And, more specifically: may (or 'should') performers play the bass notes and/or harmonic improvisations with shorter value than notated, punctuating the continuo bass with rests?

  • A practical and personal approach: introduction
  • What, then shall we play?
  • Two additional examples: Bach's arrangement of a Vivaldi recitative, and a concerto of his own
  • Resources and further reading, and practical influences
  • Additional thoughts: Empiricism vs Art (and Faith)
  • Conclusions

A practical and personal approach: introduction

Here is a summary set of principles, an approach compiled by an 'early-keyboard' specialist. I worked from the sources presented by Peter Williams (especially) [1969], and the article in The New Grove Dictionary of Music and Musicians [2001], and a book by Laurence Dreyfus [1987], and articles by Arthur Mendel [1950] and Luigi Tagliavini [1986]. Bibliographic details are below.

These writers have organized the primary (17th to 19th century) sources, and brought together evidence for the existence of "shortened accompaniment." The convention existed, and Bach almost surely knew about it. The evidence is found both in treatises (prescriptive and descriptive commentaries) and in the music itself.

Among these I am most convinced overall by Williams' discussions, both in his own article and in the New Grove entry. I admire his balanced presentation of the sources, his empirical neutrality, his emphasis on practicality, the amount of credence he gives to CPE Bach as a witness (and Telemann, Heinichen, etc.), and his recognition of any differences between sacred and secular music. He is also more careful than Dreyfus and Mendel not to conflate the sources that say the bass itself should be released, with the sources that say only the chords are released while the bass continues to sound. And, he goes much more thoroughly into relevant sources on organ registration, not focusing simply on note lengths, but on overall sound.

Another clincher, for me, isn't even a prominent part of these modern articles. It's additional evidence that (in my opinion) makes all of this obvious and almost a non-issue. And that is: these German composers--including Bach and the 18th century writers cited in these modern articles--grew up creating church music that was wholly or mostly improvised as part of the regular job. And furthermore, some of the music they did write down was in keyboard tablature: a shorthand where rests and note-lengths are not indicated at all, but only the beginnings of the notes. (Examples: some of Bach's Orgelbüchlein and other keyboard music, where he used tablature either to save space or for drafts; most of Buxtehude's keyboard music; Hassler's 31 variations "Ich gieng einmal spatieren"; and a tablature by J & C Plotz.) Such tablature did not prescribe octave choices, articulation, or tempo either. The player's musicianship and sense of voice-leading, along with experience and practical listening, told him how to articulate and release the notes at appropriate times. And those who didn't improvise well enough were derided as merely "paper organists", not fully qualified for their jobs.

As recitative notation and other accompanimental practice grew from this mostly improvisational art, it was simply a swap of one shorthand for another without getting much more specific on paper. The organist is given only the barest essentials, and all else is experience and taste (plus oral instruction). Any expectation that the basso continuo notes would have any prescribed or immutable lengths is a modern one looking backward, not one growing out of that type of normal church music practice. Listening, and all-around musicianship thinking as a trained 18th century German improviser/composer, takes precedence ahead of notation.

The following, then, is my own list of principles, as a professional performer of this repertoire. It is the way I choose to play continuo in plain recitative in the Bach vocal works: both from the set of sources noted above, and (at least as importantly) from practical experience in the music. That is, to me, the description below is a reasonable approach balanced between research and experience. I try to the best of my ability to play it "Bach's way" as far as can be determined. There is no way to know 'for sure' (see also the concluding comments below), but I feel this method works very well in 'real-life' situations.

What, then shall we play?

- Clarity of the words is paramount: both from the singer (presenting the text as declamatory, heightened "speech" in notes, freely flowing in speech-rhythms), and from the accompanists (staying out of the way on anything that would obscure the direct, clear, straightforward presentation of that text).

- That is, "less is more." The accompaniment's main job is to punctuate what the singer is already doing, letting the singer lead, and merely highlighting the occasional surprises (harmonic shocks, and the most important words, and bass motion)...and the rest of the time staying out of the way, being deliberately uninteresting, because it would detract from the singer and the words. Never should the accompaniment be too loud, or too fussy, or too busy. Changes of registration during a recitative are both unnecessary and distracting, drawing attention away from the singing.

