Perception & Psychophysics 1974. Vol. 15. v« 1. 21·25 Effect of different delayed auditory feedback intervals on a music performance task*

ANNE GATbT, JOHN L. BRADSHAW, and NORMAN C. NETTLETON Monash Unil'ersity , Clayton, Victoria, 3168, Australia

The effects of 12 different intervals were studied in a musical task involving performance on an electronic organ. Disruption was found to occur to a degree comparable to similar studies involving verbal and rhythmic tasks under OAF. Maximal disruption was found with a delay of 0.27 sec, a value rather greater than that typically found to be most disruptive in speech. Three of the 12 Ss showed speeded performance under part of the range of OAF intervals, as compared with performance under immediate feedback; however, their performance also reached a peak of slowing at a delay of about 0.27 sec. No significant differences were found between male and female Ss' performance. These findings were discussed without the context of control processes operating in music performance, and compared with the possibly analogous mechanisms of speech.

Speech performance is often greatly affected by and woodwind instruments), and mechanical problems delayed auditory feedback (DAF) (Yates, 1963). Similar posed by each specific instrument (reeds, mouthpieces, patterns of disruption have also been shown for single mutes, etc.). speech sounds, with resulting increase in intensity and The first application of the DAF technique to music length of each sound, as well as repetition of the single performance was reported by Lee (1951), the initiator speech sound (Chase, Harvey, Standfast, Rapin, & of the procedure in speech. A skilled tympanist Sutton, 1961; Fletcher & Yates, 1971); for key tapping, performed "erratically" and with "little or no rhythm." with increase in force and length of each tap, as well as No mention is made of pitch irregularities, and repetitions and grouping errors (Chase et al, 1959; disruption of rhythm was the only reported indicator of Karlovitch & Graham, 1967; Rapin, Costa, MandeL & erratic performance. Fromowitz, 1966); and for handclapping, with Havlicek (1968) has shown that on a musical repetitions (Kalmus, Denes, & Fry, 1955). Morse code, a sight-reading task, performances on representatives of rhythmic representation of speech, is also adversely each of the classifications of instruments, brass, affected by DAF (Lee, 1951: Yates, 1965). woodwind. string, and percussion (piano), were Music performance is a skilled task, in many ways significantly disrupted by DAF; the delay interval used comparable to speech. Our first contacts with both are was 0.12 sec. Bradshaw, Nettleton, and Geffen (1971) listening experiences: lullabies, nursery rhymes, etc. Just found a delay of 0.20 sec to be more disruptive to piano as we string words together into meaningful sentences, playing than the other delays tested in a pilot study so we experiment with music, trying different (0040, 0.55. 0.75, 1.10 sec). The same interval was also combinations of notes to make up melodies. Then we found to be disruptive in the playing of an electronic learn to "read" music-converting printed symbols into organ with DAF alone to one ear and immediate sounds, specific as to pitch and loudness, and at a certain auditory feedback (lAF) alone to the other ear (speed). Finally, we learn to "write" music, (Bradshaw, Nettleton, & Geffen, 1972). However, the putting into symbolic form our own musical ideas. use of the electronic organ in a situation analogous to Singing is the combination of both speech and music: speech under DAF, that is DAF and IAF together, has the singer must think not only about the words, but also not been studied. about the pitch, the intensity, the rhythm, and the Studies in speech DAF have examined the effects of tempo, dictated by the score (musical notation). Music several different DAF intervals; included among these performed on instruments, such as piano. viola, French are 10 intervals between 0.03 and 0.30 sec with peak at horn, bassoon, tympani, etc., includes not only all of the 0.18 sec (Black, 1951), four intervals between 0.085 and nonverbal aspects of singing, but also the intricacies of 0.340 sec with a peak at 0.17 sec (Butler & Galloway. bimanual coordination, awkward fingering, embouchure 1957), and four intervals between 0.18 and 0.80 sec (positioning of the mouth in forming the note in brass with a peak at 0.20 sec (Fairbanks, 1955). The present study was designed to investigate further the effects of *This research was supported by the Music and Psychology varying DAF intervals on a musical performance task, Departments of Monash University, and by an award from the Australian Research Grants Committee to the second author. and to ascertain the maximally disruptive DAF intervaL The authors wish to thank Terry Vincent for his assistance, particularly in the making of the audio am plifier and switching METHOD device. tRequests for reprints should be sent to Anne Gates. Subjects Psychology Department. Monash University. Clayton. Victoria. 3168, Australia. There were six female and six male paid adult volunteers. all

