Alma Mater Studiorum University of Bologna, August 22-26 2006
The effect of pitch, tempo and proportional pitch and tempo manipulation on memory of familiar musical excerpts David Brennan
Catherine Stevens
MARCS Auditory Laboratories, School of Psychology, University of Western Sydney, Australia
[email protected]
MARCS Auditory Laboratories, School of Psychology, University of Western Sydney, Australia
manipulations of pitch or tempo there is no significant difference between musicians and nonmusicians. However, identification of the original excerpt was poorer for nonmusicians when the variant excerpt included the proportional manipulation of both pitch and tempo. These findings suggest that musicians use different memorisation strategies for musical information than those not reporting musical training and also suggests the perception of ratio relationship between the frequencies of tonal and infratonal auditory events.
ABSTRACT The aim of the study was to investigate participants’ ability to identify the correct pitch and tempo of musical events from long-term memory. Seventytwo participants listened to a series of comparison pairs of excerpts from well known television themes. Each comparison pair included the original excerpt and a variant of that same excerpt that had the pitch, tempo or pitch and tempo manipulated. It was hypothesised that variant excerpts whose pitch and tempo had been varied proportionally are more difficult to differentiate from the original than variant excerpts that have only their pitch or tempo manipulated. It was also hypothesised that participants reporting more than three years musical training (musicians) are more able to identify the origjnal pitch and tempo than participants not reporting musical training (nonmusicians). Results suggest that for the differentiation of individual
Keywords Pitch, Tempo, Long-term Memory, Infratone.
INTRODUCTION
In: M. Baroni, A. R. Addessi, R. Caterina, M. Costa (2006) Proceedings of the 9th International Conference on Music Perception & Cognition (ICMPC9), Bologna/Italy, August 22-26 2006.©2006 The Society for Music Perception & Cognition (SMPC) and European Society for the Cognitive Sciences of Music (ESCOM). Copyright of the content of an individual paper is held by the primary (first-named) author of that paper. All rights reserved. No paper from this proceedings may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information retrieval systems, without permission in writing from the paper's primary author. No other part of this proceedings may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information retrieval system, without permission in writing from SMPC and ESCOM.
ISBN 88-7395-155-4 © 2006 ICMPC
1771
Recently Schellenberg and Trehub (2003) and Nakata, Schellenberg and Trehub (2004) reported an experimental method that presented participants with a series of brief excerpts from popular television programs. Each excerpt was presented twice, once at its original pitch and once with the pitch of the excerpt, raised, or lowered, by one or two semitones. The purpose of their experiment was to investigate whether absolute pitch perception, at least in an implicit form, was more prevalent in the general population than was generally supposed. The experimental method avoided the difficulties associated with naming or physically reproducing a tone and opened the assessment of pitch difference more fully to musically naïve participants. The
ICMPC9 Proceedings
probably a dominant part, in the coding of both tonal and infratonal musical events in memory. An investigation of whether similar evidence can be found for the relational storage of tone/infratone frequency ratio relationship would therefore lend support to the perception of tone/infratone frequency ratio relationship.
experiment "demonstrated considerably greater accuracy in pitch memory" (Schellenberg & Trehub, 2003 p. 265) than had previously been reported. This method is also suited to investigation of the relational processing of certain musical parameters, such as the perception of frequency ratio relationship between tonal frequency and tempo (infratone) frequency (Cowell, 1930; Pound, 1934). If humans do perceive frequency ratio relationship between tonal frequencies and temporal (infratonal) frequencies (Brennan & Stevens, 2002) it seems probable this relationship would form a component of an individual’s memory of a musical excerpt. Consequently proportional manipulation of both tone and tempo frequency of a musical excerpt should render it more difficult to distinguish from one’s memory of the original than for an excerpt incorporating manipulation of only one of those parameters.
The use of the relational and nonrelational manipulation of already memorised musical excerpts as the stimulus basis for this experiment offers a number of advantages. The most important is that the absolute tone/tempo (infratone) frequency ratio relationship and its effect are not relevant to the experiment. All that is being examined is whether a changed tone/ tempo (infratone) frequency ratio relationship is easier to differentiate from the original than an excerpt in which the tone and tempo (infratone) frequencies are changed but their ratio relationship remains the same.
