Webster (1992) has noted that formal or systematic assessment of children's compositions, in terms of both the process and the final product, has received relatively little attention. He suggests that advances in computer and sound technology, as well as software development, may promote more investigation of this important creative activity (p. 273). To date, it would seem that the use of technology in studies concerning children's compositions has been restricted, for the most part, to providing the tools necessary for completing the compositional task and, in some cases, recording the musical events of a composing session in a linear and passive manner. Younker and Smith (1996), for example, used sequencing and notation software combined with a digital keyboard as tools for a melody writing task, but assessment of the composition process was realized by means of verbal protocol analysis. Hoffman, Hedden, and Mims (1990) also used a digital keyboard and sequencing software to capture children's compositional activities. Assessment of the process was carried out by judges who categorized each five-second segment of recorded data with respect to the type of composing activity occurring in each segment.

Studies such as these provide valuable information about the process of composition that can be used as a basis for developing effective instructional strategies. The purpose of the present project has been to develop additional research tools for use in studies of children's compositions that not only provide the means for completing the creative task, but can also record additional information about the composition process. To date, two systems have been developed, both for use with children ages nine to eleven years of age. Both systems use HyperCard in conjunction with XCMDs and XCFNs contained in HyperMIDI 3.0 to coordinate user interactivity and sound management in the composition setting, and to capture and archive user data. Data archiving routines are invisible to the user, since they map on to operations needed to complete the composing task.

Each system has been modeled on an existing compositional study or group of studies. The first (called, "MelodyMaker") accommodates the procedures used by Kratus (1985, 1989, 1991) and replicated by Hoffman, Hedden, and Mims (1990). Working with students individually, Kratus asked each child to compose a new "song" that sounded good to him/her. Students used a small electronic keyboard to work out the composition. At the end of the ten-minute composing period, the student was asked to play his or her new song twice. Both the composing period and the final melodies were recorded on audio tape for subsequent analysis.

"MelodyMaker" is also designed primarily for monophonic composition studies, although multiple-part composition can be accommodated as well. As in the model studies, a digital keyboard is used to work out musical ideas. In addition, the user has access to several operations, via the computer screen, that mimic features found in a traditional composition setting. Just as a professional composer would record ideas s/he deemed of value by means of musical notation, the child using "MelodyMaker" can record up to four different "Neat Ideas." Each Neat Idea can be auditioned at any time by clicking on the appropriate button (which appear after a Neat Idea has been recorded) on the computer screen. When ready, the user may record the final product by clicking on the "Melody" button. All musical data obtained from the composing session are stored in both text and Standard MIDI File formats. Mouse clicks are also recorded with time stamps, thus enabling the researcher to reconstruct the session at some later time.

Figure 1. "MelodyMaker" screen. One Neat Idea has been recorded.

The Neat Idea feature is the most important addition relative to procedures used in previous studies for two reasons. First, it provides the user with a means of recording musical ideas that the student her/himself deems important in a manner which is similar to a traditional composition setting. The researcher now has the opportunity to study material the user has identified as important relative to the musical contents of the final composition product. That this information has been obtained in a fairly non-disruptive manner (the child simply clicks on the screen to record the Neat Idea) is viewed as a positive attribute of the information-gathering capacity of "MelodyMaker." Second, the recording of Neat Ideas is used in "MelodyMaker" to organize the archiving of musical data. In actual operation terms, the signal to record a Neat Idea triggers the storage of all musical information generated prior to the record signal as a single and separate file. In Figure 1, for example, all musical activity, from the start of the session up to the signal to record the first Neat Idea, has been stored as one file. Storage of ensuing data will occur when a second Neat Idea is to be recorded. Thus, whereas previous studies involved the assessment of an unstructured stream of musical information (usually divided arbitrarily into equal time frames during analysis), the structuring of the data stream in "MelodyMaker" can be said to be, at least to a certain extent, user-defined. With regard to both the composition process and the final product, the Neat Idea permits observation of the following.

"MelodyMaker" was used recently in a study involving 30 nine-year-old and 25 eleven-year-old children from two schools in London, Ontario, Canada. Using a digital keyboard and "MelodyMaker," the children were asked to compose a "new song," as in previous studies by Kratus (e.g., 1989, 1991) and Hoffman, Hedden, and Mims (1990). "MelodyMaker" was used to record the entire composition session, as well as final melodies. Although the Neat Idea feature was demonstrated to each child, its use was optional.

In all, 21 of the 55 children (twelve nine-year-olds and nine eleven-year-olds) recorded Neat Ideas. The Neat Idea recordings ranged in length from 6 to 55 seconds, with an average length of 21.8 seconds. The children tended to record complete musical ideas (either whole phrases or, in many instances, complete melodies) as opposed to melodic fragments. No recordings suggested an attempt to work separately with either pitch or rhythmic material. Motivic material contained in the final melody frequently appeared in one of the first Neat Ideas, and development of these ideas could be observed in subsequent recordings. This was particularly noticeable with the children who used all four opportunities to record a Neat Idea. Figure 2 describes the first two (of four) Neat Ideas, as well as the final melody, recorded by one of the eleven-year-old students in the study, and illustrates the development of musical ideas observed in the work produced by many of the children. The first Neat Idea contains a dotted rhythmic figure, as well as sequential treatment of a predominantly ascending motivic figure. In Neat Idea 2, a descending pitch pattern is added. The third and fourth Neat Ideas involve elaboration upon the material identified in the earlier recordings. The final melody exhibits all of the motivic elements found in the Neat Ideas. By analyzing the music data that preceded and followed each Neat Idea recording, it was possible to trace how motivic structures were initially shaped and subsequently altered. Identification of which structures to focus on was determined by elements found in the Neat Ideas recordings. In other words, identification of important musical ideas was guided by information provided by the user.

Figure 2.

The second composition system ("TuneBlox") is based on Bamberger's (1977) study of decision-making processes in melody writing. In her study, Bamberger also used a computer-based composition system with which users created melodies by arranging a series of five "tune blocks" (melodic fragments) into various orders to produce a complete melody. The tune blocks were sounded on a small synthesizer by typing two-character codes (one code for each tune block) on the computer keyboard. The tune blocks themselves could not be altered, but could be combined in any arrangement, including repetition of the same block. Subjects could listen to and/or change the arrangement of tune blocks as often as desired. Working individually with eight college students, Bamberger used protocol analysis in recording how each subject assembled and re-assembled tune block combinations in constructing a melody. Based on this study, Bamberger formulated some important considerations concerning the decision-making processes she observed and the manner in which subjects formed and manipulated their representation of a melody.

"TuneBlox" accommodates all of the melody-building requirements found in the Bamberger study. All operations needed to construct a melody are available via the computer screen. There are five tune block buttons, each of which–when clicked–will cause a corresponding melodic fragment to be sounded. Sixteen boxes in the middle of the screen are used to assemble the melody. After clicking on a numbered button, the user clicks on one of the boxes; the selected button number appears in the box. The "Play Tune" button can be used at any time to hear the arrangement of tune blocks as ordered in the boxes. An "Eraser" button is used to remove numbers from boxes should the user decide to alter the order. When the user is satisfied with the melody that has been constructed, s/he clicks on the "Final Tune" button to hear the finished product and to automatically quit the program.

As with "MelodyMaker," all mouse clicks are time stamped and archived. In addition, the current order of the tune blocks, as assembled in the melody boxes, is recorded each time the "Play Tune" button is accessed. These two data storage features enable the researcher to re-construct the sequence of user actions at some later time and to trace the development of the melody.

Figure 3. "TuneBlox" screen.

"TuneBlox" was also used in the same study referred to earlier in connection with "MelodyMaker." Working individually, each of the 55 children in the study was asked to construct a melody using "TuneBlox." The number of blocks used to form a melody ranged from five to sixteen. Indeed, a considerable number of the students (fourteen) used the maximum number of tune blocks (sixteen) in combination in forming a melody. This would suggest that at least some of the children may have felt restricted in completing the task by the limited number of possible block combinations.¹

In working with eight college students, Bamberger (1977) was astonished "to discover the varieties of strategies and levels of engagement" among the subjects (p.286). A range of different tactics was also observed among the children using "TuneBlox," although certain trends did emerge. For example, reviewing melody construction based on listening procedures (i.e., use of the "Play Tune" button) suggested at least one differentiation in approach between the two age groups. The eleven-year-olds tended to start by combining a small number (two or three) of tune blocks, listening, and then building the melody gradually by adding one or two blocks at a time. The nine-year-olds, on the other hand, tended to start with a larger number of blocks in combination, and then add large numbers of blocks before listening to the result. Most children in both age groups were quite decisive in tune block selection. After listening to all five blocks at the outset of the session, little time was spent comparing fragments during melody construction. The "Erase" button was used infrequently. The children seemed reluctant to remove a tune block once it had been installed in the melody, and, instead, would focus on extending the tune based on the current combination of blocks. Students tended to work from the beginning of the melody through to the end (i.e., in one direction only); there was little in the way of skipping back and forth to deal with tune blocks in isolation. Indeed, the most common "working unit" (Davidson and Welsh, 1988) seemed to be those consecutive tune blocks that, for the moment, were at the end of the melody.

DeLorenzo (1989) notes that music programs have varied considerably considering the kinds of music problems regarded as "creative." She also notes that the structure of a problem will have a bearing on the "creativeness of the problem-solving process and resulting product" (p.195). Identification of meaningful problem structures, coupled with an understanding of how children respond to such tasks is essential in developing music curricula that demonstrate an effective approach to creativity. It is hoped that technological devices such as "MelodyMaker" and "TuneBlox" can contribute to this identification process by profiling student response to a variety of creative tasks in such a way as to provide additional information about problem-solving behavior that was not previously available.

______________________________

¹. In the Bamberger study, there was no limit as to the number of blocks that could be used in melody construction.