This article explores the interconnection between grammar and the
performance of preferred and dispreferred responses in Japanese. As is
well known, dispreferred format turns are structurally more complex than
preferred format turns, regularly delayed, accompanied by prefaces and
accounts, mitigated, or made indirect. Owing to the flexibility of
Japanese grammar, participants have expanded intra-turn capacity to
maximize or minimize compliance with such formats. On one extreme, a
dispreferred action can be massively delayed until near the turn-ending
through opting for so-called canonical predicate-final word order and
minimization of ellipsis. On the other extreme, a preferred action can be
expedited to the very opening of a turn through non-canonical
predicate-initial word order by taking advantage of word order variability
and ellipsis. Such syntactic practices are interactionally managed for
calibrating the timing of social action. It emerges that the canonical
word order – assumed to be the generically unmarked alternative
– is actually optimally tailored for the implementation of marked
(dispreferred) responses, as opposed to a non-canonical word order for
unmarked (preferred) responses, in the given sequential environment.