John M. Holahan, Yale University
T. Clark Saunders, University of Hartford

This presentation is designed to illustrate how we have used microcomputer technology to conduct carefully designed studies of tonal cognition in schools and other non-laboratory settings with young children and adults that would otherwise be impractical to carry out. Using that technology, we have been able to (a) construct a large taxonomic database of tonal pattern pairs and their characteristics for the construction of batteries of microcomputerized tasks, (b) create software for presentation of stimuli and recording subjects' responses, and (c) maintain and analyze the data derived from the research.

Research bearing on aspects of the cognitive representation of pitch, tonal patterns, and melodies has been conducted by cognitive psychologists, music theorists, and music educators using a variety of research methods and techniques, typically with young adults serving as subjects. Studies of musical pitch perception from the discrete perspective are those studies of the mental representation of pitch summarized in Krumhansl (1990). Cognitive studies of music decision making have been designed to examine aural discrimination and decision making processes using tonal patterns, rhythm patterns, and brief melodies with standard chronometric (RT) methods (Fiske, 1992).

In contrast to the cognitive studies cited above, Gordon's (1974, 1976, 1978) taxonomic studies of tonal patterns and rhythm patterns were designed with attention to pedagogical concerns. Specifically, Gordon sought to determine the aural recognition difficulty levels and growth rates of tonal patterns and rhythm patterns by schoolchildren in grades four and eight. For the tonal studies, pairs of "same" tonal patterns were presented following a sequence of tones designed to outline a key and tonality. The relative difficulty level of the patterns was taken to be an indication of the ease with which the patterns are retained as mental representations by young children.

Gordon's (1993) use of the term "audiation" serves both as a definition of the representational "inner hearing" process in music and further, as a general description of the cognitive processes that give music its syntactic structure. Audiation describes the mind's ability to perceive, retain, compare, synthesize, reproduce, and create tonal patterns and rhythm patterns. Although Gordon's taxonomic studies were designed to guide the selection of tonal pattern and rhythm pattern content most appropriate for instruction, they also provide indirect evidence of the cognitive mechanisms involved in the retention and comparison of tonal patterns. In contrast, the aforementioned studies of pitch representation were designed to describe aspects of the cognitive structure of pitch representations.

The purpose of our research is to extend the taxonomic understanding of tonal pattern audiation by identifying the cognitive processes involved in the aural recognition of sameness and the identification of difference when listeners attend to pairs of tonal patterns. Our research strategy combines standard chronometric reaction time (RT) methods typical in cognitive psychology with the content and methods used in Gordon's (1974, 1976, 1978) taxonomic studies of tonal patterns and rhythm patterns.

The number of tonal patterns that can be constructed from a finite set of pitches such as the diatonic set in major tonality is indeed large. To achieve a finite set, constraints must be applied to the selection of tones for a given pattern (e.g. harmonic function, melodic direction, pitch range, number of tones in the pattern, exclusion of repeated tones). Gordon's (1974, 1976, 1978) taxonomic studies provide substantial empirical evidence of the relative difficulty of the aural recognition of sameness of tonal patterns in many tonalities (major, minor, dorian, mixolydian, phrygian, lydian, and aeolian). Because we are interested in the cognitive processes involved in the aural recognition of sameness and the identification of difference, the pattern content and empirical findings of Gordon's (1978) final taxonomic study of patterns in major tonality were used as input to a database to create a taxonomy of both "same" pairs and "different" pairs of tonal patterns in which the pairs have the same number of tones (from 2 to 5 tones). The database (R:Base, 1992) consists of a description of the individual tonal patterns characteristics (pitch content, harmonic functions, item difficultly levels (both proportion correct and classification [i.e. easy, moderate, and difficult], and growth rate) from Gordon's (1978) study. Furthermore, the database includes the population of all possible pairs of "same" and "different" three-tone tonal patterns in major tonality from the taxonomy and their item characteristics (i.e. the relationships between the patterns of each pair [similarities and differences of harmonic function, melodic direction, and pitches in common]) (see Table 1). Those item characteristics have been used to select item content for criteria measures in three investigations (Holahan & Goldberg, 1994; Holahan & Saunders, 1994; Holahan, 1994) designed to obtain valid measures of the speed and accuracy of "same" and "different" judgments made by young children and adults to sets of tonal pattern pairs that vary systematically in terms of their characteristics.

