212 María de los Ángeles Gómez-González
Grammar (FPG) and Mackenzie’s (1998, 2000, this volume) Incremental
Functional Grammar (IFG), which draw closer attention to work in lan-
guage production and comprehension. These represent two alternative,
though compatible top-down attempts to accommodate grammatical de-
scription to the dynamic nature of discourse organization, conceived in
both programmes as an interactive, flexible process that unfolds in real
time in ‘chunks’ or in mentally digestible units. FPG claims that linguistic
structures emerge from abstract cognitive factors or conceptualizations
which are fundamentally different from linguistic (lexico-semantic) or-
ganization, yet are still strongly language-bound (based on predicate-
argument constructs). The whole process of language production is claimed
to be a matter of gradually singling out packets of information which get
coded in single utterances (rarely full clauses, especially in speech), using
mid-to-short-term memory to adapt long-term conceptual knowledge to the
actual communicative situation and to the process of language production.
And it is only the last part of this process, the single utterance, that gram-
mar actually deals with (Jackendoff 1987, 1990). IFG, in turn, linking up
with research into starting points (MacWhinney 1977), posits an incre-
mental or gradual left-to-right build-up of chunks of information
(utterances or discourse acts), from minimally a P1 in Focus position, the
first increment or subact (the holophrase) which exploits the primacy effect
or prominence of first mention, to rightward expansions or linear se-
quences of subacts, with late, possibly last placement of the Focus in
conformity with the recency effect, according to which information offered
last is best retained (cf. Gómez-González 2001). Concomitant intuitions are
found in Bakker and Siewierska’s (this volume) rendering of expression
rules as dynamic, and working top-down, depth-first, and left-to-right as in
actual language production.
Hengeveld’s Functional Discourse Grammar (FDG; this volume) seems
to accommodate the insights gleaned from the aforementioned broad-brush
accounts. With a view to constructing a new architecture for FG, FDG re-
veals itself as a top-down hierarchical modular approach, in which entities
of a purely pragmatic Interpersonal Level (IL) at the top (acts/moves of ref-
erence and ascription) and purely semantic entities of varying orders at a
Representational Level (RL) next down are mapped by expression rules
(ER) onto the Expression Level (EL) at the bottom, i.e. the phonological
string that eventually triggers articulation. Speakers – Hengeveld claims –
draw on the cognitive component, covering (long-term) knowledge including
communicative and linguistic competence, at each of the three levels, while
the communicative component, containing (short-term) linguistic co(n)textual