Francis Dhomont's Novars

Francis Dhomont’s Novars Andrew Lewis School of Music University of Wales, Bangor Gwynedd LL57 2DG Wales, UK Tel: +44 (0)1248 382188/382181 E-mail: [email protected] Web: www.andrewlewis.org Abstract This paper examines Francis Dhomont’s tribute to the Musique Concrète tradition, Novars. A graphic score of the work is provided, using a combination of representational and symbolic graphic objects, which not only creates an additional frame of reference for the reader, but is itself an integral part of the analytical process. A classification of sound types is undertaken and a structural plan proposed on the basis of an analysis of the roles of these sound types and their relative importance within the work. This plan is related to a possible structure based on multiple Golden Sections, and is then modified to take account of changes in sound type behaviour, based on a second classification. Plotting the frequency of behaviour changes reveals the work’s main Golden Section to be of further structural significance. Francis Dhomont’s Novars Andrew Lewis Francis Dhomont’s Novars is a celebration of the birth of musique concrète, and a tribute to its ‘inventeur infortuné’, Pierre Schaeffer. It proposes a parallel between the pioneering development of Ars Nova in the fourteenth century, and that of musique concrète in our own. Quotations from Machaut’s Messe de Nostre Dame (1364) and Schaeffer’s Étude aux objets (1959) further this premise by providing the source material for Novars, along with sounds reminiscent of doors slamming and creaking (a reference to Pierre Henry’s Variations pour une porte et un soupir of 1963). Novars was composed in Paris at the Groupe de Recherches Musicales (the studio founded by Schaeffer) with the help of Syter, a computer-controlled system able to perform, among other things, continuous transformations between different sound sources. It is thus not surprising to find that, as will become clear, cross fertilisation of various kinds between different sound pools is a key feature of the work. Graphic score In order to make an analytical discussion of the work feasible a score has been prepared which gives graphical representations to the most important and prominent sonic features of the work (see Appendix). Unlike a conventional score it is not intended for realisation in sound, but is itself a symbolic realisation of sounds that already exist. As a result it should be regarded as an essential part of an analytical process in which “...the perceived object has priority over the conceived object” (Dhomont, 1989). The horizontal axis of the score is a time-scale against which the durations of events are shown proportionately Andrew Lewis (1998) – Francis Dhomont’s Novars (within reprographical limitations). The graphical representation of the music conveys (in general terms only) the pitch area of the material (vertical placement within the system) and its loudness (overall size of the symbol). The timbral identity of the sounds is conveyed by the shape of the symbols and the shading pattern used within them. Where possible, sounds are consistently represented by the same symbols, and different versions of the same sounds are shown by different versions of the same symbol - for example, one shape may appear with different shadings. It is not possible, however, to convey all the nuances of variation within the sound material, which is often in a state of constant evolution. It is also the case that, from time to time, notational devices have been adopted which are more intuitive than systematic. For example, sounds with predominantly low frequencies tend to have darker shading. Classification of sound types As a first step to unravelling the mechanisms of Novars it is necessary to attempt some sort of classification of the material of which it is made. In order to be analytically useful such classification must be based on aural impression rather than arcane infra-musical considerations. The classification proposed below identifies ten ‘sound types’ or timbres. The word ‘timbre’ here refers not simply to ‘tone colour’ as with conventional instruments, but to those characteristics which give a sound object or sound structure its unique identifying ‘stamp’ in the ears and mind of the listener. In each case an example number is given which refers to annotations on the score. These examples represent the first clear instance of that particular sound type. Approximate timings for the start of the examples are also given, and here, as throughout this discussion, these timings relate both to the score and to the CD track time (see discography). a. Filter-swept chords/resonances (Ex. 1, 0’00”) Derived from resonances taken from Objets exposés, the first movement of Schaeffer’s Étude aux objets, these sounds form a running thread throughout the work. The deep quasi-vocal sound which accompanies some of the initiating attacks is also part of the original source. Although at first hearing just a single pitch is present in the resonant sounds (the E above middle C), it soon becomes apparent