Inner hearing and direct listening

The difficulties that have been encountered recently by orchestral conductors when confronting new works are indicative of the importance of inner hearing. This particular faculty was put to the test during research conducted around The Angel of Death, in a situation where the author went from being a subject to being a co-worker. The present brief article confronts personal experience with observation of the professional musical milieu. The article will suggest that the gap between inner hearing and direct listening bears mainly upon tempo and timbre combinations. Hearing with the inner ear is not a particularly easy terrain for scientific investigation, nevertheless it should attract researchers' interest, because it has considerable contributions to make in constituting and defining musical proficiency.

Reading a tonal musical score is a highly complex process supposed to enable the reader to hear with the "inner ear" what is being read, without too much difficulty. Several variables can make the reading easier or on the contrary can hinder it: the number of staves (in the case of orchestral scores) or the presence of transposing instruments. The texture may be complex, or the reader might not possess sufficient knowledge to read a score specific to a particular instrument. If the reader is very advanced, inner hearing will make use of internal anticipation, exactly as in direct listening. Very often in such a case, the reader will be able to anticipate what comes next, especially if the composer's language is familiar. But with atonal music, the situation is very different: internal hearing becomes difficult and will only function after a long and detailed study of the musical text.

Right at the start of the 20th century scores began to appear that placed orchestral conductors and performers in an embarrassing position. The most famous example is the way the great composer and conductor Gustav Mahler reacted when reading the score of the First String Quartet Opus 7 by Schoenberg:

I have conducted the most difficult scores of Wagner; I have written complicated music myself in scores of up to thirty staves and more; yet here is a score of not more than four staves, and I am unable to read them.¹

This surprising admission on the part of a conductor of the stature of Mahler showed a humility and an honesty that we can only admire. Basically, it is not difficult to understand Mahler's embarrassment. Schoenberg understood very well why his illustrious colleague found himself in difficulty: the highly contrapuntal writing and the independence of the instrumental parts of the work in question generate unexpected combinations of sound very difficult to understand on a conventional harmonic basis. And when Schoenberg's later writing eliminated all traces of tonality, the difficulty grew greater still. Reading a score such as Erwartung demands intense and prolonged application of inner hearing, and even a good reader, with advanced training, may not gain anything more than an approximate idea. As for certain scores that abounded in the decades following the Second World War, their unprecedented complexity is such that the orchestral conductor who wishes to master them has to devote an enormous amount of time to the task. Here, the question of inner hearing becomes even more acute. The problem also concerns listeners in direct listening situations and, naturally, composers. When Roger Reynolds talks about "[…] unrealistic views regarding what listeners can hear” he puts his finger on a sore spot. [Cf. Collaborating with Perceptual Psychologists]. The scientific project that was built on and around the Angel of Death possesses the great merit of addressing listener perception in a frontal manner. It might be thought that in this context, an essay on inner hearing could be marginal or even irrelevant. Nevertheless, we wished to open the discussion, even though it puts us in a somewhat uncomfortable position: inner hearing is at the very heart of the preoccupations of the professional musician and the musicologist, and has a major influence on how music is performed. But of all the actors who take part in putting together a new piece, the orchestral conductor is the most exposed.

In the past, there have been remarkable conductors in the service of new music, which includes works from the First Viennese School: Hermann Scherchen and Hans Rosbaud are among the great pioneers to whom Pierre Boulez justly paid tribute.² Furthermore, we know it is because there were not enough conductors that Boulez himself became one. In France, the Domaine Musical³ was created in order to answer a need for high quality performances in Paris. But what was the true nature of the obstacles confronting most conductors? We do not believe that the difficulties were exclusively visual, especially in a country like France where traditional solfeggio training has long been rigorous and intense. Rather, as Boulez found out, the problem was the difficulty or even the impossibility of generating inner musical imagery from visual information. Music could no longer be navigated using tonal techniques. Most conductors found it impossible, with inner hearing alone, to hear the scores before them in a satisfactory manner.

