Logical refutation of Noam Chomsky’s famous trees – the essence of his theory of language

The fascination of Chomsky’s theory of language is due to the fact that it seems to derive linguistic diversity and complexity from a simple starting point. After Chomsky, a whole generation of linguists was busy with drawing all these elusive inverted trees. Let us stick to a simple example:

                                S

         NP                                           VP

det               N                     V                     det       N

The               boy                  eats                the       ice cream

The derivation is fascinating because of its apparent proximity to the approach of the natural sciences, where complex events are similarly derived from simple basic elements. No wonder that many praised Noam Chomsky’s approach as a revolution that finally turned the study of language into a science. The tree, with its simple peak of “S” for s(entence), seemed to define the rules that a speaker must obey in order to “generate” a potentially infinite number of grammatically correct sentences in English or any other language (hence the name “Generative Grammar”).

But right at the beginning linguists

should have asked the crucial question, what “S” at the top of the derivation is meant to represent? “S” cannot be an entity void of any content, because nothing can be derived from nothing. It must be something, but what exactly?

Of course, “S” cannot be identical with the formal end product, i.e. the English sound sequence “The boy eats the ice cream”, because then there would be no derivation at all, but the whole thing would amount to a mere tautology. Nor can “S” be a mixture of meaning and form, in the way the English word “boy” represents a phonetic form on the one hand and a carrier of meaning on the other. Then we would end up with partial tautology. The only possible interpretation is that “S” refers to something quite different: a structure of meaning. In the speaker’s brain, the real event is represented in conceptual shape, which in the act of speaking he translates into a structured linguistic form.

But then “S” as a term for s(entence) or formal structure turns out to be a misnomer. We have to replace it with another expression, say “M” as an abbreviation for M(eaning).

                                            M

                     **************************

         NP                                                       VP

det               N                                 V                     det      N

The               boy                              eats                 the      ice cream

However, once we perform this necessary distinction, it becomes obvious that we must separate the starting point “M” with a line from the following derivatives, because NP and VP represent something different from meaning, namely elements in temporal order. “The boy” proceeds in temporal sequence “eats ice-cream”. Such temporal order is not found in the conceptual structure. A unit of meaning such as “The tree is green” is independent of time. An action such as “The boy eats ice cream” is of course a temporal event like any action, but this has nothing to do with the sequence of words in the English sentence.

And “M”, which we have to substitute for “S”

exibits still one more distinguishing feature. The expression “S” suggests unity and simplicity, which, however, does not exist on the level of meaning. “The tree is green” denotes the modification of a substance by a quality. “The boy eats” or “The boy eats ice cream” refers to the modification of a person (living substance) by an action. The “Logical structure of Meaning” (see my work “Principles of Language”), portrays the most important types of such units of meaning. Each of these can take the place of “M” on the top of the tree.

Since, furthermore, the conceptual analysis of reality begins in the animal kingdom and is subject to evolution in human societies as well, only the basic types are present in all societies, not their more complex forms. In other words, evolution already comes into play at the level of “M”. Hence we must define “M” as elementary types of the “Logical and Informational Structure of Meaning” and understand them as products of evolution.

What about the components N and V of the formal level

below the structure of meaning, which, according to Chomsky, belong to General Grammar, so that we may apply them to languages as different from each other as English and Japanese? Are these terms universal? If we understand verb in the sense that it should denote a formal class in which actions occur, then this statement is certainly correct, because we may be sure to find actions like walking, speaking, striking etc. expressed by words in every natural language. So, we may of course give the name of “verb” to any formal class in any language where they occur. Doing so, we will, however have difficulties with words such as running, speaking, striking, etc., which in English and German formally belong to the class of nouns although they express actions. The conclusion seems evident: it is impossible to define verb or noun in a general (universally valid) way. All we know is that certain language-specific formal criteria make eat, run, hit etc. verbs in English, Japanese or any other language. Again, we have to modify Chomsky’s deceptively simple scheme:

                                            M

                     **************************

         NPengl                                                  VPengl

detengl           Nengl                            Vengl                 detengl  Nengl

The               boy                              eats                 the      ice cream

                     Running                      tops                             walking

Consider another example to understand this basic correction. In English we may say “Running tops walking”, which we understand in the sense that someone prefers to run rather than just go walking. In many languages this content cannot be expressed in a similar formal way. This means that already on the conceptual level, we are faced with different kinds of analysis. In some language, people must, for example, say, “I like to walk, but I’d rather run.” The agent “I” cannot simply be effaced like in English.

