International Symposium on Performance Science ISBN 978-94-90306-02-1 The Author 2011, Published by the AEC All rights reserved Annotation and the coordination of cognitive processes in Western Art Music performance Linda T. Kaastra Media and Graphics Interdisciplinary Centre, University of British Columbia, Canada This paper examines the role of performance annotations in coordinating Western Art Music (WAM) performance. Annotations are classified by their function in supporting cognitive and meta-cognitive processes of performance in relation to the printed score (visual salience, repair/correction, and anchoring). The classification supports theory in distributed cognition by demonstrating a clear functional relation between the annotated score and both internal and external performance processes. Keywords: music performance; distributed cognition; coordination devices; annotation; instrumental case study Cognition in ensemble performance is many-layered and multi-faceted, including processes that have internal and external, individual and group, cognitive and meta-cognitive components. This paper presents an analysis of annotations on a performance score drawn from a case study of ensemble performance in the Western Art Music (WAM) tradition. The observational case study consisted of nine videotaped rehearsals and a performance of Torû Takemitsu s Masque for Two Flutes (1959-60). A prior analysis of the conversation, body motion, and instrumental play revealed four domains of coordination in music performance (Kaastra 2008). This paper presents a detailed exploration of one of those domains, annotation, using theory drawn from the field of distributed cognition. For this analysis, every pencil mark on the rehearsal score has been studied in the context in which it was made, though only a summary of the analysis appears here. In the most comprehensive ethnographic study of music annotation to date, Winget (2008) collected over 25,000 annotations created by chamber and orchestral musicians at a range of skill levels. Through an examination of rehearsal scores and some accompanying interview data, she analyzed the
676 WWW.PERFORMANCESCIENCE.ORG characteristics of the annotations, the purpose for making the annotations, and the knowledge required to use the annotations in context. She classifies annotations by mode (text, symbol, number) and function (technical, conceptual, technical-conceptual). This organization of the performance world is built upon the notion of the musical score as a standardized-type boundary object. The boundary object holds information in a standard form for communication across diverse communities. This implies a linear transfer of musical information from composer (conductor ) performer audience. The score is viewed as an object of communication, and the annotations are viewed as evidence of breakdowns or clarification in the communicative event. In order to delve more deeply into the cognitive processes of performing music, I propose studying the score as a coordination device rather than a communicative artifact. This shifts the unit of analysis from the contents of the score to the relation between the score and the performers. In this work, I am interested in how a mark is used in performance, rather than its abstract musical value. The emphasis on the dynamic interaction between the score and the performers comes from the field of distributed cognition (Hollan et al. 2000, Kirsh 2005), an area of research in cognitive science that extends the unit of analysis for cognitive activity beyond the brain of an individual to the functional relations between an individual and her tools and environment. From this systems perspective, the score is not just a storage device for musical content, but a persistent external representation (Kirsh 2010) functionally tied to cognitive processes in music performance. Thus, the annotations help to make the internal processes of performing (i.e. counting) easier, and they provide external visual anchors for placing expressive sound in the context of a live performance of the work. The implications of that shift of perspective are made clear in the discussion that follows. Finally, Winget s (2008) study demonstrated that skilled instrumentalists tend to annotate more than less-skilled players. This finding corresponds with Kirsh s (2010) theory that external representations increase and enhance cognitive processes by providing a persistent external representation of internal cognitive processes. The more advanced performer, this suggests, uses more external support because she is engaged in more complex performance activity. Based on these findings, we would expect to see many annotations on a performance score for a complex or novel piece of music like Masque. In fact, the first movement contains a total of 128 pencil marks. Out of 37 bars, only 11 are free of markings. The second movement contains a total of 65 pencil marks. Out of 40 bars, only 17 have no markings. I have classified the marks by their function in supporting cognitive processes of musical coordi-
INTERNATIONAL SYMPOSIUM ON PERFORMANCE SCIENCE 677 nation in relation to a printed score: visual salience, repair/correction, and anchoring. MAIN CONTRIBUTION A musical score, from the point of view I am taking in this article, is not just a storage device for musical information. It holds links to cognitive processes required for performance. Viewed this way, every aspect of the printed page signals some process or state through which performance is coordinated. The most obvious link from the score to an internal cognitive process lies in the numerical indications for subdivisions of the beat. The rhythms (especially in Mvt. 1, Continu) are very uneven, the tempo is constantly shifting, and the pulse is deliberately obscured. Each measure has a different time signature ( 2 /8, 5 /16, 5 /8, etc.), and within each measure, subdivisions of the beat layer duple and triple and sometimes quintuple (or other uneven) counts between parts. Numerical cues indicate how uneven divisions of the beat should be counted (e.g. 3, 5, 6, 7, 5:4, 3:2). A 3 over a combination of notes indicates a triplet subdivision (see Figure 1). A 5:4 over a sixteenth note rest, one eighth note, and two sixteenth notes indicates that there are five sixteenths to be counted in the space of four (bar 8 not shown). These precise numerical cues establish a link between the internal process of counting and the external coordination of sound. Indeed, every aspect of the score can reveal a connection between a performance process and the page. Masque is a precisely notated, very detailed example of modern Western Art Music. Every note has an indication for articulation, dynamics, timbre, and shape (as well as, of course, pitch and rhythm). The likelihood of this score being memorized is small, solidifying its role as a persistent external representation. A classification of performance annotations I observed each annotation as it was made in the rehearsals, including the playing, stoppage of play, and conversation that accompanied each mark. I then classified the marks based on the reason they were used. The annotations fall into three categories: Class 1: Visual salience Class 2: Repair/correction Class 3: Performance anchors The first group of annotations make existing aspects of the score more visually salient for performance. These include widening dynamic wedges