Opinion as to whether musical parameters are memorised in an absolute or relational form is divided (Attneave & Olsen, 1971; Narmour, 1977; Premack, 1978). Fewer researchers propose a solely absolute form of musical memory than do a solely relational form. This is not surprising as humans’ ability to identify tunes in a different key from the original would make a theory based only on absolutes problematic. Arguing for a relational form of storage, Attneave and Olsen (1971) proposed that long term memory is based only on pitch relations, or intervals. This view was also supported by Monahan (1993) who wrote that musical pitch and time are described most naturally in relative rather than absolute terms. Taking a pragmatic position, Narmour (1977) suggested that musical listening involves using both schematic reduction and unreducible idiostructural information and Hintzman (1986) concurred with this view when he proposed that musical memory encodes absolute features along with abstract relations. Premack (1978) introduced the concept of dynamic allocation of memory between absolute and relational forms when he suggested that abstraction is only induced as a response to overburdened memory.
Aim The aim of the experiment was to investigate whether participants find it more difficult to identify proportional transposition of both pitch and tempo than the independent transposition of pitch or tempo of memorised auditory events (television themes). The experiment also investigated whether musical training moderates the effect. The overall design consisted of a 3 x (2) mixed factorial. Independent Variable 1 (IV1): Levels of transposition The first independent (within subjects) variable was level of transposition (IV1). Three levels of transposition were used, pitch (P), tempo (T) and proportional pitch and tempo (P/T). As this experiment investigated the difference between forms of transposition rather than the effect of amount of transposition there was no need for more than one level of transposition to be used. The degree of transposition employed in all cases was 150 cents and its temporal equivalent. Schellenberg and Trehub (2003) employed two levels of transposition in their experiment (100 cents, 200 cents) and achieved results across the range felt to be effective for this experiment (60-70% accuracy) so 150 cents was regarded as a good compromise.
Taking a similar view but not restricting the employment of relational processes to the condition of overburdened memory, Levitin and Cook (1996) wrote, “It seems increasingly clear that human memory encodes both the abstract and the relative information contained in musical pieces and that people are able to access whichever is required based on the given task” (p. 933). Levitin (1994) stated that “It seems likely that one’s internal representation of the song contains many components, such as timbre, tempo, lyrics and instrumentation; indeed, the entire spectro-temporal pattern of a song may well be represented” (p. 420). Whatever the answer to the above question is, the overall view appears to be that relational coding plays some part and indeed
Independent Variable 2 (IV2) Musical training The second independent (between subjects) variable was musical training (IV2). Most studies examining the assessment of musical parameters report musicians to have greater acuity when reporting the degree of change in musical parameters than nonmusicians (Geringer & Madsen, 1984; Wapnick, 1980) and it was of interest to examine whether this effect occurs for the paradigm used in this experiment. For this experiment, musicians were defined as participants reporting at least three years of musical training and nonmusicians were defined
1772
ICMPC9 Proceedings
as participants reporting no formal musical training and no organized musical participation.
Dependent Accuracy
Variable:
were allocated to one of two groups (musician/nonmusician) depending on whether they reported having had formal musical training. Twenty four participants reported formal musical training of three or more years (M = 6.33 years, SD = 2.45 years) while forty eight participants reported no formal musical training. No one reporting a hearing impediment was tested.
Recognition
The dependent variable was the proportion of times the excerpt presenting the original pitch and tempo of a television theme was identified as more like the original theme when compared with an excerpt presenting a manipulated pitch (P), tempo (T) or proportional pitch and tempo (P/T) combination in a comparison pair.
Stimuli The six television themes used as stimuli were similar to those used by Schellenberg & Trehub (2003) with minor adjustment to accommodate local (Sydney, Australia) programming content and the inclusion of a control theme. An informal survey of students (N = 30) from the group expected to take part in the experiment (Psychology 1A students) produced a list of their 12 most popular television programs. The themes from four of these television programs were rejected because they included vocalisation in the first five seconds of the theme. Vocals were excluded because vocals were thought to create more obvious artefacts when transposed because of humans’ greater familiarity with the timbre of the human voice.