- Lightly touching the harmonies is the norm: play them briefly, then get off them, to let the singer's words emerge most clearly. All that is necessary is to establish (very briefly) the harmonic context, and then get out of the way. This is true whether it's organ or harpsichord playing the keyboard part.

- Only very occasionally (usually not more than two or three times in an entire recitative) should special points be made: making certain progressions or strokes more emphatic/strong than the norm, or (sometimes) more gently played than the norm. These decisions always arise from the music itself, not imposed on it by the player: if the composer did something surprising with harmony or the words, bring it out. But the default is the overall neutrality: the same number of notes in each chord, the same light neutral touch, except when making a rare point one way or another, using more or fewer notes, or shorter or longer strokes.

- In preparation: a good way to think of these keyboard 'strokes,' and get them to be a decent length, is simply to make a quick sweep of the hand (like a conductor) while speaking a word. The natural speed of the hand (slow-fast-slow) during the stroke shows the general dynamic profile, and the the length of the spoken word determines a good length. Also, it is noticed that all these natural strokes have some degree of 'roundness' to them; nothing terribly sharp, and nothing artificially drawn out either.

- If there are other bass-line player(s), or if organ pedal is available, it might (or might not) sustain the bass line itself somewhat longer than the keyboard player is doing; it's independent. It can sometimes play as short as the keyboard chords, especially where the point of the dramatic text is urgency; or it might play somewhat longer, emphasizing that it's a line. Everything comes from the context of the words, and from the position of this recitative within the whole piece! Where does it come from, and where is it going? That is the determination of how long, or how short, to play. There can be crescendos or decrescendos during the notes, if that is what is called for by the directionality of the text, and the singer's presentation...moving the drama forward, according to the meaning.

- The point of the bass line is to provide enough context for the singer and listeners, but not to overwhelm anything, or to seem unnaturally sustained longer than necessary (as in, sitting there sounding bored). No single note, or group of notes, should ever exceed the length of a normal human breath...whether it's played by melodic instruments, or by organ pedals. A good guideline is: assume the listeners can breathe at the same places the bass line does, and help them to do so!...never do anything in the bass line (or the organ) that would make it difficult for the listeners or the singer to breathe.

- If the figures change during a long or tied bass note (as it appears on the page), or over a rest (yes, it happens!), the keyboard player usually doesn't need to do anything about it. Check to be sure, but it's probably simply describing what the singer's notes are already doing, a new harmony over the previously played bass note. Just watch it go by. All it is is a visual cue of the composition's structure, not something to play. In the exceptional cases where it really is some new harmony not made clear in the vocal part, do consider playing it...lightly!

- The bass-line player(s) also don't need to do anything special with that situation (figures changing over a note or a rest). They might already be holding or swelling their note for other reasons (to fit the dramatic declamation), or they might have already released it; these things come easily by context and experience. [That assumes, of course, that the players and the keyboardist understand the text that is being sung, which is of course a requirement! How else can they figure out what to do with it?]

- Occasionally long notes ("full value" according to the notation!) can also be effective, especially where the musical meaning to be brought out is exceptional gentleness. Since the default sound is a somewhat shorter rendition than this, a long note here and there can sound very special. (That is, the default sound with rests allows this "wet" end of the range to exist as well, as a contrast.)

- None of the accompaniment, from anybody, should ever sound dogmatic or rule-bound. It should all seem like a natural reaction to what the singer is doing, in the moment, and not call attention to itself. The correct weight of every stroke should arise naturally from the way the passage is composed, subtly varying (naturally) as the sentences and paragraphs progress forward in the singer's delivery.

- Yes, this all involves split-second reaction to what the singer is doing: playing by the ears much more than playing from the paper. The paper simply tells us when the notes start, and suggests the relationships among them; everything else is determined by the context of this particular performance. Obviously there are plenty of factors to consider: the acoustics of the building, the singer's confidence, the singer's strength, the singer's dynamic range, the loudness of the organ stop used, the speed with which it (or the harpsichord) speaks, the confidence of the bass-line player(s) [if any], and the poetic clarity (or denseness) of the singer's text to begin with. As noted above, clarity of the text is paramount. A good general rule is: the accompaniment should be as simple as possible so as not to distract from that conveyance of meaning by the singer.

