This tier provides a measure of perceived finality of each intonation phrase (break index 3). At present this is a simple binary choice between ``final'' and ``not final''. A phrase which is judged as ``final'' will have at its edge a strong sense of disjuncture, stronger than that of a non-final intonation phrase boundary. This percept of finality should be marked in this tier with the ``final'' label, which is aligned with the break index label. Those phrases which the labeller perceives as non-final should have no marking.
The notion of ``finality'' is subjective by nature, and will depend on several acoustic and stylistic factors which, in combination, cue that a phrase is final. The labeller should take into consideration the following phenomena (and possibly others too) when assessing the finality of a phrase:
The labeller should listen the phrase in question and ask her/himself the simple question: ``Does the speaker sound done?''. If the waveform were to be cut immediately after that break, would it sound as if the speaker had finished her/his turn (or completed an information unit)? Undoubtedly the meaning of the words in the phrase will also play some role in making this judgement, but labellers should concentrate on the sound.
The utterance
akete
provides an example of
finality marking. Here it is the last intonation phrase /sita ni
okima'su/ `below-LOC put' which is marked with the finality label at
its right edge. This utterance also provides a good example of the
so-called stylized ``finality'' contour, which is often employed to
signal the end of a turn or unit (common in narrative or instructional
sequences). In this type of stylized contour, there is typically a H%
boundary tone at the edge of the phrase just before the final
predicate (note the H% on /akete/ `open up' here), followed by an
optional pause. The final phrase (i.e. predicate) is realized in a
very reduced pitch range. This particular combination of tone, pause,
and pitch range serves to cue the finality of an utterance.
Another example of finality marking is given in
ueshita
. This utterance shows that
fragments, and not just sentence-final phrases containing verbs, can
also carry the percept of finality. At three points (aside from the
actual end of the utterance) the speaker uses cues such as lowering,
lengthening, etc. to signal finality.
Labellers should keep in mind that utterance final intonation phrases
are not always marked with a final label. In the utterance
shikakui
, the final (and only)
intonation phrase is not marked with a ``final'' label --- as the
listener will notice, this phrase was cut out of a larger utterance,
and thus lacks the finality which is characteristic of end-of-turn
phrases.
Sites which choose not to include this a finality tier in the J_ToBI transcription may mark the finality of intonation phrases by a break index 4 on the break index tier. This is essentially equivalent to a BI 3 marking on the break index tier and ``final'' label on the finality tier. However, we recommend that a separate finality tier be used. We anticipate that marking in this tier will be modified and further developed by sites whose focus is on the various degrees of finality and relationship with discourse structure.