Hypotheses Four hypotheses were proposed. Hypothesis 1 states that participants report having seen the six actual television programs listed in the questionnaire significantly more often than the two fictitious television programs listed in the questionnaire. This was seen as a way of confirming the credibility of participants’ response to the questionnaire. Hypothesis 2 states that the proportion of correct identifications of the original excerpt of theme F (control) occurs at chance level. It was reasoned that if the digital manipulation of pitch and tempo did not create artefacts that served as cues to identify the original from the processed version, then identification of an unfamiliar theme, whose pitch and tempo could not have been memorised, would occur at chance levels.
The eight remaining titles were presented to participants in a pilot test (n = 16) questionnaire. Participants were asked to indicate which programs they had watched. The final set of six themes comprised themes from the five programs reported to be watched by the most participants in the pilot test questionnaire and a television theme not broadcast in Australia (control theme). The control theme was an actual television theme commissioned for a European television program but never broadcast. The final list of six television themes employed in the experiment was E.R., Friends, Law and Order, M.A.S.H., X-Files and the control theme.
Hypothesis 3 states that the proportion of correct identifications of the original excerpt is greater when the original excerpt is compared with an excerpt transposed by either pitch only or tempo only, than the proportion of correct identifications of the original excerpt when the original excerpt is compared with an excerpt transposed proportionally in pitch and tempo. Hypothesised difference (P > P/T, T > P/T).
Stimulus Preparation The six original TV theme excerpts were approximately five seconds in length and were in all cases the first five seconds of the theme. Actual length of the original excerpts ranged from 4.87 seconds to 5.23 seconds, but this was felt to be acceptable as it allowed each excerpt to stop at a point of relative tonal resolution. The original (musical) endpoint was maintained when tempo manipulation was employed which meant that the duration of manipulations including a tempo component varied by +9.05 % or - 8.30 % of the original duration (4.47 seconds to 5.70 seconds).
Hypothesis 3 addresses the central focus of the experiment, whether proportional transposition of both pitch and tempo of an alternate version of an already memorised television theme makes it more difficult to differentiate the transposed version from the original than to differentiate between the original and an alternate version that had only its pitch or its tempo transposed. Hypothesis 4 states that musicians are more accurate in their identification of the original excerpt than nonmusicians.
METHOD
All themes were then manipulated. For pitch shifted stimuli (P), the original excerpt was either transposed up 150 cents (+P) or down 150 cents (-P). For tempo shifted stimuli (T) the original excerpt was either increased in tempo by 9.05 % (+T) or decreased in tempo by 8.30 % (-T). Excerpts representing a proportional shift in both pitch and tempo (P/T) incorporated either a pitch transposition of +150
Participants The 72 participants were all Psychology 1A students enrolled at the University of Western Sydney who participated for course credit. The mean age of the participants was 19.90 years (SD = 3.69 years) with a gender distribution of approximately twice as many females as males (49 female, 23 male). Participants
1773
ICMPC9 Proceedings
cents and a tempo increase of 9.05% (+P/T) or a pitch transposition of -150 cents and a tempo decrease of 8.30 % (-P/T). Stimuli presented in the experiment as the original excerpt were also processed, initially with a pitch transposition of +75 cents and a tempo increase of 4.53% the original excerpt and subsequently with a pitch transposition of -75 cents and a tempo decrease of -4.15%. This served the purpose of introducing as many processing artefacts from the pitch change and tempo change processes as possible into the original sample so that participants would not be identifying processing artefacts rather than changes in pitch or tempo.
version of the theme differed from the original. Participants were instructed to indicate at the conclusion of the second repetition which of the two versions they considered to be most similar to their recollection of the original. Participants did so by pressing the “F” button on their data entry pad if they considered the first version to be most like the original and by pressing the “S” button on their data entry pad if they considered the second version to be most like the original. Across the 72 participants, half recorded their responses on data entry pads whose “F” button was placed to the left of the “S” button and the other half recorded their responses on data entry pads whose “F” button was placed to the right of the “S” button.
