Bruno Nettl (5)

Chapter 4

TRANSCRIPTION

There are two main approaches to the description of music: 1) we can analyze and describe what we hear, and 2) we can in some way write it on paper and describe what we see. If human ears were able to perceive alI of the acoustic contents of a musical utterance, and if the mind could retain alI of what had been perceived, then analysis of what is heard would be preferable. Reduction of music to notation on paper is at best imperfect, for either a type of notation must select from the acoustic phenomena those which the notator considers most essential, or it will be so complex that it itself will be too difficult to perceive. But since human memory is hardly able to retain, with equal detail, what was heard ten seconds ago along with what is being heard in the present, notation of some sort has become essential for research in music. This does not imply that analysis and description based on sound-on what the researcher hears-is unessential; quite the contrary .But such analysis must almost always be supplemented by analysis of the material as it appears in notated form. In ethnomusicology , the process of notating sound, of reducing sound to visual symbol, is called transcription. The idea of putting music on paper is based on certain assumptions which need not be generally accepted. According to Charles Seeger (1958a:184): Three hazards are inherent in our practices of writing music. The first lies in an assumption that the full auditory parameter of music is or can be represented by a partial visual parameter. ..upon a flat surface. The second lies in ignoring the historicallag of music-writing behind speech-writing, and the consequent traditional interposition of the art of speech in the matching of auditory and visual signals in, music writing. The third lies in our having failed to distinguish between prescriptive and descriptive uses of music-writing, which is to say, between a blue-print of how a specific piece of music shall be made to sound and a report of how a specific performance of it actually did sound.

Seeger ( 1958a ) has distinguished between two purposes of musical notation, prescriptive and descriptive. The former has as its aim the direction of a performer, and the adequacy of prescriptive notation is judged by the adequacy of the performance, or by the degree to which a performer perceives, through the notation, the composer's wishes. The symbols of prescriptivenotation may be no more than mnemonic devices, as seems to have been the case in the medieval neumes which indicate little beyond the general direction in which the melody is to move. Exact pitches and rhythms were presumably known to the performers, or they were left to their discretion. The Westem system of notation-and the various Oriental systems as well-was developed in order to be prescriptive, and while the amount of detail which it shows has gradually increased, it is still essentially a mnemonic device. There are many things which a composer expects the performers to know and to take for granted. Curiously, of course, the fact that composers do take certain performance practices for granted is responsible for the many disagreements among scholars regarding the precise manner of performance of early music. If composers had been, or could have been, more specific, the work of many music historians would be unnecesssary. In this chapter we are concerned not with prescriptive but with descriptive notation. Historically, descriptive notation developed along with the prescriptive. Descriptive notation, after all, is intended to convey to a reader the characteristics and the details of a musical composition which the reader does not already know. The Baroque composers who used folk songs as the basis of polyphonic compositions performed both descriptive and prescriptive notation. As long as notators of folk songs insisted on fltting the folk styles into pattems of cultivated musical styles, there was not much difference between the two kinds of notation. Only when the great difIerences among the world's musical cultures began to be evident, and when the purposes of the two notations became distinct in the thinking of musicians and scholars, did external difIerences between the two also begin to appear. And these difIerences are not great, since most transcription, for better or worse, is produced in Westem notation. The advantages and disadvantages of this are discussed below.

Transcription is performed in several ways. Earliest, and perhaps least adequate, is fleld transcription, that is, writing down a song as one hears it; it is difficult to write down a song even in musical shorthand in one hearing, especially if it is in a musical style which is strange to the investigator and inimical to Westem notation. But early students of non-Westem music had to use this very moderately accurate method, and even after the invention of recording, fleld workers without recording cquipment brought home fleld transcriptions. In spite of all the obvious disadvantages of notating an African song while it is being sung and danced to, with all of the confusion of the situation tending to distract the transcriber, such a method is presumably better than that of the anthropologist who tried, in an apocryphal story, to leam an Australian aboriginal song and. through long practice on the way home to Europe, made such changes in it that it was transformed into a German folk song. Such a lamentable fate would not have overtaken this song if the anthropologist had become "bi-musical." But for the fleld worker who did not specialize in m~sic, or whose period of contact with the culture he was investigating was short, fleld transcription was the best way of bringing home musical material. ReIated to transcription of music as it is performed in a bona fide cultural context is the transcription of a piece from an informant in the Iaboratory.In such a situation it is possible to ask a singer to repeat songs many times, or even to single out parts of songs for repetition. And while a singer cannot ( by physical and physiological necessity) sing a song twice in exactly the same way, he can come reasonably close, and he will usually be accurate in those matters which are, to him, the most significant in the song. This is presumably what was done by Carl Stumpf in his 1884 study of Bella Coola music.

With the invention of recording, transcription done directly from performance, in the field or the laboratory, became obsolete. Nevertheless, techniques developed for laboratory transcription with an informant may still be used profitably in conjunction with transcription from recordings. Transcribing involves consideration of what is significant and what is incidental in a music-considerations which are related to problems faced in the study of language. Ethnomusicologists have made some use of the theory and techniques of linguistics for approaching some of these problems, and the use of an informant for transcription in the laboratory is one of these. The informant may be asked to comment on the accuracy of the recorded performance, and to identify insignificant or unintended sounds, for example. As recording progressed from the relatively primitive forms of cylinder and disk in the late nineteenth century to the more sophisticated wire and tape recordings in the post-World War II era, transcription became more convenient as well. The main advantage of transcribing from recordings is, of course, the possibility of hearing a piece many times in a single rendition, of comparing sections which are separated by intervening material, and of returning, day after day, to the same piece.