>They came to the podium with only a very imprecise inner image of the work, making it impossible to conduct the music with any clarity, having to make do with hasty last-minute rhythmical adjustments. The performers under their baton could only struggle with their own parts, unable to gain an idea of the whole, navigating blindly so to speak. Naturally, performances left much to be desired. New works suffered from not being conducted with a sure hand. The most evident symptom of the general weakness was the slowness of the tempos, severely mutilating the music. Abnormal slowness, severely criticized by Boulez,⁴ was not the result of choice or of an aesthetic decision. It happened because overwhelming technical difficulties were disrupting the entire cognitive process.

What takes place when great conductors premier new works? We know that they do not have the advantage of direct listening before the initial rehearsals, so we can be certain that these conductors have nothing but score reading when building up an inner image of the work they are going to conduct. Certain conductors " Leonard Bernstein for example " demonstrated an extraordinary faculty for taking possession of a complex score in a very short time. In conductors of this stature, very quick visual information processing is combined with extremely precise inner imagery and exceptional memory. Yet there is nothing miraculous in these amazing performances: it results from training from an early age in certain cases and unrelenting hard work in all.

The great diversity in writing styles and certain codes found in the music of today can pose a very serious problem for conductors. The solution lies in extensive analytical preparation to lay bare the underlying musical grammar and the way a work functions, before starting in on the conducting itself. A conductor-composer like Pierre Boulez partly draws his strength from this analytical phase. It opens the way to succeeding stages. Boulez has an infallible ear, held in awe by instrumentalists, and it is proof of the high quality of his inner listening: musicians know that the slightest error or omission will immediately be picked up at the very first rehearsal. They no longer risk voluntarily playing anything other than what is in the score.

Rebellious behavior of this kind, unthinkable in Germany or the US, was not rare in France in the turbulent times when contemporary music ensembles⁵ were springing up. Although ethically inexcusable,⁶ such facetiousness was not without grounds, cognitively speaking: the musicians, specialized in performing contemporary music, as excellent and highly experienced technicians, knew that the degree of control exercised by the conductor at the first rehearsal depended entirely on his inner ear. They were aware that the variety and complexity of the textures, as well as the diversity of writing styles, could be formidable obstacles to achieving a workable aural inner image. So it was at the first rehearsal that the conductor's credibility was established or destroyed: in music of this kind, professional experience alone⁷ cannot compensate for deficient abilities in musical imagery.

The control exercised by the conductor, which in Boulez' case was to become absolute, is therefore the fruit of enormous experience as well as an unswerving determination to master the score, both architecturally and down to the fine detail. Very few conductors achieve that level. For most music professionals, the gap between direct and inner hearing remains considerable. The gap only closes during the interactive phase of the rehearsal. The surprises that await one at the first direct hearing of a piece, compared to what one had imagined, probably arise out of the normal limits of the human brain capabilities, given the complexity of the score presented to the reader.

In the wake of the Domaine Musical, specialized music ensembles were constituted and the training of highly competent professional musicians became common. This was an adequate and pragmatic response to a general problem. Today, young musicians who specialize in contemporary music or who devote much of their time to it, are in a different situation. They enjoy the benefit of experience accumulated by two preceding generations and thanks to recordings of the more significant works, can easily get to know the overall sound types associated with the various texture types.⁸ They have the necessary global overview and they know how to take advantage of an abundant documentation " unlike their forerunners, who found themselves in less favorable conditions. Nevertheless, approaching a new score from scratch, by reading only, remains very difficult, for reasons that today are well known.

