THEODORE GRACYK

Meanings of Songs and Meanings of Song Performances

Distinct performances of the same song can mean reference away from the intended male referent. very different things. Yet the meaning of the song If the woman is married, the truth of the sentence may be fixed. My immediate purpose is to show depends on her husband’s behavior. If the actual how the interplay of semantics and pragmatics husband is cruel to her, then the assertion is false. generates different meanings in different perfor- Or suppose the woman is not married. Following mances. Context-dependent aspects of a song’s our semantic conventions, the sentence is false, for meaning should not be confused with context- it is false if she is not married. Yet, married or sin- dependent aspects of a performance’s meaning. gle, the intended male referent is the man actually My more general purpose is to advance our un- observed being kind to the woman in question. So derstanding of musical meaning. Musical perfor- long as the person to whom the speaker addresses mances are rich in meaning because semantic in- the sentence understands the speaker’s intended formation, contingently associated with a musical reference, the speech act will successfully refer to structure, generates myriad possibilities for gen- that person and it will succeed even when the in- erating pragmatic contextual supplementation of tended audience for the sentence recognizes the the work during its performances. failure of semantic reference. Thus, a successful understanding of a speaker often requires us to set aside a strict adherence to the semantic refer- i ence of the referring expression, and the speaker’s utterance can be true even if its proper semantic My central proposal is an application of core ideas reference should render it false. in philosophy of language. Audiences interpret art For my purposes, the important point is one that and music, and audience interpretation involves Kripke mentions only briefly, but which is central decisions about which properties belong to the to his analysis: works being interpreted and which belong only to a particular performance of that work. The notion of what words can mean, in the language, is Consider the following case. Suppose someone semantical: it is given by the conventions of our language. utters a sentence and thereby makes a claim, but What they mean, on a given occasion, is determined, the referring expression of the sentence is a name on a given occasion, by these conventions, together with or description that does not refer to the speaker’s the intentions of the speaker and various contextual fea- intended referent. Is the resulting sentence true tures . . . together with various general principles, appli- or false? In a well-known article, Saul Kripke dis- cable to all human languages regardless of their special cusses cases such as “Her husband is kind to her,” conventions.2 said of a man who is not married to the woman in question.1 On Kripke’s analysis, a narrow focus on the semantic content of the sentence tells us that In other words, Kripke distinguishes be- the spoken sentence is true only if the woman (that tween the sentence’s semantic reference and the is, the woman referred to by the gendered pro- speaker’s reference by distinguishing between the noun) is married and the man in question is kind sentence type and its contextualized tokens on to her. But this focus on literal meaning may divert particular occasions of use. Our understanding of

C 2013 The American Society for Aesthetics 24 Song, Songs, and Singing a specific context of utterance can prompt us to sis” to a Beethoven symphony or any other mu- overlook the established semantic reference, al- sical work.6 Consequently, most philosophers of lowing us to achieve a pragmatically correct un- art endorse a bipartite or dualist theory accord- derstanding of the speaker’s reference for that to- ing to which the products of some art forms are ken of that type. Correct understanding of “her singular and others are multiple in their physi- husband” is not merely a grasp of how context cal instantiation. I will proceed on the assumption fixes the referent of “her” (so-called “semantic that musical works are types and I will concen- pragmatics”) but also of how “husband” operates trate on musical performances that are tokens of in this case (so-called “pragmatic pragmatics”).3 types.7 As such, I will focus on works for musi- This distinction between sentence reference cal performance, namely, songs that are meant to and speaker’s reference can be extended to other be instantiated through the action of performing cases where the meaning of a sentence type di- them for an audience. The audience may be small, verges from the meaning of its utterance tokens. as when “Happy Birthday to You” is sung by ev- In short, the meaning of a sentence type does not eryone in the room except grandmother, to whom fully determine the meaning of each and every to- it is sung because it is her birthday, or it may be ken of that type.4 As a result, utterance meaning quite large, as when the national anthem is sung frequently diverges from sentence meaning. The to a crowd of 100,000 fans during the opening cer- same gap emerges for other meaningful types with emonies of a university football game. Audience multiple iterations, as when a song has multiple size aside, the cases are alike in that the songs are performances. types and their performances are tokens of the In philosophy of language, this divergence is ex- types.8 plained by recourse to the distinction between se- My analysis of song performance draws on four mantics and pragmatics. As William Lycan puts it, elaborations of the proposal that musical works pragmatics explains how changes in speaker con- are types and performances are tokens of types. texts make it possible that “one and the same sen- First, types limit the range of possibilities for ac- tence with an already fixed propositional content ceptable tokens of that type. The act of creat- can still be used to do interestingly different things ing or indicating a type is the act of determining in different contexts.”5 Semantic content is a mat- which properties are normative, that