- Overall: the accompaniment is there to heighten the intensity of the singer's delivery...which itself must already be committed and intense, like the spoken word that has already crossed into the realm of pitches, because a merely spoken delivery is not enough to contain it! The point is for everybody to put the message across as vividly as possible: the singer as (by far) the most important, supported by accompaniment that punctuates and affirms it. The accompaniment is like an eager group sympathetically nodding along with the points the singer is making, interjecting the equivalent of "amen!" "preach it, brother!" to heighten what is being said/sung.

- There must always be respect for the gravity and seriousness of the text (especially in church music). But, that is not the same thing as a uniform solemnity or excessive 'reverence' (an awe-struck "I am not worthy" self-denial, just from being in a church). Indeed, this respect for the music's seriousness may call for a dramatic approach, particularly where the singer is making a surprising or vehement point. 'Seriousness' is not a demand that everyone's emotions should be flattened to null.

- At the same time, the music (even at its most serious) is play. If it doesn't sound like play, if it's too inflexible, if the unpredictable is never to be trusted, I feel it stifles the numinous magic out of it! But, that's coming dangerously close to an aesthetic or theological judgment....

I state these principles, fully aware that in recordings they are not followed to the extent I suggest. Sometimes, yes; but not often. Most of the delivery I hear is too polite, subdued, un-dynamic, perhaps trying to be too 'musical' with it, or too limited by needing 'permission' from treatises, or whatever. Or, perhaps, it might seem 'overdone' when the same performance is heard a dozen times on recording playback. Perhaps not. Perhaps listeners appreciate and resonate with the added clarity that a closely reactive delivery offers!

The reason to do all this recitative delivery with such intensity is not because any treatise says so (giving permission), but because it is basic human communication: putting across a message vividly. Everything is context. Everything arises directly from the words and music, not from the performers 'adding' anything to it (other than feeling it as deeply as possible, and then reacting naturally).

Two additional examples: Bach's arrangement of a Vivaldi recitative, and a concerto of his own

Bach's solo organ concerto in C major, BWV 594, is a transcription of an orchestral concerto by Vivaldi: D major "Il grosso Mogul," from c1716. The second movement is a recitative for violin, a very elaborate one. In Vivaldi's version the bass part is [conventionally] notated in long notes, tied across the barlines, no rests...the same way that most of the 'secco' recitatives in Bach cantatas look on the page. But, in Bach's transcription for organ here, all those accompanying notes are notated as short! The pedal plays the bass note and the left hand plays a chord (a simple continuo realization), and then they have rests until the next change of harmony which is often two or even three bars later. The melody spins on its own, with no harmonic "support" at all; merely the brief harmonic pillars spaced wherever they are, and rests everywhere else.

Then, at bar 20, there is a long dominant pedal point that is tied until bar 23, at which point it resolves to the tonic and that's the end of the piece. And at that same bar 20 where the pedal starts to hold the long note, Bach introduces in the left hand a new melodic bit that intertwines with the main melody, instead of playing continuo chords. So, to sum up, Bach spends the first 19.5 bars with a 'secco' sound and then changes to an 'accompagnato' texture for the big finish. (There's our friend Bach again, adding "improvements" to whatever he can get his hands on....)

An adequate scan of a score is here:

To draw a tentative conclusion from this example: it suggests to me that Bach as a player relished this variety and flexibility. He (presumably) did so with the faith that it would not ruin the composition...and might even improve it. Obviously, to Bach, the shortening of the notes was an option and he chose to exercise it here.

Taking the next step, I believe that players of Bach's own music may use a similar range of techniques in the realization of his recitatives, even when it merely looks like long tied notes on the page. Following Bach's example here, I as a player recognize that the "long notes" of a typical continuo line in recitative are merely a convention, a shorthand, for "do something musically intelligent and tasteful." That is, the composer trusts me to do my job, without spelling out everything for me. It is the performer's duty to give the music (any music) a sensitive rendition...and it might not be the same all the way through a piece, or from performance to performance!

