thesis”), but the externalization syntax isn’t. The “internal computational system” efficiently generates expressions interpretable at the “semantic/pragmatic (conceptional/intentional) interface”, with linguistic diversity coming from the “externalization system” (the linguistic process that converts internal syntactic objects to the entities accessible to the speech/sign sensorimotor system appears varied and subject to accidental historical events). Therefore, the primary function of language is internal thought. The process of externalization is secondary.

Basically, the argument is this: internal linguistic computation and externalization are different procedures, all languages have an optimally simple deeper common structure but large superficial non-optimized variation, therefore the primary function of language is recursive thought. Communication is secondary. Language must have first evolved to serve as an “internal mental tool”.

On this point, I find Tomasello’s “shared intentionality” counter-argument much more appealing. The authors base a good part of their argument on the claim of optimal internal computational efficiency, but I didn't find this claim convincing. Here is how they argued for it:

According to the authors, in the sentence “Which one of his paintings did the gallery expect that every artist likes best”, “his” refers to “every artist’s own paintings”. This is because the mind actually perceives “[Which one of his paintings] did the gallery expect that every artist likes [which one of his paintings] best” but the second copy of the phrase “[which one of his paintings]” is omitted by the “copy property of Merge”. Furthermore, they claim that “If both copies where pronounced in such examples, perception would be much easier. [...] There is a good reason why only one of the copies is pronounced: to pronounce more of them would yield enormous computational complexity”. Then they assert “Though there is no time to go into it here, there are many other cases of competition between computational efficiency and efficiency of use (parsability, communication, and so on). In all known cases, the latter is sacrificed: language design keeps to computational efficiency.” I honestly can’t understand why that omission makes for greater computational complexity, and i think it is unfortunate they don’t provide further explanations (or some reference) because this is all the proof they give for their “Strong Minimalist Thesis”.

So, to be honest, I wasn't able to fully wrap my head around the authors’ argument for internal syntax being “optimized” (they do accept this idea is not fully validated), but, at this point, it seems to me they are building abstract assumptions on top of abstract assumptions.



Here are some of my thoughts:


A fundamental assumption of the authors’ approach is that language generates “a digitally infinite array of hierarchically structures expressions with systematic interpretations at interfaces with two other internal systems, the sensorimotor system (for externalization) and the the conceptual system (for inference, interpretation, planning, organization of action, and other elements of what is informally called “thought”).

I think this fundamental assumption can be called into question: perhaps these are not three distinct “systems”; there is fair evidence pointing to externalization being somehow separate (people who suffer left hemisphere strokes and lose the ability to speak and/or understand speech, can still function in the world and even understand prosody), but thought could well be inherently sensorimotor (the sensorimotor system of externalization —the speech muscles or sign-language gesturing— could well be a specialized/different system from the sensorimotor system that produces interactions with the world, which is what would be used in thought; also see the entire embodied cognition program). Besides, how is the “internal computational system” (with the “Merge” operation) different from thought (there is evidence for other animals —like primates or Caledonian birds— thinking in what/if steps)? As a matter of fact, closely sticking to the “computation” analogy (speaking of discrete/digitized operations) may in itself be limiting.

In order words, the idea of “thought” here doesn’t seem well-defined (I can’t see how it is grounded to reality/biology). I think this is made evident when the authors are faced with the question: what are the items/objects/“atoms” of this computation (what exactly are words/concepts… how are they formed)? The answer they give is the following: “...the “atoms” of computation, lexical items/concepts, appear to be uniquely human. Crucially, even the simplest words and concepts of human language and thought lack the relation to mind-independent entities that appears characteristic of animal communication.” (according to the authors, humans live in the phenomenal world of Kant, whereas animals directly experience the world as a set of mind-independent entities). Beyond that comment, they only state that “the atomic elements pose deep mysteries”. I think building a “computational theory” where the items of computation are completely unknown (and perhaps even unknowable) is not only an evident weakness to their thesis, but also renders their entire approach meaningless.

Another implication of the book’s thesis is that word order in a sentence is semantically inconsequential, as long as Merge works underneath: “Notice again that the optimal computational operation, Merge, imposes no order on the merged elements. It follows, then, that the mental operations involving language should be independent of order, which is a reflex of the sensorimotor system. We have to impose linear order on words when we speak: the sensorimotor system does not permit production in parallel, or production of structures. The sensorimotor system was substantially in place long before language emerged, and appears to have little to do with language.” This is in stark opposition to Lakoff and Johnson’s study of metaphors in language and thought (in “Metaphors We Live By”), where they argue that a different order of words will inevitably have a different (even if very slightly) meaning.

I think the authors’ point that phrases like “the desire to fly instinctively appeals to kids” are ambiguous (it can be interpreted as “[the desire to fly instinctively] [appeals to kids]”, and as “[the desire to fly] [instinctively appeals to kids]”) is indicative of the limitations of their approach: they see language as a computer script; this phrase is only ambiguous in written form, but language is spoken (writing is only a very recent invention) and there is certainly more to language than words (the meaning of this sentence wouldn’t be ambiguous if one heard it instead of read it). It seems to me this is quite a fundamental detail the authors miss when they subsequently claim: “In cases like the one we mentioned, the child has no evidence at all to show that the simple property of linear distance must be ignored in favor of the complex property of structural distance”. As a matter of fact, the child has a lot of additional evidence: prosody, body language and facial expressions, context, and so on.