that this pitch is actually a quasi-fundamental, above which float other ‘partials’ based on a modal arrangement. This complex amalgam of pitch nodes can be represented approximately in conventional notation, where the black note is the ‘fundamental’ (Fig. 1): Andrew Lewis (1998) – Francis Dhomont’s Novars From time to time, this modal harmonic scheme is disturbed by the introduction of resonances based around pitches foreign to the mode. These form the more extended, resonant, evolutionary events which periodically occur in the main line, and of which the opening event is one example. The first three of these events introduce the following pitches (Fig. 2 - Again, the notation shows only a basic outline): b. Machaut “Messe de Nostre Dame” (Ex. 2, 1’43”) This family of sounds is based on single, frozen chords taken from the Messe. Often various treatments - most often a shuffling, ‘brassage’ effect - detach the perceived result from its source, yet a vocal quality is always retained. When the original timbre is more clearly heard, the intervals are usually bare fifths, alluding to the original material through harmonic as well as timbral similarity. Andrew Lewis (1998) – Francis Dhomont’s Novars c. Door creaks and slams (Ex. 3, 3’49”) In its initial form this comprises two elements. The first is the door creaks which have a repetitive structure, with successive iterations becoming either lower, quieter and slower (as originally recorded?), or higher, louder and faster. This simple and audible retrograding is no doubt a direct reference to the technique much in favour during the ‘classical’ era of musique concrète . The second element is the door slams, and it is these which quickly become the more dominant of the two, their ability for stark, violent punctuation providing an ideal foil to the more extended contours of the other two main sound types. (It should be noted that the identification of these sounds as ‘door’ sounds is based on a suggestion from Dhomont (1991). They are not necessarily aurally identifiable as such, nor for the purposes of this analysis is their original source of any real relevance.) As will become clear, the next five sound types have a more auxiliary function, serving to support, develop or accompany the first three: d. Short ceramic attacks with left-right repeated echo (Ex. 4, 4’31”) e. Taut-spring agitations and shockwaves (Ex. 5, 5’11”) f. Glass against glass impacts producing rapid bounces/vibrations (Ex. 6, 5’41”) g. Longer, sustained single pitches and glissandi (Ex. 7, 7’20”) h. Granular noise bands (Ex. 8, 8’00”) The final sound type has a special role: i. Miscellaneous collection of ‘Classic’ sounds (Ex. 9, 17’01”) This is not really a single sound type, but a collection of sound types related not by source but by implied historical provenance. In fact, these sounds come from the first movement of Schaeffer’s Étude aux objets, and in the main whole sequences of events are preserved exactly as they occur in the original. Their timbre (unique identifying stamp) is the evocation of the sound world of classical musique concrète. Although the individual sounds themselves are highly disparate - including such diverse elements as electronic sounding beeps and squawks, single suspended cymbal and gong strokes, and manic improvisation on a Hammond organ - they nevertheless possess an unmistakable common origin through the manner of their juxtaposition and execution. Structural plan based on sound types Fig. 3 shows a schematic plan of sound types as they appear in Novars. The shading used for the different types loosely reflects that used in the graphic score. Andrew Lewis (1998) – Francis Dhomont’s Novars Golden section Theoretical GS scheme Actual structural scheme a expo. b expo. c expo. (dev?) (d, e, f expos.) slow section a, b, c develop. c dev (+b) a recap & dev i ‘partial expo’ c recap & dev with i a/b re- i expo cap (with a & b) Sound types 0 (min.) 1 Time 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Figure 3: Structural Plan Based on Sound Types From this plan it is possible to judge the relative prevalence of different sound types in the piece. A casual glance is enough to establish that types a, b and c are most consistently present, the other types being more sporadically represented (along with other types which appear too briefly to be represented at all). A more accurate assessment of these relationships is shown in Figure 4, which records the total duration of all occurrences of each sound type as a percentage of the overall duration of the work: Andrew Lewis (1998) – Francis Dhomont’s Novars 19 This illustrates clearly the dominance of types a, b and c, and points to a less important role for the other six types. Furthermore, since this graph shows the sound types in order of their first appearance in the piece, it is immediately apparent that the most used types are also the first three to be heard. This suggests some kind of exposition of the main material towards the beginning of the piece, and the Fig. 4 would seem to confirm this. In the first six minutes of the