Another example: the harpsichord cadenza at the end of the Concerto BWV 1044 simply has a right-hand melody over a very long pedal point bass note--which dies out quickly. I believe this too is an invitation to improvise or work out an interesting/dramatic left hand part to go with the right hand part, occasionally refreshing the bass note but also adding rhythmic interest and bits of counterpoint. It fits the dramatic build-up and energy of the concerto to that point--where the simply notated version is (arguably) a "let-down" before the concluding tutti. Accordingly, I have played this concerto in June 2005 filling out that cadenza with a left-hand realization, treating the whole cadenza like a speech in recitative! This concerto is itself already an elaboration of the Prelude and Fugue BWV 894, and my attempt here has been to follow that same direction of elaboration.

With the conventionally sketchy notation of plain recitative, the composer has simply "decided not to decide," not restricting the player's practical options any more than necessary.

An obvious objection is: there are no words here in BWV 594 or 1044, so, what 'meaning' is the continuo line responding to? Are vocal and instrumental compositions subject to the same precepts? How do we know? Empirically it can't be proven that instrumental and vocal works must be treated the same (or differently!)...Bach did not provide a statement that is normative for all performances of his music, present and future. Empiricism doesn't solve this, and in good faith we must use other methods of determination.

Resources and further reading, and practical influences

  • Peter Williams, "Basso Continuo on the Organ", Music and Letters #50 [1969], pp136-54 and 230-45
  • Peter Williams and David Ledbetter, "Continuo", The New Grove Dictionary of Music and Musicians [2001]
  • Laurence Dreyfus, Bach's Continuo Group: Players and Practices in his Vocal Works [1987], Harvard University Press
  • Arthur Mendel, "On the Keyboard Accompaniments to Bach's Leipzig Church Music", Musical Quarterly xxxvi [1950], pp339-62
  • The bibliographies of the above articles
  • Luigi Ferdinando Tagliavini, "Bach's Organ Transcription of Vivaldi's 'Grosso Mogul' Concerto", 1984; republished in the author's revision and translated by Bruce C Macintyre, in J S Bach as Organist: His Instruments, Music, and Performance Practices (ed. George Stauffer and Ernest May), Indiana University Press, 1986; reprinted 2000.
  • Scores...

My performance experience as a continuo player (both on harpsichord and organ) includes both Bach passions, the Mass in B Minor, numerous cantatas, and parts of the Christmas Oratorio; plus vocal works by Handel, Purcell, Charpentier, Vivaldi, and others; plus many dozens of instrumental works, including Bach's; plus ten years of professional service as a church organist. I am not arguing that the same style should be used in all these 17th and 18th century works, by any means. It is merely an observation that I have been able to experiment in a range of music, picking up practical 'tricks of the trade' from gigs.

To anyone interested in a scholarly presentation, a firm place to begin, I would urge consultation of the New Grove article "Continuo" (and its predecessor in the 1980s edition of New Grove), plus all of Dreyfus' book, not only his chapter "The Accompaniment of Recitatives."

In addition to the endorsement by the musicological community (and its citation in other reference books), an informal advocate for Dreyfus' book is Andrew Lewis, in this online discussion of Cantata #103:

Empiricism vs Art (and faith)

What is my own stake, and possible bias, in all this? Well, for better or worse, I have one foot firmly in each "camp." Perhaps that's a liability, or perhaps it's an asset for balance.

I have degrees in mathematics and historical musicology: both are fields that are ruled by logical empiricism as a primary method of determining truth. (Meanwhile, that method is not the only respected paradigm of truth, especially in the social sciences. And music resists the hard-science approach anyway; it has plenty to do with subjective perception and social values, etc., etc.) Empiricism doesn't always deliver neat "yes" or "no" answers; there are sometimes third or fourth logical values such as "insufficient data to reach a conclusion" and "this incongruous data type cannot even be tested by the given method." And, some empirical results are true or false for only a limited amount of time.... Nevertheless, empirical projects pay my bills nicely, and those other possible values after "true" and "false" keep us on our toes.

