ERRATIC STRUCTURES.

Reading a copy of Syntactic Structures I found what I thought was a mistake, not a big one, and subtle enough not to easily catch the eye.

A rather trivial matter one might think. Typos are, after all, quite common and, therefore, relatively unremarkable. Why make an issue of it?

I was reading Chomsky for my research. I am an artist interested in text generation. Chomsky is a major influence on this area through his work on grammars. A considerable part of Syntactic Structures is spent repudiating the relevance of random methods of sentence production. These methods go under technical sounding names such as Markov finite state processes, or finite state machines.

A Markov process is basically, in this context, a probabilistic way of generating a sentence. Essentially, it is a calculation of the likelihood of sequences of words occurring. (See my essay, Markov Chain Algorithms. A not very technical explanation, at http://in-vacua.com/markov_text.html.)

It seemed possible to think of an error in Chomsky’s text as a random substitution based on the probability of a phrase. (This might perhaps appear to contradict the general argument Chomsky was putting forward: that what he calls transformational grammars can account for all possible sentences.)

This is the passage in Chomsky (1969):

ON THE GOALS OF LINGUISTIC THEORY

6.1 In §§ 3, 4 two models of linguistic structure were developed: a
simple communication theoretic model and a formalized version of
immediate constituent analysis. Each was found to be inadequate,
and in § 5 I suggested a more powerful model combining phrase
structure and grammatical transformations that might remedy these
inadequacies. Before going on to explore this possibility I would
like to clarify certain points of view that underlie the whole approach
of his study.

Our fundamental concern throughout this discussion of linguistic
structure is the problem of justification of grammars.

The problem is “his study” (it should be “this”, as in “this discussion” that follows in the next sentence.)

Mine is a 1969 edition. The most recent I know of is 2002. The error had not been corrected in the elapse. But was it credible that no one had noticed in the more than thirty years that had passed? The answer might be simple. I could be wrong – and perhaps in danger too of making a mountain out of a molehill.

Nor could I attempt to build a theoretical edifice on my mistake: if anyone’s it had to be built on Chomsky’s, or at least that of his text. So I decided to email Chomsky.

As a famous Professor at a renowned institution his address was easy to find.

This is the text of my mail:

Dear Professor Chomsky,

forgive me for mailing you on a subject that may seem quite trivial.

I am a final year research student at Chelsea College of Art and Design, London. My research relies, in part, on several of your texts.

On page 49 of Syntactic Structures (1969 edition) there is the sentence:

"Before going on to explore this possibility, I would like to clarify certain points of view that underlie the whole approach of his study"

Should it not be "this study", as in "this discussion" that occurs in the next sentence? (There is no 'him' to refer to in this first paragraph of Chapter 6).

I say in my thesis:

"Presumably, Chomsky’s sentence must be written according to a transformational grammar, as he claims (at least in Syntactic Structures) that such a grammar is adequate to account for the production of all English sentences (including, therefore, his own). But the error in it, the substitution of “his” for “this” is the irruption of noise into order. Thus we have one ‘grammar’ breaking into another. We have a single state machine disrupting the transformational machine. Why do I say this? I say it because the presence of "his” is determined by the preceding state of the sentence. In Chomsky’s text, I suggest, “his study” is approximately as probable as "this study" (the true state of the sentence as it should be, I believe, in the transformational grammar.)

The presence of "his" can be seen as a probabilistic substitution that a Markov finite state grammar can account for. Therefore, we need at least two grammars to account for this sentence."

I would be very pleased to hear if my observations are at all accurate or interesting if you have time to reply.

Yours Sincerely,

Wayne Clements

Rather to my surprise and pleasure, Professor Chomsky replied almost immediately.

He wrote:

I don't think I have the 69 edition, but sure sounds as though it

should be

"this." The original was never even proof-read.  Dutch publishing in

those

days cut all possible costs, and was dirt cheap -- which is why this

could

appear in the first place.  No one else would touch it.

Noam Chomsky

Encouraged by Professor Chomsky’s kind response, I decided to email him again:

Dear Professor Chomsky,

Thank you very much for a speedy reply. I’ve checked the Mouton de Gruyter, 2002 edition and the text is the same there.