As a matter of fact, it seems to me that externalization of language being secondary/ancillary to internal thought seems to point to a different conclusion: language is conducive to thought, allowing thought to become abstract (providing the scaffolding for abstract set formation, and allowing for metaphorical structuring of abstract concepts), but is not the basic “computation” of thought. If, as the authors note, there are cognitively deficient but gifted savants, that shouldn’t indicate that the thought (the “semantic/pragmatic interface”) part of language is primary while the externalization part is secondary, but rather that thought is not an inherent part of language in the first place. I think the reason the authors conclude that “externalization is secondary” is because they begun with the axiomatic commitment that language has three parts (externalization, computation and the thought-action interface).

I think Feldman-Barrett, in her “How Emotions are Made” book, describes a plausible idea for these “atoms” of thought: concepts (as categorized, statistically-correlated, sensory data), where words can work as scaffolding for these statistical relationships (probably with some pre-prepared —a priori— learning mechanism for word-object correlations). It seems to me the grounded cognition approach relates to reality/biology, while respecting the idea of the synthetic a priori (the “items/atoms of thought” are built through the interaction of experience and a priori cognitive mechanisms, and are indeed not “mind-independent entities”), and without having to pose any hard human-animal divides (although, the ability for symbolic abstraction makes for a really vast distance). But that would entail that human thought, while greatly enhanced by language, is not categorically different from the thought of other animals, and this may be against Chomsky’s intuitions —which, I think, could be the ultimate reason why he concludes the evolution of language has to have been non-gradual (“not long before 60,000ya, language may not have existed at all”: hard to blame him, the evolution of culture really does look like an explosion... but I think he may be confusing thought for collective intelligence). So, it doesn’t seems reasonable to me, to focus on a “Universal Grammar” as the crucial step towards human language, rather than the ability to use symbols to construct abstract concepts (which then allows for “set formation”, which is also what “Merge” does). It is precisely the fact that, following the authors’ approach, “human-specific concepts and the “atoms of computation” that Merge uses” remain shrouded in mystery, that makes their approach meaningless.


In short: it is clear to me that this approach fails at providing meaning. I think a main reason is that the authors are restricted by the use of computation as the basis of thought and by basically considering thought as a part of language (rather than language as a part of thought). Having said that, written/spoken/signed language is serial and every language contains a systematic way of organization (grammatical rules) that we learn in the process of language acquisition. The systematicity of grammar (of how serial language codes differences in meaning) is indeed interesting and would possibly require some sort of a computational approach (although, even there, I think associative approaches might also be viable), but that would then have to somehow be tied to/grounded on a theory of meaning; otherwise I don’t see how its validity could be rendered testable. It seems to me that, as it is, this computational approach, where the items/atoms of computations are accepted to be unknown (and, perhaps, even unknowable), just cannot possibly provide an explanation of how language relates to meaning. Therefore, this entire approach may be an interesting intellectual endeavour (a necessary and productive step in the scientific process) but I don’t see how it can be useful as a part of someone’s philosophical worldview... if anything, it reinforces an abstract computational metaphor for thought, which, I think, is restrictive to one’s thinking.


If you'd like to get an idea of Chomsky's most recent (and probably final) take on language, this is a good book to read. If you're interested to read something that can change/affect/improve/deepen/enlarge the way you think and the way you understand yourself and the world, I'd say you can skip this one.

[Othman · Rating 4 out of 5]

I think the book is a good summary of the generative accounts for the evolution of language. Berwick and Chomsky in this book defend the Discontinuity Hypothesis, which argues that the evolution of language was a catastrophic event, involving a tweak to prior systems of vocal communication and mental processing. The evidence provided in favor of this hypothesis is also used against the competing hypothesis (the Continuity Hypothesis), which argues that the complexity of language can only be explained if language has emerged over a very long period, under the guiding hand of natural selection. I highly recommend the book.

[Larry · Rating 2 out of 5]

Reading Chomsky when you have studied analytic phil of mind & lang is the most redpilling experience you will ever have. Here is a guy who is 100 years old, as famous as the Bible, who is going to go: So that's what you think is true about language? Well, 100% of it is dead wrong. Let's start again. Whatever you end up thinking about Chomsky (and I'm not a fan so far, and I'm going to try to explain why), it's pretty clear that there's not a whole lot of stuff out there remotely as challenging and inspiring as what he has to say. Having said that,

B&C stress in the first chapter the role of other factors besides natural selection in evolution. The second chapter is where they launch their attack on communicationism (see Balcarras 2020). But the arguments given seem sketchy, at least at first:
- Animal communication is different from human language in structure and function (63), though “Lewontin’s remarks… illustrate how difficult it can be to assign a unique function to an organ or trait” (63 too): apparently not in the case of animal communication? If it is so difficult to assign a function to a trait, how can we tell if animal communication is different from human language in its function?
- “Human language does not even fit within the standard typologies of animal communication” (id.): this is a typical philosopher’s argument! I made a typology of communication types, human language is not in it, so it can’t be a type of communication! LOL
- “there has always been an alternative tradition, expressed by Burling (1993, 25) among others, that humans may well possess a secondary communication system like those of other primates, namely a nonverbal system of gestures or even calls, but that this is not language, since, as Burling notes, ‘our surviving primate communication system remains sharply distinct from language.’” (63-4) But “our surviving primate communication system” = “this” = “a nonverbal system of gestures or even calls”. So in essence, the ‘argument’ is that our communication system is not language, because of how sharply distinct it is from language! THAT’S NOT AN ARGUMENT
The basic minimalist idea is that language evolved for thought, and, to explain how, we need to account for two factors: one is the emergence of the lexicon (the “atoms of computation”, 66, and “their origin is entirely obscure”, 90), and the other is the computational principles of combination (the “generative procedure”). Since we don’t really know how the lexicon appears, we have no reason to think that it doesn’t predate Merge, and that language evolved from communication (even if it didn’t evolve for communication).
The generative procedure itself is very simple: take two previously constructed objects (words) X and Y, and form an unordered (98, 102) set {X, Y}. That’s it. It’s called Merge, and it’s something recursion can be reduced to. When MGG says that VP - V NP, this is Merge spelled backwards (113), but with some differences: as already said, the merged set is unordered, and generation rules like PP - V NP are allowed.
Merge handles displacement in the following way. Let us recall that back in the 90s, generative grammar was struggling with whether binding happens before or after movement, with raising sentences suggesting ‘after’, as in
*Chrisi said [CP that himselfi was appealing]
Chrisi wants himselfi to be appealing
where binding crosses clause boundaries, and wh-movement suggesting ‘before’, as in
[Which pictures of himselfi] did Chrisi like?
Chrisi did like [which pictures of himselfi] (= D-structure, before movement) (Carnie 2007: 424-5)
And Chomsky (1993) suggested that movement was really a form of copying where you don’t pronounce the original copy, so that we end up with
[Which pictures of himselfi] did Chrisi like [Which pictures of himselfi]
And binding is no longer a problem, because all binding principles hold at LF. Now to fastforward to 2016, B&C say Merge can work in two ways: External Merge (EM), where X and Y are disjoint, and Internal Merge (IM), where one is part of the other. For the two sentences
John is eating an apple
Guess what John is eating
If we have Y = what, and X = John is eating what, then we have Y inside of X, and IM can add something as long as it comes from within the expression, yielding what John is eating what. In the next derivation step, we add something entirely new from the lexicon, guess. Then we have our Internally Merged X = what John is eating what, and Y = guess, and they are disjoint, so this time we have EM, yielding guess what John is eating what (paraphrased from B&C, p. 73). The two whats are necessary to articulate two semantic roles: one is the quantifier, the other, the direct object (‘There is a thing (what thing?) such that John is eating that thing’). But we don’t pronounce the two whats, B&C go on to point out, and “the property follows from elementary principles of computational efficiency” (73). Another example, given in ch. 3 (100), is
The boys expect to meet each other
The boys expect [I wonder who]displaced to meet each other
I wonder who the boys expect …who to meet each other
Where each other refers back to the linearly remote element who, not the local the boys, because it is the mentally local element (erased during externalization).
This approach maintains the myth of the richer inner structure (the mental sentence, “the internally generated expression”, 74) that would motivate setting unarticulated constituents apart from other forms of deletion (e.g. ellipsis, or, as in this case, displacement, insofar as it involves, as Chomsky speculates, a form of silent copying), but I want to do just the opposite: show that what is true of unarticulated constituents is true across the board. Indeed, B&C go on, in a passage that is absolutely key for me:
The suppression of all but one of the occurrences of the displaced element is computationally efficient, but imposes a significant burden on interpretation, hence on communication. The person hearing the sentence has to discover the position of the gap where the displaced element is to be interpreted…. There is, then, a conflict between computational efficiency an interpretive-communicative efficiency. Universally, languages resolve the conflict in favor of computational efficiency. These facts at once suggest that language evolved as an instrument of internal thought, with externalization a secondary process. (74)
We would then expect phonology, in contrast to syntax, to have rules largely dependent on the constraints of communication, roughly like what is outlined in Pinker (1994/2000: 177; see also B&C, 82). But as we are going to see in the rest of the chapter, the same is true of everything but syntax. Conclusion: language is syntax.
Let’s reconstruct the argument:
1. There is a mismatch between how articulated the utterances are compared to the internally generated expression” or thought
2. Interpretation or communication actually requires the reconstruction of that more fully articulated expression (from the outside in)
3. So, if language had evolved for communication, we would not expect such a burden to have been placed on interpreters, but language has evolved in just such a way, so it must have evolved for thought instead
We can attack virtually every thing in this argument, from the premises to the conditional/abductive reasoning at the end:
1. There is no evidence that the internal thought is more articulated than the overt expression
2. (a) There is no evidence that communication requires samethinking (as opposed to, say, joint intention, attention, action etc., all more likely as evolutionary goals), and (b) there is no evidence that understanding requires reconstruction of the syntax: there is convincing evidence pointing to the contrary in the case of ellipsis, but then Chomsky might want to say that ellipsis and displacement work differently; (c) still, his claim seems to apply to language in general, not to sentences with displacement in particular, i.e. it is a universal claim, therefore falsifiable by existential statements, such as the one about ellipsis
3. Even so, we would actually expect a language that has evolved for communication to find ways of making communication more efficient than what Chomsky makes it out to be (i.e. a painstaking process of syntactic reconstruction), in keeping with points (1) and (2a-c).
With regards to point (2b) however, Chomsky could say that what reconstruction (Merge) does is not interpretation per se, but a mere step in the interpretive process, providing something binding applies to. Forms like I wonder who the boys expect who to meet each other would twinkle in the mind of the parser, just so that he figures out that who is the antecedent of the boys, and then moving on. Even so, as pointed out in (2c), this wouldn’t work for other cases of alleged copying mechanism aside from displacement (e.g., ellipsis), where the reconstructed form plays no role at all whatsoever at any point during the interpretive process. And besides, the reduction of the ‘reconstruction’ (my terminology, not B&C’s) to a mere step on the way to interpretation doesn’t square with what B&C later write:
If both of the copies were pronounced in such examples as these [given above], perception would be much easier [sic]. (101)
I very much doubt anyone not severely autistic would have a better time parsing *I wonder who the boys expect who to meet each other than I wonder who the boys expect to meet each other. And this is just what is wrong with Chomsky: he imagines that what it would be easier to understand is what it would be easier for a computational engine to parse. This is only true if all we have to work with when interpreting sentences is our (innate) knowledge of grammar. But we have more. (Don’t start telling me that ‘perception’ in the passage cited above only means ‘perception of the binding relations’, because Chomsky goes on to generalize to communication, so unless he thinks that all there is to communication is figuring out the syntax, I don’t think ‘perception’ can be read in this restricted sense.) In fact, it is much more plausible to think that we came up with additional, non-syntactic (but pragmatic) heuristics to parse sentences, i.e. that we can understand underarticulated (including elliptical, displaced, etc.) utterances as such, because the point of underarticulation is not computational efficiency on the side of the speaker alone, but communicative efficiency on both sides at once. In some of the cases discussed by B&C, like (see also Chomsky 2018)
Birds that fly instinctively swim
Apart from AdvP extraction, as in
Instinctively, birds that fly swim
We can also use accentuation to disambiguate. But that’s phonology, that’s auditory, externalized, secondary stuff!

There is more: a further, unstated premise, is that we think in some language-like medium, but this is surely a very specific form of thought. The evolution of language, in the Chomskyan sense, can’t explain the evolution of thought in general, because of how very specific a kind of thought it is. So we would need an additional account of the evolution of this form of thinking from other, perhaps older, visual, and other, forms of thinking, which we may share with nonhuman animals to a greater extent than we share ‘language’ (which, as Chomsky is keen to repeat, we may not share with them at all).
We can also turn Chomsky’s argument against him: if language evolved for thought, shouldn’t we expect communication to be much more difficult than it is? Why aren’t we all completely autistic? Is it just a matter of coevolution of the core language faculty on the one hand, and all our semantic and pragmatic skills that had to compensate on the other, totally functionally separated, hand?

B&C go on to give what they present as further, independent arguments for the idea that externalization is secondary: one is its modality-independence. Indeed, there are sign languages with the same structure as spoken languages. Insofar as the structure itself is equated with ‘language’, this commonality supports the idea that language evolved for thought (74-5). But this doesn’t work: communication does not cease to be communication just because it moves from one modality to another, and besides, it is quite naïve to think communication even in spoken languages is not multimodal. So the fact that linguistic communication is multimodal can’t show that language didn’t evolve for communication, it can only show that it didn’t evolve for unimodal communication, but there is no such thing (or very little of it)! Maybe language evolved for communication, communication being a multimodal process, with very good evolutionary reasons to be so! This would be like saying that language didn’t evolve for communication, because there are different languages! Of course, it makes a lot of sense to think that thought evolved prior to communication (see Miyagawa et al 2025 for recent research on this), and therefore to language, but that is very different from saying that language did (too). The evidence from aphasia shows that there is thinking besides linguistic externalization, which everyone in his right mind would agree with, not that that thinking is language-like.
Totally speculative mode now, but: maybe what makes externalization difficult in some language disorders (not aphasia, but autism maybe) is not so much a problem at the level of what organizes the externalization process, but rather a problem with the fact that the thought is not linguistically structured in the head to begin with: it is rather, say, visually structured. And so the ‘externalization’ problem becomes a ‘translation’ problem: it is difficult to put a picture into words! I would have to read Louise Cummings and Temple Grandin to test this.
B&C then make complete fools of themselves afterwards by peddling 70s theory about how each species has “a finite number of articulatory production” (78), when we know, for instance, that apes (bonobos and gorillas) can combine symbols in creative ways (see Schoenemann 2022 for a recent study). It is also well-known that there are phenomena of cross-species communication.
B&C and Chomsky’s don’t differentiate the production from the comprehension side at all. They don’t need to: it’s all externalization, secondary, has nothing to do with the core language faculty. Language comprehension disorder only impairs pragmatic or semantic skills, however defined, those are on the side of externalization (we would never be able to think ironic thoughts, and Fodor seems to think exactly that). Externalization being secondary, we would have to consider both language production and comprehension as secondary, since externalization plays a role in both. Aphasia, and language disorders in general, are very ill-named: they don’t teach us anything about how language works!
Even more interestingly, there are suggestions in the literature that visual thinking might actually be required for language comprehension, to the degree that individuals with “weak gestalt imagery” have some kind of language comprehension (not production) disorder (Bell 1991).
Basically, the Chomskyan take can be boiled down to something quite inoffensive: that syntax is not for communication, that it is a universal ability, innate, and useful for thought. None of this is controversial, and it is compatible with the idea that syntax (not ‘language’) evolved from or on the basis of an already established system of (linguistic) communication, using its lexicon as input for Merge. What is controversial is to take that small part of language, and call it ‘language’. If we stop doing that, then it turns out Chomskyan linguistics is pretty much irrelevant to how we use language. And then, if it is only relative to syntax that human and nonhuman communication or languages differ, there is no reason not to look into the semantics and pragmatics of nonhuman animal communication for insights into that of human communication.
At the end of ch. 2, B&C, far from trying to be inoffensive and conservative, launch the wide-scale attack on referentialism:
The symbols of human language and thought are sharply different [from those of nonhuman animals]. Their use is not automatically keyed to emotional states, and they do not pick out mind-independent objects or events in the external world. For human language and thought, it seems, there is no reference relation in the sense of Frege, Peirce, Tarski, Quine, and modern philosophy of language and mind. What we understand to be a river, a person, a tree, water, and so on, consistently turns out to be a creation of what seventeenth-century investigators called the human ‘cognoscitive powers,’ which provide us with rich means to refer to the outside world from intricate perspectives. (85)
This is interesting, actually, not because of how controversial it is (basically, all philosophy of language is wrong), but because it suggests a way of explaining away context-sensitivity in all its stripes: there is context-sensitivity precisely because there is no reference, because the reference relation is a myth. (This seems to be Chomsky’s point in his (2000)). So it is interesting, but it is also crazy: there seems to be things like nonhuman-human interaction, in spite of the large gap in ‘cognoscitive’ whatever powers.