piece the first six sound types are introduced one by one, and each sound type occurs only once. This ‘exposition section’ can be further subdivided by noting that the sound types are presented in two distinct groups, the first consisting of types a and b, the second of types c, d, e and f. Types within each group are allowed to overlap, but types from different groups are not. These two groups consist of material of contrasting character. The first group consists of extended resonances or textures, the second of shorter, attack based sounds. Thus the opening of Novars seems concerned with clearly establishing two opposing pools of sonic resource on which the rest of the piece can draw. In very general terms one might draw a parallel between this approach and the use by earlier composers of subject groups in the exposition of extended Sonata Form, although in this instance the first ‘subject group’ consists of two equally important sound types, whereas the second consists of just one main sound type supported by three lesser ones. This gives the second ‘group’ something of a developmental flavour, especially in view of the tendency of this section to move from a simple mono-layered texture with many silences to a complex multi-layered and more continuous texture. Whereas both the sections so far identified have expository functions, (though the second also embarks on some early development) the next section (beginning at 5’57”) is more overtly developmental and brings together for the first time all three main sound types. This is the first opportunity of exploiting in a direct way the contrasting nature of the two groups of material, with the longer trajectories of the first two types entering into a contrapuntal dialogue with the more percussive type c (supported by another member of the second ‘subject group’, type f). This energetic confrontation is followed by a ‘slow section’, a period of relative calm and repose displaying a tendency towards longer phrase lengths and a greater sense of breadth than has hitherto been the case. It is in this context that two new, previously unheard sound types, g and h, are introduced. Both types lend to the meditative mood of this section their potential for creating longer, sustained spans of less energetic activity. After this, there follows a series of sections all concerned with some kind of development or variation of the material which has so far been exposed, and in fact some kind of evolutionary development is present throughout this latter half of the work. Two elements run counter to this general developmental trend, however. The first is the Andrew Lewis (1998) – Francis Dhomont’s Novars introduction (around 13’20”) of the final sound type, i, (‘classical’ musique concrète material). Actually, sounds taken from Schaeffer’s Étude have been present right from the very start of the piece, although in transformations which make them very much a part of Dhomont’s own language. Here, by contrast, there are the first hints of the original character of the material emerging, although not enough to allow the listener positively to identify them as ‘foreign’. Consequently this is something of a ‘partial exposition’ of type i. The second non-developmental element is the appearance of material which seems to have a deliberately conjured recapitulatory function, at 14’12” (type c), and at 16’30” (types b, and a). These are not recapitulations in any classical sense, since they are generally speaking not a return of complete musical structures. Rather, they are brief reminders of the states in which the three main sound types began, and thus of the ground covered during the course of the piece. The recapitulation of the c material does include verbatim repetition of 9’57” to 10’40”, but efforts are made to disguise this. The sense of return in all these instances is, however, powerful and compelling, and effectively signals the fact that proceedings are being brought to a close. The actual closing section of the piece finally reveals the untreated version of the Schaeffer material as unabashed direct quotation, forming what is in effect a postdevelopmental exposition. This is accompanied by sotto voce reminiscences of the opening of the work, together with a brief, and apparently untreated, cadence from the Machaut Messe. With both Schaeffer and Machaut emerging for the first time from behind the veil of Dhomont’s own stylistic meta-language, and perhaps evoking a sense of wistful nostalgia for devotees of early musique concrète, this section forms an apposite and moving coda, bringing together the triptych of sound worlds - and eras - which the piece is seeking to unite. Based on sound types alone, the structural plan which emerges is that marked ‘Actual structural scheme’ in Fig. 3. One striking characteristic of this plan is its use of sections whose differing lengths appear to be in approximately the same ratio with respect to each other. This suggests a scheme based on Golden Section principles, and a theoretical Golden Section scheme is also shown in Fig. 3 for comparison (‘Theoretical GS scheme’). It would be nice to think that Dhomont had adopted Golden Section principles with reference to the mathematical techniques of the Ars Nova, but clearly Novars is not consciously based on such a scheme, since if it were it might be expected that the durations of the sections would adhere more accurately to the theoretical values. Nevertheless, the fact that there is some approximate adherence suggests that this may be a principle at work on a subconscious level, and if the Golden Section is indeed the aesthetically compelling phenomenon it is held to be, it would be natural to observe its Andrew Lewis (1998) – Francis Dhomont’s Novars presence in works of art quite independently of their creators’ intentions - and particularly so in acousmatic music, where intuitive judgement and a sense of what is ‘right’ are often the final and supreme arbiters in the creative process. The closest correspondence between the theoretical Golden Section scheme and the actual structural plan occurs at the first of the two possible Golden Sections at 7’15”. This is indeed a significant moment, since it marks quite an abrupt change of mood, from lively, combative development to more contemplative rumination, but this Golden Section is also significant for reasons which are not apparent from an analysis of Novars based purely on sound types. In order to explore this further it is necessary to examine the same material using different criteria, and to move from considering sound types to a discussion of sound behaviours. Classification of behaviours Thus far, sonic material in Novars has been classified using a system of sound-typology which uses as its basis the perceived timbral identity of sounds. Although useful, this approach has the disadvantage of relying on the idea that timbre is a static, measurable quantity, an idea which is clearly untrue. Even in music for conventional instruments, timbre is not simply a static measure of spectral content, but is a description of the way that a particular spectrum changes over time, across different registers, in different dynamic ranges, and in different rhythmic and articulatory contexts. Similarly, in Novars, as in much acousmatic music, the musical identity of sounds is not simply a matter of ‘tone colour’, but is derived from a number of other variables, including the temporal structure, evolution and morphology of sounds, as well as their musical context and their interaction with other sounds. These extra variables may grouped together under the generic heading of ‘behaviour’ since they describe how a given timbre exhibits or is subjected to different combinations of these variables in different contexts. Identifying various kinds of behaviour then presents the possibility of considering behaviour as something interacting with but distinct from sound-type - sound-types exhibit behaviours; behaviours are imposed on sound-types. This step is crucial to a clear understanding of Novars, since the piece makes extensive use of the technique of submitting sound types to variation by altering their behavioural characteristics, without destroying the essential qualities which make each sound type viable as a distinctly perceivable musical entity. This process even extends to the transfer of behaviours from one sound type to another, something which has only really become feasible since the advent of composer-friendly computer systems like Syter. It is therefore possible to draw up a classification of behaviours which complements that already undertaken for sound types. In part this overlaps with the sound-type classification already undertaken because some sound-types rely heavily on their behavioural characteristics as their main distinguishing feature. Such sound-types Andrew Lewis (1998) – Francis Dhomont’s Novars cannot therefore have their behaviour modified without also completely altering their identity as sound-types. In other cases, however, there is a clearer distinction between sound-types and behaviours, and it is here wherein lies the most fruitful area for exploring the transfer of behaviours from one sound-type to another. The behaviour classification consists of ten distinct elements: 1. Stable pitch(es) This behaviour is characteristic of sounds based either on a single pitch or an amalgam of pitches, whose temporal stability enables those pitches to be clearly perceived. The opening sounds of the piece are among those exhibiting this behaviour. It is possible to impose this behaviour upon sound-types which do not normally exhibit it, such as noisebased or textural sound-types. (This could be done, for example, by the extreme use of resonant filters tuned to specific pitches). 