Sunday, August 17, 2008

A Comparative Analysis of Generative and Cognitive Grammar in light of Popper’s meta-theory

This proposal aims to compare two opposing accounts of how language and the brain interact; these go by the names of generative linguistics and cognitive linguistics. I will outline the basic assumptions common to most mainstream adherents of these fields, together with the conclusions they are obliged to accept in light of their respective theoretical commitments. After assessing various aspects of these theories, they will be compared using the standard criteria of what ought to constitute a scientific theory, and thereby conclude that the cognitive approach seems to be more empirically and logically plausible for various reasons.

Generative Linguistics
The school of linguistics known as Generative Linguistics (henceforth GL) was started by numerous collaboraters, many of whom have since either abandoned the enterprise as initially outlined, or gone on to expand the theory in novel ways. Those who have abandoned the ship include thinkers like George Lakoff and Haj Ross, whereas those who expanded the theory in ways not originally envisioned include Ray Jackendoff, Steven Pinker and Jerry Fodor.

The central figure since the inception of the GL enterprise in the late 1950’s has been Noam Chomsky, and his position regarding the nature of language, how it ought to be studied, and its relation to human nature, has remained fundamentally unchanged over the years.

The main arguments within the GL framework have come from Noam Chomsky and Steven Pinker, and therefore their work will be drawn upon more heavily than others in outlining the said theory. Others, who have accepted the core assumptions and methodologies, will be alluded to as well, but will not constitute the core of this thesis.

The key assumptions that pervade this approach are as follows:

Language is innate, in the sense that it is part of an autonomous module (a mental organ) in the mind that grows in the child the same way your limbs grow.
Language learning or acquisition are therefore misnomers; language growth is a more appropriate term.
The physical manifestation of language (our linguistic performance) is only to be studied insofar as it provides clues to the language in our minds (our linguistic competence). We should therefore assume that a speech community is homogenous, and language should be studied as if it were part of such an idealized community.
For this reason, we are to draw a distinction between surface structure and deep structure linguistic forms.
Language is a discrete combinatorial system, and this fact allows us to use language in novel ways, which they refer to as creativity.
Language is at the core a syntactic phenomenon, which employs various finite rules which enable the user to produce a potentially infinite number of utterances. This process works like an abstract algorithm, and therefore functions independent of context and meaning. Hence, semantics should not only be seen as secondary, but as epiphenomenal.
As a genetically based module in the mind, language must have evolved at a single point in our history, either by Darwinian natural selection (Pinker) or by some kind of small genetic mutation, which somehow interacted with the mind’s computational principles (Chomsky), which gave rise to language.
Language has no direct influence on our thoughts, as thought and language are two separate processes which function separately in the mind.
Since language is part of our biological endowment, there must be a Universal Grammar (UG) present in all languages, which manifests as a result of being part of our minds at birth, providing the scaffolding that enables language to grow in the mind.
Just as our physical organs interact on some level with other organs, they are nonetheless structurally and functionally independent of them. Language, being a mental organ, should also be expected to be understood independently; language should not be intertwined with other aspects of our mental life, any more than the heart is intertwined with the functioning of the liver.
Language is understood in light of Rationalism, whereby thought is believed to be conscious and literal.
Generative Linguists are vehemently opposed to Empiricism and behaviourism.
The human mind is a uniform, modular organ.

These principles are certainly not exhaustive, but I have selected those which I seem to be the most definitive.

There are certain things that follow from this approach to the study of language and mind. Generally, accepting these premises obliges one to accept the conclusions which follow. Regarding the study of language, this includes:
Having to concede that second language learning is not at all related to the growth (sic) of your first language.
Language teaching is not related to the primary goal of linguistics.
Socio-linguistics, prescriptive grammar, speech therapy, etc. are not of primary concern to the linguist.
Reading and writing not part of language; they are cultural artefacts just like painting and dancing.
Documenting corpora is a pointless exercise, since the range of potential utterances are infinite. The goal should be to try and discover what the universal, finite rules are which all languages employ. The goal of linguistics as a whole has to be redefined, and those who pursue corpus linguistics are wasting their time.
Language is not for communication. Language appeared in the mind somehow, and we discovered that it could also be used for communication. Where it came from, and the real reason it manifested, are unknown.
Language is to be seen as part of the physical world, and needs to be studied as such.