is, which ter of general conventions, whereas a speaker’s ought to be present in tokens of that type. Thus, meaning involves sensitivity to pragmatic contex- eighteenth-century composers initiated the prac- tual supplementation. That context sometimes in- tice of specifying instrumentation and provid- volves the speaker’s actual intentions on that occa- ing notation and other instructions in order to sion. While numerous arguments have been raised indicate how their intentions supplemented the against the relevance of a speaker’s intentions, standard performance conventions of the musico- the thesis that interpretation must sometimes con- historical context.9 The type can also exclude some sider a speaker’s intentions is highly plausible, and properties. For example, most modern works pre- I endorse it. scriptively exclude “some sonic possibilities” from their correct performances.10 This point invites consideration of whether the absence of linguis- ii tic texts and other semantic content from a work necessarily excludes certain interpretations of its Let us return to philosophy of art. Quite apart performances. However, the established semantic from issues of meaning and interpretation, ontol- content of sentences containing the phrase “her ogy of art has long grappled with a fundamental husband” does not preclude using that phrase to distinction between two basic forms of art. Some refer to someone who is not married. So we should arts—most notably painting, drawing, and some not be surprised to find that a work’s semantic con- forms of sculpture—result in artistic products for tent, or lack of it, does not preclude communica- which the work in question is identified with a par- tion of ideas that the composer would not sanction ticular physical object, so that the Mona Lisa is a or endorse for it. particular piece of painted canvas and Michelan- Second, the actual properties of a musical work gelo’s Pieta` is a shaped block of marble. Yet no one are determined by the original context. A work’s seriously extends this “physical object hypothe- original context is the musico-historical context in Gracyk Meanings of Songs and Meanings of Song Performances 25 which it is created.11 Ifaworkhasameaninges- mance of “Happy Birthday to You” is nonetheless tablished by the time and place of the work’s com- a performance of it. As a consequence, it is pos- position, that meaning is a work-meaning, and it sible that few musical works have had a perfect is a property of that work. By virtue of the previ- instance, for it may happen that no performance ous point, these meanings ought to be attributed will provide access to every actual property of that to all tokens of that work. As with the previous work. Thus, it is theoretically possible that every point, this one has important implications relating performance will contain deviations from the type to the semantics/pragmatics distinction. Seman- that will facilitate an utterance meaning that con- tic properties that are fixed by a work’s musico- flicts with the work’s own meaning.15 historical context constrain but do not fully deter- mine the meaning of all subsequent performances. iii With songs, the semantic content of the sung words may be exploited in many ways, generating dis- This overview of ontology was offered as a prelude tinct pragmatic supplementation in different per- to making better sense of how a single song can formance contexts. Hence, the requirement that have many meanings in its various performances.16 all tokens will share meanings established by the My strategy is to exploit the obvious point of initial context is not an obstacle to allowing for intersection between philosophy of language and divergent meanings when performed. this ontology of musical works: I assume that Third, a type underdetermines the properties the semantics/pragmatics distinction maps onto of its tokens. This relationship generates the one- the type/token distinction in the same way for over-many principle: there are many tokens of one both topics.17 Semantics is not concerned with and the same type, and each token will be dis- utterance context, whereas pragmatics examines tinctive in some way. Each token will have more how contexts of utterance affect the meaning of properties than does the type it instantiates, yet an utterance. Therefore, semantic properties be- each token will have properties in common with long to sentences, but both semantic and prag- both the type and with other tokens.12 One way of matic properties belong to utterances.18 On my expressing this point is to say that works are on- parallel analysis, many musical works have seman- tologically thinner than their tokens, and because tic content through their association with specific some works delineate more of the token’s proper- linguistic structures or because their syntactical ties than do other works, some works are ontolog- structures function symbolically due to musical ically thicker or thinner than other works.13 If this conventions.19 However, musical works for per- were not so, a performance’s meaning would not formance lack pragmatic content. Only specific diverge from the work’s meaning as established by performances have pragmatic content. As a re- its original context. Songs are generally very thin, sult, one and the same work with an established allowing considerable variation from performance semantic content can be used to do different things to performance and thus permitting considerable in different performance contexts. tailoring to the performance context. Whereas it My analysis endorses Kripke’s point that utter- is legitimate to complain that one is not hear- ance context can include a speaker’s intentions ing Beethoven’s Violin Concerto in D Major (Op. on that occasion. In contrast, semantic proper- 61) if the first movement cadenza is played on a ties are normally independent of any speaker’s kazoo, there is no parallel complaint if a chorus intentions on any given occasion. Applying this of “Happy Birthday to You” is accompanied by insight to musical works, a work’s semantic prop- kazoos and bongos. erties are normally independent of a composer’s Fourth, a genuine token can deviate from the intentions concerning it. Does it follow that a com- type by failing to embody some properties that poser’s intentions play no role in determining a should be present. It makes no sense to sup- work’s meaning? Due to the entrenched prohi- pose that every token must possess every stan- bition against committing the intentional fallacy, dard property,for that would make it impossible to followed in turn by the supposed “death of the have instances/tokens with even a single flubbed author,” authorial intentions are frequently re- note.14 In the case of musical works meant for garded as off limits when determining the mean- performance, normative constraints must be rea- ing of a literary work.20 However, there has been a sonable. A sloppy, ramshackle, out-of-tune perfor- renewal of interest in ways that authorial intention 26 Song, Songs, and Singing contributes to the context that guides interpreta- Jimi Hendrix’s instrumental performance of “The tion of literary texts. I will not review the literature Star-Spangled Banner” on August 19, 1969, at here.21 I call attention to it to make it clear that the Woodstock Festival. Later, I provide a brief there is nothing amiss in holding that intentions discussion of Bob Dylan’s performance of “It’s play a role in explaining how utterance meanings Alright, Ma (I’m Only Bleeding)” as performed diverge from semantic meanings. on February 14, 1974. There is nothing very spe- Robert Stecker offers a robust defense of an cial about these examples beyond the fact that in utterance meaning account of artistic meaning. each case we know the day of the performance He is not alone in defending the view that, and there is ready access to high-quality, unedited whether we are interpreting a Beethoven con- recordings. In both cases, the year of the perfor- certo or a Shakespearean sonnet, the meaning of mance is relevant to grasping the proper context any work of art is its utterance meaning. Focusing for interpreting it. In Hendrix’s case, the perfor- on literary meaning, Stecker proposes that an mance is an act of political protest, challenging artwork has whatever meaning that it has due to the continuing involvement of the United States a combination of “the actual intentions of artists in Vietnam. In Dylan’s case, the performance calls and the conventions in place when the work is attention to the congressional impeachment hear- created.”22 Work meaning is utterance meaning, ings underway for President Richard Nixon. Yet a function of both semantic and pragmatic neither song does those things. My goal is to ex- meaning.23 In other words, the meaning of any plain how the performances mean what they mean particular artwork is a function of the semantic without it being the case that the songs have some- content appropriate to its structure in its original how altered their meanings. linguistic and art-historical context, together Before we turn to the Hendrix performance, with the composer’s intentions for interpreting consider the reference to the United States that is that structure in that context. Those intentions implicit in all performances of the American na- determine whatever “is pragmatically conveyed” tional anthem, “The Star-Spangled Banner.” Why by the work, extending the work’s utterance do performances refer to the United States? No meaning beyond its semantic content.24 Stecker’s country is named in the lyrics. Instead, the words primary motivation for this view is that literary describe a flag. However, the flags of Uzbekistan, texts normally mean something other than what- New Zealand, and Cape Verde also have both ever can be derived from their literal, semantic stripes and stars. Taken at face value, Francis Scott meaning. The presence of a work meaning is one Key’s referring description refers equally well to of the things that distinguishes the basic text, a those flags too. However, Key intended the lyrics structural type, from the artwork, which is the of “The Star-Spangled Banner” to refer to the flag text produced in a particular historical context.25 of the United States, and he further encodes this Literary and musical works are, therefore, intention into his lyrics by using the term “our” utterances. They are structures-in-use.26 in two stanzas, as in the line, “gave proof through However, Stecker goes astray in assigning both the night that our flag was still there.” Taking Key semantic and pragmatic properties to the type. to be the speaker of the indexical term “our,” pre- Only the former belong to the type. Pragmatic vailing norms tell us that the flag is that of Key’s properties intended by an author or a composer country, the United States. By respecting autho- belong to subsequent tokens only if so intended by rial intention as a relevant part of the originating the performer who constitutes the “speaker” on a context, we preserve the idea that original context given occasion. In the next two sections I offer ex- determines the type’s semantic content. Hence, a amples of song performances that will clarify my “semantic pragmatics” for indexicals says that the assertion that musical works are not structures-in- words of this song refer to the flag of the United use, the meanings of which are fixed and norma- States, no matter how many other flags are equally tive for all correct instances of the work. well described or who sings the words. However, it is not obvious that Key’s author- iv ship is sufficient to establish that “our flag was still there” refers to the United States in every perfor- To motivate my analysis, I want to consider two mance. The use of the pronoun has not changed in particular song performances. The first example is two hundred years, and the literal meaning of this Gracyk Meanings of Songs and Meanings of Song Performances 27 expressionisthesametodayaswhenhewroteit. of utterance are not jointly sufficient to account Key intended others to sing the song, and he may for our understanding of the pragmatic content even have intended all those singers to be Amer- of every use of it. Whenever utterance meaning icans. But what if they are not Americans? What is unclear, our grasp of the intentions directing happens if a New Zealander sings it while saluting