Generative grammar with a human face?
Commentary on Jackendoff's Foundations of Language

Shimon Edelman
Department of Psychology
232 Uris Hall, Cornell University
Ithaca, NY 14853-7601, USA
http://kybele.psych.cornell.edu/~edelman

Abstract. The theoretical debate in linguistics over the past half-century bears an uncanny parallel to the politics of the (now defunct) Communist Bloc. The parallels are not so much in the revolutionary nature of Chomsky's ideas as in the Bolshevik manner of his takeover of linguistics [Koerner, 1994] and in the Trotskyist (``permanent revolution'') flavor of the subsequent development of the doctrine of transformational generative grammar (TGG) [Townsend and Bever, 2001], pp.37-40. By those standards, Jackendoff is quite a party faithful (a Khrushchev or a Dubcek rather than a Solzhenitzyn or a Sakharov) who questions some of the components of the dogma, yet stops far short of repudiating it.

In the Foundations of Language, Jackendoff (2002) offers his version of TGG, in which the primacy of syntax (``an important mistake'', p.107) is abolished, the related notions of Deep Structure and Logical Form (``the broken promise''; p.73) are set aside, the links to other domains of cognition are discussed, and a hand is extended in peace to psychologists and other cognitive scientists. FoL is an enjoyable, thought-provoking and useful book that fulfills the promise of its title by presenting - and attempting to tackle - foundational issues in linguistics. It is an excellent overview of the ground that must be covered by any serious contender for a linguistic ``theory of everything''. Its non-dogmatic style engages skeptical readers of cognitive and empiricist persuasions (``can my theory explain this set of facts better?'') instead of alienating them.

Among the more positive aspects of Jackendoff's stance in FoL are the emancipation of semantics as one of the three equal-status components of the ``parallel architecture'' (p.125), the realization that not all rules are fully productive (admitting constructions; p.189), and the construal of meaning as a system of conceptual structures (p.306). The pervasiveness of TGG dogma is, however, very prominent throughout the book. On the most abstract level, the dogma manifests itself in the bizarre mentalistic nomenclature (f-knowledge, etc.) that Jackendoff uses instead of the standard explanatory machinery of representation found in all cognitive sciences. Jackendoff shuns a representational account of linguistic knowledge because of his (understandable) wish to avoid joining Fodor and Searle in the philosophical quagmire of intentionality. There exist, however, psychophysically and neurobiologically plausible accounts of symbolic representation that hinge on counterfactual causality and manage to stay clear of the Fodorian mire [Edelman, 1999,Clark, 2000].

The preponderance of Chomskian bricks in the Foundations of Language is revealed in Jackendoff's official insistence, in the introductory chapters, on rule-based combinatoriality. The initial formulation of this concept (pp.38-57) is so strong as to be incompatible with his own views on constructions (pp.152-187) and on their graded entrenchment (p.189), expressed later in the book. It is satisfying to observe that those latter views are on a convergence course with some of the best-known and most promising work in cognitive linguistics [Langacker, 1987,Goldberg, 1998]. As such, they can stand on their own: computationally explicit construction-based accounts of linguistic productivity need no extra propping [Solan et al., 2003]. In any case, Jackendoff should not count on any help from TGG, a Protean theory that despite decades of effort failed to garner empirical support for the psychological reality of the processes and entities postulated by its successive versions, such as movement and traces [Edelman and Christiansen, 2003,Edelman, 2003]. In a recent attempt to obtain psycholinguistic evidence for traces, for example [Nakano et al., 2002], only 24 subjects out of the original 80 performed consistently with the predictions of a trace/movement theory, while 39 subjects exhibited the opposite behavior (the data from the rest of the subjects were discarded because their error rate was too high). Jackendoff's continuing clinging to TGG (complete with movement and traces), despite its empirical bankruptcy and despite his self-proclaimed openness to reform, is difficult to explain.