21 22 GATES, BRADSHAW AND NETTLETON

Table 1 Percent Effect of Different Feedback Delays on Playing Time Delay Intervals

,10 .13 .18 .20 .27 .35 .48 .55 .65 .75 .85 1.05 1 1.5 3.1 9.2 5.0 12.3 8.1 13.7 12.6 28.6 28.5 29.5 14.6 en'" 2 9.8 12.8 15.0 17.5 27.3 22.0 17.9 15.5 12.7 13.1 15.0 14.8 "i" 3 7.7 17.4 18.9 19.4 33.4 29.9 7.4 13.1 15.6 29.2 14.8 11.1 E 4 4.6 19.7 19.3 24.4 15.3 9.7 14.0 11.9 3.7 .? 3.7 15.1 9.6 5 9.4 15.9 23.5 12.0 23.2 21.5 14.0 12.2 8.1 9.3 13.4 15.2 6 3.3 4.2 9.4 28.5 26.9 30.6 32.7 18.6 15.5 10.2 11.5 21.8 1 -3.1 -1.9 -2.4 -7.0 -3.1 -2.9 -7.7 -7.8 -10.9 -2.5 -1.2 -1.4 2 -0.3 5.0 en'" 3.1 3.5 2.0 1.4 -0.4 -3.9 -7.1 -4.5 -1.2 9.1 3 5.1 14.3 21.1 . 41.6 29.0 50.9 65.3 52.1 33.6 42.2 23.4 14.9 "i" 4 13.3 10.4 12.4 9.5 8.5 4.1 1.2 3.3 0.0 6.7 5.1 8.5 :::e 5 6.6 14.0 17.0 20.3 18.6 15.5 2.0 0.1 5.9 5.8 8.3 7.2 6 -3.4 -2.0 0.7 3.4 7.5 -9.1 -8.7 -6.0 -3.9 -12.1 -13.8 -4.5