In addition, GL uses some words in deviant ways. For example:
- creative refers to the simple act of speaking
- language is no longer seen as something spoken by a community; it is more accurately seen as an underlying mental competence, which grows in the mind and is the same in every human being. Recently Chomsky has taken to distinguishing between i-language, e-language, and Language, which adds to the confusion
- mind is merely a synonym for brain, but seen more as aspects of the brain which we know must be there
- the use of the word grow to refer to the development of a first language in the child; if the child is exposed to three languages as a young child, then three languages will “grow” in him. Generative linguists do not see this as odd, and they would be quite comfortable with an analogy along the following lines: if a child grows up gathering rocks and hunting, his hands will be stronger and more versatile; likewise, if a child grows up hearing different languages, his innate language faculty grows to be more versatile.

Also, it is not a coincidence that people like Pinker tend more towards a conservative outlook, whereas people like Lakoff tend more towards a progressive outlook. The idea that we are all the same at the core, that our linguistic and cultural disparities are just superficial, that our minds come fully equipped with genetic information, etc. means we do not need a government to see to our individual needs, since we are capable of using our innate potential to make a success of our lives. It is this mind-set that justifies the right-wing extremist view that the world should be run under one government, and any country or nation which questions it simply does not realize that we are all the same. This is the reason why democracy needs to be superimposed on Middle Eastern countries, whether they like it or not. Saudi Arabians, who are perfectly happy with their king, simply do not realize that American democracy is the best way of running a country.
Why democracy? Or why America’s brand of democracy? Well that is just a coincidence arising from the fact that they happen to be the first country to realize this. Countries like Dubai and Saudi Arabia just do not know what is good for them. One may wonder why, if “self-evident” civil liberties are innate, do they not know this; the answer: if a child is raised with his hands tied behind his back, and that is all he knows, he will not realize that they were meant to be free until some benevolent person (or institution) unties his hands, and forces the society to ban such practices. One of the conservative arguments justifying the war in Iraq goes along these lines.

This mind-set is the very same one which insists that some of the rules which pertain to American English must be part and parcel of human nature. Not only should this apply too every other dialect of English, and to the some six thousand other languages in the world, but every human being is born with this UG. If it does not appear that way, it is only because that principle must not have a surface manifestation – or maybe that principle is actually a parameter. In fact, at one point, if you did not accept this, you could hardly expect to find a job or get a research grant in the USA.

I am certainly not insinuating that the conservatives in Washington read The Blank Slate, got over excited, and invaded Iraq in an attempt to spread democracy and freedom in the Middle East, knowing now that the notion of universal liberty can be justified. I am merely drawing a parallel between these ways of thinking, and the methodologies they employ in reasoning. Just as you get different forms of democracy in the world, the principles underlying a democratic government are the same, just as with language.

Pinker refutes this by saying that political equity simply means being treated equally, not being literally equal in the biological sense. In fact, Pinker does not believe that we are all equal. In addition to his conviction that men and women have different skills and abilities (though he is careful not to hierarchialise this), Pinker also believes that there could be a racial correlate regarding our skills and abilities. During an interview with John Brockman, he said that science may one day discover that certain race groups are better at certain things than other things, which may be something we are not ready for (interview available at The interview was based on the theme Dangerous Ideas, and Pinker based his statement on a book by Daniel Dennett called Darwin’s Dangerous Idea. After pointing out how Darwin was ostracized, and how the theory shook the very foundations of the church, he says that the next shock to us will be the said discovery. He was very careful not to spell out the details, but I doubt he was implying that we may discover that Black Africans are more intellectual, and that White American males are better at manual labour.

As an aside, it is a well-known fact that the HSRC used to publish very convincing statistics during the apartheid days ‘proving’, for example, that Black people are biologically designed for intellectually inferior tasks.

The various forms of evidence cited to support these views will be discussed and criticized scientifically. I contend that the ideas on which GL is based are methodologically and empirically flawed. The same criticism applies to other paradigms which utilize the same lines of reasoning, like the conservative politicians.

