"Relational Morphology in the Parallel Architecture”.
Ray Jackendoff, Tufts University and MIT
Abstract: The Parallel Architecture (Jackendoff 2002) is a grammatical framework in which phonology, syntax, and semantics are independent generative systems that are correlated by means of interface links. Words, which contain stored pieces of structure in all three domains, serve as interface links, and so do stored phrasal structures such as idioms and complex constructions. Relational Morphology extends this analysis to the internal structure of words.
However, morphology is distinguished by the existence of numerous patterns that do not achieve productivity. RM argues that these patterns are captured not by derivation, but by establishing relational links among items in the lexicon. For instance, the word piggish has a relational link between its stem and the word pig, as well as a relational link between the suffix -ish and a schema that expresses what all the -ish words have in common in phonology, syntax, and semantics. The resulting formalism has the flexibility to account for a wide range of morphological patterns that cannot be described in derivational terms.
The most significant outcome of this approach is its treatment of productivity. Most morphological patterns (at least in English) are nonproductive, and therefore their instances must be stored in memory. In contrast, instances of productive patterns such as English plural -s can be constructed on the fly by what are usually thought of as generative rules. Nevertheless, some instances of such patterns must be stored, e.g. raining cats and dogs, best regards, odds and ends. These have to be related to the -s plural schema by relational links in the lexicon. RM concludes that all schemas can be used in this relational role, but that only a subset of them can also be used generatively. This outcome upends the traditional view in which grammar is fundamentally productive, while nonproductive patterns are of little interest. Rather, nonproductive patterns are the central case, and productivity is to be thought of as a special add-on. I will suggest that this radical realignment of the grammar is not confined to morphology: it may well pertain to syntax as well.
Time permitting, I will offer some speculation on how this configuration might apply to other cognitive faculties such as knowledge of music, knowledge of geography, and social cognition. If there is anything to this speculation, it presents an opportunity for linguistic theory to teach us something new and unexpected about the mind in general.