proficient in the playing of a keyboard instrument. None faster than 60 sec played it two or three times to get a sample claimed any hearing impairment. performance of about the same length of time as Ss who played the test piece more slowly than 60 sec. Two or three identical Apparatus pages of music were spread across the music stand for those Ss who played the test piece more than once. to obviate the Four tape recorders were used to give the 12 delays: a necessity for their counting the number of times they played the Tandberg 11 gave delays of' 0.10 sec at 7% ips and 0.20 sec at piece through. Ss were told that, in the different delay 3* ips; a Tandberg 64X gave delays of 0.13 sec at 7% ips and conditions, they would hear what they had played immediately 0.27 sec at 3% ips; a Revox A77gave delays of 0.18 sec at 7% ips through the speakers, and simultaneously, what they had just and 0.35 sec at 3% ips; and a Revox G36. modified by means of played, delayed, through the earphones. Ss were instructed to a around a roller between the record and playback playas quickly and as accurately as possible. They were told that heads. gave delays of 0.48, 0.55, 0.65, 0.75, and 0.85 sec at if they made any mistakes or slowed down, they were to keep 7'12 ips and 1.05 sec at 334 ips. All tape recorders were connected going or to try to go faster. to a 3-W audio with a switching device which gave Each 5 played two trials under each of the 12 delay intervals, output to Maico auraldomes fitted with TDH-39 . in an ascending-descending order, with DAF conditions Output from an electronic organ (National Model SX-88) was alternating with normal conditions. Half the Ss started with the channeled. to the tape recorders, and also to a Leak amplifier shortest delav interval; the others started with the loneest delay (Stereo 30), which was connected to Wharfedale speakers (Model interval. A stopwatch was used to record playing times. Thes~ Linton). A ftfth (Uher 4400) independently times were verified from the independent tape recording. recorded performances-of each S. Total reading time has previously been used as an indication In the normal condition, Sheard IAF from the speakers. In of disruption under speech DAF (see Yates, 1963). In this the DAF condition, Sheard IAF through the speakers and DAF context, playing time was used for the present experiment. It has through the headphones. Sound pressure levels of the delayed been shown that playing time reveals only a slightly greater signals through the headphones and immediate signals through performance decrement under DAF than does a combination of the speakers were equalized, using a Bruel and Kjaer precision playing time and error scores, "correct note rate" (Gates & sound level meter (Type 2203 with l-in. ). so that in Bradshaw!). the DAF conditions, the IAF was attenuated by approximately 15-20 dB by the headphones, thus making the DAF louder than the IAF by that amount. This condition is much like speech RESULTS under DAF, where S hears his voice almost immediately by bone- and air-conduction while receiving his speech, delayed and amplified, through the headphones. It was felt to be important The percentage effect of each delay interval is given in that the situation in this respect should closely mimic that of Table 1. These values were calculated as follows: Each speech. delay condition was compared with the normal condition played immediately before it and the normal Material condition played immediately afterward, and the Each S played the same piece, an adapted version of 1. S. differences expressed as a percentage of the mean of Bach's Minuet in G Major (BMV Anh. 114) from the those normal conditions, i.e., 100 X [(mean playing time Klavierbischlein filr Anna Magdalena Bach. 1722 an perial under delay - mean playing time without delayj/mean Edition No. 479, Allan and Co. Pty. Ltd., Melbourne). The playing time without delay] . right-hand part is mainly melodic (122 notes); the left-hand part (96 notes), although accompaniment, is also melodic. Depending upon the nature of their susceptibility to DAF, the Ss were divided into two groups: those whose Procedure performance was speeded up (three) and those whose performance was slowed (nine). In the latter group, the Each 5 first practiced the piece under normal conditions of magnitude of the effect ranged to a maximum of a 65% immediate, air-conducted feedback, until he could play it through without pausing to get the correct notes. Then increment in playing time. The mean effects of the successive performances of the test piece were timed, with S different experimental delay times on the two groups are playing as fast as possible. Those Ss who played the test piece given together with those of the total sample in Fig. 1. EFFECT OF AUDITORY FEEDBACK INTERVALS 23

25 ...... -..... 20 -..., -"-'- ...... 15 .. I­ -'-'-- ..---' U --.. W L.. L.. 10 W Fig. l. Percentage DAF effect overall and separately for those Ss who were slowed down or speeded up...... \ ,.. NO OAF EFFECT ~ 0 . 5? ... w • .... , Q. 5

10

10 13 III 20 27 35 '8 55 65 7585 lOS OCLAY INTERVALS (SECONDS)

Although the curve for those Ss whose performance was DISCUSSION speeded by DAF is located lower on the graph, its shape is very similar to the curve for the remaining Ss, whose These findings confirm that DAF is disruptive to a performances were slowed by DAF, with the peak at the musical performance task. The delay interval found to delay of 0.27 sec. cause greatest increase in playing time across all Ss was A division of the Ss according to sex revealed the 0.27 sec. Even for the group of Ss who otherwise differences shown in Fig. 2, with a peak disruption for showed some speeded performance under DAF, at this the male Ss at 0.20 sec and another at 0.35 sec; with the delay interval there was disruption and more slowing female Ss, there is a peak at 0.27 sec. An analysis of than for any other delay interval. While no significant variance showed that the main effect of sex was not sex differences were found in this study, Timmons significant, F(1,10) = 2.23: neither was the interaction (1971) has shown that in a comparable experiment of Sex by Delay, F(1I,IIO) =0040. However, the main measuring reading time under different DAF intervals, effect for delay proved to be significant, F(10,120) = females tended to adapt significantly more than males 2.0, p < .05. when the conditions were ordered. Mahaffey and

25 • .... -----~ 20 y. '~ ." , 15 • -----., '." , ~ 'Ii -... " l- "'. U 10 «,'. / ...... _- ...... --- --... W ~, u, .. -. -'- ...... , , u, .; , w I ..., , 5 , ..... , , ... effect for male , . , Fig. 2. Percentage DAF , , , and female Ss separately. ~ ...... - 0 • 0 ....z w o---

