lectura 8 lasso an intelligent computer based tutorial in sixteenth century counterpoint

LASSO: An Intelligent Computer-Based Tutorial in Sixteenth-Century CounterpointAuthor(s): Steven R. NewcombSource: Computer Music Journal, Vol. 9, No. 4 (Winter, 1985), pp. 49-61Published by: The MIT PressStable URL: http://www.jstor.org/stable/3679622Accessed: 25/02/2010 12:40

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/action/showPublisher?publisherCode=mitpress.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Computer MusicJournal.

http://www.jstor.org

http://www.jstor.org/stable/3679622?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/action/showPublisher?publisherCode=mitpress

Steven R. Newcomb Center for Music Research Florida State University Tallahassee, Florida 32306 USA

The Need for a Computer-based Tutorial in Sixteenth-Century Counterpoint

The LASSO tutorial in sixteenth-century counterpoint is a software system intended to provide an efficient learning environment. The project was un- dertaken in order to answer the need for new means of building skills in the compositional discipline of counterpoint (Newcomb 1983). There is no sub- stitute for training that requires neophyte composers to deal with the problem of multiple voices according to rules that (1) ensure an agreeable motion and form within each voice, (2) maintain a bal- ance of importance among all the voices, (3) keep all voices in independent motion, and (4) set the voices free from slavish dependency on prearranged chord changes.

It is emphatically not the purpose of the LASSO system to imbue undergraduates with an apprecia- tion for systematic musicology, or to turn students into composers of irrelevant sixteenth-century music. The purpose is not even necessarily to give them an absolutely accurate account of sixteenth- century style, although the rules are intended to do that as much as is consistent with the overriding objective of simplicity and consistency of the judging criteria. The proper purpose of sixteenth- century counterpoint instruction in the twentieth century is to acquaint young composers with one kind of intellectual exercise that Western composers have found valuable for centuries. The best way of providing this kind of training is still through the use of some version of the venerable "species" counterpoint method first introduced in the early eighteenth century (Fux 1725). Like Fux's Gradus ad Parnassum, LASSO is a latter-day attempt to make instructional sense of the style by codifying

Computer Music Journal, Vol. 9, No. 4, Winter 1985, ? 1985 Massachusetts Institute of Technology.

LASSO: An Intelligent Computer-based Tutorial in Sixteenth-Century Counterpoint

its elements in a fashion that a scholarly composer or musicologist might feel amounts to an over- simplification, or even a misrepresentation.

Several problems are commonly encountered by music teachers who provide species counterpoint instruction. LASSO is intended to answer or at least ameliorate the issues discussed in the following paragraphs.

First, few professional composers (or musicolog- ists, for that matter) can agree on a particular species counterpoint text, and they generally end up choosing a text they consider the best of a bad lot. It is quite difficult to write a textbook on species counterpoint; problems result from the conflict between the intended purpose of counterpoint instruction (to develop compositional skill) and the normal academic procedure used when teaching any historical subject (examination of original sources, such as the works of Palestrina, Victoria, Lassus, et al.). Some scholarly texts tend to be long on historical examples and short on simple advice for students (e.g., Jeppeson 1939); others, presum- ably in the interest of conciseness and clarity, fail to account for all the possibilities (e.g., Krenek 1959). The Schoenberg text (1963), while having the virtue of communicating the general features of the style by means of hierarchies of stylistic acceptability, can be a nightmare for a student trying to make a list of rules by which he can pass a course. While in many respects the Soderlund (1947) text appears the most useful for classroom purposes, students sometimes find even it confusing and self-contradictory, due to the fact that some of the general considera- tions outlined in the introductory section do not apply to certain species that are introduced later.

Regardless of which text is being used, for each homework assignment the instructor must explain to the class what will be accepted and what will be rejected in student exercises despite (or in addition to) what is written in the text; to fail to do so is to

Newcomb 49

invite mass consternation. Some traditional counterpoint instructors do fail to provide this humble and specific kind of guidance. Students are discour- aged when their exercises are shown to be laden with stylistic transgressions that they could have avoided if they had understood the context of the assignment better. It is all too easy for an experi- enced music theorist or musicologist to take the obvious for granted, or to dwell on details and "in- teresting exceptional instances" in the works of the masters, at the expense of instruction in fundamen- tals. If for any reason students fail to grasp one or more of the basic concepts or rules, and if they have not been provided with explicit guidance regarding which rules are to be applied to a given exercise, they are often left with the feeling that there is lit- erally no way to write an exercise that is truly acceptable. A tested computer program such as LASSO, capable of pointing out stylistic deficien- cies in student exercises, necessarily embodies an explicit and achievable set of objective and know- able criteria.