the act of producing a particular token of a sen- a New Zealand flag? Why should Key’s intentions tence “picks up the slack” in linking pragmatic preclude the New Zealander’s use of the phrase content to the utterance.30 I am proposing that “our flag” to refer to her own country’s flag? the same holds of song performances construed as Rather than assume that authorial intentions analogues of utterances. Hence, a performer’s in- decisively preclude this use, consider a nonmusi- tentions are analogously relevant to performance cal case. Blessed with sudden good fortune, some- meaning. Thus, “our flag” should be interpreted one might say, “now is the winter of our discon- in terms of the singer’s intentions that are, pre- tent made glorious summer.”27 Many lines from sumably, to refer to the American flag, as did Key. Shakespeare are used by people who do not know In the case of Hendrix, we take that phrase to be that he authored them, and this speaker might relevant even in its absence from the particular, use the line without knowing that its source is purely instrumental performance. Shakespeare’s Richard the Third. In its original Second, this approach reaffirms Stephen and intended context, its semantic reference is Davies’s proposal that performers only perform Richard and, perhaps, the house of York. How- a particular work by intending to honor a com- ever, there is no reason to suppose that someone poser’s work-determinate intentions for it.31 We who repeats the line out of context, in “real life,” consult the speaker’s intentions to decide whether thereby refers to Richard and the house of York. a particular use of “Now is the winter of our dis- In the play, it means what it does as a result of content made glorious summer” is a theatrical use, our standard interpretive norms together with an preserving Shakespeare’s intentions, or whether it acknowledgment that the context is theatrical fic- is a nontheatrical, “real” reference to current cir- tion. Quoted outside a performance of the play, cumstances (or both, where the former is means to there is no reason to suppose that its origins in the latter). Consistent with Davies’s proposal, the Shakespeare’s play are relevant to determining its importance of considering the intention of the per- current range of application.28 former is already implicit in discussing Hendrix’s In light of this example, reconsider “The Star- instrumental performance of “The Star-Spangled Spangled Banner.” What happens when Celine´ Banner” at Woodstock. As an instrumental per- Dion, a Canadian citizen, sings it? Why is her per- formance, semantic referential content is present formance about the United States (which, I take only because of its status as a song in which Key it, it is), rather than a reference to Canada and associated particular words with particular mu- to some event or other in Canadian history? Two sical phrases. Those who know the song will be explanations might be offered. At best, “our flag able to attach particular lines to particular melodic was still there” might be a case where reference phrases. However, the mere fact that Hendrix has become so habitual in the community that the sounds as if he is playing “The Star-Spangled Ban- phrase has evolved a special semantic reference.29 ner” is not sufficient to demonstrate that his per- On the other hand, a simpler explanation is that formance is an instrumental performance of that the reference to the United States is established work, and that his performance is to be interpreted pragmatically, by a singer’s intentions to honor the in light of Key’s lyrics. Why was it a performance composer’s intentions. of that song at all, rather than “To Anacreon in The latter solution is more plausible for two Heaven,” the earlier English song upon which reasons. First, it parallels the idea that prag- the American anthem is based? Hendrix’s per- matic meaning can involve consideration of the formance of “To Anacreon in Heaven” might be speaker’s intentions. More specifically,I am adapt- perceptually indiscernible from his performance ing Kripke’s point that the speaker’s intentions of “The Star-Spangled Banner.” The identity of are an element of the context that allows “her the musical work being performed is surely de- husband” to refer to someone who is not her hus- termined by reference to Hendrix’s performance band. Absent this concession about intentions, the intention. If we are wrong about his intention, phrase’s semantic content together with a context we are not justified in supposing that Hendrix’s 28 Song, Songs, and Singing musical performance involves a semantic refer- the twenty-four notes have a highly specific il- ence to the United States. locutionary function. During his Woodstock per- We want to determine this reference, in part, be- formance of the national anthem, Hendrix inter- cause at one point Hendrix’s guitar pyrotechnics rupts the musical melody of the line “Oh, say (following what would be the sung line “bombs does that star-spangled banner yet wave” by in- bursting in air”) are then plausibly construed as serting the opening phrases of “Taps” into the musical representations of bombs bursting in air. middle of the anthem’s melodic line. Hendrix ex- This recognition is facilitated by Hendrix’s pro- pected his audience to understand this interpola- viding enough of the melody to make it clear tion as a reference to death without their possess- that the absent lyric has just referred to bombs. ing shared knowledge of the official fixed meaning This reading generates new interpretive questions. of the bugle call. Thus, Hendrix’s arrangement of Who is bombing whom? The British, shelling the music introduces musical material for prag- Fort McHenry in 1814? The consensus is oth- matic interpretation without relying on audience erwise: Hendrix is calling attention to the on- knowledge of an associated English text. (His- going war in Vietnam. He refers to bombs, but torically, the lyrics associated with “Taps” were not Key’s bombs. Where Key’s utterance meaning added many years after its meaning was firmly is, broadly, pride in the resilience of the United established.)34 Independent of words, musical de- States, Hendrix