Unlike in English intonation, where the use of a downstepped accent
(!H* or H+!H*, as opposed to H*) is a paradigmatic choice made by the
speaker, downstep in Japanese is completely predictable from the
lexical accent specification of the preceding phrase. An accentual
phrase (whether itself accented or not) will be downstepped if (1) the
preceding accentual phrase bears an accent, and (2) both phrases are
in the same intonation phrase. (The terms ``accentual phrase'' and
``intonation phrase'' are described in more detail below in the
sections on tones 3.2, 3.3.1, and sections on break
indices 4.3, 4.4.) Downstepping is seen, for example, in
the utterance 
sankaku
, in which the
second phrase /ya'ne no/ `roof-GEN' is downstepped relative to the
preceding accented phrase /sa'Nkaku no/ `triangle-GEN'. Since the
presence or absence of downstep is predictable from the information
given in the word tier (i.e. the lexical accentuation of words) and
the break index tier (i.e. the type of prosodic boundaries), there is
no need to mark downstep in the tone tier of J_ToBI.