But I also have degrees in performance (and historical performance practice) on harpsichord, organ, and fortepiano: the artistic and practical side apart from empiricism. In a field of art, empiricism can tell us a great deal, but I believe it can never tell us everything; opinions of respected musicians, and considerations of taste, do still matter. I do trust my own experience as a composer, improviser, and performer, as valid information to be used alongside the empiricism. And I trust the taste and conclusions of my peers in the field of performance.

And, even my empirically-based projects (outside music) have an artistic flair to them, because they are designed for understanding and use by untrained people. Such a blend is necessary in such projects. Empiricism by itself is not enough for me, either in design or execution of a project. Not everything can be quantized in the ways a pure empiricist would demand, or try to demand. When something is to be used or contemplated by ordinary people with no specialized preparation, it must be as immediately clear as possible, even if that compromises some empirical "purity." Sociologists and psychologists know very well that people and groups do not receive information in a purely rational manner. Communication and negotiation are not "zero-sum" games whenever real people are involved. The artistic performer who wishes to communicate must understand that, and allow for it! (Yes, I am married to a social scientist and we have lively discussions.)

So, there is a lot to balance here. If there is any chance of error, I believe we should err on the side of artistic conviction and strong communication: that is, putting the music across as clearly and vividly as we are able to. I believe most listeners to music are not there to receive empirical facts, but to receive a musical experience that moves them.

Empirically, the arguments in favor of shortened accompaniment are inconclusive. We cannot "prove" that Bach absolutely expected or demanded that his musicians apply that principle to his music. But, crucially, it also cannot be proven that he forbade it, or even disapproved at all. The empirical sword, if it cuts at all, must be allowed to cut both ways: and not only used to forbid practical imagination!

Bach was a practical musician, working under tight deadlines (especially while composing cantatas in Leipzig). There is no way to notate all details of a composition, all practical considerations for performance, enough to fully satisfy an empiricist; music does not work in that manner, and especially when the composer is in a hurry. (That is, performers must always go beyond the page, not merely 'following instructions' that tell us exactly how to do our jobs.)

Therefore, through no fault of Bach's, some empiricists might never be convinced that a practice of shortened accompaniment ever existed at all, for Bach, in his sacred works. (Or, similarly, the existence of a consistent and practical keyboard tuning across all or most of Bach's career! And such a tuning on organs and harpsichords might also tell us some important things about emphasis through articulation and timing, as to the relative crunchiness or calmness of the harmonies....)

But: neither can an empiricist prove that it did not exist. Again, the empirical sword, if it cuts at all, must be allowed to cut both ways: and not only used to forbid practical imagination! Additional methods of inquiry must also be brought to the table: especially direct performance experimentation with the music of Bach's own repertoire to see what works well in practice, according to any aesthetic clues and notational clues we can find about his milieu and his performances/teaching.

So, what is a practical musician to do? Play it musically, and use the techniques that Bach clearly knew himself, as far as can be determined. I am especially swayed by the example of BWV 594, where we see him interpreting another master's music by shortening the bass notes of recitative. For Bach, in c1720, that was one option that he obviously knew about and considered musically valid, for at least this example of plain ('secco') recitative.

Is it normative for his own music? We don't know. Is it something Bach might have done in the performance of (at least some of) his own music? I'd say 'probably'; at least, we cannot prove that he did not do it.

As pointed out by Williams and Ledbetter, J. F. Daube [1756] described three playing styles for continuo keyboard: "1) simple, with closely connected right hand chords and all chord factors present; 2) natural, which adjusts this strict manner to be more melodious and like a composition; and 3) complex, similar to the elaborated accompaniment given by Heinichen (1728), or the concertante patterns of Mattheson's Exemplarische Organisten-Probe (1719). Daube noted that J. S. Bach excelled at the third style (and this is probably more in regard to chamber music performance, with harpsichord, than church cantatas).

