[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [AUDITORY] Perceptual basis of evolving western musical styles

Hello Robert,

I agree with you that it is very important to get feedback from the musicological and music-theoretic communities. Even though I work with a musician on these matters, it is really hard to cover all technical details of the three aspects of the research: mathematical, cognitive and musical.

Regarding the composers, we noted what you said, and reported it in the Supporting Information. It is true that the Baroque period seems to be under represented. However, if Bach copied the Italian work, then some of the statistics should kept.

On the other hand, the melody perception is an issue by itself. It is in fact something we can not control, since we don't know how data was pre processed. Depending on the frequency of the cases you mention, it could perfectly bias the analysis. We believe that it is not the case, because we made two controls. The first one is published in the Supporting information with an alternative corpus, and the second one is a preliminary result with a corpus which consists only of melodies.

Anyway, that is not a guarantee of anything, but seems to point in the same direction as our intuition.

Lastly, and not less important, I want to thank you for the work you put in your mail. Not only for the examples, which were really good, but for the criticism on the corpus. As you say, this feedback is indispensable for the growth of this line of research.

Best wishes for you,


On Thu, Jun 6, 2013 at 1:39 PM, Robert O Gjerdingen <r-gjerdingen@xxxxxxxxxxxxxxxx> wrote:

        Thanks for your paper on the "Perceptual basis of evolving western musical styles." I was especially happy to see your engagement with the IR theory of my own teacher, Eugene Narmour.

        I know it can be tiresome for people to complain about the corpus, but after downloading the Peachnote corpus I did notice some very odd things (admitting that I may have misread or misunderstood the Excel files for each musical era). There appear to be 600+ composer names for "Romantic" and just 4 for "Baroque." Of those four, three are German protestants, one is French, and none are Italians, even though 17th and 18th-century musical style was almost entirely determined by Italian composers (Bach, who accounts for %53 of "Baroque" pieces in the corpus, assiduously copied Italian manuscripts to learn their famous art, whereas the Italians did not even know he was alive). The "Classical" set, by contrast, has 15 names, almost half of which most musicians would put in the "Romantic" box:  Schubert, Weber, Rossini, Mendelssohn, Auber, etc.  It may be worth pointing out that only "Romantic" was a term known and used as a style designation by any of the composers from 1700-1900. The Romantic-Classical-Baroque terminology only becomes widely used in the 1920s.

        Interval counts are not a straightforward calculation in musical scores because of the frequent streaming effects and the importance of discontiguous tones. Take, for instance, the following example by Bach from his first suite for solo cello:

Perceptually, there are a couple of large intervals at the very beginning, and then one realizes that the "musical" intervals are within each of three auditory streams. For ex., the B3 at the beginning connects to repetitions of that tone and then ascends one step to C4. This is the small-interval phenomenon that you mention as characteristic of this era, but would the Peachnote corpus have registered this emergent perceptual phenomenon?

You also mention Gestalt pattern recognition. Take, for instance, the following excerpt from a Haydn string quartet:

Listeners tend to hear the tonic in the melody of measure 1 "go" to the second scale degree of m. 2, and then to the 3rd scale degree of m. 3.  In Gestalt terms, the evident similarity of those three tones helps to group them as a salient foreground against the background of more rapid tones. Note, however, that there is no small interval connecting degree 2 to degree 3. The "step" that connects them is an emergent property of the figural complex.

I've gone on at length because, since corpus studies offer so much to computational musicology and music theory, it is important that pioneering studies like yours get some feedback from the musicological and music-theoretic communities.

Best wishes,
Bob Gjerdingen
Northwestern Univ.