OK. Let’s wrap this up.
Part Nine. Beyond Chomskyan Linguistics.
Let’s say that we wanted to create a general theory of linguistics to compare to Chomsky’s ideas. Rather than taking as our starting point the rather dubious proposition that language is hard-wired into the brain, would it not make more sense to try to use as our foundational idea a fact which is indisputable and which is supported by evidence found in every language in the world, living or dead – better still, in every word of every language? And then would it not make sense to try to connect all of our theoretical postulates to that foundational idea? If that seems like a reasonable way to proceed, then the foundational idea I would choose reads like this:
In all human language, the connection between meaning and representation is arbitrary.
One thing we can say with certainty about the brain is that it is capable of associating any chosen word with any chosen idea. Thus, there is no reason to think that the Tanzanian fisherman’s samaki is any better or any worse than the Samoan sailor’s i’a as a word that indicates a particular animal that we speakers of English happen to call a fish. As the above is true, then this must be equally true:
There are no limitations on how many or how complex the meanings associated with a word can be.
Thus, the word firm can be a noun, a verb, or an adjective, and coronation can have a large complex of meanings associated with it. I once read that the Chinese word “i” has eighty-three meanings. Cauliflower, on the other hand, seems to have but one meaning. Ask Chomsky to explain what anarcho-syndicalism means. I guarantee the explanation will take a while.
As a third fundamental idea, I would add this:
It is the arbitrary nature of language which gives it its labile, or unstable, quality but which at the same time gives it the power of creativity.
Let’s see if we can take these few uncontroversial facts and elaborate on them somewhat in order to explain more about how language works than Chomsky can explain with the biolinguistic model:
· Given, as seen above and in Part Four, the range of complexity and breadth of ideas which can be found in vocabulary words, there is no reason to believe that complex and wide-ranging ideas cannot operate as grammatical constructs as well;
· From there it follows that, since language is a creative process, and since people can be quite ingenious in their use of language, there is no reason to think that we cannot encode into our grammars ideas which represent human relations between speaker, hearer, and others (the seems construct);
· From the above we can reason that, since there is no reason to think that highly specific ideas such as complex human relations (seems) are encoded by the process of evolution any more than we would think that highly complex vocabulary words such as coronation are encoded by evolutionary processes, the notion of language as an intricate construction and agglomeration of ideas makes far more sense than does the notion that language is hard-wired into the brain;
· This leads us to the notion that, taking into account that ideas can parallel one another, can work harmoniously, can be contradictory, ambiguous, or even nonsensical, we now have the basis for the wide range of variation in language production – from the highly poetic to the merely grammatical to the barely comprehensible to utter nonsense;
· From there we can go to the notion that if language is a set of ideas, then logical contradictions in language produce ungrammaticality. Thus there is a grammatical/ungrammatical divide, Chomsky and Wittgenstein notwithstanding;
· If we assume that a commonality of experience accounts for a commonality of ideas, the so-called universals which have been found in languages can be attributed to the effort on the part of all speakers to express similar grammatical ideas. Human ingenuity allows people to encode a wide range of grammatical ideas; thus, the universality of function in languages is a concept with far more legitimacy than Chomsky’s professed but failed endeavor, which was to find a universality of form;
· Additionally, because the connection between meaning and representation is arbitrary, giving to language its unstable nature, there must exist braking mechanisms which hold languages together in order to ensure common understanding. There are two major forces which do this: sociopolitical forces and creativity itself;
· Sociopolitical forces act as a brake on linguistic change. These forces can be benign, as in the case of parents or teachers; benign or neutral for those who are in a position of social superiority. On the other hand, malign social forces such as conquest or political suppression induce language change for the less powerful and provide language stability for the more powerful;
· The creative nature of language, deriving as well from its arbitrary nature, is an important source of language change. However, some creative works, such as plays, poetry and other works, including religious works, through publication and wide dissemination, help to preserve linguistic traditions, thus stabilizing the language. For speakers of English, good examples of this phenomenon would be the Oxford English Dictionary, which uses the literary history of England for its vocabulary citations; the King James version of the Bible; and a huge output of literary works.
