The typology of hearing and general musical experience

- The latest on absolute pitch -

Originally published in Zeitschrift für Musikwissenschaft, 13, 21-8.

Albert Wellek, Wien.

Heinrich Hein's informative report [1] on Lothar Weiner's laudable "Analysis of absolute pitch" invites additional comment regarding the question of "types" of absolute perception; as Hein emphasizes, even after the Weinert experiments the question remains thoroughly open.

Weinert encountered a distinction between "unipolar" and "bipolar" types which was also recognized by O. Abraham [2].  These distinctions don't seem to indicate altogether different abilities, but rather different levels of the same ability.  However, I would not consider these levels of ability as quantitative.  Certainly, the range of the ability is excellently described (such as with the ability of active pitch reproduction), but there are certain aspects of the ability which may or may not appear together with other aspects.  Weinert himself assumed that the "unipolar" type is incapable of actively producing pitch sounds and is highly dependent on timbre, more or less "chained" to one or two timbres; although my observations have typically confirmed this association, this is not always the case.  If the ability were to be distinguished in a quantitative manner, the division would probably result in still more identifiable types and subtypes, which is already implied by Weinert's distinction between "total" and "partial" (i.e. constant and occasional) absolute hearing.

My acceptance of multiple qualitative types-- which Hein mentions specifically in his paper-- is not merely a philosophical acceptance, but is fully proven by experiments which I have already conducted even prior to Weinert's report.  So far, there are 20 cases that I have experimentally examined, and numerous additional cases which I have personally interviewed.  Of the former, there were 15 cases examined in Vienna, 3 in Prague, and 2 in Hamburg, some of whom were also Weinert's test subjects and were amiably referred to me by Dr. Weinert.  My investigations resembled those of G. Anschütz as well as Weinert's, but I expanded on the procedure in the following three ways:

1. Second octave

over the entire piano range (73 per trial)

2.  Major-

Triads (4 consistent in narrow octave range)

3.  Minor-

My evaluations of both Weinert's test subjects in Hamburg are now confirmed in this series of triad tests, with a decided preponderance of semitone errors-- at least 2/3 to 3/4 of the total number of errors.  Indeed, hardly a single case of identical ability arose in any trial in Vienna or Prague.  Here there are errors with harmonious intervals, in the first trial confusions with fourths and fifths [3], and in the second trial thirds present the largest problem.  The most extreme case shows about 44-46% errors with a series of triads, and even up to 50% errors with fourths.  This suggests a boundary where the majority of errors takes place within the fourth interval, but the consonant intervals (fourths and thirds) produced the greatest number of errors.  On the other hand, the fourth error has the majority, outweighing the semitone and other consonant errors overall.  Sometimes the halftone errors are more numerous than whole tone errors, and other times the halftone error is more common, but there seems to be a certain commonality to third errors.

The ear seems to orient itself to certain classification systems:

First, the scales
a) chromatic
b) diatonic

2a) the circle of fifths,
2b) the series of thirds in our harmonic system.

The bare fact that cases of the second type exist indicate that the rules of absolute pitch represented statistically by Weinert do not define absolute pitch universally, but describe instead only one type of absolute pitch.  The types of error tended toward by Weinert's subjects ("1") versus the dissimilar errors committed by my subjects in Vienna and Prague ("2") forces the interpretation that the basis of their absolute pitch ability is not equivalent.  One type orients itself to a broad spectrum of sound, to the linear function of brightness [3] ("tone height" in the narrower sense as Révész puts it); the other detects a specific subjective tonal characteristic [4] (Révész's "tone quality") which could potentially be represented by the circle of fifths, or curves, or irrational functions.  I call these two types of hearing linear and polar [5]:  "linear" if neighboring tones seem similar to each other, and "polar" if neighboring tones seem to be opposite.

These two dominant modes of absolute hearing recall Révész's theoretical distinction between "types of absolute listening" [6].  I would add to his statements with one substantial amendment:  that absolute sound recognition can orient itself predominantly either to the "height" or the "quality".  Also it will have to be understood that these abilities can manifest themselves very strongly even in extremely complex sound clusters.  In the total impression of the sound, the combined sound is "itself" above all its components, existing apart from the components (particularly spatial, visual, motor, tactile) and the abundance of "arrangemental" [7] conditions or the listener's musical experience, from which a pitch assumes its identity within a system.  Hein specifically mentions this.  Révész, however, makes no such separation, and above all offers no criterion for the distinguishing of the two types; he merely indicates that there where there is no precise "quality recognition" then a more unsteady, more uncertain "height recognition" must be present [8].  Révész rejects the proposition that quality recognition occurs in the central octaves and height recognition occurs at peripheral locations.