A more fundamental revolution than the development of recording has been the invention of electronic devices which perform transcription automatically. Such devices have always been both the dream and the terror of many ethnomusicologists. They would save the ethnomusicologist hours of labor and agonizing decisions while increasing his accuracy, but they would also allow him less control over his own work and make him the slave of elaborate machinery which produces graphs and similar communications of no interest to musicians, and they might, incidentally, put some of the techniques which he has laboriously leamed out of business. At the time of writing, these devices have fulfilled neither their promise nor their threats, but they are being improved, and there is no doubt that they will soon play a considerable role. They are discussed below, but we mention them here because their existence leads us to examine some very fundamental assumptions regarding transcription by humans.

Transcribing music by hand and ear, as it were, is hindered by the situation in which the transcriber is a native of one musical culture trying to write down the music of another culture, a transcriber using a notation system devised for one culture and foreign to the styles in others. Thus, a concept such as the note, whic.h forms the basis of Westem musical thinking, might be erroneously applied to another musical culture in which the glides between notes are the essential feature. Slight deviations from pitch, hardly audible to Westem ears used to the tempered scale, might be essential distinctions in another music. Again, several obviously distinct pitches could be considered merely different versions of a single tone. The point is that human transcribers, using a notation which is always to some extent selective of the musical phenomena it reproduces, and having a background in a specific musical culture which is also selective of the musical phenomena which it uses as communication, might have great difficulty in first perceiving and then reproducing on paper the music of another culture in such a way that the essential distinctions are indicated in a way comparable to that which would be required by descriptive notation. Assuming that no human transcriber could reproduce a11 of the acoustical phenomena of a musical utterance, he should reproduce those which are essential, and deciding this is probably the most agonizing part of transcription. Electronic devices which transcribe probably cannot be made selective in this way. They record everything regardless of its importance, and selection of the essentials must be made later by the scholar. Thus, even with machine transcription, the informed human interpreter must be available; conversely, even the best human transcriptions can be improved by machines.

Approaches to Transcription The inadequacy of Westem notation as descriptive notation is readily admitted by most authors on the subject, and its failures are vigorously attacked by Seeger (1958a). Nevertheless, even as strong a champion of change as Seeger was prepared to recommend that Westem notation and graphs (see p. 126) be used concurrently. The practical advantages of transcribing in a system at least based on Westem notation are considerable.

Transcribing music is itself an excellent way for the scholar to leam the details of a musical style. There are other ways of doing this -studying by means of performance is one –but transcribing imposes on the student a kind of discipline which could hardly be exacted by mere listening to recordings. Thus transcribing has also an educational function. Until electronic notation devices are readily available and perfected, transcription with some sort of manual notation system remains one of the indispensable tools of the ethnomusicologist. And although attempts at providing other notation systems have been made, the Westem system has traditionally been preferred.

The transcriber is usually faced by musical phenomena too detailed to be notated, and by others which do not fit the notation system. For the latter, special symbols have been devised; for the former, however, there is basically no solution. After all, musical sound (as it appears in the stroboscope, for example) is extremely complex. The slight Buctuations in pitch which occur when a singer performs one tone-the vibrato; the tones he moves through when gliding from one note to the next; the slight differences in length among notes of approximately the same value: alI of these should be perceived by the transcriber. Whether they should be written down or not depends on the possibility of distinguishing, in a musical style, between the essential and the nonessential phenomena.

Assuming that music is a form of communication, there must, in each musical style, be signals which communicate something to the initiated. Just what it is that is communicated we need not discuss here; it may be a nonmusical message or something intrinsically musical. But we can assume that a song sung, say, by a Blackfoot Indian and heard by one of his fellow tribesmen communicates something to the latter, and that this "something" is not perceived by an outsider who is totally unacquainted with Blackfoot songs. Now we may also assume that certain aspects of that song must remain as they are in order for the song to retain its identity for the singer and the native listener, but that there are other aspects of the song which can perhaps be varied in a way which will not disturb the song's identity.

The problem is similar to one in linguistics, where the distinction between phonetics and phonemics has long been recognized, the former being the study of speech sounds as they occur and the latter being concemed with those distinctions among speech sounds which produce, in a given language, distinctions in meaning. A transcription of speech may be phonetic or phonemic-and there is a theoretical possibility that it could be both. The student of an unwritten language will normally begin by making a phonetic transcription, i.e., noting all distinctions in sound which he can, and then try to deduce the phonemic system of that language from the distribution of and relationships among the phonetic symbols. Presumably a similar approach could be followed in transcribing music. It should be possible to move from transcription of all musical phenomena perceived by the transcriber to another transcription which gives only the essentials. But one element of language which is a great aid to the linguist is absent here: meaning, in the lexical sense. For in transcribing music, one can usually do little to persuade an informant to distinguish between ( to him) correct and incorrect renditions of songs, phrases, or intervals. Nevertheless, linguists can sometimes deduce the phonemic system of a language without recourse to meaning; they can identify particles of speech from their structure and distribution alone. Similar procedures ought to work for music, at least to a degree.

Actually, little progress has been made toward developing a phonemic method of transcription for music. But in examining transcriptions by authoritative scholars it is possible to distinguish those who have tried to notate alI aspects of a musical utterance in detail from those who evidently were willing to commit themselves to the difference between essential and unessential distinctions. To be sure, one cannot blandly assume that a transcription full of detail is "phonetic" while one with less detail is "phonemic." The former one may indicate a style whose minute distinctions are essential, while the latter one may be disregarding important details without having deduced a "system" for the style. Finally, before examining specific examples and methods, we must emphasize that details which are not "phonemic" in a language or a music are not, of course, unimportant. They may be unessential to communication, but they stil1 contribute to the character of the style. Their non-phonemic quality-and their omission in a transcription which is based on phonemic principles-is the result only of the fact that their presence can be predicted from their environment.

