Kopiez, R. (2001). Book Review. MUSICÆ SCIENTIÆ V(1) 110-115 .Peter Desain and Luke Windsor (eds), Rbythm percepion and production. Lisse: Swets & Zeitlinger, 2000. (ISBN 9026516363) (Book including CD-ROM, ref: 9026516363).

In recent years rhythm research has become an important facet of music psychology. One may assume that a better understanding of the mechanisms of rhythm perception and production leads to a better understanding of music perception in general. In direct contrast to the importance of rhythm for the perception of music is the obvious lack of systematic rhythm research. Music psychology seems to be a far cry away from a unified theory of rhythm perception and production. This justifies the need for the book under review, which reflects the recent progress in rhythm research and summarises often widely spread results in a monography. This volume also partly compiles the results of the "Rhythm Perception and Production Workshop" organized by the Nijmegen Institute for Cognition and Information (NICI) in 1998. Despite the probably unattainable aim of a unified theory of rhythm perception, the editors give an introductory statement of their aim: "[... ] we seek to give some small indication of the ways in which scientists with different methodological backgrounds and perspectives have approached the diversity of human rhythmic behaviour." (P. xii).

In terms of structure, the book is divided into 5 parts, Part I beginning with contributions to the topic of "Mental timekeepers, internal clocks, oscillators and complex dynamics". The first chapter of Part I tackles the question of mental timekeepers and starts with an essay by Beek, Peper and Daffertshofer titled "Timekeepers versus nonlinear oscillators". The theoretical framework of Part I is sketched by Summers in the Introduction: on the one hand proponents of a cognitive approach assume a control of movement timing by temporal codes stored in human motor programs, whilst on the other hand the dynamic approach argues that timing is a property of self-organising processes in the neuromotor system itself. This first chapter is intended as an introduction to the subject of rhythmically timed bimanual movements. The authors compare very clearly the properties of the "timekeeper" approach for bimanual rhythm production with the oscillator approach. Despite the numerous equations this essay provides an excellent overview of discrete and dynamic models of polyrhythmic production. The second chapter, titled "Considerations regarding a comprehensive model of (poly)rhythmic movement" by Peper, Beek, and Daffertshofer is a continuation of Chapter 1. Shortcomings of both oscillator and timekeeper models are discussed against the background of issues such as learning processes, transitions between oscillation patterns, multistability without an external reference signal, or an explanation for typical temporal variability. This results in the conclusion that neither of the two competing models can alone provide an adequate explanation for the above mentioned phenomena. Thus an integrated model is presented which describes the two limbs of a tapping subject as a two-tiered oscillating system. The third chapter by Peper, Beek and Daffertshofer on "Limitations of the scanning procedure in assessing changes in coordination dynamics due to learning' tackles methodological problems of switching behaviour from anti-phase to in-phase tapping. This experiment demands phase-shifted tapping with both hands under the condition of extended training. The results show localized areas of greater stability which are interpreted as chaotic attractors. Within this framework motor coordination and learning might be interpreted as a process of the strengthening of these attractors. In the closing chapter of the first part Summers describes the learning process of multifrequency patterns such as 5 : 3 beats. The author is interested in transfer effects to the performance of other multifrequency patterns which would refer to a parallel model of motor organisation characterized by independent hand movements. Alternatively, an integrated model of motor organisation is suggested. Results reveal large differences in achievement between musicians and non-musicians. He concludes that no transfer of bimanual tapping skills could be observed and an hierarchical integrated model fits best to the results found.

