Commentary on James R. Hurford
Abstract: 55 words
Main Text: 1089
words
References: 186 words
Total Text: 1365
words
While the search for the neural basis of the language of thought is a laudable enterprise, and the article by Hurford a valiant first attempt, we argue that in investigating the argument structure of natural language it will ultimately prove more fruitful to consider the restrictions forced on the system by its inherently syntactic character.
The success of this kind of project is devoutly to be wished, though, at variance with the Author's own characterization of it as "reductionist", we prefer to characterize it as constructively "translationist". In this brief commentary, we intend to explain why, in our opinion, it is only partially successful, and what would constitute success for a better project.
We start by focusing on what is arguably the central problem (CP) in the study of arguments and argument structure: the paucity of their number and the rigidity of their character across the languages of the world. This datum stands in stark contrast with the myriad properties and relations that may be relevant to the most ordinary commerce of living creatures with the surrounding world. We may well expect a living creature to entertain judgments concerning an action or an object that are sensitive to properties such as: edibility, appeal, danger, source, rarity, risk, ease, just to name a few. Yet, the number of thematic roles that may be associated with any single predicate is severely restricted: at most three, possibly four. All other judgments can easily be supplemented linguistically by means of adjunction or conjunction: on Tuesday, with alacrity, and it was made of bamboo, and the warrior handled it deftly.
The Author says: "In the view adopted here, a predicate corresponds ... to a judgment that a creature can make about an object". He duly acknowledges the existence of the (in our terminology) CP: "The simple clauses of human languages are constrained to a maximum of about 4 or 5 core arguments; indeed most clauses have fewer than this. Presumably this reflects the structure of the underlying mental propositions." The explanation offered is crucially in terms of pre-linguistic (perceptual, mnemonic, computational constraints, allegedly supported by neurobiological data).
In this paper, predicates are presented as characteristically derived from propositions. This is basically the Aristotelian conception, which does not withstand serious scrutiny. Hale and Keyser (2002), leading proponents of the "constructionalist" approach to argument structure, have explained the scarcity and the rigidity of arguments in terms of the very limited number of possible syntactic nodes projected by predicates (notably verbal predicates) in hierarchical phrasal trees, and the intrinsic syntactico-semantic value associated with these nodes. If this account is even approximately correct, the explanation of the CP lies in the development of the linguistic apparatus. The constraints are language-internal, hierarchical and structural, not perceptual, conceptual, mnemonic, computational or otherwise imposed by some extra-linguistic system. They indeed "reflect... the structure of the underlying mental propositions", though in a sense rather different from the one suggested by Hurford.
A different theory of argument structure, a "lexicalist" one (Jackendoff is rightly quoted by Hurford) lends itself more to the picture suggested in this paper. In a Jackendovian semantics, a lexical conceptual representation is composed of several irreducible conceptual elements (predicates), each of which may contribute one or more argument positions to the final lexeme. Jackendoff argues that these conceptual primitives owe their existence and character to extra-linguistic 3-D perceptual representations. However, Jackendoff is no closer to accounting for the CP than Hurford is. The construction of a lexical conceptual structure could in principle involve an arbitrary number of primitives, and hence an arbitrary number of arguments. The numerical limitation is not imposed by formal logic, where the expectation for n-ary predicates is that n could equal any natural number, any more than it follows from restrictions imposed by our perceptual system.
Concerning reference, Hurford says "Information about the relative location of the objects in a visual scene, or about the properties of those objects, represented in a perceiver's brain, has the same essential quality of 'aboutness'... that linguists and philosophers identify with the semantics of sentences. Those ... who have insisted that semantics is a relation between a language and the world, without mediation by a representing mind, have eliminated the essential middleman ... The vision researchers have got it more right, in speaking of the 'semantics' of neural representations, regardless of whether any linguistic utterance is involved. It is on the platform of such neural representations that language can be built".