Clearly, an expert pianist, in possession of a good mastery of tonal grammar, will readily pick out overall outlines, harmonic patterns and instrumental gestures (scales, arpeggios etc). Therefore, it is the identification, subdivision, and anticipation of structural units in successive visual fixations that condition the ease of sightreading at the piano. The same good musician, without a keyboard, will nevertheless succeed in hearing in his or her head what was seen in the score, even though we cannot really know just how exact that inner image might be. But when it comes to reading a score in which the cues normally furnished by harmonic and metric structure are absent, the effectiveness of the hand-eye span is reduced. The more novel and complex the musical language contained in the written score, the greater the mental energy required by musicians to hear it in their heads.

Performers who play a lot of contemporary music use texture cues as much as possible, and display remarkable agility when sight-reading. It would seem that there are new common instrumental gestures that can take the place of those in the tonal register. For example, interval structure and statistical assessment of the number of notes in a run, or knowing a trajectory well when performing lateral movements on the keyboard, improve performance. Performers who return to the tonal repertoire do so with enriched perceptions, thanks to their practice in new music.

There is an obvious link between being able to sight read at the piano, for example, and being able to hear with the inner ear what one is reading. It is possible to learn the mechanical process of transforming visual information at great speed into instrumental gestures, without attempting to first hear the music in one's head. However, this is only part of the process. A purely mechanical approach that did not take inner hearing into account would result in a very unmusical performance, not to say a meaningless one. There must be some kind of mental image, however tenuous, of what one wants to play. This is the crux of the problem. An inner image can only be acquired with long and intense practice. Any reader facing a musical text based on a new kind of musical grammar will experience the greatest difficulty in constructing such an image. Their inner hearing will be no more than a narrow temporal window and they will quickly experience a feeling of saturation and fatigue.

Nevertheless, reading at the table has an advantage over instrumental sight-reading: silent readers can go over a difficult passage as many times as desired. They can go back over their own tracks with impunity. By contrast, when sight reading with an instrument, especially if more than one musician is involved, there is a golden rule that the musician should never go back. Otherwise the other musicians are disrupted and the work of the entire group is hindered " it may be a chamber ensemble or an orchestra.

That being said, the question remains as to exactly what is heard when reading at the table. In order to offer some answers to this question, and sketch out possible avenues of research, we would like to draw upon our own experience and sensations. In the course of working with the psychologists in the context of the Roger Reynolds composition The Angel of Death, the author went from being a subject in the initial experiments to co-worker in succeeding stages of the project. This change afforded some surprising observations, and we would like to share them while they are still fresh in memory.

Our first discovery of The Angel of Death took place under interesting conditions. Brief extracts (subsections) of the thematic materials from the piano part, judiciously selected and printed on 34 pieces of paper, first had to be classified on the basis of their musical similarity as imagined. This was a score-reading counterpart to a similar listening task described in Vieillard et al. [Cf. Perception of musical similarity among contemporary thematic materials in two instrumentations]. We started out by grouping them into five categories, explaining the reasons for our choice and our classification criteria without however concluding that this necessarily meant they sprang from five "themes”. The next task consisted of bringing together our "similarity appraisals” in a single tree structure, on three levels. We found it impossible to complete this second task. The use of the word "theme”, which the psychologists proposed should designate little compositions made up of thematic material, and the difficulty in assessing the time scale over which the observed musical phenomena took place, were very probably what caused a misunderstanding. The point to be stressed here is that we based our choices upon an initial silent reading, at the table, of the proposed extracts, without first sight-reading them at the piano. We immediately picked out a number of common pianistic gestures. We were also aware of the high degree of unity in pitch relationships. However, we were unable to pin down what might be the rules governing that unity.

When asked, after that first task, to classify the 34 extracts during a direct hearing session in the laboratory as had the listeners in Vieillard et al (this volume), we were greatly surprised by the rapid tempos. And yet, the tempos had been clearly indicated on the printed fragments that we had read in the course of the first phase, and we knew exactly what is meant by "MM = quarter note at 120 or 150”. In the following phase of the project, having in the meanwhile become the psychologists' active co-worker and no longer a subject, we had access to the complete piano score, identical to that used by the soloist Jean-Marie Cottet. Our curiosity had now been aroused by the difference between the real-world tempo and our inner image of it. Consequently, and without saying anything to our co-workers the psychologists, we arrived at two decisions: firstly, we would proceed with reading the score at the table, until such time as a recording became available to us. The second decision was that until the task of silent reading was completed, we would avoid any kind of documentation concerning Roger Reynolds' work. This was a necessary constraint if we were to approach the score without preconceived notions. Our inner ear was going to be on its own.

When the time came around again for another first direct hearing session, we observed once more the same phenomenon: the tempo appeared to be much faster than what we had imagined. It was only at this point that we made known our observations to the psychologists, in the course of a work session in the presence of the composer. What might the explanation be for the difference between direct hearing, and using the inner ear to read the score? We will examine two hypotheses that do not preclude each other.

The first hypothesis concerns the initial reading of the cut up sequences. Having to classify the extracts using specific criteria changed the way we carried out our first reading. The fact that we had to apply analytical reflexes to fragmented material, while sight-reading, slowed down our inner ear. As a result we later remembered the music as being significantly slower. It was only when we listened to the same extracts in the laboratory that we realized the extent to which we had unconsciously misrepresented the tempo. We do not know whether this would have been the case had we read the music with no intention to analyze. In a musicologist, analytical reflexes tend to take over, and the desire to understand how a language functions becomes a natural reflex. However, we do not make the mistake of confusing this type of analytical work, carried out upon fragments of music while operating under constraints, with the complete analysis of an entire score, as we will show below. The latter approach would be a very different matter.

The second hypothesis concerns the continuous reading of the entire piano part, while at the table. In the course of this new stage, we became conscious of the above-mentioned slowing down. And yet, when listening to the recording, we were obliged to recognize that we had once more misrepresented the tempo, even though we had in the meantime become aware of that risk. Because we were no longer attempting to carry out the task of classification, we conclude that another process was drawing energy away from the inner ear. In our youth, we had spent so much time working at the piano that now, when reading a piano part, we have an irresistible reflex to figure out the technical difficulties. In other words, although the motor mechanisms that control muscular activity are normally at rest while we are reading only, a faint mental motor activity⁹ is nevertheless present, and it interferes with the reading process. As it happens, the solo part in The Angel of Death is studded with minute asymmetrical elements, as well as more spectacular obstacles, such as the torrential unfolding of the core element in Theme 4, Jagged Rips.

A mountain guide, when taking his first look at a cliff face he is about to climb, will allow his eyes to dwell at greater length on the spots where he expects serious difficulty, slowing the speed at which his eyes move over the obstacles. In the same way, technical obstacles in the score slowed our reading: working out the fingerings, the hand movements and the asymmetrical gestures all took place involuntarily, sometimes at a conscious level. While trying to hear in our head what we were reading, and although we were not getting ready to sight-read the music at some later time, we were unable to stop the above-mentioned motor reflexes, and this significantly slowed the inner ear. No such slowing happened while reading the scores of instruments we do not play.

However, the above remarks are subjective and should be placed in perspective. Furthermore, pianists are accustomed to reading two or even three staves at the same time and so will usually read faster than other instrument players. In addition, slowing was not uniform throughout the piano part. It mainly concerned faster tempo and apparent higher density passages. However, we did notice the phenomenon in some slower passages, such as Other, in which the glissandi and the lateral movements were obvious sources of difficulty.

Finally, in considering the slowing down phenomenon, we must not overlook the fact that we are dealing with a new or unknown musical language. As mentioned, this requires greater concentration, which, added to accumulating difficulties, will always put a brake upon internal mental processing. The flow of the inner ear inevitably slows down, compared to what it will do with a musical language well known to the reader.