The second part is guided by the topic "Tapping and synchronization'. After a summarized introduction from Michon (a pioneer of performance research who investigated the timing in Satie's "Vexations" already in the early 70s) there follows a chapter "On the nature of variability in isochronous serial interval production": by Madison. The author argues that drift in isochronous tapping should not be viewed as an "inconvenience" but rather as a feature of the interval generating system itself. He points out the dependency of the shift from the inter-onset interval (IOI). Deviations are nearly absent in short IOIs and increase (a) alongside an increasing IOI duration and (b) as a function of distance from the starting point. This drift in IOI variability can best be explained in terms of 1/f (pink) noise, which is also characteristic for most biological mechanisms. As a methodological consequence the author concludes that it is problematic to use linear methods for the analysis of a timing series. This paper is very inspiring and sheds new light on the capacity of rhythmic real-time performance of human beings, as can be experienced, for example, in the ability to synchronize in an ensemble performance and to keep a constant tempo with a remarkably low variability - even despite the mentioned shift. The paper by Wohlschläger and Koch on "Synchronization error" tackles the question of synchronization with a trigger signal always precedes the trigger (the so-called negative synchronization error); usually of an amount between 20 and 60 ms. As the authors convincingly prove by use of experimental data, explanations of this effect based on external factors such as signal properties (for example used in the P-center theory) are insufficient, because the synchronization error can even be produced by deafferented subjects without any visual or auditory feedback. In the authors' view the synchronization error nearly disappears if in-between contact-free taps are performed or similarly, in the case of tapping to percussion music and not to clicks. This means that the synchronization error only appears under artificial laboratory conditions when the time lapsed between taps is viewed by the subject as "empty" time. In the following essay "Subliminal temporal discrimination" Repp investigates the interesting question, whether synchronization errors in phase and period are compensated for, even if they do not exceed the temporal detection threshold of roughly 20 ms. indeed, a review of the literature reveals that periodic changes of a stimulus below the detection threshold were compensated for by subjects. Furthermore, Repp confirmes that the lengthening or shortening of a single interval in an isochronous series triggers a compensation mechanism in the subjects' synchronous tapping. Additionally, he points out that the perceptual mechanism for the extraction of temporal information which is important for timekeeping in synchronous tapping, must be different from the perceptual mechanism on which judgements of temporal order of events is based. Against the background of Epstein's theory of proportional tempi, FraÀek, Mates and Nártová investigate the question whether there is any experimental evidence for the theory, which might be found in the production of temporal ratios. Tasks involving doubling and halving of tempi were used and significant deviations from exact proportions were found in both tasks. Deviations were strongly influenced by base tempi. The last paper in Part 2 by Eck, Gasser and Port, tries to develop a model of beat induction in human perception. The model's quality was to be assessed by its ability to make a robot tap alongside to music from the radio. The model is based on a network of oscillators and it is interesting to read about the experience of the authors with various network architectures: for example, the system of frequency-adaptive oscillators by Large and Kolen (1996) ran "unstable when coupled together in the large numbers necessary to support network learning" (P. 162). Thus it was decided to use a different network architecture of 40 fully-connected Fitzhugh-Nagumo neurons. Beat-tracking works acceptably with this model but is lacking in the behaviour characteristic of human beat tracking such as the continuation of tapping after switching off the external trigger.

The essays of the third part are dedicated to the topic "Time perception and estimation". This part begins with the contribution from Suetomi, Nakajima, Sasaki and ten Hoopen on "Time shrinking". The authors investigate an illusion which can be described as follows: the subjectively perceived length of the second duration is influenced by a first duration followed by a characteristic underestimation of the second. Despite the authors' claim that "this illusion may affect the perception of various rhythm patterns in our daily life" (P. 180), it remains to be seen whether the illusion is more than a psychoacoustic phenomenon resulting from specific artificial experimental conditions. Scott, Mattingley, Manly and Wise present an investigation of the influence of right hemisphere damage on the perception of inter-aural rhythmic sequences. Similar to the recognized deficits in spatial perception, patients perceived a temporal shift of 100-200 ms as subjectively regular. Due to the very limited number of subjects (patients: n = 2, controls: n = 2) it is my opinion that results should be interpreted cautiously in the sense of a pilot study. The last essay of this third part by Brown and Frieh presents a study of the influence of a concurrent memory task on temporal production and vice versa. Results support

the assumption of a bidirectional interference and the authors conclude that timing uses the same processing resources used by other executive level tasks" (P. 196) as was previously assumed in Baddeley's model of "working memory". Additionally, it remains doubtful if a time estimation task as used in the experiment (tapping with interonset intervals of 5 s) can be regarded as relevant to rhythm production in music. The sound examples on CD remain unclear because the stimuli for the subjects (track 19) and for the controls (track 18) sound identical.

The fourth part presents papers related to the topic "expressive timing in music" and begins after Vorberg's introduction with a paper from Desain, Honing, Aarts and Timmers titled "Rhythmic aspects of vibrato". This part is most closely aligned to the aspect of musical performance and is not as related to motor performance as most of the previous chapters. The chapter begins with a contribution from Desain, Honing, Aarts and Timmers on "Rhythmic aspects of vibrato". The authors investigate factors which could affect the parameters of vibrato such as tempo, note transition or meter of music. Although experimental data from the aforementioned study shows significant differences between instruments, the question remains unanswered whether such comparisons are legitimate as long as tenor, theremin and strings use completely different vibrato systems. In order to extend this study I would suggest an investigation of the biomechanical features of each system. Additionally, the tenor in Sound exemple 24 does not give one the impression of sufficient technical skill and seems to be of mediocre voice quality. Windsor, Aarts, Desain, Hejink and Timmers present a paper on "Timing of grace notes". The authors investigate the context- and tempo-dependent timing of these ornaments. In my opinion this is a significant contribution because it brings ornamental musical events into focus, which are usually neglected in performance research. Regression analysis reveals that grace notes are not invariant to changes in overall tempo, and also takes features of the individual musical "character" of a melody into concern. The chapter on "Rhythm in music performance and perceived structure" by Penel and Drake investigates the question of to which degree the players intentions are communicated to the audience. As in many previous investigations, Schumann's "Träumerei" is employed as the test composition. The decision to use "Träumerei" as the stimulus is the weak point of the investigation and is indeed an unfortunate one: due to the high popularity of the piece I am doubtful if the segmentation of the mechanical melody sequence perceived by the listeners is caused by the inherent compositional features or by the listeners' expectations. (For an extended version of this paper see Drake, Penel and Bigand in Music Perception, 2000, 18(l), 1-23.)

