The goal of this study is to reintegrate the theory of generative grammar into the cognitive sciences. Generative grammar was right to focus on the child's acquisition of language as its central problem, leading to the hypothesis of an innate Universal Grammar. However, generative grammar was mistaken in assuming that the syntactic component is the sole course of combinatoriality, and that everything else is “interpretive.” The proper approach is a parallel architecture, in which phonology, syntax, and semantics are autonomous generative systems linked by interface components. The parallel architecture leads to an integration within linguistics, and to a far better integration with the rest of cognitive neuroscience. It fits naturally into the larger architecture of the mind/brain and permits a properly mentalistic theory of semantics. It results in a view of linguistic performance in which the rules of grammar are directly involved in processing. Finally, it leads to a natural account of the incremental evolution of the language capacity.

I challenge Jackendoff's claim that semantics should not be paralyzed by a failure to solve Brentano's problem of intentionality. I argue that his account of semantics is in fact paralyzed because it fails to live up to his own standards of naturalization, has no account of falsity, and gives the wrong semantic objects for words and thoughts.

In this commentary I discuss the role of types of knowledge and conceptual structures in lexical representation, revealing the explanatory potential of frame-based knowledge. Although frame-based lexical semantics is not alien to the theoretical model outlined in Jackendoff's conceptual semantics, testing its relevance to the analysis of the lexical evidence presented in his book has been left out of consideration.

I reject Jackendoff's view of Universal Grammar as something that evolved biologically but applaud his integration of blackboard architectures. I thus recall the HEARSAY speech understanding system—the AI system that introduced the concept of “blackboard”—to provide another perspective on Jackendoff's architecture.

Jackendoff's major syntactic exemplar is deeply unrepresentative of most syntactic relations and operations. His treatment of language evolution is vulnerable to Occam's Razor, hypothesizing stages of dubious independence and unexplained adaptiveness, and effectively divorcing the evolution of language from other aspects of human evolution. In particular, it ignores connections between language and the massive discontinuities in human cognitive evolution.

Jackendoff's Foundations of Language: Brain, Meaning, Grammar, Evolution has many points of similarity with Skinner's analysis of verbal behavior, though the former emphasizes structure whereas the latter emphasizes function. The parallels are explored in the context of a selectionist account of behavior in general and of verbal behavior in particular. Part of the argument is that behavior drives evolution and therefore also drives brain organization. Another concerns itself with the nature of explanation. Recent experimental developments in behavior analysis are reviewed as potential contributions to an understanding of language that incorporates its functional as well as structural dimensions.

Brain activity data prove the existence of qualitatively different structures in the brain. However, the question is whether the human brain acts as linguists assume in their models. The modular architecture of grammar that has been claimed by many linguists raises some empirical questions. One of the main questions is whether the threefold abstract partition of language (into syntactic, phonological, and semantic domains) has distinct neural correlates.

Jackendoff's “mentalistic” semantics looks more radical than it is. It can best be understood as a necessary corrective to the traditional oversimplification that holds that psychological variation “cancels out” on the path from word to world. This reform parallels the “evo-devo” reform in evolutionary biology.

Recognizing limitations of the “syntactocentric” perspective, Jackendoff proposes a model in which phonology, syntax, and conceptual systems are each independently combinatorial. We can ask, however, whether he has taken this issue to its logical conclusion. The fundamental question that is not fully addressed is whether the combinatorial aspect of syntax originated in, and derives from, the indeed “far richer” conceptual system, a question to be discussed.

The theoretical debate in linguistics during 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 & Bever 2001, pp. 37–40). By those standards, Jackendoff is quite a party faithful (a Khrushchev or a Dubcek, rather than a Solzhenitsyn or a Sakharov) who questions some of the components of the dogma, yet stops far short of repudiating it.

Instead of commenting directly on Foundations of Language: Brain, Meaning, Grammar, Evolution, I provide some remarks from an interdisciplinary view. Language theory is examined from the perspective of the theory of complex systems. The gestural-vocal dichotomy, network theory, evolutionary mechanisms/algorithms, chaos theory, and constructive approach are briefly mentioned.

Jackendoff claims that current theories of generative grammar commit a “scientific mistake” by assuming that syntax is the sole source of linguistic organization (“syntactocentrism”). The claim is false, and furthermore, Jackendoff's solution to the alleged problem, the parallel architecture, creates a real problem that exists in no other theory of generative grammar.

Foundations of Language (Jackendoff 2002) sets out to reconcile generative accounts of language structure with psychological accounts of language processing. We argue that Jackendoff's “parallel architecture” is a particularly appropriate linguistic framework for the interactive alignment account of dialogue processing. It offers a helpful definition of linguistic levels of representation, it gives an interesting account of routine expressions, and it supports radical incrementality in processing.

In an attempt to provide a unified model of language-related mental processes, Jackendoff puts forward significant modifications to the generative architecture of the language faculty. While sympathetic to the overall objective of the book, my review points out that one aspect of the proposal – the status of the lexicon – lacks sufficient empirical support.

Based on neuropsychological evidence of nonverbal impairment accompanying aphasia, I propose that the neural prerequisites for language acquisition are shared with a range of nonverbal capacities. Their commonality concerns the ability to recognize a limited number of finite elements in manifold perceptual entities and to combine them for constructing manifold entities.

In this commentary, I concentrate upon Ray Jackendoff's view of the proper foundations for semantics within the context of generative grammar. Jackendoff (2002) favors a form of internalism that he calls “conceptualism.” I argue that a retreat from realism to conceptualism is not only unwarranted, but even self-defeating, in that the issues that prompt his view will inevitably reappear if the latter is adopted.

Jackendoff's criticisms of the current state of theorization in cognitive neuroscience are defused by recent work on the computational complementarity of the hippocampus and neocortex. Such considerations lead to a grounding of Jackendoff's processing model in the complementary methods of pattern analysis effected by independent component analysis (ICA) and principle component analysis (PCA).

Psychologism is the attempt to account for the necessary truths of mathematics in terms of contingent psychological facts. It is widely regarded as a fallacy. Jackendoff's view of reference and truth entails psychologism. Therefore, he needs to either provide a defense of the doctrine, or show that the charge doesn't apply.

Contra Jackendoff, we argue that within the parallel architecture framework, the generality of language does not require a rich conceptual structure. To show this, we put forward a delegation model of specialization. We find Jackendoff's alternative, the subdivision model, insufficiently supported. In particular, the computational consequences of his representational notion of modularity need to be clarified.

Jackendoff (2002) argues that grammatically relevant and irrelevant components of meaning do not occupy distinct levels of the semantic system. However, neuropsychological studies have found that the two components doubly dissociate in brain-damaged subjects, suggesting that they are in fact segregated. Neural regionalization of these multidimensional semantic subsystems might take place during language development.