These upshots and underlying assumptions are not accepted universally by linguistics scholars for various reasons. Scholars like Geoffrey Sampson rejects every argument and every logical implication of this approach; he has argued very effectively against GL in his various works over the years, like in his book Educating Eve, and various papers like Popperian Language Acquisition Undefeated, Minds in Uniform, Revisiting the ‘poverty of stimulus’ argument, etc.

His approach is a negative one, where the nativists say they have evidence and arguments for an innate language module in the brain, and Sampson points out that their evidence and arguments are flawed. These will be outlined here, but the problems with Sampson’s own account will be highlighted. His representation of Popper is not accurate, his alternative account of language acquisition and human nature is simplistic and reductionistic, and there seems to be a disparity between his academic work and his political views. For example, he got into trouble whilst an MP for saying that Asians are better at Maths because of their genes; in Minds in Uniform he takes issue with forcing a little country to abolish advocating the death sentence for speaking out against their country’s leader (cultural relativism).

Sampson is however a liberal, and his alternative theory can be supplemented by what has come to be known as Cognitive Linguistics (henceforth CL). As an aside, Sampson, like Chomsky, admits that he does not know anything about CL.

CL as a whole was born as a reaction to what is still referred to as “mainstream linguistics”, ie. Generative Linguistics. The question I will ask here is whether this reaction was justified, as is entailed rejecting all the key assumptions upon which GL rests, and in fact virtually all of Western philosophy as well, if you accept the import of George Lakoff’s and Mark Johnson’s Philosophy in the Flesh. For such a paradigm shift to be justified, the following has to be clearly demonstrated:
Each theory will have to be independently evaluated in light of evidence; refuting evidence should be taken seriously
Each theory will also need to be assessed in terms of other methods of argumentation employed to justify their stance
Where there is a deviation from the norm, either by way of reinventing terminology, postulating counter-intuitive hypotheses, or postulating more hypotheses than necessary, it needs to be adequately justified
The implications of each theory, both within its own field and in cognate disciplines, need to be assessed as well
Aspects of each theory which may be metaphysical in nature, need to be assessed in terms of its relative explanatory power
Are there aspects outside of the theory and evidence which influences the theory, like ad homonim considerations, socio-political factors, etc.?

These are the standard criteria used when deciding which theory is a better one. It is only a coincidence that I will use the works of Karl Popper in doing so, as he is one of the few philosophers of science to systematically outline what the logic of scientific inquiry entails, though few would disagree with his approach. Had I used Einstein, Newton, Feynman, Penrose, Putnam, or any other serious scientific thinker, the criteria would be the same.

The evidence and methodology used by the GL school specifically will be assessed. If their hypotheses make predictions, I will test these predictions, and see how the theory handles counter-evidence. If the theory can be always adapted to assimilate counter-evidence, then it is problematic. If the theory can construct ad hoc hypotheses without compromising testability, it would be better. If new concepts and terms are invented to account for this, then these new concepts have to be justified in light of the evidence, not just to protect the theory. If counter-evidence is ignored all together, then the theory is not an empirically responsible one, and therefore not scientific.

Before doing this, we need to understand a little about the alternative theory, which we will turn to now:

Cognitive Linguistics
CL was born as a reaction to GL, after exponents like Lakoff initially tried to expand Chomsky’s theory to include semantics, which at the time went by the name Generative Semantics. As more things came to be known, the idea of trying to incorporate all evidence within a GL paradigm became more and more difficult, and a ‘breakaway faction’ ensued, resulting in what was documented by Harris as The Linguistics Wars. Lakoff was a student of Chomsky’s, and the latter seemed to take this as a betrayal. After Lakoff’s criticisms, Chomsky is on record as saying that Lakoff does not understand linguistics very well. Today, Chomsky claims that he knows nothing about Cognitive Linguistics!

Anyway, while Lakoff is one of the exponents, CL cannot be attributed to a single ‘founding father’. It is a collaborative movement which is always evolving, though there are theorists who are prominent within the field. As a result, there are various branches of CL, and not every linguist who calls himself a CL has to accept all the assumptions that others espouse. Hence, delineating criteria which distinguish the field can lead to some fuzziness.