Stimuli were presented in comparison pairs. A comparison pair consisted of the television theme with its original pitch and tempo and the same theme with digitally manipulated pitch, tempo or pitch and tempo. Order of presentation was counterbalanced. This generated 72 manipulated/original television theme comparison pairs (6 x theme, 3 x form of transposition, 2 x order of presentation, 2 x direction of transposition).
Participants were told that if they were unfamiliar with the excerpt they could either not respond or indicate which excerpt they suspected to be the original. Participants were told there would be an eight second pause between each comparison pair and that they should record which excerpt from a pair sounded more like the original during that period. Participants then responded to a sequence of six manipulated/original television theme comparison pairs. The sequence of six themes lasted approximately two minutes. At the conclusion of the sequence, participants spent the ensuing 20 minutes completing a series of tasks from an unrelated experiment and then after a two minute break completed a second sequence of six manipulated/original television theme comparison pairs. At the conclusion of this sequence participants spent another 20 minutes completing a further unrelated experiment and then after another two minute break completed a final (third) sequence of six manipulated/original television theme comparison pairs.
Equipment and Materials Stimuli were obtained from a variety of TV theme compilation CDs and were manipulated on a Macintosh G4 Powerbook (887Mhz) computer operating Logic Audio Platinum (V 6.3.2) software. The manipulated stimuli were played back through a MOTU 828 Firewire audio interface that fed a purpose built 4x4 headphone amplifier driving four sets of Sennheiser HD580 stereo headphones. Additional hardware consisted of a Roland CS-20 music entry pad and four purpose built data entry pads to encode the participant’s response. Two written response questionnaires were supplied to each participant prior to commencement of the experiment. Questionnaire 1 gathered information regarding age, sex and musical training of the participant. Questionnaire 2 presented a list of 10 television programs.
At the conclusion of the experiment its general aims were explained and participants were invited to ask questions about the goals of the experiment.
Procedure
RESULTS
Participants were asked to complete a questionnaire that consisted of a list of the names of ten television programs. Participants were asked to record a “yes” for each of the ten programs listed if they had ever watched that particular program and a “no” if they had never watched that particular program. They were further instructed that if they knew of a program but had not watched that program they should respond “no” for that program.
To investigate H1 (reported watching of actual television programs is greater than for fictitious programs) scores for each of the television program titles listed on the response sheet were compared using a paired sample t test. The assumption of normality was met, and with alpha set at .05, a statistically significant t value was returned, t(71) = 29.55, p < .001. This indicated that the mean proportion of times participants reported experience of the actual television programs (Cheers, E.R., All Saints, Friends, X-Files, M.A.S.H., Law & Order, Blue Heelers) (M = .70, SD = .20) was significantly greater than the mean proportion of times participants reported experience of the fictitious television programs (Drayton’s Justice, Second Chance) (M = .00, SD = .00). These results support H1. In fact, as can be seen from the descriptive
Participants were then played four examples of manipulated/original television theme comparison pairs. The themes used for demonstration were popular themes but did not include those used in the actual experiment. It was explained that each five second theme would occur twice in succession, with a two second pause between versions, and that each
1774
ICMPC9 Proceedings
.28, nor was the main effect for level of musical training F(1,70) = 3.02, p = .09. However, the level of musical training by form of transposition interaction was significant F(1,70) = 4.64, p = .01. The interaction is illustrated in Figure1.
statistics, none of the 72 participants reported having seen either of the two fictitious programs. Scores for participants’ identification of the original version of the television theme during the experimental trials were then used to calculate the proportion of correct responses (for each television theme) out of those trials attempted and also to calculate the proportion of times each television theme was attempted overall (Table 1). Table 1 Number of Correct Responses, Incorrect Responses, Total Responses, Proportion Correct and Proportion of Total Responses Compared with the Maximum Possible Number of Responses by Television Theme.
