The free access to this article was made possible by support from readers like you. Please consider donating any amount to help defray the cost of our operation.
Reflections on Cover Songs: Replies to Polite
P.D. Magnus
Abstract
Brandon Polite poses a number of interesting criticisms of my account of cover versions. Some I am happy to acknowledge and take on board, but with others I feel he goes too far. I take this chance to clarify and extend my account.
Key Words
cover versions; “Hallelujah”; music ontology; philosophy of music; songs; Todd Rundgren; versioning
1. Introduction
I want to thank Polite for engaging with the book in the way he has done. I interpret his criticisms as friendly additions and amendments, rather than as crushing take-downs. Yet we do not entirely agree on the details.
2. On covering performances
Polite raises questions about my claim that a cover has as its target—the thing that it covers—an earlier recording. Rather, he suggests, a musician might cover an earlier, unrecorded performance.
Partly this just is a report of usage: Sometimes people use the word ‘cover’ when there is no earlier recorded version.[1]
As a definition rather than a necessary condition, Polite’s proposal faces several other kinds of counterexamples that I discuss in the book.[2] So, his hypothetical could be taken to illustrate the futility of trying to define ‘cover.’ For any would-be definition, there are putative counterexamples. If the definition is revised to accommodate those, then the revised definition faces further counterexamples. And so on, in a regress of what Imre Lakatos calls “monster barring.”[3] Lakatos argues that monster barring can be a productive strategy in formal disciplines. The concept ‘cover,’ however, does not survive that kind of rigor.
Even my necessary condition is not intended to be a necessary condition for ‘cover’ as such, but ‘cover’ as it interests me in the book. Other uses of the term, I write, “do not describe the kind of covers that this book is about.”[4]
Still, one might argue that my book ought to address covers of performances. Intuitions here are vexed, but let me comment on one of the cases that Polite mentions: covers of Leonard Cohen’s “Hallelujah.”[5]
The song became popular in the 1990s due to Jeff Buckley’s cover, but Buckley had never heard Cohen’s version. Rather, he was covering the track that John Cale recorded for a Leonard Cohen tribute album. (So, it is an example of Polite’s Scenario 1.) Interestingly, Cale had originally heard the song in concert. He contacted Cohen about it, and Cohen sent him the lyrics. The lyrics Cale used included alternate lines that Cohen had written after recording his album version. Moreover, Cale had perhaps never heard Cohen’s album version, since the record label had such a dim view of it that they had not released it in the U.S. (This might be construed as Polite’s Scenario 2 or 3.) Despite all of that, I am happy to call Cale’s record a cover.
That just leaves Scenario 4. I draw a line here, because I think there are some important differences between versioning when there’s an earlier recording (that is, covering in the sense that interests me) and versioning when there are only earlier, unrecorded performances by other people.
First, playing songs the way you have heard other people play them is a way music has always worked. Playing them the way they sound on the record is something novel that arose in the mid-1900s. The fact that the word ‘cover’ was coined at that time is no coincidence.[6]
Second, you can go back and listen to the original version instead of the cover, so the cover has to offer something extra—either in terms of availability or artistic value. A performance now does not compete with a past performance in the same way.
Third, a recording is repeatable in its thick particularity. So, we can come to expect the song to sound just like that. This allows the cover version to be compared to a permanent record, most vividly when an original and a cover are played together—crossfading back and forth from one to the other. That is simply not possible when there is no recording of the original.
So, I still think there is something especially interesting in the cases that satisfy my necessary condition—that is, where the cover covers an earlier recording.
3. On referential mimics
Polite points out that I am mistaken in the book when I say that referential covers are all rendition covers. I think he is absolutely right about this, and I take it as a friendly amendment.
One might wonder whether it is possible to have a transformative mimic cover—that is, a cover that is meant to sound exactly like the original but that is not an instance of the same song. I cannot think of any actual example. But consider a musical counterpart to Pierre Menard, author of the Quixote in Borges’ story.[7] Suppose the musical Menard tried to live a life that led him, by separate inspiration, to compose a song that was note-for-note and word-for-word identical to Cohen’s “Hallelujah.” If he were to succeed, then on most philosophical accounts he would have a distinct song. Perhaps, though, one might still call it a mimic cover.
With all that in mind, I now think it is best to treat the mimic/rendition distinction as primary.
Independently of its status as a mimic or rendition cover, we can ask: Does the cover refer to the original or not? Is the cover an instance of the same song or not?
Polite suggests that mimic covers can offer aesthetic rewards beyond merely providing an imperfect lens through which to hear the original. In some sense, that is correct.
Nevertheless, I think that appreciating a mimic cover is always in relation to the original. At one point in the book, I call this the “etiological mode of evaluation, appreciation informed by provenance.”[8] I only use this label once, because I do not intend it to be a refined technical notion. In general, I rely instead on the homier description of appreciation in relation to the original.
There are several ways this can go in the details.
First (and most obviously) an unsuccessful mimic cover fails precisely insofar as it fails to sound like the original. There is no artistic significance to respects in which a mimic cover deviates from the original. I have said before that any deviation is a failure of craft, and usually this is right. I will return to exceptions in a moment.
Second, a successful mimic cover allows one to hear the interpretation that was instantiated in the original. Even if the listener does not have the original in mind, the listener hears the same interpretation in the mimic that one would hear if one listened to the original. Judgments about the mimic thus reflect on the original. And a knowing listener might, in the absence of access to an actual recording, listen to a successful mimic as a way to appreciate the original.
Third, a mimic cover can be made so as to refer to the original. This reference may be part of the artistic content of the cover. I do not discuss this in the book, but I am happy to accept Polite’s suggestion that Todd Rundgren’s covers from Faithful can be understood in this way. A curious thing about this dimension of appreciation is that it almost does not matter how the covers sound. The artistic move is more conceptual than sonic. One can appreciate it in terms of etiology alone.[9]
Yet what about cases where deviation from the original is not a defect of craft, but instead a deliberate difference? Polite uses the example of Taylor Swift’s re-recording of her earlier albums. Taylor’s Versions do not sound precisely like the originals. I find many of them hard to tell apart even when they are paired up, crossfading between the two, but avid fans and critics hear differences. Turning this point around: Given the resources she had available, Swift could have made them more similar if she had wanted to. There are deliberately some differences in sound and production.
What I want to argue is that this is not primarily and maybe not even at all an artistic matter. Swift’s goal with her re-recordings is to commercially replace the originals, because she wants the versions that people play to be ones she owns. So, the goal for the project is to make sonic replicas of the originals, just different enough that discerning listeners can tell that they are hearing the preferred version. In a forthcoming paper, my coauthors and I refer to the differences as sonic watermarks.[10] They are just intended to be subtle marks for verification purposes. So, attending to them in appreciation is really missing the point.
One might disagree with this reading of Taylor’s Versions and argue instead that the deviations are artistic choices, so that one might listen to Taylor’s Versions as something else besides mirrors of the original. That is tantamount to saying that the re-recordings are not mimics, but rather deceptively straight renditions. If we take them to be renditions, then the two-mode account of appreciation applies.[11]
So, I agree with Polite that there are artistic and aesthetic possibilities to mimic covers that go beyond what I discuss in the book, but I disagree that any of these allow for artistic appreciation of the mimic cover simply as its own object (apart from the original). That is an important difference between mimic and rendition covers.
4. On lineages and single instances
Polite extends the example of “My Way” to note the connection to David Bowie’s “Life on Mars.” That is really neat.
Whether “Comme d’habitude” and “My Way” are called different songs or the same song depends on context. I wonder—I genuinely do not know—whether there are any contexts in which “Life on Mars” is counted as the same song because of its connection to the others. If not, then perhaps this illustrates a point that I should make explicit. Even on the pluralism that I advocate, there will be same-song judgments that are too much of a stretch to be tenable in any context. Pluralism is not anarchy.
Polite raises a different point. What about “extemporaneous song-like entities” that he improvised, words and all, but did not record or write down? Note that these are single versions—free-standing performances that, as it happens, have no influence on anything. Surely (he says) these can count as songs!
Note that on my account, they could count as songs. As performances, they are versions. Since he did not make them in isolation from the whole history of music, they are in some lineage of influence and inspiration. A singleton set is still a set, and a selection of one version from a broader lineage is still a selection. So, we could treat them as songs.
Yet each is just a terminal branch of the broader lineage. I suggest that a widely diverging instance of a particular song tends not to be counted as a distinct song if it is a one-off. The temptation to count it as a separate song is stronger if it inspires further versions. Recall Aretha Franklin’s “Respect.” It follows from this that we would tend not to count one-time improvisations as songs.
This ambivalence seems right to me. Improvised ditties lack clear identity conditions. Suppose Polite remembers an earlier improvisation and plays something that is similar in some respects. We ask him if he is playing the same song as before. He might say that he is. He might instead say that he is not, but that he is using parts of what he played before. It is tempting to think that either answer is compatible with the musical and artistic facts.
What I want to say about these improvised one-offs is that, in line with my pluralism about songs, we might call them songs or not. Polite can say that there are dozens of songs that he improvised and then forgot. But, attentive to different concerns, he might instead retreat to calling them proto-songs or song-like entities: performances that could have become instances of full-fledged songs, if only they had spawned a lineage of their own.
P.D. Magnus
pmagnus@fecundity.com
P.D. Magnus (he/him) is a Professor of Philosophy at the University at Albany, State University of New York. He works in the philosophies of science, music, and technology. He has written scores of articles, and also Scientific Enquiry and Natural Kinds: From Planets to Mallards (Palgrave MacMillan, 2012) and A Philosophy of Cover Songs (Open Book Publishers, 2022).
Published February 9, 2025.
Cite this article: P.D. Magnus, “Reflections on Cover Songs: Replies to Polite,” Contemporary Aesthetics, Volume 23 (2025), accessed date.
Endnotes
[1] Henry Pratt pressed a similar point to me after a talk I gave in Spring 2022.
[2] See what I call the problem of demo versions, the problem of songwriters, and the problem of genre; P.D. Magnus, A Philosophy of Cover Songs (Open Book Publishers: Cambridge, U.K., 2022), 4-6.
[3] Imre Lakatos, Proofs and Refutations (Cambridge University Press: Cambridge, U.K.,1976).
[4] Magnus, A Philosophy of Cover Songs, 28.
[5] For more details, see Alan Light, The Holy or the Broken (Atria Books: New York, N.Y., 2012) and Ray Padgett, I’m Your Fan: The Songs of Leonard Cohen (Bloomsbury Academic: London, U.K., 2020).
[6] Andrew Kania even takes learning from the record to be definitive of covers. Andrew Kania, “Making Tracks: The Ontology of Rock Music,” Journal of Aesthetics and Art Criticism 64, no. 4 (2006): 401-14.
[7] Jorge Luis Borges, “Pierre Menard, Author of the Quixote,” in Labyrinths, ed. Donald A. Yates and James E. Irby (New Directions: New York, N.Y., 1964), 36-44.
[8] Magnus, A Philosophy of Cover Songs, 65.
[9] A similar oddity arises regarding Mostly Other People Do The Killing’s note-for-note remake of Kind of Blue. See P.D. Magnus, “Kind of Borrowed, Kind of Blue,” Journal of Aesthetics and Art Criticism 74, no. 2 (2016): 179-85, https://doi.org/10.1111/jaac.12272
[10] Cristyn Magnus, P. D. Magnus, Christy Mag Uidhir, and Ron McClamrock, “Tell Me Why This Isn’t a Cover,” in Taylor Swift and the Philosophy of Re-Recording: The Art of Taylor’s Versions, ed. Brandon Polite (Bloomsbury: London, U.K., 2025).
[11] My coauthors and I argue that Taylor’s Versions do not fare well if judged as renditions, but see the paper for the details.