Types of Recording, Types of Performance and the Ontological Identity of Musical Works

Abstract

While all types of musical works have been affected by the advent of recording, it has not always been in the same way or to the same extent. The aim of this paper is to delve into these differences and to clarify the role played by recording in defining the ontological identity of musical works. Starting from the discussion of some ontological approaches, it considers music recording in light of the conceptual tension between its meaning as a document of a performance of a pre-existing musical work and its meaning as a construction of an autonomous musical work. It argues that the diverse roles that recording can play depend not only on a grasp of the ontological identity of works but on an adequate understanding of the (type of) performance to which they refer. An expansion of the “logical space” of the notion of performance is ultimately necessary to account for the ontological identity of musical works, whether documented or constructed through recording.

1. Introduction

Long neglected by academic research, the question of recording has come to play a crucial role in the field of musical ontology. This is not surprising, since—as scholars have recognized since the 1990s [1,2,3,4]—it is due to recording that we owe a new mode of being for musical works. But while every kind of musical work or event has been affected by the advent of recording, it has not always been in the same way or to the same extent. This article is intended to probe these differences and to clarify the role played by recording in defining the ontological identity of musical works. Following Stephen Davies’ approach, I will employ this term to designate any piece of music that can be played and broadcast—or, more exactly, any “sonic entity the identity of which persists over its repetitions” [5] (p. 1). This basic definition should be taken in a descriptive and not evaluative sense. Although very broad, it leads us to exclude musical improvisations—for the obvious reason that these, by definition, do not consist in sonic entities intended to endure over repetitions. I am aware that sometimes the boundaries between these two ways of organizing musical practices are not clear (just think of all the cases, often discussed in the literature1, in which classical works contain parts or sections that need to be improvised). In the short space of this article and for the sake of clarity, however, I will limit my examination to the cases in which recordings refer to pieces conceived to endure. To do so, I will discuss three ontological approaches to the relationship between works, interpretations and recordings summarized in a concise article by Roger Pouivet [6] and consider some reflections developed by other authors [7,8,9,10]. The goal is to show that the variety of roles played by recordings depend on our understanding not only of the ontological status of works but also of the type of performance to which they refer.

2. Three Ontological Models

One of the first functions assigned to a musical recording was to document a performance and, in so doing, to give us access to a pre-existing musical work. The idea that a recording might present itself as a faithful document—and even as a better document than a notational transcription—was particularly important to musicians and musicologists who, in several countries of Europe, in the early 20th century, became interested in folk music traditions. In collecting 450 folk songs in England, for example, Percy Grainger emphasized how the use of the phonograph had given “an enduring picture of the live art and traditions of peasant and sailor singing and fiddling” [11] (p. 150). While reproducing a musical performance, recording, in this sense, remains an “external” act—external to the performance and, by extension, also to the musical work (i.e., the song or the piece) expressed through these means. This understanding of the act of recording is encapsulated in an observation by Gérard Genette, who argued that “a record, a tape, a cassette, a film […] is never more than a more or less faithful document (by imprint) of a performance, and not a state of that performance […]” [12] (p. 77). In that sense, one of the greatest merits of recording is that it “erases itself”, meaning that it is as “transparent” as possible with respect to the performance—which, in turn, constitutes a way of presenting a musical work.

Apart from the fact that the performance (which is the subject of the recordings) is not always the presentation of a pre-existing work but can draw attention to itself (as is the case with free improvisation), a basic objection can be made to this view. If we look at some musical genres, we can easily realize that documentation is not necessarily the primary function of recording. This applies to jazz [13] (p. 42) and certainly also to popular music. As Evan Eisenberg wrote in a celebrated book, some studio recordings, strictly speaking, “record nothing”; rather, they present themselves as the “composite photograph of a minotaur” [14] (p. 89). The formula aptly reflects the implications of an artifact that, while assembling bricks of reality, has no need to conceal its artificiality. But one can even go further, to the point of recognizing that in the studio productions of rock, pop and many other genres or sub-genres, and to a greater extent in much electronic and electroacoustic music, works simply could not exist without the act of phonographic recording. In these cases, it is fair to acknowledge that recording has the primary function of constituting (or crucially helping to constitute) these works [1,4,6].

