Visible Syntax

This blog has lamented several times the absence of an attentional syntax of some sort. We need, or at least I need, a way to discuss the relation between words in terms of the attentional triplet (speaker, listener, topic). Thus, I have been pleased to find an essay in the September issued of Trends in Cognitive Science that suggests what such a syntax would be like.

The essay, “The Simpler Syntax Hypothesis” (abstract here)  by Peter. W. Culicover and Ray Jackendoff is a 5-page summary of a 600-page book published a year ago, Simpler Syntax.

This blog has shied away from syntactic analysis because (a) it is not my strongest suit, (b) its technical nature is bound to scare away some visitors to the blog, (c) its very high level of analysis does not permit many concrete statements about particular languages, and (d) it is very removed from evolutionary credibility. There has also been a kind of distracting deus ex machina in the form of attentional semantics. The latter is too central to the focus of this blog to pass over, even though it was no where in my mind when I began preparing to create Babel’s Dawn.

Syntax is central to language, however, and cannot be set aside forever. We can never understand what evolved to make us a linguistic species without knowing the nature of our syntactic powers. So this week I want to take a good look at Simpler Syntax. Simpler than what? Simpler than the elaborate trees developed by Chomsky and his disciples over the past fifty years.

“The Simpler Syntax Hypothesis … holds that much of the explanatory role attributed to syntax in contemporary linguistics is properly the responsibility of semantics.” (p. 413) It is a natural draw for this blog because it makes good sense in an evolutionary context. It “claims that the foundation of natural language semantics is combinatorial thought, a capacity shared with other primates.” (p. 417) I will confess that it also fits better with my own empirical observations as a writer and editor, but if it did not make evolutionary sense I would probably still let it pass by.

Simpler Syntax rejects the notion that the meaning of a sentence is completely contained within the meaning of its parts and the syntactic rules that combine them. For example, if somebody says to you Ozzie tried not to drink, you will understand that Ozzie was tempted to drink and, probably, you will suspect that Ozzie has been drinking.

If you think about the sentence in attentional terms, it seems straightforward. The speaker/writer draws your attention to Ozzie and then tells you something about him. You suspect Ozzie did drink because otherwise there would be no reason to tell you he tried not to. If he didn't, the speaker probably would have said something along the lines of Ozzie managed not to drink.

If you analyze the sentence formally, however, not to drink is a separate syntactical structure from Ozzie tried, but the subordinate idea is not complete. You cannot tell from its three words who is not drinking. Even though the sentence is likely to strike most readers as transparently simple, its meaning cannot be drawn from its syntactical form alone. If you want to argue that the whole meaning emerges from its words and syntax, a simple sentence like this one is a challenge.

The solution proposed by Chomsky and widely accepted has been to insert an unvoiced “deep structure” (now called a logical form) in which a silent pronoun is inserted. The two deeper parts of the sentence are [Ozzie tried] + [PRONOUN not to drink]. (The implicit message that he did drink is not covered in formal analysis.) Chomsky's solution contradicts Giorgio Marchetti’s proposal that words are pilots directing attention and that the meaning of a sentence comes from that piloting. The pronoun in this deeper structure is unspoken so our attention can never be steered toward what it refers to.

Prior to Chomsky, speech had seemed like the most conscious of human activities. I speak, you listen, and we share a conscious experience. With the introduction of deep structure, much vocal interaction was declared to be unconscious as the listener moved through a hierarchy of structures, some of whose elements are inaudible to the listener’s ear. "Simpler Syntax, by contrast, regards linguistic meaning as largely coextensive with [conscious] thought.” (p. 416) The complexity that Culicover and Jackendoff are seeking to simplify is to reduce the number of hierarchical levels required and to abolish hidden elements in syntactic structure. Thus, even though attentional semantics is nowhere part of their concern, they have provided a syntax in which attentional semantics appears to be possible.

The radical nature of this approach is clearly stated by the authors. They opened their article by asking and answering a question:

What roles do syntax and semantics have in the grammar of a language, and what are the consequences of these roles for syntactic structure? These questions have been central to a theory of grammar for close to 50 years. We believe that inquiry has been dominated by one particular answer to these questions and that the implications have been less than salutary both for linguistics and for the relation between linguistics and the rest of cognitive science. (p. 413)

The one-particular, less-than-salutary answer they refer to has been Chomsky’s. The unsalutary results they refer to include an inability to cover many “empirical linguistic phenomena” (i.e., inattention to many of the sorts of things people say and imply) and poor “integration into the cognitive scientific enterprise” (that is to say linguistics has been off in its world, unperturbed by the contrary evidence found by students of how we think, how other animals think, and how humans evolved). It’s a startling indictment of what is commonly cited as one of the leading intellectual achievements of the twentieth century. More tomorrow.

-------------------------------------

NOTE: Dr. Jackendoff has written to say: "You quoted our book: "Simpler Syntax, by contrast, regards linguistic meaning as largely coextensive with [conscious] thought.” (p. 416). Your insertion actually violates my sense of the situation. In my book "Consciousness and the Computational Mind" as well as several subsequent publications, I've argued that meaning is *never* conscious except via linguistic imagery, which is primarily phonological in form, or through other forms of imagery such as visual, tactile, and auditory imagery.
This is too complex a story to reproduce in an email. the point only is that I suspect that you have a different notion of thought than I do."

Comments

[Giorgio Marchetti]

If I have understood well what Bolles’ observation implies, we have to do here with a particular instance of a more general case.

Referring to the sentence:

“Ozzie tried not to drink”,

Bolles observes that:

“If you analyze the sentence formally, however, not to drink is a separate syntactical structure from Ozzie tried, but the subordinate idea is not complete. You cannot tell from its three words who is not drinking. Even though the sentence is likely to strike most readers as transparently simple, its meaning cannot be drawn from its syntactical form alone”.

The problem highlighted by Bolles is that certain parts of speech are not (always) explicitly stated (“The pronoun in this deeper structure is unspoken so our attention can never be steered toward what it refers to”).

In my opinion, this lack of explicitness is a very common trait of many languages: so common that you can find it in simpler, albeit very important, structures, such as for example the noun-adjective correlation, the subject-verb correlation, and the verb-object correlation.

If you consider the noun-adjective correlation, you can realize that apparently there is no part or element that explicitly links or combines the noun with the adjective, and that pilots your attention so as to form a noun-adjective correlation. I say “apparently”, because this function is actually carried out (at least in English) by the order of the words, and by their proximity/nearness.

Say “red” and then after a few seconds say “hat”: the two words have nothing in common: they are isolated. But then say “red hat”: immediately the two words become correlated by means of their position and order (the order is very important: in Italian, an adjective can either follow or precede the noun, giving rise to two different meanings).

Likewise, the case considered by Bolles, the subordination of clauses, is not always explicitly spoken or stated by a word, but sometimes is implicitly left to the order of words (or to some other linguistic device or means).

Not all languages behave in the same way. Each language chooses and adopts its own way of symbolizing (or not symbolizing) meanings: some languages express them explicitly by means of words (or suffixes, prefixes, and so on: see the Italian “fiori rossi” = “red flowers” and “fiore rosso” = “red flower”); some others, leave them to the order of words, to their intonation, and so on.

Nonetheless, whether a language has adopted words, prefixes, the order of words, or whatever else, both the speaker and the listener of a given language know how their language function: that is, they produce and re-produce, respectively, a certain meaning by performing the attentional operations conveyed through the specific means (words, the order of words, suffixes, etc.) adopted by their language. Even a pause, then, can assume a fundamental meaning.

Giorgio Marchetti