Second, no text can be expected to serve as a sole source of guidance to students. The purpose of counterpoint instruction is to inculcate a way of thinking about and composing music, but no textbook is capable of examining student work and pointing out errors. Often graduate music students are found to have an inadequate background in species counterpoint, and they are required to re- mediate themselves. Armed with a good text and LASSO, these people can avoid having to take an undergraduate course for which they cannot receive graduate credit.

Third, for maximum effectiveness, counterpoint instruction demands small classes. Only in the context of an intimate group of interested students can each student's work be examined by the class. The instructor's role is to point out the problems in each exercise and to encourage all the students to solve them. In most institutions, it is impractical to offer an undergraduate course with an enroll- ment limit of seven or eight. In a large class, LASSO can be used to simulate the small class tutorial environment.

Fourth, if the class must be large, it is impractical to examine each student's work during class time,

and the sheer tedium of marking many student exercises (not to mention the quantity of information that an instructor should write on student compositions) militates against adequate guidance from the instructor. If homework assignments are made via LASSO, the students are certain to receive flawless and complete guidance. Moreover, with LASSO, the instructor is in a position to demand perfection, and the student is in a position to achieve it, taking as much time and submitting as many exercises as necessary. The organizational problems involved in teaching a counterpoint class (e.g., "if this is the fourth week we must be using the second species rules for last week's overdue homework") are minimized with LASSO.

Building the LASSO System

An instructional music editor, Ottaviano, was the first software module created for the LASSO system (Newcomb 1978). The music editor had to be easy for students to understand and use. It had to be able to support four voices (only two are currently used in LASSO), pieces up to 50 measures long, and im- mediate playback via computer-controlled music synthesizer. In addition, it had to work efficiently enough for large numbers of students to use it without sandbagging the timeshared computer. The current version of Ottaviano met or exceeded all these specifications in 1981.

The next software module written was the system that supports student records, instructor operations, and other housekeeping chores and utility displays. This system serves as the user entry point to the entire system and as the transfer mechanism between all the separate programs (editor, help display, judging, etc.) among which the student un- knowingly shuttles. A large database of musical examples, exercises, and help sequences, and the personal records of a large number of students, can be accommodated.

Finally, a list processing system for judging operations was added, together with the rules themselves and their associated judging routines and help sequences. This was the heart of the project. (The text portions of three rules are reproduced in Table 1. For

Computer Music Journal 50

Table 1. Three representative rules supported by the LASSO system, chosen mainly for the brevity of their help pages (in keeping with the space limitations on this article). For a complete list, see Newcomb (1983).

Rule 19:

Mnemonic title of rule: CERTAIN MELODIC INTERVALS MUST NEVER OCCUR

Brief description: Augmented and diminished intervals, compound intervals, sevenths, and descending sixths are not permitted.

Help page 1: In sixteenth-century counterpoint, there are NEVER any melodic intervals of the following kinds:

Augmented or diminished intervals, including the tri- tone (which may be described as an augmented fourth or as a diminished fifth)

Compound intervals (intervals larger than a perfect octave)

Sevenths of any description Descending major or minor sixths

Press NEXT to see the definition of melodic interval. Help page 2:

A melodic interval is the interval between two adjacent pitches in a single voice. Another kind of interval is frequently mentioned in this lesson: the harmonic interval. A harmonic interval is the interval between two simultaneous pitches sung by two different voices. A melodic interval, then, is the interval between two successive pitches sung by a given voice, while a harmonic interval is the interval between two pitches which are heard at the same time.

Help page 3: Hint: If you have written a skip of an augmented fourth or a diminished fifth, see if you can avoid it by flatting a B.

Applicability of this rule: The first time this rule appears is as a new displayed rule, in the first exercise group category. It appears in all subsequent exercise group categories as an old displayed rule.

Error message(s) generated by the judging routines for this rule:

This melodic interval is NEVER used: [name of melodic interval student wrote].