Part 5 seeks to broaden the book's horizons by introducing a collection of papers related to "Rhythm and meter in music and speech." It begins with a short introduction by Repp and is continued with a chapter on "Complexity measures of musical rhythms" by Shmulevich and Povel. The authors develop several models to calculate rhythmic complexity, however the validity of such models have to be tested through the use of less standardized input. In the following chapter by Drake Penel and Bigand which endeavours to answer the question, "Why musicians tap slower than non- musicians", data from a tap-along experiment is presented. A synchronization task was given to subjects to reveal the most salient metrical level induced by the particular rhythmic stimuli. It was observed that musicians synchronized with longer IOIs (1042 ms) compared to non-musicians (858 ms), a result which indicates different "referent levels . Musicians also tapped alongside a trigger pulse with a greater number of metrical levels. The results were interpreted as showing the different degress of complexity in rhythm perception to be dependent on the degree of the subject's musical competence. Surprisingly, Parncutt's "Perceptual model of pulse salience in music" was not mentioned in this paper (see Music Perception, 1994, 11(4), 409-64). Part 5 continues with a contribution from Müller, Aschersleben, Esser and Müssler on "Effects of delayed auditory feedback on speech". The authors refer to the disturbance of fluent speech production by delayed auditory feedback (the so-called Lee-effect). Their primary investigation looks at, "what aspect of the feedback produces the disturbance" (P. 256). An explanation of the Lee-effect by Howell's "displaced rhythm effect" takes only the rhythm of the delayed speech into concern whilst not addressing the semantic content of the signal. The authors conclude that displaced rhythm is not the only variable responsible for the Lee-effect. Additionally, it is suggested that the coincidence of two different contents causes perceptual confusion. The final chapter of the book by Lee, Todd, Foster and Lomlu on "Speech rhythm in French and English" investigates the influence of syllable stress on perceived tempo. From the accent structure in French (syllables are characterized by nearly equal stress) and English (stressed and unstressed syllables alternate) the authors predict that due to the higher rate of stressed events in French ("rate of dominant periodicity"), spoken phrases should be rated faster than in English. However, the results showed no differences in stress frequencies between languages and in a subsequent rating experiment an opposite result to the predictions emerged: English spoken sentences were rated subjectively faster than French ones. Despite several explanations offered by the authors, subjectively perceived flow of time seems to be influenced by numerous factors of which we are not yet completely aware.

In general the book's editors could have been more careful. For example, the entire lower paragraph on p. 10 has been copied onto p. 11, and the relationship between sound examples on CD and the corresponding chapters is not clearly indicated in the text for the most part. In the contribution by Eck et al. the reader has to search for the passages with which to fit the sound examples. Finally, the reviewer would like to add a general comment on the editors' intention: in the introduction the editors suggest that, due to the lack of unified theory, rhythm should be modest in the way it uses laboratory research and artificial stimuli under strictly controlled laboratory conditions, instead of investigating real-life settings (P. xiv). The reduction to controllable laboratory experiments is simultaneously the strength and weakness of this approach and it remains to be seen, how the gap between the experience of rhythmic phenomena (such as the experience of apparent motion) and artificial stimuli may be filled in future reseach. For example: if the Julliard experiment hypothesizes that off-beat tapping to a trigger pulse switches to synchronized on-beat tapping at a rhythmic frequency of about 2 Hz (p. 25), this is contradictory to the style of syncopated playing at extremely high tempi as in the Amadinda xylophone music of Uganda. It remains a great challenge to apply all findings of the book to an explanation of rhythmic phenomena which occurs in real music. Although it seems to be plausible judging by the recent state of rhythm research, to focus on a small area of rhythmic phenomena which are investigated under strictly controlled conditions, the question of validity should not be forgotten in the paradigm of controllability. In conclusion the book gives an excellent summary of the "state of the art" research in rhythm production and is a quarry of ideas and a "must" for anyone interested in this topic .

Reinhard Kopiez

Hochschule für Musik und Theater