In ch. 4, B&C stress that the difference Merge makes is how it enables humans to build unbouded hierarchically structured expressions (132).

In the end, what seems to be happening is that what makes our ‘language’ faculty special is syntax/recursion (but see Chemla, Schlenker, and other people working on nonhuman syntax), but we don’t use that aspect of our language faculty all the time (in communication), and, thus, there is a fairly broad extent to which our capacity for communication is, after all, common to nonhuman animals. Now Chomsky is entitled to say that that’s not what he calls language, and fair enough, because it isn’t, but what follows? Nothing!

Bell N. (1991). Gestalt imagery: A critical factor in language comprehension. Ann Dyslexia. 41(1): 246-60. doi: 10.1007/BF02648089. PMID: 24233768.
Chomsky N. (1993). A minimalist program for linguistic theory. In: KL Hale, SJ Keyser (eds). The View from Building 20: Essays in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press, 1-52
Chomsky N. (2018). Two notions of modularity. In: RG de Almeida, LR Gleitman (eds). On Concepts, Modules, and Language. Oxford: OUP, 25-40
Miyagawa S, DeSalle R, Nóbrega VA, Nitschke R, Okumura M and Tattersall I. (2025). Linguistic capacity was present in the Homo sapiens population 135 thousand years ago. Front. Psychol. 16: 1503900. doi: 10.3389/fpsy

[PepeLagarto]

está inleible

[Ans Schapendonk · Rating 1 out of 5]

Neither Chomsky nor Berwick came up with this theory themselves!
In fact, they stole it! The actual author comes from a Dutch-speaking corner. When Robert Berwick in 2011 with J.J. Bolhuis wrote an article about the linguistics of birdsong (Trends in Cognitive Sciences) they had no idea about the 'universal sound helix' (soundhelix). This is a sound pattern that is the same for all languages, but not simultaneously, which means that there are countless different languages. JJ Bolhuis published in 2013 together with M.A.C. Huijbregts again a book about birdsong, but this time supplemented with the Design Principles of Natural Language. The authors were informed about the soundhelix, which, in addition to the well-known three rows P, T and K, also contains a fourth vertical row W, which automatically leads to a fourth horizontal row of the M. The author of this soundhelix formulated sound rules as adjectio, detractio , metathesis, permutatio and delivery. Sounds, syllables and words become longer at the back (adjectio), dissolve at the front (detractio) and rotate on their own axis (metathesis). This course is exactly the opposite of the well-known course that Jacob Grimm invented, which has a lot of hard consequences since now, out of Dutch not only German and English, but also French, Spanish and other languages are ‘helixing’. Sometimes you also have to read from back to front (permutatio, retrogarde) or repeat (permutatio, repeat). When word parts are losing the contact like in AFVAL > AFVALLIG > BEVALLIG LICHT LICHT OP the Dutch author speaks of delivery. She (!) also found that vowels helix alphabetically, Out of Dutch DOOR the English THROUGH is helixing with already a U. Grammar is changing from an noun into a adjective into verbs (oplichten). Now ‘oplichten’ means ‘luminescence’ but also ‘betrayer! These insights are not only new, but also completely surprising, because they appear to have a predictive value. You can use this technique in all languages, but it works best in Dutch. When Berwick, Bolhuis, Huijbregts and Everaert realized that they were at risk of being accused of plagiarism, they involved Noam Chomsky in their research, assuming that no one would want to accuse Chomsky of piracy. The article they published together calls: Structure, not strings: linguistics as part of the cognitive sciences. Trends in Cognitive Sciences. 2015; 19(12):729–743. https://doi.org/10. 1016/j.tics.2015.09.008 PMID: 26564247. In a (P)ROOF or a (P)ROVE of PIRACY, the word ROOF means in Dutch to steal (which means also IRON like in ISRA-EL), but in English it means the highest part of a house that looks like a triangle (corner), with three dots. Together with a quarter (tetra) with four dots, one can build a fivehouse (pentagram). In the form of a five star you get ten dots by taking two of them. So, it is all about ROOMS which is also a religion (roman catholic). But in Dutch ROOM is helixing in MELK which is helixing in English MILK which brings us to the MILKYWAY (a very big room). Room in Dutch means KAMER (ment is the brain). So in DEKAMERONE we see ONE which is spoken like WAN (a number), but in German ONE or OHNE means ZERO (a number) which are algoritmes! The semantic meaning of numbers deals with inference (interference). The name of the devil in Dutch is LUCIFER which is helixing in CIJFERS (numbers), not thought out by male humans but by female since MA-the-MA-tics mention the word MAMA. Our language is called MOTHERTONGUE. In the article of Rikker Dockum and Caitlin Green they write about ‘a big tent’ whith place for everybody who has a relation to linguistics. The Dutch author Huijbregts will like this article, afraid of being accused of piracy?
Religion of GODS LAM helixing in DALAI LAMA shows the adjectio, but in Spanish, a lama is called AL PACKA which is helixing out of Dutch since AL (s)PACKA means SPEEKSEL (salive). So, this devil's system of moving of clouds (salive is helixing out of devil!) has indeed a connection with mathematics. This is the reason why YINGXU WANG was involved to ‘figger’ out the structures of the sounds changing in words which deals with THERMAL CRACKING (crack) which deals with chemistry, physics and biology, but also with ASTROPHYSICS (kosmology, topology). It were the prehistoric MIDWIVES who did discover this system. Long before Chomsky females did make paintings on walls to show that they knew a FINGERSYSTEM with a connection to the stars, which are DOTS, which is helixing in DUTCH. But making a DESIGN of DOTS like in the starconstellations in Dutch is calling a FRIES (which is also a name of a language), but it deals with FREYA, i.c. VENUS with her pentagram course.
In my book The Universal soundhelix of the prehistoric midwives stolen by famous scientists one can read more about this amazing rediscovered pattern, which can show us the way how to manage problems like the climate that is warming up.
In The slings and arrows of comparative linguistics (2018) remember, this is not an ‘original draft’ of Johan J. Bolhuis, Marinus A.C. Huybregts, Robert C. Berwick and Martin B.H. Everaert. Ingo Plag and Jenny Audring are mentioning names as PHONAESTHEMES or SPLINTERS for sounds which are disappeared like the K in I KNOW which you do not hear, but write. This K deals with Dutch IK (I) which is helixing in a sound ending with J since the fourth row of W goes like W > H > J > NJ. For many of you it may sound silly, but the Dutch language (i.c. Flemish with a FLAME) is the oldest spoken language in the world, which has consequences for the country where the ‘first’ Jews did live. Doggersland means in Dutch HOND, but DOG also means GOD (metathesis). The MADOGS were the mother and daughters, a matriarchy, that was turning around in DOGMA, the patriarchy, which brought wars. Time is now changing in peace, but there is a long way to go. We think we do have chatGPT now, but how long we will have computers is the question. You better learn to use the soundhelix by yourselve. It is easy to learn (like reading and writing).

