You are on page 1of 30

Nelson Goodmans Languages of Art, Notation, and Artistic Representation: An Analysis of Music Notation

Nelson Goodmans Languages of Art, Notation, and Artistic Representation: An Analysis of Music Notation ...........................................................................................................................1 Introduction ................................................................................................................................2 Goodmans Theories and Terms..................................................................................................3 Modes of Reference ................................................................................................................4 Symbol Schemes, Systems, and Notation ................................................................................6 Artistic Identity: Autographic and Allographic Arts ................................................................8 Single and Multiple Arts .......................................................................................................11 Music Notation .........................................................................................................................12 History..................................................................................................................................12 Elements of Common Notation .............................................................................................13 Music Notations Place within Goodmans Notational Theory ..................................................15 Criticisms of Goodmans Theory of Notation........................................................................17 Common Usage.................................................................................................................17 Musics Notationality........................................................................................................18 Musical Expression ...........................................................................................................19 Work in Information Technology with Notation and Musical Scores.........................................22 MusicXML ...........................................................................................................................22 Elements of MusicXML ....................................................................................................24 Conclusion................................................................................................................................29 Bibliography .............................................................................................................................30 Megan Winget Page 1 7/25/05

Introduction
What is the relationship between score and performance? Between a computer program and its product? Or between a databases design, the records that constitute it, the system which produces records, and its interface design? What constitutes a work? How does a set of instructions for performance become the performance? How many changes can you make to a work, and have it retain its identity? What elements of a work are essential to retain in their original form, and which elements are variable? When a work is inherently variable, what does the term authenticity imply? These issues of identity, representation, and authenticity, while originating in aesthetics, are valid for digital preservation as well. Before the advent of ubiquitous computing, these questions were limited to esoteric matters like variable art forms such as music and drama; and conceptual and performance art. Today, however, nearly everything is variable, from the most mundane Word documents to the most complicated databases. Not only are there difficult issues concerning preservation methods , the more fundamental issue of identity; i.e., what has to be preserved in order for documents to retain authenticity also remains to be answered. Implausible as it might seem, aesthetic theories of identity (what constitutes a work) may have some bearing on present concerns. Not only have people been working on this problem since the classical era, their theories and explanations regarding representation, notation, interaction, interpretation, and expression are surprisingly relevant to present discussions regarding digital authenticity and reliability. The representational schemes of music are particularly germane to this discussion. Musical scores and computer programs are unexpectedly similar. They both represent, in a highly symbolic and notational language, rules for presentation of a variable and manifold product; they both rely on intermediaries to translate and interpret their instructions; and their Megan Winget Page 2 7/25/05

output is measurable, in that the score or program defines a correct representation, whether that correct object is aesthetically pleasing or not. In this paper, I will provide a brief review of aesthetic theories regarding notation and the representation of variable works as developed by Nelson Goodman. I am hoping that a review of these concepts will bring a deeper understanding of variability, representation, and interpretation for the purpose of authentic depiction and preservation of todays omnipresent variable works, digital media.

Goodmans Theories and Terms


It would be difficult to overemphasize the influence of Nelson Goodmans 1968 classic, Languages of Art on twentieth century aesthetics (Goodman, 1976). In this small book, he developed a new version of aesthetic theory that was grounded in the philosophy of language, and in so doing, reframed many of the questions, and answers, of contemporary philosophical thought. Goodmans primary contribution was to define works of art as symbols within symbolic systems, and treat the problematic issues of artistic representation and expression as semantically based questions of reference and denotation. Its subtitle, An Approach to a General Theory of Symbols, implies that Goodmans book is not only concerned with artistic issues, but also with gaining an understanding of symbols generally, both linguistic and non-linguistic; in science as well as in regular life and the arts. Goodmans general attitude is that symbols are pervasive and fundamentally important to conception. We use symbols to recognize, comprehend, and even create our world; and both science and the arts work together to help make sense of that world. Indeed, one of the most noteworthy effects of this book is that it brought science and art together in a significant and meaningful way. Megan Winget Page 3 7/25/05

Regarding art particularly, and symbolic activity generally, Goodman promotes a form of cognitivism that is, because we discover and even create the worlds we live in through our interaction with symbols, our interest in those symbols is manifestly cognitive. Paintings, sculptures; and music and dance pieces are all objects composed of symbols; each artistic form functions differently and has distinct relationships to the referent world. Because of these distinctions among art forms, understanding depends on interpretation, which means understanding what an art object refers to, how it refers, and within which system of rules it makes that reference. There are three issues that Goodman emphasizes. First, for Goodman, symbolization and reference are synonymous. Second, there are different modes of reference; and third, when something is a symbol, it is a symbol of a certain kind. This can only be so within a symbol system of that kind, which is governed by the syntax and semantic rules distinct to symbols of that kind. Robinson gives the example, just as the noun boot is a boat-description in German and a boot-description in English, so a portrait might be an ordinary-woman-picture in a Cubist symbol system and a grotesque-woman-with-a-serious-eye-deformity-picture in the symbol system of the late nineteenth century academicism (Robinson, 2000, p. 214).

Modes of Reference
The core of Goodmans theory is reference , the idea that one thing can stand for another thing. Reference is expressed in different modes, specifically denotation and exemplification. Denotation is the relationship between a label, like William Jefferson Clinton, or The 43rd President of the United States, and the labeled object. In fact, according to Goodmans approach, possessing a feature or property, like being blue, merely amounts to being denoted by a particular label, like blue. Exemplification, typified by tailors swatches, requires Megan Winget Page 4 7/25/05

that the exemplifying symbol refer back to the label that denotes it. For example, while the label blue denotes any blue object, only those things blue color swatches for example that also refer to, or stand for, blue are samples of it. An important characteristic of samples is that they are selective in their functioning. A tailors swatch, for example, does not exemplify all the features it possesses but only those for which it is a symbol e.g., those labels related to color and texture, but not those related to size and shape. The properties a sample exemplifies are dependent on the system in which the sample is being used: color and texture of the swatch are appropriate for the tailoring-system; size and shape are not (Giovannelli, 2005).

Figure 1. Edvard Munch. The Scream (or The Cry). 1893; 150 Kb; Casein/waxed crayon and tempera on paper (cardboard), 91 x 73.5 cm (35 7/8 x 29"); Nasjonalgalleriet (National Gallery), Oslo.

In terms of art, Goodman uses the idea of exemplification to explain artistic expression as well as style and form. Artistic expression is analyzed as metaphorical exemplification, and style as literal. For example, Munchs The Scream (figure 1) literally exemplifies swirling shapes, lurid colors, dramatic contrasts, and powerful brushwork. It metaphorically exemplifies that is, it expresses feelings of anguish and alienation (Robinson, 2000). Exemplification becomes a particularly useful concept in explanation of how abstract arts, which normally do not represent Megan Winget Page 5 7/25/05

or describe anything, can carry meaning nonetheless. Another positive outcome from Goodmans book is that it was one of the first descriptive systems that made allowances for expression of non-emotional qualities. His work on exemplification allowed art historians and theorists to talk about the expression of qualities that are metaphorically expressed by artworks but arent specifically emotional (Robinson, 2000). Another advantage of Goodmans theory of exemplification is that it takes into account the fact that the metaphorically expressive quality of a work of art is dependent on the symbol system in which it works. For example, Piet Mondrians Broadway Boogie-Woogie (figure 2) expresses gay abandon partly because of its place within the symbol system of Mondrians oeuvre. If, on the other hand, someone else had painted it, the same painting might express cool aloofness (Gombrich, 1972, Chap. 11).

Figure 2. Piet Mondrian, Broadway Boogie Woogie 1942-1943. Oil on canvas. 50 x 50 in. (127 x 127 cm) The Museum of Modern Art, New York.

Symbol Schemes, Systems, and Notation


As earlier noted, interpretation of artistic reference whether a symbol denotes or exemplifies; what it denotes, or which of its features it exemplifies; and whether it does so literally or metaphorically depends on knowing the symbolic system in which it functions. Megan Winget Page 6 7/25/05

Another issue to take into account is that a symbol takes its identity (linguistic, musical, pictorial, diagrammatic) from its inclusion in a certain kind of symbol system, each with its own syntactic and semantic rules. Symbol systems basic mode of reference is denotation: characters denote items in the field of reference. Syntactical rules determine the acceptable forms and combinations of characters, which, all put together, is called the symbolic scheme. Semantic rules determine how the symbols in the scheme refer to the objects in their field of reference, and they determine the symbolic system. Notation is the means by which Goodman explains the different syntactic and semantic rules of symbol systems. Basically, notation is a symbol system where each symbol corresponds to one item in the field of reference, and each item corresponds to only one symbol. A musical score, the whole of it, is for Goodman a character in a notational system if and only if it determines which performances belong to the work, and at the same time, is determined by each of those performances (Goodman, 1976; 129-30). There are two syntactical rules to which a scheme must adhere in order to be notational: The first rule is that all members of a character are interchangeable, i.e., theres character indifference, and theyre disjoint. A good example is a musical score, where any quarter-note symbol can be exchanged with any other (Godman, 1976; p. 132-34). The second syntactical rule is that characters disjointness should be testable. That means that characters should be finitely differentiable. This rule then excludes dense systems (like painting) where any two characters can have infinitely more characters between them. Goodman compares notational symbol schemes to digital instruments measurement. Digitally, measurement is unambiguous and easily defined. Non-notational schemes are compared to analog systems of measurement: For their complete lack of articulation, those systems can also be said to be dense throughout: given any mark (e.g., a mark in a scale) it could

Megan Winget

Page 7

7/25/05

stand for virtually an infinite number of characters, hence of measurements; or, equivalently, given any two marks, there is a virtually infinite number of possible characters between them (Giovannelli, 2005). Symbol systems, which are defined by semantic rules, require more than this to be notational: the characters in a notational system must be: 1) unambiguous; 2) the characters must be semantically disjoint (meanings cannot intersect); and 3) the system must be finitely differentiated (its always possible to know to which item a symbol refers). Musical scores qualify as notational systems, with some qualifications that will be discussed later. Natural languages have a notational scheme but fail to be a notational system because of ambiguities (in English, a bank refers to a piece of land on the side of a river, as well as a place where people conduct financial transactions) and instances of semantic disjointness (the words woman and teacher often refer to the same thing). Pictorial systems fail for both syntactic and semantic reasons.

Artistic Identity: Autographic and Allographic Arts


Goodman uses his theories of symbol systems and notational adherence as the basis for an outline for identity conditions for different kinds of art works. Based on the syntactic and semantic characteristics of notational systems, the different art forms can be placed on a continuum that spans between pure notation, where there is perfect preservation of identity between replicas or performances of the work; and a purely analog form, which is densely represented, and every work is an original. The question of a works identity is related, for Goodman, with a works history of production. If that history is essential in determining the value of the work, then that work is called autographic. Autographic arts are those like painting and sculpture, where there is only Megan Winget Page 8 7/25/05

one instance of a work; as well as etching and woodcut, where there can be multiple instances of the same work. Basically, all those arts where the final artistic output is descended directly from the creator, without mediation from any other entity, are said to be autographic. Goodman places this definition within the context of authenticity and forgeries of art works: Let us speak of a work of art as autographic if and only if the distinction between original and forgery of it is significant; or better, if and only if even the most exact duplication of it does not thereby count as genuine (Goodman, 1976; p. 113). Therefore, only the actual canvas painted by van Gogh in 1889 counts as the original Starry Night , and anything else is a copy, no matter how indistinguishable it is from the original. If a works history of production is not essential to the value of the work, then it is called allographic. Music, dance, theater, literature, and architecture all allow for instantiations of a work that are independent of a works history of production. You can watch a production of Shakespeares Henry V even if its performed from a contemporary version of the script. Every performance of Beethovens Fifth Symphony is authentic, given its performed in line with the specifications set out in its score. So this question of identity is bound up with whether a given form of art allows for a notational system a score that identifies the essential properties a performance must have to belong to the work. (Goodman, 1976; p. 212). In this spectrum, music and painting end up as diametrically opposed, the former being allographic, its notation system allowing for different authentic instantiations of a work; the second being autographic, its non-notational dense representation not allowing for any kind of authentic secondary copies. A musical score is in a notation and defines a work. (Goodman, 1976; p. 210). Only those performances that correspond to or comply with the score count as performances of that

Megan Winget

Page 9

7/25/05

work, although there is no value judgment involved: the most miserable performance without actual mistakes does count as [a genuine instance of the work], while the most brilliant performance with a single wrong note does not (Goodman 1976; p. 186). Although presented as the most purely notational form, musical scores do have elements that are not notational. Published expression marks: indications of tempo, mood and articulation, give the performer great autonomy, for example, are consequently not definitive of the work and therefore by definition are un-notational. Dance is similar in its notationality to music. Although dance did not have a cohesive notation system at the time Goodman was writing the Languages of Art, Rudolf Laban had recently developed a tentative notation system for dance, eventually called Labanotation, which Goodman believed had the likelihood of becoming fully notational perhaps even more purely notational, and with fewer departures from notational systems, than even music. In dramatic works, the text of the play is an amalgam of the score, the written dialog which is notational; and script, the stage directions which are not. The parts of the text other than the dialog count not as integral parts of the defining score but as supplementary instructions (Goodman, 1976; p. 211). Only those performances that comply with the dialog written in the text are compliant with that work. This distinction is related to that in musical scores between the notational score and the non-notational expression marks, mentioned above. At the other end of the spectrum, painting and sculpture are defined as analog, which means their characters are in syntactically and semantically dense systems, and are unable to be reproduced by a notational scheme whatsoever. This does not mean that there are not notational systems that classify these art forms: a library classification system for paintings is a good example of a notational system that describes paintings. However, given the history of painting,

Megan Winget

Page 10

7/25/05

library style classification would be incompatible with artistic practice; e.g., artists would not create artistic works using a library classification scheme; and the library classification of a given painting could not represent, in any significantly artistic way, the work.

Single and Multiple Arts


Goodman has an additional discussion of single and multiple arts, a distinction that is independent of whether an art form is autographic or allographic. Etching is given as an example of an autographic art form which is also multiple: many prints can be made from one plate. In this case, the work is the etching plate, and the prints are instantiations of that work. In the hypothetical situation where someone were to clone a painting molecule by molecule that technique would transform the art of painting from single to multiple art without changing it from autographic to allographic. The original painting, the one from which the copies were made, would still be the work. Another related, and in fact formative element of Goodmans discussion of notation is the concept of authenticity. In short, for Goodman, authenticity only matters when there is no notationality. For example, it doesnt make any difference whether a dramatic piece is performed from the original text or from a copy matching it, because the text (commonly called the script) is in a notational system. However, it does matter whether a painting is an original Rembrandt or a copy. Because paintings are analog, only the original counts as the work. Goodmans theory of notation and his investigation of the different ways in which different art forms fit into that theory establish a kind of system for the arts, which has been fundamentally influential in the way people think and talk about art. Whether they agree or disagree with the specifics of his theories, the distinction between autographic and allographic arts is a powerful idea; and his development of notation theory, while not as familiar a concept, is useful as a Megan Winget Page 11 7/25/05

descriptive tool in studying these art forms, and how users interact with them. The remainder of this paper will provide historical background and analysis of formation and use of the most common symbolic notation system: music, and will conclude with a brief exegesis on how this research could be used by the information science community in thinking about digital preservation, interaction, and interpretation.

Music Notation
History
As Goodman points out, music started out as an autographic art, each musical piece being unique, not represented by any notational system. Some time around the 9th Century, Catholic monks developed methods to record sacred songs (called plainchant) in written form. The earliest of these systems dont have a staff, but use a system of dots and strokes that were placed above the text, called neumes (figure 3). Although these symbols expressed considerable musical complexity, they were unable to convey pitch or tempo, and mainly functioned as a reminder to someone who already knew the tune, instead of as a means by which someone who had never heard the melody could sing it exactly right just by looking at it.

Figure 3. Fragment with neumes taken from the Laon 239 manuscript (Metz), written around 930.

Megan Winget

Page 12

7/25/05

The inability to express pitch and tempo was probably less problematic for those early monks than it would be today, because spoken Latin has innate cadences and rhythms that the monks wouldve used naturally. As scope of music expanded, though, these shortcomings became more pronounced, and the monks worked on various modifications of the system. The breakthrough came in the 10th Century, when Guido of Arezzo, developed the staff notation system. In this system, each staff is made up of four horizontal lines, with the vertical position of each mark on the staff indicating the pitch of the note it represents. This system, the four-line staff, remains in use to the present day for plainchant, although other musical styles use staffs with differing numbers of lines. The modern system of notation, with its standard five-line staff was first used in France and became widely used by the 16th Century.

Elements of Common Notation


The staff in Western notation is generally begins with a clef, which represents the range of pitches included by the staff. A treble clef tells the viewer that the notes in that staff are the higher ones (lowest note in the staff is the note E above middle C, and the highest line represents the note F, one octave higher); and the bass clef represents the lower notes (second G below middle C to the A below middle C) (These are illustrated in figure 4). Directly to the right of the clef on the staff is the key signature, which specifies the notes that should be held flat or sharp throughout the piece. The time signature appears next on the staff, communicating the pieces rhythmic characteristics to the musician. The time signature generally has two numbers, one on top of the other; the upper number indicates the number of beats per measure (or bar), and the lower one shows what sort of note constitutes a beat. A time signature of 4/4, also called common time, says that there will be four beats per measure, with each beat being a quarter note. A time signature of 2/2, or cut time, represents two beats Megan Winget Page 13 7/25/05

per measure, with each beat being a half note. Because the first beat of each bar is generally stressed, its important for the musician to understand the complexities of this information. A staff system is simply multiple staves grouped together. This happens when two staves are necessary to cover the range of the instrument being played (like in piano), or where many instruments are played together (like an orchestral score). Various expressive directions are commonly added above or below the staff, often in abbreviated Italian. Those related to tempo: adagio (slow, leisurely); andante (moving with a moderate tempo); allegro (moderately fast); allegretto (a little slower than allegro); and vivace (lively, animated, brisk). Those related to dynamics: pianissimo (abbreviated pp, meaning very soft); piano (abbreviated p , meaning soft); mezzo piano (abbreviated mp , meaning moderately soft); forte (abbreviated f , meaning loud); crescendo (abbreviated cresc., meaning increasing in loudness); and decrescendo (abbreviated desc., meaning decreasing in loudness). Those related to style: animato (animated, lively); con brio (with vigor and spirit); dolce (sweetly); giocoso (humorous); and legato (smooth and connected). Finally, if the music has a vocal component, the lyrics are written below the staff. Notes (decreasing in length) Rests (decreasing in length) Clefs
Figure 4. Musical Symbols. From Wikipedia. (http://en.wikipedia.org/wiki/Musical_notation)

Megan Winget

Page 14

7/25/05

Music Notations Place within Goodmans Notational Theory


As a complex, practical, and commonly understood notation system, the notationality of musical scores is unquestioned by Goodman, but he also he believes that only after the notes had been defined with precise time lengths and their placement on a staff represented exact pitch, did music notation become fully notational in his sense. Goodmans syntactic requirements are met: the notes are differentiated and unique, and theres character indifference between notes of the same type. Most characters of a musical score, whether numerals or letters or neither, are syntactically disjoint and differientated. The symbol scheme is thus substantially notational, and the language of scores truly a language (Goodman, 1976; p. 181). In the case of his semantic rules, musical scores become a bit more problematic, although ultimately not to the exclusion of its notationality. Rember that Goodmans semantic rules are: 1) meanings must be unambiguous; 2) meanings cannot intersect, i.e., theyre sematically disjoint; and 3) it must always be possible to know to what an item refers; meanings are finitely differentiated. First there are issues related to semantic disjointness: in piano scores, for example, the same sound-event complies with the characters for c-sharp, d-flat, e-triple-flat, b-double-sharp, and so on. In a violin score, however, the characters for c-sharp and d-flat do not have any complaints in common (Goodman, 1976; p. 181). Goodmans response is that these characters do not violate the rule of semantic disjointness in a meaningful and exclusionary way; they violate that rule only in that these characters are redundant. Redundancy, while not optimal, is not cause for exclusion from notationality. Another problem Goodman identifies is that of finite differentiation. If we suppose the series of whole note, half-note, quarter-note, eighth-note, etc. to be continued without end, the

Megan Winget

Page 15

7/25/05

semantic requirement of finite differentiation will be violated (Goodman, 1976; p. 182). The problem, essentially, is that time is indefinite, although musical notes represent it in specific lengths. Forby tying note-signs together we can construct characters for notes differing in duration by less than any given fraction of a beat. Hence no sounding of a note could be determined to comply with at most one character (Goodman, 1976; p. 182). He recognizes that, in any score or corpus, the number of note-signs, and flags on them (which denote length), is finite, but also claims that there must be a rule delimiting the number of flags permitted by the system at all; otherwise, recovery of score from performance will not even be theoretically possible, identity of work from performance to performance will not be ensured, and the primary purpose of a notational system will not be served (Goodman, 1976; p. 183). He notes that in modern music practice there seems to be a traditional limit set at five flags, the 1/128 note. With two qualifications, Goodman accepts that the peculiarly musical characters of the system (Goodman, 1976; p. 183) meet the semantic as well as syntactic requirements for a notation. As mentioned earlier in this paper, this cannot be said for the other numeric and alphabetical characters that also appear in musical scores. He cites three areas that are specifically problematic: instances of figured bass, the free cadenza, and expression marks. Figured bass (or basso continuo) is a common notation in baroque music where only the bass line, inversion symbols, and chromatic alterations are written. Keyboard players then improvised an accompaniment. Free cadenza is similar. Commonly its an improvised musical flourish, which takes place when an aria or section of an aria is coming to a close its cadence spot. The free cadenza, until the time of Verdi, was seldom notated precisely by the composer, and, like the figured bass, gives the musician great leeway in performance. Goodmans problems with these elements of the score are subtle. First he recognizes that if the basso continuo or

Megan Winget

Page 16

7/25/05

cadenza are truly improvised, these instances can be defined as sections where improvisation occurs, variation is possible, and notational rules have not been violated. However, if there has been some attempt to notate these sections, but performers or artistic directors have chosen to partly improvise then notation is no longer defining the work, and any given performance cannot be said to be an instance of the work. Verbal notation of expression represents a different kind of problem. Goodman does not object to the use of ordinary language. The problem is whether the language meets the semantic requirements for notation. Apparently, almost any words may be used to indicate pace and mood. Even if unambiguity were miraculously preserved, semantic disjointness would not be. And since a tempo may be prescribed as fast, or slow, or as between fast and slow, or as between fast and between-fast-and-slow, and so on without limit, semantic differentiation goes by the board too (Goodman, 1976; p. 185). Therefore, the verbal language of tempos is not notational, and cannot serve to help identify a work from performance to performance. No departure from the indicated tempo disqualifies a performance as an instance however wretched of the work defined by the score (Goodman, 1976; p. 185). It should be noted, though, that the metronomic specifications of tempo (generally written somewhere near the top of the first stave), do count as notational.

Criticisms of Goodmans Theory of Notation


Common Usage
Understandably, musicians and music theorists found Goodmans discussion of musical scores highly problematic. Most of these problems Goodman recognized as the disjunction between languages common usage and his philosophical application of it. However, Goodman never expected musical practice to comply with his stated philosophical obligations. One hardly Megan Winget Page 17 7/25/05

expects chemical purity outside the laboratory, (Goodman, 1976; p. 186) for example. One of their major concerns, mentioned earlier, is Goodmans contention that a musical performance with one wrong note is not a performance of the work at all. Wouldnt it be possible to bring our theoretical vocabulary into better agreement with common practice and common sense by allowing some limited degree of deviation in performances as instances of a work? (Goodman, 1976; p. 186) Goodman recognizes the problem, but responds that because transitivity of identity, if we allowed for even the most mundane deviations to still count as instances of the same work, then all performances whatsoever are of the same work. If we allow the least deviation, all assurance of work-preservation and score-preservation is lost; for by a series of one-note errors of omission, addition and modification, we can go all the way from Beethovens Fifth Symphony to Three Blind Mice (Goodman, 1976; p. 186). So, while a score may leave many features of a performance unspecified, and allow for considerable variation by the performers, one must comply with those instructions actually set down in the score in order for a performance to serve as an instantiation of a work.

Musics Notationality
There are two other specific criticisms of Goodmans notational theory. The first is that music is not in fact an allographic art form. There are two parts to this issue. First, some critics believe that a musical score, rather than defining a work, is merely an intermediary step between primary sound structures and performance. Boertz (Boertz, 1970) contends that musical notation doesnt specify sounds, but musical-structural components, like pitch, relative attack times, relative durations, and whatever othercategorical information is functionally relevant (Boertz, 1970; p. 543). He feels that sound successions are the real symbolic languages of music, and that notes, in that they require prior musical knowledge and interpretation, are essentially an Megan Winget Page 18 7/25/05

intermediary step between the actual notation the sound successions and the performed work. Instead of a performance acting as an instantiation of the score (or work); the score notation determines the interpreting musical works, and the performances thereof (Boertz, 1970; p. 543) Furthermore, Boertz, like others (Webster, 1971), asserts that music is a dense system that is suggestive rather than fully denotative. In fact, Boertz believes that modern notation isnt fundamentally different than the neumes system of the early 9th century: our present pitch notation is not necessarily more precise relative to the piece it notates than, say, that of preGregorian chant but only that what counts as compliance to it of interpreting sounds may be inferred as being more highly constrained with respect to their pitch components (Boertz, 1970; p. 544).

Musical Expression
The second issue regards the relationship between Goodmans theory and musics expressive quality. Both Boertz (1971) and Pearce (Pearce, 1988), express concern that Goodmans theory does not explicitly allow for the existence of expression in music. Both of these writers admit that Goodmans autographic / allographic division is, in general, a useful one, but that its integration with his theory of exemplification leaves music as a medium unable to express anything. Remembering that expression in Goodmans theory is metaphoric: i.e., the literally exemplified swirling brushstrokes and jarring colors of Munchs The Scream metaphorically exemplify, or express, anguish. Evidently, a performance exemplifies a score, but does not express it, for the exemplification is literal rather than metaphoric (Pearce, 2000; p. 234). Goodman does hold that a musical performance can be expressive, but because those instructions that indicate expression are nonconstitutive and can vary from performance to performance, the work itself cannot be expressive. Pearce attempts clarification of this problem Megan Winget Page 19 7/25/05

by comparing two multiple art forms, one autographic etching; and one allographic music. In both of these cases, the work is identified with a class. Performances are compliant with the score, and impressions are compliant with the etchers plate; the score and the plate constituting the work. However, because music is allographic and notational, and etching is autographic and dense, the analogy is unsatisfactory because the identity of a piece of performed music is dependent on the compliance between performance and score, whereas the identity of any given etching impression is only dependent on its history of production not compliance with the plate. As regards expression, the fundamental difference between music and etching is that for music, notationality is intrinsic to the nature of the work. The properties exemplified by the composers score are possessed literally. Properties exemplified by an etchers plate can be possessed either literally or metaphorically. Once you accept that, its easy to argue that since some properties will be metaphorically exemplified by all genuine instances of any given etching, the work itself is expressive. It is impossible to say this about music, because music is essentially and fundamentally notational, and therefore all of its properties are only exemplified literally. This is problematic, because although there is a formalistic tradition that denies musics expressiveness, it is generally understood that music is expressive, whether that expression is emotive, thematic, or conditional. Pearce brings Jerrold Levinson into the discussion (Levinson, 1980). Levinson, a philosopher primarily interested in aesthetics and music, and, like Boertz, defines music as a sound structure, which includes rhythm, timbres and tempo indications. In addition to this soundstructure, though, Levinson believes a musical work is a specific thing created at a certain time and place by a certain composer, and is not creatable by anyone else in any other situation. Music, for Levinson, cannot be purely abstract sound, but is also a creative art form whose

Megan Winget

Page 20

7/25/05

definition must also integrally include a certain means of performance or sound production, and a specifically related musico-historical context. For Levinson, a musical work is a sound/performance-means structure as indicated by composer X at time t. Additionally, if a given performance is to be a valid one, not only does it need to fit into the sound/performancemeans structure of the composers work, there also needs to be some connection between the performances sound event and the composers creative activity (Levinson, 1980; p. 25). Whether this connection would be determined by musicians performative intentions or by some causal chain linking the composition of the piece to its eventual performance, this definition would lead ultimately to the conclusion that music production and the identification of the work is dependent on the history of production, just as etching, an autographic art, is. So, while Goodmans theory is valuable in that it precisely defines a work, and provides a robust means to identify a genuine work from instance to instance, it does not allow for expressive properties to be part of the definitive nature of the allographic work generally, and music specifically. Levinsons theory allows for expression in music, but does not have a robust means to identify instances of a musical work qua work, and essentially makes the argument that musical systems are just as dependent on historical production as paintings and sculptures. Pearce and Boertz have similar solutions to the issue of Goodmanian exemplification and expressivity. They both recognize that listening to music is not necessarily a passive act: while it is possible to let music wash over oneself, many people actively and intellectually listen to music, and recognize elements that are not specifically in the score. Some of those elements might be contextual; some might be emotional or even theoretical. Once one is able to instinctively perceive those extra-musical elements, and can innately recognize correct or incorrect performative interpretations of them, those properties become additional means by

Megan Winget

Page 21

7/25/05

which authenticity or validity is attributed. This addendum to Goodmans theory is related to purely artistic criticisms of Goodmans theory of exemplification in the field of painting and sculpture: both Wollheim (Wollheim, 1980) and Walton (Walton, 1993) make an argument for a viewers intelligent eye, which, instead of letting a paintings exemplified properties wash over it, is seeing specific objects or people in images. Viewers, as well as listeners, are not passive, but actively engage with the work. Goodmans theory seems to ignore that fact, both in the autographic and allographic arts. In the case of music, Pearce and Boertz allow for a kind of intelligent ear, which recognizes music as conveying specifically emotional or narrative themes. Even if those themes are not transmitted through the score itself, it is no less part of the authentic work.

Work in Information Technology with Notation and Musical Scores


In terms of musical scores, most of the research concerning notation is located in the field of music information retrieval, which depends on robust musical representation for reliable results. Without a representation method that allows for searching, editing, and close comparison of scores, the development of useful music digital libraries remains far behind the development of text-based, or even image digital libraries. Despite decades of research, meaningful score representation is still problematically instituted at best. The most promising development in music notation digitization is the recent (2002) release of MusicXML and the opportunities it affords for search and retrieval, new interface design, and frameworks for dissemination, collaboration, and preservation in musical digital libraries.

MusicXML
MusicXML is the most recent attempt to develop an open-source music encoding language for use in musical representation systems, both commercial and non-commercial. Thus Megan Winget Page 22 7/25/05

far, the digital sheet music market has been hindered by its dependence on proprietary binary formats, like the Portable Document Format (PDF), which has no musical semantic functions and can only be viewed on screen or printed on paper. Various companies (like Sibelius) do have proprietary semantic music formats, but only proprietary players can read and edit them, leading to their limited usability and application. In the 1980s, electronic instruments like synthesizers faced a similar problem, when there was no way to get instruments made by different vendors to work or play together. Only the development of the Musical Instrument Digital Interface (MIDI) format (MIDI Manufacturers Association, 1996) resolved this problem the resolution providing an atmosphere that led to rapid growth of the electronic musical instrument industry. Development of general MIDI led to even further levels of compatibility, interchange, and growth in the electronic instrument market (history provided by (Good, 2001)). MIDI still remains the only symbolic music interchange format in wide use today. MIDI, however powerful it might be for the electronic music market, was designed to solve problems of music performance not representation, and therefore does not have the appropriate tools for encoding sheet music. MP3 and other audio formats represent the aural component of music recordings, not the notation; and despite years of research, computers are not able to automatically derive accurate music notation from sound recordings. Other music interchange formats have been developed, but MIDI remains the most successful. Notation Interchange File Format (NIFF) was developed to exchange music between scanning and notation applications. Although NIFF contains more notation information than MIDI, highly graphical representation made it inferior to MIDI for performance and analysis applications. Standard Music Description Language (SMDL), a subset of the Standard

Megan Winget

Page 23

7/25/05

Generalized Markup Language (SGML), was another effort to create a generally applicable, formal specification for music but its complexity and wide-ranging aims design stalled implementation on a wide scale (Roland, 2000). Michael Best, the president of Recordare, the firm that has developed MusicXML, has defined it as an attempt to do for online sheet music and music software what MIDI did for electronic musical instruments (Good, 2001). MusicXML basically serves as an exchange format for applications in music notation for the purpose of analysis, information retrieval, and performance. Its designed to represent complex, structured data in a standardized way. The same characteristics that make XML appealing in other areas, including straightforward Internet usability, easy document handling, and human readability are true for MusicXML as well.

Elements of MusicXML
In an attempt to give a brief introduction to how MusicXML represents musical scores, here is an example analogous to Cs hello world exemplar. This is the simplest music file imaginable: one instrument, one bar, and one note the whole note middle C.

Figure 5. Hello, world in Music XML. From Good, 2001: http://www.idealliance.org/papers/xml2001/papers/html/03-04-05.html

Heres the score represented with MusicXML

Megan Winget

Page 24

7/25/05

Reference to the dtd

Music has one part, named P1.

Beginning information on measure 1.

Information about the key Time information: 4/4 Clef: Treble Music Attributes Information about the note Pitch: Step C & Octave 4 (=middle C) Duration of note = 4 beats Type of note = whole note Note Attributes Figure 6. "Hello World" in Music XML format. From Good, 2001: http://www.idealliance.org/papers/xml2001/papers/html/03-04-05.html

Scores can be represented either partwise, as measures within parts; or timewise, parts within measures, with XSLT stylesheets to go back and forth between them. The above example uses a partwise score. The attributes elements describes the musical attributes like key signature, which are represented by the number of sharps or flats; the time signature, which includes the numerator (the beats per measure) and the denominator (the types of notes which make up a beat); and the clef, whose representation here shows that the second line from the bottom of the staff represents a G, and is therefore a treble clef. After the attributes comes the one note in the score, a C in octave 4, which is the octave Megan Winget Page 25 7/25/05

that starts with middle C. Both duration and type are included as children of note, even though the information is somewhat redundant, because notation programs have an easier time dealing with the written type, while MIDI programs deal more easily with duration. Additionally, in some cases, like jazz, sounding duration is different from the written duration, so having both would be preferable in those cases. In an illustration of how MusicXML gives better results than MIDI for music interchange, Best (2001) provides an example of typical differences in a practical situation. In his example, he scanned the fourth song of Robert Schumanns Liederkreis, Op. 24, on poems of Heinrich Heine, using SharpEye Music Reader v. 2.16. After correcting scanning mistakes, they then saved the SharpEye files to MIDI and MusicXML. They then imported the MIDI files into Finale 2002 and Sibelius 1.4 usingthe default MIDI import settings; and imported the MusicXML file into Finale 2002 using the Recordare MusicXML Finale Converter Beta 1. Best assures us that that song is not very complicated, so all of the musical features could be captured within SharpEye and saved to MusicXML. Figure 7 shows the last four measures as scanned by SharpEye:

Figure 7. Excerpt from Schumann Op. 24, No. 4, as scanned into SharpEye. From Good, 2001: http://www.idealliance.org/papers/xml2001/papers/html/03-04-05.html

Megan Winget

Page 26

7/25/05

Figure 8 shows what the last four measures look like when imported into Finale using MusicXML:

Figure 8. Importing SharpEye data into Finale via MusicXML. From Good, 2001: http://www.idealliance.org/papers/xml2001/papers/html/03-04-05.html

Figure 9 shows the last four measures as imported into Finale using MIDI:

Figure 9. Importing data from SharpEye into Finale via MIDI. From Good, 2001: http://www.idealliance.org/papers/xml2001/papers/html/03-04-05.html

The song lyrics are in the MIDI file, but the Finale reader did not import them. Figure 10 shows the last four measures as imported into Sibelius, which can read the lyrics, using MIDI. So while its an improvement in that the lyrics are included, Sibelius uses treble instead of bass clef for the left hand of the piano part. Megan Winget Page 27 7/25/05

Figure 10. Importing data from SharpEye into Sibelius via MIDI. From Good, 2001: http://www.idealliance.org/papers/xml2001/papers/html/03-04-05.html

Its relatively easy to see that the files that use the MIDI import are much less accurate than the MusicXML import. MIDIs shortfall can be attributed to a number of issues, which essentially boil down to the fact that theres a lot of guesswork involved in a MIDI program thats not necessary in MusicXML. For example, MIDI has no note element, but NoteOn and NoteOff events surround notes. Rests are not represented at all, but are simply inferred by the absence of notes. This works well for synthesizers and other electronic musical instruments, but is not robust enough for representation of music notation. Furthermore, in MIDI theres no way to tell the difference between a D-sharp and an E-flat; the one above middle C has a note value of 63 in both cases. In this example, Sibelius guessed the correct note, while Finale did not. MIDI also has no way to represent beams or stem direction, and both programs got that wrong in the voice part. Finally, MIDI misses clefs; Sibelius guessed wrong on one part, where Finale guessed correctly.

Megan Winget

Page 28

7/25/05

Conclusion
Musics notationality is fundamental to the issues surrounding its digital representation. Not only is musical representation exceedingly complicated itself; its symbolic representation system makes it difficult for computers to deal with in a meaningful way. Development of a semantically meaningful representation system for music is a huge step forward for music information retrieval and digital libraries. Not only does it allow for new levels of interaction, editing, and collaboration among users, it provides designers with the opportunity to realistically develop new functionalities for currently unsupported users, like practicing musicians [Bellini, 2004 (Bellini, Nesi, & Spinu, 2002)]. Furthermore, development of MusicXML and its attendant technologies allows researchers to study the use and behavior of music digital libraries users in a way not supported before. Thinking about the musics notation system and the concept of notation generally might also help us gain insight into seemingly unrelated issues in the digital sphere. There are similarities between Goodmans allographic, multiple arts and normal digital documents. His discussion of authenticity and reliable reproduction might be useful for issues surrounding digital preservation and archiving. The complexities surrounding development of a semantically meaningful music representation system might be illuminating for those wishing to develop representation systems for other variable media.

Megan Winget

Page 29

7/25/05

Bibliography
Bellini, P., Nesi, P., & Spinu, N. B. (2002). Cooperative visual manipulation of music notation. ACM Transactions of Computer-Human Interaction, 9(3), 194-237. Boertz, B. (1970). Nelson Goodman's Languages of Art from a musical point of view. Journal of Philosophy, 67(16), 540-552. Giovannelli, A. (2005). Nelson Goodman's Aesthetics. Stanford Encyclopedia of Philosophy, 2005. Gombrich, E. H. (1972). Art and Illusion: A study in the psychology of pictorial representation. Princeton, NJ: Princeton University Press. Good, M. (2001). MusicXML: An Internet-friendly format for sheet music. Paper presented at the XML Conference 2001, Orlando, Florida. Goodman, N. (1976). Languages of art: An approach to a theory of symbols (2nd ed.). Indianapolis, IN: Hackett. Levinson, J. (1980). What a musical work is. Journal of Philosophy, 77, 5-28. MIDI Manufacturers Association. (1996). The complete MIDI 1.0 Detailed Specification. Document version 96.1. Los Angeles: MIDI Manufacturers Association. Pearce, D. (1988). Musical expression: Some remarks on Goodman's theory. In E. Rantala, L. Rowell & E. Tarasti (Eds.), Essays in the philosophy of music (Vol. 43, pp. 228-243). Helsinki: Acta Philosophica Fennica. Robinson, J. (2000). Languages of Art at the Turn of the Century. Journal of Aesthetics and Art Criticism, 58(3), 213-218. Roland, P. (2000). XML4MIR: Extensible markup language for music information retrieval. ISMIR 2000, 2000(1). Walton, K. L. (1993). Mimesis as make-believe: on the foundations of the representational arts (1st Harvard University Press pbk. ed. ed.). Cambridge: Harvard University Press. Webster, W. E. (1971). Music Is Not a "Notational System". Journal of Aesthetics and Art Criticism, 29(4), 489-497. Wollheim, R. (1980). Art and its objects: with six supplementary essays (2d ed.). Cambridge, MA; New York: Cambridge University Press. Megan Winget Page 30 7/25/05

You might also like