—————————————————————————————————————

A. The Structure of the Database

Part I Data from Gordon (1978) for each pattern

1) Classification

2) Category

3) Patterns (notation converted to Pitch names)

4) Pattern Difficulty (proportion correct and level (easy, moderate, and difficult)

5) Growth rate (positive, static, and negative)

Part II Additional descriptive information for each pattern

1) Melodic intervals (in semitones) from pitch to pitch in the pattern

2) Pattern contour (direction of the melodic intervals within each pattern)

Part III Characteristics of pattern pairs (items)

1) Number and position of identical pitches in the same relative position in the pair

2) Index of commonality of melodic contour between the two patterns

3) Function (category) of the two patterns (same or different)

4) Composite a priori item difficulty (based on pattern difficulty levels)

B. Item Content Currently in the Database

Data for tonal patterns in Major tonality

Number of Tones Number of Patterns

2 87

3 277

4 18

5 6

Number of 3 tone item pairs= 38503= (277 * 278)/2

——————————————————————————————————

The customized software ICATS (Holahan, 1992) was written specifically for test administrations using tonal patterns extracted from the taxonomic database (Saunders & Holahan, 1993). The executable program is compiled from source code written in Turbo Basic (version 1.0) and runs on IBM compatible computers using DOS version 3.3 or later. The tonal pattern items are appended as data at the end of the source code and become part of the .EXE file. The tonal pattern pairs (test items) are presented to the subject through the built-in speaker of the microcomputer. The subject indicates a "same" or "different" judgment by pressing a "same" key (<---) or a "different" (--->) key. The subject's choice is indicated on the computer monitor by the appearance of either (a) a pair of "same" smiling faces when the "same" key is pressed or a pair of "different" contrasting smile/frown faces when the "different" key is pressed (for young children), or (bs) the word "same" when the "same" key is pressed or the word "different" when the "different" key is pressed.

Through a series of practice trials, the subject is given instructions to (a) produce a pair of smiling faces (or a "SAME" response) by pressing the "same" key and to produce the smile/frown faces (or a "DIFFERENT" response) by pressing the "different" key with the index finger of the preferred hand; (b) listen and then indicate whether the two patterns are "same" or "different" by pressing the appropriate key; and (c) listen and respond as soon as the answer is determined for each item. The subject is reminded throughout the practice trials that only one choice can be made for each item and to return their index finger to a finger rest between the two response keys. Pairs of patterns, not included in the test portion of the task are used in the practice trials. The test portion of the program is presented immediately following the practice trials. The computer records the child's choice (same or different), determines whether the response is correct or incorrect for the given item, and measures the time required for the response (in milliseconds) measured from the onset of the first pattern of the pair until the moment the response key is pressed. The outline of the program is presented in Table 2. The general structure of the program follows the pattern suggested by Dowling (1989), with minor differences. First, our stimuli are presented in the format of a conventional standardized test, with test items occurring in the same order for every subject. Therefore, a procedure for stimulus randomization is not included in the program. Second, in the practice and test trials, our program provides feedback to the video screen indicating whether the subject has responded "same" or "different." Third, our tonal patterns are presented through the internal speaker of the computer, rather than through specialized sound synthesis equipment. Fourth, response time in measured by the built-in internal clock rather than by specialized timing equipment.

We recognize that the task program is relatively simple when compared to such systems as MIDILAB (Todd, Boltz, & Jones, 1989). We have found, nonetheless, that the tasks administered on our unadorned laptop computers have functioned with subjects from preschool age through young adult undergraduates. The portability, ease of use, and minimum expense of the system have met our immediate research needs. Furthermore, the content and design of the pattern database makes possible the construction item pools for use on any future sound generation system, MIDI or otherwise, because the pattern data can be translated into any pitch-naming convention prior to export from the database. As we continue our cognitive studies, we look forward to utilizing the relatively new portable sound capabilities that will enable us to bring high-quality-sound stimuli out of the laboratory and into the field and classroom for research.

——————————————————————————————————

Main Program

Read number of Items N (first line of data block)

Dimension Data Arrays for number of items

Subroutines

Call Title Screen

Call Login (Subject Data)

Call Practice

Read Test data

Call Test

Call Output Data

Subroutines - Description

Login - Subject Variables - ID, Name (First, Last), Gender, Grade/Class, Test administration number, Date of birth, Date of test, hand used for responses

Practice - train association of response keys to "same" and "different,"

Practice Items

Test

Loop Structure For items 1 to N

Start Timer

Play Pattern 1 Silence Pattern 2

Read Keyboard Buffer for response

Stop Timer on Response

Display faces or word that corresponds to key pressed

Check whether response is correct (1) or incorrect (0)

Subtract time end from time start (in milliseconds)

Loop until item N

Item data (Pitches and item characteristics)

——————————————————————————————————