The truth lies somewhere between Révész and Weinert.  There is no single type of absolute listening; nor is there a type which switches between dominant modes of recognition.  There are instead different types with different bases for their absolute judgment.

Diagnosing which type is present by using the data provided by the subjects' errors is often difficult.  An extraordinary case, with whom I met only once, made no errors in any octave.  Most frequently, however, there is an area which remains error-free regardless of the method:  that is, of course, the middle octave.  This is the area where the tone character is most purely evident, and within which tone quality is most obviously apparent; additionally, the brightness and distance in the peripheries of the musical range fluctuate and distort so that it becomes more difficult to detect the heights which are so evident in the stable middle octaves.  The nature of the central octaves becomes even clearer in cases where it is not distinguished by lack of error, but with a different statistical occurrence of errors than in the periphery.  There are cases whose raw error count produces an entirely obscure picture; if, however, one arranges the errors from the exterior octaves inward, then in this polar arrangement the errors will diminish as one approaches the center, up to the complete disappearance of the semitone error.  A musician, in real life, will generally produce music only within a handful of octaves, creating a well-defined "central position", while the peripheral areas are more or less infrequent and chaotic.  Nevertheless, Révész's hypothesis states that pitch recognition would occur from the recognition of sound quality even for these extreme pitches, but this has no validity even for the polar type of absolute pitch; there are cases where the polar type is apparently making such judgments of the border pitches, but cases with the middle octaves removed cannot generally be differentiated.  Statistically, judgments of the outer octaves regularly resemble the spectralinear type of listening, with its attendant insecurity of judgment.

On piano particularly, especially with the baby grand piano and pianino, the bass register is provided in the purely mechanical sense of number of covered strings.  But there is a treble limit marked by where the mute pedal starts (or stops), i.e. the point where the sounds are too weak to be significant.  Actually the zone bounded by these two points, for instance from middle B to F of the third octave up, have the statistical characteristics of the "middle octaves".

A special demarcation between middle octaves on the one hand, and high and low octaves on the other, is absolutely necessary to recognize the difference between the types of absolute pitch, and hereby the necessity arises of making such a distinction.

It has already been mentioned how the listener type cannot be identified without such divided statistics.  Because no such distinction is evident in the material published by Weinert regarding the distribution of error magnitude in individual octaves, it is possible to suggest a linear type of height listening for all 22 of his subjects who do not have strong identification skills; those who have flawless or near-flawless results leave the question open.

Fortunately, for the two dominant modes of absolute hearing, there are still other criteria other than error size distribution.  One of these is different reactions to unaccustomed transpositions.  The threshold of sensitivity in this kind of absolute discrimination is much stronger in the polar type of listener; the polar listener will also register more clearly any mistuning on the piano.  If naming skill is accounted for, the ability to indicate the correct octave of a given pitch is usually weaker, and every now and then the listener will be completely off [9]; the same applies to octave distances between tones (of the same pitch) and the determination of the exact octave of same.  The subjects' descriptions of their inner processes cannot be considered authoritative.  The differences expressed in the experience of the pitch character is still so meagerly clarified and so difficult to grasp conceptually that people who are not aware of the theoretical alternate method of judgment will not understand what they are being asked to attempt.  The blurred distinction between "brightness" and "tone quality" already creates difficulty, as Weinert's experiments show in his having to explain himself to his test subjects.  A better approach than the theoretical description would seem to be the subjects' direct experience of pitch, but here also the reply is limited to the subjects' skill at introspection which naturally is of unequal value.  At the same time, the difficulty of the question itself should not be underestimated.  Trials to determine pitch similarity (excluding octave classifications) generally produce negative results [10]; similarities between the tones are therefore still in agreement with Hornbostel [11].

Further typological criteria are also found on the far side of absolute hearing.  The question of pitch similarity can be successfully answered by non-absolute listeners, which is likely a result of stable systems of tonality and harmonic ratios, e.g. in the old Indian metric (where they are highly individualized) [12], which conforms to the Hein hypothesis of a general latent absolute hearing ability.  In all cases, absolute hearing is only one border which can be drawn within the boundaries of musical hearing and musical life, albeit an unusually evident and thus measurable skill.  So it should not be surprising that there are similar types to be found among relative and harmonic listeners, in which the distance and width of brightness, interval color, and chord color may be assigned certain qualities (a higher-order quality), or may instead be judged predominantly on distance and width even if these are recognized as the "color"-- and with progressively weaker ability, the more uniformly the chordal structures are perceived.  One is a linear, another a polar type of listening, and both types of absolute listeners may show either type of attitude toward intervals and chords.