Default penalty applied to student exercises per infraction:

42 points. Discussion:

This rule covers more ground than might be expected at first glance. For example, a student who writes a melodic line using an unraised leading tone followed by a raised leading tone will be reminded that augmented unisons are NEVER used.

Rule 59:

Mnemonic title of rule: LIMIT OF 3 ADJACENT PARALLEL 3rds and 6ths

Brief description: Do not use more than three successive (adjacent) harmonic thirds or sixths in any given passage.

Help page 1: For the time being, limit your use of parallel thirds and sixths to three in succession. Later, this limit will be relaxed, but it is never a good idea to overdo the use of parallel intervals (even thirds and sixths!), since parallel motion lessens the independence of the voices, and weakens the counterpoint.

Applicability of this rule: This rule appears in the rule lists for five exercise group categories. The first time this rule appears is as a new displayed rule, in the second exercise group category; it subsequently appears as an old displayed rule in the third, fourth, fifth, and seventh categories.

Error message(s) generated by the judging routines for this rule:

Too many adjacent parallel ["thirds" or "sixths"]. You were penalized for [excess number] of the [total number] adjacent ["thirds" or sixths"].


5.5 points for each excess parallel third or sixth in each succession of parallel thirds or sixths.

Discussion: The limit of three is fairly strict, since the database of examples which were analyzed (see Table 2, below) typically contained passages of four or five parallel thirds and/or sixths. The reason for the strictness of the limit is that parallel thirds and sixths can become a crutch for students who look for an easy way to complete an exercise.

Rule 100:

Mnemonic title of rule: WRITE UNACCENTED QUARTER NOTE PASSING TONES

Newcomb 51

Table 1. (cont)

(This title is displayed on rule lists, and when the rule has been fully satisfied in the exercise group in which it is introduced, as the rule associated with the congratulatory message.)

Brief description: Each of these exercises must contain at least one unaccented quarter note passing tone dissonance treatment.

Mnemonic title of rule: UNACCENTED QUARTER NOTE PASSING TONE PROBLEM

(This title is displayed only when an ill-formed dissonance, which LASSO guesses is an attempt at an unaccented quarter note passing tone, is detected.)

Brief description: LASSO has guessed that you were attempting to write an unaccented quarter note passing tone dissonance.

Help page 1, accessible through rule list display or from an error message:

At least one unaccented quarter note passing tone dissonance must appear in each of these exercises. A major goal of this exercise group is to make you proficient in the use of this dissonance type.

(Student may press LAB at this display to see the detailed help lesson described below.)

Help page 1, accessible only when this rule is fully satisfied:

Congratulations! You have written a correct unaccented quarter note passing tone dissonance treatment.

(Student may press LAB at this display to see the detailed help lesson described below.)

Detailed help lesson, page 1: Dissonant quarter note passing tones which appear on the second half of any beat are called "unaccented quarter note passing tones." They are called this in order to distinguish them from "accented" quarter note passing tones, which may appear on the first half of the weak beats only (see "accented quarter note passing tones").

Detailed help lesson, page 2 (music display not reproduced here):

Here are some correct unaccented quarter note passing tones.


THESE ARE DISSONANT UNACCENTED QUARTER NOTE PASSING TONES. An unaccented quarter note passing tone may be dissonant or consonant. The following rules deal only with dissonant unaccented quarter note passing tones. All of

the above unaccented quarter note passing tones are dissonant.


PRECEDE BY QUARTER NOTE OR DOTTED HALF. An unaccented quarter note passing tone must always be preceded by another quarter note, or by a dotted half.


PRECEDE AND FOLLOW BY CONSONANCES. The unaccented quarter note passing tone dissonance must be preceded and followed by consonances.


PRECEDE AND FOLLOW BY STEP IN THE SAME DIRECTION. The unaccented quarter note passing tone must be preceded and followed by step, and both the preceding and following step must be in the same direction (both ascending or both descending).

Applicability of this rule: The rule "WRITE UNACCENTED QUARTER NOTE PASSING TONES" appears as a new displayed rule in the seventh exercise group category. The rule listings for "Congratulations .. ." and ". . . PROBLEM" are hidden rules in the same category.