I wonder if Markov processes might be adequate to a theory of the typo in general? I would be very interested to hear what you think.

Best wishes

Wayne Clements

Again Professor Chomsky replied promptly. I had tried to draw him out, it is true, on my theory about Markov processes and textual errors. Professor Chomsky was quick to put me right not only on the responsibility for the crux in his text, but also on my theoretical ambitions.

Out of curiosity, I looked up the earlier edition, using the 1966

edition

which is, I believe, identical to the original 1957 edition.  That has

it

correctly.  So evidently the typo was introduced in later editions,

which

are never given to the author to proof-read.

I doubt that Markovian processes would be of much if any help in

account

for typos.  This one, for example.

Noam Chomsky

I emailed him thanking him for his consideration. Somehow I did not think there was much more mileage in taxing him with my ‘typo theory’ at the time, and the exchange seemed to have come to a natural conclusion.

…

However, reading on a few pages I found what looked like another printer's error. This time I decided to check it for myself. I arranged with the British Library to look at their first edition copy of Syntactic Structures.

Having seen it, I wrote:

Dear Professor Chomsky,

You may remember, I wrote to you recently about a textual error in Syntactic Structures. I have now found another in later editions.

On page 55 of my 1969 edition: “…the set of grammatical sentence (sic) and the set of observed sentences.” (That is, the singular.)

In the first (1957) edition on page 55 it is “…the set of grammatical sentences (sic) and the set of observed sentences.” (Now the plural.)

Therefore, presumably the error was the printer’s, as was the earlier case.

Has this mistake come to light before? If so, I am sorry for troubling you with it.

Should I find any more (there may be none), shall I pass them on to you?

Best wishes,

Wayne Clements.

University of the Arts.

Chomsky replied:

Thanks. I checked my 1966 edition and it's correct. I suppose they must have reset the 1969 edition. I don't recall ever being told, or asked to look at it. I'll register it in case I'm ever asked by the publisher for a corrected edition. Please do send more if you find them.

Noam Chomsky

…

Paper printed texts conventionally are of a higher status than electronic texts. However, as the above correspondence makes plain, it is possible for mistakes to be introduced into them by other hands at a later date and for these to pass unnoticed for several decades, if noticed at all.

If you have the proofreader’s eye, these mistakes seem to abound in many texts you might read. But would more examples lead to one being thought a crank? Perhaps this fear is the reason no one seems to have developed this most pedagogic of all themes. We still lack a theory of the typo.

An obvious potential candidate as a theory could be the psychoanalytic: an adaptation of Freud’s theory of parapraxis (the “Freudian slip”) from the spoken to the written. But I am not convinced. We do not need a theory in which Chomsky subconsciously suppresses “this study” and replaces it with “his”, perhaps with a desire to distance him from what was then a still untried work. This is primarily because the error is not Chomsky’s, but the printer’s.

Of Chomsky’s text I wrote at the time I found the first error and before contacting the author:

“I think we have here a case of what Kittler (1990) is referring to when he writes of: "Writing and writers as accidental events in a noise that generates accidents" (p. 184). There is, in my understanding, little space in Chomsky's theory for these accidental irruptions of noise into the structuring of grammatical utterance, even when Chomsky's own is marked by it. This is perhaps because of his interest in syntactic structures, whilst probabilistic models have been ruled out by him of producing any truly useful insights  (Chomsky p. 17).” 

Noise disrupts the signal whether the channel is electronic or paper and ink. This still seems to me to be a viable route for investigation, although I may not have yet persuaded Professor Chomsky.

…

I wish to thank Professor Chomsky for his patience and time in dealing with my queries.

Bibliography

Chomsky, N. Syntactic Structures,’s-Gravenhage: Mouton & Co, 1957.

Chomsky, N. Syntactic Structures, The Hague: Mouton & Co, 1969.

Chomsky, N. Syntactic Structures, Berlin; Hawthorne, N.Y. : Mouton de Gruyter, 2002.

Kittler, Friedrich A. Discourse networks 1800/1900, trans. Metteer, M, and Cullens, C. Stanford California: Stanford University Press, 1990.

.