It may be redundant to emphasize that there are also cases of mixed types and subtypes, some of which are suggested above.  Overall, it is essential to note that the total attitude of hearing shifts in relation to higher complex qualities.  Immediately evident is the change of confidence in judgment of triads-- sometimes considerably larger or smaller, other times the same, sometimes only for major, or only minor, sometimes one versus the other unequally.  The relative hearing ability in the middle octaves can be changed, just as the linear or polar listener can learn to better identify triads (percentage-wise).  It is not unusual that a predominately linear listener can recognize the quality of a triad, or of a particular type of triad, in a predominately polar manner.  Likewise, the linear listener may not identify the proper octave, but the security of their judgment will nonetheless increase toward the middle octaves as will a polar listener's.

Now there is also a third type of absolute listener which I have identified and assessed.  Among the absolute listeners I have previously experimented upon, there were two who analyzed the tones as colors, that is, with whom the pitches were associated with specific colors; there was also a case of synthetic imagery, which I must separate out because its synaesthesis was not attached to individual tones.  Generally, I understand that the third type is contingent on overtonal quality, or perhaps is only a pathological complaint of the absolute ear, especially through the hearing of actual colors.

Cases and methods of more indirect (more logical) development of judgments from overtonal qualities have already been compiled by Abraham, such as the case of an R. Hennig, in which only one pitch class was accompanied by an imaginary image and by this was recognized.  None of this results in an ability somehow comparable or equal to absolute hearing; mainly, it comes off as some kind of partial absolute hearing.  But even in the hypothetical case in which all 12 sounds are assigned a special color and judged from this would not indicate absolute synaesthetic hearing.  In actual synesthetes the association is involuntary, with no apparent conscious decision involved; it is more probably a malfunction of the hearing ability.  The color cannot therefore be the exclusive means of recognizing the sound and therefore cannot be the essential sine qua non for the absolute pitch ability.  It would be difficult to show that the color appears first and the pitch judgment follows; even if the process seems to be a conscious flow, the soul is not so easily monitored.  Both experiences could be intricately intertwined in a "functional interdependence"; the apparent flow could be conditionality, not causality.  In this sense, the color becomes meaningfully associated with the pitch, while the actual characteristic representing the pitch is felt only subconsciously-- but without this deeper experience, the synaesthetic experience would disintegrate.

These problems have already been reported elsewhere, provisionally [13], describing the case in which, with progressing years, the absolute color hearing fades and pales.  Taking this type into consideration, it is probable that this type may produce a whole new set of subtypes and mixed types; not merely the first two types as described, but also with a factor dependent on the strength of the imaginary images.  The synaesthetic absolute hearing could be either linear or polar; it could be totally or partially conditioned by the visual impression; it could be wholly dependent on or only occasionally supported by the phantom image-- the latter of which perhaps would be via uncertainty in extreme situations, in unaccustomed timbres, etc, in which the judgment is made by fitting known information into a new template.  The liveliness of these images and degree to which they influence the listener stipulates further typological possibilities.  Generally speaking, these third types are extremely rare, and in fact each case seems a type-- or rather, a subtype-- all its own.

Also, this third type of absolute listener is only the borderline case of a more general type of musician who prefers to think visually when it comes to music.  In this direction there are further and far-reaching consequences arising from the types outlined here.  The linear listener seems more prone to analytical behavior or character type, and the polar listener seems more to be an integrated or holistic thinker.  In addition, there is an unmistakable connection between linear hearing and linear polyphony (a happy coincidence), between polar hearing and harmonic musicality (also increased sensitivity for sound).  One type seems to be more common in the north of Germany, and the other in the south [14]; and, with this, another entirely new philosophical outlook on music is opened.

To support this outlook, we designed the following questionnaire.  The questionnaire is not to be taken as a formal experiment; it is meant only to help guide future experiments in larger geographical surveys when they can be practically attempted.

In the following, the views solicited are meant to provide only an entirely preliminary sketch and introduction to a comprehensive phenomenology of musical comprehension.


The questionnaire presented here is concerned with only the most relevant information in order to be easily answered by as many people as possible.  Therefore, no attempt is made to surpass the valid and highly precise questionnaire created by Abraham.  Because our principal interest is the clarification of the subjects' psychological condition, it was unfortunately unavoidable to also ask about the origin of the answers.  This question should not place me under suspicion!

