On Tue, 06 Nov 2007 07:17:51 -0800, Christopher Green wrote: >Frankly, I think that Mike raises a bunch of red herrings here.
I can understand why you might think that but only because you seem to place such importance on syntax. >The issue is not whether human language is ALL syntax or not. Perhaps we should start by asking a fundamental question: What is syntax? There are a variety of answers but let's consider two mundane ones: (a) syntax refers to the ordering of elements (e.g., words) in an utterance or sentence. or (b) syntax refers to the rules that allows one to organize elements (e.g., words) in an utterance or sentence. One way of viewing what (a) is saying is that we are providing a description of the patterns of elements in an utterance or a sentence and contrast this to random groupings. Information theory and associative chain theory (which would be consistent with a Skinnerian account of how sentences get formed) both attempted to provide such an account by looking at sequential probabilities for the co-occurance of words and whether there appeared to be empirically apparent contraints on word orderings. This may be a simple-minded way of thinking about syntax but this is a data-driven approach that attempts to synthesize the rules/regularities from actual utterances. Prior to Chomsky, this would have been referred to as a Bloomfieldian approach to syntax. Chomsky's view is represented by (b) and it is a theory-driven view, that is, syntax is defined as as a finite set of specific rules, comparable to the rules that were originally described for computational devices in automata theory. Chomsky's initial triumph was to show that phrase structure syntax and transformational grammar could account for a number of ENGLISH sentences. Through the 1960s it became clearer to everyone that Chomsky's standard theory was heavily biased toward accounting for patterns in English and not other languages, which is why Chomsky had to ultimately chuck the original theory in favor of government and binding and the recent minimalist program (which has such strange rules such as "move anything anywhere" -- is this the kind of rule one thinks of when thinking about syntax?). The unanswered question is why choose one theory of syntax over another? The problem with this approach is that it assumes that language require syntactic rules comparable to those used in automata theory and that other factors, such as semantic relationships, could not account for the structure of utterances and sentences. The development of generative semantics (see chapters 5 and 6 in Randy Allen Harris' "The Linguistic Wars") showed that semantic relations could potentially account for such structure. The development of neural networks have shown that these can also learn the "rules" for producing utterances without explicit knowledge of syntactic rules. That is, associations among elements could be detected and this infor would then be represented in the weights of the network. Gary Marcus and others have looked at infant learning in this context and one can get a taste of this research from the following: http://www.sciencemag.org/cgi/content/full/284/5413/433f Ultimately, because we do not know what syntax actually is (after all it is inferred from the structure of utterances and sentences and many different rules can give rise to similar patterns), I believe that one is forced to endorse a particular theory of syntax in order to instantiate the types of rules one is referring to. And then one needs to explain why that theory's rules are used instead of a, say, Generalized Phrase Structure syntax, a generative semantics relationship framework, a neural network model, and so on. Do animals have syntax or syntax like that in human utterances? First we need to define what syntax is, preferably a syntax that can account for all human languages that have ever been or ever will be. But I think we are a long way from that point. >The issue is whether a given model of language captures ALL >the various aspects of human language. Syntax is a notable >aspect of human language, and it is the one that is notably >lacking (or at least questionable) in studies of "ape language." Again, when you say syntax, do mean that ape utterances have a pattern that match human utterances or that apes use syntactic rules similar or different from those used by humans? If the latter, what is the definitive evidence that the "syntactic rule" is "real" and not just a conjecture in context of a particular theory of syntax? >Nevertheless, they are obviously able to communicate to >some degree via other means, and they are smart enough to >mimic some simple syntactic phenomena through general >cognitive (as opposed to specific linguistic) abilities. Communication does not require a syntax, at least not as specified by automata theory. -Mike Palij New York University [EMAIL PROTECTED] >Regards, >Chris Green ---
