Footnotes for Four tones and downtrend – –

1.  I would like to express my gratitude to J. Etung and N. Martin for their help in constructing the dataset and their willingness to participate in the recordings.
2.  This measurement was taken as close as possible to the end of the utterance but avoiding any possible ‘tail-off’ effect. In a great many cases a tail-off was observable as a very steep drop in F0 as phonation ceased, i.e. in the last 15 ms or so of te utterance, and was excluded from the measurement.
3.  Two sentences originally felt by my assistant to consist entirely of T4 later proved to have a T3 word in them. A second potential source of confounding influence on the results involving T4 is discussed below.
4.  Despite being restricted to the last 40–80 ms of the utterance, this drop in F0 cannot be attributed to a ‘tail-off’ effect (cf. fn. 1).
5.  Note, however, that Clements’ study was not a controlled experiment, but one based one a monologue. Neither is it clear precisely what is meant by downdrift; the usage in Clements’ paper seems to subsume different downtrends. It is therefore not immediately obvious that what is referred to as ‘downdrift’ in that paper is comparable to ‘declination’ in the present study.
6.  The inherent or ‘normal’ level of the tone is taken to be either the F0 value of the tone as it occurs earlier in the sentence or, where the structure of the sentences did not permit this comparison, the F0 value of the tone at a comparable place in the like-tone sentences. In cases where both comparisons were possible, there was agreement between the two.