110 John Connolly
plified above is not intended to be overly prescriptive. The discourse ana-
lyst is expected to include what is relevant to his or her purposes. The
notation as exemplified above is intended to be fairly explicit, and in some
circumstances it would doubtless be possible to omit some of it as being
obvious, although this is less likely if the notation is being used in the con-
text of Computational FG. In any case, though, it is quite acceptable to
omit information that is not relevant to the task in hand, or alternatively to
add further information if necessary. For instance, if it desired to label acts
as ‘central’ or ‘subsidiary’, after the fashion of Kroon (1997), then this can
easily be done (by means of lines like ‘ENTER MAJOR ACT 301 ’ or ‘EN-
TER SUBSIDIARY ACT 302 ’, or by some suitable alternative). Another
point of clarification concerns the units within the interaction. It is assumed
that each member of the hierarchy (apart from the act) consists of a se-
quence of one or more units of the rank below. Thus, for instance, a move
can be composed of just a single act.
Let us now consider the coverage of the interpersonal level description.
The latter is constructed around the discourse interaction, revealing the hi-
erarchical structure of the latter, but it also accommodates statements about
attitudes, for example ‘Attitude(A 153 ) = surprise’. It incorporates broad-
scope parameters such as discourse topic, discourse tense and large-scale
illocutionary force, as well as recognising ILLA, ILLS and their values. In
addition, it accommodates both adjacency sequences (for instance ‘A 152 =
Elaboration(A 151 )’) and rhetorical relations (for example ‘A 152 = Elabora-
tion(A 151 )’).
It should be noted that the overall framework accommodates certain
discourse-related concepts that are not strongly associated with the FG tra-
dition. These include ideology (alluded to in the above example by the
statement ‘ideology = neutral’), as well as voice in Bakhtin’s sense of the
term, reflected in the statements ‘AV 1 = Ella’ in the contextual description
and ‘CONTAIN(A 154 , BLEND(PSp, AV 1 ))’ in the interpersonal level rep-
resentation. Following the example of authors in, for example, DRT and
Situation Theory, the referring elements are picked out (for instance by
means of the statement ‘REFER(x 251 , x 252 )’) in the interpersonal level de-
scription, which have corresponding entries at the contextual level (in this
case ‘Referents of: x 251 (JayN), x 252 (KayN)’). Note that the words in brack-
ets here are intended merely as glosses for predicates which would
normally be incorporated, instead, into the representational level and ex-
pression level description. All the other information in Devlin’s
characterisation of a speech act, as illustrated earlier, also finds a place in
one or other of the levels of description envisaged here.