Accompanying the Beethoven Violin Sonatas

This weekend, I attended a marathon performance of all of Beethoven's violin sonatas in a single concert, organized by one of my McGill colleagues as a studio project. Each of her students (plus two participants from other violin studios) paired up with a pianist to learn and polish one of Beethoven's ten magnificent violin sonatas. Along the way, a wide range of other colleagues attended studio classes to coach the students (including a modern piano professor, our historical keyboard professor, and me, as a historical violin and musicology professor), and then, this weekend, the whole gang got together to present the cycle in public, moving chronologically from Op. 12 no. 1 all the way through Op. 96. (Studio teachers, take note! This was a fabulous experience for everyone involved.)

I love the Beethoven sonatas dearly, and I know them well, having performed the cycle a few times on period instruments. But as a performer, I've always split the sonatas into three concerts, which in turn could be separated by days or even months. I had never heard the whole cycle live in a single event--and this listening experience alone was edifying and instructive. There is the sheer creative ingenuity Beethoven exhibits across the set. It feels like each work is a fresh attempt to solve the problem of how to write for these two instruments. Not once did my attention flag. I also marveled at Beethoven's creative development across the set. Unlike his piano sonatas, symphonies, and quartets, which are spaced more evenly across his career, the violin sonatas are chronologically lumpy. He composed the first nine within about five years (c.1798-1803) but waited nearly a decade before writing the final sonata, in 1812. This, too, is remarkable. I still find it hard to believe that the first sonata and the "Kreutzer" have only half a decade between them--that an artist can undergo that much growth in such a short span of time, redefining so many formal and expressive features of the genre. And of course, it was equally uplifting to see ten different students, each a true musical individual at a singular stage of development, grappling with the composer, the music, and the instrument.

Watching this inspiring performance, I found myself reflecting on what it is that makes these sonatas so difficult. Flip through the score: there's nothing in the music that "appears" at first glance to be technically impossible--certainly nothing that even begins to approach the challenges posed by, say, Paganini's Caprices, composed 1802-1817 and thus contemporary to the last five of Beethoven's violin sonatas. And yet, despite the seemingly simple notation, these pieces are incredibly hard.

Many of the challenges Beethoven sets for us violinists in these works are expressive rather than technical. If I had to identify the single most important thing for a violinist to keep in mind while playing this music, it would be this: that we are, for much of the time, accompanying the piano. The late 18th-century violin sonata was a genre in which the piano soloist would take center stage, with the violinist often playing quiet whole notes in support; and although Beethoven does expand the role of the violinist beyond mere accompaniment, very often we are there to bolster the pianist. This is even reflected in the way Beethoven and his contemporaries referred to the genre. Although today we think of these as "sonatas for violin and piano," in the late 18th century they were known as "sonatas for piano and violin."

If you're a violinist starting to dig in to these pieces, one way to begin thinking about your role is to ask your pianist play various passages without you. Sometimes, as in the opening phrases of Sonatas 1, 3, 4, and 8, you'll see that absolutely nothing is missing, that the piece is "complete" even without the violin part. Ask yourself, then: in such cases, what exactly is your job? Why did Beethoven bother writing a violin part? One answer is that the violinist's function is to provide aural "background" so that the pianist can act like a soloist. In Sonata no. 3, for instance, the piano part alone sounds like a coherent solo sonata. Add the violin playing those half notes, though, and you suddenly have an "orchestral" background from which the pianist emerges, like a concerto soloist. (In fact, once you see it this way, isn't the opening just like the beginning of the "Emperor" Concerto?) At other times, as in Sonatas 1 and 8, the unison helps the pianist sound more orchestral. And in Sonata 4, the violinist gets to manufacture the illusion of the piano's resonance, so that the pianist is free to play a clear left hand without obscuring the eighth notes with the pedal. With the violinist's help, the pianist can have it both ways, articulating the left hand while also producing a halo of sound that supports the long slur and adds warmth.

Here's how I usually describe all this: the violinist's job in 80% of this music is to make the pianist sound better. Once you take this outlook on board, so many interpretive matters clarify themselves. Vibrato, tone color, articulation, and the like are suddenly to be used in the service of blending with the piano and creating resonance that is unavailable on that instrument alone. Try, as an exercise, having your pianist play just the left hand along with the violin part, so you can coordinate these matters: you'll find, for instance, that if you really focus on supporting the piano, you'll vibrate a lot less on those long notes than you might have otherwise. One of my favorite passages for this exercise is the theme in the slow movement of the "Spring" Sonata (no. 5). Those interjected quarter-note sighs in the first iteration of the theme, and the syncopated eighth notes and gentle sixteenth-and-eighth-note rhythm when it repeats, need to be both audible and truly in the background, supporting what the soloist does without taking attention away. When the violin plays a dissonance that is absent from the piano part (the G flat in m.35, for example), it's a moment to reclaim aural focus. And even when the task isn't to play an accompaniment "with" the pianist, you can benefit from imitating the pianist's style of executing similar figures. In the slow movement of Sonata 6, don't try to sing out every sixteenth-note triplet in the arpeggiated accompaniment passage in the second half of the movement; instead, ask your pianist to play their version of that accompaniment for you, and try to imitate the lilt so easily achieved when a keyboardist plays that figuration.

Perhaps because the violinist spends so much time accompanying, I've always felt a little strange standing in front of my pianist when I play these pieces. So I generally set up the stage with the pianist in front, while I stand behind and read over their shoulder. The very nicest way to do this is to have the pianist actually facing the audience, the end of the instrument pointed directly out, with violinist standing by the pianist's left side. This is how musicians generally set up in the late 18th century, and it works beautifully in this repertoire. It makes it easier to play in the background, since the piano is, quite literally, in front--and it carries the added benefit of allowing you to actually see the pianist's left hand and adjust your playing in response. The benefits accrue everywhere, but are especially palpable when the violin and piano left hand carry joint accompaniments. Of course, being a historical performer, I'm ok doing wacky things like radically rethinking the stage setup for this music, since I'm not contending with the weight of a modern-instrument performance tradition. But as HIP practices become increasingly mainstream, even modern players might want to experiment with this setup. (And, to their great credit, many of the McGill students did this past weekend!) It really allows both players to make these pieces into the chamber masterpieces they are, rather than putting the accompanist out front while the piano soloist, in the back, does much of the work.

Another set of ideas that can help performers find their way through this music involves understanding the gestures that make up Beethoven's expressive arsenal. I hinted briefly already at the value of thinking this way, when discussing the opening "concerto" passage of Sonata 3. Once you recognize the opening four bars as sharing some elements of the "concerto" genre, your pianist might feel emboldened to play those bars out of tempo, like the quasi-cadenzas they appear to be. This idea can be generalized as follows: always ask whether the texture of a given phrase implies some performance directives. To me, the opening of Sonata 1 looks like the start of a symphony; and this means that my job as a violinist is to help the pianist sound like a full orchestra, complete with strings, winds, trumpets, and drums. This means that I'll limit the vibrato and adopt a different tone color than I would in a more melodic setting. Sonatas 2 and 6 open with what seems more a string quartet texture, with the violinist playing either second violin or second violin + viola, and this in turn carries a different set of associations for phrasing and rhythmic feel.

Nor are Beethoven's signals purely textural. Other rhetorical or expressive gestures come in the form of rhythmic patterns that suggest various dance types, which can also offer insight into tempo and phrasing. The last movement of Sonata 2 is a minuet--so, don't play it too fast, and be sure those lovely syncopations tug against the more usual downbeat-centric hierarchy. Likewise, the second movement of Sonata 8 is a minuet--in this case, don't play it too slowly! And make sure the unslurred quarter-note upbeats to the second melody are light. Other dance-like patterns found throughout these works include gigues (last movement of Sonata 1); contredanse (last movement of Sonata 3); gavotte (last movement of Sonata 8). Other rhetorical markers, meanwhile, are broader and say something about the atmosphere of a piece. Particularly well represented in this cycle are features of the "pastoral" style (drone basses, 6/8 meter, woodwind textures), which show up in Sonatas 4, 5, 8, and 10, and may suggest a less virtuosic, and more muted and intimate style, than what you often hear in modern-instrument recordings.

