Society for Music Theory

Epistemology and Procedure in Aural Training: In Search of a Unification of Music Cognitive Theory with Its Applications Author(s): Kate Covington and Charles H. Lord Source: Music Theory Spectrum, Vol. 16, No. 2 (Autumn, 1994), pp. 159-170 Published by: University of California Press on behalf of the Society for Music Theory Stable URL: http://www.jstor.org/stable/746031 . Accessed: 01/05/2013 06:28
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

University of California Press and Society for Music Theory are collaborating with JSTOR to digitize, preserve and extend access to Music Theory Spectrum.

http://www.jstor.org

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

and Epistemology
In

of a Search with Its Applications

Procedure Unification of

in Aural Training: Music Cognitive Theory

Kate Covingtonand Charles H. Lord

INTRODUCTION

Research in music cognition has seen significant growth in recent years, transcending many disciplinaryboundaries. Such organizations as the Society for Music Perception and Cognition and the Foundation for Human Potential, which sponsored the 1992 symposium on Music and the Brain, reflect the degree to which this researchhas penetrated not only the domains of psychology and music but also such areas as neurobiology, neuropsychology, and computer science. Another measure of the maturingnature of the field is given by the number of important volumes of reported research thereof which have appeared on the scene in the immediate past.1
1See, for example: Jeanne Bamberger, The Mind behind the Musical Ear (Cambridge:HarvardUniversity Press, 1991); David Butler, The Musician's Guide to Perceptionand Cognition (New York: SchirmerBooks, 1992); W. Jay Dowling and Dane L. Harwood, Music Cognition (Orlando, FL: Academic Press, 1986); Robert Frances, The Perceptionof Music, trans. W. Jay Dowling (Hillsdale, NJ: Erlbaum, 1988); Carol L. Krumhansl, Cognitive Foundationsof MusicalPitch (New York: OxfordUniversityPress, 1990);and John Sloboda, ed., GenerativeProcessesin Music (Oxford: ClarendonPress, 1988).

On the other hand, one laboratory in which training in music cognition takes place- the aural-training classroom- is beset with numerous frustrationson the part of both learners and instructors. Students, for example, find at least three endemic problems. First, the training they receive does not seem to match what they will do as professionals. There is usually an emphasis on musical events isolated from their natural context, together with pencil-and-paper responses. Studentsfrequentlyfail to see how this methodology prepares them for the world of contextually based experience in which they will function as professional musicians. Second, there is usually a focus on the musical aspects of pitch and rhythm to the virtual exclusion of others. Yet what makes a passage musical depends heavily on factors beyond pitch and rhythm -dynamics, articulation, timbre, register, texture-aspects for which a keen awareness is critical for performers, conductors, and composers. And third, a direct connection to other aspects of the music curriculum itself is difficult for students to discover, particularlyrelatingto solo performance and ensemble work. Since aural training takes place in a learning environment so isolated from real-world contexts, and since paper-and-pencilis the primarymeans of response,

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

160

MusicTheory Spectrum

students find aural training to be disconnected from their immediate musical needs, as well as those of longer range. Shouldn't they instead be recognizing the connections with everythingthey do in the rest of their everyday musical lives? From the perspective of the designer and deliverer of instruction, there are also several perennial problems. One basic dilemma is that assessment seems to demand short, quantifiable examples. Because of the need for numericalresults, instructors search for mechanisms which are conducive to quantitativeevaluation. Short, isolated examples are a match for that need; thus much of what is usually done is driven by the evaluation mechanism. As an example, consider that the identificationof intervals is easily rated, whereas measuring sensitivity to intervallic intonation is much more difficult. Thus, the former receives attention in aural training, although the latter is equally important as a music-cognitive skill. Second, isolated drill tasks preclude work with full musical contexts. Traditionalpedagogy centers on a bottom-up approachin which elements are treated in isolation until well enough masteredfor integrationinto largercontexts. Because mastery of these may take several years, the fully contextual listening integratingall aspects into the kind of listening students need as professionals rarely gets developed within the curriculum.Finally, there is the ever-present dilemma of the highly skilled performerwho does not succeed in aural training situations. Can we safely assume that this student has poor cognitive skills despite success in other areas? Doesn't this situation raise questions either about what is being taught or at least whether it is compatible with how students already perceive and relate to music? It would certainlyseem that the time has come to reconsider both what is done in the auraltraining curriculumand how it is done. Motivated by these frustrations, therefore, we have undertaken over several years a fundamentalreexamination of both the goals of aural trainingand the means by which they are reached. An obvious startingpoint was the study of music