Stromsta (I965) showed differential reactions to the intervals of 0.10,0.20, and 0.30 sec, whereas syllables of different delay intervals. with maximal disruption at long duration were different from the immediate 0.27 sec for females and at 0.18 sec for males. feedback condition for only 0.20 and 0.30 sec. The The same kind of mistakes made in speech under OAF effect of reading rate under OAF was investigated by (Yates, 1963) have been shown to be made in music Kodman (1961), who found fewer errors with the slower under OAF (Havlicek, 1968). For example, Ss tend to rates, while Robinson (1972) showed that the delay adopt an increasingly slower tempo, to vary the tempo, producing maximal speech disruption was nearly equal to add extraneous material. to increase loudness, and to to the duration of the average spoken syllable. add heavy accents. Tempo variations and addition of Bradshaw- has suggested that the similar time constants extraneous material were noted in the present study, of syllable duration, of maximal disruption in speech although Ss were instructed to playas quickly and as under OAF, and under interaural switching may be due accurately as possible. It was hoped that any effects due to interruption of the sequencing of normal speech. to intensity changes would be eliminated by the use of There may be some musical analogue of the spoken the electronic organ, as no dynamic changes were syllable, or even the isolated phoneme, possibly the possible without a change in the amplitude or timbte single note, in that the information of a single note can controls, which were fixed for this experiment. only be regarded in the context of the melodic or Mahaffey and Strornsta (I965) found intensity to be a harmonic sequence. A post hoc comparison of critical significant factor in reaction to speech OAF, i.e., an OAF intervals for the Ss when divided into two groups, increase in intensity caused longer times. those who played overall faster compared with those Although errors 'were not specifically recorded and whose general performance was of a slower tempo, scored, notes were made of mistakes during the indicated no obvious difference; in fact, both groups performance and of the S's comments after the showed comparable variation in the effects of the performance. Some Ss tended to repeat notes, different delay intervals. Inspection of Table I suggests particularly where there was already one repeated note that, for at least half of the Ss, the critical OAF intervals in the music. Such perseveration has been reported for occur in two regions, around 0.20 or 0.37 sec and keytapping by Chase et al (1959) and for handclapping around 0.75 or 0.85 sec. This distribution may be in by Kalmus et al (1955), whose Ss exceeded the rhythmic some way related to the perceptual unit of music. An responses required under OAF; however, Fletcher and experiment studying the effects of different delay Yates (1971) found fewer repeated sounds under OAF. intervals up to 1.00 sec on a rhythmic keytapping task Specific comments by Ss may prove useful in showed an increase in function with no obvious peak evaluating the results. The one S who was consistently (Rapin, Costa, Mandel, & Fromowitz, 1963). The speeded by OAF said after one trial that "it was easy demand characteristics of a repetitive keytapping task because the beat came with the note." What he meant are likely to be somewhat different in kind from the was probably explained by another of the. Ss who was more complex sequencing required for the control of speeded by OAF, who said, after the experiment, that he speech and music. tried to pace his playing so that the delayed note Such sequencing for music must be investigated in coincided with the note he was playing. A similar terms of the perceptual unit of music to see whether it is technique may have been used, perhaps unintentionally, the delay interval which causes disruption to by those Ss, reported by Chase (1958), who repeated the performance or the interaction of the delay interval and speech sound [b] more times under OAF than under the music itself. IAF. There seemed to be three strategies adopted by Ss in playing under OAF: to play so fast that they tried to REFERENCES "get ahead" of the delay; to play more slowly or pause Black, J. W. The effect of delayed side-tone upon vocal rate and and wait for the delay to "catch up"; or to ignore the intensity. Journal of Speech & Hearing Disorders, 1951, 16, delay and play on. However, with the one exception of 56-60. the S who was always speeded up by OAF, all other Ss Black, J. W. Systematic research in experimental phonetics: 2. Signal reception: Intelligibility and side-tone. Journal of showed some relative slowing in their playing under the Speech & Hearing Disorders, 1954, 19, 140-146. three OAF conditions on or near 0.20 sec. Bradshaw, J. L., Nettleton, N. C., & Geffen, G. Ear differences The delay interval causing maximal disruption for and delayed auditory feedback: Effects on a speech and a speech has been shown to be about 0.20 sec, depending music task. Journal of Experimental Psychology, 1971, 91, 85-91. on the various delays tested (Black, 1951; Butler & Bradshaw, J. L., Nettleton, N. C., & Geffen, G. Ear asymmetry Galloway, 1957; Fairbanks, 1955). The question arises and delayed auditory feedback: Effects of task requirements as to the reasons for the longer critical delay for music and competitive stimulation. Journal of Experimental than for speech. Black (1951, 1954) suggested that the Psychology, 1972,94,269-275. syllable duration might be an important factor in the Butler, R. A., & Galloway, F. T. Factorial analysis of the delayed speech feedback phenomenon. Journal of the Acoustical effects of OAF on speech. Frankenberger (1968) found Society of America, 1957, 29, 632-635. that syllables of short duration were affected by OAF Chase, R. A. Effect of delayed auditory feedback on the EFFECT OF AUDITORY FEEDBACK INTERVALS 25

