Of the four speakers investigated in the previous experiment, three had participated in experiment two, producing rhythmic speech. In the previous experiment the P-centres of their utterances were compared to these produced intervals, to establish that differences in P-centre explained the timing variations.
There were four other speakers in experiment two; to extend the information on P-centre differences across speakers, and to relate differences in produced timing to P-centres. P-centres were calculated for examples of "ones" and "twos" for each of these four speakers.
6.11 Method II
The procedure described above in Method I resulted in P-centre values for all eight stimuli. P-centres were also calculated for a further eight stimuli - the "ones" and "twos" from SM DG WC and SR gathered in Experiment 2b, that is
Chapter 6 - Experiment 3
all the other speech items not used in method I. Oscillograms of these speech items are shown in Figure 6.7. A reduced version of a dynamic rhythm setting task was used for these speech items, in order to gather further data.
6.11.1 Design II
The speech pairs of "one" and "two"s from SM DG WC and SR were set against each other and against the common reference sound in four dynamic rhythm setting tasks - a separate set of experimental blocks for each pair of stimuli. In each set, there were thus 3 stimuli and 9 experimental blocks of trials.
The subjects performed five trials in each block, seven blocks in each rhythm setting experiment (two blocks having been carried out in experiment 2), for four speakers’ speech. The reference sound - reference sound settings were always identical and were therefore only carried out in two of the dynamic rhythm setting tasks to reduce the duration of the experiment. For each speaker" speech there were 9 sets of trials. All of the (one - two) and (two - one) pairings had been set in a dynamic rhythm setting task in the perception part of experiment 2; these were not therefore repeated; the data gathered in experiment 2 was used again. This was possible because the same three subjects who performed experiment 2 were used in this experiment, and the data gathering was contemporaneous.
The A - A interval was 1250ms, and the initial A - B interval was randomized.
The data from these dynamic rhythm setting tasks was treated separately from that data collected in Method I, due to the differences in the way that the P- centres of the stimuli were determined.
II' '! 111 1!M. ;1 j'1: " „ i i i l i ' l l i ' ' ' ' ' » I 1, , i . • M 1 1 r 1 ) u I ! ! I U 1 1 , mw
Figure 6.7 Oscillograms of examples of "ones" and "twos" from speakers SR, WC,
Chapter 6 - Experiment 3
6.11.2 Subjects
Three subjects completed these experiments. Three of these were from the four subjects who participated in Method I.
6.11.3 Analysis
As described in the Design II section, the P-centres for these four speakers were not based upon the 5X5 matrix of trials as was done in the previous experiment. Instead reduced P-centre setting experiments were used. Each speaker was treated separately; their "one" "two" and the reference sound was used in rhythm setting task - a 3X3 matrix of stimuli. The intervals set in this experiment, plus the (one - two) and (two - one) interval from experiment 2, were used to calculate P-centres for each of the speech samples. The P- centres for each speaker have been determined only with respect to speech from that speaker and the reference sound; this reduces the degrees of freedom of the P-centre values.
6.12 Results II
The tables below show the mean final intervals for each of the rhythm setting tasks for each of the speakers. The shaded cells are the conditions where the same sound is set against itself (and the mean interval should thus be equal to [A - A]*0.5, that is 625ms).
In each of the four Tables 6.9, 6.10, 6.11, 6.12, the stimuli combinations where the same sound is set against itself (the shaded cells) show final intervals that are similar to 625ms. This indicates that the subject set intervals of near physical isochrony when setting identical sounds to a rhythm. Elsewhere in the
tables large deviations from isochrony can be seen, which implies that the subjects did not set these stimuli to physical isochronous rhythms in order to acheive perceptual isochrony. This indicates that the stimuli vary in their P- centres.
"one" "two" ref
"one" 624.94 611.76 674.24 (9.48) (11.00) (25.27) "two" 635.83 627.50 676.89 (17.96) (11.50) (28.30) ref 571.28 579.82 624.22 (18.08) (27.14) (9.72)
Table 6.3 of mean final intervals in rhythm setting task - speech from SR
"one" "two" ref
"one" 620.06 614.59 680.73 (10.47) (13.13) (16.99) "two" 633.00 616.88 673.33 1 (15.66) (10.61) (33.25) ref 570.82 581.89 (23.35) (26.73)
Table d.W of mean final interval settings in rhythm task
-
speaker WC"one" "two" ref
"one" ! 628.00 593.67 632.41 (13.27) (17.74) (19.21) "two" 657.56 626.78 645.62 (19.50) (14.90) 17.42 ref 614.67 597.94 623.78 (23.79) (20.31) (10.69)
Chapter 6 - Experiment 3
"one"
"two"
ref
"one"
619.71
612.29
659.29
(9.71)
(8.24)
(22.25)
"two"
639.06
624.4
655.24
(12.91)
(6.57)
(18.78)
ref
589.94
583.11
(16.37)
(19.86)
Table 6.1 Z o f mean final inten/al settings in rhythm setting task