· And so if a teacher recommends a literary classic to a student, language stability is being enhanced by the sociopolitical status of the teacher, who recognizes value in a work of linguistic creativity, thus reinforcing the traditions of the language. But if the student decides to blare his rap music in class, the student is asserting a new form of creativity, usually employing language which is alien to the teacher, who would most likely suppress the student’s initiative. The teacher would see the matter as disruptive to the purposes of the class, whereas the student would view the suppression as a political act. So goes the world.
· Finally, because there has been so much political fragmentation in the course of history, there is now a large – but shrinking – number of languages in the world. And because of modern communications and, in recent past times, colonialism, the world has seen the emergence of a number of very widely spoken languages, such as English, Spanish, and French. Chinese is spoken by a great many speakers due to political consolidations of past millennia in Southeast Asia, and Arabic is spoken in a large number of culturally diverse countries due to the spread of Islam.
All of the ideas expressed above – ideas with enough cohesiveness, I would suggest, as to constitute a theory of language – are derived directly from the simple and undoubted fact that in all languages the relationship between meaning and representation is arbitrary. Biolinguistic theory, on the other hand, takes, as its fundamental presupposition, a clear and unambiguous pathway by which language is processed through the brain. A casual study of this concept reveals the following deficiencies:
· Assuming that such a pathway could be found, biolinguistics would then be at a loss to explain how it is that a single organic pathway is capable of producing such a wide range of productive result, to say nothing of why it is that reason, judgment and experience must then be invoked in order to choose the proper level of grammaticality, formality, register and tone which an individual employs in daily speech;
· Biolinguistics fails to explain many grammatical constructs, particularly those which can be seen as codifying human relationships, because biolinguistics does not take into account matters which are external to the function of the brain;
· Biolinguistics has nothing to say about the vital role of sociopolitical forces in language change and stability; indeed, the theory minimizes the importance of such forces (poverty of the stimulus);
· Biolinguistics has nothing to say about the relationships among the forces of creativity, language change and language stability;
· Biolinguistics treats syntax as a world unto itself, and fails to see the obvious connection between syntax and morphology as complexes of ideation. Furthermore, the biolinguistic model ignores the ingenuity of humans, as seen in their various forms of literature, and completely ignores the possibility that syntax is also the product of creativity and ingenuity;
· Biolinguistics, by not seeing language as ideation, with the inherent implication of ideational conflict, fails to see a grammatical/ungrammatical divide, even though such evidence is abundant. Consider Sentence (A) from Chomsky’s speech:
(A) *There seems a man to be in the room.
(B) There seems to be a man in the room.
Chomsky himself can’t tell you why we don’t employ Sentence (A) (nor can he tell you why the biolinguistic process produces flawed products) but no one would dispute the fact that (B) is fine. Since both sentences contain the exact same elements, then this means that the difference must be explainable by logic alone. One doesn’t need to know how the brain is wired, or what the language of thought is or how language evolved over millions of years in order to explain the difference, any more than one would need to know those things in order to solve a problem in geometry. We have already seen abundant evidence in Part Six that the biolinguistic model is not concerned with finding logical solutions to problems such as the above.
Throughout history, the way in which the truth of ideas is verified is through logic; the biolinguistic model, however, does not trade in logic, but in circuitry. In fact, I would suggest that the reason the Chomskyans are having such a hard time figuring out how language works is that they decided a long time ago that the grammaticality/ungrammaticality divide was an unfashionable concept. It seemed too much like right versus wrong. For the Chomskyans, the age of Aquarius has never ended.
In conclusion, we can explain far more about language by using the most basic law of language – the arbitrariness of meaning and representation – than we could possibly explain with the biolinguistic model. Chomsky’s model is novel and innovative but ultimately it is not intellectually well-founded; in addition, it is self-limiting in the scope of its inquiry. On the contrary, the ideational model takes a very commonplace and universally employed property of human language and follows that idea to its many logical and cohesive conclusions.
John Williamson is a contributor to The Anti-Chomsky Reader and is writing a book about linguistics.
Footnote from Part Seven: (Addressed to Chom Ling)
1. First of all, let me say that even if I am able to convince you of my claim, I will not deserve any credit for solving this specific problem. That is because this problem was solved a long time ago; in fact, the problem has probably been rediscovered, solved and recorded numerous times over a number of centuries by various people in English-speaking countries.
