The question of discourse representation 111How do the present proposals compare with those of Hengeveld (this
volume)? On the one hand, Hengeveld’s treatment of the interpersonal
level of discourse is more concise, and will no doubt suffice for many cir-
cumstances in which discourse description is not the primary topic of
interest. On the other hand, the present proposals are more appropriate
when a detailed treatment of discourse is required, particularly since they
include phenomena such as adjacency relations, which are omitted from
Hengeveld’s (admittedly non-exhaustive) presentation.
There is another significant difference between the two sets of propos-
als, which concerns the recognition or otherwise of structures such as
paragraphs above the sentence layer. The stance tentatively taken here is
that paragraphs can be seen as the expression-level counterpart of moves.
Similarly, chapters and sections can be seen as the expression-level foot-
prints of interpersonal-level units; thus a chapter expresses a transaction
and a section expresses what we may term a sub-transaction. In this way,
the hierarchical structure of discourse can be simply attributed to the inter-
personal level, in which case its expression at lower levels is purely
derivative, and therefore does not really need to figure in the lower levels
of the linguistic description, where it adds nothing fresh. This stance ac-
cords with Verschueren's (1999: 130) view that there are no purely
structural units above the sentence. It also simplifies the description, in
dispensing with what Dik calls the organisational level. In these respects,
then, it seems a preferable alternative to other approaches. However, to re-
peat, this proposal is tentative and will need to be examined more fully on
another occasion. If it should prove to be untenable, then there would be no
inherent difficulty in incorporating the higher-ranking structures into the
expression level description.
Finally, we need to consider how the four levels of description all fit to-
gether. Ideally, it would be satisfying to set the discourse description out in
four parallel columns. However, limitations of space may well prevent that,
in which case we may set out the contextual and interpersonal levels along-
side each other as in Table 2, and then deal with the other two levels as in
Table 3. The first column of Table 3 allows for a cross-reference to each
act, which is dealt with in more detail in Table 2. In the second column of
Table 3 we place the representational-level description of the propositional
content of the act concerned, and in the third column its expression-level
description.