Figure 1. Mean proportion correct, and standard errors of the mean, for identification of original theme recording for pitch transposition, tempo transposition and proportional pitch and tempo transposition across nonmusicians and musicians. To investigate H2 (the proportion of correct identifications of the original excerpt of theme F (control) occurs at chance level) the proportion of times the original version of theme F (control) was identified correctly was compared to chance using a single sample t test. The assumption of normality was met, and with alpha set at .05, there was no significant difference between the proportion of correct identifications of the original unbroadcast (control) theme and chance (.5), t(49) = -.18, p = .86. Results for the (control) theme were excluded from all subsequent stages of the analysis.
Three post hoc comparisons (paired t tests) were conducted on the accuracy scores using a Bonferroni adjusted alpha of .017. There was no significant difference in accuracy between pitch transposition and tempo transposition for nonmusicians t(47) = .92, p = .36. However, the proportion of accuracy was significantly greater for pitch when compared to proportional pitch and tempo transposition for nonmusicians t(47) = 2.77, p = .008. There was no significant difference in accuracy between tempo transposition and proportional pitch and tempo transposition for musicians t(23) = .94, p = .36. Thus H3 was supported for nonmusicians but not for musicians and H4 was only supported at the proportional pitch and tempo level of form of transposition.
To investigate H3 (the proportion of correct identifications of the original excerpt is greater when the original excerpt is compared with an excerpt transposed by either pitch only or tempo only, than the proportion of correct identifications of the original excerpt when the original excerpt is compared with an excerpt transposed proportionally in both pitch and tempo) and H4 (the proportion of correct identifications of the original excerpt when compared with excerpts manipulated by pitch, tempo or proportional pitch and tempo is greater for musicians than nonmusicians, scores from the 72 participants across the three levels of form of transposition (pitch, tempo, proportional pitch and tempo) were analysed using a mixed repeated measures analysis of variance (ANOVA) with degree of musical training as the between-subjects factor. There were two levels of degree of musical training: nonmusician (no reported formal musical training or experience, n = 48) and musician (more than three years reported musical training or experience, n = 24). The ANOVA test assumptions were met. Using an alpha level of .05, the main effect for form of transposition was not significant F(1,70) = 1.28, p =
DISCUSSION The results for pitch transposition were in keeping with those found by Schellenberg and Trehub (2003) and were similar for both musicians and nonmusicians, with musicians achieving a slightly higher proportion of accurate responses than nonmusicians. Results for tempo transposition were similar to those for pitch, although in this instance nonmusicians recorded a slightly higher proportion of accurate responses. Interestingly, recognition of tempo change was as effective as recognition of pitch change when one considers that the generally accepted just noticeable difference (JND) for tempo is considerably greater for tempo than pitch. The most salient results however were those for proportional pitch and tempo transposition. There was no significant difference between the proportion of correct responses for transposition of pitch, tempo and proportional pitch and tempo for musicians.
1775
ICMPC9 Proceedings
Hypothesis 3 predicted an overall advantage in the identification of differences in pitch and tempo from recalled themes for musicians and while the difference for proportional transposition of pitch and tempo was sufficiently severe to force the overall comparison into significance it was not evident in any of the other comparisons. Indeed nonmusicians showed greater accuracy for tempo than musicians. The outcome for tempo, while not significant, is of interest even as a trend. Research suggests that nonmusicians pay more attention to tempo than pitch (Le Blanc, 1981; Madsen & Geringer, 1976; Wapnick, 1980) and Great Base Theory (Pound, 1934) argues that tempo is the fundamental frequency upon which all other frequency relationships are based.
However, the proportion of correct responses in the proportional pitch and tempo condition was significantly lower than either the tempo or pitch condition for nonmusicians.
Musicians/Nonmusicians It would seem probable that the results for nonmusicians were representative of the more “natural” perceptual condition and that the results for the musicians were a product of the perceptual/musical training and rehearsal that a musician undertakes. This might well include the independent memorisation of pitch and tempo rates. Musicians place a high priority on such information. On the other hand, relational coding might be a strategy employed by a person who does not place as high a premium on storing individual parameters of musical information, such as a nonmusician. This is not to suggest that one form of musical information is memorised by musicians and another by nonmusicians. As stated previously, storage of musical information includes both abstracted relational information and irreducible absolute information (Narmour, 1977). However, it may be the balance between these two strategies differs between musician and nonmusician. Premack (1978) suggested that abstraction is only induced as a response to overburdened memory, if that is the case it could be argued that abstraction could also be the product of lower prioritisation of memory resources with regard to musical information.