What we are dealing with is a problem which has been noticed by people, mainly serious writers, I suppose, who care about precision in writing. And so a special structure was invented in order to solve this particular problem. I don’t know who invented it, but it’s been kept alive for all these years because people keep noticing the problem and they want a way to deal with it.
I first took note of this special structure over twenty years ago while reading a grammar book in a public library in Alexandria, Virginia. I was very interested because, although I had seen sentences containing it, I had never seen an explanation of it. Although I can’t recall the exact semantics of the sentence, the syntax is contained in this similar sentence:
(B1) These are the arguments which it is impossible for me to refute.
Now you are no doubt wondering what that it is doing there after which. You and the great majority of people in the English-speaking world would probably have said:
(B2) These arguments are impossible for me to refute. or,
(B2b) These are the arguments that are impossible for me to refute.
The potential problem with structure (B2) is that the question has presupposed the structure of the answer, the presumed question being:
(B3) Which arguments are impossible for you to refute?
But suppose the real answer is not (B2) but (B4):
(B4) This argument is impossible for me to refute. or,
(B4b) This is the argument that is impossible for me to refute.
And so the plurality anticipated in (B3) is not answered by the singularity expressed in (B4). The question presupposes multiple arguments, but the answer refers to only one. Or vice versa:
(B5) Which argument is impossible for you to refute?
(B2) These arguments are impossible for me to refute.
You see the problem, of course; but suppose the question were the very elegant:
(B7) Of these arguments, which is it impossible for you to refute?
Or these alternates:
(B7b) Which of these arguments is it impossible for you to refute?
(B7c) It is impossible for you to refute which arguments?
Either the singular (B4) or the plural (B2) will answer (B7). This is because of the following advantages of this special structure:
(a) The question’s main subject and main verb agreement is between it and is and that relationship is fixed;
(b) Which can stand for either singular or plural responses; from the Oxford English Dictionary: “Which: Expressing a request for selection from a definite number: What one or ones of a stated or implied set of persons, things or alternative?” I would add that which can also anticipate none.
(c) The other verb form in the sentence (to refute) is an infinitive, and thus not to be conjugated, and so agreement isn’t an issue.
Now, if you word your answers in a subject plus copula plus adjectival phrase form, you get:
(B2) These arguments are impossible for me to refute.
(B4) This argument is impossible for me to refute.
where are must agree with arguments and is must agree with argument.
But the important thing is that both (B2) and (B4) are anticipated by Question (B7), and yet neither the singular nor the plural is presupposed.
It is a quite useful structural element; whether people choose to take advantage of that utility is another issue.
2. Let’s postulate that, in order for a question to be grammatically correct, any potential answer should be able to be substituted for the interrogative, in either a reformulation of the question or in a reformulation of the question as a statement. By the term any potential answer, I mean any answer which is not constrained by the semantics of the question sentence. Thus, if the question were:
Who was the guy I saw you talking to when I passed you in front of the bank on Saturday?
then we know that there can be one and only one potential answer: that one person who I was talking to, etc. Let’s say that that person’s name is Jim. Thus, reformulating the question:
Was Jim the guy I saw you talking to when I passed you in front of the bank on Saturday?
And then reformulating the question as a statement:
The guy I saw you talking to when I passed you in front of the bank on Saturday was Jim.
But now suppose the question is this:
Who were those two guys I saw you talking to when I passed you in front of the bank on Saturday?
And suppose the answer that comes back is Jim. Substituting Jim in both reformulations, we get:
* Were Jim those two guys I saw you talking to when I passed you in front of the bank on Saturday?
* Those two guys I saw you talking to when I passed you in front of the bank on Saturday were Jim.
Both of these answers are both semantically and syntactically flawed. Semantically, because the question itself demanded an answer which named two people; syntactically, because a single person substituted for the interrogative in either reformulation produces a syntactic conflict in the form of the verb. And so the question itself wasn’t ungrammatical; rather, the answer given could not have satisfied the semantic requirements of the question. Furthermore, the semantic requirements for any correct answer required a plural form of the verb. This answer, however, would have sufficed:
Jim and Roy were those two guys I saw you talking to when I saw you in front of the bank on Saturday?
Those two guys I saw you talking to when I saw you in front of the bank on Saturday were Jim and Roy.