How does the opposition between documentary and constructive recordings relate to the definition of the ontological status of musical works? In a short article that details the theses introduced in his book Philosophie du rock [15], Pouivet provides some useful pointers to answer this question. The article begins by distinguishing three ontological models: (1) the emanation model; (2) the type/token model; (3) the constitution model. The first supports the full autonomy (and transcendence) of the work, which pre-exists its multiple executions—a principle like that enforced by the neo-Platonic idea of an “emanation” from the unity of being to the multiplicity of entities. Such a conception can be found in some interpretations of the type/token pair, where the first term has a purely ideal value. In this view, performances and recordings appear as somehow “degraded” versions of the original ideal entity. This is denied by the “immanentist” conception of this model: if the work is a type, it “has no metaphysical precedence over the token” [8] (p. 165), but it is realized in its tokens. These can be of two kinds: performances (interpretations) and recordings. The latter, in turn, can be classified into two categories: “simple” captures of performances and “sound reconstructions made possible by recording and mixing techniques” [8] (p. 166). The third model, finally, is based on the idea of a non-numerical identity implied by the relation of “constitution”: just as a piece of paper (x) can constitute a EUR 20 banknote (y), a recording can constitute a musical work. The recording, which in this case is of the second type (R2), constitutes the work without identifying numerically with it [8] (p. 168). The correspondence of these three ontological options with the two types of recording mentioned above is defined as follows: the first and second correspond to R1, the third to R2.

Before reflecting on this correspondence, it is worth considering how Pouivet came to identify these three models. From the very first analysis, he makes it clear (and ironically so) how far removed he is from a historical (or historical–philosophical) approach: if the idea of emanation is the hallmark of Neo-Platonism, it would be quite arduous to find a Neo-Platonic philosopher who has questioned what a musical work is. Admittedly, some ontological perspectives—such as the work of Julian Dodd [16], who Pouivet explicitly cites, but Peter Kivy [17] might have been equally appropriate—have referred to Platonism and have indeed interpreted the work as an abstract “type”—namely, an ideal entity that does not exist in time and in space and whose reality is similar to that of a number or of a geometric figure [17] (pp. 210–223). It is not certain, however, that one can ascribe to them such a radical position regarding the (inevitably) degraded value of any interpretation—unless one interprets precisely in this sense the pluralist thesis regarding the question of authenticity of musical interpretation formulated by Kivy [18]. As for the “immanentist (non-Platonic) interpretation of the distinction between type and occurrence”, according to which the identification of type can only take place in its occurrences, and the third model, based on the distinction between “the ‘is’ of identity and the ‘is’ of constitution” [8] (p. 167), no reference is given. Regarding the second option, one might recognize in some respects Davies’ Aristotelian and immanentist view [19] (pp. 30–46); about the third, the way in which Danto’s philosophy of art enforced the problem of indiscernibility, possibly in the version revived by Levinson’s moderate Platonism [20] (pp. 63–88). But readers of Pouivet [15] will be able to more easily recognize the author’s own position, based, as far as the type/occurrence model is concerned, on an original interpretation of Goodman’s arts theory. In this perspective, a musical work—as well as, more generally, a work of art—far from being an ideal entity, is an “artifactual substance whose aesthetic functioning determines its specific nature” [21] (p. 66).

Let us come to the approach followed to distinguish these three ontological options. This is the metaphysical approach introduced in the Philosophie du rock [15] (pp. 22–24): it involves questioning the mode of existence of ordinary things, i.e., the things we consider to be real in our everyday lives—and this is precisely the case with musical works and recordings—to identify our most common ways of conceiving them. In a nutshell, the three ontological options, therefore, constitute a formalization of the three directions we tend to take when we try to explicate our (usually implicit) beliefs about the nature of these objects. So, when we try to explain the distance between a performance or a recording and the assumed properties of the work, we might say that we end up subscribing to an emanatist model. Some composers of the past have suggested this in their own way: one can think of Brahms, for example, who is said to have preferred enjoying the Don Giovanni by comfortably sitting at home reading the score rather than at the opera; or, of Stravinsky, who asked performers to give up any desire for interpretation and let the work appear in all its objectivity. Even if it seems difficult today to share such a radical point of view—the antithesis of which could be constituted by Christopher Small’s perspective [22] (p. 2), inviting us to rethink the meaning of music starting from the priority of a performative act2—a presupposition of this kind could continue to act as a kind of regulating ideal in the practice of some performing musicians, conscious of the impossibility of realizing all that the work may contain or manifest. The corollary of this way of thinking is to consider that the recording is no more than a poor simulacrum, an image that is doubly distant from the work. While it may be true that the performance or recording may possess properties that the work lacks, these nonetheless remain alien to it: they “are not constitutive of the work” [8] (p. 164).