678 WWW.PERFORMANCESCIENCE.ORG (top of staff, bar 20 in Figure 1), adding existing markings (e.g. dynamics, tempo) to the second staff, circling something, or translating a verbal cue into English. The intent behind these marks is not to revise the content of the music but to assist the performers in accomplishing a performance goal already indicated in the score. Only nine such marks appear on the rehearsal score for Masque. The second group of annotations repairs or corrects some aspect of the score, or a cognitive process implied by the score. For example, bars 20, 23, and 32 are presented with time signatures that are uneven: 3½ /8, 2½ /8, and 2½/8. The uneven time signatures appear to require subdivision at a unit below what is written. To correct the undesired outcomes from counting in such a lop-sided manner, the flutists revised the measures to 7 /16, 5 /16, and 5 /16, respectively. For the purposes of making music, it is much easier to count in equal units to 7 than to count to 3 and add a half-beat. The flutists revised the score to indicate the required subdivision. Other Class 2 annotations include verbal cues that try to mitigate negative performance outcomes. For example, go indicates keep moving in a passage that has a tendency to slow down; SLOW in block letters serves as a red flag to slow down; note names written in an attempt to prevent note errors; and words like fast, slower, more time, accel. molto (+ wavy line) all try to drive the musical activity away from natural outcomes that were determined to be undesirable by the flutists. The third group of annotations I call, performance anchors because they hold and link internal and external performance processes. This group comprises the vast majority of annotations on this score, due to the rhythmic complexity of the music. Beat marks were used extensively in early stages of rehearsal. The marks form a visual click track so that the flutists can see (while they count, listen, and play) how the beats line up. Other anchors include triangles to indicate the count in an uneven bar (e.g. 7 /16 with a triangle over the last part indicates 1234-123 count). In addition, arrows are used to coordinate motion or pitch. For example, curved arrows over the end of the staff means that the flow of the music should continue into the next line. Straight arrows indicate motion (forward=faster; backward=slower) or pitch (up or down). Not surprisingly, most Class 3 annotations are symbols, although some of these symbols are also accompanied by words (e.g. go ). This annotation combines marks from Classes 2 and 3. While this article presents only a summary of the annotations, it may be helpful to view a few heavily annotated bars from the score. Figure 1 presents an excerpt of the score that shows some annotations from all three categories. Note the expanded dynamic wedge in bar 20 (Class 1), the revised time sig-
INTERNATIONAL SYMPOSIUM ON PERFORMANCE SCIENCE 679 Figure 1. Rehearsal score bars 18-20 from Movement 1, Continu, of Takemitsu s Masque for Two Flutes ( Editions Salabert France, used by permission). nature and arrow (Class 2), the short and long beat marks, and triangle (Class 3). IMPLICATIONS This classification of annotations supports the notion of the score as a persistent visual representation (Kirsh 2010). Perhaps the most persuasive case comes from Class 3, the Performance Anchor. The flutists used beat marks to line up their parts in the early stages of rehearsing the work. The flutists used the marks to coordinate the main beats and to coordinate their subdivision between beats, bridging internal and external coordination domains. Similarly, the wider dynamic wedge (see Figure 1) is not a revision of the composer s intention, but a signal to perform the dynamics. While Classes 1 and 3 support the notion of the annotated score as a persistent visual representation linking internal and external domains of coordination, Class 2 annotations more obviously link metacognitive processes (i.e. of monitoring, repairing, and/or correcting performance outcomes) with a mark on the score. The arrow in Figure 1, for example, is a reminder to push the tempo. The flutists will monitor the tempo and make adjustments to their performance of it based on the presence of the arrow. The beat marks, on the other hand, allow the flutists to anchor the flow of notes as they work through the passage. In the dress rehearsal, the flutists forgot the performance score and used a substitute score with no annotations. I observed a noticeable increase in the volume of bodily gestures in that rehearsal (Kaastra 2008). The flutists also reported that they were more attentive to the other cues that they used to coordinate performance in that rehearsal. Therefore, we cannot say, for example, that the beat marks were solely responsible for coordinating the
680 WWW.PERFORMANCESCIENCE.ORG rhythms in the performance of Masque, but we can say that they present one layer of coordination within a domain. WAM performance is coordinated by a combination of counting, marking, listening, watching, and moving. Instrumentalists attend to multiple cognitive processes and representational systems, sometimes leaning more on one than another, depending on the needs for performance. Future work will continue to compare cognitive processes between domains (e.g. annotation and gesture) to further examine the coordination of cognitive processes in music performance. Address for correspondence Linda T. Kaastra, Media and Graphics Interdisciplinary Centre, University of British Columbia, FSC Building, 2424 Main Mall, Vancouver, British Columbia V6T 1Z4, Canada; Email: lkaastra@magic.ubc.ca References Hollan J., Hutchins E., and Kirsh D. (2000). Distributed cognition: Toward a new foundation for human-computer interaction research. ACM Transactions on Computer- Human Interaction, 7, pp. 174-196. Kaastra L. (2008). Systematic Approaches to the Study of Cognition in Western Art Music Performance. Unpublished doctoral thesis, University of British Columbia. Kirsh D. (2005). Metacognition, distributed cognition and visual design. In P. Gardenfors and P. Johansson (eds.), Cognition, Education, and Communication Technology (pp. 147-180). Mahwah, New Jersey, USA: Erlbaum Associates. Kirsh D. (2010). Thinking with external representations. AI and Society, 25, pp. 441-454. Winget M. A. (2008). Annotations on musical scores by performing musicians: Collaborative models, interactive methods, and music digital library tool development. Journal of the American Society for Information Science and Technology, 59, pp. 1878-1897.