The simple style was the common teaching method of the 18th century, i.e. for beginners getting the rudiments of harmony and voice-leading, from which one could branch out into idiomatic organ or harpsichord accompaniment (and composition). In practice good professional harpsichordists probably played, as Quantz (1752) recommended, a nuanced accompaniment with texture varied to enhance the musical effect of the moment. This is probably what Daube had in mind for his natural style, a style which according to Löhlein (1765) was only for those with experience of composition and the sensitivity to do it properly. Löhlein also tells us that by his time the complex manner had gone out of fashion."

That suggests: good keyboard players today, for 'authenticity' in this music, should also strive for the 'natural' and 'complex' styles known to Bach and his colleagues. At least, I know I do! What better model is there than trying to play as Bach himself reportedly did, and as illustrated in his keyboard and ensemble compositions?

Is there a distinction between secular and sacred music? Yes, such a distinction exists; but is it a line that absolutely calls for separate practices, no 'mixing' whatsoever? It can't be proven...especially for a composer who freely used and reused his own music in both sacred and secular contexts. He 'converted' compositions from sacred to secular, and vice versa, whenever a need arose or whenever it was a convenient solution to a problem.


So, to sum up: it has not been proven that Bach ever explicitly forbade the application of this improvisational/notational convention to his sacred music. All we can do is to have faith that he expected musicians to do something tasteful and intelligent. The empirical research shows us what the available techniques and expectations probably were, as far as can be determined. Bach knew this conventional evidenced both in the BWV 594 example, and in the material by Friedrich Niedt that Bach used in his own teaching. And according to the extant evidence, he didn't explicitly forbid or condemn it.

As I noted above, I believe this validates a performer's choice today to allow it as a possible option in Bach's sacred music. It is not something to do with a mindless and literalistic regularity, but something to allow if the musical context seems to warrant it...which, in my musical opinion, is most of the time! It's an improvisational art built on extremely sketchy notation: notation that wasn't and didn't need to be specific about note-lengths. It's not any more specific or restrictive than guitar chords (labelled harmonies, and/or tablatures) are in single-line songbooks today, as to telling the accompanist(s) what lengths and emphases of notes are appropriate to play. Let the ear, taste, and experience do their jobs!

That is, as both an empiricist and a performer I am convinced strongly enough. Where it clarifies the musical message, shortened accompaniment is allowable as an option. It is up to the performer's good taste and experience to employ it intelligently and effectively. That is, it is a valid technique to be used at some level greater than or equal to zero. A performer cannot be faulted for not using it at every opportunity! But, neither can a performer be faulted for using it as a technique, if the goal is to put the music across vividly and with conviction. (The performer might not be enterprising enough in this and other areas of expression, leading to a bland delivery that fails to communicate; but that's a different and much broader issue.)

If performers 'must' be restricted to merely following some supposedly complete set of instructions, I believe the burden of proof would be on those who would impose such confinements. Meanwhile, we serious performers should be allowed to do our jobs according to our training, experience, taste, and convictions! This is not 'anything goes' permissiveness, but merely the acceptance of a possible option that is not fully prescribed in the score, delivered with the conviction that Bach did likewise.

That is still not to say that everybody will like this particular technique, or agree on the appropriateness. Because the empirical findings (alone) are inconclusive, an empiricist might be displeased by a performance on the basis of this issue. Did Bach use shortened accompaniment occasionally in performance of his own sacred works? The mathematical probability is greater than zero and less than 100%; perhaps that 'less than 100%' is too disturbing for an empirical skeptic. But: one wonders what such a supposedly pure empiricist (or over into "cynic") is hoping to get out of the music, at all. Music is an art, not an absolutely empirical venture. "Gott sei Dank!"

Bach's cantatas were written by a man of faith, to edify and instruct people of faith. And performers charged with their presentation also have to have some faith and conviction in the task: giving our best guesses and efforts to the goal, and having faith that we're accomplishing something worthwhile. If not, why bother? This is an art about numinous principles and experiences. It's not completely understandable. "Gott sei Dank!"

© Dr Bradley Lehman, Sept 2003 plus some later revisions