Those ethnomusicologists who have written about the methods and problems of transcription have not, to this writer's knowledge, identified themselves with the so-cal1ed "phonemic" approach, that is, they have not omitted features of music because these were unessential; at any rate, they have not cal1ed such features unessential. That many have ( consciously, we must assume) made such omissions is evident from their transcriptions.

Among the best transcriptions done – evidently - with the phonemic approach are those of George Herzog, who, while not neglecting details which characterize a musical style, avoids including such a mass of detail as to make a song unintel1igible to the eye. An opposite approach appears to have been taken by Béla Bartók, who included al1 of the details he could, and who believed that it was the small, barely audible effects produced by the voice which actually characterized a musical style. Here of course we encounter a flaw in the analogy between music and language as objects for visual transcription. For while those elements in a language which are so characteristically present as to be predictable in their occurrence and location are considered non phonemic, in a musical style it may be those very elements which give the style its character and which must be represented in a transcription. For transcriptions of music must do two things: they must include the elements which serve to distinguish musical utterances so far as their communicated content is concerned within their style or their musical culture; and they must contain those features which distinguish a whole musical style from another. Only the first of these needs is present in transcription of language. At any rate, Bart6k's transcriptions are exceedingly difficult to comprehend, and they are examples of descriptive notation which could at the same time be used as prescriptive notation. There seem to be times when Bart6k was conscious of the phoneticphonemic distinction. In at least one of his works ( Bart6k and Lord 1951) he gives detailed transcriptions above which he places less detailed versions of the melody which presumably represent songs as they appear in the listener's first impression.

Instructions and advice to the beginning transcriber are not lacking in the literature, and much of this advice may well be heeded today. Otto Abraham and Erich M. von Hornbostel ( 1909/10 ) , in an early paper, admit the deficiencies of Western notation but advise the ethnomusicologist to use it because it is so widely known. They recommend modification of that system where it definitely does not fit the musical style which is being transcribed, and they present a table of supplementary symbols which have since been widely used and generally adopted. The main contents of that table are reproduced in Fig. 1; some symbols generally used but not included in Abraham-Hornbostel are also given. Above a note-approximately a quarter tone higher than written above a note-approximately a quarter tone lower than Written These two signs used next to a key signature mean that the modification occurs consistently throughout the song; if they are in parentheses next to a signature, the modification is only occasional.

Pitch uncertain

Pitch quite indefinite, in the neighborhood of where the stem ends grace note dynamically weak tone long pulsating tone without actual breaks strong tie glide, glissando above a note-tone slightly longer than n.oted, the lengthenÍng being no more than half of the value indicated above a note-tone slightly shorter than noted, but shortened by no more than one-third of the value indicated major structural subdivision or minor structural or rhythmic subdivision Supplementary symbols frequently used in transcription.

Hombostel and Abraham also suggest some procedures. For instance, when a tune or phrase with three tones is identified, but the intervals among the tones are not clear, they suggest using a stafl of only one line and placing the three notes, respectively, above, on, and below it, until more exact measurements can be made. In transcribing polyphonic music, they say it is sometimes advisable to give the over-all melodic impression, even when this is not the product of a single voice. They also recommend, for indication of small intervals, the insertion of additional lines within the staff:

______________________

______________________

______________________

______________________

______________________

______________________

______________________

______________________

And in spite of the early date of their work, they recognize the importance of graphs as they were then used by the American scholar B. I. Gilman.

In a pamphlet designed to standardize transcription of folk music, the Committee of Experts of the Intemational Folk Music Council ( International Music Council 1952) gives directions. Recommending the use of the Westem stafl, the pamphlet suggests the use of special symbols identical or similar to those of Hombostel and Abraham. It suggests dividing the intervals in a song into two groups, those which remain constant, "implying the existence of a definite scale," and "those that vary" (p. 1). Here one might ask whether we can justly make this distinction. Two intervals will appear as separate units in either arrangement, and in the case of intervals which remain constant there will still, in most cultures, be considerable variation among different renditions. The Council recommends a footnote arrangement for variation among the different stanzas of a song, and it urges the transcriber to retain the original pitch of the performance rather than transposing to a more convenient pitch level.

Jaap Kunst ( 1959:39-41) gives valuable advice to the transcriber. In contrast to the present writer, Kunst believes that absolute pitch is an important part of the ethnomusicologist's gift. He believes, also, that the availability of the words is a great asset in transcribing vocal music, not only because of the value of the text for research, but because its structure may illuminate and thus facilitate perception of the rhythmic structure of the song. He cautions the student (1959:38) to work with the realization that the transcriber cannot know whether a given rendition by a member of a non literate culture sounds as it was intended to sound, and he presumably would not have subscribed to the concept of "phonemic" described above. He believes that the best procedure is to perceive the rhythmic structure of a piece by repeated hearings, then to notate all of the different pitches which occur (in a single scale), and then to proceed to transcription of the piece itself. Although admitting the value of mechanical and electronic devices for transcription, Kunst does not consider that such devices can be used alone. He considers them rather as occasional aids to be used when special problems arise.