Error messages generated by the judging routines for this rule:

You must write at least one correct unaccented quarter note passing tone dissonance. You didn't. This is a correctly handled unaccented quarter note passing tone dissonance. Bravo! Unaccented quarter note passing tone? The dissonant note is not preceded by a quarter or a dotted half. Unaccented quarter note passing tone? The dissonance is not preceded by consonance. Unaccented quarter note passing tone? The dissonance is not followed by consonance. Unaccented quarter note passing tone? The dissonant quarter note is not approached by step. Unaccented quarter note passing tone? The dissonant quarter note is not left by step. Unaccented quarter note passing tone? The dissonant quarter note is not approached and left in the same direction.


100 points.

Computer Music fournal 52

a complete list, see Newcomb 1983.) It was necessary first to make graded lists of specific rules based on objective criteria that could be communicated simply and easily to the student.

A number of standard texts were used as guides (Jeppeson 1939; Krenek 1959; Schoenberg 1963; Soderlund 1947). The Soderlund text proved the most useful in outlining a procedure by which students should be introduced to counterpoint. For each rule, an algorithm for detecting infractions (and, in the case of newly introduced dissonances, instances where all criteria are met in an exemplary fashion) of the rule in student exercises, for setting pointers to the exact notes and/or rests involved, and for concatenating sensible remarks in good En- glish about the rule infractions (or observances) was developed and rendered into software. When the software for a judging routine was finalized, a simple explanation of the criteria by which it judged student exercises was written into the help sequence for the rule. It is obvious to students using LASSO why they are penalized for some infraction of a rule; they are shown a clear explanation of exactly what the judging routine was designed to detect, often with graphic and audible musical examples.

LASSO's judging feature may be a poor example of knowledge engineering, since the representation of each rule consists mainly of software that is in- separable from the judging engine. It was tedious and repetitious to represent the rules in this fashion, but there was no other way given the technical constraints of the PLATO system and its unique hodgepodge of a computer language, PAL (nee TU- TOR). However, even though LASSO does have an execrable knowledge representation scheme, it may be considered "artificially intelligent" because it is able to guess, for example, which dissonance treatment a student was probably trying to write even when the treatment is ill-formed (and incorrect), and it is able to give appropriate advice on the basis of its guesses.

There was no obvious way to represent certain features of the style in software. No one had quan- tified the notion of secondary rhythm, for example, or even fully described the elements of that feature of the style. Therefore, for some rules, it was necessary to apply computer analysis to a body of lit-

Table 2. The answers to these questions were sought by means of computer-aided analysis of the thirty compositions written for two voices in 4/2 meter contained in the Soderlund and Scott Examples (1971).

1. What is the allowable range of ratios of skip to non- skip melodic intervals?

2. What is the allowable incidence of ascending major and minor sixths and ascending and descending perfect octaves?

3. What is the allowable incidence of compound harmonic intervals?

4. How much time can be spent with the voices crossed? 5. How long may a passage of parallel thirds or sixths

continue? 6. How long may a melodic (or harmonic) passage con-

tinue with all the long notes appearing on odd- numbered beats (or all appearing on non-odd beats)?

7. How often may pairs of white notes of equal (or un- equal) value sound simultaneously?

8. How large can the differences in the distributions of note values between the two voices be?

9. How many eighth note passages can be expected to be found in a piece of a given length?

erature to extract the specific information that was needed (for example, exactly how many compound intervals should be permitted between the two voices in a two-voice composition). The body of lit- erature used consisted of the thirty compositions written for two voices in 4/2 meter contained in the Soderlund and Scott Examples (1971). A list of the questions that this research answered is given in Table 2. After this research, considerable com- mon sense still had to be applied in order to represent a reasonable set of rules in the software.

Instructional Design

Each graded exercise group consists of a text description, a list of rules for which students will be held responsible, and the exercises themselves. Ad- vancement from one exercise group to the next is controlled by LASSO according to criteria set by the instructor. At the main index, students may (1)

Newcomb 53

read the description of the exercise group, (2) study the list of rules with their associated helpful expla- nations and/or other displays, or (3) work on the exercises themselves. Students may do these activities in any order, and they may move between them at will.

LASSO permits instructors to create and individ- ualize exercise groups within each of the seven exercise group categories, and to create exercises within each group. They may also specify class file names, give various persons and classes various kinds of access to LASSO, inspect and print out student records, specify the criteria whereby students may pass from one exercise group to the next, specify whether certain students will work with two or four different clefs (both the help lessons and the exercises themselves are affected by this specifica- tion), and set the penalties to be applied for infractions of the various rules.