Regardless, these are some of the key assumptions shared by CL as I understand it:

Language may or may not be innate. It remains to be seen how much is innate if so. Regardless, evidence thus far shows that language is part and parcel of other higher cognitive functions, and is neither an autonomous mental organ, nor does it grow – it is learnt via interaction with the environment and the mind, using all its capacities.
Language as expressed and transcribed is one our most important sources of data, and should be studied as such without assuming that each speaker is homogenous, and that his language does not change in any way.
The idea of every sentence having a deep structure is thereby also rejected, since the reasons for postulating it in the first place are not clear.
Language is at the core a semantic phenomenon which cannot be divorced from context. It makes evolutionary sense that parable and narrative evolved first, followed by other aspects of language. Otherwise, nothing precludes a computer from simulating human language.
Language could not have evolved the same way our physical organs evolved, since that assumes that natural selection has a vision to the future; if it evolved with other cognitive faculties and manifested initially as parable, it would have immediate survival value.
Language and thought are inextricably linked, so much so that without language our level of consciousness would be reduced.
Though there is an innate basis for language, it has to be learnt via hypothesis formation. Hence, the will be commonalities constrained in trivial ways by our biology, but nothing to suggest that there is a UG hard-wired into our brains.
Language is an abstract cognitive faculty, which is neither functionally nor cognitively separate from the rest of our other cognitive faculties. This predicts that damage to our language centres in the brain would affect other things as well, and that language proficiency is conducive to clear thought (and of course, that linguistic proficiency varies from individual to individual)
Language is by and large figurative (or metaphorical, in the larger sense of the word), and thought is mostly unconscious.
Empiricism and behaviourism are not rejected by fiat, but aspects of either are assessed on their merit.
Language and mind are not necessarily part of the material world. While not saying so explicitly saying so, they leave a gap in their paradigm to cater for the possibility that they might be good candidates as possible denizens of Popper’s World 3.

Once again, these assumptions are not exhaustive, and the evidence will be duly scrutinized, as with GL. As mentioned, this school was born as a reaction to mainstream linguistics, and as such will be expected to handle the evidence more responsibly, as they claim, should be commensurable with common sense, and the rejection of things like the primacy of syntax will be looked at.

Stylistic issues like the novel use terminology will be assessed as either bad style or bad science, since CL scholars do not do this. The question will be asked: are there any theoretical implications leading to pseudo-problems[1] as a result of this? If so, would the problem still be there if these additional assumptions were not made? (For example, the concept of creativity in the conventional sense does not exist anymore, for generativists – though this is more of a semantic reductionism, which is the converse…)

To avoid redundancy, the entailments of these assumptions will not be reiterated as they are pretty much the converse of the opposing paradigm, as expected.

Regarding politics, CL seem to have a less euro-centric view of the world, and are progressive in their thinking. They do not feel the need to create a hierarchy, either in the academic world or the political world. This is not only because of George Lakoff’s now defunct progressive think tank, The Rockridge Institute, but because there are explicit links between the two (unlike GL, where the links are implied and never stated explicitly; in Chomsky’s case, he explicitly denies the link).

As mentioned, there are parallels between conservative political arguments and the way GL argue. The correlation does not entail causation, as any statistician would confirm. This merely shows a parallel in thinking, with possible causation. For example, the reason gay marriage is opposed is argued for as follows:
If they allow this, then what is next? What’s to stop from condoning bestiality, and then allowing them to get married!
Or their argument against raising the minimum wage:
If you give them a two dollar raise, what is to stop the unions from demanding a hundred dollar raise the next time round!
For some reason, it has to be one extreme or the other. Gay marriage not creating a terrible problem in South Africa is not considered as counter-evidence. A colleague told me: Wait and see what happens in five years time! When I pointed out that this has been happening for five years, he did not feel the need to reevaluate his position, but without reason. Likewise, Pinker would tell you that you either accept that there is a UG, with built in x-bars and island constraints, hard-wired into our minds (and in fact many other alleged things not related to language, as he explains in The Blank Slate, How the Mind Works, and The Stuff of Thought), or that we believe that the mind is a blank slate no different from a rock – he says we must either believe there is a human nature (and therefore accept his argument in its entirety), or totally reject the idea of the existence of human nature. He is careful to quote John Locke as advocating the “blank slate” idea, knowing that few people have read his works[2], but then adds that Locke does not deserve the blame for the idea that most intellectuals today believe that our minds are empty buckets and that people are hapless victims of their environment. As with the “gay marriage” argument, the alternative does not follow; likewise, you can reject GL and still believe that there is such a thing as human nature. Likewise, Chomsky has stated that the existence of UG is unquestionable, unless you believe in magic; clearly, it is quite tenable to reject both given the evidence.