[Rhys · Rating 2 out of 5]

I thought this would be written more for public consumption. I found that the text warped from loose discussions around evolution to very specific and jargon filled discussions around brain functions and development. For whom was this book intended, I wonder ...?

[Nevzat]

Most books on linguistics are focused on primatology or theories of cognitive evolution without a deep structural understanding of language, and many are focused on linguistics without a detailed approach to evolution. This book bases its core arguments mostly on `computer sciency` linguistics and a detailed understanding of evolution. It's important to note the difference between human language (including sign language) and other forms of communication animals use.

The core idea they propose is the simplest form of recursive operation, which they call Merge (basis of universal grammar) is special to humans, and this led to the internal language (i-language), that have later turned into communicative language as we now it today. So as a summary, what this book suggests is that, language didn't start as a means of communication but as an `organiser of thought`.

Naming complex internal thought skills `i-language` (it's not a language) creates a confusion and even more when, I think, Merge is proposed as Chomsky's famous 'universal grammar' because when there is the word grammar involved, most people, even linguists tend to think of it not as a cognitive feature but more like the grammar of spoken languages today. This leads to a waste of energy in a field already very complex. (Tomasello's recent `universal grammar is dead` article is a good example: (https://medium.com/@dan.milway/dont-b...)

While the authors describe Merge and the hierarchical nature of language in very much detail, the examples they give on these topics are solely based on spoken language. As their main argument is the basis of language had nothing to do with communication, I think examples unrelated to spoken language would serve the argument better. It’s pretty hard to imagine how Merge would work in a world without words and the writers agree. It's an area to be thought upon.

If Merge is purely an internal process and had nothing to do with communication, why this operation is not also usable in other types of cognitive processes besides language -like tool making-. If it is used, why call it UG and make it only about language?

There is also the between-the-lines fact that, if you accept the theory of Merge as basis of human language and accept that it has evolved relatively rapidly in evolutionary terms, you also have to accept that there is no 'semi-merge' and therefore no proto-language. This is a bold claim and I think it needed to be put more strongly in the book.

But the advantage of this book's theory is that because the book starts its language history from mental operations (an earlier step in the history of human evolution) rather than theories about the usages and advantages of communication, it's 'one step behind' from most of the other popular linguistics books and this can help their arguments to be used as a based upon other language theories. For example, when Tomasello says language is based on gestural communication and I can't see why gestural communication could't be based upon Merge.

[Michael · Rating 4 out of 5]

Makes an important case for a narrow definition of what has evolved as language: Universal Grammar and the Merge syntactic operator, enabling internal language as a cognitive tool—importantly, not as a communication tool. This leaves to the side both externalization (whether speech, writing, or signing) and other general thought capacities. These other components of what we commonly call language must have evolved separately to UG, as they are at least partially shared with other animals. The argument relies heavily on comparisons to the Bengalese Finch, to delineate complex auditory externalization, and signing chimpanzees (Nim), to demonstrate the lack of Merge. The idea is that those other components could be tied together in language in a sort of evolutionary recombination of whatever was at hand.

One important work accomplished in this book is correcting many common and subtle misunderstandings of evolution, through the lens of other attempts at understanding how human language could evolve.

However, I would say it is most useful as a bibliography: if I need to know more about any tangentially related subjects, such as language externalization, computational grammar, evolution of different linguistic and cognitive capacities, theories of UG, etc., I would reach for this book, which at times seems like a survey of these disparate areas of ongoing research.

I have only one real criticism of this account. I felt it offered a handhold for each subject discussed, and crucially each aspect of the nature of language relevant to its central argument. Even when the research on a given aspect is far from being definitive about anything, the authors were brave enough to put forward some speculation, offering an avenue to read up on other researchers' attempts and even guide future research.