An alternative explanation An alternative explanation for the results is that rather than having their identification of the original pitch and tempo confused by the identical frequency relationship between the pitch and tempo of the altered excerpt, participants were confused by having to respond to two parameters which differed from their memory of the original, rather than being able to focus on just one differing parameter. Experiments that have manipulated multiple parameters (pitch, reverberation, amplitude and spectrum) of auditory icons (Stevens, Brennan & Parker, 2004) found that participants exhibited less accuracy when asked to recognise multiple auditory parameters compared with recognition of a single auditory parameter. Although this does not seem as plausible an explanation as that of relational coding, it could be investigated with an experiment similar to that just described in which all manipulations include changes to both pitch and tempo. These manipulations could be conducted to varying degrees and in varying directions (e.g. pitch higher/tempo slower or pitch lower/tempo faster). If the reason for reduced accuracy in identifying the original is because of the need to track changes to more than one parameter, then an excerpt with increased tempo and lowered pitch should be as difficult to separate from the original as one with proportional pitch and tempo changes.
Presumably, musical training would lead a musician to be more thorough in their identification and memorisation of individual musical parameters. Indeed, many aspects of musical training are designed to develop the ability to deal with individual components of music independently, such as the accommodation of the separate pitch and tempo constraints imposed by their instruments/voices and other physical limitations. Also the conventions of Western musical notation predispose musicians to practices that deal with pitch events and temporal events independently. Conversely, nonmusicians might be more pragmatic and store musical information in the most economical (relational) form available. Conversation with musicians and nonmusicians often bears this observation out, with musicians referring to individual instruments and harmonic structure while nonmusicians tend to a more unified, though in many ways just as detailed, recollection of a piece of music.
Implicit Knowledge of Iconic Themes Another finding of this experiment was that the accurate memorisation of auditory parameters such as the pitch and the tempo of iconic TV themes extends beyond participants’ ability to report memory of those themes, or indeed report having watched the television programs. Good examples of this are the M.A.S.H. and X-Files themes. While these two themes were reported as being watched by approximately 80% of participants in the questionnaire, their response rates were 99% and the proportion of correct responses was greater than that for Friends which was reported as being watched by 95% of participants. As participants were quite willing to employ the "no response" option in 51% of
The finding of significantly greater accuracy in identifying the original excerpt by musicians over nonmusicians is therefore somewhat problematic. The results are better described as showing a decrease in the performance of nonmusicians for the proportionally transposed pitch and tempo condition than for any overall advantage in identification of the correct pitch and tempo shown by musicians.
1776
ICMPC9 Proceedings
presentations of the control theme, a 99% response rate with accuracy superior to the more frequently reported themes would seem to indicate a greater memorisation of the themes than was reported in the questionnaire. The fact that these two programs are not presently being televised suggests that participants may have forgotten the names of the programs, yet still retained pitch and tempo information in memory. It also seems probable that even for participants who do not watch these programs, the programs popularity and the fact that they are advertised during commercial breaks at other program times would lead to some level of familiarity with their musical themes.
ability to deal at times with the parameters of pitch and tempo independently. Also, as stated above, it can be argued that nonmusicians’ results represent a more natural perceptual outcome than musicians. Results for nonmusicians were as would be expected if there is perception of frequency relationship between pitch and tempo.
REFERENCES Attneave, F., & Olsen, R. K. (1971). Pitch as a meduim: A new aproach to psychophysical scaling. American Journal of Psychology, 84, 147-166.
This information is important when designing future experiments because of the apparent decrement in accuracy when the same theme is repeated even after a break in repetitions. The more themes that are sufficiently well remembered by participants to be used in an experiment such as this, the fewer times a theme needs to be repeated in the experiment.