is offering a “musical protest tails contribute to a performance’s pragmatic con- against the United States’s war in Vietnam.”32 tent through both conventional and unconven- Because Hendrix’s performance refers to some- tional implicature (for example, “Taps” and the thing that is not referred to by Key’s initiated “pictured” bombs, respectively). type, Hendrix’s utterance meaning involves addi- In summary, Hendrix’s Woodstock perfor- tional pragmatic considerations arising from the mance stands to the song as an utterance of a structure-in-use in 1969. The song composed in sentence stands to that sentence. Hence, he could 1814 is a structure-for-use, not a structure-in-use, exploit the cultural context of 1969 in order to and there is no reason to suppose that later tokens generate pragmatic implications that are not part cannot have pragmatic meanings beyond those an- of the musical work. ticipated by the artist who indicated the type and thus fixed the semantic conventions that were to be relevant to its subsequent use. To offer another v version of my central point from §I, “semantics concerns sentences, not utterances.”33 That is, a At this point, skeptics will object that, “Taps” sentence is a type, and the type possesses semantic aside, song lyrics are doing all the heavy lifting of content. This semantic content is compatible with the analysis. Lyrics contribute the semantic con- an array of utterance meanings. The full meaning tent that sets up the pragmatic dimension of the of each linguistic utterance—meaning as a combi- performance. The objection reasons that because nation of semantics and pragmatics—depends on illocutionary meaning arises from words, my dis- the context in which the utterance, the particular cussion of songs does nothing to illuminate the token, is produced. I am contending that the same contribution of music, nor can it extend to music is true of songs and their performances, as exem- that does not feature words as essential properties plified by Hendrix’s performance of this song. of the work type. However, I am not proposing that pragmatic This objection can be defused with any num- implications always depend on the presence of ber of examples demonstrating that musical prop- lyrics and other associated texts. Many musical erties make an independent contribution to our phrases carry conventional implicature apart from understanding of the lyrics of songs and other any such association. For example, the melody sung texts. Consider Randy Newman’s hit record- known as “Taps” is an official American mili- ing of his own song, “Short People” (1977). Lyrics tary bugle call that signals the end of the day. aside, the aggressive staccato lines that dominate However, it is frequently played at military fu- the music later give way to a gentle, soaring pas- nerals and memorial services, in which context sage in the “bridge” section. Simultaneously, New- listeners can understand it to be an expression man’s harsh nasal timbre is smoothed with accom- of grief and farewell without understanding that panying vocal harmonies. Not surprisingly, these Gracyk Meanings of Songs and Meanings of Song Performances 29 musical decisions signal a shift in the lyrics, sug- dal and ensuing political crisis, the crowd went gesting that the meaning of the previous verses wild at the lyric “(E)ven the president . . . Some- and choruses is not the whole statement. The times must have/To stand naked.” In that con- change in the music communicates that a com- text, the referring expression “the president of peting voice has entered into conversation with the United States” called attention to Richard the dominant one, setting the new lyrics against Nixon in a way that it did not call attention to what has been previously “said.” Chester A. Arthur.38 Taken literally, Dylan’s lyric Furthermore, purely musical humor seems in- refers to each and every president, and thus to explicable unless humorous passages are con- Nixon, Barack Obama, and Arthur. Pragmatically, strued as speech acts.35 Although these acts are in the political context of 1974, Dylan singled out sometimes built into the work itself, as with Nixon. However, general context always under- Mozart’s Ein musikalischer Spaß (K. 522, “A Mu- determines pragmatic implications, so it remains sical Joke”), that is not always the case. Perform- possible that Dylan did not emphasize Nixon.39 ers can interject humor that is not inherent in the Deciding that his performance singled out Nixon type. Frank Zappa often did so with the satirical requires a decision about his intentions in per- musical embellishments he supplied when “cover- forming that song on that occasion. ing” music written by others.36 Hence, my analysis As a general principle, it may be the case that of the legitimate divergence of work and perfor- our ability to produce an acceptable paraphrase mance meanings extends to cases where pragmatic of a song performance’s pragmatic implications meaning does not arise from associated language. is inversely proportional to the degree that we Lyrics aside, a performance of sad-sounding music must consider specifically musical decisions. Musi- can be the illocutionary act of expressing sadness, cal nuances can overwhelm our capacity for para- given the proper context, and therefore the illocu- phrase. tionary act can be subject to sincerity conditions Second, utterances are acts. In ordinary speech, even if there is no propositional content that as- they are “acts of producing tokens of sentences.”40 serts truthfully or falsely. By injecting musical el- Musical performances are likewise acts: acts of ements that undercut sincerity, Zappa subverted producing tokens of musical works. The actions authorial intentions about a work’s illocutionary of uttering and performing are the true bearers of force. pragmatic content.41 This result might be regarded as a mere elaboration of the point that sensitivity to a speaker’s intentions allows us to make sense vi of an utterance in a context. However, it merits separate attention because it clarifies the point Two additional points arise from the parallel be- that intentions and pragmatic content are proper- tween linguistic and musical types (for example, ties of an act of producing a token of a type. They between sentences and songs). are never properties of sentences, which is to say First, pragmatic content is often difficult to that we should not attribute intentions and prag- paraphrase. Although it is plausible to seek the matic content to the sentence that is being uttered. pragmatic dimension of every performance or mu- For that would be to collapse semantic meaning sical utterance, it is unlikely that we are in a posi- and pragmatic meaning. Therefore we should not tion to produce a precise linguistic gloss on Hen- attribute pragmatic meaning to the musical work drix’s intended meaning when performing “The being performed. The type has semantic content Star-Spangled Banner.” For example, while it is that informs yet underdetermines the pragmatic plausible to say that Hendrix’s introduction of content of any of its utterances or performances. “Taps” is a reference to “the death of American This second point is a strong reason to re- soldiers in Vietnam,” there is little reason to re- ject Stecker’s position that a musical work is a gard it as a reference to “the death of the ‘Ameri- structure-in-use. Some properties simply cannot can Dream.’”37 In contrast, a somewhat more de- be properties of some types. Pragmatic content is terminate interpretation appears to fit parts of such a case. Hence, composers’ known intentions Dylan’s performance of “It’s Alright, Ma (I’m about those meanings are not work-determinative; Only Bleeding)” when he sang it on February 14, they do not determine anything about their works. 1974. Sung at the height of the Watergate scan- For example, Mozart’s intentions determine that 30 Song, Songs, and Singing his Piano Sonata in C Major (K. 545) is to be I am not denying that many or even most perfor- played in C Major. However, the sonata cannot mances successfully convey pragmatic meanings have the property of being striped teal and mauve. that composers hoped would be present. How- It is not impossible for a created type to have this ever, when performances do carry the pragmatic among its normative properties. A wallpaper de- implications intended by the composer, those im- sign might be built around these properties, but plications are not present by virtue of their being not a piano sonata. Hence, Mozart’s intentions, among the work’s properties. They will be there whatever they were, could not make being striped because the work is performed in an appropri- teal and mauve a property that its tokens ought ate context or with appropriate intentions. Con- to have. Likewise, pragmatic content cannot be versely, a performance could be a sonically perfect a property of musical types. Consequently, musi- instance without possessing any of the prag- cal works are not full-fledged types-in-use. Finally, matic meanings that its composer intended for it. whatever is not an actual property of a work is not Haydn’s “The Seven Last Words from the Cross” normative for its instances. So a composer’s in- is an instrumental work with an associated reli- tentions about the pragmatic implications of its gious text. Performed during Lent as part of a re- performances cannot be normative for any of its ligious service, it encourages and guides Christian performances. Mozart’s intentions could not give reflection on Christ’s suffering and sacrifices.43 the Piano Sonata in C Major any pragmatic con- However, it is not inevitable that, because tent that its tokens ought to have. At best, the com- Haydn designed it to be used in that way, every poser can intend a preferred use of the work. But use has the illocutionary force that he intended the composer cannot do more than hope that per- for it. Performed in a secular setting without inter- formers will respect and reaffirm those intentions. polating the intended spoken sermons, a sonically I have endorsed the general idea that a sentence perfect instance of the piece can be appreciated and a song are alike in being types whose semantic as absolute music, as a set of Haydn adagios. De- content is governed by shared conventions. Fur- spite knowing the work’s history, the audience for thermore, song performances are like sentence a secular performance need not construe it as an utterances. Both are structures-in-use. Because invitation to reverently reflect on Christ’s passion. sentences are not structures-in-use, they cannot Although I know of no such case, a particular per- have pragmatic content. In parallel fashion, no formance might be used to mock Haydn’s inten- song has pragmatic content. At best, its perfor- tions for it. Consider, for example, The Roches’ mances will have the pragmatic content intended performance of G. F. Handel’s “Hallelujah Cho- by its composer(s) whenever subsequent perform- rus” for the television program Saturday Night ers choose to respect those intentions. Stecker is Live (November 17, 1979). Their performance, ar- right to propose that literary works are contextual- ranged for three voices and sung with an over- ized structures-in-use, where that context includes enunciated New York accent, displays a combi- relevant authorial intentions. However, it does not nation of bravado and disrespect that makes the follow that the creative act uniformly bestows ut- performance playfully sacrilegious. Given an audi- terancemeaningonthetype. Literature and mu- ence who knew it well enough, the same could be sic may work differently. Relatively few modern done with Haydn’s “The Seven Last Words from literary works are works for performances.42 In the Cross.” contrast, songs are works for performance. Perfor- Performances, like utterances, are acts. Because mance practice is at odds with the ideal of each to- the performer’s act of generating a token can oc- ken’s having the same pragmatic meanings as the cur in a very different context than that which musical work. Because he intended others to use informs the composer’s act of creating the type, “The Star-Spangled Banner,” Key’s authorship of full understanding of a performance can require the words “our flag” does not unequivocally de- attention to two distinct contexts. In the Hendrix termine semantic and pragmatic reference to the case, there is a gap of 150 years; in Dylan’s case, United States in every subsequent performance. a mere decade. Either way, the site of pragmatic Authorial intentions guide but do not decisively meaning is the act of producing the token. The type limit the pragmatic meanings associated with a contributes semantic but not pragmatic meaning. song’s incarnations as a structure-in-use, namely, Even if the same pragmatic content happens to its performances. emerge from every token of a particular type, it Gracyk Meanings of Songs and Meanings of Song Performances 31 does not follow that the type possesses a corre- THEODORE GRACYK sponding property. Taken together, these points Department of Philosophy indicate that it will always be possible to produce Minnesota State University Moorhead a perfect token of a musical type while simulta- Moorhead, Minnesota 56563 neously deviating from its customary pragmatic internet: [email protected] implication, as when a properly contextualized song performance is ironic when the song itself is not. 1. Saul Kripke, “Speaker’s Reference and Semantic Reference,” Midwest Studies in Philosophy 2 (1977): 255– vii 276, at p. 261. Kripke’s disagreements with Keith Donnellan are not germane to my argument. Furthermore, other sem- inal articles in philosophy of language could take the place Pursuing the implications of the analogy with sen- of Kripke here. tences and their use, I have argued that the musical 2. Kripke, “Speaker’s Reference and Semantic Refer- analogue of a sentence is a work, an indicated type. ence,” p. 263, emphasis added. The analogue of the sentence-in-use (an intention- 3. This way of distinguishing two distinct contributions guided act of producing a token of the sentence) is of context is from Max J. Cresswell, Logic and Languages (London: Methuen, 1973), p. 238. the work-in-performance (an intention-guided act 4. I have been influenced by Herman Cappelen and of producing a token of the work). In the same way Ernest Lepore, “A Tall Tale: In Defense of Semantic Mini- that sentences have actual semantic content and a malism and Speech Act Pluralism,” in New Essays in the Phi- wide array of possible pragmatic uses, songs and losophy of Language, eds. Maite Ezcurdia, Robert J. Stain- ton, and Christopher Viger (University of Calgary Press, other musical works that possess semantic content 2004), pp. 3–28. will always permit an array of possible pragmatic 5. William G. Lycan, Philosophy of Language: A Con- uses.44 Accounts of meaning that ascribe prag- temporary Introduction (London: Routledge, 2008), p. 142. matic content to songs and other musical works 6. The hypothesis is so called by Richard Wollheim, Art are mistaken in expecting works to fix properties and Its Objects (Cambridge University Press, 1980), p. 4. 7. I am excluding consideration of pure improvisation, of performances that they cannot determine. Acts where no work constrains the performing. of generating tokens of a type convey pragmatic 8. Instances or tokens of works for musical perfor- content, but that content should never be counted mances can also be secured through playbacks of recorded among the actual properties of the type. The actual music. To simplify the analysis, I will treat recordings as transparent representations of particular performances. properties of a musical work will always underde- However, I have argued elsewhere that we must qualify the termine the meaning of each performance because achievement of transparency (see my Rhythm and Noise: An a work has no pragmatic content to bequeath to Aesthetics of Rock [Duke University Press, 1996] and “Doc- its instances. umentation and Transformation in Musical Recordings,” in Although our arguments are very different, I Recorded Music: Philosophical and Critical Reflections,ed. Mine Dogantan-Dack [Middlesex University Press, 2008], have therefore been supporting Stan Godlovitch’s pp. 61–81). general position about the relationship between 9. “Composers take for granted the musical practices musical works and their performance: “the fix- they share with the contemporary musicians to whom they ity of the work must typically be consistent direct their instructions. As a result, they say no more than is necessary.” (Stephen Davies, Musical Works and Perfor- with the opportunities for novelty expected in mances: A Philosophical Exploration [Oxford University 45 performance.” We both stress that the actual Press, 2004], p. 60.) properties of musical works underdetermine their 10. Stephen Davies, “John Cage’s 433:IsItMusic?” performances. However, I have not said that the Australasian Journal of Philosophy 75 (1997): 448–462, at semantic content of works underdetermines the p. 458. 11. See Jerrold Levinson, “What a Musical Work Is,” in semantic content of their performances. I have Music, Art and Metaphysics (Cornell University Press, 1990) claimed that the actual semantic content of a work pp. 63–88. Indexical determination of the original context and its performances will coincide in correct in- generates a work’s properties without thereby belonging stances of the work, setting the stage for pragmatic to the resulting set of properties; see Stephen Davies, “In- terpreting Contextualities,” Philosophy and Literature 20 content that exists only in the confluence of work, (1996): 20–38, at p. 22. performing context, and, frequently, performers’ 12. See Nicholas Wolterstorff, “Toward an Ontology of intentions.46 Art Works,” Nousˆ 9 (1975): 115–142. Wolterstorff denies 32 Song, Songs, and Singing that tokens literally share properties with types. He pro- 28. Linguistic conventions at the time of composition poses that they share predicates that direct us to analogous will assign semantic properties to the language of a play properties (pp. 125–129). or the words to a song. Consequently, a performance of 13. Stephen Davies, “The Ontology of Musical Works “The Lass of Aughrim” that employs the standard lyrics and the Authenticity of Their Performances,” Nousˆ 25 always refers to a girl from a town in County Wicklow, Ire- (1991): 21–41. land. However, as with “her husband” in Kripke’s example, 14. For a contrasting approach, see Jerrold Levinson, a performer may use that reference to Aughrim to refer to “What a Musical Work Is,” Journal of Philosophy 77 (1980): some other place (for example, Ireland in general). Similarly, 5–28. Against Levinson, see Davies, Musical Works and Per- songs embedded in plays generally represent specific illocu- formances, pp. 159–160. tionary acts, but that does not preclude innovative pragmatic 15. “Many sentences are commonly used only to con- exploitation, setting up a different pragmatic implication in vey something other than their literal content, or would be that performance context. so used if they were used at all.” (Nathan Salmon, “Two 29. Kripke, “Speaker’s Reference and Semantic Refer- Conceptions of Semantics,” in Semantics versus Pragmat- ence,” p. 271. ics,ed.Zoltan´ Gendler Szabo´ [Oxford: Clarendon, 2005], 30. Kent Bach, “Context ex Machina,” in Semantics ver- pp. 317–328, at p. 320.) sus Pragmatics,ed.Zoltan´ Gendler Szabo(Oxford:Claren-´ 16. An alternative position says that the changes in don, 2005), pp. 15–44, at p. 38. meaning reflect post-creation changes to the properties of 31. Davies, Musical Works and Performances, pp. 163– the work. This position is defended by Sondra Bacharach, 175. In terms of Davies’s analysis, the relevant intention “Toward a Metaphysical Historicism,” The Journal of Aes- would be a higher-level rather than a lower-level one, where thetics and Art Criticism 63 (2005): 165–173. However, only the former involves “the intention to perform X’s work, Bacharach’s analysis is undercut by a failure to explore the where ‘X’ is a name” (p. 171). distinction between works and their instances in the per- 32. Eric F. Clarke, Ways of Listening: An Ecological forming arts. Approach to the Perception of Musical Meaning (Oxford 17. Exploration of this point differentiates my analy- University Press, 2005), p. 46. See also Douglas Dempster, sis from that of Thomas Carson Mark, “Philosophy of Pi- “How Does Debussy’s Sea Crash? How Can Jimi’s Rocket ano Playing: Reflections on the Concept of Performance,” Red Glare? Kivy’s Account of Representation in Music,” Philosophy and Phenomenological Research 41 (1981): The Journal of Aesthetics and Art Criticism 52 (1994): 299–324. 415–428. 18. Reasons favoring this characterization of the se- 33. Bach, “Context ex Machina,” p. 22. mantics/pragmatics distinction are summarized by Zoltan´ 34. Clarke erroneously overemphasizes the role of as- Gendler Szabo,´ “The Distinction between Semantics and sociated language in interpreting Hendrix’s quote from Pragmatics,” in The Oxford Handbook of Philosophy of “Taps” (Clarke, Ways of Listening,p.58). Language, eds. Ernest LePore and Barry C. Smith (Oxford 35. Justin London, “Musical and Linguistic Speech University Press, 2006), pp. 361–389; it corresponds closely Acts,” The Journal of Aesthetics and Art Criticism 54 (1996): to the view of Robert Stalnaker (“Pragmatics,” Synthese 22 49–64. [1970]: 272–289), except that Stalnaker restricts semantics 36. For example, Frank Zappa, TheBestBandYou to the study of declarative sentences. Never Heard in Your Life (Barking Pumpkin D2–74233, 19. The conventions can be established and long- 1991), recorded in concert, 1988. standing, but they may be ad hoc on a particular occasion. 37. Clarke, Ways of Listening, p. 57. 20. William K. Wimsatt and Monroe C. Beardsley, 38. Lyrics as printed in Bob Dylan, “It’s Alright, Ma “The Intentional Fallacy,” Sewanee Review 54 (1946): 468– (I’m Only Bleeding),” Writings and Drawings by Bob Dy- 488, and Roland Barthes, Image–Music–Text, trans. Stephen lan (New York: Knopf, 1973), p. 172; the recorded perfor- Heath (New York: Hill and Wang, 1977). mance is available on Bob Dylan and the Band, Before the 21. For a summary, see Sherri Irvin, “Authors, In- Flood (Asylum Records AB-201, 1974), recorded February tentions and Literary Meaning,” Philosophy Compass 14, 1974. Nixon’s presidency ended on August 9, 1974. 1/2 (2006): 114–128; available online, doi:10.1111/j.1747– 39. Dylan rejects determinate, “specific” interpreta- 9991.2006.00016.x. tions of his songs and regards them as open to ongoing 22. Robert Stecker, Interpretation and Construction: pragmatic reinterpretation; see Kurt Loder, “Bob Dylan,” Art, Speech, and the Law (Malden, MA: Blackwell, 2003), in The Rolling Stone Interviews: The 1980s, ed. Sid Holt p. 42. (New York: St. Martin’s Press, 1989), pp. 93–104, at p. 96. 23. Stecker, Interpretation and Construction, p. 17. 40. Bach, “Context ex Machina,” p. 18. 24. Stecker, Interpretation and Construction, p. 17. 41. Bach, “Context ex Machina,” p. 18: “The speaker’s 25. Stecker, Interpretation and Construction, p. 64. act . . . is what brings extralinguistic information into play.” 26. Stecker, Interpretation and Construction, p. 88. Im- 42. Although I do not find it persuasive, the contrary portant criticisms of Stecker are found in David Davies, “Se- view is defended by Peter Kivy, The Performance of Read- mantic Intentions, Utterance Meaning, and Work Meaning,” ing: An Essay in the Philosophy of Literature (Malden, MA: in Contemporary Readings in the Philosophy of Literature: Blackwell, 2006). An Analytic Approach, eds. David Davies and Carl Mathe- 43. This work falls within the broad category of the son (Peterborough: Broadview, 2008), pp. 167–181. characteristic symphony. See Richard Will, The Charac- 27. William Shakespeare, The Tragedy of King Richard teristic Symphony in the Age of Haydn and Beethoven III, ed. John Jowett (Oxford University Press, 2000), p. 147. (Cambridge University Press, 2002), pp. 83–128. Gracyk Meanings of Songs and Meanings of Song Performances 33

44. Potentially, this is all music, provided we follow 45. Stan Godlovitch, Musical Performance: A Philo- Bach and regard semantics as more than “computational sophical Study (London: Routledge, 1998), p. 86. semantics” and as embracing anything communicated as 46. My thanks go to Jeanette Bicknell for invaluable a projection of syntax when interpreted in light of gen- advice at several stages, to Christopher Bartel and John A. eral conventions. See Bach, “Context ex Machina,” pp. 24– Fisher for pushing me to greater clarification, and to Andrew 25. Kania for getting me to write it down in the first place.