Even Jackendoff's highly commendable effort to treat semantics seriously may be undermined by his continuing commitment to TGG. Conceptualist semantics is an exciting idea, but to develop it fully one must listen to what cognitive psychologists have to say about the nature of concepts. Instead, Jackendoff erects his own theory of concepts around a scaffolding left by the generative linguists, which, in turn, is only as sound as those decades-old intuitions of Chomsky and Fodor. In particular, incorporating Marr's and Biederman's theories of visual structure (pp.346-347), themselves patterned on TGG-style syntax, into the foundations of semantics cannot be a good idea. Jackendoff's acknowledgment, in a footnote on p.347, that Marr is ``out of fashion'' with the vision community holds a key to a resolution of this issue: current perceptually grounded theories of visual [Edelman, 1999,Edelman, 2002] and other [Barsalou, 1999] concepts are a safe, additive-free alternative to TGG-style semantics.

In summary, Jackendoff's book is one of several recent manifestations in linguistics of the equivalent of the Prague Spring of 1968, when calls for putting a human face on Soviet-style ``socialism'' began to be heard (cf. the longing for ``linguistics with a human face'' expressed by [Werth, 1999], p.18). Jackendoff's stance, according to which the ``mistakes'' that were made do not invalidate the TGG framework, amounts to a bid to change the system from within. In a totalitarian political system, this may only work if the prime mover behind the change is at the very top of the power pyramid: Czechoslovakia's Dubcek in 1968 merely brought the Russian tanks to the streets of Prague, whereas Russia's Gorbachev in 1987 succeeded in dismantling the tyranny that had sent in the tanks. In generative linguistics, it may be too late for any further attempts to change the system from within, seeing that previous rounds of management-initiated reforms did little more than lead the field in circles [Edelman and Christiansen, 2003]. If so, transformational generative grammar, whose foundations Jackendoff ventures to repair, may have to follow the fate of the Communist Bloc to clear the way for real progress in understanding language and the brain.

Bibliography

Barsalou, 1999
Barsalou, L. W. (1999).
Perceptual symbol systems.
Behavioral and Brain Sciences, 22:577-660.

Clark, 2000
Clark, A. (2000).
A theory of sentience.
Oxford University Press, Oxford.

Edelman, 1999
Edelman, S. (1999).
Representation and recognition in vision.
MIT Press, Cambridge, MA.

Edelman, 2002
Edelman, S. (2002).
Constraining the neural representation of the visual world.
Trends in Cognitive Sciences, 6:125-131.

Edelman, 2003
Edelman, S. (2003).
Naturalizing linguistics.
-, -:-.
submitted.

Edelman and Christiansen, 2003
Edelman, S. and Christiansen, M. H. (2003).
How seriously should we take Minimalist syntax? A comment on Lasnik.
Trends in Cognitive Science, 7:60-61.

Goldberg, 1998
Goldberg, A. E. (1998).
Patterns of experience in patterns of language.
In Tomasello, M., editor, The new psychology of language, pages 203-219. Erlbaum, Mahwah, NJ.

Koerner, 1994
Koerner, E. F. K. (1994).
The anatomy of a revolution in the social sciences: Chomsky in 1962.
Dhumbadji!, 1(4):3-17.

Langacker, 1987
Langacker, R. W. (1987).
Foundations of cognitive grammar, volume I: theoretical prerequisites.
Stanford University Press, Stanford, CA.

Nakano et al., 2002
Nakano, Y., Felser, C., and Clahsen, H. (2002).
Antecedent priming at trace positions in Japanese long-distance scrambling.
Journal of Psycholinguistic Research, 31:531-570.

Solan et al., 2003
Solan, Z., Ruppin, E., Horn, D., and Edelman, S. (2003).
Unsupervised efficient learning and representation of language structure.
In Alterman, R. and Kirsh, D., editors, Proc. 25th Conference of the Cognitive Science Society, Hillsdale, NJ. Erlbaum.
in press.

Townsend and Bever, 2001
Townsend, D. J. and Bever, T. G. (2001).
Sentence comprehension.
MIT Press, Cambridge, MA.

Werth, 1999
Werth, P. (1999).
Text worlds: representing conceptual space in discourse.
Pearson Education Limited, Harlow, UK.

About this document ...

Generative grammar with a human face?
Commentary on Jackendoff's Foundations of Language

This document was generated using the LaTeX2HTML translator Version 99.2beta8 (1.43)

Copyright © 1993, 1994, 1995, 1996, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999, Ross Moore, Mathematics Department, Macquarie University, Sydney.

The command line arguments were:
latex2html -split 0 edelman.tex

The translation was initiated by Shimon Edelman on 2003-05-16


Shimon Edelman 2003-05-16