Of course, despite their often simple appearance on the page, these sonatas are exceptionally difficult. But I hope this brief overview of some of their expressive features helps others find a productive point of entry into Beethoven's writing for these two instruments. Although thinking of the music in such ways does not automatically disarm the technical challenges, I've found time and again that the technical difficulties become less acute with these adjustments of mindset--that accompaniment passages become a little easier to play when we stop trying to emphasize every note, and that even the flashier phrases become more approachable when we recognize them as part of an intimate dialogue with the other performer rather than as soloistic flights that demand high-octane delivery. The dance-like movements, too, become easier when we allow the gestures to have some strong notes and some weaker notes, and when we relax our sound and arms on the lighter parts of the bars. Needless to say, these are just a few of the relevant expressive issues, and there are others as well, for instance the sense of humor that imbues these sonatas, and other kinds of expressive characters. I may revisit them in a future blog--but for now, happy practicing!

Embellishing Mozart's C-major Quintet

 For many of us who play music by canonical composers like Mozart, the notated score is sacred. Our job as performers, we might think, is to realize the notation as faithfully as possible--and this means in particular that we can't just change things without Mozart's permission. But this isn't how musical scores were treated during Mozart's lifetime. It's now clear that Mozart and virtually all his contemporaries expected performers to make all kinds of changes to the music they played, especially through the insertion of elaborate cadenzas, lead-ins ("Eingänge") and embellishments.

Research into these topics has been carried out most thoroughly in studies of Mozart's keyboard music. And the practice of improvisation and embellishment in modern-day performing culture are most closely associated with keyboard players, especially Robert Levin, who has recorded highly embellished readings of Mozart's complete piano sonatas and piano concertos. (There are very few exceptions; one of them is Nils-Erik Sparf's lively and daringly unconventional recording of Mozart's violin concertos.) One reason for this keyboard-centrism is the simple, historical fact that Mozart himself identified primarily as a virtuoso pianist for much of his career, and studies of his performing practices therefore often lead back to his activities as a keyboardist. In my own recent book on Mozart, which devotes a chapter each to improvisation and embellishment, keyboard music features more centrally than string music because there's more direct evidence for Mozart's stylistic preferences, in the form of manuscript models, published variants, and pedagogical samples, in his keyboard output.

But I wonder whether there's also another, perhaps more interesting, reason that most Mozartean embellishers are keyboardists. Much of Mozart's keyboard music is written not just for a soloist playing alongside accompanying forces, but rather for an individual player, as in the piano sonatas or in so many concerto passages where the orchestra drops out and the soloist forges on alone. In such context, embellishment presents the fewest possible practical obstacles. The performer who plays entirely alone can do all sorts of things to mess with the musical text, and never needs to worry about the effect this might have on collaborators. Thus, Levin's hilarious reading of the last movement of the Sonata K.283 features madcap textural changes and transpositions, and Andreas Staier's brilliant recording of the last movement of the Sonata K.331 introduces some unexpected contrapuntal tricks during repeated passages--and these kinds of textual interventions would be impossible in the context of a work for multiple performers. For us string players, however, everything we play by Mozart is a work for multiple performers! What are we to do?

I started asking such questions in earnest last year, when I recorded Mozart's violin-viola duos (and I blogged back then about many of the performance decisions, including the embellishments and cadenzas). One of the reasons I selected those pieces as my entry-point into embellishing Mozart's string music is that the players are comparatively unconstrained. Although there are many instances of real textural complexity, there are also plenty of phrases where the violinist plays the tune and the violist accompanies. (This did not stop me from inserting a surprise viola Eingang in the first movement of the B-flat duo!) The practice of writing and performing embellishments for the duos was highly instructive--but at no point did it challenge my fundamental understanding of the topic.

This month, however, I'm performing the great String Quintet in C major, K.515, perhaps the most daring piece Mozart wrote. Part of the pleasure of doing K.515 is, of course, simply to be inside Mozart's mind at this high-point of his compositional life. A nice add-on, however, is that I get to try to embellish in a limit-case: a context where it is always stylistically appropriate but extremely difficult to pull off in a way that seems musically appropriate.

One of the hardest things about embellishing in K.515 is that counterpoint features centrally throughout the work--and this means that very few melodies can be innocently tweaked without wreaking havoc among the other players. For instance, the final theme in the first movement exposition might look like a perfect candidate for embellishment:

However, following those first four bars, the theme is immediately played in octaves between the first and second violins, and imitated contrapuntally by the first viola and cello. The first violinist can embellish mm.132-35, but short of pre-coordinating some embellishments with the other players (a no-go, since it would shatter any sense of improvisatory freedom) this would mean that the embellished theme is followed by a necessarily unembellished restatement. That's something Mozart himself rarely gives us in his composed embellishments. Embellishments are meant to ratchet up the intensity of a melody, not vanish and let the intensity wane.

Then there are passages like this:

Here, it's entirely possible to insert embellishments! First off, the violinist has four bars of sitting on a D dominant 7th chord, and can easily turn those bars into a stylistically-appropriate Eingang of some kind. I jotted this down, but the possibilities are endless:

However, problems pile up in the following phrase. Beginning in m.86, we get another melody that should be eminently embellishable: there's no immediate contrapuntal imitation, and the first violinist is just as free as any soloist, with all four other players holding long notes beneath the tune. But look closer, and once again constraints appear. The voice-leading is such that, when we move from tonic to dominant in m.87 and m.89, the first violinist still needs to hover somewhere around the fifth scale degree, as in the unembellished original. Move anywhere else and we'd hear parallel octaves, since the second violinist controls the third scale degree and the two violists control the first scale degree. What might one do? A possibility would be something like this, a chromatic wiggle that coexists with the original melody for m.86:

As far as embellishments go, I like this one--though it hardly draws attention to itself as a florid embellishment.

This previous example points to the most pervasive difficulty in embellishing Mozart's chamber music. Mozart favored an embellishment style full of chromatic sinews and twisty gestures that circle around the notes they embellish rather than connecting them in a direct scale. But introducing such winding, circuitous embellishments often brings problems of voice-leading in music so densely notated as these string quintets. This is not to say that embellishment is impossible; but the performer who wants to embellish is certainly on a leash.

In the slow movement, these effects are slightly diminished. The recurring theme calls out for embellishments, as to all repeated themes in Mozart, and here the soloistic nature of the first violin line makes intervention easier than it was in the first movement. One of the interesting questions here is how to treat the many short rests (an eighth note, a quarter note) in the first violin part--whether florid embellishments can just cross over those silences, as Mozart's own embellishments often do in his keyboard music, or whether the silences need to be observed in order to clear aural space for interjections from other players. My solution has generally been to cross over the rests; thus, this cadential figure returns twice and can be treated with various embellishments--perhaps a diatonic arpeggio on its first recurrence and a twisting, chromatic scale on its second:

Then there are passages where playful imitation occurs between the players, and embellishments in the first violin part will be a spur to creative invention for the first violist:

So far I've composed embellishments for the first violin part in the first two movements. The minuet, like the first movement, offers very few possibilities (though I'll certainly come up with some chromatic variants for the cadence gesture in m.9 (and m.23 of the Trio). But it's the last movement I'm most excited to do, where the rondo theme repeats a few times across the movement and feels very much like the soloistic rondos elsewhere in Mozart. It should offer plenty of opportunities for embellishment--and, as with the previous movements, I'm sure I'll learn a lot in the process of writing them!

In the meantime, here's a messy draft of my embellishments for the slow movement, in case others are interested in seeing them and playing them--or taking inspiration from these ideas and trying their own hands at composing different embellishments. These generally follow Mozart's melodic style as closely as possible--though I'm certain that these will change once we start rehearsing in two weeks:

Diary of a Recording: Mozart's Violin-Viola Duos

I've been a Mozart fanatic for as long as I can remember, and his music is absolutely central to my life -- outside my career as a violinist, that is. I've conducted his operas, symphonies, and concertos; I've written a book on interpreting his instrumental works; I jump at any opportunity to coach his chamber music; I routinely try to pick my way through the solo parts of his piano concertos at home when nobody's listening...and yet for some reason I've rarely found occasion to actually perform his string music, preferring instead to engage with it as a conductor, teacher, or author. Maybe I've been just a wee bit scared of the enormous challenges, both interpretive and technical, he throws at us violinists!

But this summer, my relationship with Mozart changed. I finally gathered the courage to embark on a recording project exploring his string writing. I experimented quite a bit along the way, and I learned a lot in the process!

