Performing Bach's St. Matthew Passion

by Bernard D. Sherman

Reprinted by permission from Early Music America, Fall, 2001.

Other articles by Bernard Sherman from The New York Times, Early Music, and elsewhere are online. (Among them are two on performing the B Minor Mass and Well-Tempered Clavier.) So are excerpts from my book Inside Early Music (Oxford University Press, 1997).

Recording labels and numbers appear in the sidebar accompanying this article, along with a brief introduction to the work.

The St. Matthew Passion is sometimes compared to an opera, and sometimes to a sermon. Neither comparison does justice to the work's complexity and depth. As John Butt writes, "with its unfolding levels of symbolism, theological interpretation, and -- most striking of all -- psychological insight, the Matthew Passion is perhaps the most challenging and ambitious Christian artwork."

Butt implicitly includes Dante and Michelangelo in the comparison. Yet unlike The Divine Comedy or Sistine Chapel ceiling, the St.Matthew Passion lives only when performed by modern musicians. Many performers try to vivify it by using the means and styles of modern performance, but for over 30 years, some have explored performance that, in various ways, try to revive performing practices of Bach's day. Critics often argue generally that true historical re-creation is impossible; but more to the point, they sometimes charge that historical performers can't plumb the depths of the St. Matthew Passion. Ever since Paul Henry Lang lambasted the pioneering 1970 recording by Nikolaus Harnoncourt, critics have complained that period-instrument performers trivialize the work.

I disagree. The difference of opinion is, of course, partly a matter of taste; to my ears, Otto Klemperer's St. Matthew Passion sounds far too weighty and slow, so it should be no surprise when his admirers find my own favorites lightweight. All the period-instrument performances use faster tempos, smaller ensembles, and less operatic voices, .

But some of the other differences between the performances contradict stereotypes. The Israeli musicologist Uri Golomb argues that recordings generally contradict the argument that historicist Bach performance is Stravinskian modernism in disguise: light, literalistic, and impersonal. Golomb agrees that this description applies to some early Beethoven, but holds that it does not characterize the most influential early music Bach performers. These performers tend to shape motifs, phrases, and movements in a more rhetorical and inflected way than their mainstream colleagues.

For example, Golomb explains how Sir John Eliot Gardiner's 1988 recording -- through the use of crescendi, diminuendi, and sforzandos -- carefully builds the St. Matthew Passion's opening movement to an overarching climax. Nothing of the sort happens in the 1954 recording by Wilhelm Furtwängler, the Romantic genius who was the supreme conductor of Wagner, Brahms, and Bruckner. In the opening of the St. Matthew Passion, Golomb writes, "Furtwängler hardly alters any parameter, and his performance contains no obvious climax at all."

In the final chorus of Part I the two conductors again contradict the stereotypes. John Eliot Gardiner "often highlights individual words and draws sharp contrasts between sections." By contrast, Golomb writes, "Furtwängler's only modifications in a rather uniform reading are structural: small diminuendi preceding some choral entrances, and a massive ritardando towards the end."

Golomb notes that conductors of Furtwängler's generation often seem to have felt that it was wrong to "tamper" with Bach (exceptions, he adds, include Ralph Vaughan Williams and Willem Mengelberg; and I should note that Furtwängler cut out many sections of the Passion in performance). Early music performers, for reasons that may or may not have to do with period practice, tend to feel freer to shape this music.

Admittedly, many early music recordings show both the advantages and the dangers of such nuance. Consider Gardiner's recording. True, it is powerfully realized, and the cast of soloists is unsurpassed. As always with Gardiner, the choral singing is astonishing, and (as Golomb shows) the interpretation is thoughtfully nuanced. But not all of Gardiner's nuances are appealing, at least to me. The second choir's questions in the opening chorus ("Whom?" "How?" and so forth) are cut very short, an effect not at all suggested by the notation. As John Steane writes, the result is to make the rhythm sound waltz-like. I wonder, too, why Gardiner must take the beautiful chorus of the disciples, "Wo willst du, dass wir dir bereiten" in so fast and clipped a style.

Or consider Frans Brüggen's 1996 recording. It features the most thoroughgoing nuances of dynamics and rhythm that I have heard in this work. Some great moments result, and much of the singing and playing is moving. Yet because of overdone inflection, the overall effect of the performance is a bit fussy.