Estreicher (1957), in a detailed essay which is presented in condensed and translated form by McCollester ( 1960) , describes some problems encountered by him in transcription, and proposes solutions and procedures. Emphasizing the need for adapting the processes of transcription to the particular characteristics and difficulties of the style of music which is being researched, he recommends the use, essentially, of the Hornbostel-Abraham techniques described above. He makes use of controlled sounds superimposed on the music to help him perceive details not easily noted by the Westem-trained listener. For example, he suggests the use of a steady tone signal to indicate changes in pitch level, and of devices such as the metronome to measure changes in tempo which may otherwise be too gradual to be noticed. He suggests to the transcriber that he prepare the tape he is transcribing by measuring it and indicating the points (by number of feet) at which certain key events of the music are heard. He thinks tape essential and will not work with disks. He makes considerable use of slowing down the tape to half speed; this transposes the music down an octave and causes some distortion of sound and timbre while clarifying some of the details which occur too quickly for grasp at ordinary speed. Parenthetically, we should say that the slowing-down technique is more useful for instrumental than for vocal music. In the case of music in which a single motif appears repeatedly, Estreicher recommends writing these various versions of the motif below each other for easier comparison.

Ethnomusicologists certainly vary in their approaches to transcription. While the techniques of one scholar may not be useful to another, it is interesting (and unusual) to find the procedures followed by one individual described in detail in print. Estreicher (1957) does give us a record of his own work, and students who are faced with the task of transcribing and are having difficulty in deciding how to attack the job may profit from his statements. Estreicher (as quoted by McCollester 1960:130) prefers working with stylistically homogeneous recordings wherever possible in order to pick up early in the game typical traits within the musical style. He suggests working at the same time each day for several hours of intense concentration, and arranging the musical sections to be transcribed that day into homogeneous groupings from the simplest, that is to say easiest to grasp and most elementary stylistically, to the most complex, often the beginning of pieces where one finds statements more richly omamented and rapid for example than interior passages. He advises the researcher to listen to the separate sections many times before arranging them in a working order.

Estreicher also suggests working first with the tape recorder at regular speed to get broad outlines, then at half speed for checking details, and finally retuming to the regular speed. He is concemed also with the transcriber's psychological reactions to his own work; spending many hours on what would appear I to be a simple and trivial matter can greatly discourage scholar, and the use of autocriticism, constant revision, and patience are essential. It goes without saying that a transcriber will use a pencil with an eraser rather than a penl Laying aside a transcription for days or weeks and returning to it is also a useful technique.

Altogether, Estreicher's techniques are probably similar to those of most ethnomusicologists. All may not, however, agree with his theoretical assumptions in which there may be confusion between what he calls "reproducing an acoustical phenomenon" and transcriptions which "could reveal the full musical style" (McCollester 1960:131). He does face, of course, the problem of reproducing all acoustical details as against providing a transcription which can be understood by the Western-trained listener acquainted only with Western notation.

But he does not indicate the relationship between the essentials of a musical style, that is, the phenomena which communicate and which correspond to the phonemes of languages on the one hand, and to the nonessential, "phonetic" ones, on the other. Perhaps an orientation toward this useful distinction would help to solve the dilemma, for it is a dilemma which grows as the possibility of an absolute transcription of phonetic-musical phenomena grows with the increased use of machine transcription.

Hints for the Transcriber

Estreicher's recommendations are of great value; a few directions should be added to them. It is useful, when transcribing a piece, to listen to it or to portions of it several times before setting pencil to paper, and .to note the over-all form, possibly with the use of letters.

The latter can be put on the music paper, spaced approximately, so that the details of the transcription - the notes-can be filled in. As transcribing begins, it is often good to transcribe the first phrase or short section in great detail, so that many of the kinds of problems to be met in the piece can be encountered and surveyed early in the game. The use of descriptive notes and whatever information accompanies a recording is of course indispensable, and the information included should be digested before transcription begins in earnest.

After doing one phrase in detail, the transcriber may write down the remainder of the piece in more schematic form so that the outline of the entire structure emerges. After that, the details should be worked out. In the case of monophonic music with rhythmic accompaniment, it is frequently advisable to transcribe the accompaniment first. In general, short bits of music ( perhaps six or eight notes) should be transcribed at a time and replayed many times in the process.

Among the mechanical aids which can easily be used by anyone is the tape loop. If a short section ( not more than two seconds at 7~ in. per sec. ) is very difIicult to comprehend, it can be cut out temporarily and made into a loop which continues to replay so that the pauses caused by rewinding are avoided. Transcribing, as is now obvious, is a very time consuming task. Even a simple song lasting from ten to twenty seconds may easily take one or two hours, distributed over several sittings, to transcribe. The first piece in a style is likely to be by far the most difficult for the transcriber, and students are advised to learn the technique by working on a body of compositions in one musical style rather than trying to sample the world's musical cultures in individual pieces. Students should also be cautioned against working too long at a time; there seems to be -at least for this writer-a decrease in accuracy after one or two hours of work.

If the student is beginning transcription without having a corpus of material which he collected in the field, he would do well to begin with a European folk style. This should be followed by one of the relatively simple, monophonic styles from a non literate culture, such as North American Indian. As a start, styles with strophic song structure are more easily transcribed than others. The student learning transcription should try to do so by concentrating, for a year or two, on a few contrasting musical styles, and by working on each one intensively. Polyphonic material should not be attempted until after a good deal of monophonic music has been transcribed.

Here the availability of field notes is extremely important, and the student should not throw himself at polyphonic material for which he does not know the number of singers or the kinds and numbers of instruments. Acoustic distortions, overtones, combination tones, etc., are likely, in the case of vocal polyphony especially, to put hazards in the way of the transcriber unless he knows a good deal about the circumstances of a performance. In all cases, transcription with tape recordings is to be preferred to work with to other media. Tape does not wear out with constant replaying, as do disks. It is easier to find one's place on a tape than by dropping the needle on a disk; and when it breaks, tape can be spliced rather easily. Some of the slower-moving recorders tend to break tape less easily than those which rewind at high speeds. Using single-track recordings and machines is usually preferable to those with dual tracks, since the former avoid the possibility of "ghosts" from the other track being audible on the track being transcribed. And speaking of the problems involved in breaking and splicing tapes, plastic-backed tape is more durable than paper tape, some of the professional brands developed since 1958 being almost impossible to tear. The exact measurement of pitches and intervals is a special problem which has long interested ethnomusicologists. Identification of the individual tone in terms of cycles per second is the first step. In the laboratory, the oscilloscope and the stroboscope are time honored devices, but they are not frequently available to the private ethnomusicologist or student.