There were other surprising elements when listening to The Angel of Death with the inner ear, for example the importance of ornamental notes. The difficulty is not in reading them, but in the way that the ornaments must flow with the musical discourse. The long ornamental example given below, taken from the Domain part of the piece, is easy to read and easy to hear with the inner ear. However, it irresistibly reminds one of Chopin: this was a surprise that momentarily interrupted our first reading of this passage.

The composer is fully aware of the equivocal nature of this kind of writing, and even sets out to create ambiguity as in the following episode:

Here there were no insurmountable obstacles, but during the initial silent reading, one found oneself wondering what this passage was supposed to sound like. Perhaps the composer expected a certain detachment on the part of the performer? We were unfamiliar with Roger Reynolds' style " at the time of that first reading " but very familiar with piano literature of the past. Clearly, we share that pianistic culture with the composer. Subsequently, when we examined and listened to Fantasy for Pianist¹⁰ (1963-64) and Variation¹¹, it became obvious to us that ornamental notes were very characteristic of Roger Reynolds' writing for piano. Below is a group of quick grace notes taken from Fantasy for Pianist, highly representative of Reynolds' piano music. The work contains many elements found in composers of the Sixties, such as chromatic clusters or unconventional sounds made inside the piano. The tempo given in the score for the following passage is MM quarter note = 180.

The importance of ornamental grace notes in the texture is immediately apparent from page one of Variation (MM quarter notes = 72). In his Performance Notes, the composer gives very precise instructions on how to interpret the ornamental notes, once again showing his deep affinity for this kind of writing and for the specifically pianistic effects that it can generate.

Had we read these pieces before seeing The Angel of Death score, we would have not been surprised at the ornaments, and so neither our reading, nor our inner ear would have been affected. The ornaments, and the historical heritage from which they spring, are fundamental elements of Roger Reynolds' style. The reader's degree of familiarity with the style of a composer can therefore strongly influence the quality of inner hearing when reading a score at the table, in the same way that it can affect precision when sight reading a score for an instrument. Sight reading at the piano has been the subject of far more extensive study than pure inner listening, which poses serious methodological problems in the experimental framework.

The conditions under which we read the orchestral score of The Angel of Death were very different from the two situations mentioned above. The analysis, the study of the documents made available to us by the composer, and listening to the recording profoundly changed the nature of our task: inner hearing was no longer the result of reading only. We were able to benefit from the information gleaned during previous study sessions: pitch series structure, temporal proportion, and textures linked to the identified thematic material. Clearly, traces left in memory by numerous hearings facilitated inner hearing. And yet, even under such favorable conditions, the reading remained initially rather slow. Natural limits to the amount of visual information taken in during successive fixations became apparent: we had to build the overall resulting sound from instrument staves visually distant from each other. We had to understand how they related, and make that perception evolve alongside the orchestration. The unconscious part of the mind advanced in fits and starts towards an accurate image, and to this was added conscious visual searching for a melodic line spread out among instruments across the score. The eye had to seek out certain timbres memorized by the ear in the course of previous hearings, which the visual reading process could not create on its own.

Hearing a score such as that of The Angel of Death with the inner ear is a difficult but not impossible cognitive task. The load is significantly lightened if there are direct listening sessions before the reading, allowing memory to facilitate the reconstitution of the sound. However, the music will be slowed, and will only pick up speed if the reader uses reading reduction techniques to better grasp the score. As far as we can judge from personal experience, reading [at the table] as well as inner hearing, flow much better if the eye can pick out thematic material. This happens with core themes and thematic elements of a strongly marked character, which act as reading guides. We say "guides” in the plural, since such elements are often combined and superimposed.

If intense effort is applied using the visual faculty and the inner ear in passages that contain multiple musical layers, the reader will quite readily "hear” the result of that superimposition, after a few readings. Peripheral elements will graft themselves onto layers that have already been formed and will become part of the inner listening dynamic. However, asynchronous elements remain a powerful braking factor. Passages that contain them require frequent re-readings if the inner ear is to grasp the multiplicity of the time layers.