Each absolute listener of each nationality is asked to send their answers to me at this address:  Dr. Albert Wellek, Vienna XIX, Pfarrwieseng 11.  Please answer the questions thoroughly; foreign language contributions are welcome.

1.  Name, occupation, exact address.

2.  Year and place of birth, nationality.

3.  Where did you grow up?  In which places did you spend the most time?

4.  Where were your parents born?  Raised?  Nationality?

5.  At what age did you recognize your absolute listening ability?

6.  Which interval sounds (besides the octave) seem to you to be most similar?  Which seem especially different?  (Not theoretically, but according to your direct impression!)  The third?  Fifth?  Semitone?  Whole tone?  Tritone?  Seventh?

7.  Do the white and black tones of the piano seem to you substantially different from each other?  Do the white give you one kind of impression and the black another?

8.  In which areas do you make the most errors with your absolute identification?

a.  in the middle octaves (e.g. with unfamiliar instruments)?
b.  with high tones?
c.  with low tones?
d.  or are all areas the same?

9.  Do you mistake octave intervals for single tones (or vice versa)?

10.  Do you recognize the exact octave of a tone without difficulty?  (Check yourself before you say yes!)

11.  Do you recognize a specific kind of limit to your absolute pitch ability?

a.  between an eighth and quarter tone
b.  about a semitone

Does this limit deviate?  What mistakes do you make as a result (and in which octaves)?

12.  Do you associate musical sounds-- either pitches, intervals, chords, or any other-- with colors?

a.  Always?  Consciously or unconsciously?
b.  Or occasionally, only when you think about it?

If yes, please name these colors!

[1] Neues über das absolute Gehör, ZfM XII, 1930, S. 503-06.

[2] Das absolute Gehör, SIMG III, 1901, S. 46t

[3]  With me, as with Weinert, a determination of the Octave height of the testing sounds was generally not asked for, as we were only looking for an outline of the possible error sizes.  An interval larger than the half-octave (the Tritone) cannot be as easily distinguished as the fourth or fifth, which in turn is more difficult than the third or sixth, etc. (Cf. below S.25, Anm. 1.)

[4]  This concept of Hornbostel's sound psychology is preferred to that of Révész, which is more likely to cause confusion.

[5] So zuerst in der ZfM XI, 1929, S. 474f. ("Das Farbenhören im Lichte der vergleichenden Musikwissenschaft").

[6] ZIMG XIV, 1913, S. 130ff.

[7]  The old expression "associative" would be sooner true of the participants previously named in this document.

[8] Vgl. auch Révész, Prüfung der Musikalität, Z. f. Psychologie LXXXV, 1920, S. 179f. Here the "octave deceit" is supposed as a feature of the "qualitative" ear type, is needed yes with this to a certain extent synonymous! (Cf. meanwhile below.)

[9] Cf. also Révész (a. a. O.) Because meanwhile the capacity to assign, the correct signs the correct heights, a practice prefaces, that is absent also from very bewanderten musicians usually, a dependable condition is hereon often impossible.

[10] So auch bei Révész im Hinblick auf die "Tonqualität". (Vgl. ZIMG XIV, 135.)

[11] v. Hornbostel, Musikalische Tonsysteme, Handb. d. Physik, Hg. Geiger u. Scheel, Bd. III, Berlin 1927, S. 428; Psychologie der Gehörerscheinungen, Handb. d. Physiologie, Hg. Bethe u. a., Bd. XI, Berlin 1926, S. 718. 4 Vgl. Wellek a. a. O., S. 480-83.

[12] v. Hornbostel, Psychologie der Gehörerscheinungen, a. a. O. (Hier wäre zu Intervall- und Akkordfarbe noch eine Melodiefarbe, zu Distanz und Breite ein Melodie-Profil hinzuzureihen.)

[13] Besonders ZfM IX, 1927, S. 583; auch Arch. f. d. ges. Psychologie, Bd. 76, 1930, S- 201.

[14] For this the previous literature of yet two criteria offers: 1. The questionnaire Abrahams that will have circulated probably predominantly in the northern half of Germany, zeitigte under 100 Beantwortern "several. . . Which itself easily around a fifth. . . error, and one, which more often mistake makes around a major third. All these observers are be mistaken such, which only exceptionally around a semitone". (A. o. a. O. S. 8.) 2. One first recently, shortly before Weinert's, appeared study of M. Gebhardt ("Contributions of absolute pitch research to compulsory preschool training") describes a case out of Würzburg with clear polar features. (Arch. f. d. g sharp. Psychology, Bd. 68, 1929, bes. S. 276 and 280.)