Where does one begin when trying to get inside Mozart's violin music? Pianists have it easier than we do, since the piano concerto served as an artistic vehicle for the entirety of Mozart's life and career, meaning that any point of entry into the repertoire is sure to yield interpretive riches. Mozart first dipped his toes into the genre of the piano concerto at age 11, with four pastiche concertos arranged from music by contemporary composers, and he wrote his last piano concerto in 1791, his own final year. This means that his piano concertos chronicle a good 25 years of his stylistic explorations. By contrast, his violin concertos are youthful pieces, all dating from his teens. They provide a snapshot of his development at a single moment, but they are not mature masterworks. And even broadening the purview to include pieces outside his official five concertos, such as the "Haffner" Serenade K.250, or the spectacular Sinfonia Concertante K.364, we remain in the early phases of Mozart's musical development. The "Haffner" dates from 1776, only a year after the A-major violin concerto, and the Sinfonia Concertante from 1779, when the composer was just 23. The music is great by any measure; yet it clusters at a single moment of his life, and it can feel limited.

Why does Mozart turn away from the violin concerto at this point in his life? It's not that he abandoned string writing altogether -- though he did announce to his father, in a letter of 7 February 1778, that he planned to stop performing publicly on the violin in order to focus his efforts on the piano and composition. And with this turn away from public performing as a violinist, his innovations with the instrument shifted away from music for the concert stage, relocating instead to the home and salon, where Mozart continued to play the violin and viola as a chamber musician. It comes as no surprise, then, that beginning in the early 1780s, the string writing in his chamber music became more daring and more complex. Thus, I elected to begin my explorations in that world -- not with the quartets or quintets, but with the two extraordinary duos for violin and viola, K.423 and K.424. I spent spring and summer of this year preparing these remarkable pieces with the intrepid Catherine Cosbey of the Cavani Quartet and McGill's violin faculty, and we took them into the recording studio in June. The other works on recording are the Violin Sonata in A major K.305 in an anonymous arrangement for two violins published around 1799, as well as selections from Mozart's final opera La Clemenza di Tito in an arrangement published around 1800. Although Catherine is primarily a modern violinist, we used a period-instrument setup. I played viola for half the recording and violin for the other half. Despite my longtime professional association with Mozart, the experience was by turns challenging and revelatory.

Perhaps the primary discovery I made during this process -- maybe it sounds obvious in hindsight, but I didn't take it for granted at the start! -- was just how good Mozart's violin-viola duos are. Of course I had played them recreationally before, but I had never actually learned them properly or interpreted them, and I was unprepared for their astonishingly high quality. Many academic discussions of the duos focus on the fact that Mozart wrote them as a favor to Michael Haydn, who was unable to finish a set of six duos and requested that Mozart provide the remaining two. In such discussions there's often an implication that Mozart tailored his duos to match the quality of Michael Haydn's less inventive efforts. But nothing could be further from the truth. I discovered, approaching the pieces this year, that there isn't a lazy phrase to be found. This is true, first of all, on the level of sheer compositional technique. Mozart's craftsmanship is always astonishing. Consider, for instance, the first movement of the G-major duo K.423, whose second theme is full of subtle wit. The melody seems unassuming enough, but look closely and you see that each zigzagging interval is one step larger than the previous one. The theme begins (m. 27) with an ascending second, followed by a descending third, ascending fourth, descending fifth, ascending sixth, and descending seventh - a fun effect in itself, since Mozart somehow manages to bring this off without warping that old standby of a chord progression, I-V-vi-IV, or undermining the tune's lyricism. After this wedge-shaped bit of melody, in m. 29 the sixths and sevenths catch on and spin out into a melodic answer in their own right, even as they continue to push lower and lower, the harmonic rhythm increasing before steering to a cadence.

So far so good. But in the recap, when the theme returns, the descending answer runs away with itself. The viola begins the gesture in m. 118. But instead of successfully bringing about a cadence, as the violin did previously, the two players get tangled up, endlessly tossing the melody back and forth and continuing to desperately transpose up so as not to crash into the bottom of their instruments' ranges. They keep this process going for so long that, if we removed the ascending leaps, the whole passage would plunge down more than four octaves. Meanwhile, as the players get lost in the tangle, chromaticism creeps in, and by the time the violist starts in on the sixteenth notes in m.121, things seem to be spiraling out of control! Yet somehow we arrive back on the necessary predominant in m.122 and come, once again, to the polite cadence a bar later.

Beyond such compositional cleverness, the duos are impressive for their expressive range. We find in them a massive store of operatic references -- there are phrases and melodies in the last movement of the G-major duo that come to feature in Don Giovanni, and throughout the B-flat duo that come to feature in Die Zauberflöte. Most of all, however, I'm struck by Mozart's textural ingenuity. Even with only two instruments at his disposal, he simulates a dizzying number of non-duo textures, from the string quartet (when both musicians play double-stops, yielding four-part writing) to horn calls, symphonic fanfares, an aria, and more. Taking this in, and considering that Mozart composed the duos in late 1783, I found myself understanding that this is where he really learned how to write for strings. By the time he wrote the two duos, he had completed only one of the mature, "celebrated" string quartets, and in the duos' immediate aftermath he would write five more in relatively rapid succession. Perhaps the burgeoning textural and instrumental ingenuity in those quartets was sparked by the creative constraints he faced here, in writing for only two instruments.

All this was on my mind while preparing the recording. But the duos, like all of Mozart's music, also present considerable interpretive challenges that performers have to face.

One of my Mozartian obsessions is musical character. It's often said that Mozart's output is fundamentally operatic, and I wholeheartedly agree. Every phrase suggests a character of some sort, and the duos are no exception, whether in the literal operatic references (for instance, Donna Elvira's aria "Mi tradì" makes a brief appearance in the last movement of the G-major duo, m. 96-97) or through more general references -- here a gesture implying the gravitas and menace of The Count, there an amorous march redolent of The Countess, elsewhere the patter of Figaro or Masetto. The question I sometimes wonder about, however, is not whether there are "characters" present in the music, but how to tell where one character ends and another begins. Do we divvy up expression at the level of the entire phrase? Of the half-phrase? Perhaps by the bar, or even the beat? I feel that this may be the most pressing challenge I face when interpreting Mozart.

In some rare cases, Mozart makes it very easy to tell where one character ends and the next begins. Consider, as an example from a different work, the familiar first movement of the G-major violin concerto, mm. 64-68. The contrast between the staccato marks and slurs, which suggests a broader expressive contrast between implied fanfares and swooping, lyrical gestures, makes it clear that the passage implies a dialogue, and that the characters shift back-and-forth by the bar. No problem there! But the situation is usually more ambiguous. How many expressive stances might we find, for instance, in this unassuming passage from the slow movement of the G-major duo K.423?

On first glance, it might seem like just one: some sort of a "singing" melody in the violin, tied together with the two rhyming downbeats of m.9 and 10, all above an unobtrusive accompaniment in the viola. But, looking more closely, each bar also carries its own expressive implications, and the whole progression tracks a series of highly distinct gestures, perhaps even a hint of dramatic narrative. The first bar shown here, m. 8, features a slinky, intense chromatic line. Indeed, when we hit G# on the downbeat of m.9, there's even a small moment of uncertainty: is it an A flat, and have we just veered into F minor? (The movement is marked Adagio, so between the slow tempo and a touch of rubato to elongate the G sharp slightly, the suspense is not negligible!) But then the note resolves up to A, and we remain safely harbored in F major. ("Phew! That was a close call!") The ascending figure later in m.9, with its delightful dotted rhythms, seems to laugh after this brief musical double entendre. ("Did someone say F minor? Not me! Hah!") The bar might be played in a way that is dainty, almost coy. And then the trill-and-scale in m.10 seems altogether more lyrical, confident, and operatic. Viewed thus, the character changes by the bar -- and this rapid succession of implied expressive states is a challenge to bring out. Both performers need to remain constantly alert for opportunities to deploy those small shifts of inflection so as to convey the whole story.

The situation is largely the same in the fast movements. In the B-flat duo first movement, the second theme runs:

Here too, it isn't obvious how many characters we should imagine. It might be just one -- another largely lyrical melodic line. But maybe the pantomime unfolds more quickly. Perhaps the upbeat (m.43) is playful, slowing down with the ascending chromatic line and hesitating ever so slightly to create another moment of suspense: something new is coming! What kind of theme will follow? In m. 44, we find out that the theme is lyrical. The arpeggiated line can swoop forward and up. The real question is about m. 45: are the trills joking, even teasing? Or do they remain lyrical? That the repetition of the theme in the subsequent phrase places the trills in both instrumental parts (m. 49) makes me think the gesture in m.45 is more delightful than lyrical, less a continuation of m. 44 than a new, interrupting idea that takes hold and humorously distracts both players away from the songlike theme that began the phrase. Then, m.46 stops the trilling delight and re-introduces melodic sincerity.

In the above examples, nothing conclusive can be said about these changes of character. They may simply be a matter of taste: how playful, how irreverent, how mercurial do you like your Mozart? Those with a high tolerance for Mozartian volatility will accept more rapid changes of character; those who prefer a more staid Mozart will find my suggestions implausible. But there's little in the music itself to adjudicate the matter. The interpretive approach I find useful in such cases, and one I explore at greater length in the final chapter of my book, is to really focus on what specific expectations are plausible at each individual musical moment. To do this, I imagine that I am, truly, encountering the piece for the first time. This isn't a cliché -- it suggests a very specific stance, one that lives in the moment and ignores our knowledge about how each phrase will actually unfold. In the case of the K.423 slow movement, with that G-sharp discussed above, the very idea that we should pretend it "might be an A flat" relies on this outlook. Sure, in some sense we know that the G sharp is, in fact, a G sharp, that it will resolve up to A, and that the F major chord will continue through the bar. But if we pretend just for a moment that we don't know that, then a whole new interpretive mindset presents itself. The viola accompaniment doesn't make it clear whether the chord is major or minor. We can easily try to imagine that the note might in fact be an A flat, and that it will move down to F and set off a soaring, dark F minor arpeggio. The entire notion that the phrase is somehow dangerous, that the character might have to change by the bar, follows irresistibly when we begin to think of music in this way, adopting a stance of make-believe ignorance, responding to each successive moment on its own terms and pretending that the interpretation isn't premeditated. Applying this approach to Mozart often leads to readings that are alert, dramatic, even wild. We are able to hear, and thus to play, each piece as though it is, in fact, unfolding in real time from a series of compositional decisions. Asking at every opportunity what Mozart "might have done" differently and comparing the options, can spark a more complex and nuanced way of approaching the issue both of character and of character-shift.

In other cases, meanwhile, the question of how quickly expressive states shift is moot: it might be clear that one character persists over a long stretch of music. Even in those cases, however, identifying a phrase's expressive world is not always a simple matter. For me, playing viola in K.424, I faced this challenge most directly in the slow movement. The violinist doesn't need to entertain any doubt: Mozart marks the movement "cantabile" and spins out one of his most operatic melodies:

For the violist, however, the nature of the phrase is far less obvious! What kind of accompaniment is this? Preparing for our recording, we experimented with a range of options. The usual approach, and the first we tried, is to play the accompaniment fully legato, in the style of a wind serenade -- as if scored for basset horns and bassoons. We also experimented with bowings, trying both linked and unlinked versions of the quarter-eighth rhythm. Then we explored a totally different conception of the piece, playing the viola part very staccato (except when there are slurs), and imagining the accompaniment as if scored for a strummed instrument like guitar or lute. I even tried it pizzicato once or twice! We tried, too, a middle-of-the-road version -- what I thought of as staccato with a dab of "fake reverb". For the recording, we ultimately settled on doing it mostly staccato: not actually plucked, but light and short enough that it would sound song-like. But, even though that's how we recorded it, it's clear that the choice might be wrong. The piece remains elusive, and I have no doubt that when I next perform it, I'll feel differently about the articulation and, by extension, the expressive nature of the movement.

Although in this particular case I don't feel that I personally hit on the one "right" answer, there's a more general principle at work. One of the fundamental goals I always pursue when interpreting Mozart is to be alert to possibilities for simulating the sounds and styles of other instruments. Sure, I'm actually playing the violin or viola (or talking to a member of an orchestra, as the case may be); but often Mozart writes music that seems ventriloquistic, where one instrument is meant to embody the timbre or attitude of another. This is the thought behind my deliberations on the slow movement of K.424 and the question of basset-horn vs. guitar. In addition to that example, there are plenty of other phrases in the duos where such antics take place: at times, for instance, the viola part approximates a horn call (K.423 last movement, mm. 34-39), and perhaps this means that the violin during this phrase is really meant to feel a bit like an oboe -- boisterous and reedy. Elsewhere, including much of the first movement of K.424, it feels like both instruments are referencing a wind serenade. And needless to say, this idea applies across Mozart's output, far beyond the duos. In the string chamber repertoire alone, we find countless horn calls (famously at the outset of the E-flat major string quintet), and the same might be said of the concertos -- consider the G-major concerto, where in the first movement (mm. 76-77) the soloist joins a melody previously played by the winds. But these questions struck me as being particularly acute in the context of the duos, where the instrumental forces are uniquely limited. In the concerto, the soloist plays with the winds but is not necessarily ventriloquizing a wind line, whereas in the duos there is a much more visceral approximation of a wide variety of instrumental writing, and thus a relentless challenge to the performers to adopt the right kind of "voice" for each phrase.

One final realm of experimentation in this recording was embellishment. This, too, has been central to my Mozartian explorations in the past -- I write about it in Chapter 4 of my book, as well as in this open-access article. Although we know that Mozart himself was a habitual ornamenter, and he expected his contemporaries to creatively intervene in his works by adding extensive embellishments, many string players have been slow to take up the challenge. Period-instrument pianists are generally more willing -- Robert Levin's complete cycle of Mozart's piano sonatas is a fascinating example of the interpretive riches that flow when a performer is sufficiently uninhibited! -- and in approaching the duos we set ourselves the task of bringing the same ethos into Mozart's string music. Of course, the situation in a duo is a little different from that in a sonata: an individual pianist can intervene in a musical text without worrying about how any embellishments might affect other performers' lines, whereas in a duo the players need to respond to each other. Even so, in this recording we went all-out.

We started by adding significant embellishments in all repeated passages, which was Mozart's own practice. This means that embellishments -- sometimes so dense that they seem more like wholesale variations -- feature in the repeated exposition from the first movement of K.423 as well as in smaller-scale formal repeats, such as reprises of the rondo theme in the last movement of K.423 and all the repeats in the variation movement of K.424. Throughout, we attempted to mirror the style of embellishments Mozart uses in the handful of movements for which he provided models. Here's what we came up with, for instance for the rondo-finale of K.423. Mozart's original theme runs:

Our embellishments for one of the melodic reprises (the handwritten passage pasted into the score):

We wrote different embellishments for each appearance of the theme, increasing the density of additional notes every time the theme recurs. In another case, in the slow movement of K.423, we drew embellishments from the first edition of the piece, published shortly after Mozart's death. The Bärenreiter edition excludes these because they can't be confirmed to have originated from Mozart; but we thought the style was convincing, so we re-introduced them! And even if they weren't written by Mozart himself, they are contemporaneous, so they certainly reflect historically-appropriate practices:

Finally, we added cadenzas at every fermata, and even placed one or two brief extra cadenzas in the outer movements of K.424 where no fermata is indicated but where we thought Mozart might have expected a touch of improvisation. We generally tried to base these, too, on models: for instance, the cadenza we added in the K.424 variation movement imitates the cadenza Mozart wrote for a piano variation set. For the other cadenzas, we drew from motifs in each movement and did our best to match Mozart's style.

It's been an exhilarating and rewarding experience to get inside this music and to try to give it fresh, creative life. The final product is due for release in November. I certainly hope this will be but the first step in a longer-term project featuring similar expressive experimentation with Mozart's other, later string chamber music -- but in the meantime, we have another round of edits to check!

Smith on Sympathy and Selfishness in the Wealth of Nations, Book 1, Chapters 1-3

I'm hardly the first person to note that "The Adam Smith Problem"--that apparent contradiction between the sympathy ascribed to humanity in The Theory of Moral Sentiments and the self-interestedness ascribed to humanity in The Wealth of Nations--is itself problematic, based as it is on a mistaken premise about what Smith actually argues in these texts. But, revisiting Smith this weekend, I'm struck by the fact that these two texts are more attuned to each other than even mainstream Smith scholarship seems to think. (I say this knowing that, despite relatively wide reading at this point, I still haven't scratched the surface of Smith scholarship...so it's both possible and likely that I've missed a source that takes a similar view to the one I'm about to outline!)

This term, I'm teaching a grad seminar on various conceptual (rather than historical or practical) overlaps between narrative art (especially opera) and economic theory in the 18th century. The purpose of the course is not to ask how the economics of performance or artistic production worked back then, but rather to investigate how the various ideas and preconceptions that gave rise to the birth of economics (especially to Smith's writings) also structured the way various artists were thinking about character, narrative, plot, and psychology at the time--that is to say, how the conceptual structure of early economic thinking enabled a certain kind of artistic output to arise. We've spent the first few weeks of term grappling with various musicological texts on market culture and its effect on the works of Haydn, Mozart, and Beethoven especially--but starting next week, we will begin to read Smith. I opted to begin with WN rather than TMS.

My romp through WN this weekend is my second time opening this book. I read it for the first time in 2021, and now I'm rereading portions specifically with an eye to class discussions this term. Reading it the first time, I was overwhelmed by the system of thought it put across, and the encyclopedic completeness with which it communicated this system. It was also the first Smith I had read. Now as I reread it, I have the benefit of also having read TMS, his History of Astronomy, a bunch of his essays (including the excellent writings on music, the imitative arts, etc.), and some of his lectures on rhetoric.

All of this is to say, I'm approaching WN with a vastly different structure of background knowledge from what I had back in 2021. And the experience of reading even the first few chapters is indeed strikingly different from what I recall from two years ago. Here are the things that stand out to me, on the level of argument and rhetoric.

First, and most broadly--and pace all those commentators (such as Russ Roberts, whose How Adam Smith Can Change Your Life is fun, but now seems a bit misleading)--Smith does not begin WN with an appeal to selfishness. There is a famous sentence, quoted by virtually all the commentators I've read, which runs:

"It is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own interests. We address ourselves, not to their humanity, but to their self-love..."

Yes, it's clear that this is the ur-statement of self-interest it is often taken to be; yet I find it significant, upon re-approaching this text, that it is placed in Chapter 2 rather than Chapter 1. Chapter 1, far from promulgating (much less recommending) a view of the fundamental selfishness of people, is all about the seeming-miracle of coordination among those who work together in individual industries. Smith marvels at the productiveness that arises with the division of labor, and implies, both in this first chapter and in Chapter 3, that the mechanism which allows the general increase in prosperity, productivity, and well-being to occur is a result of mutual, imaginative sympathy as much as it might be a result of self-interest. Chapter 1 reads like proto-Hayek on the distribution of knowledge through society (and anticipates, also, the famous "I, Pencil" essay by Leonard Read. Smith portrays the most successful people as those who work together to complement each others' needs, and even in his appeal to the "limit of market power" in Chapter 3, he suggests that an awareness of other people's interests and desires is itself the factor that determines what professions each individual can pursue. In itself, this framing of the first chapter, and the fact that Chapter 2 (with the "self-interest" statement quoted above) is tucked quietly between Chapters 1 and 3, undermines the apparent distinction between the worldview encoded in TMS and that encoded in WN.

There are some other interesting quirks of rhetoric and argument I noted as I made my way through these chapters, as well. For instance, I am struck by the mode of presentation of the division of labor idea. Smith could easily have begun the book with a statement along the lines of what we read in Chapter 2; he could also have begun Chapter 1 with a clear, overarching statement of the idea he will come to by the end of the chapter. Instead, however, he begins immediately with an example. He states that it will be easier for the reader to grasp the overall concept if he begins this way--a bit inductivistic, alas, but unsurprising given the Zeitgeist--but what he actually does, right on the first page, is to make a basic point about invisibility and evidence. He states that the effects of the division of labor are greatest precisely where they cannot be observed directly: in small enterprises, he tells us, it's easy to see people working on individual components of a project, whereas in large societal enterprises the work is so widely distributed that nobody can actually see all of it happen, nor grasp the extent of division it takes to complete it. Reading this passage after having read the History of Astronomy, I'm struck by the fact that this explanation is a concrete manifestation of a point he implies w/r/t the philosophy of science, namely that the task of science is to explain the seen in terms of the unseen. He immediately appeals to, and strengthens, his reader's tolerance for arguments invoking invisible mechanisms. The effect of this is not only to lead up to the famed Invisible Hand (though I think it does that as well!), but to bolster the subsequent arguments and anecdotes concerning coordination and the extent/power of the market, which relies on a sense that an invisible network of sympathies and imaginings connects all of the individual people in society.

Even the statement concerning self-interest, in Chapter 2, is not what it seems when quoted out of context. Smith begins the chapter by distinguishing the behavior of animals from that of humans. Animals have to look cute, he says, and appeal to humans' good nature, if they want a human to pamper them. Humans, he says by contrast, are forced to think rationally about what other people want. The ensuing statement about the self-interestedness of the butcher, baker, and brewer looks on the surface like it defends a view of humanity as intrinsically selfish. But it seems to me that it rather urges people to sympathize more with those around them, for the very mechanism by which we could even imagine the desires of the baker, brewer, or butcher is precisely the sympathetic impulse taken up in TMS. Smith speaks explicitly of self-interestedness, but implicitly gives us an explanatory structure that depends entirely on sympathy, coordination, and an imaginative effort towards fellow-feeling.

Finally, I'll just note a few fun things that occurred to me while reading today. First, Smith anticipates the argument (end of Chapter 1) that it's better to be a poor person in a wealthy nation than a wealthy person in a very poor nation. He puts the point in terms of Britain vs. Africa; but of course what comes to my mind, also, is the quite uncontroversial claim that I'm better off living in middle-to-lower class America in 2023 than I would be if I could change places with even the wealthiest nobleman in 1600. (This point is underscored, also, by my current reading of Katherine Rundell's John Donne biography, which paints a grim picture indeed of many aspects of life back then.) Second, I was struck by Smith's observation that automation generally makes things better for workers--a point anticipating Milton Friedman's argument to the same effect. Friedman said in some lecture or other that the invention, say, of running water did more to alleviate the lot of the poor than of the ultra-rich, since, as he puts it, the wealthy have always had running water (carried on the backs of their slaves or servants), and that it is in fact those carrying the water who are saved from their toil by the advance of technology. For Smith, too, it is the boy who wants to play with his friends who benefits from ingeniously devising a mechanism to do his work for him. This anticipates the Suitsian Utopia laid out in The Grasshopper, in which only those who wish to work need to work, and everyone else is essentially playing games. Finally, I will just gleefully note that Smith, too, treats animals (Chapter 2) as being essentially automata, a point that slots nicely alongside my earlier, cursory musings about Descartes and ChatGPT.

Favorite books and music of 2022

 As the season of "Year in Review" posts comes to an end, I'm thinking not only about the projects I saw through to completion this year (the biggest of which were the final revisions of my book manuscript), but the wonderful array of books and music I've consumed this year. Here's the roundup, along with some comments and observations.

Music

Strangely, my music consumption this year was modest compared to what it's been during previous years. Perhaps I was so absorbed with my various writing tasks, as well as with grant applications, prep for my CD recording in June, and the like, that I just didn't have it in me to engage in the depth of listening to match what I did in 2021. However, there were a few noteworthy albums I enjoyed. In classical music, the biggest discovery was Benjamin Allard's recording of the complete Bach organ works (with a few clavichord performances thrown in). I'm trying to listen to this relatively closely, so I've been slow to progress through the huge set; however, even just judging by the first couple of discs, the achievement is amazing. The playing is spectacular, and the music--much of which is new to me--is really brilliant. I've also been listening somewhat obsessively to the Peabody Trio's recording of Beethoven's Op.70 (these are on modern instruments, but are wonderfully flexible and deep), as well as to Robert Levin's complete Mozart sonata cycle, which I'll be reviewing next year. Outside of classical music, I've really enjoyed Ethan Iverson's jazz album "Every Note Is True" (especially for its large-scale pacing).

Books

But it was a good year for reading! As ever, I've been somewhat lazy about keeping up with academic literature in my field. I tend to think that the bad stuff, which isn't worth reading anyway, will turn out not to have mattered, and that the good stuff will keep rearing up over the coming years, and that if this is true it makes sense to hold off and read things a few years after they're published. Nonetheless, I did enjoy some new or new-ish academic books. Highlights in this category were:

• Matt Strohl, Why It's OK to Love Bad Movies (not exactly in my field, but there were enough resonances with my academic interests that I was able to review it for the Journal of Aesthetics and Art Criticism)
• Adeline Mueller, Mozart and the Mediation of Childhood (a brilliant and exacting study, the kind that makes me realize just how lazy a historian I actually am)
• Rebecca Cypess, Women and Musical Salons in the Enlightenment
• Mark Ferraguto, Beethoven 1806
• Yoel Greenberg, How Sonata Forms (an excellent book, written in fabulous prose. The hardcore music-theory portions I'm not really qualified to comment on, but the premise, framing, and methodological discussions are revelatory and pertinent)
• Nicholas Mathew, The Haydn Economy
• Tom Beghin, Beethoven's French Piano
• John Butt, Bach's Dialogue with Modernity (can't believe it took me so long to get to this!)
• I also read a bunch of Benedict Taylor, whose stuff is pretty amazing (The Melody of Time remains a highlight)

Outside of academic writing, I had a particularly rich year of reading novels. Because I don't tend to keep track of how many novels I read per year, I don't actually know whether there were more than usual in 2022 (and in any case, I'm usually reading some sort of fiction)--but I have the sense that there were. Interestingly, although I didn't explicitly set out to read novels by women (I had tried this sort of thing in 2017 or '18, deciding to read only female novelists for a year), much of what I read ended up being by women, just by accident. There were some revelatory new finds for me:

• I read three novels by the wonderful Tessa Hadley - Free Love, Late in the Day, and The Past. I suspect that The Past may have been, in some absolute sense, the "best" of these; but Late in the Day was the one I found most moving and important, for my own life
• Octavia Butler, Blood Child - I didn't really enjoy this collection, but I'm told that my taste is faulty
• Hilary Mantel, Beyond Black - this was an amazing book, with some of the most beautiful sentences I've read in a long time. I didn't ultimately care for the story or characters, but the truly gorgeous prose was enough to keep me involved
• Elizabeth Bowen, The House in Paris - spectacular novel. If Mantel's sentences are the most beautiful I've read recently, Bowen is perfection at the level of the paragraph. A gorgeous novel, in every way.
• J.M. Coetzee, Foe - it took me about 66% of the book before I really got into it, but once this happened, it was excellent. And it's interesting to see seeds of Coetzee's trademark terseness in this early book of his. It was less sparse than his mature work, but there were passages of real beauty, where one had the sense that each word had been chosen with astonishing care, and this seemed to prefigure the experience of reading, say, Disgrace or Elizabeth Costello.
• Nabokov, Mary - again, astonishing prose, and remarkable book
• Rachel Cusk, Aftermath - is it fiction, or not? I found it moving but perplexing
• Proust, Swann's Way, in the Lydia Davis translation. This was great, and I tweeted a bit (in May? June?) about some favorite passages. I didn't love Proust when I read the cycle (in the old translation) some years ago, and though I see that this new edition is a big improvement, I still mostly didn't enjoy the text. There were some wonderful, revelatory, moving passages, though! But I found myself thinking back to that quip about Wagner, that there are lovely moments separated by terrible half-hours.
• Amor Towles, A Gentleman in Moscow - recommended by a friend. Very easy to read, feels like literary candy. I didn't like the prose style, which felt smarmy and smug...but the last third of the book had some lovely things in it, and the ending was moving.
• Orwell, 1984
• Skinner, Walden Two

I also read some wonderful non-academic nonfiction:

• Russ Roberts, Wild Problems - highly recommended, very humane book with lots of wisdom in it
• Dan Moller, The Way of Bach - also highly recommended, a rare success at turning some detailed philosophical discussions of music and art into a reader-friendly trade book with a narrative arc. A model for what I'm trying to write at the moment, and far smarter, better, and deeper than some other attempts at this genre
• Tyler Cowen, An Economist Gets Lunch - tons of fun! I was less thrilled with his Discover Your Inner Economist
• Matt Ridley, The Rational Optimist

I also read some good academic books from other fields, most notably Joel Mokyr, A Culture of Growth, Aaron Hanlon, A World of Disorderly Notions, Cowen, In Praise of Commercial Culture, and a few others.

All of this probably means that the "Recommended Books" tab on this blog needs to be updated!

Finally, I started and then abandoned a bunch of books that I just couldn't get into, despite (in all of these cases) wanting to be able to get into them. Perhaps my tastes just aren't quite up to the task, or I'm missing something?

• Ali Smith, Autumn
• Susan Blum, Ungrading
• Oliver Roeder, Seven Games (this, for instance, was a book I really wanted to love)
• Larry Lockridge, The Great Cypress Think Tank
• Jeremy Denk, Every Good Boy Does Fine
• Nick Riggle, On Being Awesome
• Laura Kipnis, Against Love
• Agnes Callard, Aspiration
• George Saunders, A Swim in a Pond in the Rain

I'm sure there were others as well, but I can't remember them.

And to end, I'll say that what I'm most looking forward to reading at the beginning of 2023 is more Elizabeth Bowen, Anita Brookner, and Maeve Brennan! Another year beginning with female novelists.

Happy new year!

ChatGPT and the animals

 Like most people I know, I've spent the last month+ experimenting with ChatGPT. I've asked it to write poetry; I've asked it arcane questions about various books and articles; I've asked it to create imaginary dialogues, podcasts, arguments, and debates between thinkers I admire from different centuries. And like most people who have put the new platform through its paces, I've come away with a mixed appraisal of its various skills. I've been impressed by its ability to handle requests that generally fall under the banner of what people think of as "creative" tasks; its poetry, for instance, especially within certain constraints, is a particular highlight. And I've enjoyed the dialogues it can come up with, spinning genuinely interesting fictional conversations between historical figures who, in reality, never met. I've been less taken with its logical abilities, which are questionable at best, and at worst make it abundantly clear that the bot isn't actually thinking, but just parroting text back at us. David Deutsch's flying-horse question gives a neat demonstration of this: the bot flatly contradicts itself and misunderstands the question in a way that can only mean it doesn't actually know what it's saying.

Playing with the bot has afforded one kind of pleasure; but nearly as fun and interesting has been to observe various public thinkers' reactions to this new technology. In particular, I've been struck by the far-reaching claims voiced by Tyler Cowen, who for much of this year has been making cryptic references to a coming revolution in AI tech that, he argues, will entirely transform how we use the internet, as well as how we work with and produce ideas. He says that ChatGPT is just the first step on this path--"bread crumbs, not dessert"--and although he could well be right, some of the individual claims he makes seem questionable to me.

For instance, he claims in a recent blog post that ChatGPT's linguistic competence likely narrows the gap between human and animal intelligence. This seems wrong to me--indeed, it's striking that one could make exactly the opposite argument just as plausibly (perhaps more plausibly). Although Tyler doesn't flesh out his version of the argument, I assume his line of thought would run something like this: ChatGPT displays enormous competency without needing to think; the competency approaches that of humans w/r/t language-use; this suggests that we aren't quite as special as we thought, and that other program-running devices like animals (which are like versions of ChatGPT, except they're programmed by evolution rather than by OpenAI) can approach the type of skills we have; thus we differ from them in degree rather than in kind.

However, that argument says much more about Tyler's "priors" than it does about humans or ChatGPT. If anything, to my eyes an argument with the exact opposite conclusion seems even more compelling! Here goes: ChatGPT shows that a program can display enormous competencies without needing to think; animals display enormous competencies, which many people want to attribute to thought; however, we can now see a demonstration that behavior that strikes a casual human observer as thought-like might not depend on thought at all; thus, it follow that animals might not actually need to think in order to perform the tasks they can perform; and if that's the case, the gulf between humans and animals is very likely to be wider than people think!

Of course, the fact that ChatGPT demonstrates that an entity can perform thought-like actions without thinking says nothing about the question of whether some other entity does need to think. So, this argument is purely analytic. CPT's ability to write sonnets doesn't ultimately settle the question of whether rats are conscious. But the version of the argument that I've laid out seems, to me, to increase the plausibility that complex behavior by non-humans can be accomplished algorithmically. Of course, just as Tyler's version of the argument says more about his priors than it does about animals, the same is true for my version. I was already inclined, intuitively, to buy the arguments of philosophers who think that consciousness exists only in people.

Tyler's other claims about the philosophical implications of ChatGPT seem equally stretched. He asks, for instance, why the aliens haven't visited us. Now that we see how easily the trappings of intelligence can be conjured by programmers, surely evolution or some other such force must have created far more intelligence across the universe than we can know? But, to build on Deutsch's point about GPT and the flying-horses, the better conclusion to draw would be that the outward suggestion of intelligent behavior is misleading, and that genuine thought remains rare, on our planet and possibly elsewhere as well.

How to write long texts (without hating yourself) - Part I

This past summer marked the completion of the second long document I've written. In the past few years I've produced a book (approx. 94,000 words) and a dissertation (approx. 100,000 words), plus a bunch of shorter pieces like articles (6,000-13,000 words each) and blog posts.

When I was a full-time professional violinist, I spent a lot of time seeking out all the self-help material I could find concerning deliberate practice and technical efficiency. I've found that the same mindset has been essential for my development as a writer. Writing, like playing the violin, is a technique. And although there may be some irreducible element of inspiration and talent in both, there are ways of approaching these activities that can make them manageable and even pleasurable. The purpose of this series of posts is to gather some ideas, both theoretical and practical, about how such processes can work for the writer of long documents such as books and dissertations.

1. Why is writing hard?

The first step to doing anything difficult is to acknowledge that it's difficult, and to understand that it's ok to experience challenges. The second step is to try to figure out why it is difficult, the better to solve the problems facing you. There are many reasons why writing is difficult--some practical, some technical, some aesthetic--but for now I'll focus on a theoretical difficulty particularly pressing when writing long texts. (My framing of this idea is based on Popper's arguments concerning empiricism, theory, and observation.)

Thoughts can take any number of shapes, and they can connect to other thoughts in any number of ways. To write thoughts down is to translate a set of largely amorphous ideas into fixed, linear form, with specific words, sentence structures, and a set order. This process is an act of interpretation: one that (like all interpretations) is carried out with an overarching theory about how and why the writing should take the shape it does. In other words, when we try to capture ideas on the page, we do not only contend with the ideas themselves; we also adopt some sort of theory about what 'the book' or 'the dissertation' will end up being--and it is this theory that allows us to know how to even begin tackling the ideas in the first place. Every sentence in a document is written with a theory about what the document is. But the difficulty is that with each sentence written, the document becomes better-fleshed-out, diverges in all its messiness from the idealized theory, and, even more important, changes the nature of the theory we might hold about the project as a whole. That is to say, every sentence is both written under a theory of 'the document' and alters the theory of 'the document' that operates in the writing of future sentences. By the time one reaches the end of (say) a 90,000-word draft, one has essentially produced a document in which most of the component parts--the paragraphs, the sentences--are part of about 90,000 different conceptual books or dissertations.

Actually, perhaps this is a simplification. The speed with which theories of 'the document' change will itself be subject to change as the draft unfolds. In my experience, the first 5-10% of the document is written under a highly consistent theory governed by an outline (assuming one is using an outline!); then change accelerates as development occurs, and each sentence/paragraph exerts a bigger pull on the overall theory; and finally, after about 80% of the draft is written, the theory of the document once again becomes more stable. Of course, it's highly unlikely that the stable theory one arrives at for the final 20% of the draft is similar to the stable theory one held while working on the first 5-10%.

So, even in this weaker statement of the problem, the basic point remains that you will probably produce a first draft in which about 80% of the sentences belong to thousands of different conceptual books, none of which is actually aligned with the finished draft, which is necessarily a jumbled mess.

2. What to do about it

This way of framing the challenge of writing may seem needlessly theoretical, but a number of practical points follow directly from it. Here are some extrapolations:

1. First drafts are necessarily awful. Many writing self-help guides (including most famously Anne Lamott's Bird by Bird) acknowledge the awfulness of first drafts. But these guides tend to do so ruefully. We are told that coming to terms with the awfulness of first drafts is a kind of self-acceptance--as though we simply need to acknowledge that we aren't smart enough to write perfect first drafts. I think that this way of framing misses the point. The problem of first drafts, as I see it, isn't that "we're not smart enough to get things right the first time"; the problem is that first drafts are necessarily awful because they are written under a huge variety of mutually-contradictory theories of what 'the document' will end up looking like. In other words, the awfulness of first drafts is inherent, and there is absolutely nothing we can do to escape it. It follows that this awfulness is something to embrace and make friends with.

2. From (1) it follows, in turn, that it's best to write an entire first draft before starting to revise or edit. People take many different approaches to this practical question; and even for me, a hardened embracer of bad-first-drafts, there is always a temptation to try to revise during the initial drafting process, for no better reason than the sheer frustration one feels at producing a 200+ pages of terrible, cluttered, aimless, unpolished prose.

However, there are a few reasons not to give in to the temptation to edit. The first two reasons are practical, and have nothing to do with the philosophical framework I'm outlining. First, the more one writes, the more momentum one gathers. It's often easier to push through the pain to the end of a document when one writes in a single sweep of productivity. Second, the mindset needed to produce words on the page is very different from the mindset needed to revise what's already there. The biggest difference between the two mindsets is that the first requires a real lack of inhibition and a confidence and belief in one's powers of creativity, whereas the second requires coolheaded discernment, self-criticism, ruthless questioning of one's ideas, etc. It's rarely easy to switch back and forth between the two mindsets, so it makes sense to disentangle them, accomplishing as much as possible under the first mindset before adopting the second.

In addition to those practical points, there are some theoretical reasons to write the entire first draft before embarking on any revisions. The most important is, simply, that revisions, just like first-draft writing, are carried out under a 'theory of the book'. In order to revise any sentence, paragraph, or chapter, you need to have some idea of how it fits into the whole, and what angle you're revising for, and why; otherwise, you don't know what to keep, what to expunge, what to fix, or how. And it's inherently impossible to know these things without having a complete theory of 'the book' as a whole. And, returning to the initial point, because that theory is changing until near the end of the first draft, revisions can't really happen until then. So it makes sense to put off revising until the initial drafting has taken place.

Note that the idea of writing the entire manuscript before revising does not translate into the process of writing individual chapters. Many people I know attempt to write and polish individual chapters before completing the rest of the manuscript. (This is particularly true of academics, who often use individual chapters as conference papers or articles, and write and polish one-at-a-time.) However, there are a few problems with this approach too. First, it presupposes that you know where in the document each argument or example will belong. In everything I've written, ideas move around significantly during the course of revisions, so that paragraphs that began in (say) chapter 1 might wind up in (say) chapter 5 or 6, or vice versa. Polishing individual chapters makes such moves difficult. Second, even when arguments don't move around between chapters, when individual chapters are polished before the entire draft is complete, it can be difficult to integrate nuances of arguments between the chapters. Finally, on a psychological level, as Ayn Rand points out in The Art of Nonfiction, it can be demoralizing to go from a polished final draft of one chapter to a messy first draft of another, and this can bring its own set of emotional and psychological challenges. Again, better to get all 90,000 words of ideas out of one's system, in all their unpolished messiness, before deciding what to do with those ideas and how to shape the next stage of the document.

3. Edit the complete manuscript in iterations. Editing, like writing, will change your sense of what the document is saying. Therefore, editing should be carried out in multiple stages, since it is unlikely (or impossible) that any one round of editing will create the final version of the document. Rather, the final version of the document emerges from various layers of editing: the revision of chapter 1 will necessitate changes in chapter 6; but those changes in chapter 6 will also cause changes in chapters 1, 2, 3, 4, and 5, which will in turn cause changes in chapter 6. These large-scale processes of bringing the book "into agreement with itself," so that the individual components have ceased to contradict each other or repeat after each other, is a fully absorbing mode of editing--and it's difficult to accomplish it while also tinkering with sentences and word-choice. So, separating out the various rounds of editing can be helpful in making each one as efficient and productive as possible.

How many rounds of editing are needed? This is probably a matter of taste and personal preference. The way I think about the issue is mostly psychological: as Eviatar Zerubavel advises in The Clockwork Muse, the ideal number of drafts is high enough that no individual draft introduces the pressure of getting things perfect all at once, but low enough that the process doesn't seem endless. Of course, these calculations will be different for different people. Zerubavel recommends doing 7 drafts; so far, for long documents I think I'm a 5-draft writer. (By contrast, when I write articles, I typically need only 3 or maybe 4 drafts, since it's comparatively easy to be coherent and integrated when you're dealing with only a single thread of argument.)

