Sight-Reading and Improvisation

In my post on Unpracticed Skills, I spent some time discussing the main findings of a paper written by K. Anders Ericsson (author of Peak) and co-author Lehmann about sight-reading. I wasn’t surprised by those main findings, but I was a bit struck by one of the subsidiary results: the significant correlation between skill at sight-reading and skill at improvisation (.626). At first glance, this is a bit surprising because improvisation (making up music) is in some sense the opposite of sight-reading. Why should these two skills be particularly connected?

Ericsson’s study does not, however, appear to be an isolated result. Jennifer Mishra did a pair of meta-analyses in 2013 and 2014, surveying the literature on sight-reading and looking in the first case for skills found to be correlated with sight-reading performance, and in the second case for interventions found to impact sight-reading performance positively. In 2013, she found that improvisation was one of the factors most highly correlated with sight-reading ability (quite a lot more highly correlated, shockingly, than the construct that included number of years spent sight-reading!) Moreover, in 2014, she found that interventions that involve practice with improvisation and composition (dubbed collectively “creative activities”) are one of relatively few intervention categories associated with a significant positive effect on sight-reading performance. This is the same meta-analysis in which a variety of common-sense interventions, including practice clapping out rhythms, were found to have no significant effect.

I’m not any kind of statistician, but I will say upfront that some of these results strike me as hard to accept on their face. Even if improvisation is deeply connected to sight-reading, it seems hard for me to believe that it is more connected than either experience sight-reading (“just sight-read” seems likely to have some merit at least) or rhythm drilling.

That said, the more I think about it, the more I think a connection between sight-reading and improvisation might make sense.

One reason for this is that improvisation might improve power to predict the next notes of musical sequences by teaching musicians to create musical sequences of their own. This reminds me of the PD I’ve ben taking over the past few days on the Science of Reading, which finds (reasonably enough, to my mind) that practice writing can improve reading skill. In the meta-analysis cited by my training documents, Graham and Hebert explain that “Because writers need to make their assumptions and premises explicit as well as observe the rules of logic when composing text, this presumably makes them more aware of these same issues in the material they read.” Likewise, practice creating musical sequences might be expected to improve a person’s understanding of similar musical sequences found when reading, and improved comprehension should lead to greater reading fluency in reading music as in text.

Another plausible reason is quite apparent in my interview with Leo: it appears that good sight-reading often directly entails improvisation, albeit improvisation within tightly bounded constraints.

you really don't have to play your part as written. You just have to play your part so that everybody else thinks that it's right.

Russell says the same thing, of course, though it’s less the focus of his remarks about sight-reading:

It almost doesn't matter exactly what you're playing; as long as you're playing the right chord, you're getting most of the information. With sight reading, you're not looking for 100% accuracy anyway. You're looking to capture the gestalt.

In short, based on the combination of evidence and reasoning, I’m prepared to accept that it might be worth trying to practice improvisation en route to improving sight-reading. This leads to the next obvious question, which is “How am I supposed to practice improvisation?”

I need to look into this farther, because so far, the answer would have to be, “I have not the faintest idea.”

I was never taught improvisation, and I am not aware of improvisation as a part of any tradition of modern [Western] classical instruction for string players. However, this includes some important qualifications. Notably, my friend Jesse (whom I consider to be a brilliant sight-reader, providing some anecdotal support for this theory) was taught classical improvisation on the piano and later studied Klezmer improvisation on the clarinet. A very quick survey of the internet suggests that classical improvisation used to be a major part of classical music training. Wikipedia, for instance, summarizes the case as follows:

Throughout the eras of the Western art music tradition, including the Medieval, Renaissance, Baroque, Classical, and Romantic periods, improvisation was a valued skill. J. S. Bach, Handel, Mozart, Beethoven, Chopin, Liszt, and many other famous composers and musicians were known especially for their improvisational skills… However, in the 20th and early 21st century, as common practice Western art music performance became institutionalized in symphony orchestras, opera houses and ballets, improvisation has played a smaller role.

I am intrigued, and am likely to try looking into this further, which may yield results to be discussed in another entry. Meanwhile, if you know anything about improvisation for string instruments in a classical style, I’d love to hear anything you know.