Do the experimental instructions affect the settings made in rhythm setting tasks?

This was investigated by Whalen Cooper and Fowler (1989). They were concerned by comments that subjects might be making rhythm alignment judgements that were influenced by the other qualities of the stimuli, rather than the P-centres. The explicitly suggestion was that subjects were aligning the

offsets of the stimuli instead not the P-centres. To test this they expressly

asked subjects in a P-centre determining experiment to align either the P- centre, the onset, the vowel onset or the offset of speech sounds. They found that subjects could not perform the offset alignment condition at all, and that in the other conditions they were making settings which were not significantly different from P-centre settings for the same stimuli. In a second experiment they altered the vowel duration of the stimuli, to make alignment by offset easier, by providing more distinguishable duration cues. They found that this was still not a possible task to perform, while the onset and vowel onset settings were still comparable to the P-centre settings for the stimuli.

From this they concluded that P-centres are the only perceptual events at which syllables can be aligned to sound regular, and that this meant that the P-centre settings were not affected by the subjects making other alignments, or by the experimental instructions.

Regrettably, given the questions that this experiment could have posed for the whole experimental paradigm of determining P-centres using dynamic rhythm setting tasks, they used only one naive subject. Three of the four participants were the authors. This is acceptable if the experiment investigates effects on responses which are driven by the stimuli. It is not acceptable in situations where the experimental variable measures whether subjects can or cannot perform a task, in addition, their initial A-B interval was always the same rather than random, leading to the possibility that subjects responses became influenced by a knowledge of how far the knob should be turned (that is, become stereotypic).

These results appear to be in contradiction to those of Seton (1989) who found that subjects could attend to different aspects of a signal in a rhythm setting task, depending on the instructions. In his case he used non-speech sounds, comprising a noise/periodic excitation compound (the periodic excitation was added into 250ms of white noise). The position of the periodic excitation within the white noise was varied. It occurred at either 0, 50, 100 150 or 250ms delay from the noise onset. Dynamic rhythm setting experiments were run twice with all four stimuli, with four naive subjects. On the first occasion the subjects were asked to make their rhythmic judgments based on the noise portion of the stimuli; on the second set of trials they were asked to concentrate upon the periodic components. If the subjects could perform both tasks, then different results would be found in the two different instruction conditions. This was indeed the case. In the noise condition, there was no significant difference in the P-centres set for the four different stimuli. In the second condition the later the periodic component occurred in the noise, the later the P-centres set for the stimuli, by all the subjects. Unfortunately Seton does not give any results for what would be found if the subjects were not told to attend to either component specifically, but merely asked to set the sounds to a rhythm as is normally

Chapter 4 - Methodology

specified. Conclusions as to which component normally contributes more to the P-centre setting cannot therefore be drawn.

Reasons why his result differs so from that of Whalen et al can be considered. His use of naive subjects suggests that his results are more reliable; in the Whalen et al study the one naive subject does make different settings from the three experimenters. This cannot explain all the difference; the naive subject in Whalen et al did not show any signs of being able to perform the different settings. Another reason is Seton’s use of non-speech sounds; while these were ‘speech-like’, they may have made the task easier by providing a clearer acoustic marker than ‘vowel onset’. The noise and the periodic sections are acoustically much easier to separate than the segments of speech sounds. These stimuli may have provided clues that are more salient to naive listeners, to whom the instruction to attend to vowel onset is not necessarily a meaningful statement. Natural vowel onsets also display longer rise times than the 5ms of Seton’s periodic component, again making their detection in the acoustic waveform harder.

The question of whether subjects can attend to different aspects of the acoustic waveform, and use these to make rhythm setting judgments remains open. The general consistency of P-centre results in the literature (with a few exceptions) indicate, as Whalen Cooper and Fowler (1989) state, that the P-centre is the most perceptually salient event to use to make rhythmic judgments, and that in different experiment subjects are not using different acoustic cues. However Seton’s findings show that subjects can use different aspects of the signal to make their settings if so instructed.