The truth-functional semantics of natural languages, in the wake of Frege and Tarski, has eliminated the subjectivity of the particular speaker, but not the speaker's tacit knowledge of language, nor internal representations of expressions in the "language of thought" (Fodor and Lepore, 2002; Fodor, 1975; Fodor, 1987). All reference is always made to entities "under a description", (Chomsky, 1988), that is, via tokens of internal standard representations in Mentalese. Neural states or processes as such have no semantics. They co-vary nomologically and causally with events in the world. Only symbols (bona fide representations) can have a semantics, and representations are descriptions accessed internally by the subject. It's very important to determine how the neural apparatus constrains those representations, but the locus of semantics is in those representations, not in their supporting neuronal states and processes.
A sophisticated neuroscience can ascertain the nature of the "raw" perceptual inputs to the representational apparatus, and the constraints imposed on the relevant mental computations by the hardware that implements them. For instance, an intrinsic sensitivity to events as primary percepts (as foreshadowed by Lettvin, Maturana, McCulloch and Pitts, (Lettvin, Maturana, McCulloch & Pitts, 1959) and competently refined and updated here by Hurford) is surely evidence in favor of the central role that events play in the semantics of natural languages. (Higginbotham, 1985; Higginbotham, 1989; Schein, 1993) The split between neuronal pathways sensitive to locational and intrinsic properties of objects and events can indeed offer the ingredients of an explanation of the chasm between closed class and open class lexical items. These are, of course, promissory developments, only dimly foreshadowed by present-day neurobiology, in spite of its relentless and awesome progress.
In sum, argument structure is presently analyzed by two distinct theories, "lexicalism" (Jackendoff 2002, Pustejovsky (1995) and "constructionalism" (Hale and Keyser 2002, Borer (in prep), Goldberg (1995). Should the considerations developed by Hurford withstand a revision along the lines proposed here, lexicalism might win some points. In the end, we suspect that the contest between theories will be determined by internal coherence, and ability to predict and explain relevant linguistic phenomena. Yet, cogent evidence from the basic neurosciences may well play a role. We are not advocating reductionism, but rather a sensitivity to the results of a productive inter-translation between domains. Probably, even Frege and Russell would have welcomed the opportunity to test their semantic theories on a terrain that was still inconceivable in their own time.
Borer, Hagit. (in prep). Structuring Sense.
University of Southern Californa, Ms.
Chomsky, N. (Ed.). (1988).
Language and interpretation: philosophical reflections and empirical
inquiry.
Fodor, J., & E. Lepore. (2002). The Compositionality
Papers. Oxford, UK and New York: Oxford University Press.
Fodor,
J. A. (1975). The Language of Thought. New York: Thomas Y. Crowell.
Fodor, J. A. (1987). Psychosemantics. Cambridge, MA: Bradford Books/MIT
Press.
Goldberg, A. (1995). Constructions: A Construction Grammar
Approach to Argument Structure. Chicago: University of Chicago Press.
Hale, K. and S. J. Keyser. (2002). Prolegomenon to a theory of argument
structure. Cambridge, MA: MIT Press.
Higginbotham, J. T. (1985). On
semantics. L.I. 16(4), pp. 547-593.
Higginbotham, J. T. (1989).
Elucidations of meaning. Linguistic and Philosophy 12(3), pp. 465-517.
Jackendoff, R. (2002). Foundations of language. Oxford: Oxford
University Press.
Lettvin, J. Y., H. R. Maturana, W. S. McCulloch, &
W. H. Pitts. (1959). What the frog's eye tells the frog's brain. Proceedings
of the Institute of Radio Engineers 47, pp. 1940-1951. [Reprinted in
McCulloch, 1988]
McCulloch, W. S. (1988). Embodiments of Mind,
Cambridge MA, The MIT
Press
Pustejovsky J. (1995). The Generative
Lexicon. Cambridge, MA: MIT Press.
Schein, B. (1993). Plurals and
events. Cambridge, Mass.: MIT Press.