In contrast, there is one subject where they leave the reader completely in the dark: the hypothesized interface been internal language and other general thought capacities. On one level, this is understandable: of the many discussed, it's probably the biggest unknown in the whole book. It seems the authors carefully avoid saying anything about this other than asserting its logical necessity for the argument, and quoting a few poetic platitudes about how language enables that most divine and awesome human ability of human imagination, or whatever. I don't think it would have been out of place to include some discussion of the possible nature of that interface and what others have said about the cognitive side of language, no matter how fraught or vague the state of the art might be right now. Not only would it thereby offer a handhold for future reading or research, it would actually bolster the argument by demonstrating the authors had performed due diligence, so to speak, in all the aspects of language required by their argument.

As it is, this lack felt quite glaring, and left me wondering what theory of mind could fit with this hypothesis. It seems the authors could land anywhere from "I-language is the basis of all thought," to "I-language is sometimes invoked for special processing," and there isn't really any discussion of what that would mean. You can start to narrow down the possibilities pretty quickly, but the fact that the authors leave that mostly as an exercise is both concerning and annoying.

[Billie Pritchett · Rating 4 out of 5]

This book by Robert Berwick and Noam Chomsky, Why Only Us, tackles the question that why humans, of all living creatures, possess language and other animals don't. They are quick to differentiate that by 'language,' they don't mean communication. The claim is that while it is very possible that several higher organisms have communication systems, only human beings have language.

So what is language? It's this system in the brain that takes words and expressions and orders them. According to Berwick and Chomsky, humans, perhaps some 80,0000 years ago through a genetic mutation, evolved this ability to organise their thoughts. They call the internal system that emerged 'Merge.' It can take any two words or expressions and form them into ever-more complicated sets. For example, read and books to form read books. Then I and read books to form I read books. And so on.

If we think about communication as something like the sounds you make with your mouth or the way a songbird sings, that's something external to the mind/brain and which relies on linear order. But if we think about language as some system internal to the mind/brain, we quickly see that it doesn't rely on linear order but hierarchical order. So here's an example. Take the sentence The man who is tall is happy. Every young English-speaking child knows that if you want to make that into a question, you take the is closest to the adjective there: Is the man who is tall happy? Virtually no child makes the mistake and says Is the man who tall is happy? and (this is important) even if they've never heard the correct utterance. This has to do with phrase structure rules inside the brain, which don't respect the linear order. If language were just a matter of communication, it would make more sense to take the is closest to the beginning of the sentence. This doesn't occur because it isn't a function of how the phrase structures work in the brain.

Very very fun book, one of my favourites of the year. Occasionally technical, but in general not extremely.

[Bernard · Rating 4 out of 5]

A robust and interesting systematisation of Chomsky and Berwick's arguments for situationg transformational generative linguistics as a convincing bedrock for the study of language evolution. There is a wealth of sources contained in this number that have made me more intrigued in the field overall, often branching out into computer science and evolutionary science more broadly. The rating is indicative of the presentation and coherence of the arguments, even if Chomsky (and I know it is Chomsky) repeats himself quite often across the four, unevenly distributed (Chapter 4 makes up more than 1/3 of the book) chapters. At times he can also come across as quite brazen, especially in situating his conception and study of language within what he conceives of as a 'Cartesian science', which invites more skepticism and scrutiny from me than I would like, but also unifies his internalist account within a particular 'rationalist' tradition, for better or for worse. I would recommend this to people interested in the subject, although quite often the linguistic concepts presuppose some basic understanding of the framework; it's not clear what a 'XP' (here being NPs,VPs,IPs, CPs, etc.) is from reading this book alone, and whilst Chomsky emphasises the 'Merge' operation as the basic building block of language, some of the examples he provides explicitly reference concepts such as the Copy Theory of Movement, and a particular analysis of 'Reconstruction' for Principle C, which are by no means simple concepts within generative syntax. That being said, I'm grateful for Robert Berwick acting as a moderating voice, as the presentation is better than a lot of latter-day Chomskyan works due to his contribution.

[Kevin Gross · Rating 5 out of 5]

A fascinating book surveying the history of current theory and understanding of the origins and mechanisms of natural language, largely the neuroscience. The title refers to a core question pursued: why are humans the only creatures that possess language capability that meets the standard of what the authors call the Basic Property. (And the book does include interesting explanations of why primates and birds are not truly using a language.)

The book is not a primer and you're very much jumping into the deep end of the pool in picking it up. For the most part, Berwick and Chomsky lay out credible reasoning for their positions on a wide range of questions. However, two recurrent problems, to my thinking, dog their expositions. The first is a rigid insistence on syntax as the sole mechanism for parsing meaning from language. My belief is that syntax alone is not enough. If we take the phrase "the deep blue sea" as an example, syntax will never make clear whether "deep" binds to "blue" or to "sea." There is inevitably a place for context, history, and semantics (etc.) in assigning meaning to language.

My second issue is the authors' repeated dismissal of problematic issues with a phrase such as "we don't have the time to delve into that here." I didn't see a clock on the wall, and at 167 pages, they could easily add a few more to properly tie up these loose ends.

Despite these quibbles, a wonderful book, a great resource with many pointers to further explore the topics it covers.

[Jeremy Massaquoi · Rating 3 out of 5]

Pretty good book regarding the question of evolution in particular regard to Language.

Does language evolve? How did language start? And why did the act of language occur?

Pretty deep question that famed Chomsky and Berwick attempt to answer.

Without fully reveling all of their thoughts and conclusions, one assumption that they make is that evolution of language was rather one time thing (albeit they state a a decent amount of time) and since that one instances of "mutation" or evolution, there hasn't been any evolution of language since.

