Lundin & Allen (1962) trained Ss to identify piano tones presented 20 sec. apart (with intertrial silence) in the two octave range above middle C.Their procedure.
Improvement of absolute pitch naming MICHAEL TERMAN CO L UMB IA UN I V E R SITY !
Ab8traet Ss improved the ir accuracy in naming sine tones in an " absolute context" in which an int ertrial dist r active procedure was used to reduce "relative" cues . Thi s result does not support the popular assumption that absolute pitch is an innate and inviolable "gift. " Introduetlon Absolute pitch (AP) has been casually defined as the ability to name or produce a pitch without reference to any other pitch. Revesz (1954) has suggested that people with this ability judge pitch on the basis of "tone color" rather than "tone height." This subjective distinction defies behavioral ana lys is since it evades operational definition . It would thus seem necessary to define AP by experimental procedure, so that stimulus, response and intertrial variables are throughly spec ified. Lundin & Allen (1962) trained Ss to identify piano tones presented 20 sec . apa r t (with intertrial silence) in the two octave range above middle C. Their procedure was r efined in the present experiment, for it was noted that stimulus tones may be recalled a nd reproduced with high ac curacy after 20 sec . of intertrial silence, and thus used a s relative cues , weakening the "absolute" construct. Method Seven magnetic tapes were programmed for use with TMI-Grolier Min/Max II teaching machines which were modified so that a three step response-answer-corr ection sequence could be followed on ea ch training trial . Sine tone stimuli were used to reduce the complex cues a s sociated with musical tones. Each tape contained 25 randomized trials, comprising the two octave range of 25 equal-tempered pitches surrounding middle C (A=440.0 cps). Ss used a musical nomenclature which referred to pitch name and octave Table 1. 5E CUN DS :
TAPE :
10 loud
sweeps
a
~
Procedural Sequence for One Training Trial
----!.4 : soft
i
SUBJECT:list e n
5 : loud
isweep9 ~sweeps
(fade in ) i advance p roMACHIN E : g r am f o r step i (1) below i
height. Individual tapes were not presented cons ecutively, but were repeated three to five times in the cour se of the experiment. The procedure for one training trial, constituting an operational definition of absolute pitch naming, is described in Table 1. A di st ractive procedure consisting of silence, random sweeps of loud and soft sine tones (similar to a fire-siren pattern), and vocalization by the S was employed to minimize intertrial cues. Eleven adult Ss , who had various musical ba ckgrounds but did not claim to "have" AP, were run in weekly I-hr. sessions, two training tapes per session. Re8ult8 and DI8eu881on In order to evaluate the judgments, criteria for "correct" and "error" responses were specified. As emphasized by Graham & Ratoosh(1962), the defined response continuum (here, the musical scale) need not reflect the physical dimensions of the stimulus continuum. Occurrence of pitch names as a function of individual stimulus frequencies did vary widely. Sample distributions are shown in Fig. I, with appropriate pitch names indicated in quotes. Since the distributions were all modal on the defined stimulus frequencies, and tended to decrease monotonically with increasing distance, accuracy scores were based on the deviation of pitch names in musical halfsteps from the defined stimulus frequencies. (The distance between neighboring tone s on the stimulus continuum is defined by music theory as a halfstep.) A general indication of group accuracy may be seen in Fig. I, where the modal pitch name for each stimulus frequency is plotted (solid line) . Pitch naming accuracy was highest in the extremes of the stimulus continuum, and a roughly V-shaped function connects the extremes, suggesting' that relative accuracy for naming the various stimuli was partially a function of a procedural artifact-the limited range of test frequencies.
10
isilence ;( a p p r o x .
llfade out )i
! ! !
! r e a d c ll isten ! aloud :
20 ~ 3 ~ ------'=-----b i stimulus i . b
!
i 92
i
: s 11 e n c e ~
dbSPLJi
i
l
:
!
