What a reply can't be
We have identified so far, besides reactions that merely
reinforce the problematic aspect of the topic, three logical types of replies :
invalidation, banalization, antagonistic reaction. What we suggest is that
there is no other possibility[6]. The following
table summarizes the possible combinations topic / reply :
topic
reply
|
paradoxical
|
improbable
|
(un)desirable
|
reinforcement
|
|
|
|
invalidation
|
|
|
|
banalization
|
|
|
|
antagonistic
r.
|
|
|
|
All
replies have an effect on the logical context : they emphasize or conversely
reduce (or even suppress) what appears as problematic in the topic. We can make
a principle of this :
Second principle of logical relevance :
Every reply aims at strengthening the problematic
side of the topic, or conversely at diminishing or suppressing it.
A "
neutral " reply will be irrelevant.
This constraint dramatically limits what can be said at any
moment during the conversation. As the preceding table shows, the most
constrained case is the paradoxical mode. It requires an invalidation (i.e.
an explanation). We observed previously one temporary failure in this
process, and as predicted by the model, the reply was rejected as irrelevant.
This was in [ex_Goffman, p.6]. We see how A and C are dissatisfied with B1, B2,
B3, and B4 : they repeatedly ask the same question " Are you in sociology ? ",
considering that it has not be correctly answered. B's replies have indeed no
effect on the problematic clause : the fact that B read the books does not
invalidate any term in the problematic clause, nor can it be considered as a "
forgotten term ". B's repeated answer is therefore irrelevant, and it is only
with B5 that A and C eventually obtain an (indirect) invalidation of their
puzzle, and are thus satisfied. Notice again that current pragmatic theories
are unable to detect any lack of relevance here, and are therefore unable to
explain the strange behavior of A and C.
[6] We will slightly moderate this statement in
the next section.