52 Matthew P. Anstey
principle and exists irrespective of any psychological theory. Dik notes fur-
thermore that LIPOC is explanatory with respect to linguistic data, but
what would explain LIPOC? He writes that “[w]e expect that such [ex-
planatory] principles can be found in psychological mechanisms bearing on
the perceptibility of constituents and on the human capacity for processing
complex information” (Dik 1978a: 212). Note that we do not start with a
psychological principle and then inject it into the FG model. By analogy,
the proof of the pudding with respect to the interpersonal layer in FG 4
(where the strongest claims for psychological adequacy are made) is
whether it accounts for linguistic data. Do languages show clear evidence
that discourse moves ‘affect’ illocutions, that illocutions ‘affect’ proposi-
tions, and so forth? If so, then irrespective of psychology, one can justify
representing this ‘chain of command’, as it were, iconically in the linguistic
model. Psychologists may have something to say about such phenomena,
or they may not. In other words, I suggest that the way Dik used psychol-
ogy to explain an already observed fact such as LIPOC, P1, and so forth,
should be applied to layering.
Secondly, an inverse situation obtains when the linguist encounters in-
disputable psycholinguistic phenomena, such as Hengeveld’s mention of
the ‘intention to articulation’ processing pathway. Does this justify iconic
representation in the model? Not necessarily, because the model is imitat-
ing linguistic phenomena and not psychological. A case in point is the
‘chronology of composition’ problem presented in example (27) above.
The so-called ‘time course’ of speech production constitutes a sine qua non
for psycholinguistic grammatical models (Van Turennout 1997) but is to-
tally absent from almost every current linguistic theory. Similarly, with
respect to the three-layered structure, I do not think that Hengeveld can jus-
tify this division on evidence from psycholinguistic research, but only on
linguistic grounds.
So then, what do we do with such research? It provides a conversation
partner to engage with, one whose views may open up new approaches to
knotty problems. Sometimes this partner may be called upon ‘to give evi-
dence’ in a disputed linguistic case, but only alongside the witnesses of
language. It is collaborative, not authoritative (cf. Berg 1998; Boland
1999).
Thirdly, the very concept of psychological adequacy itself, as applied to
linguistics, involves a confusion of categories. It is like saying that a theory
of psychology should attain linguistic adequacy – one is simply talking
about two non-contiguous academic disciplines, which nevertheless, like
physics and biology, are in many ways deeply intertwined. The mistake is