Hornbostel's Reisetonometer was a whistle with a slide which was graduated in terms of cycles. Here the ability of the scholar to judge the identity of simultaneous or successive pitches is an essential factor. This is true also of Jaap Kunst's monochord (see Fig. 2a), which consists of a stretched steel string over a board which is calibrated to vibration rates, and a movable block of wood used for stopping the string. Pitch can be identified by ear and then "measured." The monochord was also expanded into a composite Figura device consisting essentially of twelve suc4 monochords so that a scale or tone system can be reproduced. The monochord seems to be a device of only moderate use to the transcriber, but successful transcriptions have resulted from its use ( see, for example, Brandel 1962) .And it is very useful to the teacher who wishes to demonstrate scales and intervals.

Having found the vibration rates of tones by any of these methods-the Melograph, Kunst's monochord, Hornbostel's Reise-tonometer-the transcriber will need to convert his figures to ex- pressions in terms of intervals. The system most widely used is the cent8 system devised by A. J. Ellis. Conversion of cycles to cents makes necessary the use of logarhythmic tables. A cent is one-hundredth of a tempered semitone, so that a minor third would be 300 cents, a perfect fifth, 700 cents, and an octave, 1,200 cents. Since the series of vibration rates progresses according to the logarhythmic principle, Sachs (1962:25-26) gives a table according to which the log of an interval as expressed in cycles can be converted to cents. The superiority of cents as a means of expressing intervals is obvious. Two numbers must be used to express intervals in terms of cycles, while in the cents system a single number is used. Morever, a difference in vibration rates such as 400 could be any interval, depending on the absolute pitch level, while a number of cents expresses a specific interval, i.e., the same ratio of cycles, at any pitch level. But conversion by means of Sachs' tables, while exceedingly accurate, is complicated. Several shortcuts have been devised; among them, that of Bose (1952a:297) is perhaps the most adequate (see Fig. 2b), though it is a bit too gross for very detailed calculation. But since the human ear evidently does not discriminate beyond perhaps 1/14 of a tone, the margin of error inherent in Bose's shortcut is perhaps insignificant.

Bose's method requires the use of a ruler with centimeters and millimeters marked. Each centimeter indicates a tempered half-tone, i.e., 100 cents. Thus, if we wish to find the interval between two cycles, we simply measure the distance in terms of centimeters. For example, we may wish to find the interval between 370 and 505 cycles; measuring, we find that it equals 51J3 cm., or 533 cents, which is very slightly more than a pedect fourth. Again, we may wish to find the vibration rate for a neutral third, i.e., 350 cents (half-way between a major third, 400 cents, and a minor third, 300 cents), above middle-c (256 cycles). We measure 3~ cm. from 256 and arrive at 312. A method of transcribing by ear, measuring intervals by the use of Kunst's monochord, and converting these measurements to cents, has been used by Brandel ( 1962) .Her transcriptions, which use ordinary notation augmented by the conventional symbols, are accompanied by charts which give the specific distances between the tones in terms of cents. Thus, for the one transcription (BrandeI1962:252), she uses the pitches g, f, e-flat, c, and a, but finds that the distances among these tones are actually 191, 227, 283, and 279 cents, respectively. If the intervals were those of the Western tempered scale, they would have equalled 200, 200, 300, and 300 cents, respectively. Here, then, we have a method which uses Western notation but .assigns to each one a pitch which differs from that which it would have had in the tempered scale, and which indicates exactly the pitch actually used in the piece. But deviation from a "tone" is not accounted for in Brande1's method. Thus we must assume that the distance between a and c is, in this piece, always 279 cents; this obviously cannot be the case, but we have no indication of the amount of variation from this standard.

It is often difficult to distinguish between formal divisions of a piece and the larger metric units. Thus, full bar lines are, in many transcriptions, intended fo separate the small divisions of the over-all form. Differentiation between major divisions and subsidiary ones can be made, and both kinds of bar lines taken together would then provide a proper picture of the metric structure of the composition involved. Above all, care must be taken not to force .the music into an isometric structure of the type common in Western cultivated music. If the transcriber encounters difficulty in deciding on the location of rhythmic divisions, he may do best by omitting bar lines entirely. In Westem notation, bar lines are, after all, among the most obviously prescriptive features. The individual differences among transcribers, their ears and methods, and even the differences between the ways in which a recording will sound to a transcriber at different times, are translated into considerable discrepancies when different transcriptions of one piece are attempted. Thus, the song in Fig. 3a and 3b is given in two transcriptions, one ( 3a ) made by the collector, Frances Densmore (1939:228), the other (3b) by the present writer. This example shows the need for recourse to devices which are more accurate than the human ear steeped in one musical tradition. But in transcribing into Westem notation, the student can at least eliminate some of the devices which are useful only for Western high culture music, for example, the use of key signatures which indicate sharps or flats for tones which do not occur during a song. Thus, the F -sha~ in the Densmore transcription of Fig. 3 is quite unnecessary.

Since rhythm involves the relationship among small segments such as note values but also the temporal relationships among larger formal divisions, it is difficult sometimes to identify rhythmic and metric divisions and to draw bar lines at sensible places. It is advisable to use small formal units as well as stresses in drawing bar lines, and it is essential to make clear to one's self the criteria being used for each type of bar line.

