True, there may be two language-processing systems, lexicon and syntax. However, could we not say more than that they are computationally and linguistically distinct? Where are they in the brain, why are they where they are, and how can their distinctness and functional properties be explained by biological principles? A brain model of language is necessary to answer these questions. One view is that two different types of corticocortical connections are most important for storing rules and their exceptions: short-range connections within the perisylvian language cortex and long-range connections between this region and other areas. Probabilities of neuroanatomical connections plus associative learning principles explain why different connection bundles specialize in rule storage versus exception learning. Linguistic issues related to language change and plural formation in German are addressed in closing.

Clahsen's compelling evidence for the dual-mechanism model of the lexicon derives in part from the use of cross-linguistic data in psycholinguistic research. This approach reflects a growing (and positive) trend toward incorporating data from several languages when analyzing and modeling human language behavior. This perspective should be expanded to include data from typologically distinct languages to develop more explanatory models of language.

Clahsen's claim to contribute arguments for dual mechanisms based on rule analysis and against connectionist proposals is refuted. Both types of modeling are inadequate for principled reasons.

Results from our laboratories show clear frequency effects for regularly inflected forms in both German and English. Moreover, there does not seem to be sufficient evidence to treat the -s plural as the default plural in German. Together, these data do not support a dual modular structure of the language faculty.

We report an experiment that unambiguously shows an effect of full-form frequency for fully regular Dutch inflected verbs falling into Clahsen's “default” category, negating Clahsen's claim that regular complex words in the default category are not stored.

I argue for an extension of Clahsen's psycholinguistic paradigm to well-known languages with more complex morphological systems. This would help to address conceptual questions such as the nature of defaults and the way in which syncretisms are coded in the brain.

The German facts are consistent with the hypothesis that the default is the most frequent allomorph. Plural -s is the least frequent allomorph and does not act as a default. There is another way to measure the frequency of perfects in which no single -n allomorph is as frequent as -t. Lexical versus computational components do not correlate with regularity.

Clahsen has presented an impressive range of psycholinguistic data from German regular and irregular inflection to support the view that lexical memory and the combinatorial operations of grammar are subserved by distinct mental mechanisms. Most of the data are convincing and important. I particularly applaud Clahsen's effort to extend this lexical/grammatical dichotomy from mind to brain. Here I discuss some problems with the evidence presented by Clahsen in support of a neural lexical/grammatical dichotomy, and offer some additional evidence to reinforce this neural distinction.

Clahsen's theory raises problems that make it seem untenable. As an alternative, a constructivist neural network model is reported that develops a modular architecture and in which a single associative mechanism produces all inflections, displaying an emergent dissociation between regular and irregular verbs. Thus, Clahsen's rejection of associative models of inflection concerns only a subgroup of these models.

This commentary concentrates on the nature of irregularity in morphology. What is called “irregular” in the target article by Clahsen is not a homogeneous class. Rather, there are areas of strong subregularities in the domains both of German participle formation and of German plural information that need to be distinguished from the irregular domain.

German noun plurals not ending in -s are not as irregular as Clahsen suggests. Feminine nouns get the -n plural, unless they umlaut and are subject to a constraint that requires a reduced final syllable in the plural. Another regular class is masculine nouns ending in schwa, which are weakly inflected. It is suggested that more differentiated psycholinguistic experiments can identify these regularities.

Dual-mechanism models of language maintain a distinction between a lexicon and a computational system of linguistic rules. In his target article, Clahsen provides support for such a distinction, presenting evidence from German inflections. He argues for a structured lexicon, going beyond the strict lexicon versus rules dichotomy. We agree with the author in assuming a dual mechanism; however, we argue that a next step must be taken, going beyond the notion of the computational system as specific rules applying to a linguistic domain. By assuming a richer lexicon, the computational system can be conceived as a more general binding process that applies to different linguistic levels: syntax, morphology, reading, and spelling.

The following discussion aims to illuminate further the way in which morphologically complex words are represented in the mental lexicon. It is argued that the dual-mechanism model can accommodate the linguistic and psycholinguistic evidence currently available, not only on German inflection (as pointed out in the target article) but also on other languages (as presented in several commentaries). Associative single-mechanism models of inflection, on the other hand, provide only partial accounts.