74 Carlos Inchaurralde
tional levels are linked to the expression level by means of expression
rules. Obviously, the influence of the cognitive component on any of the
levels should have an observable result on the structures produced, and the
information that can have such an influence may be of many different
kinds. If we assume, as Hengeveld claims, that it “represents the (long-
term) knowledge of the speaker, such as his communicative competence,
his knowledge of the world, and his linguistic competence” (this volume:
3), it is important to bear in mind that we are dealing with many different
types of information that may even have different psychological reali-
zations.
In modular approaches such as Fodor’s, our linguistic competence, that
is, our capacity to create and interpret language, has many layers that work
independently of others: our use of syntax, phonetics, morphology is auto-
matic, with very little influence from central systems. The most important
properties of modular systems, according to Fodor (1983), are information
encapsulation (they perform the operations that they have been assigned
independently of other systems) and domain-specificity (they use specific
information that is not shared with other modular systems or central sys-
tems).^2 It is the autonomous, domain-specific nature of the different
‘modules’, which are specialized in different kinds of tasks, that distin-
guishes this conception from the holistic one, in which all mental processes
share mental resources.
In a manner parallel to this modularity of the mind, the architecture of
the FDG model assumes some degree of modularity for the different levels
in language production, but this structure is described in such a way that
there is also the possibility of having some influence from a general cogni-
tive component. Thus, this model allows some room for the claims of the
interactionist perspective, which assumes that there is interaction across all
mental processes, including the linguistic ones. This is thus an intermediate
point of view. But, in any case, the status of ‘linguistic competence’ (or
even that of ‘communicative competence’) will need to be very different
from that of ‘knowledge of the world’, which has a direct influence on the
choice and the interpretation of vocabulary and on the inferences that are
possible in discourse. For this kind of information, long-term knowledge
could easily be equated with long-term memory.
2.1. Long-term knowledge and encyclopaedic knowledge
There is considerable evidence that supports the well-known distinction be-
tween two types of memory: short-term and long-term (Loftus and Loftus