Extracting Drum Patterns in Traditional Folk Songs Among East Japan

Introduction
The musicologist Fumio Koizumi analyzed the beats of traditional Japanese music (consisting of a pattern based on the alternation of strong and weak beats) and found that the strong and weak beat corresponded to the concept of
omote (the front beat) and
ura (the backbeat) of traditional Japanese music, respectively. He revealed that space has been created for a set of two notes with those front beats and backbeats. In addition, Koizumi classified the rhythm pattern of Japanese traditional music into two types: the freeform Oiwake style and the clear rhythms of the Yagibushi style. Since the Yagibushi style has a rhythm pattern with a constant interval, this style tends to appear in song accompaniments using Japanese drums (Koizumi, 1958; 1983).

However, studies that capture the rhythm patterns of traditional Japanese music theoretically, through empirical analysis, have not been conducted. At present, not only rhythm studies but also studies that capture the melodic characteristics have not been performed either; therefore, it is unclear how the characteristics of Japanese music propagate and transform.
In this research, in order to empirically capture the rhythm patterns hidden in traditional Japanese music, we focused on folk songs (having the most primitive characteristics of Japanese music) and analyzed the rhythm of the Japanese drums. This study aimed to compare the rhythm patterns of Japanese drums in the folk music of East Japan quantitatively and to clarify the rhythmic features of each region.

Overview of the data
We analyzed 589 songs using Japanese drums from the East Japan folk songs published in “
Nihon Min’yo Taikan” (
Anthology of Japanese folk songs) (see Fig.1). Specifically, we selected all songs using one of the following:
oh-daiko (big drum),
ko-daiko (small drum),
koshi-daiko (waist drum),
daibyoshi (wooden clapper), and
taru (barrel).

Fig.1: Map of East Japan and number of songs used from each region

Procedures
We converted all 589 songs into MusicXML format and created a music corpus. Musical notes and rests of the drum parts were extracted from each XML data and were replaced with a symbol string. Then, N-gram analysis was executed on this symbol string, and frequent fixed-length patterns were counted. Additionally, by creating co-occurrence networks between notes and conducting network analysis, we clarified the connection of elements constituting rhythm patterns in each region. By comparing the above analysis results with the findings of Koizumi, we clarified the characteristics of Japanese drum patterns in the folk songs of East Japan, which have been passed down over generations.

Results and Discussions
Comparing the results of the N-gram analysis, we found that the rhythm patterns of Hokkaido and Hokuriku were different from the other three areas. From the results of this summary, Hokkaido’s rhythm pattern had characteristics of notes and rests being combined at the same degree, unlike in other areas. In Hokuriku, from the result of the unigram, the pattern with consecutive notes was overwhelmingly high, which differed greatly from other regions. Also, rhythm patterns combining 1/8 and 1/16 notes occurred frequently, indicating that there was a tendency to tick small rhythms, unlike the other three regions of Honshu (Tohoku, Kanto, and Tokai). These results were similar to features found in chanting and Buddhist music, which strike at regular intervals. It can be pointed out that there is a high proportion of people who believe in Jodo sect and Jodo Shinshu (Buddhist sects) in Hokuriku.
The result of the degree centrality for the whole of East Japan showed that 1/4, 1/8, and 1/16 notes were central elements within the drum patterns. This result overlaps with that of Fraisse, showing that 80 percent of the melodies in Western music are occupied with only two kinds of note values (Fraisse, 1982). Regarding rests, we found that 1/2, 1/4, and 1/8 rests were high in terms of degree centralities, but that the 1/2 rest was low for betweenness centrality. Therefore, unlike other rests, a 1/2 rest was more likely to have the role of concatenating coherent rhythm patterns than as an actual component of the rhythm patterns in traditional Japanese folk songs.
As a future task, we will extend the song data and conduct an analysis for each region of West Japan. By conducting a comparative analysis with the features of neighboring countries, we will develop basic research to demonstrate the relationships, as well as the propagation and transformation of rhythm patterns, between regions.

Bibliography

Fraisse, P. (1982). Rhythm and Tempo. In Deutsch, D. (eds),
The psychology of music. New York: Plenum Press, pp. 149-180.

Koizumi, F. (1958).
Study on traditional Japanese music I. Tokyo: Ongaku-no-Tomosha.

Koizumi, F. (1983). Breathing ethnic music, Tokyo: Seido-sha.