!lis ten :
!listen :
'i advance program ! for steps (2 ) land (3) below
j{l ) write R ! (2) compare with an swer ( 3) wri te correct R
DIST RAC TIVE PRUCEDURE aRan dom sweeps of s ine tones (ap prox. 92 db e xtending ab ove a nd below SPL)' the test ra nge ( 1 30 . 8 - 52 3 .3 cps) . b Ta p e print-thro ug h was occasionally he a r d briefly before and after pre sentation of the stimulus tone. c To reduce possible intertrial covert vocal cues.
Psychon .
set., 1965.
Vol. 3
243
DEFINED STIMULUS FREQUENCIES
w
u ~80
0::: 0:::
=>60
u
u
°40
I-
Z
w20
u
0:::
w
o,
/30 .8
155 .6
185.0
220.0
261.6
311.1
370.0
440 .0
523.3
STIMULUS TONES (CPS) F ig . 1. P er cent occ urre nce of pitch names as a func tion of stimul us fr equency pr es ent ed . Group dat a based on th e l ast eig ht tra i ni ng tape s . Solid lin e connec ts modal pi tch name for eac h s tim ul us fr equency .
Sample ind ividual learning cur ves a re shown in Fi g. 2, based on four sc ore level s with the following cr iteria for " correct" r es pon se s: r e sponses naming only the defin ed sti mulus fre que ncy (A); and r espons es occurr ing within one (B), two (C) or thre e (D) halfsteps of the define d sti mulus frequency, plus octave errors . Octave errors r arely occurred afte r the fi rst few t raining sessions , but were included in scor e s B-D becaus e of Reve sz ts (1954) r eport th at they are common for peopl e who "have" AP. Accuracy sc or e s , as well a s rates of score increment, show gr eat difference s for each S. For exa mple , both SF and JM began training with s core A le vels of appro ximately 10-15% cor rect , but at th e conclusion of training J M scored 38% and SF scored 75%, mu ch close r to the fin al s core of SS (82%), who wa s the
overall high- s corer . Ten out of 11 Ss did show some scor e impr ove ment. Ther e wa s no clear r el ation between the exte nt of improvement and S's mu si cal exper ie nc e. Compa r ison of score level s A- D indicates th at th e gen eralization of pitch name s was mo st often close to the defin ed stimulus frequen cy. All Ss achieved sc or es of 84% cor rect or better on th e level of score D. In some cases a steady level of performance was r eached in sc ores C or D while scores A and B continued to rise , but extended training would have been necessary to a scertain th at a s ympto te s were, in fact, ac hieved. On th e anecdotal level, at le ast, the di stractive procedure was found to be ve r y effe ctive in se pa rating trials. Only SS r eported r emembering sti mulus ton es through th e di stractive procedure and using th em as relative cues .
Relereneell Gra ham, C . H .. & Ratoosh , P . Not es on some i nte rr el ations of senso ry psychol ogy , perception, and behavior. I n S . K och (Ed ), Ps y chol ogy . a s t udy oj a s cie nce. Vo l. 4 . New Y ork : McGr awHill , 196 2. Pp , 483-5 14 . Lundin , R. IV.. & Allen , J . D . A tec hni que for tra ining per f ect pitch . Ps uchol . Rcc ., 196 2, 12, 139-146. Rev esz , G. Ent roduc ti on t o th e psy chology o i mus ic. Norman : Un i versity of Ok lahoma Pr ess , 1954 .
Note' TRA INING TAPES (BLOCKS OF 4) Fig. 2. Mean per ce nt of co rrec t respons es for s uc cessive bl ock s of four tra i ni ng tap es . Sco re crite ria A- D des cr ib ed i n text.
244
1. N ow at B ro wn University . 1 t hank H . S . T err ac e and D. F . Johns on for t heir guidance , J . IV. Kling for hi s critical reading of th e manusc ri pt, and C . M. Harri s and V . us sache vskv f or laboratory fa ciliti es.
Psychon . Sci. , 1965, V ol. 3