The Angel of Death is a complex work, in the strict sense of the word, and is richly orchestrated but never overloaded. The composer, aware of past excesses, has written with a certain restraint in order to make the music as transparent as possible. This naturally has made listening to the piece with the inner ear more effective. Nonstandard notation was not used for the ensemble or the piano part, so no new codes are needed to read the score.

Until now we have stressed differences between direct hearing and the inner ear, emphasizing the slowing of the tempi. We should also mention points of convergence.

- The first point of convergence is the greater general legibility of the Sectional part. The Domain part, having fuzzier borderlines, required greater attention, which is after all what the composer intended. It agrees with results obtained from the subjects involved in direct listening (McAdams et al., this volume).

- The second point of convergence concerns the electroacoustic part. Its content is not easily heard by inner hearing, although based upon common material, elements of which can be identified in the score. It carries considerable emotional force, and had a disturbing effect on the listeners during the direct hearing. We are forced to admit that it caused us certain difficulties.

This second point deserves some explanation. We noticed our tendency to "forget” the electroacoustic part when reading the score, or at least to view it as background sound, whenever it was at a certain distance in time. Because our study centered on the piano and the instrumental ensemble parts, we did not acquire sufficient knowledge of the transformations wrought upon the electroacoustic material to be able to construct an inner image of it. Co-workers specifically involved in the electroacoustic part would most certainly have built up a more precise inner image than we did, within transformation limits allowed by the composer. So, whereas it particularly salient in direct hearing, we relegated it to a fuzzy background sound when inner hearing was called upon.

It would not be wise to draw generalizations from the introspective approach that we adopted in exploring the limits of our own inner listening. However, it does raise a number of questions. What would have been the quality of our inner listening if we had read the orchestral score without listening to the recording of the Angel of Death, as was the case when classifying the 34 extracts and the entire piano part? We have no hesitation in stating that it would have been less precise while nevertheless requiring greater time and energy. In addition, we would probably have had a lesser appreciation of the subtle combinations of timbres by which the composer solved the problem of moving the pianistic material to the instrumental ensemble. A clear comparison between inner listening and direct hearing can hardly be made " even when very fresh in our minds, it is almost impossible to pit the memory of that inner hearing against what we hear in the direct listening situation. In our case, the main consequence of comparison was to restore proper tempos. But we are convinced that, without the benefit of a recording, there would have been two further consequences: we would have had to go through the timbre combinations all over again and take a totally fresh look at the electroacoustic part.

Our personal experience in the research surrounding The Angel of Death, and the resulting self-assessment do not have any claims to be of paradigmatic value. But they do bring together a sufficient number of observations and accounts¹² that might serve in setting up the basis for future research.

It would seem that reading with the inner ear requires a thorough knowledge of the underlying musical grammar of the work. Familiarity on the part of the reader with the composer's style also makes reading and inner hearing easier. When these two conditions are not met, the act of reading becomes laborious and the inner ear seriously weakened. Analysis, combined with close scrutiny of the constituent parts helps to overcome these difficulties, although the inner ear will still only give a deformed version of the music, affecting mainly the tempo, and the timbre combinations in the case of orchestral scores. Comparison with direct hearing may hold surprises¹³, but will also make the necessary corrections possible. The latter tend to be numerous when dealing with contemporary works.

It would be a vain Utopia to hope for inner hearing to be as precise as direct listening. We do believe however, that it plays a determining role in proficient musical activity: interpretation, improvisation and above all composition and orchestral conducting. As is shown by the achievements of certain great conductors and performers, inner hearing, with proper training, can attain a very high level indeed. The mental processes surrounding it still need to be understood and verified under controlled experimental conditions. This is not a domain that lends itself to easy or comfortable investigation. However, the stakes are as high as the conditions are difficult: the excellence of inner hearing, insofar as it is verifiable, could very well be an important component of musical proficiency.