Another Dutch early music pioneer, Ton Koopman, lays on inflection of a different sort -- that is, ornamentation, mostly in the instrumental, especially continuo, parts, but also in the solo singing. Much of this singing is outstanding. For me, however, the choral singing is not distinguished (I would say the same of Brüggen's chorus), and some of the inflection and ornamentation seems precious or interferes with the expression: it actually can make Bach seem more lightweight.

All my judgments are, of course, debatable. Both Gardiner's and Koopman's recordings of the St. Matthew Passion have won major awards and are regularly chosen as the best available. What's significant is that my reservations are not about impersonality or literalism but about interpretive nuance that feels, to me, excessive or unintegrated.

Not that I dislike interpretive nuance. Consider three recent St. Matthew recordings that, in my idiosyncratic experience, convey the range of this masterpiece especially fully. Insightful nuance lets these versions do justice not only to the gravity of the work but also to its humanity.

Three Great Period-Instrument Versions
Nikolaus Harnoncourt was the first to record the St. Matthew Passion on period instruments; that 1970 attempt sounds immature today. But his third recording of the piece, made in 2000, is focused, accomplished, and expressive. As in much of his Bach, Harnoncourt aims to convey the emotional and rhetorical implications of every detail. He often articulates motifs and dissonant notes emphatically. Not a line or harmonic change goes unprobed. Yet in this recording, he makes his approach work in a more natural, integrated way than he has managed in certain other works. Some of his inflections - e.g., cutting long notes shorter than written (listen, for example, to the opening chorus or to the accompanying strings in "Erbarme dich") - may be debatable from a historical perspective, but they work well here. And the power, sweep, and meaning of the work are nowhere better served. This is an extraordinary St. Matthew Passion and one of Harnoncourt's best recordings.

A number of Harnoncourt's interpretive insights are memorable -- for example, the inwardness of "Komm, seusses Kreuz," with its moderate tempo perhaps illustrating the motion of someone bearing a cross. The soloist in this aria, Dietrich Henschel, sings the same aria very differently in Philippe Herreweghe's second recording , which suggests the influence of Harnoncourt on the newer interpretation. Henschel is, by the way, less successful in some of the other arias, but most of Harnoncourt's soloists respond superbly to his approach. Harnoncourt follows the practice specified in Bach's score of having two sets of soloists; so do Brüggen, Gardiner, Leonhardt, and Jeffrey Thomas. Most other recordings use only four main solo singers, whether for financial or artistic reasons.

Early music conductors shape the Lutheran chorales expressively, but one point on which they differ is how to treat the fermatas written over the end of each line of the hymn text. Harnoncourt holds them each for about three beats; so did Herreweghe when he recorded the work in 1984. I believe that this was probably the historical practice. I like the expressive effect when it's done well, as is it is here. But many scholars, and most early music conductors, have come to the opposite conclusion and move through the fermatas with barely any extra holding.

Herreweghe's 1984 recording is much admired; but his 1998 attempt seems to me better. His orchestra and chorus are more accomplished today (compare the oboe obbligatos) and, more importantly, Herreweghe aims now for more vivid expression of the emotions Bach is trying to illustrate. By his own description (found on a multimedia CD-ROM that comes with the recording) he also aims for greater continuity from beginning to end. The performance has flaws, as all performances do; my favorite aria, "Mache dich, mein Herze, rein," for example, is a bit too fast and its singing is undistinguished (in this aria I prefer Jeffrey Thomas's William Sharp). But overall, Herreweghe's St. Matthew Passion is a powerful response to the work.[Click here to read an earlier review I wrote of it.]

There there's Masaaki Suzuki's 1999 recording. Its strengths and flaws are perhaps most easily explained by comparing it to Herreweghe's 1998 recording, since the differences are evident from the opening chorus.To Herreweghe this huge tableau seems to represent the crowd frantically seeking Jesus. The tempo is on the fast side and the expression intentionally unsettling. Suzuki's tempo is slower and conveys a profound sense of lamentation. But what's most striking is the dramatic entry, a few minutes in, of a third soprano in ripieno singing a Lutheran chorale, "O Innocent Lamb of God." Bach's listeners had heard a version of this chorale in morning services a few hours before hearing the St. Matthew Passion. John Butt notes that its image of Christ as sacrificial lamb was associated with "the Apocalypse, when Christ as a lamb rules the new Jerusalem." In Bach's own performances, as Christoph Wolff has noted, this third soprano ripieno would have been stationed apart from the other choirs, in an organ loft on the eastern wall of the church - and listeners understood that East was the direction of Jerusalem. As the Bach scholar Michael Marissen notes, Suzuki conveys the entry's Apocalyptic implications better than anyone else, partly because Suzuki or his engineers give this line a more unearthly acoustical stage than their other two choirs have.

On the other hand, the choral interjections in "So ist mein Jesus" are more ferocious in Herreweghe's 1998 recording, as are some crowd scenes. Yes, ferocious -- Herreweghe is sometimes criticized for being smooth and controlled to a fault in Bach, but not in this recording. And no one has surpassed the expressive depth he finds in the string lines of the chorus that ends part one.

Herreweghe's is the more dramatic performance, Suzuki's the more meditative. But Suzuki is by no means less involving. The role of the Evangelist illustrates the distinction. Herreweghe's Ian Bostridge, a noted lieder singer, seems to be acting the text out, while Suzuki's lighter-voiced Gerd Türk seems to be recounting a story. But both interpretations are compelling.

I don't want to overstate my theme of drama versus meditation. Herreweghe's Jesus, Franz-Josef Selig, has a powerful voice and gives a moving performance; but Suzuki's Jesus, Peter Kooij gives a portrayal that is particularly human and vulnerable. After the heart-rending passage where Jesus says he is "sorrowful unto death" (his emotion is portrayed in heavy string chords), he sings, "keep watch with me." Kooij turns these last words into a cry of anguish. And Suzuki's soprano soloist, Nancy Argenta, sings in a warmer and more heartfelt way than Herreweghe's singer.

The choice between Herreweghe 1998, Suzuki 1999, and Harnoncourt 2000 is hard, but each of them, I think, rebuts any charges of superficiality that might still be lodged against period-instrument performances of the St. Matthew. They also disprove any charge of uniformity. Each conductor sheds a very different light on this work, and each illuminates different facets. My preference wavers between Suzuki and Harnoncourt, depending on my mood, but some people prefer Herreweghe, and I can understand their choice.

Other Notable St. Matthews

Until the above three recordings were released, my favorite St. Matthew was a flawed but profound one by Gustav Leonhardt. Leonhardt uses a boys' choir, as well as boy soloists for the soprano arias. All the other conductors save the 1970 Harnoncourt use women for the soprano parts (although all the conductors use boys for the soprano in ripieno in the opening chorus). Boys' choirs may suggest extra authenticity, since Bach's church did not allow females to perform. But the claim to authenticity loses something on inspection. Boys' voices "change" earlier today than in the 18th century, for one thing, so modern boy singers are younger than Bach's singers and may sound less mature as a result. Besides, some argue that Bach may often have used mature falsettists, rather than boys, for soprano parts. Nonetheless, Leonhardt's boys bring considerable power to the crowd scenes. His German boys' choir has an earthier, less angelic sound than that cultivated by English boys' choirs, and the style is well suited to Bach.

The recording sometimes suffers from a peculiar inflection that mars Leonhardt's cantatas as well -- in a few places, such as certain chorales, he overdoes each downbeat, an effect one critic compared to a "squeezebox." The boy soprano soloists are more accomplished than you might expect; but in the great aria "Aus liebe," one of them breaks up the melodic line annoyingly, probably at the conductor's insistence. And I have yet to acquire a taste for the countertenor René Jacobs. To me, his portamentos and odd emphases sometimes sound overwrought. But then countertenors often sound that way to me. History aside (and, again, it does support the use of falsettists in Bach), I tend to prefer light-voiced mezzos. Need I add that many listeners have exactly the opposite preference?

Anyway, the strengths of Leonhardt's performance far outweigh the weaknesses. Among these strengths are the baritone soloist Klaus Mertens (who also sings for Koopman) and the Evangelist Christoph Prégardien (also superb in this role in the new Harnoncourt recording). Leonhardt's tempos are more moderate than those in other early music recordings, and help him bring suitable weight to the performance.

I have been unable to get copies of the period instrument versions of the St. Matthew Passion by Paul Goodwin and Hermann Max. Both conductors fit the work onto only two CDs, but I wouldn't assume that they are unduly hasty; Gardiner's performance could also fit on two CDs, and many others come close. Max's version has been highly praised. Jos van Veldhoven's recording, made in concert with a strong cast, is very fine in many ways; perhaps even better is the recently issued recording by Jeffrey Thomas. Thomas's cast, mostly American, may not be as consistently strong as those of my favored three, but its best members, such as William Sharp and Catherine Bott, are among the best on disc. Thomas not only conducts the performance but also sings the part of the Evangelist, quite powerfully. The dual function is no small feat. The recording captures a concert performance from 1996, and that circumstance may contribute some power and spontaneity. Other contributing elements are the outstanding orchestral soloists, and the interpretive insight we have come to expect from Thomas. Note, for example, the hushed singing in "Herr, bin ich's?" by the disciples who have just heard from Jesus that one of them will betray him. On the other hand, the opening chorus is the fastest I have heard, and it may rule this performance out for some listeners. The recorded sound, too, is compromised by the live-concert origins.

With so many fine recordings, does the catalog have enough versions of the St. Matthew Passion? At least one obvious gap exists: by contrast to the Mass in B Minor, no one has released a version of the work with only one singer per choral line, as Joshua Rifkin argues Bach performed the work. In my view, his argument has strong support with respect to the St. Matthew Passion. And several writers have pointed out that a one-per-part approach would reveal theological meanings in the St. Matthew Passion that are lost in choral performance. John Butt notes that when the same man sings the part of Jesus at one moment and joins with his tormentors in the next, it "helps reinforce the typically Lutheran point that we are all to be held responsible for Christ's death, however Christ-like any of us may seem."

Rifkin and Andrew Parrott have led a number of one-per-part performances of the St. Matthew Passion over the last few decades, but never in the studio. Recently, Paul McCreesh became the first conductor given the opportunity to record the piece one-per-part. I await the release with interest.

Of course, it will almost certainly have some flaws. Given the work's emphasis on human frailty, it seems poetic justice that all performances of the work are imperfect. Capturing the multiple facets of the St. Matthew Passion is beyond the ability of any single interpreter. But early music explorations have brought nuances and insights to the work that might enrich future performances by conductors of all stripes.

[2003 Update: McCreesh's St. Matthew Passion turned out to be outstanding; you can add it to my list of essential recordings. If you love this piece, don't miss it.]

Author Credit:
Bernard D. Sherman writes about music for national publications [read samples here] and is the author of Inside Early Music: Conversations with Performers.

Butt, John. "Bach's Vocal Scoring: What Can It Mean?" Early Music 26/1 (Feb 1998), pp. 99-107.
Christoph Wolff, "Musical Forms and Dramatic Structures in Bach's Saint Matthew Passion," Bach (1988) 19/1, pp. 6-20.
Butt, John. Liner notes to Jeffrey Thomas's recording of the St. Matthew Passion.
Golumb, Uri. Unpublished paper, "Modernism, Rhetoric and (De-)Personalisation in the Early Music Movement.".
Parrott, Andrew. The Essential Bach Choir (Rochester, NY: Boydell & Brewer Publishers, 2000).


A Brief Introduction to Bach's St. Matthew Passion

The Lutheran tradition of setting the Passion narrative to music reached its peak in Bach's St. Matthew Passion. No other setting equals it for length, dramatic complexity, and artistic genius.

Like many Bach sacred works, the St. Matthew both recounts biblical events and tries to apply their lessons to our lives. In the process, the work regularly alternates between different time frames. The central narrative, taken directly from the Gospel of St. Matthew, takes place in biblical times. In it, an "Evangelist" (a tenor) narrates the story of Jesus's arrest, trial and crucifixion, with various solo singers performing the parts of Jesus, Judas, Pilate, et al.; one or both of the work's two choirs sing the roles of the priests, disciples, soldiers, and "crowd."

But another set of musical pieces regularly takes us out of the biblical time frame and into modern times (whether ours or Bach's). This set includes arias and duets for soloists, choruses, and many settings of Lutheran chorales. In these pieces, the individual or congregation contemplates the implications of the biblical events for modern believers. The St. Matthew's complex mix of time frames is part of what places it squarely outside the realm of Baroque opera. Indeed, we sometimes forget that the work was meant for use as part of church services, specifically Good Friday Vespers.

For over a century, scholars believed that Bach premiered the St. Matthew Passion in 1729. But more recent research by Joshua Rifkin and others has shown that the premiere was actually two years earlier, in 1727. The 1727 version differed in some respects from the one typically performed today. This earlier version would make an interesting recording project.

Bach may well have had a sense of the greatness of the St. Matthew Passion. He meticulously prepared a polished score and parts for the now-familiar version of the work in 1736. A facsimile of this score is available on a bonus CD-ROM in the newest recording by Nikolaus Harnoncourt.