A Suggested Procedure

While individual differences among transcribers and among musical styles to be transcribed require modification of any set procedure, the following one might perhaps serve as a point of departure.

1.    Listen to the piece carefully; read all notes and available material about it, determine the number of singers, instruments, etc. If the piece is strophic in structure (i.e., if it consists of repetitions of a fairly short segment of music), determine the strophe to be transcribed first. The first one on the recording may not be the best one for starting, since it may deviate markedly from the rest.

2.    Decide on the broad structural divisions, and note them with the use of some kind of letter scheme. (Doing this is more properly part of the description of style, but it is useful as a preliminary step to transcription because the transcriber benefits from having the over–all outline of a piece in mind before becoming immersed in its details.)

3.    Determine the number of pitches and kinds of intervals. Transpose the piece to a key requiring few accidentals or ledger lines, but indicate the actual starting pitch somewhere.

4.   Notate the first phrase in detail. Make special note on points which present unusual difficulty.

5.    Notate the rest of the piece or stanza in less detail. If several strophes are recorded, variations of the later ones from the first can be indicated by a kind of footnote arrangement.

6.   If words are available, fill them in. Then use them, wherever possible, to help solve problems of rhythmic detail.

7.    Slow the tape to half speed, and check the entire transcription, especially the points found to be problematic in step 4.

8.     Using normal speed, check the transcription again; then go on to another piece.

9.    A day or two later, check the transcription again, but begin at a point other than the beginning. Possibly one should start with the parts which presented difficulty, or with the ending. Rechecking of this sort should, ideally, be undertaken several times. But care should be taken not to negate, at a moment's notice, what was done at the beginning of the transcription procedure. This writer has occasionally made changes in a transcription and continued revising it until he finally returned to what he had originally transcribed.

Sometimes it is useful to make a rough, undetailed transcription of several songs or pieces in a style, and then to return for more thorough work. Acquaintance with a style in advance of transcribing is, of course, important; this can be gained in the field or by thorough listening to the collected recordings which are to serve as the basis for transcription.

Automatic Transcription

Among the objections to the use of Western notation for transcription is the view that it was invented as a tool for prescription, and that possibly the concept of "note" is not a valid one either for description of music or even for prescription of music in other cultures. As a remedy, the use of hand graphs has been advocated. Seeger (1958a:188) cites the advantages of graphs over notes in detail. According to him, graphs have far greater potential for achieving accuracy even when they are drawn by hand with the use of the same methods as those ordinarily applied by the transcriber in making notation. Exact measurement of tempo, rhythm, and pitch can be more easily approached if the transcriber can cast aside the concept of the articulated note as the main point of order. Rhythm can be represented better in a system which does not depend on dividing units into halves as does Western rhythmic notation. The phenomena between the "notes" can be better indicated in graphs than with notes. Fig. 4 shows a hand graph.

But a hand graph would not indicate the musical features such as timbre and vibrato which Western notation also fails to include. Here the use of electronic or automatic transcribers is important. The two reasons for using graphs – the failure of the Western notation system as a descriptive tool, and the ease of adapting graphic techniques to automatic transcription – should not be confused. The second reason is responsible for a long, and it by far not yet complete, history of technical development to which ethnomusicologists, psychologists, physicists. and engineers in several countries have contributed.

Among the first attempts to transcribe with electronic apparatus was that of Milton Metfessel (1928), who worked in the 1920's in association with the famed music psychologist Carl Seashore. By using a stroboscope which indicated vibration rates and photographing the oscillation of the pitch against a control frequency, and finally superimposing a graph against the photograph, he was able, by calculating each pitch, to arrive at a sort of graph notation. The method was complicated and expensive, and except for the work of a few scholars, it went largely unnoticed. Another attempt, using an oscillograph which eliminated the need for detailed calculations, was made by Juichi Obata and Ryuji Kobayashi (1937).According to Dahlback (1958: 8), this apparatus registered pitch, rhythm, and intensity, but was unsatisfactory for lower frequencies.

In the case of Metfessel and some other attempts, the automatic transcription device was invented not for the purpose of transcribing per se, but for studying some special aspect of music performance. Metfessel wished to show significant differences between Negro and white singers, an idea also pursued in a project initiated by Fritz Bose (1952).In the latter, an attempt was made to measure differences in vocal tone color with the use of apparatus similar to that used by Obata and Kobayashi (1937).Bose (1952) describes his apparatus in some detail. His device, evidently a rather primitive one, does not seem to have played an important role in the later and more successful developments in machine transcription.

The two most important developments in automatic transcription during the 1950's were those resulting from the work of Charles Seeger and of the Norwegian Folk Music Institute, Oslo, under the 1eadership of Olav Gurvin. Seeger, who has pioneered American work in this area since Metfessel, began his attempts in the 1930's, and in 1950 he presented a paper describing a rudimentary "instantaneous music notator" (Seeger 1951). He describes the structure of the apparatus, which is based on the principle of a frequency net provided by an oscillograph. According to Dahlback (1958:7), Seeger's work is based on principles similar to those used by Obata and Kobayashi. The recorder produces separate curves for frequencies and amplitude (melody and rhythm).The recorder in Seeger's first model was a Brush Development Corp. Double Channel Magnetic Oscillograph, BL – 202. The pen of the oscillograph moved up to 120 cycles per second. This model, which could reproduce only whistling, served as the basis of a more advanced type: the addition of filters has made possible the transcriptions of singing as well. According to Seeger, his device, later named the Melograph, is superior to the Norwegian apparatus discussed below in its "directness, speed, and simplicity of handling" (Seeger 1953: 63). But according to another statement of his, the device awaits various developments. Its accuracy and discrimination are much greater than those of the typical Western – oriented ear, and its recording of material on graphs allows much greater accuracy than does Western notation.