cognition as reflected in the research literature. Unfortunately, the experimental findings which comprise this literature still represent only the beginnings of a long-term research task-that of building an elaborate but well-formed knowledge base. This effort is requiringmultiple testings of many related hypotheses, each of which is only one component of what will ultimately become the larger knowledge base. As a result, currentthought has not yet reached a state of full development, since there is still so much about musiccognitive processing which is unknown or at least has not been conclusively demonstrated to date. We have therefore also sought sources outside the music disciplinesin our quest for solutions, research which addresses fundamental issues of the nature of the acquisition of knowledge, the resulting knowledge base, and the design of training. What we have discovered has challenged us. We find that the desired knowledge base itself, the theoretical basis for constructing it, and the means of evaluating the success of training are all in need of significantrethinking. In what follows, we will begin with an outline of the most prevalent model of music-cognitive knowledge and the strategies for teaching its acquisition-what we will call the objectivist model. Parallelswill then be drawnto the paradigmson which most music-cognitive research is based. Next, having cited the limitations of this model for instructional designers, pedagogues, and researchers, we will pose an alternative model for aurallyperceived music paradigm-a constructivist We will conclude with a few thoughts on the ramcognition. ificationsof the constructivistparadigmfor designers of both classroom and technology-based learning for researchers in music cognition.
THE OBJECTIVIST MODEL

To begin, then, we identify the most pervasive model of cognitive training, which educational theorists call the ob-

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

Epistemologyand Procedurein AuralTraining 161

jectivist model. We take this model as a point of departure, as it forms a consensus position, or what Denis Phillips refers to as the criticaltraditionwhich is shared by members of our disciplinein general.2The objectivist knowledge base consists of a distinct set of facts and skills-a combination of what psychologists call declarative and procedural knowledge.3 If one were to write out this knowledge base as a series of behavioral objectives, a sample of them might read: "the ability to identify accurately a heard minor sixth or diminished-seventh sonority," or "the ability to recognize pitch and rhythmic errors in performed melodies, given the notated version." One view is that this knowledge is encoded and represented in the brain in some sort of schematic network or hierarchy. Alternatively, it might be viewed as an expert system, with self-contained production rules, frames, and so on. Objectivist training enables discrete components of the network to be gradually refined in structure and better anchored into their locations. A similar increase in the strength of links between schemas develops with training, so that the entire fabric of the network gradually evolves into a well-formed and responsive whole.4 The primary goal of instruction, therefore, becomes the development of this clear and well-organized schematic network in the minds of the learners, or, in effect, the transmissionof the expert knowledge base of the person directingthe instructioninto the mind

2Denis C. Phillips, "Subjectivityand Objectivity:An Objective Inquiry," in Quantitative Inquiryin Education:The ContinuingDebate, ed. Elliot Eisner and Alan Peshkin (New York: Teachers College Press, 1990), 19-37; cited Revisited: A Search for Common Ground," in Peggy Cole, "Constructivism Educational Technology32, no. 2 (1992): 27. 3W. Jay Dowling, "Form and Memory," paper presented at the annual meeting of the Society for Music Theory, Kansas City, 1992. 4David E. Rumelhart, "Schemata:The Building Blocks of Cognition,"in TheoreticalIssues in Reading Comprehension,ed. Rand J. Spiro, Bertram Bruce, and William Brewer (Hillsdale, NJ: Erlbaum, 1980), 33-58.

of the trainee. The long-term ideal is that this nouveau expert schematic network be cognitively available whenever an educated musician is at work, providingthe necessary facts and skills for whatever musical context is at hand. The prevalent strategies for such traininghave their roots in behaviorist psychology. If the goal of instructionis to develop a well-structuredschematicnetwork, it would seem that an optimal method would be to hypothesize the component schemas, then teach each one in isolation until it is learned. Objectivist instructional design, therefore, concerns itself with identifying the facts and procedures to be transmitted and determining such factors as the order and frequency of task presentation. And indeed, the isolated drilling of intervals, pitch patterns, rhythms, sonorities, and so on, has been a hallmark of our critical tradition for decades. Not coincidentally, the growing predominance of this type of traininghistoricallyparalleled the increasing influence of behavioristicpsychology on training across all disciplines, consisting as it does largely of repetitive drills derived from a stimulus-responseparadigm. Assessment of objectivist training is primarily based on retrieval; that is, the responses requested generally require direct retrieval of the information learned or its cognitive relatives-such as recognition, recall, comparison, and identification. While examples used for assessment differ in specific details from those actuallytaught (for example, different pitches for a sonority type), the essential tasks are chosen to be similar enough to those through which the training was given that potential confusion from case novelty is minimized. Testing need not examine the learning activities themselves; it need only measure the retrieval of knowledge to satisfy its objectives. Moreover, this measurement is highly quantitative. Responses are generally classified as correct or incorrect, though at least minor gradations of the significance of different errors are often provided in instructors'evaluation scales. On the whole, the assessment process is, as could be

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

162

MusicTheory Spectrum come rigid and compartmentalizedrather than flexible. For example, if a student uses the NBC motto as a crutch to identify the major sixth, limitations are imposed on the context in which he or she will readily recognize the interval. All newly learned knowledge is integrated into prior knowledge, and if that major sixth is tied in prior knowledge to NBCthat is, 5-3 of a major scale-the learner will have to untie it cognitively in order to recognize it if it occurs as 2-i (which of course also carries a conflicting harmonic implication). Moreover, isolating elements from their natural context emphasizesthe separatenessof the elements ratherthan their integration. If learning activates prior knowledge, drillingsonorities separately from their context may cognitively link them more to other isolated sonorities ratherthan to the roles each sonority might play in a phrase. In fact, research has shown in other domains that such training can actually develop barriersbetween the schema types instead of developing an awareness of their interconnectedness.6

expected, rooted in objectivism just as is the instructional design.

INHERENT LIMITATIONS IN OBJECTIVIST PROCEDURES

Does this model of aural training produce the desired results? Our answer is both yes and no. If in fact the desired results are (1) a knowledge base of specific facts and skills, (2) format-dependent retrieval ability for the specific information trained, and (3) success within the limited context of isolated aural-trainingexercises, we will accept that those students scoring well seem to have developed the type of schematic network or expert system desired. If, on the other hand, the desired result is the long-term goal mentioned earlier-application of these schemas to real-world contexts outside of the aural-traininglaboratory-the results are not so convincing. We believe there are several reasons for this lack of success-reasons rooted in the epistemology of the objectivist model. For one thing, the ability of the learner to transfer knowledge from one context to another is simply assumed in objectivist theory. In practice, this transfer does not occur very often or very readily. That is, while it may be true that students can memorize the sounds of intervals well enough to respond correctly on an information-retrievaltest, their abilityto transferthat identificationabilityto actualmelodies, much less actual pieces with complete textures, is far from guaranteed. In addition, studies in other domains have shown that training in isolation from natural context can develop schemas which are of less than ideal quality.5 Instead, they be5See, for example: Richard L. Coulson, Paul J. Feltovich, and Rand J. Spiro, "Foundationsof a Misunderstandingof the UltrastructuralBasis of JourMyocardialFailure:A ReciprocationNetwork of Oversimplifications,"

nal of Medicineand Philosophy 14 (1989):109-46; and Paul J. Feltovich, Rand J. Spiro, and RichardL. Coulson, "TheNature of ConceptualUnderstanding in Biomedicine:The Deep Structureof Complex Ideas and the Development of Misconceptions,"in CognitiveScience in Medicine:Biomedical Modeling, ed. David Evans and Vimla Patel (Cambridge:MIT [Bradford]Press, 1989), 113-72. 6Forstudiesof issues relatedto "inertknowledge,"see, for example:Mark H. Bursteinand Beth Adelson, "Issuesfor a Theory of Analogical Learning," in ArtificialIntelligenceand the Futureof Testing,ed. Roy Freedle (Hillsdale, NJ: Erlbaum, 1990), 137-72; Rand J. Spiro, Paul J. Feltovich, Richard L. Coulson, and Daniel K. Anderson, "MultipleAnalogies for Complex Concepts: Antidotes for Analogy-induced Misconception in Advanced Knowledge Acquisition,"in Similarityand Analogical Reasoning, ed. Stella Vosniadou and Andrew Ortony (Cambridge:CambridgeUniversity Press, 1989), 498-531; and Carl Bereiter and Marlene Scardamalia, "Cognitive Coping Strategiesand the Problemof 'Inert Knowledge,' " in Thinkingand Learning Skills: CurrentResearchand Open Questions,vol. 2, ed. JudithSegal, Susan F. Chipman, and Robert Glaser (Hillsdale, NJ: Erlbaum, 1985), 65-80.

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

Epistemologyand Procedurein AuralTraining 163

LIMITATIONS OF MUSIC-COGNITIVE RESEARCH

Where then do we search for answers to these difficulties? Is behavioristic training wrong? Are retrieval-oriented tasks of little use? Is measurement by quantitative means improper? Is the desired knowledge base the wrong one? Logically we might look to research in music cognition for answers. As this literature grows, we should expect a corresponding increase in the degree to which the results of this research address the problems we face in training. However, there are several reasons why answers of the scope needed have not yet been forthcoming from this research: (1) Due to the exigencies of experimental design, nearly all cognitive research focuses on isolated behaviors. If one is to control variables properly, the research involved must isolate a single aspect of a musical situation for treatment. The process has its analog with the objectivist trainingmodel above the study of isolated phenomena without reference to their context, and in ways which rarely reflect their actual usage in a real-world situation. (2) Measurement of the results of research needs to be scientificallyvalid; hence there is an unavoidable motivation to use quantitative assessment. Again an analog to objectivist training appears. Measurement of single, isolated events by single or similar quantitative measures cannot reach beyond its own parameters;it cannot serve as a basis for studyingthe effects of experimental training done beyond the context of the experiment, nor can it deal directly with the processes involved in the learner's production of the results. (3) The knowledge base which is being developed by cognition research consists of fragments of information, each generally self-contained and, by the necessities of experimental design, somewhat isolated from other results-at least until a bigger picture begins to emerge. It is as though a huge

communal network of results is in the process of formation. (Will it become a metasystem in which each validated fragment serves as a schema?) Though metahypotheses are now emerging from analyses of experimental data, debates such as that between Krumhansl and Butler over the emerging model of tonal relationships serve to reinforce the observation that we cannot yet be sure of the overall shape of this network.7We are simplytoo early in the gradualdevelopment of the overall picture to expect clear directionsfrom cognitive research at this point. (4) Finally, as noted, the very model on which cognition researchersbase their work is the same as that of instructional designers-the behaviorist one. In fact, a research experiment generally involves a training situation in microcosm. The stimulus-response paradigm, done with repeated presentations, constitutes prototypical behaviorist training in itself. The unavoidable consequence is that research done within the same paradigms as that of traditional instruction simply cannot address issues which are outside those parameters-those which require stepping back from those paradigms to reexamine their very basis.
PERCEIVED MUSIC AND ILL-STRUCTURED DOMAINS

We have, however, found at least the beginnings of some answers among educational theorists and disciplineindependent instructional designers. Spiro's notion of the well-structurednessor the ill-structurednessof a content
7Fora few examples, see: Carol L. Krumhansl,"PerceptualStructuresfor Tonal Music," Music Perception1 (1983): 28-62; David Butler, "Describing the Perceptionof Tonalityin Music:A Critiqueof the Tonal HierarchyTheory and a Proposalfor a Theoryof IntervallicRivalry,"MusicPerception6 (1989): 219-42; Carol L. Krumhansl,"TonalHierarchiesand Rare Intervalsin Music Cognition,"MusicPerception7 (1990): 309-24; and David Butler, "Response to Carol Krumhansl,"Music Perception7 (1990): 325-38.

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

164

MusicTheory Spectrum

relevant here.8The principlein essence domain is particularly is this: the objectivist model may succeed in a domain which is inherently well-structured, wherein facts and procedures are highly rule-based so that little confusion can result from the attempt to transfer knowledge from one context to another. Examples of such domains are arithmetic or spelling of notated intervals. The problem comes where the domain is inherently ill-structured, where the presentation of new contexts creates a complexity which cannot easily be addressed by the well-structuredand somewhat rigid schematic facts and procedures developed by objectivist training. Some of the features of ill-structureddomains are: andsubsumption areinverted relations of dominance hierarchical fromcase to case; tend to often be misleading; prototypes of significance when assumedifferent the samefeatures patterns placedin differentcontexts;and interactions an explosionof higher-order amongmanyrelevant featuresintroduces aspectsof case novelty.9 To illustrate these points in regard to music as aurally perceived, let us reflect on the omnipresent debates in which we as music analysts engage. For example, rhythmic theoretic
8Rand J. Spiro, Walter P. Vispoel, John G. Schmitz, Ala Samarapungavan, and A. E. Boerger, "Knowledge Acquisition for Application: Cognitive Flexibility and Transferin Complex Content Domains," in Executive Control Processes in Reading, ed. Bruce K. Britton and Shawn M. Glynn Jehng, (Hillsdale, NJ: Erlbaum, 1987), 177-99; Rand J. Spiroand Jihn-Chang "CognitiveFlexibility and Hypertext: Theory and Technology for the Nonlinear and MultidimensionalTraversalof Complex Subject Matter," in Cognition, Education, Multimedia,ed. Don Nix and Rand Spiro (Hillsdale, NJ: Erlbaum, 1990), 163-205; and Rand J. Spiro, Paul J. Feltovich, Michael J. Jacobson, and Richard L. Coulson, "Cognitive Flexibility, Constructivism, and Hypertext: Random Access Instruction for Advanced Knowledge Acquisition in Ill-Structured Domains," Educational Technology 31, no. 5 (1991): 24-33. 9For a summaryof this position, see Spiro et al., "Knowledge Acquisition," 184.

discussions routinely become heated over issues such as the "predominance" or "subsumption" of variouslevels of metric hierarchy. Linear analysis regularlyreveals examples of hidden repetitions in which different levels of activity predominate at different times, or passages in which a prolongation of one primary tone may conflict with that of another on a different level. Moreover, certain low-level features can be misleading when considered as prototypes in relation to their contexts. If, for example, students were trained to recognize a given interval in relation to a particularcontext (for example, the major sixth as part of the NBC motto), their cognitive encoding of the interval sound and its name will be linked to that context. As noted above, research is revealing now that such single-contextual encoding creates rigid and even transfer-inhibitingschemas. In addition, varying contexts may alter the significanceof a given musical element. One need only consider the witticisms of Haydn in so many aspects of his style-phrase lengths, tonal designs, thematic restatements, and his pervasive experimentationswith formaldesign-to recognize the gamut of ways in which composers may utilize contextual differences as a means of manipulating listeners' expectations. And, if these contextual games can unsettle even the expert listener, how well can we expect the less-than-expert to do? On a more basic level, simplyvaryingthe metric placement of chromatic sonorities in a phrase can create enough change in significance to confuse many students. In sum, then, music as aurally perceived is far from fully predictable; it is what Spiro would call an ill-structureddomain. Indeed, if music as an art form were well-structured, it could not support the varied hearings and readings of the masterworks which are essential to its aesthetic value. Repeated hearings continue to generate varied responses for each of us, at least partiallydue to the "explosion of higherorder interactionsamong many relevant features"mentioned

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

Epistemology and Procedure in Aural Training

165

above. One might well ask if in fact the domain in which any art is expressed can be well-structured. Such a statement would appear to be oxymoronic.
AN ALTERNATE PARADIGM FOR KNOWLEDGE ACQUISITION IN ILLSTRUCTURED DOMAINS

the final product-that fully integrated set of prepackaged schemas-than on the process of applying preexistent knowledge in new situations. From the constructivist perspective, one may conclude (1) that theories in the cognitive sciences have produced a far better understanding of cognitive process in well-structured domains (WSDs) than in ill-structureddomains (ISDs); (2) that theories appropriatefor WSDs are in many ways inappropriate for ISDs-that, in fact, optimal conditions of learning and instruction in the two kinds of domains are opposite in several important respects; (3) that one of the most serious problems with training ISDs as if they were WSDs is an inability to establish a basis for knowledge transfer, for the application of preexisting knowledge to new situations; (4) that transfer in ISDs is best promoted by knowledge representations that possess the following features: multiple interconnectedness between different aspects of domain knowledge, multidimensional or multiperspectivalrepresentation of examples/cases, and allowance for various forms of naturally occurring complexity and irregularity; (5) that in order for knowledge structures to possess the characteristics described in 4, emphasis must shift from the retrievalof a precompiled schema to the assembly of a situation-sensitiveschema from knowledge fragments-the features described in 4 and 5 are characteristicsof what we call cognitiveflexibility; (6) that the best way to learn and instructin order to attain the goal of cognitive flexibility in knowledge representation for future application is by a method of case-based presentations which treats a content domain as a landscape that is explored by "criss-crossing" it in many directions, by reexaminingeach case "site"in the varying contexts of different neighboring cases, and by using a variety of abstract dimensions for comparing cases.12
12Spiroet al., "Knowledge Acquisition," 178.

This distinction between well-structuredness and illstructuredness reflects an alternative view of knowledge and its acquisition, one which addresses the features of illstructured domains. This view is commonly classified as constructivism. To a constructivist, the act of learning itself is not simply the receiving and encoding of information. Learning which is significant must be usable by the learner as different situations are encountered. In other words, the schematic network envisioned by constructivists is less static than dynamic, with each element connected in such a way that it can be disconnected from its original location and reassembled with other elements to apply to a new context. The goal thus shifts from schema selection to schema assembly.10 The role of prior experience is also significant. It is generally agreed among cognitive psychologists that learning does not take place in a vacuum, and that new information is integrated with that already present."1 In this sense, all knowledge is constructed as it is acquired. However, recent theory proposes that there is a second constructive aspect, namely, that the prior knowledge brought to bear in a new context is not simply retrieved intact; rather it is itself reconstructed specifically for the case at hand. Thus the goal for developing a learner's knowledge base should rely less on
178. ?OIbid., 1For a review of this notion, see Rand J. Spiro, "ConstructiveProcesses in Prose Comprehensionand Recall," in TheoreticalIssues in Reading Comprehension, ed. Rand J. Spiro, Bertram C. Bruce, and William F. Brewer (Hillsdale, NJ: Erlbaum, 1980), 245-78.

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

166

MusicTheorySpectrum in multiple ways reinforces the interconnectedness of concepts. For another, there is a certain front-end cognitive investment in becoming acclimated to a new context; where anchoredinstructionis employed, the need for such repetitive acclimationsis minimized. And thirdly, throughsituated cognition many subtasks can be addressed while leading toward a larger overall task, providing not only a lifelike variety of tasks but also the opportunityfor the learner to plan a strategy for completing the entire task, that is, to manage the
learning.
15

In line with constructivist thinking, therefore, decisions in instructionaldesign should shift away from identifying in advance the precise responses to be desired (that is, retrieval) and towardencouragingthe development of the learningprocess, including the means of supporting student-controlled learning. One such model with a long traditionin the musical world is that of apprenticeship. It was the primary means of theoretical training until well into this century, and of course still exists in performance and composition studios. The notion of cognitive apprenticeshipis now gaining favor among constructivistsas they recognize the type of teacherguided but learner-controlledexplorations which this model
provides.13

Constructivistmodes of instructionwhich are appearingin other content domains are variously termed "anchored instruction," "situated cognition," or "multiple criss-crossings of conceptual landscapes."14 They share a concern with learnenvironment contextual in a (either real or simufully ing lated), in order that the interconnectedness of the various features of that context are present to the learner at all times. This feature appears to improve the transferabilityof the knowledge acquired. Constructivistresearchers also demonstrate that spending a good deal of time in one context is highly beneficial. For one thing, dealing with one situation

13See,for example:Allan Collins, John S. Brown, and Susan E. Newman, "Cognitive Apprenticeship: Teaching the Craft of Reading, Writing and Mathematics,"in Knowing, Learning,and Instruction,ed. LaurenB. Resnick (Hillsdale, NJ: Erlbaum, 1989), 453-94; and Allan Collins, "CognitiveApprenticeship:Making Things Visible," American Educator 15, no. 3 (1991): 6-11, 38-46. and Technology Group at Vanderbilt,"AnchoredInstruction 14Cognition and Its Relationship to Situated Cognition," EducationalResearcher19, no. 3 (1990): 2-10; John S. Brown, Allan Collins, and Paul Duguid, "Situated Cognition and the Culture of Learning," Educational Researcher18, no. 1 (1989): 32-41; and Spiro and Jehng.

One clear advantageto a constructivistapproachis that the most essential problem of traditional training-that of developing knowledge transfer between varying contexts-is addressed directly. The very activities one performsin learning this way consist of transferitself-working extended types of exercises within a single context, only to confirm one's success by performingsimilaractivities in a different but similar context. If the contexts and activities are chosen wisely, a natural progression of maturing application of knowledge will emerge. In this manner, then, assessment is built into the design in two ways. First, the facts/skillsto be taught are tested as integral subtasks of solving the cognitive problems presented; the learner must both have successfully encoded the knowledge and also be able to apply it correctly to the context at hand. Second, transfer of knowledge is similarly assured, since the assessment is done in a manner which embeds transferinto the process itself, presentinga new context in which to perform the same types of skills which have been learned.

15TheCognition and Technology Group at Vanderbilt University conducted a series of mathematicalprojects in which multiple tasks were embedded in a "How do you get from here to there?"problem-solvingscenario. See their report "Technology and the Design of Generative Learning Environments," Educational Technology31, no. 5 (1991): 34-40.

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

Epistemologyand Procedurein AuralTraining 167 be aware of the interrelationshipsbetween the various elements of the musical fabric. And perhaps it could enable students to see how essential and vital is the role that aural training plays not only in their music-theory experience, but also across their entire curriculum-in private study, ensemble work, future conducting, teaching, coaching, and even in music analysis!

ANALYSIS TRAINING: THE ALTERNATE PARADIGM IN CURRENT USE

While the use of constructivistmethodology in the auraltrainingcurriculummay be a novel concept, it is actually not so novel to instruction in our discipline overall. Consider music analysistrainingof various types-formal, stylistic, linear. Here our critical tradition reflects the view that the domain is indeed complex and ill-structured.Multidimensional analyticaldecisions can lead to different interpretations, and in many cases there can well be more than one valid interpretation. The instructionalgoal in such situations focuses as much on the process of analysis as on the production of correct answers-on the rationale as much as on the final conclusions. And analysis can only be taught well by providing full contexts in which to work. Of course, the instructional methodology may well involve the provision of general guidelines for proceeding at the outset, but the focus is on the learner's control of the process of knowledge acquisition, on solving problems in context, and on the learner's revealing his or her mental processes to the instructor through providing rationales for the analytical decisions involved. Assessment too is done in context, by asking for an application of the learned analyticalskills in a similarbut distinctcontext. Therefore, to the extent that (1) we encourage students to do their own learning, (2) we allow for multiple possible answers (provided that sound reasoning supportsthem), and (3) we work from one context to the next in order to apply the knowledge base to new contexts, we are engaging in essentially constructivistmethodology when we teach analysis. Could we not, therefore, envision an appropriationof this constructivistmethodology, both from other domains where it is proving beneficial and from elsewhere in our own domain, for use in the aural-trainingaspect of our curricula? Perhaps such a change would better equip students for the kinds of tasks they will face once launched into their professional careers. Perhaps it could better train students to

FIRST STEPS TOWARD UNIFICATION

At the University of Kentucky, we have been gradually developing over the past several years an approach to aural trainingfounded on the constructivistparadigm.Our work to date has been informaland our reports can only be anecdotal. Yet we have succeeded in implementing a number of the features associated with Cognitive Flexibility Theory and anchored instruction. For example, in our music technology laboratory,we have employed a number of interrelated activities within a casebased training model. At the outset of this portion of the curriculum,we train students to focus on specific aspects of a musical fabric. One way this is accomplished is by providing a complete formal unit in its original texture, say a classical double period, on one track of a MIDI sequencer. We then ask students to extract individuallines from that texture by playing them back on their keyboard, recordingtheir performance on a separate track while they do so. The melody serves as a natural first focal point, but we then proceed to extracting the bass line-not an automatic skill for many students. Studentsrecordthese individuallines separately, on distinct sequencer tracks, so that the accuracy of their perceptual focus and memory can be verified. Certain cases may also lend themselves to other linear focusing as well. A multitimbral sequence, such as a brass solo with keyboard accompaniment, could lend itself to extracting not only the

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

168

MusicTheory Spectrum

brass line but also any complementary keyboard lines of interest. Once students display sufficient ability to discern individual lines-as measured by their success in extracting these lines-we progress to harmonic considerations. With the same context at hand, we ask that students determine the sonorities used in the piece. They demonstrate their understanding by recording the harmony in whatever manner and at whatevertempo they can physicallymanage. We encourage them to use timbral distinction between the sound they perceive from the original context and that with which they play their extracted sonorities, to facilitate their comparison of their own version with that of the original. Our evaluation of activities such as these has been enlightening. By allowing students the freedom to select not only their sonorities and inversions but also the harmonic rhythm itself, we often observe vastly different responses from different students. This variation in turn leads to fascinating class discussions about such issues. And these discussions of subjects such as harmonic rhythm, which had always seemed ratheracademicto our classes when run under an objectivist model, has taken on a new life when addressed in a fully contextual musical problem-solving situation. In addition, different combinations of extracted textural components can lead to still further exploration of the interrelationships between melody, bass, harmony, and texture. If, for example, students have recorded the bass and chords for a segment of a Mozart aria, they can then be asked to turn off the original context and compose a new melody appropriateto the now isolated bass and harmony. Or, they can study at the keyboard how to reharmonize an extracted melody or bass line, recordingthe results for later evaluation and discussion. As a means of involving the student in a manner which, while technology-based, is as intuitive as possible, we utilize a sequencingprogramwhich only records and plays back. No

editing of the digital data is permitted, nor is any extratemporal step recording. Thus we ensure that the responses we evaluate are ones that have been performed, again fostering a more direct link between aural training and the other aspects of the curriculum.And it is instructiveto observe students' learning strategies for the activitieswhich they do during class periods. The varied learning styles of different students are reflected not only in the results they achieve but also in their approaches to solving the problem at hand. Another means of incorporatingintuitive and familiar elements into auralinstructionis provided by our use of acoustic performancemedia. In a traditionalclassroom setting, we explore many of the above-mentioned tasks with students' own instruments,portable keyboards, or readily-learnedOrff xylophones. Working with a deliberately patterned context, such as that of a folk song or the Pachelbel canon, students explore the bass line, middle voices, melodic lines, and even improvisedmelodic lines of these harmoniccontexts in a sort of jam session. At such times they learn the interactions of the various lines by personally experiencing them and developing intuitive strategies for reacting and performing successfullyin real time. Carefullyselected archetypicalcontexts, such as the Pachelbel canon, also provide cognitive templates from which patterns may be extracted, varied, and presented again for recognition in other contexts. Thus a structurewhich requires the reassembly of prior knowledge in new situations is built into our curriculum. Yet another strategy which we employ extensively is to assign performancesto be done by pairs of students. These are designed to challenge students to cooperate musically as they develop performanceskills in melody, rhythm, and harmony. We develop their sensitivity to the line, rhythm, or sonority which their partner carries as a means of preparing them for the multiple situationsin which they will be required to perform professionallywhile listening and reacting to others' performancesaroundthem. Not only do we requirepairs

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

Epistemologyand Procedurein AuralTraining 169 of students to sing melodic and/or rhythmic duets together, but we also have them explore multiple interconnections in the harmonicdomain through accompanyingeach other with simple harmonizationsof sight-singing melodies, linearizing each other's harmonic successions, and consciously considering the influence of the harmonic domain on both foreground and middlegroundmelodic lines in transcribingmelodies or playing them back from memory in a variety of keys. Various types of musical problems can be addressed in a similarmanner. For example, compositional problem solving can be introduced by judicious choice of contexts to study. One instance, devised by our colleague Peter Simpsonfor use in our technology facility, presented a sequence from an Ives song, which incorporatedan alternating3 and 2 meter. Simpson required his students to record the melody in a straight 3 meter, indirectlydisplayingwhat strategies they were using. He had predicted a particular configuration as most likely, only to find that the actual result was almost universally different. Like us, he found himself in the midst of a lively and productive discussion of significantcompositional issues, one which could hardly have occurred in a traditional auraltraining environment. velop the type of knowledge representations which will be more cognitively flexible, or better able to transferfrom one cognitive situation to another, since they are learned themselves in context. Next, the use of case-based learning more closely follows models that succeed in other domains and that we employ in our own development of students' analyticalskills, providing a more unified approach to training across the musictheoretical curriculum.The use of full contexts, traversed in multiple ways, with study anchored in a single context to permitthe assimilationof all of its inherent interrelationships, is the hallmark of constructivist methodology. Finally, the issue of transferability of knowledge, one which is not even directly addressed in objectivist methodology, is embedded in our approachfrom the outset. All the activities of this type share the concept that what is learned from one context will be cognitively reassembled and reapplied as appropriatefor the next one. In that sense the methodology is identical with that for most analysis training, and in this manner the question of transfer of knowledge is answered by the very nature of the learning process itself.

NEXT STEPS? AURAL TRAINING BY THE CONSTRUCTIVIST PARADIGM

In sum, then, our approach serves to implement the constructivistmodel in several ways. First, it does not emphasize retrievalof factual knowledge. Instead, it requiresapplication of knowledge to new situations. Second, by working within the same complete musical context for extended periods, students gain an appreciation of the interconnectedness of the various musical parameterswhich cannot be gained from traditional context-free exercises. Further, emphasison the multidimensionalityof each case, together with an allowance for and even encouragement of varied responses, helps to de-

Therefore, in view of the difficulties which we all experience with aural training under its current methodology, in view of the call for change which comes from other domains, and in view of the fascinating responses we get from our students, we believe that the time has come for considering a significant paradigm shift as regards aural training.16We believe that a change in emphasis from the objectivist focus
16ThomasS. Kuhn develops the concept of the paradigm shift in The Structureof ScientificRevolutions, 2d ed. (Chicago: University of Chicago Press, 1970).

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions

170

Music Theory Spectrum

on information retrieval of static schemas to the constructivist focus on the flexible assembly of situation-specific schemas will better equip students with the type of knowledge which is available for ready application to the multifaceted situations with which they will have to deal spontaneously in their professional musical experience. We would call for more use of full musical contexts, both in terms of length and variability of performing media. We would call for sustained study within each of these contexts, criss-crossing the conceptual landscape to focus on the multifaceted nature of our art medium, encouraging students to develop a sense of the interconnectedness and interrelatedness of the various elements of music. Finally, we would call for a reconnection of the epistemology undergirding our teaching of music analysis with that of aural cognition, thereby presenting a unified approach to the study of music-theoretic concepts -a step which should in turn produce a more unified, better-structured, more flexible knowledge base in our students. Further, in line with this paradigm shift, we find the need for rethinking the actual goals of aural training and their implications for computer-based instructional design.17 For cognition researchers, might we also call for an analogous expansion of the paradigms of scientific inquiry? While it is true that study of individual responses to full contexts is difficult, such "messy" research has become widespread in

areas such as anthropology and women's studies. A further shift from an almost exclusively quantitative to a more triangulated evaluation process-including or even emphasizing the qualitative-may enable a wider scope of research projects to be undertaken.18 The optimal benefit would be that, rather than relying on the induction of cognitive processes from the study of its products, we could more directly address the processes themselves. ABSTRACT
This paper reflects a reexamination of the desired knowledge base of aural cognition, the theoretical basis for constructing it, and the means of evaluating the success of its training. Reasons for the frustrations generated by the gap between current cognition research and its applications are presented. The pervasive model of knowledge acquisition, one rooted in objectivist theory, is examined, and the limitations of the training effected through this model are noted. A distinction between well-structured and ill-structured content domains is introduced. Next an alternate, constructivist paradigm for training in ill-structured domains is proposed, and its advantages are cited. A further inconsistency in the epistemological/procedural relationship is cited regarding teaching of analysis vs. teaching aural cognition. The article concludes with a call for a broadening of the epistemological basis of all the interrelated areas of aural training, instructional design, and cognition research itself.

'7We have addressed these issues elsewhere. A revised set of aural development goals is outlined in: Kate Covington, "An Alternate Approach to Aural Skills Pedagogy," Journal of Music TheoryPedagogy 6, no. 2 (1992): 5-18; and Kate Covington and Charles H. Lord, "RethinkingAural Training," paper presented at the annualmeeting of the Society for Music Theory, Kansas City, 1992. Issues relating to courseware design for computer-based instruction are addressed in Charles H. Lord, "HarnessingTechnology to Open the Mind:Beyond Drill and Practicefor Aural Skills,"Journalof Music Theory Pedagogy 7 (1993): 105-17.

18For examples of the use of qualitativemethodology in our domain, see: Richard D. Ashley, "Computer-basedLearning: Models and Lessons for ComputerMusic ComputerMusic Systems," Proceedingsof the International Conference1989 (San Francisco:Computer Music Association, 1989), 9-12; Richard D. Ashley, "A Computer Programfor Analyzing Computer Musicians' Problem-solving,"Proceedings of the InternationalComputer Music Conference1990 (San Francisco:Computer Music Association, 1990), 14446; and Susan M. Piagentini, "Skill Acquisition in Undergraduate Music Theory: Analysis as ProblemSolving,"paper presentedat the annualmeeting of the Society for Music Theory, Cincinnati, 1991.

This content downloaded from 193.136.173.41 on Wed, 1 May 2013 06:28:59 AM All use subject to JSTOR Terms and Conditions