Also, as with conservative politics, there always seem to be hidden, ulterior motives. For example, in Thinking Points[3], the authors quote various right-wing journals and magazines as stating before the “war” in Iraq that:
- Iraq’s oil reserves are open for the taking, and that profits should go to American companies since Iraqis are unable to handle it
- A puppet government is what is required to make this possible
- Iraq is an ideal military base for an attack on Iran
- An invasion can be justified by villainising Saddam Hussein, and using 9/11 as an excuse
- The move is justified since the Iraqi people have nothing anyway, and a semblance of democracy would make them happy; however, Americans should do what maximizes profit for themselves…

Despite being published in dozens of dated pre-war sources, they would all deny such an agenda. They pretend it was to find WMD’s, and then to remove a brutal dictator. Only in light of the above would we understand why America never invaded to remove Robert Mugabe from power, also a brutal dictator, and certainly worse in many ways. Likewise, Chomsky was out to make a name for himself. His ulterior motive was exactly that. His representation of Empiricism is very inaccurate, as is his representation of behaviourism. When I took Verbal Behavior out of the Wits Library in 2002, it had NEVER been taken out before, despite being a very old book. The spine was not even cracked. What I found was a very sophisticated and well-written account of language learning, not very different from Piaget’s account. Skinner never believed that a parrot could learn language, because of syntactic and contextual nuances. One of the reasons he proposed was a lack of attention span. The point is, though Skinner’s assumptions may be questioned, Chomsky’s famous review was not accurate, and most people do not know this because they trust his authority and veracity. In the past, Chomsky used give many lectures on his ideas on language, preceded by polemical remonstrations against the Vietnam war; many contend that the latter drew the crowds, not the former. When I wrote to Chomsky, asking some rather sycophantic questions, he replied in the form of two hand-written letters, signed. He also used to reply to every email I sent. When I raised some critical questions, his replies got more terse. My last question was regarding CL, and his one line reply was: I don’t know much about it.

Pinker also feigns ignorance about Sampson’s Educating Eve, which demolishes his (and Chomsky’s) arguments very systematically and logically. Strangely, he uses some of the very arguments used in that book in his subsequent book. He claims that he was too busy with his subsequent projects to “dwell on critiques of old ones” (personal communication). He also uses Lakoff’s theory of metaphor in How the Mind Works, saying that many researchers, including himself, have discovered it; this was not long after pointing out in very vicious and inaccurate review that the very idea is wrong.

I cannot help but draw parallels between Bush’s rejection of Scott McCelland, former White House spokesperson, after he wrote a critical book on the Bush administration, and Chomky’s rejection of his disciples in the early 1970’s when all they did was try and improve the theory. Lakoff also alleges that many of his ideas were adopted by Chomsky, without credit.

These parallels may or may not be coincidental, but this idea will be explored in more detail to ascertain its viability. At the moment I conjecture that the methodology used by CL drives one to a progressive mode of thought, and GL drives one towards doing what the conservatives do.

It is thereby clear that GL and CL are mutually exclusive, and the conventional hybrid solution would therefore not be a viable cop-out. Both theories aim to provide a comprehensive and holistic analysis of language and mind, together with implications of how to go about doing it.

Hitherto, it seems like CL has the upper hand; its alleged empirical accountability together with the relevant political and intellectual implications seems more palatable. GL needs to adopt a less dogmatic approach, as the assumptions guiding it mostly precede the research and therefore the evidence, which is not the way it is meant to be.

It remains to be seen whether this remains true in light of a more in-depth analysis or not.


[1] This is a problem which follows only as a result of the assumptions inherent in your particular theory.
[2] And in fact Locke did indeed believe in innateness, which Pinker very well knows.
[3] A progressive handbook, available from the Rockridge Institute’s web-site: