Part One
Songs
1
Meanings of Songs and Meanings of Song Performances
Distinct performances of the same song can mean very different things. This is because, in the pithy description of Jeanette Bicknell, âmeaning is a product of three factors: the songâs text, its music, and the performance context.â1 Yet the meaning of the song may be fixed. If this sounds paradoxical, my immediate purpose is to show how the interplay of semantics and pragmatics generates different meanings in different performances.2 The basic idea is that context-dependent aspects of a songâs meaning are different fromâand supplemented byâcontext-dependent aspects of a performanceâs meaning. Popular song performances are rich in meaning because the semantic content of the lyrics is wedded to a very âthinâ musical structure. This combination of semantic information, musical flexibility, and shifting performance contexts generates myriad possibilities for generating pragmatic contextual supplementation of the song during its performances.
Sentences and Songs, Utterances and Performances
Pragmatics examines how the meaning of an utteranceâone particular communication at a particular momentâdepends on the context of utterance. Utterance meaning often diverges from the standard or face-value meaning of a phrase or sentence or larger text. Applied to popular song, it implies that audience interpretation should distinguish meaning-making properties that belong to the song from those that are specific to a particular performance, recording, recontextualization, or some other use of that song.3
Consider the following case. Suppose someone utters a sentence and the referring expression of the sentence is a name or description that misdescribes the speakerâs intended referent. Is the resulting sentence true or false? In a well-known essay, Saul Kripke discusses cases such as âHer husband is kind to her,â erroneously said of a man who is not married to the woman in question.4 A narrow focus on the semantic content of the sentence tells us that the spoken sentence is false if the woman is unmarried. If the woman is married, the truth of the sentence depends on her husbandâs behavior. If the actual husband is unkind to her, then (again) the utterance is false. Yet, married or single, the intended male referent is the man actually observed being kind to the woman in question. So long as the person to whom the speaker addresses the sentence understands the speakerâs intended reference, the utterance will successfully refer to that person, and it will succeed even when the intended audience for the sentence recognizes the failure of semantic reference. Thus, a successful understanding of a speaker often requires us to set aside a strict adherence to the semantic reference of the referring expression, and the speakerâs utterance can be true even if its proper semantic reference should render it false. Thus, utterance meaning diverges from semantic content.
For my purposes, the important point is one that Kripke mentions in passing, but which is central to pragmatics:
The notion of what words can mean, in the language, is semantical: it is given by the conventions of our language. What they mean, on a given occasion, is determined, on a given occasion, by these conventions, together with the intentions of the speaker and various contextual features ⊠together with various general principles, applicable to all human languages regardless of their special conventions.5
In other words, our grasp of the context of utterance can prompt us to amend a sentenceâs established semantic reference. We do so by distinguishing between the sentence type and its utterance on a particular occasion of use, which allows us to achieve a pragmatically correct understanding of the speakerâs reference for that use of that sentence. Correct understanding of âher husbandâ is not merely a grasp of how context fixes the referent of âherâ in the normal way but also of how âhusbandâ operates in an abnormal manner this case.6
This distinction between sentence reference and speakerâs reference can be extended to other cases where the meaning of a sentence type diverges from the meaning of a specific utterance. In short, the conventional meaning of a sentence type does not fully determine the meaning of each and every instance (token) of that type.7 Semantic content is a matter of general conventions, whereas speakerâs meaning (utterance meaning) demands sensitivity to context-sensitive pragmatic implicatures.
The implication for songs and song lyrics is straightforward. Because utterance meaning frequently diverges from sentence meaning, song performances frequently communicate context-sensitive meanings that diverge from the semantic content provided by the songâs words. The performance context sometimes involves a performerâs actual intentions on that occasion. As William Lycan puts it, pragmatics explores how changes in speaker contexts make it possible that âone and the same sentence with an already fixed propositional content can still be used to do interestingly different things in different contexts.â8 Substitute âsong lyricâ for âsentenceâ and one and the same song can be used to do interestingly different things in different performance contexts.
In summary, the meaning of a songâs words contributes to the meaning of each of its performances, but that is merely one factor among several in generating the meaning of any particular performance. Although our arguments are very different, I am expanding on Stan Godlovitchâs general position about the relationship between musical works and their performance: âThe fixity of the work must typically be consistent with the opportunities for novelty expected in performance.â9
Songs as Structural Types
Quite apart from issues of meaning and interpretation, philosophy has long grappled with a fundamental distinction between two basic forms of art. Some artsâmost notably painting, drawing, and some forms of sculptureâresult in artistic products for which the artistic achievement is identified with a particular physical object. The Mona Lisa is painted on a particular wooden panel and Michelangeloâs PietĂ carved into a particular block of marble, and a viewer has only seen the actual artwork if they have been in the physical presence of that wooden panel and that carved marble. Yet no one seriously extends this âphysical object hypothesisâ to musical works.10 When Dolly Parton wrote âJoleneâ (1973), she may have written it out on paper, but no one equates the piece of paper with the song. No one thinks that access to the song depends on access to that one piece of paper. It depends, instead, on access to performances and recordings that follow the music-with-words structure that Parton created. Consequently, most philosophers of art endorse the thesis that the products of some art forms are singular (there is only one âoriginalâ), but others involve works that are multiple in their instantiation, where none of the instances are more genuine than any others.
I will adopt the standard assumption that songs are structural types and I will concentrate on musical performances and recordings that are instances or tokens of those structures.11 The present chapter focuses on works for musical performance, namely songs that are meant to be instantiated through the action of performing them, most often for an audience. The audience may be small, as when âHappy Birthday to Youâ is sung by everyone in the room except grandmother, to whom it is sung because it is her birthday. Or it may be quite large, as when the American national anthem is sung to a crowd of tens of thousands of fans during the opening ceremonies of a university football game.12 Audience size aside, the cases are alike in that the songs are structures that guide performances. Conversely, their performances are instances or tokens of the structural types.
My analysis of song performance draws on four elaborations of this distinction between types and instances.
First, types place constraints on instances. The act of composing a song includes the songwriterâs determination of which features are normative, that is, which ought to be present in performances and other tokens of that type. This point is sometimes made by saying that the musical type is a norm-kind. In the tradition of classical music, the type is usually indicated through the creation and publication of a musical score. In this tradition, scores specify all instrumentation and provide notation and other instructions in order to indicate how the composerâs intentions supplement the standard performance conventions of the musico-historical context.13 However, in the field of popular music, the basic type can be initiated and passed down through oral tradition (as with many âtraditionalâ and folk songs), or generated in the act of recording it, with a minimal score extracted later and published (e.g., much of what we find in Alicia KeysâNote-for-Note Keyboard Transcriptions). Because the identity of a popular song is ultimately score-independent, many variants, and many wildly different arrangements of it, will count as instances of the same basic song.
Most scored, classical works are highly prescriptive and constraining, excluding âsome sonic possibilitiesâ from their correct performances.14 The same cannot be said about popular songs. If you add a vibraphone to a string quartet by W.A. Mozart, itâs not a performance of Mozartâs work. Not so with Kurt Weillâs âMack the Knifeâ (1928). With lyrics by Bertolt Brecht, the song was written as a musical number for the show The Threepenny Opera, a production which Weill and Brecht intentionally undertook as an exercise in popular music theater.15 While the published score does not indicate use of a vibraphone, no one who has any familiarity with the conventions of popular music would say that a performance cannot be a performance of âMack the Knifeâ because it includes vibraphone. And singers often supplement the words of popular songs, as in Ella Fitzgeraldâs famous 1960 live recording of âMack the Knifeâ with extemporaneous lyrics. The Last Town Chorusâs mournful interpretation of âModern Loveâ (2007) features a crawling tempo and a slide guitar. Sonically, it sounds almost nothing like David Bowieâs original, uptempo R&B version (1983). To the casual listener, there is little agreement beyond the lyrics. In this respect, we find that popular song, rather than classical music, exhibits a strong license for meaning alteration by allowing considerable flexibility in the music. With art music, the performance norm is for performers to act in service to the composer and the score.16
Second, every type underdetermines the properties of its instances. This relationship generates the one-over-many principle: there are many instances of one and the same type, and each instance will be distinctive in some way. Each instance will have properties associated with the type and consequently shared with other instances; yet, every instance will have more properties than can be specified in the type.17 If the latter were not so, a performanceâs meaning would almost always align with the typeâs meaning as established by its original context. Another way of expressing this point is to say that types are ontologically thinner than their instances, and because types vary in how many performance features are set as invariant in its performances, some types are ontologically thinner (specifying few) or thicker (specifying many) relative to their determinative properties.18 Popular songs are generally very thin, allowing considerable variation from performance to performance and thus permitting considerable tailoring to the performance context. Whereas it is legitimate to complain that one is not hearing Beethovenâs Violin Concerto in D Major (Op. 61) if the first movement cadenza is played on a kazoo, there is no parallel complaint if a chorus of âHappy Birthday to Youâ is accompanied by kazoos.
