Composer’s Notebook #6: Bowing and Articulation in Violin Playing


At the beginning of the year, I promised that I would try to write more regularly. This has clearly not been achieved! In my defence, studying mathematics full-time requires much dedication, patience, and practice — not unlike learning a musical instrument. But now I have time to write since I have completed my semester 1 exams.

(Main article is below)

A few months ago, I was asked to contribute an article on composition for the October 2017 issue of Stringendo, which is the bulletin of the Australian Strings Association. (Incidentally, I have already contributed an article on focal dystonia, which appears in their April 2017 issue). The theme of the upcoming issue is “Add-ons”, which I assume refers to “other” activities in which string players are engaged, besides the usual performing and teaching. Since it is known that I am a self-taught composer, I was asked to write a piece describing my approach to composing for the orchestral stringed instruments. This is far too broad of a topic, so in the end, I decided (with the editor’s approval) to limit the scope of my contribution to writing for the string quartet. Obviously I cannot reproduce the piece on my blog at this time, but I do want to share some thoughts specifically about composing for the violin in this blogpost.

For obvious reasons, writing for violin comes most naturally to me, and as a result, I find it impossible to give impartial advice for composing for the violin. Hence, in this post, I will not attempt to give general “tips and tricks” for writing music for the violin. Instead, let me express what I want to do using this metaphor: imagine “violin music” as a self-contained universe, in which inhabit many violinists and composers. Then I am one of the many tour guides of this world, and what I decide to show you will depend on my (necessarily limited) travel experience in this world, and on what I find most interesting. You only need to ask a different tour guide to see potentially entirely different features and landscapes (or should that be soundscapes?). In this particular tour, I would like to focus on some details of bowing and articulation.

Anyone who wants to write serious music for orchestral stringed instruments should be knowledgeable about how the bow is used. After all, it is the primary means of sound production, and how well a string player performs is determined by their mastery of the art of bowing. I wonder if non-string players find the proliferation of different terms (mostly in French and Italian) used to describe bowing techniques rather confusing: détaché, legato, spiccato, sautillé, flautando, portato… and so on. We will later clarify exactly what these kinds of instruction mean. An essential question to consider is how specific should the composer be with regards to bowing. Often in baroque and classical scores, very little bowing instruction is specified, but that is because there is a well-defined style of playing associated with music of that era, and an experienced string player will intuitively know what to do. For composers nowadays, I suggest giving enough bowing information so that the performer knows, or at least can infer, what kind of sound you wish to achieve. (Of course, if you’re writing something with lots of extended techniques, you’ll need to specify all of these very accurately, but I will be thinking more “classically” in this blogpost). How might one go about this? Instead of just going through the list of bowing techniques, I suggest it is more helpful to think physically about what bowing entails, and connect that to the notation. Therefore I must emphasise my opinion that is impossible to talk about bowing without including articulation.

Every kid who learns violin starts by playing “separate” notes (détaché) and “slurred” notes (legato). As a provisional definition, let’s say that détaché means something like “each note is played on a separate bowstroke” and legato is “several notes are played on the same bowstroke”. So détaché is a faster bowstroke (one note per bow), while legato is a slower stroke (many notes per bow), right? This is fine in many cases, but the reality is that bowing is a lot more flexible than such a definition would imply. Consider the following simple exercise. CN6_fig1

Naïvely, we might identify that the first two bars demonstrate legato playing, while the next passage consists of détaché playing. This is actually correct, but perhaps not for the reasons suggested by the provisional definition above. Now think about what the bow is actually doing — you should realise that in both passages, the bow does exactly the same work! In each passage, it makes a total of two “return trips” from the frog to the tip and back. So this shows that whatever détaché and legato mean, it cannot just be a bowing pattern, otherwise there would be no difference between the two passages above. A more accurate definition is as follows: détaché means that the notes have to be articulated separately, while legato refers to smoothly connecting the notes. The emphasis has been shifted to how the notes are articulated, and thus the terminology does not necessarily indicate a specific bowing pattern. Rather, it describes a manner of “speaking” with the instrument, which is how I think about articulation. Consider the following exercise:


“Now you’re being obnoxious,” I hear you grumble, “what difference does that make? You didn’t even write a slur in the second exercise!” The point is that I expect the violinist to make two different kinds of articulation. In the détaché exercise, the notes should be clearly articulated as separate, but in a “neutral” kind of way (i.e. without any kind of accent). I think this would be the default way to play such a passage, if no other context was given, so the designation détaché is a bit redundant. But the legato marking is definitely not redundant — in this exercise, I expect the violinist to make good effort to disguise the bow change, so that the notes sound smoothly connected, as if they could be played all in one bow, or sung in one breath.

This begs the question: what is the point of writing a slurred line then? I find that this quite a fascinating and somewhat difficult point of discussion. The short, unsatisfying answer is that it depends. Most of the time, a slur over a group of notes indeed indicates that the group should be played in the same bow stroke, but this is not the only function. There are many examples even in standard repertoire where this clearly cannot be the case:

Brahms: String Quartet No. 3, Op. 67, 4th movement, bars 53-56

Given that the tempo for this section is usually around crotchet = 52, have fun trying to fit all those notes in the same bow (especially the cellist)… If you are still wondering, then this more extreme example should clear the doubts:

Brahms: Symphony No. 1, Op. 68, 1st movement, opening

(These were the two examples that stood out to me. Purely by coincidence, I chose works with consecutive opus numbers!)

Very often in Brahms’ string parts (and also in Richard Strauss and Wagner), slurs indicate phrasing as well as legato. This is no surprise to a pianist of course, but perhaps it is an overlooked fact when it comes to composing for strings. While we know the phrasing, Brahms does not give explicit instruction on how to bow the passage. In fact, it is not at all obvious in general how to construct a good bowing — those pesky sequences of up- and down-bows which have probably sparked many arguments in rehearsals. However, I think both performer and composer can agree that a good bowing is one that finds the best compromise between respecting the phrasing and articulations indicated, and being relatively uncomplicated to execute. In many cases, like the opening passage of Brahms’ First Symphony as shown above, the policy is to leave the bowing to the performers. Of course, if you want a very specific effect, then by all means specify a bowing, but in general, I tend to prioritise phrasing, and give the performer some degree of freedom (and hence responsibility!) in selecting a bowing. Let us consider one more example before moving on:


Mendelssohn’s beloved Violin Concerto in E minor needs hardly any introduction, but I draw attention to the opening passage to illustrate further how it is more useful to think about détaché and legato as describing articulation first (and then the bowing follows). One could play the first two detached B’s on separate bows, and likewise the E in the following on a separate bow, but traditionally violinists have performed a more “fluid” bowing, such as the one shown below:


The important thing to note is that the two B’s should still sound detached, even though they are played on the same bowstroke (up-bow). Likewise, if we use the particular bowing above, the down-bow E in the third bar should sound detached from the slurred B to G… although perhaps not too separate. I believe many violinists will not mind including an audible but tastefully executed portamento between the G and E… but I will not delve into the details of interpretation for now.

The difficulties discussed above are present due to the overloaded function of the slur — it has evolved to somehow convey legato, bowing, and phrasing. The dot above a note is another symbol that is overloaded, especially in violin music, and it can be quite a headache to work out what exactly is meant when there is a combination of slurs and dots! Fortunately the ambiguities are often resolved in context — often by additional instruction from the composer, or appeal to a well-known performance tradition. To discuss these notational difficulties will take us too far from where I intended to go with this blogpost, thus I will now focus the discussion back to bowing technique. As I stated at the beginning, it is instructive to consider the physics (in a qualitative way of course). So far, we have only considered détaché and legato, which are strokes that are played on the string (alla corda). I think most kinds of bowing can be characterised by the following fundamental parameters:

  1. Type of contact between bow and string
  2. Contact point along the string
  3. Contact area along the bow
  4. Bow speed

The first parameter, “type of contact”, actually consists of several “sub-parameters”:

  • The amount of friction between the hair and the string. This can vary from hardly any friction (a “floaty”, or airy sound, flautando) to so-called “over-pressure” (excessive friction), which is sometimes found in modern works. The result is a harsh, scratchy, distorted sound (e.g. as shown on the cello in this video). Speaking of modern techniques, you could extend this definition to include col legno (playing with the wood of the bow) — that is certainly another type of interaction between the bow and the string.
  • The length of the contact period, that is, long (e.g. legato) or short (e.g. spiccato, a “bouncing” stroke played off the string), and everything in between;
  • The kind of attack on the note. An analogy is often made with consonants in spoken language, so assuming standard English, “k” would be a stronger attack than “b”, which is stronger than “m”, for example.

The second parameter, “contact point along the string”, can vary from being over the fingerboard (sul tasto) to being at the bridge (sul ponticello). These are special effects, but it turns out that we don’t need any more specifications. The contact point along the string will vary naturally throughout the course of regular playing, in response to different dynamics and expressive markings, so it is unnecessary and probably impractical to prescribe the contact point exactly.

The third parameter refers to where along the bow a particular bowstroke is executed. For slow legato playing, for instance, the contact area will often be the whole bow, while for spiccato, it is generally a small section of the bow around the middle third to upper-lower-third (i.e. the upper portion of the lower third). I don’t think these designations are exact, i.e. I don’t think anyone has tried to set numerical measurements to define the “middle third”, but string players will know intuitively what is meant. Sorry to everyone else! It is more useful to see it in action:

Finally, variation of bow speed is not usually something composers need to specify, since it should vary naturally, like the contact point along the string. However, I include it here for completeness, as well as the fact that it is a very important aspect of bow technique, and hence useful for composers to be aware of it.

The four parameters above pretty much span the space of all possible bowings (and extended techniques can be incorporated by extending the definitions… this is sounding a bit mathematical now). Certainly we have characterised all the bowing techniques one is likely to encounter in standard violin repertoire, including chamber music and orchestral parts too. We can finally look at the some of the terms given to bowing techniques! What I will show is that any bowing-related term can be easily described in detail using the four fundamental parameters. In this way, I hope that the reader will gain a detailed insight into the how the bowing is executed, as well as the resulting sound.

  • Détaché and legato we have already discussed at length
  • Spiccato — from Italian spiccare, “to separate”, which doesn’t give much information. This is in fact a “bouncing” stroke, and generally the contact between bow and string is light. It is played most often in the middle-third to upper-lower-third, as mentioned above. The interaction time between the bow and string is obviously short. The bow starts slightly above the string, then makes contact briefly before being lifted up again. When done well, the notes should sound crisp and clear. A great (and notorious) example is the Scherzo from Mendelssohn’s Midsummer Night’s Dream, which is traditionally played this way.
Mendelssohn: 1st violin, Scherzo from Incidental Music to A Midsummer Night’s Dream, Op. 61
  • Sautillé — from French sautiller, ” to hop”. The key difference here is that the bow is made to jump by its own accord — it may be said that this is a “passive” stroke — and hence this can only be achieved when the speed of the notes is sufficiently fast. See this article for more information. The spiccato on the other hand is “active” — each note is articulated separately by the bow arm. The sautillé is lighter than the spiccato, and the contact area along the bow is even more restricted, since the player has to find the “bounciest” bit of the bow, which is generally around its balance point. Once again, the articulation should be crisp and clear, and due to the speed, this is often a very virtuosic, impressive effect. We have another example from Mendelssohn, this time from the third movement of the Violin Concerto in E minor. As in the preceding example, note that sautillé is not an explicit instruction here, but it is traditionally played this way. The speed of the music and also the instruction pp leggiero suggest this manner of playing quite naturally.
Mendelssohn: solo violin part, Violin Concerto in E minor, excerpt from 3rd movement
  • Ricochet — self-explanatory (probably?). The player throws the bow onto the string, letting it bounce freely, the result is like a rapid-fire spiccato or sautillé. However, this stroke is usually played on a few notes at a time, since this action can be controlled precisely. I include a famous excerpt from Rossini’s William Tell overture by way of example. Again, there is nothing specifically asking for ricochet in the score, but it is traditionally played that way, and it certainly makes sense due to the speed of the music. Another way this stroke is used commonly is in arpeggiation of chords. With the appropriate impulse from the bowing arm, the bow can bounce across the strings easily. Paganini’s Caprice No. 1 consists almost entirely of this highly virtuosic bowing.
Rossini: famous excerpt from the overture to William Tell. The repeated semiquaver notes are the ones that are traditionally played as a controlled ricochet bowing.
Paganini: opening of Caprice No. 1 from 24 Caprices for solo violin, Op. 1
  • Staccato — Italian for “detached”. This is a very general term indicating that notes are to be played shortened and detached. For stringed instruments, this is done on the string, i.e. the bow is at first stationary on the string, and then released. The initial friction between the bow and string creates the attack, which is like a consonant. With some skill, staccato can be played pretty much anywhere along the bow.
  • Martelé — from French, meaning “hammered” (Italian version is martellato). The physical mechanism is the same as for staccato, but the attack is strong — i.e. start with a significant amount of friction between bow and string. It is often not necessary to write this specifically — for example, if a violinist is instructed to play staccato notes at the dynamic fortissimo, what will happen is martelé. Unlike staccato, which has the built-in implication that the notes are short, the martelé works very well on long notes too. However (see below), I feel that martelé is limited to louder dynamics.
  • Marcato — from Italian, meaning “marked”. This is another imprecise term, with the same physical mechanism as staccato. I think many would agree that it is somewhat stronger than staccato, in other words, the attack is sharper, more pronounced, and perhaps the decay of the note quicker. But I don’t think it is as strong as martelé. In fact, staccato and marcato will work in any dynamic marking (e.g. you could write pp ma marcato), but to me, pp martelé looks very bizarre.

For a modern case-study, many of the above bowing techniques can be found in Stravinsky’s Dumbarton Oaks Concerto (see the end of the blogpost).

Remark: It is useful to compare the attack of a bowed note to pizzicato. To pluck the string, the finger must “catch” the string firmly, then release to allow the string to vibrate, Similarly, in staccato and martelé, the bow “catches” the string (with varying amounts of friction) and then releases.

Of course this is not an exhaustive list, but I think I’ve covered the most important ones. I could have just started with this list, but then this entire blogpost would be pointless. My aim has been to equip the reader with a physical understanding of bowing, rather than merely collect a bunch of terms in Italian, French and German. I hope the usefulness of this approach has been made clear. For composers in particular, I suggest that thinking about the four main parameters of bowing identified above enables a closer connection between the composer’s conception and the act of performance. From the performer’s perspective, the worst kind of piece to play is one that is so terribly difficult and unrewarding, because the composer has little understanding of what is physically involved in playing the instrument. Where the composer and performer come together is both in real life (yes, composers and performers should actually talk to each other) and, especially if the composer is long dead, in music notation and terminology. If the composer has a good understanding of how to play the instrument, and if the performer has a good understanding of notation and terminology, in spite of the inherent limitations, a meaningful and fruitful collaboration may be established.

There is no better way to learn about bowing than to see and hear it in action. Here are two recordings (I think of the same performance) by the Passacaglia (after Handel) for violin and viola by Johan Halvorsen, one of them with score, the other with video (so you can see how the bowings are performed). Along with the Stravinsky, these videos contain pretty much all the bowing techniques mentioned in this blogpost — how many can you identify? And can you find examples where a single technique, such as spiccato, is played in different ways?

Bonus example!

Some of the most effortless ricochet I’ve ever seen — and the left-hand pizzicato is bordering on magic.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s