To sum up, Chomsky’s scheme does not in any way describe the generative linguistic capacity of human brains. On the one hand, Chomsky’s “S” is either tautological or has to be replaced by “M”(eaning) – and then becomes much more complex, since “M” consists of different conceptual types (described in the Logical and Informational Structure of Meaning). And on the formal level below the separating line, categories such as V and N are not universal – when used as such, they obscure the existing differences in the formal realization of language instead of explaining it.

That is because the transformation of structures of meaning into structures of sound does not only result in differences of syntax, i.e. in different temporal sequences (like SVO in English, SOV in Japanese), but creates differences in paratax as well, which concern the classification of concepts as formal categories like English verbs, Japanese verbs, etc.

With their deceitful simplicity Chomsky’s trees – the essence of what is new in his linguistic theory – all but obscure our understanding of language. The question why Chomsky created a scheme that so blatantly disregards basic logic, is of interest, but it must here be relegated to a footnote.*1* 

His trees need still one further correction. All the expressions above the dividing line belong to meaning, ie the immaterial conceptual structure, while all expression below belong to the acoustic chain or its representation on a sheet of paper. Now, there is no cogent reason why “eats the ice cream” or “tops walking” should be put in one class named VP rather than “The boy eats” or “running tops”. There is no justification for such classification neither on the formal level nor on the conceptual level (above the dividing line). But there is such a justification (on both levels) for grouping “dirty cloth” or “chanting joyfully” (see footnote one) each in a special class. So, we again modify Chomsky’s tree leaving out NP and VP altogether:

                                            M

                     **************************

detengl           Nengl                            Vengl                 detengl  Nengl

The               boy                              eats                 the      ice cream

                     Running                      tops                            walking

After this final transformation, Chomsky’s modified and reduced tree corresponds exactly to the general formula I had already used back in the eighties:

M formally realized as F

where M refers to meaning and F to its transformation in symbolic form.

Now lets get back to the present example. It represents a conceptual structure consisting of Agent and Patient and an Action. Its members must be separated by commas as on the conceptual level there is no temporal sequence. According to the specific rules governing English syntax and paratax the conceptual structure is then transformed into the following acoustic chain or sentence (with the arrow signifying formal realization in English):

Ag, Pt, A in English symbolically realized as The boy eats the ice cream

or, if we prefer the shape of a tree:

                                Ag, Pt, A

                     **************************

detengl           Nengl                            Vengl                 detengl  Nengl

The               boy                              eats                 the      ice cream

.

1 Chomsky inherited his position from his teacher Zellig S. Harris, the founder of distributionalism, who had excluded the semantic dimension. Harris restricted the description of language to the study of recurrent formal elements. Let us consider the following utterance:

Bird-s are chanting joyful-ly Mary wash-es all dirty cloth big cloud-s cover the sky

N                 V                Adv          N             V           Adj        N    Adj    N         V     det  N

NP                             VP                NP                       VP                   NP                VP

                  S                                                        S                                                     S

Supposed the analyzing linguist knows beforehand what elements in the unbroken spoken chainrepresent English nouns, verbs, adjectives, then he may by purely formal analysis dissect this chain into three sentences. Knowing furthermore that he may substitue any noun like for instance cloth with a larger expression like dirty cloth, he may write NP for N. Such a purely formal distributional analysis may, of course, be turned upside down. Then “S” is placed at the top but that doesn’s change its nature: it remains strictly tautological and is still a mere abstraction representing no more and no less than the respective formal chains in English. By a mere sleight of hand Chomsky turned an analytical process – a tautology – into a derivation. The apparent miracle was nothing more than a logical error.