Towards a speaker model of FG 343new to the conversation. Such units typically take 1 to 2 seconds at most,
and are embedded between intermediate pauses (cf. Givón 1975: 202f.).
The most common unit is of the Topic-Comment type, prototypically ex-
pressed in Subject-Predicate form. Quite often however, utterance units do
not consist of the complete intended Move or Act, and may also be ‘in-
complete’ from the perspective of GrammarN.^15 This incompleteness may
be based on processing constraints. It may also be caused by the fact that
part of the topical information is thought by the speaker to be redundant. In
the latter case, this ties in neatly with the discussion in Mackenzie (1998),
which is a plea for adapting FG such that it also produces holophrases like
those in (9) below.
(9) Tomorrow.
To Mila.
Liam too.
Five of these, please.
The green one, if possible.
Skating.
Indeed, many utterances that appear to be ‘incomplete’ from the per-
spective of a sentence grammar turn out to be complete from the
perspective of a discourse grammar. Speakers, it is assumed, only produce
‘complete’ sentences in order to create more cohesion in their discourse
when the speech situation calls for this. In such cases they opt for what
Hannay (1991) calls the Topic Mode. However, in informal discourse
speakers will try to limit themselves as often as possible to new, focal in-
formation, thus producing chunks that need only the P1 (clause-initial)
position of the traditional FG template for their expression. The latter is
above all the case in the Reaction Mode, where speakers are frequently tak-
ing turns in the conversation. Such chunks may fit into working memory in
their entirety, which makes for ease of processing in the highly interactive
and competitive set-up of a face-to-face conversation. On the other hand, it
is precisely the size of working memory which puts constraints on what
can be maximally dealt with during speech production in the first place. So,
in the case of weightier Moves and Acts, spoken language typically comes
in Sub-Acts and Sub-Clauses, or even parts of these, i.e. mere fragments of
GrammarN complete clauses. Quite frequently, in spontaneous speech con-
secutive Sub-Clauses do not even combine to form GrammarN-complete
and well-formed ones. We think this is also a strong indication of ‘incom-
pleteness’ at the higher levels of linguistic planning, of the role of feedback