46 Matthew P. Anstey
7.1.2. Does it address objections raised against previous models?
Focusing on the problem of structure (PR1), we may concede that the ex-
pression level is an interesting addition to Hengeveld’s model, but his
examples are not clear as to the precise nature of this level. Hengeveld (this
volume: 7) characterizes it as “a representation of constituent structure”;
the expression rules fill this constituent structures in generating linguistic
expressions, since as Hengeveld observes, “the expression level ... is the
product of grammatical and phonological encoding” (this volume: 7)
The problem of structure involves deciding if there is any non-
functional structure in language, and if so, deciding where it belongs in
FG 4. From the perspective of speech production, the question is whether
the Speaker himself contributes constituent structures from a sort of ‘syn-
tacticon’ in the way he contributes lexemes from the lexicon (and
functional primitives from the ‘semanticon’). Hengeveld hints at such a
process when he writes, “Top-down decisions ... concern the decisions the
speaker takes with respect to (i) the semantic content ..., and (ii) the ex-
pression category necessary to successfully transmit his communicative
intentions” (this volume: 10) Yet consider his example (2), here renum-
bered as (22):
(22) Damn!
(A 1 : [EXPR (P 1 )Sp (P 2 )Addr (C 1 ) ] (A 1 ))
-----
(Lex 1 )The choice of expression category, in this case EXPR, is not a matter of
constituent structure but an abstract illocutionary type. This presumably re-
sults in an isomorphic function-to-form correspondence between the type
and its (grammatical and prosodic) expression (cf. Van Buuren 1985; Lied-
tke 1998). Similarly, in footnote 2, Hengeveld (this volume: 18) lists
among the “encoding possibilities” of the illocution “prosodic encoding,
morphological encoding, and conventionalized lexicalization patterns”. So
although the expression level is clearly a representation of constituent
structure, it seems that, as in previous FG models, only the expression rules
manipulate such structure – the Speaker can only influence the structure
indirectly by choosing semantic and lexical primitives.
On this interpretation of FG 4 we would have to say that it does not re-
solve the problem of structure,^38 but the absence of detailed examples