Recordings discussed (in chronological order by date of the recording, not of the release. All are 3-CD sets. Notice that certain singers appear in a number of different recordings.)

Nikolaus Harnoncourt (1970). Evangelist: Kurt Equiluz; Jesus: Karl Ridderbusch, Soloists of the Wiener Sängerknaben (s); Paul Esswood, Tom Sutcliffe (a); James Bowman, Nigel Rogers (t); Max van Egmond, Michael Schöpper (b). King's College Choir Cambridge/ Concentus Musicus Wien. Teldec 2292-42509-2

Philippe Herreweghe (1984). Evangelist: Howard Crook; Jesus: Ulrik Cold. Barbara Schlick (s); René Jacobs (a); Hans-Peter Blochwitz (t); Peter Kooij (b). La Chapelle Royale & Collegium Vocale de Gand/La Chapelle Royale Orchestra. Harmonia Mundi HMC 901155.57

John Eliot Gardiner (1988). Evangelist: Anthony Rolfe-Johnson; Jesus: Andreas Schmidt. Barbara Bonney, Ann Monoyios (s); Anne Sofie von Otter, Michael Chance (a); Howard Crook (t); Olaf Bär, Cornelius Hauptmann (b). Monteverdi Choir, English Baroque Soloists. Archiv: 427 648-2

Gustav Leonhardt (1989). Evangelist: Christoph Prégardien; Jesus: Max Van Egmond. Christian Fliegner, Maximilian Kiener (s); René Jacobs, David Cordier (a); Marcus Schäfer, John Elwes (t); Klaus Mertens, Peter Lika (b). Men's Choir of La Petite Bande and Tölzer Knabenchor / La Petite Bande. Duestsche Harmonia Mundi 7848-2-RC

Ton Koopman (1992). Evangelist, Guy de Mey; Jesus, Peter Kooij. Barbara Schlick (s); Kai Wessel (a); Christoph Prégardien (t); Klaus Mertens (b). Netherlands Bach Society Choir/ Amsterdam Baroque Orchestra. Erato 2292-45814-2.

Frans Brüggen (1996). Evangelist: Nico van der Meel; Jesus: Kristinn Sigmundsson. Maria Cristina Kiehr, Mona Julsrud (s); Claudia Schubert, Wilke te Brummelstroete (a); Ian Bostridge, Toby Spence (t); Peter Kooij, Harry van der Kamp (b). Netherlands Chamber Choir / Orchestra of the 18th Century Philips (not available in the US).

Jeffrey Thomas (1996). Evangelist: Jeffrey Thomas; Jesus: William Sharp. Catherine Bott, Tamara Matthews (s); Judith Malafronte, Dana Marsh (a); Thomas, Benhamin Butterfield, David Munderloh (t); Sharp, Nathaniel Watson, James Weaver (b). American Bach Soloists. Koch 3-7424-2.

Jos Van Veldhoven (1997). Evangelist: Gerd Türk; Jesus: Geert Smits; Johannette Zomer (s); Andreas Scholl (a); Hans Jörg Mammel; Peter Kooij (b). Netherlands Bach Society Chorus and Orchestra. Channel Classics CCS 11397

Philippe Herreweghe (1998). Evangelist, Ian Bostridge; Jesus, Franz-Josef Selig. Sibylla Rubens (s); Andreas Scholl (a); Werner Güra (t); Dietrich Henschel (b); Chœr et Orchestre de Collegium Vocale. Harmonia Mundi 951676.78. With a bonus multimedia CD-ROM on the St. Matthew Passion.

Masaaki Suzuki (1999). Evangelist: Gerd Türk; Jesus: Peter Kooij. Nancy Argenta (s); Robin Blaze (a); Makoto Sakurada (t); Chiyuki Urano (b). Bach Collegium Japan. Bis 1000/1002.

Nikolaus Harnoncourt (2000). Evangelist: Christoph Prégardien; Jesus: Matthias Goerne. Christine Schäfer, Dorothea Roschmann (s); Bernarda Fink, Elisabeth von Magnus (a); Michael Schade, Markus Schäfer (t); Dietrich Henschel, Oliver Widmer (b). Arnold Schönberg Choir, Concentus Musicus Wien. Teldec 8573-81036-2. With a bonus CD-ROM of the 1736 autograph manuscript.