Third, the basic properties of a song are determined by the musico-historical context in which it is composed.19 Every song has meaning established by the time and place of its composition. That meaning is its work-meaning and it is a property of that song. By virtue of the first point, that work-meaning ought to be attributed to all instances of that work. However, any such meaning can be nullified by power of pragmatic implicature. Semantic properties are fixed by a workâs musico-historical context. They may inform and contribute to the meaning of subsequent performances, but they do not fully determine the meaning of its performances. The semantic content of the sung words may be exploited in many ways, generating distinct pragmatic supplementation in different performance contexts. In the same way that sentences containing the phrase âher husbandâ permit use of that phrase to refer to someone who is not married, it is unsurprising when a popular songâs semantic content is exploited to communicate meanings that the songwriter would not sanction or endorse for it. Hence, the supposition that all performances will share meanings established by the initial context is not an obstacle to allowing for divergent meanings of different performances. (I discuss this at greater length in Chapter 2.)
Fourth, song instances can count as recognizable instance even when they fail to embody some properties specified by the type. If we do not make this concession, then it would be impossible to have instances/tokens with even a single flubbed note.20 As I said earlier, the musical type is a norm-kind. It is a pattern that can have better and worse realizations, and so an error-ridden or minimally recognizable performances can be counted as instances. A sloppy, ramshackle, out-of-tune performance of âHappy Birthday to You,â or The Replacements performing âLike a Rolling Stoneâ as âLike a Rolling Pinâ (1990), along with other choice word alterations, are nonetheless performances of those two songs.
Ironically, given the variants that arise from contextual supplementation, performance arrangement, and performance style, it is theoretically possible that every performance of a particular popular song might contain enough variation from the composerâs expectations and design that every performance of it will fail to convey the composerâs intended meanings for it. Unlikely in practice, but a theoretical possibility.21
Authorship, Semantic Content, Intentions
This technical overview of ontology has been a prelude to seeing how a single song can communicate many different meanings in its various performances.22 The key issue is whether, or to what degree, a song (as a type) has invariant meaning.
Normally, semantic properties are independent of any speakerâs intentions on any given occasion. Thus, a songâs basic semantic properties are determined by prevailing semantic conventions at the time of composition. If a blues fan thinks that Robert Johnsonâs âTerraplane Bluesâ (1936) is about some kind of airplane, that interpretation is wrong. But we do not have to peer into Johnsonâs private thoughts to know this. In the 1930s, âTerraplaneâ was the name of a brand of automobile, and this bit of public, historical information explains why his lyrics reference the hood, battery, and other car parts, and fails to reference propellors and wings.
Does it follow that a composerâs intentions play no role in determining the songâs meaning on the grounds that semantic conventions override those intentions? As I explained in the Introduction, the celebration of the âdeath of the authorâ and the common prohibition against committing the intentional fallacy have created an environment in which authorial intentions are dismissed as irrelevant to the interpretation of literature and other meaning-structures.23 However, there has been a counter-movement that explores ways that authorial intention contributes to the context that guides interpretation of composed texts.24
Consistent with recourse to pragmatics, I will offer a number of reasons to think that the counter-movement is correct and that performersâ intentions play a role in explaining how utterance meanings diverge from semantic meanings. In terms of the pragmatics of popular song performance, the crucial point is that we must distinguish authorial intentions of a songâs compose...