2. Sustained granulated ‘brassage’ texture This is derived from a characteristic computer music technique in which sounds are sliced into small segments and reassembled in a different order, often with varying degrees of randomness affecting the order of the reassembly, the scope of the selection of sound segments within the original sound, and the pitch, duration, loudness and spatialisation of the individual segments. The sonic result varies depending on the original sound, but the general behavioural characteristics are of a trembling, scintillating mass of tiny sound objects with a common timbral identity. The effects of this behaviour are most obvious when it is applied to the Machaut source material (Ex. 2) since the original state of the material is already known in general terms. 3. Rapid judder Although its appearances are too brief to be shown on any structural diagram of the piece, this behaviour is a significant idée fixe which crops up throughout the piece. It consists of a rapid rhythmic modulation of otherwise continuous sounds, and its first unmistakable appearance is shown in Example 10 (2’49”). 4. Percussive attack An innate behavioural characteristic of some sound-types, this behaviour can also be imposed on continuous or resonant sounds by the careful use of re-enveloping. This must be extreme enough to induce a sense of percussive causality, yet must preserve sufficient of the sound-type’s ‘native’ behaviour to be perceived. Andrew Lewis (1998) – Francis Dhomont’s Novars 5. Glissando This is similar to behaviour 1, in that it is both an innate feature of some sound-types and may also be imposed upon sound-types which do not ordinarily display it. 6. Repeating left-right echo Sounds with this behaviour contain rapid repeated iterations of the same material, first on the left side of the stereo image, then on the right (or vice versa), each iteration being quieter than last (Ex. 4). Although this would be regarded as an echo effect in some areas of studio work, it is important that those with studio experience do not simply regard this or other behaviours as effects, treatments or processes, as though they were mere decoration added to an already formed musical fabric. The way that these ‘treatments’ change the behaviour of these sounds alters their musical identity and function within the piece. They are not, therefore, a matter of ‘production’ - as might sometimes be the case in the commercial music sector - but are as valid a subject for analytical discussion as any aspect of the music. In this case the echo introduces rhythmic and spatial elements which are used as characteristic behavioural patterns for a number of sound types, and forms an important means of cross referencing between them. 7. ‘Pebbledash’ - agglomeration of sharp attacks into coarse but sustained texture This is an example of a behaviour involving the coordinated activity of a number of individual sound events. In this case, attacks group together to form a rough granular texture related to, but distinct from, behaviour 2. This is a process which involves a number of sound types, and is a characteristic device in Novars. 8. Sustained fluid flanged/phased pitch(es) Again, what might be an ‘effect’ in another context is here elevated to the status of a musically significant behavioural characteristic. The introduction of moving bands of phase shift with greater or lesser degrees of resonance turns otherwise stable structures into mobile and flexible entities, and relatively simple spectral profiles into complex ones. 9. Resonant filter Another spectral modification, this behaviour can be applied wholesale to entire phrases without disturbing the essential features by which the original material can be recognised (Ex. 11, 11’26”). Although it introduces a stable pitch element, it falls short of the stability characteristic of behaviour type 1, giving more emphasis to the rhythmic or articulatory features of the original material. Andrew Lewis (1998) – Francis Dhomont’s Novars 10. ‘Classical’ montage - many different behaviours both concurrently and simultaneously This is really identical to sound type i but classifying it as a behaviour is probably more accurate than seeing it as a sound type, since it is its behavioural characteristics - rhythmic construction, use of dynamics, juxtaposition of sound types, and so on - which distinguish it most from the rest of what happens in the piece. Structural plan based on behaviours Having identified the types of behaviours involved, it is now possible to modify the original structural plan to take account of behavioural characteristics (Fig. 5). The shadings used indicate which behaviours are involved, according to the key, but they do not relate very closely to those used in the graphic score: Figure 5: Modified structural plan based on behaviors What is immediately clear from such a plan is that there is a great deal of behavioural transformation in Novars. The nine basic sound types are subject to the ten behaviours in a variety of combinations, making the totality of the material in the piece very diverse indeed (even though only a fraction of the possible combinations is employed). What is also clear is that the trend is one of increasing complexity as the piece progresses. During the ‘exposition’ section (0-6’), each sound type is consistently linked to one behaviour pattern. Later, however, behaviours are exchanged freely between sound types with increasing frequency. The