My process for going through the drafts looks something like this. First, I write a complete, messy, awful first draft. I attempt to do this as quickly as possible, both because I want to ride the wave of enthusiasm, and because I know that the writing will need a lot of revision anyway, so there's no point in taking even a minute longer than is absolutely necessary to get the awful first draft on paper. I attempt to write approximately 1,000-1,200 words a day during the first draft stage, and I stop myself once I've reached this goal, even if I'm midway through a sentence (especially if I'm midway through a sentence!) because I don't want to exhaust my ideas and have nothing to say the next day. By stopping myself at a given word limit, no matter how excited I am or how well things are flowing, I guarantee that I'll be able to hit the ground running the next day. At the rate of 1,000-1,200 words a day, you can write a 90,000 word draft in approximately two and a half or three months.

Having written the draft, I then take stock. I think about the flow of argument, I think about ways to re-order the pieces, and I think about what might be missing, which sections need to be lengthened or excised, etc. I then write a second draft, moving significantly more slowly than in the first draft. My goal for the second draft is always to make sure the paragraphs and chapters are roughly in the right order. At this stage, I still do not worry about the awful prose. The sentences are a mess, but I try to get the ideas in vaguely the right place in the manuscript. When writing my book, I also experimented with retyping the entire manuscript in a new, blank document from scratch for the second draft (following the advice from Zerubavel). I found this to be extremely helpful for two reasons: first, because printing the MS and retyping it from scratch gives me simultaneously a written, fixed first draft to work on as well as the freedom of a blank document. I feel safe making changes and trying experiments knowing that I won't lose work I've already done. Second, it allows me to (subconsciously) revise some of the sentence-level prose as well, since it's very difficult to retype a terrible sentence without making tiny tweaks that improve it. Although fixing the prose isn't my goal at this point, I can make little changes in the process of typing the new draft that I probably wouldn't think to make (and perhaps wouldn't even notice) if I were just picking through the same document as the original first draft.

Having written the second draft and gotten the ideas vaguely in the right order, I then repeat the process at the level of the prose. First I take stock; then I retype the entire document once again.

For drafts nos. 4 and 5, I no longer retype the document; at this point, the main ideas are in the right place, and my goal is to make the prose flow as well as possible.

To be continued, with thoughts on outlining, stamina, and other aspects of the process of writing long documents.

Utopia and its discontents

What comes to mind when we hear the word "Utopia"? Perhaps a particular set of texts, mostly from the worlds of philosophy or political theory. More likely, we think of a place--one that is Edenic and flawless, a paradise for its inhabitants, albeit one that is impossible to actually construct. But what I've always found strange about the idea of Utopia is the divergence between these idyllic associations that the noun-form of the word has accumulated, on the one hand, and on the other, the negative associations called up by its adjective, "Utopian." Popper and many of his followers in the tradition of classical liberalism leveled this term as a criticism against Marxism. Someone who engages in Utopian thinking chases fantasies of societal perfection while subjecting actual people to all manner of injustices. Indeed, one of the problems with Utopian pursuits is that they seem to offer a blank check to those held in their sway; since if one is pursuing what one genuinely believes to be an infinite good end, then any means needed to achieve that end will come to seem justifiable.

These critiques of Utopianism apply to a surprising number of dystopian worlds--not only the Marxist societies that were the subject of Popper's arguments, but the Sparta-like police-state described in Plato's Republic (a subject of different discussions by Popper!) as well as the fictional Oceania of Nineteen Eighty-Four. Indeed, encountering these dystopias, we may be struck by the fact that they are Utopian for at least some of their inhabitants (for instance, for those who directly benefit from the societal structures, i.e. those at the top of the ladder who are able to maintain power by controlling the thoughts and actions of others). And even some ostensibly Utopian fictions--for instance, Skinner's Walden Two--may easily come to seem distinctly dystopian when viewed from another angle--say, in the case of Walden Two, when viewed as something akin to what Oceania might feel like to its happier, Big-Brother-touting inhabitants. (Even if we dismiss this particular possibility for Walden Two, and claim that the society described in those pages actually is "perfect" in some sense, the fact would remain that it seems like an incredibly dull place to live, and thus would be Utopia only for a people with specific personalities....)

The fact that utopianism can so easily seem like the other side of the dystopian coin makes me wonder: is there a deeper attribute that all of these imagined societies share? I think the answer is yes--and that attribute is: these are societies entirely devoid of problems. In the case of the worlds that are supposed to look like utopias (Republic, Walden Two, and so forth), the problem-free environment is meant as a "feature"--one that will make life easy and pleasant for the inhabitants. For the worlds that are supposed to look like dystopias, the problem-free environment, tenaciously enforced through (in Orwell) torture, brainwashing, Newspeak, the rewriting of history, the erasure of truth, and so forth, is precisely what makes the fictional societies look so bleak. But, despite the different valence these and other authors give to their worlds, the underlying logic is the same. A superficial form of "contentment" is maintained by preventing people from seeing problems and trying to solve them. The enforced avoidance of problems may be well-intentioned, yet whatever the rulers' intentions, the picture that emerges is of a static society, in which growth, discovery, novelty, excitement, and further exploration and understanding are impossible. This underlying logic is why the very idea of Utopia as it is frequently understood is, in fact, so dark.

There is a twist. If dystopias (and dystopic "Utopias") aspire to be problem-free worlds, then it would follow that an actual Utopia--not a society simply called that, but a society that actually does allow for flourishing in its fullest sense--would be full of unsolved problems, which the inhabitants would pursue freely and as they pleased. This dovetails with Robert Nozick's view, described in the final part of Anarchy, State, and Utopia, of Utopia as not a single kind of place, but a "meta" world in which vast numbers of different societies and associations, each tailored to the different interests of different kinds of people, would emerge. Nozick's Utopia, too, is full of problems, both for those individual societies (whose task, he says, is partly to facilitate the discovery of what good societies might look like--thus implying that this is not yet known, and thus can't be perfectly instantiated even in principle) and for the meta-Utopia, whose problem will involve, for instance, the peaceful interactions and integrations of its constituent societies. Of course, most importantly, the problems that arise in such a meta-Utopia and its constituent Utopias would also exist on the level of individual inhabitants, who, unlike the inhabitants of the Republic, Oceania, Walden Two, and so many other "Utopian" worlds, would be free to imagine different, perhaps better worlds, and thus would be faced constantly with the personal problem of how to reshape their circumstances in pursuit of those dreams.

Of course, one cannot help but be struck by how decidedly unglamorous this account of Utopia sounds. Indeed, a "society full of problems, in which people go about trying to solve their problems" sounds not only unglamorous, but downright mundane. No wonder fictionalized false-Utopias and Dystopias seem to outnumber fictionalized real-Utopias.

However, here too, there is a twist. Yes, there are a handful of actually-optimistic Utopian fictions (some Le Guin comes to mind); but, there is also a sense in which most fiction is Utopian in the positive meaning of the term--that is, most fiction is about everyday people with lots of problems trying to do stuff that will make their lives better. In this view, the genuine champions of the Utopian vision are novelists like Jane Austen and George Eliot, whose characters may "rest in unvisited tombs," but who nonetheless show us a model of striving in which problems are present, but problem-solving is not thwarted by some oppressive autocrat. Indeed, perhaps this explains my abiding love of rom-com movies, whose plots tend to track the same processes. Rom-coms often highlight the problems and dissatisfactions of individual people, yet they also do so in a way that is fundamentally comic, and in which the overcoming of those problems is allowed even as success is never guaranteed.

Perhaps this is also why my intuitions about works like Così fan tutte or Into the Woods diverge from the intuitions of other listeners with whom I've compared notes. Many over the years have been tempted to see in these narratives a fall from grace--a loss of perfection and a confrontation with the tarnished reality of human life. And yes, that is one view: no character in these two works survives unscathed. But in both cases, what we witness is a process in which problems denied in Act I are recognized in Act II--and, in both cases, the end of Act II brings a loosening of authority and control, and with it a sense that, perhaps, problem-solving in the future will be possible for the characters. In Così, Don Alfonso steps back and allows the lovers to pursue their problem-filled lives without interference. Perhaps, after the end of the opera, the lovers' relationships will grow into authenticity. And in Into the Woods, the narrator is, literally, vanquished, as are the Witch's meddling, supernatural powers. What we are left with at the end is, in some sense, a fallen world. In another sense, however, it is a world that, for the first time in the musical, will allow its inhabitants to "just pursue [their] lives." That their strivings will never come to an end--that Cinderella begins to say "I wish..." even as the music stops--is itself a point of hope for the nascent, optimistic Utopia we see taking shape.