146 Advances in spoken discourse analysis
Predictive Predicted
I+ -
R/I + +
R- +
F- -In subsequent discussions, however, the notion of predictability has been
associated with the closely related concept of obligatory/optional. Making
the same association, we state the rules as follows:
(a) I is always obligatory and predictive;
(b) As I is predictive it must be followed by R/I or R;
(c) If R/I occurs it must be followed by R;
(d) F is always optional.
The above rules raise two problems for analysts. Firstly, the following typical
classroom exchange can no longer be coded in the ‘traditional’ manner:
Example 20 move e.s
T: What’s this? elicitin gI
P: A saw informin gR
T: Yes, it’s a saw acknowledging F
The third utterance in this exchange cannot be F, as F is by definition
always optional/unpredicted yet the acknowledging move of a teacher-
initiated Elicit exchange is obligatory/predicted in that pupils expect to
be told whether their answers are ‘right’ or not (Sinclair and Coulthard
1975:51).
We approach this problem through the following reasoning. Any Elicit
exchange which is initiated by a primary knower must consist of at least
three parts. This is consistent with Berry and with the Sinclair and Coulthard
(1975) notion of a Teacher-Elicit exchange. As the third part is obligatory
it must be predicted by the second part. The structure of the Elicit exchange
of Example (20) is therefore I R/I R, with the moves remaining the same
as in the Example. This correspondence between moves and elements of
exchange structure is indicated in the diagram on p. 141 above.
The second problem is that if R is predicted it cannot be optional. In
other words, the structure I (R) is a contradiction in terms. Unfortunately,
this suggests that Berry’s example, reproduced below, is an incomplete
exchange:
Example 21
Guide (conducting party round cathedral): Salisbury is the English cathed-
ral with the tallest spire
(Coulthard and Montgomery 1981:126)
Such one-part exchanges are frequently found, yet if I is predicting, the
exchange must be two-part. Neither is it possible to propose that I is not