There are seven graded categories of exercise groups. The arrangement is most similar in spirit to Soderlund (1947), although certainly all texts on the subject of sixteenth-century counterpoint owe much to Fux (1725), who originated the idea of organizing the presentation of the basic concepts of the style by means of exercises in five increasingly difficult "species." Indeed, the second and fourth exercise group categories in LASSO are virtually identical to Fux's first and second species of counterpoint. The first exercise group category deals with melody in "white" notes (i.e., breves, whole notes, and half- notes); the second with whole notes in two voices; the third with whole notes and half-notes in two voices in a rigid rhythmic pattern; the fourth with all white note values in two voices in no particular rhythmic pattern, including half-note passing tone dissonances; the fifth with all white note values in two voices in no particular rhythmic pattern, including both half-note passing tone and suspension dissonances; the sixth with all note values in one voice only; and the seventh with two voices and six additional dissonances, including accented and unaccented quarter-note passing tones, eighth-note passing tones, quarter- and eighth-note auxiliaries, and the nota cambiata.

The rules accumulate as the student proceeds from one exercise group to the next, although some,

those used temporarily for instructional reasons, are later withdrawn, and their replacements have help sequences that explicitly declare the earlier rules to be inoperative.

Each exercise group has associated with it an internal rule list, which is used by LASSO to deter- mine which rules are to be displayed as old, and which are to be displayed as new. The internal rule lists also contain listings for "hidden" rules that do not appear on the displayed rule lists at all. One reason that a rule may be hidden from the student is that the internal lists control the judging process, and some of the entries merely initiate judging procedures that do not have rule descriptions associated with them. (Another reason for hiding a particular rule from a student is that it may be in- sultingly obvious to mention it, as, for example, the rule that compositional exercises ["attempts"] must be completed before they can be judged. Yet another reason for hiding a rule is that the rule may have nothing to do with counterpoint, such as the rule that no more than 127 "specific remarks" and 127 "general remarks" can be made about any given attempt to complete an exercise.) A listing in an internal rule list, then, can initiate a judging procedure or not, can be displayed or not, and can, if displayed, be listed on either the old rule list or the new rule list. The simplest case would be where a rule is displayed, and where the same listing in the internal rule list also initiates a judging routine that checks for infractions of that rule only. More typically, however, infractions of several displayed rules are detected by a single judging routine, the identifying number of which appears on the internal rule list as a "hidden" rule. In such cases, the routine refers the student automatically to the correct displayed rule when an infraction of that rule is detected. This method of combining the detection of infractions of several rules in a single judging routine made it possible to avoid cases where rule interactions would have caused needlessly repeti- tive or conflicting remarks to be generated.

Each exercise group contains from one to sixteen exercises. The index of exercises available within a given exercise group informs students of the number of exercises in the group, the number of times they have attempted to complete it, the best grade


they have received for each exercise, the grade they need to attain on each exercise in order to pass on to the next exercise group, and whether or not they have in fact passed the exercises.

Having selected an exercise on which to work, the student sees a text description of the exercise. The first four lines of the text description are computer-generated from data stored with the exercise when it is created by the instructor. The instructor sets the mode and transposition, the minimum acceptable length (three measures is the absolute minimum), the maximum acceptable length (fifty- eight measures is the absolute maximum), the number of candences that the student must mark (the absolute minimum of zero cadences is possible mainly to permit the author to reduce the tedium of debugging, while the absolute maximum of twenty cadences is probably more than enough), and the voice(s) for which the exercise is to be written. The instructor can add remarks at the bottom of the first page, and continue them indefinitely on succeeding pages. The instructor can also write a cantus firmus in one of the voices, and prevent the student from altering the notes of that voice.

Having read the description of the exercise, students come to the index of their attempts to complete it (Fig. 1).

To begin work on the exercise, students press the "a" key. The computer retrieves the entire exercise from its database of exercises and copies it into each student's personal file. Then the LASSO version of the Ottaviano music editor is invoked, so the students can write notes and rests and mark candences on their copies. Figure 2 shows the screen display exactly as students see it using the editor.