repetition of speech sounds. Journal of Speech Disorders, feedback as a function of frequency, intensity, time, and sex.) 1958.23,583-590. De Therapia Vocis et Loquelae, 1965,20.233-235. Cited by Chase. R. A., Harvey, S., Standfast, S.. Rapin, I.. & Sutton, S. B. A. Timmons, Sex as a factor influencing sensitivity to Comparison of the effects of DAF on speech and keytapping. delayed auditory feedback. Perceptual & Motor Skills, 1971, Science, 1959, 129.903-904. 32, 824-826. Chase, R. A.. Harvey, S., Standfast, S.. Rapin, I., & Sutton, S. Rapin, I., Costa, L. D., Mandel, I. J., & Fromowitz. A. J. Effect Studies on sensory feedback: I. Effect of delayed auditory of varying delays in auditory feedback on key tapping of feedback on speech and keytapping. Quarterly Journal of . Perceptual & Motor Skills, 1963, 16,489-500. Experimental Psychology, 1961, 13, 141-152. Rapin, I.. Costa, L. D., Mandel, I. J., & Fromowitz, A. J. Fairbanks. G. Selective vocal effects of delayed auditory Keytapping and delayed feedback. Journal of Speech & feedback. Journal of Speech & Hearing Disorders. 1955,20, Hearing Research, 1966,9,278-288. 333-345. Robinson, G. M. The delayed auditory feedback effect is a Fletcher. J. F., & Yates, A. J. The repetition of single speech function of speech rate. Journal of Experimental Psychology, sounds under delayed auditory feedback. British Journal of 1972,95,1-5. Psychology, 1971,62,73-79. Timmons, B. A. Sex as a factor influencing sensitivity to DAF. Frankenberger, R. L. Disruption of speech under delayed Perceptual & Motor Skills, 1971, 32, 824-826. auditory feedback: A function of syllable duration and Yates, A. J. Delayed auditory feedback. Psychological Bulletin, duration of delayed auditory feedback rates. Doctoral 1963,60,213-232. dissertation. Ohio University. 1968. Yates, A. J. Effects of delayed auditory feedback on Morse Havlicek. L. L. Effect of delayed auditory feedback on musical transmission by skilled operators. Journal of Experimental performance. Journal of Research in Music Education, 1968, Psychology, 1965,69,467-475. 16,308-318. Kalmus, H.. Denes, P.. & Fry. D. B. Effect of delayed acoustical feedback on some non-vocal activities. Nature. 1955, 175, 1078. NOTES Karlovich. R. S.. & Graham, 1. T. Visually and auditorily paced keytapping performance during synchronous, decreased, and 1. Gates, A., & Bradshaw, J. L. Interactions of ears, hands, and delayed auditory and visual feedback. Journal of Speech & cerebral hemispheres in a musical performance task requiring Hearine Research. 1967. 10.865-875. bimanual coordination. In preparation. Kodman.-F., Jr. Controlled reading rate under delayed speech 2. Bradshaw, J. L. Interaural switching and delayed auditory feedback. Journal of Auditory Reserach. 1961, 3. 186-193. feedback. In preparation. . Lee. B. S. Artificial stutter. Journal of Speech & Hearing Disorders. 1951, 16.53-55. (Received for publication May 23, 1973; Mahaffey. R. D., & Stromsta. C. P.