So we see that, in order for any answer given to be correct, it must be responsive to the semantic requirements of the question – requirements which in turn determine the syntactical form of the question. It must follow logically then, that no correct answer, defined as an answer which meets the semantic requirements of the question, should be prohibited by the syntax of the question from being able to be substituted for the interrogative, either as a reformulation of the question or as a reformulation of the question as a statement. And so let’s this test idea with these examples:
Who did you see at the party last night?
The subject-verb agreement is between you and see; who is unrestricted as to singular or plural; the semantics of the question do not restrict in any way the number of people who could have been seen at the party.
Potential answers: (a) Sarah (b) Sarah and Louise. (c) Sarah and Louise and Mary. etc.
Let’s substitute (a) and (b):
(a) Reformulation of question, substituting potential answers for interrogative:
Did you see Sarah at the party last night?
Did you see Sarah and Louise at the party last night?
(b) Reformulation of question as statement, substituting potential answers for interrogative:
You did see Sarah at the party last night.
You did see Sarah and Louise at the party last night.
Now let’s try this with the LI question: Who is easy to get Bob’s mother to talk to?
There is nothing in the semantics of the sentence which limits the number of potential answers. Thus, if we asked the question of Mary, she might answer “Louise” and if we asked Louise she might answer “Hilda”. On the other hand, if we construe the semantics in this way:
Who is always the easiest person to get Bob’s mother to talk to?
then there is only one correct answer and the verb would have to be singular.
Now, why can’t we just assume that who in (LI) is restricted to singular since we are using a singular verb? Well, the extremely obvious answer to that is that the subject determines the number of the verb, and not vice versa:
The girls are out shopping.
The girl is out shopping.
Thus, the verb is can’t just restrict the scope of who to make it singular. Since the scope of who can’t be restricted by the verb, then it has to be restricted by the semantics of the sentence; however, as we see above, the semantics of the (LI) sentence are not restrictive as to the number of potential answers. So here’s the test:
Potential answers: (a) Larry (b) Larry and Muriel (c) Larry and Muriel and Mabel. etc.
(c) Reformulation of question, substituting potential answers for interrogative:
Is Larry easy to get Bob’s mother to talk to?
*Is Larry and Muriel easy to get Bob’s mother to talk to?
(d) Reformulation of question as statement, substituting potential answers for interrogative:
Larry is easy to get Bob’s mother to talk to.
*Larry and Muriel is easy to get Bob’s mother to talk to.
As seen, there is an answer which is not prohibited by the semantics of the question but which is in fact restricted by the syntax of the question from being able to be substituted for the interrogative in both the reformulation of the question and the reformulation of the question as a statement. Thus, the (LI) sentence is ungrammatical. QED
On the other hand, the sentence below, which I suggested in my first article as being correct, has no such problems:
Who is it easy to get Bob’s mother to talk to?
Potential answers: (a) Larry (b) Larry and Muriel (c) Larry and Muriel and Mabel. etc.
(e) Reformulation of question, substituting potential answers for interrogative:
Is it easy to get Bob’s mother to talk to Larry?
Is it easy to get Bob’s mother to talk to Larry and Muriel?
(f) Reformulation of question as statement, substituting potential answers for interrogative:
It is easy to get Bob’s mother to talk to Larry.
It is easy to get Bob’s mother to talk to Larry and Muriel.
A few notes:
1. (a) The semantics of easy to get Bob’s mother to talk to determines the scope of who. Since there is nothing restrictive about those semantics, who is unrestricted.
(b) It is the scope of who which determines the form of the verb. Since who is unrestricted, it can license neither is nor are, since both of those forms are restricted. Thus, the failure of the (LI) sentence is at the semantic-syntactic interface.
2. The structure I have suggested:
(Q2) Who is it easy to get Bob’s mother to talk to? works this way:
(a) The question’s main subject and main verb agreement is between it and is and that relationship is fixed;
(b) Who is unrestricted as to scope;
(c) Who is the object of the other verb form in the sentence (to refute), an infinitive. It is hard to get more unrestrictive than an infinitive.
3. However, all is not lost. If your purpose is to ask the question:
What is the name of just one person who is easy to get Bob’s mother to talk to?
there is an available solution:
Who is someone who is easy to get Bob’s mother to talk to?
Larry comes readily to mind, among many others.
Larry, but if he’s not in town, try Nancy.