Ottaviano is a music editor designed for instruction only (Newcomb 1978). It is not powerful enough for serious composers, but, on the other hand, it is simple and can be intuitively grasped by students without any special instruction (Newcomb, Weage, and Spencer 1981). Notes are written by touching the screen with the finger. This simple editor pre- serves a rigid relationship between elapsed time in the music and horizontal position on the staff. Stu- dents can write any notes they like, unless they would force the editor to behave illogically for the purposes of counterpoint, as, for example, when the

student attempts to tie two notes together that are not the same pitch. Students may at any time enjoy an automatic audio playback of their work via a Gooch Synthetic Woodwind instructional music synthesizer. Sometimes called the PLATO Music Box or, more affectionately, the Gooch Box, this de- vice was developed in 1974 by Sherwin Gooch at the Computer-based Education Research Labora- tory at the University of Illinois (see Scaletti 1985). Gooch Boxes were the first instructional music synthesizers to be used at all widely (as widely as was PLATO). They have four voices (all square waves) and eight volume levels; they remain more useful for music instruction than several other inexpen- sive synthesizers made since then because of their combination of crystal-controlled clock and 15-bit pitch resolution-they are always in excellent tune. They are used at the Center for Music Research mainly because we have 12 of them, and they keep working despite their age and obsolescence.

Having completed their attempts, the students return to the index of their attempts. Students may elect to initiate entirely new attempts to complete an exercise (as discussed previously), to revise earlier attempts (whether they have been graded or not), to submit an attempt to the computer for grading, to see the remarks that the computer has made in the course of grading an attempt, and/or to destroy unwanted attempts. In all likelihood, students will choose to have their newly completed attempt judged.

The length of time required for the judging op- eration depends on several factors. On the Florida State University PLATO system in 1983, judging operations for a thirty-two measure, two-voice composition by Lassus, at the seventh and most com- putation-intensive exercise group level, required four to six minutes at a time when the demand for system resources was low (fewer than twenty users signed on).

In order to see the remarks generated by LASSO during judging, students must begin by making a copy of the attempt that has just been graded. While editing the new attempt, students may see the remarks generated by the computer when it judged the previous attempt of which the new attempt is a revisable copy. The display of remarks

Newcomb 55

Fig. 1. Index of a student's attempts to complete an exercise. This exercise is the eleventh of twelve exercises in Exercise Group

#39. The first two attempts have been judged; both received a grade of zero. If the student presses the "a" key at this point,

attempt #4 is created in such a way that (unlike attempts #2 and #3) it will not be a revision of an earlier attempt.

INRDEX' OF ATTEMPTS TO COMPLETE E',EF..RCISE * 11

ii-lttlmt j Date Ulher F.ev i sin ot f l , ,: :ra,:ide Att Last Edited At tmpt * _ em. Received

1I .Lf 9.. 3 .1 _, t. . %

=' 3 f.' 9, '3.02 .. 2:-3 n

a

b

HELP

BACK

to START this exerrcise " frcom scratch"

to REVISE a previou-s attempt (and/o r see remarks generated b,y jud ging:

to have the computer JUDGE an attempt

to DELETE an attempt (to make more room)

for an EXPLANATION of th is page

for a DESCRIPTIONP o f Exerc i se 11

for the INDEX:: o f exerc i .-se

(Fig. 3) tells the student how many "general" and "specific" remarks were made, the overall grade received, the remarks themselves, and the penalty applied for each infraction. A "general remark" is a remark that applies to the attempt as a whole and that does not apply to a specific note or a single passage within the piece. An example of a general remark is "Too many skips in proportion to steps." A "specific remark," on the other hand, applies to a specific place in the music, and LASSO is able to show students the exact point in the music where the rule infraction (for example) occurs.

LASSO is capable of generating a very large number of different remarks in good English. As mentioned earlier in another context, LASSO can store only 127 general and 127 specific remarks about an attempt. Any attempt that exceeds these limits is automatically graded zero, because a grade cannot be accurately calculated once the remark storage area has overflowed and judging is forcibly inter- rupted. In such a case, a remark saying that too many remarks were generated is the last remark stored.

At the display showing the list of remarks, stu-

Computer Music Journal

shi ft-BRCK

56

Fig. 2. Ottaviano editor. The student has pressed the "a" key at the display shown in Fig. 1, and now is ready to write a new attempt. There is insufficient

room on the screen to show key signatures on the staves, and therefore there is no B-flat shown in the key signature, despite the fact that this exercise

is in a transposed mode; instead, key signatures are shown by appropriate in- dications inside the letter- name boxes at the bottom left of the screen.