There is more to get into regarding this book, its assumptions, and its insightful conclusions; However, I would like to attack the book on the basis of its structure. The book is structured as it's for those who are not linguists or biologist. However, upon reading the book, one who's not acquainted (like myself) with such topics can become lost. This leads also into the fact that this book isn't exact a book that is targeted to scholars (although this is for them).

This in term lessens the enjoyability of the book, even if the book present a new and interesting way to look at evolution and language. Compare a book like Freaknomics (Economics for non-economics people) to this to explain the last sentence.

Overall, the content presented is pretty interesting, but the way its presented can dampen the experience. Should be a must read though for the even crazier propositions made in here, that when pondered on, makes sense!

[Jamshid · Rating 4 out of 5]

This book is was a concise and informative analysis of the evolution of language. It argues that the faculty of language is not a mere communication ability but an important tool for thinking and imagination, which separates humans from other animals (hence "only us"). It then gives a somewhat detailed reasoning that the evolution of language must have happened in a much slower pace than classic Darwinian process. The discussion includes evidence from biology, genetics, linguistics and even computer science fields. It's a great book for those who seeks to know more about the origins of language as a general human ability, and even more generally, whether we, as a species, are really too different from other animals.

The only reason that I didn't give full score to this book was that it somewhat had a heavy language (especially when they talked about genetics, where I have null background). For such a detailed analysis that the authors tried to make, a ~166-page volume is too short. I wish they had made the book little longer and the language simpler. It took around three months for me to finish this small book.

[Rashaad Sidique · Rating 5 out of 5]

The Language Capacity

Apparently unique to our species, the language capacity seems to have evolved relatively recently in evolutionary time. How such a significant change could have come about in such a short period contradicts the gradualist view of evolution, but the authors convincingly argue that large phenotypic changes may result from small genetic changes. The explanation that stochastic factors besides natural selection play a large part in evolutionary change is fascinating. I found the argument that our language capacity is qualitatively different from anything existing in other species convincing, namely our ability to use the Basic Property of Merge to recursively generate a digitally infinite array of possible sentences and the idea of language as primarily an instrument for thought rather than communication. Parts of the book get a bit technical, especially the discussions of specific genes that may be related to language, but overall it’s an engaging and fascinating book.

[Sam Bowman · Rating 3 out of 5]

This was—indirectly—a nice overview of the broad worldview surrounding modern Minimalist syntax. However, the actual discussion of the book's new, UG-oriented, take on the origin of language is repetitive and confusingly structured. The four chapters seem to have come from separate papers written for similar audiences, and there is no introduction or conclusion. Further, several of the choices about what to include are a bit strange, and it's often hard to reconstruct the full argument behind each chapter.

I can't say I fully buy the main claim. This academic review* puts it nicely:
"There is imbalance in this book between covering evolutionary biology vs. syntactic theory. B&C proceed as if minimalism is a given, well known, and hardly controversial, while Darwin’s natural selection is in serious trouble."

* https://muse.jhu.edu/article/641213/p...

[David M]

Chomsky and Berwick argue that language did not necessarily evolve due to natural selection but in fact was the result of some stochastic process. The problem is that this assigns randomness an even larger roles than it has in a more conventionally Darwinian account but randomness isn't really an explanation at all, just the lack of one.

As a young man Chomsky recognized that it wasn't possible to reconcile with behaviorism, Skinner's paradigm simply had to be destroyed; it's a bit disappointing then to see him try and seek a rapprochement with the equally untenable materialist paradigm of evolution in his old age.

[Santiago · Rating 4 out of 5]

Es el tipo de libro que no leería si no fuese porque uno de los autores es el gran Chomsky (soy fan de sus libros políticos y quiero introducirme a su otra faceta). A menudo es muy técnico y los autores no se detienen a explicar muchos conceptos. Además, la versión en español no parece estar muy bien traducida. Y sin embargo me parece una buena lectura porque el tema de por qué el lenguaje humano es único es fascinante, y los autores se toman en serio la tarea de encontrar respuestas.

[Cex · Rating 5 out of 5]

A short and sweet entertaining and comprehensive exploration into where linguistics is currently and what scholars have learned on the subject, a worthwhile read for anybody interested in the history and evolution of language.

[Mesut Bostancı · Rating 5 out of 5]

this book is a peyote trip's worth of mind-expanding ideas, discussing recent advances in evolutionary theory, the biological basis of human language, and one of the most sober explanations of what makes us uniquely human

[Dominic Neesam · Rating 2 out of 5]

Takes a while to get your head around the main concepts & streams in this academic book - Merge, hierarchical vs linear structures, FOXP2, etc. If you're looking for a gripping or fascinating read which offers layman's terms, this isn't for you!

[Дмитрий Давыдов · Rating 3 out of 5]

Читалось очень тяжело: книга скорее научная, а не научно-популярная, с множеством непонятных слов и ссылок на источники. Но все же было интересно узнать предпосылки возникновения языка и некоторые его аспекты. Плюс эта книга сломала мое представление о том, чем занимаются современные лингвисты

[Anu Migom Panging · Rating 5 out of 5]

A surprisingly lucid book for a counterintuitive thesis about language and biology. I could understand most of the literature without having any decent background in biology or genetics. I can see myself revisiting this book a few more times.

[David Snower · Rating 4 out of 5]

They’re almost certainly wrong…but such beautiful ideas. This book definitely could have been broken up/organized better, but a really amazing exploration of the foundations of language and its relation to complex thought.

[Anahit Hakobyan · Rating 4 out of 5]

An interesting approach to the evolution of human mind.