Indeed, the Seeger frequency analyzer has "a top discrimination of about 1/14 tones" (Seeger 1958a:188 – 89). Rhythm and tempo can be reproduced in ways which also show changes with a margin of error of only 1/100. Accentuation can, on the grarh, be indicated for the time being only by "manual superscription of notational symbols, such as, for example, of meters, bars, etc. But it can show, with surprising accuracy, the fluctuations of a basic pulse so symbolized" (Seeger 1958a: 188-89).

The Norwegian method is described by Dahlback (1958:7-17) in what is so far the most detailed report on a project involving machine transcription. The project was an attempt to study the singing style of 125 Norwegian folk singers. For this chapter, the equipment used in the project is of greatest interest; but the importance of the project in the history of ethnomusicology should be emphasized as well, for it is unique in its detailed coverage of personal singing style and in its dependence on automatic transcription apparatus. The Norwegian device consists of a double cathode ray tube, the lower ray of which measures frequencies and the upper, amplitude. The series is projected on a screen, which has logarhythmic gradations to approximate the tempered scale, and is then filmed. A period meter marks seconds. Three filters allowed the fundamentals to pass but cut out other frequencies. It is the filtering system of the Norwegian method which enables it – in contrast to the early Seeger Melograph – to record singing as well as whistling. And it is the problem of distinguishing between fundamentals and other frequencies (overtones) on a machine which is the main stumbling block in the development of really practical automatic transcription. According to Seeger (1960:42), the superiority of the Norwegian device consists largely of the elaboration of the filter system. Fig. 5 is a sample of automatic transcription made by the Norwegian method.

In spite of the very considerable accomplishments made to date, the field of automatic transcription must be said to be in its infancy, awaiting great achievements which are bound to come in the near future. Ethnomusicologists have reacted variously to these developments. Seeger hails them as bringing about a revolution in musicology, saying that «from now on, field collection and study of music of whatever area, occidental or oriental, and of whatever idiom of primitive, folk, popular, or fine art cannot afford to ignore the means and methods of the work outlined" (Seeger 1960:42).Herzog, while admitting the importance of the Seeger Melograph, writes in a somewhat less enthusiastic vein: «But the profusion of detailed visual data will have new problems of their own...and all this will have to be re – translated into musical reality and musical sense" (Herzog 1957: 73).He also questions the utility of graphs in publications which are intended to convey information about non – Western and folk music to readers who are members of a public acquainted with Western notation but who would balk at the idea of reading graphs. Kunst (1959:38) is even less optimistic: It is possible, by applying a mechanical – visual method of sound – registration... tocarry the exactitude of the ~anscription to a point where one cannot see the wood for the trees, so that the structure of the piece transcribed has got completely out of hand. In my own view, the transcription by ear, in European notation, as nearly exact as possible, combined with the measurement of the actually used intervals, is nearly always sufficient for ethnomusicological purposes.

The «Committee of Experts" of the International Folk Music Council took a decisive stand in 1950 in favor of Western notation: The deficiencies which this system presents for the notation of folk music can be overcome by the use of supplementary signs. This is all the more necessary because a notation tending to mathematical exactitude must necessarily depend on physical principles and would therefore entail the use of signs intelligible only to the specialists (International Folk Music Council 1952:1).

The Future of Transcription

Speaking very broadly, one cannot assert that the automatic transcription devices have made a great or obvious impression on the ethnomusicological literature of the 1950's. Except for the Dahlback study and Seeger's papers, which are primarily descriptions of apparatus, there have been few publications based on these devices and techniques. It is possible that the developments in automatic transcription have had a negative effect on the publication of transcriptions, for there has been, since about 1950, a decrease in the amount of printed music published by ethnomusicologists in their research papers. Possibly this is also due to the increased expense of printing music, but there is at least a possibility that the degree of perfection in transcription which is promised by the inventors of automatic devices has discouraged scholars from making transcriptions by ear. Such discouragement is not warranted. To be sure, the automatic devices cannot be ignored; they must be used whenever and wherever possible. The Seeger Melograph promises eventually to be only moderately expensive, so that many institutions could obtain it when it is generally available.