I ? I I ? ? 1 I ? ? I i ? ? ? I I ? ? II-? I I?1? ?I1?

e Press BACK when you waant tc:. lea.e. If :.Lu p el--s ' c._.U

sh i ft-STOP, '.,u.- I.jork l.. i 11 nt be _:ed! ! ! !

T C tie

+-

*l( C)

_CC C _ C2 3 , 4

. d- .e f g- a b

dents can type the number of a remark and see either the help sequence explaining the rule involved, or the exact passage that generated the remark, with rounded frames around the first and last notes of the offending passage, while at the bottom of the screen the remark itself is also shown (see Fig. 4). Students have direct access to the general explanation of the rule involved.

In Fig. 4, the tiny 8s under each of four of the quarter-notes in the lower voice indicate that those

ti e IITE

Nr...A.. E.1 t i ing rlm. 1 -3

-sh i f't -HELP f: .-r . -pt i ::.n -~ X k .

oCt t: 1:E

BACK AHEACD TO TO:

LAST NET::T

notes sound an octave lower than written. Inciden- tally, this is the same attempt to which the display shown in Fig. 3 refers. As any student of sixteenth- century style will quickly realize, there are problems in these three measures that are more serious than the comparatively trivial one which this remark mentions; all these problems are addressed by other remarks. No penalty is applied because of remark 30; this remark is made simply to remind the student of an important overall feature of the style.

Newcomb 57

dJd

Fig. 3. Four of the 68 remarks generated by LASSO when judging a student's (very poor) counterpoint exercise.

Remarks generated when Attempt *2 was j udged: 8 general remarks 6. speci fic remarks grade = 8.8%

measure *6 2 9 Th- i s i nterva 1, a m i nor .second,

I -1 f points I

i s a m i Mshandl 1 ed d i ssonance.

beginning in measure #* | no penalty | 3.8 A melodic interval of a third is fo1ll owed by stepwise

mot i on in the same d irect i on.

measure *7 | -1.1 points | :31 This pitch, c , is at the lower limit of the

range of the soprano voice and is :rarely used.

measure *7 | - 1J points I 32 Accented quarter note pass i ng tone? The di ssonant quarter

note is not pre-ceded by descend i ng tep.

rcore inf'. con w. which -remmark? > :30J :shi ft - NEXT t, C: see the mus i c LAB to move d i spla 1 to

One might wonder why the A after the B-flat is not highlighted instead of the B-flat, since the "stepwise motion in the same direction" continues to the A. The reason is that LASSO considers the stepwise motion from the middle C to the B-flat suffi- cient to establish the fact that stepwise motion in

NE:XT or BACK for more remarks 5shi ft-DATA to go back to editing

the same direction has occurred, and it looks no further.

Students can toggle between editing their current attempt (the copy of the judged attempt) and examining the remarks and help displays associated with the judged attempt by pressing a single key.


Fig. 4. Optional display of a portion of a student's judged exercise, highlight- ing the passage to which a remark (30) refers.

I. t j J 1 4 J_

mot i on i n the same d irec.tion.

The passage begins :_ in bar 6 in the third beat, and ends in bar 7, in the first beat. Vc. ice i nvo I ved (count i ng fr -:m to,p) 2

rINEX'T for more deta il Bl ACK f r i rndex o f remarks LAB E-r sh i ft -LAB t o pla 1 a,

Conclusion

The intent of the LASSO system is to provide an interactive learning environment for iterative improvement in students' ability to write in sixteenth-century sacred vocal style. Students can freely move between instructional displays and a

.h i ft -NE::'T fo r ne.<t remark shi ft-BACK fo- r prev i ous- remark

s:hi ft -DATA to -i: t: back to editing:

music editor where they put their learning to im- mediate use. Each student's learning process is guided in that the exercise groups are carefully ar- ranged in graded levels of difficulty. The student cannot pass into the next exercise group without having first met the criteria of the current one. Learning is guided in that the student's attention is

Newcomb 59

directed by the remarks to those stylistic features with which the student has the most difficulty. The student's learning process is self-directed to the ex- tent that the study and work activities can be done in any order. Perhaps most significantly, the student can write any notes or rests at all (there are none of the "guessing games" so prevalent in much of today's instructional software).