increasing use of changes of behaviour within sound types is illustrated in Fig. 6, in which the number of changes of behaviour within successive two minute periods is plotted for each of the sound types a, b, c, f and g. (There are no changes of behaviour within the other sound types.) There is a further plot showing the total number of changes: Andrew Lewis (1998) – Francis Dhomont’s Novars Figure 6: Frequency of sound type changes This chart clearly shows the increasing frequency of changes of behaviour. It also shows that for each individual sound type there are peaks of activity in which the changes become more rapid, and that these peaks generally occur at different times for different sound types. This means that while one sound type is particularly unstable in terms of its behaviour, the others are more static, thus allowing the ear to make sense of the situation. The most striking aspect of this chart, however, is the role of the Golden Section. From the study of sound types it was apparent only that the Golden Section coincided with a change of ‘tempo’, from energetic activity to more contemplative music. The 13study of behavioural changes has revealed its greater significance in dividing the behaviourally consistent opening section from the changes in behaviour which characterise the rest of the piece. The ‘slow section’ can now be seen as something of a watershed, not only changing the mood and energy level, but introducing the idea that behaviours can be exchanged between sound types. It does this through several instances of metamorphosis - in particular Exx. 13 (7’39”) and 14 (8’23”) - where one sound type passes through a number of different behaviours in a more or less seamless transformation. Conclusion Based on the conclusions drawn from the study of sound types and behaviours in Novars it is possible to suggest a number of alternative ways of viewing its structure. The detailed structural scheme already seen in Fig. 3 is one such view, in which a complex pattern of multiple ‘expositions’, ‘developments’ and ‘recapitulations’ is based on two contrasting ‘subject groups’. This scheme bears a fair likeness to one possible theoretical structure based on multiple nested Golden Sections, and is based purely on an analysis of the piece in terms of sound types. Andrew Lewis (1998) – Francis Dhomont’s Novars Two further views are illustrated in Fig. 5, both of which add to the sound type analysis aspects of behavioural typology. The first (Scheme I), is a simplified version of that shown in Fig. 3, and groups expositions, developments and recapitulations together, linking them to the incidence of behavioural changes. Scheme II simplifies Scheme I even further, suggesting a ternary form based on behavioural change, with most changes taking place in the long central section and a behaviourally static coda bringing the work to an end. Even a casual hearing of the work might tend to confirm the idea that Novars moves from some form of simplicity towards increasing complexity (and Fig. 6 is particularly convincing in this respect). Familiarity with the original Schaeffer material allows the listener further to perceive the work’s progress as gradual shift from the voice of Dhomont (derived from Schaeffer) to the voice of Schaeffer (in the light of Dhomont), passing through a middle ground where the two become intriguingly entangled, and with the ghost of Machaut ever present. It will also be clear to most listeners, however, that Novars is structurally much more complex a work than perhaps any analytical scheme might suggest, particularly as what has here been condensed to a few pages of graphics actually takes place over nearly twenty minutes of highly demanding listening time. Nevertheless, the overriding impression which Novars leaves as an immediate aural experience is one of organised artistry and craftsmanship of the highest level - a conclusion which is by no means at odds with a more considered ‘non-real time’ analysis such as this, and one which would seem to confirm Novars as an appropriate homage to Vitry and Machaut, Schaeffer and Henry, and to the great traditions which they pioneered. References Dhomont, F. (1989) Mais est-ce de la musique?. In >convergence< GUIDE, Montréal, CEC. (Tr. Laurie Radford & Claude Schryer). Dhomont, F. (1991). Programme note to Novars. See booklet accompanying the CD Mouvances-Métaphores, Disk 2 (listed in the discography). Discography Schaeffer, P. (1991) L’oeuvre musicale intégrale, Schaeffer: ‘Étude aux objets’ (1959 rev. 1971). (Compact Disc INA C 1006-9). Montréal: empreintes DIGITALes/INA•GRM Dhomont, F. (1991). Mouvances-Métaphores, Disk 2: ‘Les dérives du signe’, Track 1: Novars (Compact Disc IMED-9107 / 08-CD). Montréal: empreintes DIGITALes Andrew Lewis (1998) – Francis Dhomont’s Novars Appendix Score: Andrew Lewis Andrew Lewis (1998) – Francis Dhomont’s Novars Novars – Francis Dhomont Andrew Lewis (1998) – Francis Dhomont’s Novars Andrew Lewis (1998) – Francis Dhomont’s Novars Andrew Lewis (1998) – Francis Dhomont’s Novars Andrew Lewis (1998) – Francis Dhomont’s Novars