4. If you want to know why (LI) “sounds OK”, that, too, is easy to see. We can restrict the interrogative subject of this structure thus:
(Q4) Which one of them is easy to get…
and we would then have a singular subject plus is easy to. Of course, we have to use one to restrict which that narrowly, (although note that we can’t say *Who one is easy to get…)
Thus (LI) parallels (Q4) pretty closely in terms of syntax, but the semantics aren’t precise enough to make (LI) good.
5. In the event you are asking whether semantic conflict makes an otherwise syntactically acceptable sentence ungrammatical, I offer this:
He is buying the book for himself.
*He is buying the book for herself.
6. If you want to say that the (LI) sentence is grammatical, then does that make my suggested sentence (Q2) ungrammatical? Does that make the it redundant? What would be your argument along those lines? Please clarify.
3. Relevance of this problem to larger issues.
1. I believe that this problem highlights the danger of using personal intuition as the basis for determining grammaticality. I would have to assume that more than ninety-nine percent of English speakers do not use the structure which I have suggested is grammatical, nor are they even aware of its existence. But that’s not particularly relevant to the grammaticality of the structure: it is logically based, and so it is correct.
Generative grammarians talk about how they describe rather than prescribe. But I think that that is the wrong distinction to make. The important thing is whether or not they understand.
I assume that you have read Chomsky’s Minimalist Program. In that book he discusses structures very much like these we have been discussing. However, at no place in the book does he show a unified, clear, understandable way of resolving issues such as those discussed above. MinPro is just a lot of theoretical speculation and one dead end after another.
In fact, if the generativists had been able to solve problems such as these, we wouldn’t be seeing the (LI) sentence in the latest issue of Linguistic Inquiry in 2004.
The irony is that the generativists threw out all the old ways of looking at language, and basically they have ended up trying to reinvent the wheel, not realizing that a good deal of what they threw out or ignored was accumulated knowledge which they would eventually be looking for. In essence, the generativists have been trying to rediscover a lost world, and not doing a very good job of it at that.
2. I know that generative grammarians bandy about terms like computational efficiency and so forth. The problem under discussion is a good test of whether you folks practice what you preach because, in this instance, computational efficiency is in direct conflict with what people say. Most people say, Who is easy to get Bob’s mother to talk to. If we say that whatever people say is grammatical, then that sentence would have to be grammatical. Yet in terms of computational efficiency, it’s a disaster. If the answer to the question is Larry and Maude, then we have to say, in effect, “I was expecting to plug in a singular answer before is, but now I will have to remove the singular form, retrieve the plural form are, and insert it.” (I am assuming that the brain works just like a computer. It’s kind of fun. I can see the attraction.)
To me, the sentence Who is it easy to get Bob’s mother to talk to? does not have the problem described above, because neither the singular nor the plural is assumed in who. No presuppositions are committed to, and therefore no corrections need to be made.
How can generativity be valid if the brain “creates” these inefficiencies for itself? It doesn’t make any sense. It seems to me, rather, that computational efficiency comes through education, motivation, practice and study, (kind of like practicing the piano). These, of course, are cultural constructs.
Speaking of practicing the piano, you may say, “Wait a minute. A concert pianist has to practice, since the musical structures aren’t already there in his brain. OK, you’re right. But I am currently studying Arabic, as I mentioned, and I can guarantee you that the only thing I know of the language is what I have learned. If anybody should be able to turn on the “Arabic” parameters pretty quickly, I should, but that’s just not the way it is. It’s a laborious process.
3. In my article I also wrote about the sentence: Me and Larry went to the races. Many people use that locution. The computation seems to go like this:
1. If a singular subject, use nominative: I went to the races.
2. If two or more subjects, override nominative and use oblique form: Me and Larry went to the races.
3. If you hear someone say He and I went to the races. you must convert both subjects to the oblique case in order to understand the sentence. Thus: Him and me went to the races.
4. Better still, invert the order of the oblique subjects: Me and him went to the races.
I would suggest that, since we are assuming that the brain generates language through hard-wiring, that the above processes are hugely inefficient. One also must wonder whether the generativists are suggesting that, since the standard sentence in (3) is more likely to be used by more educated, higher income persons, and the substandard sentences in (2) and (4) are used by those who are less educated and in lower income groups, that we should conclude therefrom that those in the lower socioeconomic brackets have brains which operate differently from others.