Neverthless, at the time of writing only one institution in the U.S.A. possesses the Melograph. Seeger himself, the most outspoken champion of machine transcription, recommends that the two kinds of notation – conventional and graph (by which he presumably means both hand-graphs and automatic graphs) – be used concurrently for the foreseeable future (Seeger 1958a: 188).The value of transcribing as a Iearning device has been mentioned above, and its importance in this function should be stressed in view of the possibility that students of the future wiII rely on the relative ease of the automatic approach. In a sense, transcription by ear amounts to careful listening which is organized so that various aspects of a musical style can be perceived in some kind of order. Listening to a piece without the aid of transcribing it is, in a sense, like hearing a lecture without taking notes – something which has its values but which results in a more general, superficial impression than does the intensive listening with the help of paper and pencil. Thus we can, for all times, recommend the use of the long established custom of transcribing by ear as a method for students to absorb styles of music even if they will later transcribe the same pieces with automatic devices. Transcribing by ear does not, of course, preclude using a hand–graphing technique rather than that of conventional notation. But few publications of the past have made use of hand graphs to a significant degree. Some of those which have done so have used the graph not as a device to inject greater accuracy and greater detail into the transcription but in order to reduce to its essentials and thus to simplify the picture gained from the use of conventional notation. (See Densmore 1918:519 as an example) Besides the two main purposes of transcription, as a way of putting music down on paper for facilitating analysis and as a way of enhancing the information gained from listening, we must take into consideration a third purpose, the presentation of material in published form for consumption by the non-specialist. This is perhaps the most problematic of the tasks faced by the transcriber. In the early days of ethnomusicology, transcription was the main vehicle for presenting the music of the world's cultures to the musician, the music historian, and the music – loving but not ethnomusicologically specialized anthropologist, folklorist, linguist, psychologist, and interested layman. Today, much of this function is taken over by good commercial recordings. The layman can satisfy his interests through such recordings much more easily than by laboriously reading notations which do not, after all, reproduce some of the most obvious features of the sound such as tone color (though these can sometimes be described in words). But the layman or the musician who wishes a closer acquaintance than recordings can afford could gain great insight from a notated transcription. He can analyze the material from viewpoints which cannot be accommodated by listening. Thus the publication of transcriptions in Western notation is of use here. There is, moreover, considerable demand for published notations of a combined prescriptive and descriptive character, i.e., collections of folk songs which can be both analyzed and sung. Here the ethnomusicologist can render considerable service, for his transcriptions, arrived at by ear and limited in accuracy and discrimination, are still of vastly greater value than the notations of certain laymen who have no interest in reproducing a song as it was collected from an informant. Regarding transcription as a whole, we may, then, conclude by saying that automatic transcription is ideal, but that it does not allow the student to make the detailed personal discovery of a music which can come only from transcription by ear. It also does not provide the non specialist with material he can readily absorb; this applies, of course, to the specialist as well, but it is up to the professional ethnomusicologist to train himself rapidly to absorb the information from automatic graphs. Transcription by ear can also be done with the use of the hand graph, but no graph system is as elaborate or as widely understood as the Western notation system; thus the use of hand graphs can be recommended only for special problems. Western notation, on the other hand, incorporates sortle of the characteristics of Western cultivated music and tends to accommodate the transcriber's subjectivity which is usually rooted in Western cultivated styles. But Western notation can be modified and, because of the facility with which it can be used, it offers the most practical method of presenting new musical data in visual form. It forms the best basis for analysis and description of music.

Each ethnomusicologist should be well versed in the art of transcription by ear, but he should be aware of, and make clear in his publications, the limitations of this technique. And he should acquaint himself also with the apparatus and techniques of automatic transcription which, in the decades to come, will have to become increasingly prominent in his work.

Bibliography

Abraham, Otto, and Erich M. Von Hombostel (1909-10). "Vorschlage für die Transkription exotischer Melodien”, Sammelbünde der Internationalen Musikgesellschaft 11: 1-25.

Bartók, Béla, and Albert B. Lord (1951). Serbo-Croatian Folk Songs. New York: Columbia University Press.

Bose, Fritz (1952). "Messbare Rassenunterschiede in der Musik”, Homo 2, no.4.

______. (1952a). "Ein Hilfsmittel zur Bestimmung der Schrittgrosse beliebiger Intervalle”, Musikforschung 5:205-208.

Brandel, Rose (1962). The Music of Central Africa. The Hague: M. Nijhoff.

Dahlback, Karl (1958). New Methods in Vocal Folk Music Research. Oslo: Oslo University Press.

Densmore, Frances (1910). Chippewa Music. Washington: Smithsonian Institution (Bulletin 45 of the Bureau of American Ethnology).

(1918).Teton Sioux Music. Washington: Smithsonian Institution (Bulletin 61 of the Bureau of American Ethnology).

(1939). Nootka and Quileute Music. Washington: Smithsonian Institution (Bulletin 124 of the Bureau of American Ethnology).

Estreicher, Zygmunt (1951). Une technique de transcription de la musique exotique.

Neuchatel: Bibliotheques et Museés de la Ville de Neuchatel. (See McConester 1960 for translation in summary form.)

Gurvin, Olav (1.953)."Photography as an aid to folk music research”, Norveg 3: 181-196.

Herzog, George (1957). "Music at the fifth international Congress of anthropological and ethnological sciences,» Journal of the International Folk Music Council 9:71 – 73.

International Music Council (1952). “Notation of Folk Music”, Recommendations of the Committee of Experts. Geneva.

Kunst, Jaap (1959). Ethnomusicology, 3rd edition. The Hague: M. Nijhoff. (Pp. 7-11 and 37-46 are suggested reading.)

McConester, Roxane (1960). "A transcription technique used by Zygmunt Estreicher”, Ethnomusicology 4: 1.29-132.

Metfessel, Milton E. (1928). Phonophotography in Folk Music. Chapel Hill: University of North Carolina Press.

Obata, Juichi, and Ryuji Kobayashi (1937). "A direct–reading pitch recorder and its application to music and speech," Journal of the Acoustical Society of America 9:156-161.

______. (1938)." An apparatus for direct-recording the pitch and intensity of sound," Journal of the Acoustical Society of America 10:147-149.

Sachs, Curt (1962). The Wellsprings of Music. The Hague: M. Nijhoff. Pp. 20-33 are suggested reading for this chapter.

Seeger, Charles (1951). "An instantaneous music notator”, Journal of the International Folk Music Council 3: 103 – 106.

______. (1953)."Toward a universal music sound-writing for musicology," Journal of the International Folk Music Council 5:63–66.

______. (1958). "Singing style," Western Folklore 17:3-11.

______. (1958a)."Prescriptive and descriptive music writing," Musical Quarterly 44:184-195.Bruno Nettl - Theory and Method in Ethnomusicology 78

______. (1960). Review of Dahlback (1959) in Ethnomusicology 4:41-42.Bruno Nettl - Theory and Method in Ethnomusicology 79

~0~