Brennan, D., & Stevens, C. (2002, July). An experimental investigation of preferred simple integer relationships between pitch and tempo. In C. Stevens, D. Burnham, G. McPherson, E. Schubert & J. Renwick (Eds.), Proceedings of the 7th International Conference on Music Perception and Cognition. Sydney, Australia: Causal Productions
The Effect of Repetition While not the basis of a hypothesis it was evident that the most accurate proportion of identifications of the original TV theme excerpt occurred during the first presentation of that theme. In their experiment, Schellenberg and Trehub (2003) conducted five trials at a session and reported “performance on the first trial of each excerpt significantly exceeded performance on subsequent trials” (p. 264). In light of this effect, repetitions in the current experiment were separated by 20 minute intervals during which other tasks were performed by the participant, however, the effect of better accuracy at the initial presentation of a theme remained. It would appear that, as one might expect, the experience of several alternative forms of the same excerpt in relatively short-term memory confuses the comparison process. It was even thought that as the current experiment presents the original theme three times and each manipulation only once that the repetitions might reinforce the recollection of the original theme. However, this does not appear to override the effect of alternative recent memories of the theme.
Cowell, H. (1930). New musical resources. New York: A.A. Knopf. Geringer, J. M., & Madsen, C. K. (1984). Pitch and tempo discrimination in recorded orchestral music among musicians and nonmusicians. Journal of Research in Music Education, 32, 195-204. Hintzman, D. L. (1986). "Schema abstraction" in a multiple trace memory model. Psychological Review, 93, 411-428. Le Blanc, A. (1981). Effects of style, tempo and performing medium on children's music preference. Journal of Research in Music Education, 29, 143-156. Levitin, D. (1994). Absolute memory for musical pitch: Evidence from production of learned memories. Perception and Psychophysics, 56, 414-423. Levitin, D. J., & Cook, P. R. (1996). Memory for musical tempo: Additional evidence that auditory memory is absolute. Perception and Psychophysics, 58, 927-935.
Conclusions
Madsen, C. H., & Geringer, J. M. (1976). Preferences for trumpet tone quality versus intonation. Bulletin of the Council for Research in Music Education, 46, 13-22.
Results for the memory task used in this experiment support the possibility of the storage of tone/tempo frequency relationship in memory, in a similar form to the generally accepted relational storage of tonal harmony and rhythm.
Monahan, C. B. (1993). Parralels between pitch and time and how they go together. In Tighe, T.J., & Dowling, W.J. (Eds.), Psychology and Music: The understanding of melody and rhythm (pp. 121-153). New Jersey: Lawrence Erlbaum.
It should be conceded that the ability of musicians to better differentiate pitch/tempo relationships based upon different frequencies but exhibiting the same ratio relationship was not predicted. However, a great deal of sense can be made of such an outcome. As stated above, in the context of the way musical notation and instruments have evolved, it is of considerable benefit to musicians to develop the
Nakata, T., Schellengberg, G. E., & Trehub, S. E. (2004, April). Cross-cultural perspectives on pitch memory. Paper presented at the International Congress on Acoustics, Kyoto, Japan.
1777
ICMPC9 Proceedings
Narmour, E. (1977). Beyond Schenkerism: The need for alternatives in music analysis. Chicago: University of Chicago Press.
Stevens, C., Brennan, D., & Parker, S. (2004, April). Multidimensional coding of auditory icons. Paper presented at the Proceedings of the 18th International Congress on Acoustics, Japan.
Premack, D. (1978). On the abstractness of human concepts: Why it should be difficult to talk to a pidgeon? In S. H. Hulse, Wapnick, J. (1980). Pitch, tempo and timbral preferences in H. Fowler & W. K. Honig (Eds.), Cognitive processes and recorded piano music. Journal of Research in Music behavior (pp. 423-451). Hillsdale, NJ: Erlbaum. Education, 28, 43-58. Pound, E. (1934). Ezra Pound and music: The complete criticism. New York: New Directions. Schellenberg, E. G., & Trehub, S. E. (2003). Good pitch memory is widespread. Psychological Science, 14, 262-266.
1778