The immanentist version of the “type/occurrence” model and the constitution model offer alternative solutions: in the former case, for the multiple copies of musical works conceived from the writing of a score; in the latter, for the copies of works whose constitution takes place in a recording studio. In both cases, the recording as occurrence is “rehabilitated”, even if its ontological identification differs: while in the first case, the work’s identity is assured by a pre-existing trace that is recognized as having normative value; in the second, it is assured by an act of constitution based on the same recording. We are thus faced with the enhancement of the two main functions of recording: that of documenting (in the first and second options) or constructing (in the third) a musical work.

3. Between Document and Construction

If we take a closer look, however, we realize that the framework involved in this correspondence is not as simple as one might think. In fact, in the immanentist version of the type/token model, we are confronted with two types of recordings: the “recordings of performances” and the “sound reconstructions” made in a studio (the example of Glenn Gould is mentioned [8] (p. 166)). What exactly is the difference here? It seems technical: we have two strategies for obtaining an aesthetically satisfactory result. But if the sound sequence we hear in the recording must above all be a good token of the work, we can easily conclude that an adequate phonographic construction is an “aesthetically preferable” token, superior to those that might be obtained in concert [8] (p. 167). A sort of “direct link” seems to be established between the recording and the work: the recording restores a global sound sequence which, thanks to editing techniques, constitutes a good token of a work. Gould’s own reflections point towards this possibility, particularly when they advocate recreating the work and transforming the act of interpretation into an act of composition [23]. From this, we can infer that it is appropriate to speak of “recordings of compositions” [3] (pp. 30–36).

It seems to me, however, that this conclusion could only apply to a very limited number of cases—or, to put it another way, just to exceptional cases, at least if we consider the recording of the repertoires of notated music concerned by the type/token model. Take the example (examined by Pierre-Emmanuel Lephay [24] (p. 120)) of the three Sibelius piano pieces op. 41 Kyllikki recorded by Gould in 1976–77: the effects of the proximity and distance of the sound source that we perceive are somewhat justified if we look at the score. These recordings could be described as ‘closer’ to the intentions notated by the composer in the score than a concert performance. But is this really the case? To put it frankly, these results sound artificial to our ears. Under normal conditions, a pianist would not be able to achieve effects like these, which are clearly the result of studio manipulation. Does this mean that recording should be limited to documenting a single performance of the work?

Some people think that this is the way to go. We agree with Pouivet [8], this choice brings out a “soft” (but still recognizable) version of the emanation model: if a recording must be faithful to a performance, it is because the latter remains closer to the source that determined the emanation process. However, the ideal of a perfect transparency in a recording, precluding any kind of technical manipulation, has problematic implications. Any audio engineer knows this: a recording requires the making of a series of choices (directional or omnidirectional microphone, cardioid or condenser, distant or closer to the source, with one type or another of reverberation, etc.). These choices determine very different sound images: the same performance will be rendered in a more or less convincing way by two recordings that could both be presented as “faithful”. The compromises that must be made to give a studio recording an aesthetically effective spatiality (see [14], pp. 90–92) show, in turn, that the most “natural” sound image in this kind of recording ends up coinciding with the result of a fine (and often complex) technical installation. Moreover, it is not certain (as argued by Joshua Glasgow [25]) that the problem of the transparency of musical recording can be solved in the purely technical terms of sonic accuracy. Gracyk [1] and Kania [26] have rightly pointed out that transparency, in this case, necessarily involves an ontological issue: it is uncertain whether a recording that sounds exactly like the sound source allows one to hear the music it is supposed to document.

At the same time, if we think about our ordinary experience of listening to recordings of music produced through an actual performance, a total skepticism about the documentary function of such a recording would certainly be inappropriate. It can be said that the recording—through all the retouching, mixing and editing operations—obeys a double constraint: not only must they correspond to the work (the type) but to a possible performance of it. There is nothing more disturbing in this kind of recording than to perceive sound discrepancies that would be impossible in a real performance. At best, you get the impression of an artificial result; at worst, of a poorly done job. One can deduce the following maxim: insofar as such a recording is (in fact) constructive, it must be able to give us the impression of an actual performance. In other words, it must make its documentary functions apparent—even in a fictitious or illusory way.

If it succeeds in doing so, this result is a “work of performance” [27] (p. 182). The recording is indeed a R1, the aim of which is to fix and “perpetuate” a performance made in the studio. It restores a specific sound image which, although obtained through recording technologies, unavoidably appears as the image of a real performance. Stephen Davies has coined an apt phrase for describing the purpose of a recording of this kind: it can be considered a “simulated performance” [19] (p. 38)—that is to say, a performance which must look like a real one.

At this point one might ask the following: where does such a constraint come from? Here is a short answer: whatever we do, we are recording a performance of a work that was designed to be (actually) played. In other words, the type to which these tokens refer was conceived to work in an actual performance, made by humans and based on the use of specific instrumental means (as stressed by Levinson [20] (pp. 76–77)). Admittedly, this constraint has not always been so explicit (it does not apply to much of the Baroque repertoire, or even to contemporary music); it is, however, relevant for most of the music in the Classical–Romantic repertoire that we have mentioned. This means that the recording relates only to an act of real performance (or to the simulation of an act of real performance)—and not, for example, to a token produced electronically. A more general conclusion that can be drawn from this case is that, as Davies [6] has seen and as Gracyk and Kania [9] have pointed out, “the functional relationship to performance practice, rather than the kind of musical work that is presented, determines which kind of recording presents the music” (p. 85). While I fundamentally agree with this conclusion, I believe that considering the relation to the performance practice should not take priority over considering the (ontological) type of work in question: they should rather go hand in hand, when we seek to clarify the type of recording that we are listening to or examining. I will try to show this through the examination of some cases.

4. Types of Work and Types of Performance

A clarification is needed regarding the object of a musical recording of type R1. Such an object, as we have just seen, coincides with a performance which, in the two cases mentioned, can be considered as actual or simulated. However, it can happen that these two types of performance must in turn respond to different functions: they can appear as the execution of a pre-existing composition or as a performance that is a creation in itself, as in jazz improvisation. The former can be called the performance of a work; the latter a performance work. How does R1 relate to these two situations? Here is an answer: in the first case, R1 relates to a performance (real or simulated), which must correspond to a prior trace in order to be ontologically correct; in the second, it relates only to a performance (real or simulated), which, in fact, coincides with the work.

Let us turn now to R2. We have noted that this type of recording calls on the ontological model of constitution, which means (among other things) that it no longer has to correspond to a previously conceived type. The freedom from the second constraint we mentioned (in relation to the “type/token” model) is evident: it is no longer necessary to correspond not only to an actual execution but even to a possible execution. In Pouivet’s terms, this is a recording in which we are not given the opportunity to observe an intentional relation—or more precisely, the intentional relation required by R1 [8] (p. 171). We can conclude from this that, when listening to a rock work (the example of a Lou Reed album is mentioned), “we listen to what we listen to, and nothing else, nothing that is given to listening and towards which we must intentionally strive” [8] (p. 171).

While perfectly consistent with the premises of the discourse, this conclusion seems slightly perplexing to me. It is true that our listening to this recording is not usually “oriented” towards a type that (ontologically) precedes it—unless, as Pouivet rightly explains, we make a “categorical error” and think we are dealing with an R1 when in fact we are dealing with an R2. At the same time, we find it difficult to accept that, in the example given, the recording can renounce any intentional relation—and, more precisely, the intentional relation to a performative act: can an AC/DC record be properly appreciated without referring to Angus Young’s instrumental prowess?

The question of the nature of such a performance remains legitimate. It is clearly not the performance of a pre-existing work, as in the case of a Beethoven’s work—in other words, it is not an actual or at least simulated performance. A solo by David Gilmour or Mark Knopfler could be considered as a work as such. These solos can be improvised and recorded as such; but they can also be (largely) constructed and/or processed in the studio: effects, electronic modifications, montages and collages can be used at any time. Also, the performance can be repeated ad libitum, until the result is achieved. According to Pouivet [15] (p. 65), these possibilities take us beyond the concept of musical performance—for the latter implies a certain level of risk and contingency, which is absent in cases where it can be repeated and perfected endlessly until a flawless result is achieved. In my view, however, they do not entirely defeat the idea of performance: rather, they expand its “logical space”, so to speak.

Davies [19] paved the way for this option by introducing the notions of “studio performance” and “virtual performance”. The first designates a kind of performance ontologically distinct from live performance; the second, the way the performance is presented in a studio recording. They look very similar and almost synonymous, but, in fact, they involve a difference in perspective that may be worth observing. As remarked by Kania [28], the first notion is ambiguously employed to designate both a process and a product. This ambiguity produces some confusion insofar as it leads us to think that the production of a record (of type R2) depends on a special kind of performance (as stressed also by Pouivet [15] (p. 65)). However, even if a studio performance can be technically quite different from an on-stage performance, it is not clear in what sense this difference has ontological relevance. I fundamentally agree with these criticisms; the most relevant difference, from my point of view, is in the product—or, more precisely, in the way it is perceived and conceived by those who listen to it. In this regard, the other category introduced by Davies seems to me more interesting: what we listen to, when we listen to a R2, is a kind of “virtual performance” [19] (p. 37). Such a performance detected in a recording is supposed to extend—and even, sometimes, “to denature”—an actual performance, without losing entirely, however, its meaning as a (kind of) performance.

To show this, we can appeal to a simple example that goes back to what, in Pouivet’s view [15] (pp. 11–12), is the birth act of rock: the release (on 26 March 1951) of the record featuring Mary Ford and Les Paul’s performance of the standard How High the Moon. As is well known, these musicians used multitrack recording to produce this song. What we hear, listening to their record, is a polyphonic (and homorhythmic) vocal performance accompanied by multiple guitars. The perfect timbral fusion of the voices (as well as the fusion of the guitars) depends on the fact that it is the same voice, and the same guitar, recorded multiple times. We know that such a performance cannot take place in real time (at least without some electronic manipulations3); nevertheless, we listen to it as a sort of (recorded) performance—and not, say, as a sound object appreciated by virtue of its morphology, without attending to the fact that it is intentionally produced by someone on an instrument or with her voice. That is what it seems to me appropriate to call it a “virtual performance”, namely, an (electronically) enhanced performance that draws its origin from an actual execution.

The category of “virtual performance” is, I believe, useful insofar as it emphasizes the idea of a performative result whose artificiality we explicitly accept. A virtual performance must still be able to be perceived as a performance—and not, say, as a “mere” sound construction. This is why listening to a Lou Reed record in the same way as we would listen to a record of acousmatic music (or, possibly, of entirely “disembodied” electro) would be committing a categorical error, just as listening to it as a record of music by Beethoven would be.

Virtual performances can involve a minor, but also a (far) greater, distance from actual performance than the above example suggests. That an echo can make a sung word repeat once or twice is very possible in the real world; that the same word keeps repeating itself in a sequence of several minutes, progressively degrading and gradually transforming into a kind of groaning, is unheard of in the real world. And yet, we take it for granted, when we listen to “Dogs”, the second track on Pink Floyd’s Animals studio album (1977). Again, although this sounds artificial, we willingly accept it since, musically speaking, it works. This sound is not unlike the sixty-feet jumps actors are shown making in science-fiction or action movies (for example, the cyberpunk Matrix, 1999) in the sense that it extends a real action or takes it beyond what is possible. It is likely that, in light of the recent spread of artificial intelligence (not only in recording studios but in social platforms and in the very common DAWs), virtual performances of this kind will continue to have wide appeal.

Admittedly, this outcome deprives us of the feeling that a genuine risk was taken to bring us into a fictitious world; it does not, however, erase the idea of an action performed by a human subject. Likewise, a virtual performance may appear to denature what we generally think of as a musical performance because it goes beyond what we may encounter in such a performance, but it is not entirely disconnected from the performative act from which it originates.

Surely it is no accident that the above examples are taken from rock history. As Gracyk, Kania and Pouivet observed, it is in this “artistic tradition”, as opposed to classical and jazz, that a musical work is identified with “a ‘thick’ sound structure encoded on a recording and properly instanced through playback of a copy of the recording” [28] (p. 401). While there is no question that rock has massively contributed to spreading this way of producing musical works, I would note that it has not been alone in this. A lot of music produced in other popular genres— Kania has rightly singled out pop, hip hop and electronica [29] (p. 228)—but also a lot of music for film and even, arguably, works belonging to the classical tradition—or, if you will, to the avant-gardes of the 20th century—might also apply here. For example, take Thema (Omaggio a Joyce) (1958) by Luciano Berio or Gesang der Jünglinge (1954–55) by Karlheinz Stockhausen. These works are both based on a vocal recording and their ontological density is certainly no less significant than a Pink Floyd album. The same could be said of one of Steve Reich’s earliest minimalist works, It’s Gonna Rain (1965), a work built on the recording of a phrase uttered by the African American preacher Brother Walter, sequenced and looped on two tape recorders running at slightly different speeds. One might object that these works fall under the purview of acousmatic listening, but I do not think this is entirely the case since, as we listen to them, we are made aware of their roots in characteristic vocal performances, even though these are subjected to extensive acts of phonographic construction.

Although I am sympathetic to the purpose of considering recordings based on musical genre [29], it seems to me that this poses significant challenges. One of them is the increasing permeability between genres in the contemporary musical world. Take a composer like Giovanni Allevi: he could be considered as a “classical” composer who writes scores intended to be performed. What we can hear in his recordings seems closer to the idea of an actual or simulated performance than a virtual performance (except perhaps for the reverberation and spatiality of the piano sound, which appears carefully constructed in the studio, everything seems achievable on the stage). However, his participation in the production of his albums as a performer is not secondary. One could argue that this case is comparable to that of many composers of the past (such as Stravinsky or Boulez) who recorded their own works. But I do not believe that these examples should be lumped together. Allevi’s audience clearly looks out for his records (played by him and not by other pianists), not his scores. They probably appreciate his works first and foremost as phonographic products and secondly as written works that can be performed by others. It is also interesting to observe that performances of his works are often presented on YouTube as “covers” rather than interpretations. Whether it is an exception or a growing trend, this reveals a notable hybridization of genre-related practices. Rather than thinking of musical works as relating to the rock or the classical tradition, it seems to me thereby more appropriate to distinguish them through the (maybe more neutral) categories of “notated works” and “phonographic works” [10]—where the former mainly corresponds to works conceived to be performed over and over again, the latter to works that are fixed and transmitted primarily through recording4.

Note: In the cases of the works of Berio, Stockhausen and Reich mentioned above, as well as in those we detected in the rock productions, we are clearly faced with a R2. If we want to stay closer to the mode of presentation of these works, however, it seems appropriate to recognize, in this R2, a tension in the direction of some documentary functions—although without forcing us to accept the normativity that explicitly accompanies R1, and which is manifest in the case of Allevi. So, instead of saying that in these cases the recorded sequence does not intentionally tend toward anything other than itself [8] (p. 171), I prefer to say that it refers to a performance that has been augmented or enhanced by studio techniques.

But what then authorizes us to depart from the rule which, as we have shown, is at work in the case of R1? (a rule which requires, it should be emphasized, that it corresponds not only to a composition or work prior to the act of recording but also to a real—or at least simulated—musical performance). The answer could sound as follows: the awareness that we are faced with a phonographic work that is a musical work constituted by a recording rather than a notated work. We “accept” and regard as “standard” this expression of musical works—an expression which, as Pouivet points out [15] (p. 15) [30] (pp. 38–40), only really became established with the creation of works explicitly aimed at the mass market. But accepting that these works can exist in this way also means accepting that a musical performance can be something other than what it is in a concert hall or on a stage. Recording has not only helped to bring into the world a new category of works but also a new category of performances.

5. Conclusions

I hope that the reflection I have “grafted” on Pouivet’s ontological analysis [8] has helped to delve deeper into the links between the two main functions of recording and the musical objects they deal with. While my observations support those of other scholars, some differences are worth recalling. Like Davies, I think it is appropriate to distinguish between recordings referring to an actual (or possibly simulated) performance and recordings referring to a virtual performance. However, I do not believe that this difference depends on the fact that in the latter case we are dealing with an ontologically different performative act (the “studio performance”). It is rather a matter of an outcome that is categorized differently by those who receive it. According to Gracyk and Kania [28], the artwork resulting from this mode of production can be considered as a “constructed track” with ontologically autonomous value. However, I would like to point out that such products, although derived from the traditions of rock and popular music, are not found only in those genres—this could refer to all musical works whose identity is supposed to be (mainly) phonographic.

The conclusion to be drawn from this is that a proper grasp of the role of the musical recording requires a strategy of ontological identification of the work in question; this, however, concerns not only, as pointed out by Pouivet, the type of work that is the object of a recording, but the type of performance, both when it appears to convey the occurrence of a previously established type, and when it functions as a stakeholder in the creative process. The complexity of this situation gives us a better sense of the heuristic value of the opposition between the constructive and documentary functions of musical recording. One could represent the latter as a “field of forces” stretched on a spectrum between the two poles constituted by these functions: the work is positioned more or less close to one or the other, depending on its ontological identity and on the type of performance to which it refers.