The inheritance properties oftake to taskandwentcan be set up in this format too:
Hence, as many have noted (e.g. Lakoff 1987; Pinker 1999), the hierarchical relations among stored lexical items fall
under more general brain mechanisms involved in any sort of memory. This accounts for the fact that lexical memory
displays prototypicality effects(some members ofa category beingmorecentral and others being more peripheral)and
family resemblance effects (there being no single set of features shared by all members of a category and no
nonmembers). Such a connection permits a welcome reduction of the aspects of language that the theory needs to
clai mare special.^91
Now we can ask, just as we did with lexical redundancy rules, whether the inheritance links in semantic networks are
explicit in memory, or whether they are epiphenomenal consequences of the nature of neural instantiation of
memories. I leave this important issue open. My inclination is to think that inheritance hierarchies, like lexical
redundancies, are implicit—that there are no overt
LEXICAL STORAGE VS. ONLINE CONSTRUCTION 185
(^91) Some people (including Lakoff) go on to assert that this shows that there is nothing special about language. I hasten to disagree. Recallingthe discussion of special versus
general capacities in section 4.5, we can conclude only that there isless that is special about language.