In my previous post I proposed a definition of language: a human artifact used for contemplating or communicating topics as though they were perceptions.
How would something like that work? Perception works by shifting attention. Language works the same way.
I have been inspired to make these posts because I read through the new book by Berwick & Chomsky summarizing their long held views of language. For them, the most important thing about language is that its sentences can run on forever. To account for this, they imagine an endlessly recursive process, i.e., a machine that can go on stringing words together without limit. Thus, they can explain how we get the sentence: She told her friends to tell their friends to tell their friends to tell their friends to tell their friends that the mechanic said pizza will be served. The fact that nobody can follow such a sentence is of no importance to the authors.
More entertaining is another sentence which they discuss a few times: Birds that fly instinctively swim. Notice the ambiguity of the sentence. Instinctively can modify either fly or swim. The authors do not point out that the ambiguity would disappear in speech, depending on whether there is a short pause just before or just after the instinctively. The rhythms of speech would make it Birds that fly / instinctively swim or Birds that fly instinctively / swim. These pauses are lost in writing and the reader is left to guess which word instinctively modifies. Many people probably would not notice the ambiguity since it seems obvious that birds like ducks, which both fly and swim, do each thing instinctively. But then the authors pull their coup. Suppose the sentence reads Instinctively, birds that fly swim. In this case, it is clear that instinctively does not modify fly.
The authors seem to believe that this result will baffle all but those who embrace their particular grammatical theories. They say that rival theorists assume that in all cases the adverb modifies the closest verb. That is probably not a bad first rule of thumb. If the sentence were Birds that fly instinctively in cold weather swim, there would be no doubt that instinctively modifies the closest verb. The original ambiguous sentence placed the adverb equidistant from the two verbs. Nonetheless, the authors have found a clear example where the adverb does not modify the closest verb. Readers who want to know how Berwick and Chomsky account for this fact are referred to page 103 of their book. I consider all their premises and conclusions to be absurd and I don’t want to get bogged down going over them. Is there a way that avoids the Chomskyan swamp and can still account for our ability to understand Instinctively birds that fly swim?
I have said already that language works by shifting attention. The so called content words of a sentence, the nouns and verbs of ordinary grammar, pilot attention. A sentence like Birds fly works by directing attention to birds and then to fly. When we perceive birds flying, however, we do not see the birds and the actions as separate things. They are bound together in a unit. In the theory of attention-based syntax, action verbs are called “binders” because they bind things and actions together. Thus, in a sentence like Birds fly over the rainbow, the binder (fly over) joins both birds and the rainbow into a single dynamic image.
In English, the basic sentence structure is pilot phrase + binder + pilot phrase, in which the first pilot is the subject and the second is the object. Pilots can be modified. For instance, we could say Birds with bright pink bills fly over the rainbow. In that case the phrase with bright pink bills all refers to the head pilot, birds. If we mark the pilot somehow, say by putting brackets at the start and end, we see what is bound into fly over: [Birds with bright pink bills] fly over [the rainbow]. Before attention shifts all the words that modify a pilot must be stated. Thus, it is grammatically impossible to mix pilots, as in [Birds with bright] fly over [pink bills] [the rainbow]. Chomsky himself has often pointed out that children never make this sort of mistake.
So let’s look at Chomsky’s challenge sentence, this time with brackets inserted: Instinctively [birds that fly] swim. The verb fly turns out to be part of the pilot about birds, so of course there is no attempt to insert an external word into the phrase. Adverbs can be separated from their verbs, thanks to our working memory which allows us to carry the adverb. But working memory can be overloaded: Instinctively birds that fly and dogs that paddle swim. In this case, by the time readers have reached the end of the sentence they may have forgotten about that word instinctively. A skilled writer would probably revise the sentence: Birds that fly and dogs that paddle swim instinctively.
On pages 115-116 the authors discuss another sentence: Put [[your knife] and [fork]] down. Here the binder is split into two parts. So attention shifts from binder to two pilots that are combined into a unit, and then working memory unites the binder. Again, such a sentence might overload the working memory: Put the knife and fork which are badly bent from the way you have been banging them on the table down. In this case Put down … is recommended.
These three things--piloted attention, binding, and working memory-- make it possible to explain how a speaker creates a complex sentence without using the merge procedure favored by Berwick and Chomsky. This approach also allows us to discuss sentences in terms of a human listener’s ability to follow them. We need not take seriously absurd sentences like How many cars did you tell your friends that they should tell their friends … that they should tell the mechanics to fix?
Note: for a fuller discussion of attention-based syntax see my paper.