The fact that students write music according to their own best understanding of what is expected, and then receive in-depth stylistic guidance, today sets LASSO apart from other software for music instruction. It is hoped that future software for music instruction will far exceed this bare-minimum level of allowing student creativity. Future authors of instructional software should consider the creation of a synthetic explorable world, where the student makes discoveries and has "learning adventures." With artificial intelligence technology, we may hope to be able to provide a far larger percentage of the population in future generations with a higher qual- ity music education (see Sleeman and Brown 1982). The PLATO technology is pushed to its limits (and, practically speaking, well beyond its limits) in the LASSO system. Better ways of applying computer power to education must be developed, probably in- volving new ways for educators and researchers to develop courseware. One can imagine systems for the development of advanced learning environments that provide for intelligent conversation between the courseware developer and the computer. Both the development environments and the developed applications will require enormous amounts of computer power, far more than most classrooms are likely to have for a few years yet.

It would be unreasonably ambitious and of little practical value to continue the LASSO project beyond the seventh exercise group category. It is unreasonably ambitious because, due to the limitations of the knowledge representation scheme employed, it is increasingly difficult to maintain the existing knowledge base whenever a new rule is added. It is of little practical value because LASSO can never be published, having been technologically orphaned by changes in the standard CDC PLATO product since the project was begun (LASSO continues to run at Florida State University [FSU] only

because the Center for Music Research enjoys, for the time being at least, special system privileges). Naturally, under different circumstances, with a supporting computer system designed around modern artificial intelligence technology, it would be reasonable and valuable to rewrite the entire LASSO system, and to extend it to include three and four voices, imitations, and so on. If such a system were properly devised, the knowledge base could be used in reverse, i.e., the system could be used to generate counterpoint as well as to write commentary about counterpoint.

LASSO was used experimentally only once, in 1982, at the FSU School of Music, just before the 16 PLATO terminals that the Center for Music Re- search had borrowed from the Florida Department of Education were sold by them to a paying cus- tomer. The instructor who used the system (Dr. William Harbison) was enthusiastic, the students less so. Several students complained that the editor was too slow; the author lays the blame for this on the fact that the PLATO terminals available were dumb and employed a forward-channel bandwidth of only about 1000 useful bits/sec. Other students complained that it was so difficult to satisfy LASSO's demands that they were forced to revise the same exercises repeatedly; the author feels that this reaction is expectable and satisfactory. After all, LASSO cannot make counterpoint inherently easier, it can only make interactive counterpoint instruction more readily accessible.

References

Fux, J. J. 1725. Gradus ad Parnassum. Vienna. Jeppeson, K. 1939. Counterpoint: The Polyphonic Vocal

Style of the Sixteenth Century, translated and with an introduction by Glen Haydon. New York: Prentice- Hall.

Krenek, E. 1959. Modal Counterpoint in the Style of the Sixteenth Century. New York: Boosey and Hawkes.

Newcomb, S. R. 1978. "Ottaviano: An Interactive Music Editing Program for Use on the PLATO System." Paper presented at the Seminar on Computers and Music, University of North Carolina at Greensboro, North Carolina.

Newcomb, S. R. 1983. LASSO: A Computer-based Tu-


torial in Sixteenth Century Counterpoint. Document #8404750. Ann Arbor: University Microfilms International.

Newcomb, S. R., B. K. Weage, and P. Spencer. 1981. "The MEDICI Tutorial in Melodic Dictation." Journal of Computer-based Instruction 7(3): 63-69.

Scaletti, C. 1985. "The CERL Music Project at the Uni- versity of Illinois." Computer Music Journal 9(1):45-58.

Schoenberg, A. 1963. Preliminary Exercises in Counter-

point, edited and with a foreword by Leonard Stein. London: Faber and Faber.

Sleeman, D., and J. S. Brown, eds. 1982. Intelligent Tutor- ing Systems. London: Academic Press.

Soderlund, G. F. 1947. Direct Approach to Counterpoint in 16th Century Style. Englewood Cliffs, New Jersey: Prentice-Hall.

Soderlund, G. F., and S. H. Scott. 1971. Examples of Gre- gorian Chant and Sacred Music of the 16th Century. Englewood Cliffs, New Jersey: Prentice-Hall.

Newcomb 61

lectura 8 lasso an intelligent computer based tutorial in sixteenth century counterpoint

Documents