Some Metaphysical Puzzles About Songs
© 2022 P.D. Magnus, CC BY 4.0 https://doi.org/10.11647/OBP.0293.05
The Dead Kennedys’ 1979 track ‘California Über Alles’ begins with Jello Biafra singing the line ‘I am Governor Jerry Brown’, and a line about ‘Carter power’ refers to President Jimmy Carter— that is, the lyrics namedrop then-current leaders. A verse about secret police and death camps begins ‘Now it’s 1984’, alluding to George Orwell’s 1984 but also looking a few years ahead. The track was successful and has been covered numerous times. A few examples include recordings by Six Feet Under in 2000, The Delgados in 2006, and Vio-Lence in 2020. These later bands sing the same words that Biafra sang in 1979, but the lyrics are out of place. They are a relic of a California long passed, but the covering bands sing them because they want to record the same song that the Dead Kennedys recorded.
The Disposable Heroes Of Hiphoprisy’s 1992 cover of ‘California Über Alles’ both changes the genre (from punk to hip-hop) and changes the lyrics. It begins with a sample of Biafra singing the title, and then Michael Franti of the Disposable Heroes sings ‘I’m your governor Pete Wilson.’ There’s no mention of ‘Carter power’, and the verse about secret police begins with ‘Now it’s 1992.’ By changing the lyrics in this way, the Disposable Heroes bring the song up to date— but do these changes mean that they are singing a different song?
One might argue: The Disposable Heroes’ version of ‘California Über Alles’ is a cover of the Dead Kennedys’ track, so it must be an instance of the same song. This supposes that cover versions are always the same song— a problematic assumption, it turns out. It also fails to capture the important sense that verbatim covers by Six Feet Under and others are importantly different than the Disposable Heroes’ version.
One might argue instead: The Disposable Heroes’ version is about different people and times than the Dead Kennedys’ version, so it is a different song. It is clearly based on the Dead Kennedys’ original, but that does not make it the same song any more than sampling three words of Biafra’s original vocal does.
Or one might argue: ‘California Über Alles’ is meant to be a politically-charged attack on the governor. Six Feet Under’s version is about someone who had not been governor for 17 years, so their version lacks the right political valence to be an authentic instance of the song. It is ‘California Über Alles’ in name only. (The original lyrics were perhaps relevant again in 2011–2019, during Jerry Brown’s second stint as governor.) Pursuing that line of reasoning, the Disposable Heroes’ version is the real thing because it called out the then-current governor.
Let’s put this specific example aside for a moment. I will not be able to offer a solution until the next chapter. The rest of this chapter raises other, related concerns about what it takes for two versions to count as instances of the same song.
Interpolated covers
In pop music, the word ‘interpolation’ is often used to contrast with cover. It means a version which uses parts from an earlier song, rather than one which uses so much as to be the same song.
In rap and hip-hop, ‘interpolation’ is used to contrast with sampling. It means rerecording a vocal or instrumental part of an earlier track instead of using a sample. For example, Wu-Tang Clan was able to secure rights to the Beatles’ song ‘While My Guitar Gently Weeps’ but not to the recording of George Harrison’s original guitar part. So they interpolated it by having Dhani Harrison play the guitar part in the studio (Montgomery 2007). In some hip-hop tracks, the interpolation is just an isolated fragment. In others, like Wu-Tang’s ‘The Heart Gently Weeps’, the chorus or refrain from the earlier recording is used as the hook which connects rapped verses with new lyrics. Claire McLeish coins the term interpolated cover to describe hip-hop tracks of this form (2020: ch. 5). Examples which McLeish discusses include three from 1988: The Real Roxanne’s ‘Respect’ (covering Aretha Franklin’s 1967 hit), the Fat Boys’ ‘The Twist (Yo, Twist!)’ (covering Chubby Checker’s 1960 hit), and 2 Live Crew’s ‘Do Wah Diddy’ (covering Manfred Mann’s 1964 hit). These have the same or similar titles to the earlier recordings as ‘an easy way for hip-hop artists to indicate which earlier songs inspired their own’ (2020: 195). McLeish takes interpolated covers to be almost but not quite the same song— more similar than a plainly new song but less similar than an ordinary cover.
Another example of an interpolated cover is Hilary Duff’s 2008 track ‘Reach Out’, which changes lyrics and adds rapped sections to Depeche Mode’s 1989 ‘Personal Jesus.’ In Depeche Mode’s original, the invocation of Jesus and the demand to ‘Reach out and touch faith’ are a metaphor for obsessive love. Riley Haas describes it as having ‘sex appeal with a sinister undercurrent of dominance and submission’ (2020). In Duff’s version, where the chorus is ‘Reach out and touch me’, the themes of sex and submission are all on the surface. Dan Burkett identifies Duff’s version as a cover and writes that fans recognize Depeche Mode and Duff as ‘performing the same rock songs’ (2015). I am not sure whether Burkett is right. Martin Gore (who wrote ‘Personal Jesus’) is credited as one of the writers on ‘Reach Out’, but there are two other credited writers. Online discussions of Duff’s track typically do not use the word cover, although some do.
It is instructive to contrast Duff’s ‘Reach Out’ with two other tracks. First, consider Johnny Cash’s 2002 cover of ‘Personal Jesus’. Although Cash sings all the lyrics from the original, the invocation of Jesus in his version is not a metaphor for anything. Cash comments in an interview with Bob Edwards, ‘To me it’s a very, very fine evangelical song— although I don’t think that’s why it was written’ (Edwards 2002). So, although Cash’s version unproblematically counts as a cover, it effaces one half of what is going on in the original just as much as Duff’s. Second, consider Jamelia’s 2006 track ‘Beware of the Dog.’ Like Duff’s track, it samples the main riff of ‘Personal Jesus’ and credits Gore as one of the writers. The bulk of ‘Beware of the Dog’ is straightforwardly a different song, especially when played live— when the riff is played by guitarists rather than being replayed as a sample. Where Duff’s version is an interpolated cover, Jamelia’s just uses samples or interpolations from ‘Personal Jesus’ in a new song.
Must a cover be the same song as the original?
In the typical case, a cover is a version of the same song as the original track. For example, They Might Be Giants are singing the same song in their version of ‘Istanbul (not Constantinople)’ that the Four Lads sang in the original. However, interpolated covers raise the spectre of extraordinary cases: Are there some covers which are not (versions of) the same song as the recordings that they are covering?
One might argue: No! A cover is a recording of a song that was first recorded by someone else, so the cover and the original are versions of the same song just by definition.
I argued against trying to define ‘cover’ in Chapter 1, so I think this argument is a non-starter. Even if one decided to stipulate a precise definition of ‘cover’, that would not help here. Someone else might just as easily stipulate a different definition.
My strategy has been to take the category of so-called covers as given, which suggests that maybe what we need is data about how audiences think and talk about covers. Christopher Bartel reports on a number of small experiments that provide some data. Here is a summary of his results:
-
62% of respondents said that a recording of a mimic cover of AC/DC’s ‘Back in Black’ is the same song as the recording of an indistinguishable performance by AC/DC.
-
Only 39% said that Whitney Houston’s cover of ‘I Will Always Love You’ is a recording of the same song as Dolly Parton’s original. The prompt told respondents that Houston’s ‘recording contains the same lyrics and the basic melody; but it sounds dramatic, powerful, and heartrending’ (Bartel 2018: 358).
-
Only 19% said that Johnny Cash’s cover of ‘Hurt’ was a recording of the same song as the Nine Inch Nails original. The prompt told respondents that the instrumentation and some of the lyrics were different in Cash’s version, and that ‘the song seems to be referencing the aging music legend’s failing health’ (Bartel 2018: 360).
Bartel was especially interested in the way changes in emotion and meaning affect judgements of whether a cover is the same song, so he constructed the three cases so that the first involves no change, the second involves a change in emotional force, and the third involves a change in meaning. However, it is striking that in every case a sizable percentage of participants thought that the cover would count as a different song than the original. Back in Chapter 2, I noted that the word ‘song’ is used loosely in everyday talk— sometimes it is used to mean the recording. Perhaps that is how Bartel’s subjects are using the word. Bartel acknowledges this possibility but notes that, although it would explain how more than a third of respondents counted a mimic cover as a different song than the original, it would not explain the increasing tendency to consider something a different song when there was a greater change in force or meaning (2018: 363). Every cover is a different recording, regardless of whether it is the most faithful mimic or the most transformative rendition.
There are several issues one might raise with Bartel’s results: It is just one study. The prompts did not use the word ‘cover.’ The participants were students in philosophy and music courses rather than experts on music. And so on. Motivated by concerns like these, one might go on to do variants of the study and obtain further results.
I am not going to do that, however. I argued, back in Chapter 2, that my use of the word ‘song’ is an explication. It is not quite what ordinary people mean by the word, even though it marks a distinction that ordinary people can recognize. More experimental results which show that people do not use the word ‘song’ this way are just what one would expect. And experimental results which showed how philosophers of music use the word ‘song’ would just recapitulate the philosophy of music.
Striking covers
The fact that covers can sound different than earlier versions raises what Andrew Kania calls the striking cover paradox. The idea is that there could be a series of covers, each making small changes to the one before it, so that the final product sounds nothing at all like the original. As Kania puts it, the outcome could be ‘a cover of “Don’t Be Cruel” [that] sounds for all the world like “Pop Goes the Weasel”’ (2006: 410). Here is the puzzle posed in explicit steps:
-
The Striking Covers Paradox
-
Take an original track. Call it A. Someone records a rendition cover of it. Call that first cover B. Someone else records a rendition cover of B. Call it C. And so on for versions D through Z, each a cover of the one before it in the series.
-
[scp-c1a] Because each is a rendition cover of the one before, each will be at least slightly different than the one before it in the series. Small changes might accumulate so that Z sounds nothing at all like A.
-
[scp-c2a] Because each track is a cover of the one before it, they are all instances of the same song.
-
[scp-c2b] It is impossible for two instances of the same song to sound nothing at all alike.
-
Therefore (from [scp-c1a]) it is possible that Z sounds nothing at all like A, and (from [scp-c2a] and [scp-c2b]) it is impossible that Z sounds nothing at all like A.
The conclusion, that something both is and is not possible, is an explicit contradiction. The inferential steps seem secure, so there must be a problem with one of the premises.
Kania’s resolution to the paradox is to deny step [scp-c1a]. If there were a series such that Z sounded nothing like A, then Kania maintains that at least one of the recordings along the way must not really have been a cover of the one before it. Suppose, for example, that differences accumulated so that Q is the first one in the series that is not an instance of the same song as A. Then Kania would say that Q is not really a cover of P. Note that it might be the case that Q sounds recognizably like P, that the musicians intend for Q to be a cover of P, and that music critics and fans call it a cover. Kania would deny, on principle, that it actually is a cover.
I am open to the possibility that philosophical results can outweigh common usage and practice like this, but it is not a happy outcome. It can be avoided by instead denying step [scp-c2a]. If some covers are different songs than the recordings they cover, then every version in the sequence can be a cover even if the final cover sounds nothing like the first original.
This is all well and good in the abstract, but can we find a real example of a cover version which is not an instance of the same song as the track that it covers?
Regarding crossover versions of doo-wop songs, shorn of their original stylings, David Goldblatt writes that ‘for those with the proper sensitivities, the differences were understood to be so great that the two were thought to be the same song only nominally’ (2013: 109). Which is to say: The pop covers were not really the same song. However, someone taking Kania’s position might say that Goldblatt is just speaking figuratively— that is, one might say that doo-wop originals and pop crossovers really are the same song.
Kid Cudi’s ‘50 Ways to Make a Record’, a cover of Paul Simon’s ‘50 Ways to Leave Your Lover’, refigures the lyrics to make a song about the craft of music. Accepting that it is a different song, someone taking Kania’s position might say that Kid Cudi’s track is not really a cover.
What we need is an example that resists both replies. There must be a strong case both that it is a different song and also that it is a cover. In work with collaborators, I have used the example of Aretha Franklin’s 1967 cover of Otis Redding’s ‘Respect’ (Magnus et al. 2013). It is standardly called a cover, and Rolling Stone calls it the ‘definitive cover’ (2021). However, as Jeff Giles writes, when ‘people think of “Respect” — hell, when they just hear the word respect — it’s Aretha’s voice they hear. Through a dizzying blend of flawless technique and raw power, she owns “Respect”’ (Popdose 2011). Ray Padgett expresses a similar thought when he writes that Franklin ‘treated it like a totally new song— which, in many ways, it was’ (2017: 50).
Franklin ‘transforms Redding’s ultimatum to a housebound woman into a demand for consideration, one which might be made between equals’ (Magnus et al. 2013: 365). This means that ‘Redding and Franklin both sing about respect, but they say importantly different things about it’ (366). She does this not just by changing the mood, but by changing many of the melodic, structural, stylistic, and lyrical features. Victoria Malawey details the differences and concludes that ‘Franklin re-authors “Respect” to such an extent that ownership transfers from songwriter Redding to Franklin’ (2014: 205). I would put the point somewhat differently. It is not that Franklin takes ownership of Redding’s song, but instead that— starting from the material of his song— she makes a new one. Among Franklin’s changes is the addition of the memorable lines ‘R-E-S-P-E-C-T / Find out what it means to me’; Redding never spelled it out.
A further thing to note is that the words in Franklin’s version of ‘Respect’ are not merely changed so that the narrator is female rather than male— rather, the narrator in Franklin’s version is understood to be the woman who is being addressed by the narrator in Redding’s. Where Redding gives his woman permission to mess around on him when he is away, Franklin says to her man that she has no interest in messing around. So Franklin’s cover is not merely in reference to Redding’s track, but in dialogue with it.
There is a tradition of answer songs— especially in the 1950s and 1960s— which used the same melody as a popular song but changed the lyrics so as to provide a response. To take just a few examples: Rufus Thomas’s 1953 ‘Bear Cat’ was an answer song to Big Mama Thornton’s ‘Hound Dog’ with lyrics from the man’s point of view. After Elvis had a hit in 1960 with ‘Are You Lonesome Tonight?’, Dodie Stevens and Thelma Carpenter both released answer songs titled ‘Yes, I’m Lonesome Tonight.’ After Dion’s ‘Runaround Sue’ in 1961, Ginger Davis and the Snaps released ‘I’m No Run Around.’ And so on.
Franklin’s ‘Respect’ has the same title as Redding’s, but it can be seen as part of the answer song tradition (Malawey 2014: 196). B. Lee Cooper surveys answer songs and notes that, although they are ‘usually humorous’ and ‘regarded as a novelty’, ‘the functions of specific answer songs vary greatly’ (1988: 57, 58). Both of the songs titled ‘Yes, I’m Lonesome Tonight’ are cheesy love songs in the same vein as the original, rather than being jokes. Franklin’s ‘Respect’ is a serious song about relationships just as much as Redding’s.
The song that Franklin sings is derivative of the song that Redding sings, because Franklin obviously did not make it all up. I am using ‘derivative’ in a genetic sense rather than suggesting anything negative. Everyone agrees that, with enough change, a derivative song can be a different song than its source. Answer songs are typically seen as different enough that they are distinct, albeit derivative, songs in this way. This is suggested just by distinguishing the original song from the answer song. However, one might argue that ‘song’ is used here to mean the track; for example, Cooper defines an answer song as ‘a commercial recording’ which is related to ‘a previously released record’ (1988: 57). More evidence that answer songs are distinct songs is provided by commercial practices. Jukebox programmers, responsible for buying records for jukeboxes, would typically not stock a jukebox with two versions of the same song (Billboard 1971). But one jukebox programmer commented, ‘I used to work in a cafe, and answer records were always played by both young and old, along with the original version of the song’ (Billboard 1973b).
A further twist is provided by Stevie Wonder’s 1967 cover of ‘Respect’ which uses Redding’s lyrics. Billboard’s Pop Spotlight mentions the version as ‘Wonder’s answer song to Aretha Franklin’s “Respect”— also titled the same’ (1967). In the month’s following Franklin’s release, it had largely eclipsed Redding’s original. A fairly straight rendition cover of Redding could be seen as an answer to Franklin.
In the next section, I offer a general argument that applies to answer songs and other referential covers.
Songs about songs
In earlier work, my collaborators and I offered an argument that referential covers are not instances of the same song as the original. Consider a cover which not only means something different than the original but also says something about the original version— for example, the Meatmen’s cover of the Smiths’ ‘How Soon Is Now?’ (which I discussed in Chapter 4). The Meatmen change the lyrics somewhat to provide commentary about Morrissey, the lead singer of the Smiths. Here is the argument: ‘The Meatmen’s cover is not merely a distinct, derivative song. It is one which is partly about the canonical track and the man who sings it. Its semantic content partly refers to The Smiths’ track’ (Magnus et al. 2013: 367). Putting this in abstract terms yields something like this:
-
The Songs-About-Songs Argument
-
In S’s original recording of H, they are not singing a song that is about their version of the song.
-
In M’s cover of H, they are singing a song about S’s version.
-
A song which is not about S’s track and a song which is about it are different songs.
-
S’s original recording and M’s cover are of different songs.
This is a valid argument. For the obvious substitutions of the Smiths for S, the Meatmen for M, and ‘How Soon Is Now?’ for H, the conclusion is that the Meatmen’s cover is an instance of a different (albeit derivative) song. (The argument also works with Redding for S, Franklin for M, and ‘Respect’ for H.)
However, one might object to premise 2 in the argument. A version of a song can mean something that the song itself does not mean. So, even though the Meatmen are singing about the Smith’s track, they might be using the same song to express something different than the original version. The reference to the Smith’s original might be part of the content of the Meatmen’s performance rather than part of the content of the song itself.
As an example of the difference, consider Marilyn Monroe’s famous performance of ‘Happy Birthday’ in 1962, sung to President John F. Kennedy. The televised performance was famously described as ‘making love to the president in direct view of forty million people’ (Goodwin 2012). Monroe ended up meaning something very different by that performance than a roomful of first-graders means when they sing ‘Happy Birthday’ to a classmate. Nevertheless, Monroe and the first-graders are singing the same song.
With the difference between the meaning of a song and the meaning of a version in mind, it could be that the Meatmen take the Smiths’ song and use it to say something about Morrissey. And perhaps— as Kania suggests— Aretha Franklin ‘takes the content of Redding’s [song] and uses it to communicate a radically different message’ (2020: 242).
Kania makes a similar point by considering Cake’s 1996 cover of Gloria Gaynor’s 1978 hit ‘I Will Survive’ (2020: 241–242). Gaynor’s upbeat disco track can serve as an anthem for survivors of all sorts, but Cake’s cover has a disaffected, almost apathetic tone. Kania calls it a ‘brilliant reimagining’ (2020: 249, fn. 35). Paul Pearson writes, ‘Cake tread the fine line between parody and reframing… with machine-gun guitar lines and a stand-alone trumpet that remains hilariously true to the original melody’ (Treble 2018). Despite the different meaning and significance of the two versions, it seems plausible to think of them as instances of the same song.
The examples which seem most clearly to be the same song (Monroe’s ‘Happy Birthday’, Cake’s ‘I Will Survive’) are cases where none of the lyrics are changed. Franklin and the Meatmen mean something different than earlier versions partly by singing different words. This is suggestive but not decisive, though, because cover versions can have different lyrics than earlier versions without thereby being transformative. To recall an example from Chapter 3, consider Willie Nelson singing Paul Simon’s ‘Graceland.’ Where Simon’s lyric is ‘a girl in New York City’, Nelson sings ‘a girl in Austin, Texas’— but nobody suggests that this substitution makes it a different song. What I suggested about that case was that the variations of ‘New York City’ and ‘Austin, Texas’ both fit within the overall meaning of the song. The lyrical changes which Franklin, the Meatmen, and others introduce in their referential covers are more substantive and— it seems to me— do not fit within the overall meaning of their original songs.
Concluding his discussion of whether covers can be instances of different songs than the original, Kania writes that ‘answering this question will depend mostly on how knowledgeable rock artists and audiences treat songs and recordings of them’ (2020: 242). Perhaps, but I think that drawing precise boundaries around songs will also be a matter of explication. As Bartel’s survey results suggest, audiences are not terribly precise about it. We have some latitude and can be guided not just by what people already say but by what purposes we would like these concepts to serve.
The Fitzgerald dilemma
I have argued that the kind of changes which Aretha Franklin makes to ‘Respect’ are enough for her to be singing a different song than Otis Redding sang. This is, partly, because I think that the Songs-About-Songs Argument is sound when applied to covers which change the lyrics so as to refer to earlier versions. One might worry that, in some cases, this will lead to an awkward dilemma.
Consider Ella Fitzgerald’s cover of ‘Mack the Knife’ (which I discussed in Chapter 4). She skips some of the material from earlier versions, and adds a verse about the fact that she is covering the song. For example, she sings in one performance, ‘Bobby Darin and Louis Armstrong. They made a record (ooo, what a record) of this song. And now Ella, Ella and her fellas, we’re making a wreck (what a wreck, such a wreck) of this same old song.’
The Songs-About-Songs Argument applies to Fitzgerald’s cover of ‘Mack the Knife’, and the conclusion is that Fitzgerald is not singing the same song that Bobby Darin and Louis Armstrong were singing. However, Fitzgerald refers to the song that Darin and Armstrong were singing as ‘this song’ and says that she, too, is singing ‘this same old song’— from which it follows that Fitzgerald is singing the same song that Bobby Darin and Louis Armstrong were singing. It cannot be both the same song and not the same song, so either the Songs-About-Songs Argument leads us astray here (because it is the same song) or Fitzgerald is saying something false (because it is not).
I can think of ways to argue that the lyrics she sings are not strictly false, but the details turn on the semantics of indexicals like ‘this.’ Even if Fitzgerald’s claim using the phrase ‘this song’ is not strictly false, it still is not quite right. So I am tempted to accept the second horn of the dilemma: She is singing a different (albeit derivative) song.
You may not feel that this is entirely satisfactory. It may seem like there is another sense in which she is singing the same song, even if there is a sense in which she is singing a different song. The easiest way to understand such ambivalence is to think that questions about which versions are the same song do not have absolute, univocal answers.
Jeanette Bicknell recommends what she calls ‘a pragmatic approach’ in cases like this. When asking whether two performances are of the same song, she writes, ‘we should first ask, “Who wants to know, and why?”’ A singer, a musicologist, an historian, or an intellectual property lawyer might be asking for different reasons, and— Bicknell suggests— ‘When people have different reasons for asking, it is not surprising if they all come up with different answers. Yet each may have good arguments for answering the question in a particular way, depending on their reasons for seeking to differentiate one song from another’ (2015: 8).
This kind of pragmatic pluralism is ultimately correct, I think, but more needs to be said. It does not, by itself, provide any guidance in telling songs apart. Saying that it depends on who I am and why I care does not tell me how to get from my identity and interests to an answer. The next chapter approaches these questions in a more systematic way.