Sunday, August 5, 2007

78. The Power of Cantometrics: 3

I'll now describe two more Cantometric parameters, lines 4 and 5, each of which present somewhat different problems. Line 4, labeled "Vocal Org." or "Vocal Organization," asks the coder to choose between five basic modes of musical presentation in the voice(s): 0 (with a slash through it) for Uncoordinated, M for monophony, U for Unison, H for Heterophony and P for Polyphony -- numbered 1, 4, 7, 10 and 13 respectively. As with the first line, on those rare occasions where a purely instrumental performance is being encoded, slash/0 is treated as a null case. Otherwise, it is coded when two or more singers are performing independently from one another, in an uncoordinated manner. Monophony is coded when only one vocalist is heard at a time, singing one note at a time. Unison is coded when a group of singers perform the same melody "in unison," i.e., all in the same rhythm, and with no harmonization. Heterophony is coded when all singers are singing essentially the same melody but not in unison, i.e., with some coming in a bit later than others, to produce a kind of staggered effect -- or some voices more active or ornamented than others, etc. Polyphony is coded when there is a relatively consistent use of multipart singing, i.e., where at least two different pitches (other than octaves) are heard at once. This can be either "chordal," with two different notes sung simultaneously counting as a "chord," or "contrapuntal," where at least two different parts are sung with independent rhythms. (The various types of polyphonic singing are coded on line 22.)

Line 5 is labeled Tonal Blend-V, i.e., Tonal Blend-Voices. Unlike lines 1 and 4, this is a scaled variable, where the coder is asked to rate the relative degree of vocal blend on a 5 point scale. Point 1 is the null case, when there are either no vocalists at all (very rare), or no vocal group -- i.e., monophony (see above). Point 4, "b," indicates minimal blend, where there is no attempt by the singers to match one another's voices, resulting in a harsh effect. Point 7, "b" underlined, stands for "medium" blend. Point 10, "B," stands for "good" blend and point 13, "B" underlined, is "maximal" blend.

Line 5 is an example of the sort of rating often criticized as overly "subjective," and also "ethnocentric," so I'd like to say a few words regarding both issues. It should go without saying that what one person may hear as "harsh" blend, another may hear as "medium" or even "good." And this might, indeed, reflect the rater's own cultural background. All raters are therefore prepared, as part of the training process, by being asked to listen to a great many examples of different types of vocal groups, blending in a variety of ways, so they will get a feel for the overall range of possibilities, worldwide -- a process intended to serve as a counterbalance to any biases they might have had previously with respect to any of the Cantometric parameters.

Raters are asked to rate blend on a relative, not absolute basis, depending on where any given performance seems to fall within the above mentioned range. The terms "minimal," "maximal," "harsh," "good," etc. are intended as general guidelines, not absolutes. While terms like "harsh" and "good" may seem judgmental and potentially ethnocentric, they are simply the most convenient and straightforward terms we have for describing various degrees of blend. It's important to understand that in certain cultural settings what we would call "harsh" blend is regarded as good, in the sense that it is what is expected in a particular tradition. All musical traditions are assumed, in the context of the social sciences, to be based on what is considered good and proper within their traditional setting.

Scales of this kind are routinely used in similar types of rating system in the social sciences generally, and have proven reasonably accurate in a wide variety of different types of research. In an independent consensus test conducted under the supervision of Norman Markel, at the University of Florida (see Chapter Five, of Folk Song Style and Culture), Tonal Blend achieved a reliability rating of 92%, indicating that different raters from different backgrounds can indeed, when properly prepared, achieve a significant degree of consistency. The mean reliability for all 13 Cantometric parameters tested was 84.7%. Subsequent consensus tests have produced similarly impressive results.

No comments: