14
1 Janos Sipos (www.zti.hu/sipos) János Sipos (Institute for Musicology, HAS), www.zti.hu/sipos Using artificial intelligence for comparative musicology on Eurasian folk music – Following the folk music of a community to different areas

1 Janos Sipos ( János Sipos (Institute for Musicology, HAS), Using artificial intelligence for comparative musicology

Embed Size (px)

Citation preview

Page 1: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

1

Janos Sipos (www.zti.hu/sipos)

János Sipos (Institute for Musicology, HAS), www.zti.hu/siposUsing artificial intelligence for comparative musicology on Eurasian folk music – Following the folk music of a community to different areas

Page 2: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

2

Important task of the archives are the reliable and durable saving of the material and the service for the general public. However there is a similarly important task: to help the researchers and scholars in scientific analysis of the material.

What do scholars do with the folk music material in the archives? Let us overview the main phases of the folk music research. In the 19th and the early 20th centuries the universalistic mode became predominant. It was searching for the origin and the evolution of everything, and from it developed the comparative musicology. In contrast to the comparative musicology came the (American) ethnomusicology, with the same main question and sometimes the methods of the „social anthropologists”: how do individual cultures function. Though these days comparative and analytical musicology is sometimes considered “old fashioned”, this branch is regaining more and more strength. And as we will see computer programs developed by Zoltán Juhász may give enormous help to it.

With Zoltán Juhász we continue to build a large computer-database of musical transcriptions - in other terms we have been developing a transcription-archive. This archive completes well the other parts of the archives and makes the analytical research much easier.

First we convert the melodies to a form understandable by computers using Finale or ABC note writing programs. Though this is a lengthy work, afterwards the transcriptions can be used in many ways. What is more, recently the international research community created lots of available and downloadable digitized transcription-databases, which increases the radius of our action considerably.

Page 3: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

3

Let me illustrate one application of this database and computer program in one of my researches concerning the music of a mystic Islamic sect, the Bektashis.

Since 1999, my wife, Éva Csáki, and I collected more than 1000 melodies in 24 villages from Turkish women and men of the Bektashi faith, whose grandparents migrated from the Balkans to the European part of Turkey. By the end of this research series it seemed so that we had reached our goal, and recorded the majority of their religious and secular songs.

The nomadic and semi-nomadic Turks did not become Muslims at any one time but rather gradually, over centuries. They adopted some Sunni, Shiite and mystic elements of Islam while continuing to cling to their traditional shamanistic beliefs and practices. Bektashism is a syncretistic folk religion connected to nature; they worship mountains, trees and heaven. We can consider it a Turkish form of Shiite religion mixed with Sufism.

Let us see an excerpt from their religious ceremony. It illustrates the fact that structural analysis reflects only one side of the big picture. For lack of time this time I can not introduce the cultural and social aspects. However in our book in preparation "The psalms and folk songs of a mystic Islamic order" we introduce the musical culture of this group from different aspects. [1] As an example let us have a glance into their religious ceremony.

1st video: Bektashi zikr from Thrace

Page 4: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

4

Now we turn to our proper subject: to the comparative analysis of the Bektashi songs. I call your attention to the uniqueness of the following method. We do not "only" compare two folk

musics, or search the existence of some specific melodies or melody type in different folk music. Now we project a whole classified material to the folk music of other people searching if there are similar melody groups, and if there are, how strongly are they represented in the folk music in question.

Page 5: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

5

A) Melodies traceable back to a single short section (I-II)

I. Melodies built up of motives rotating around the middle tone of a trichord

II. Melodies traceable back to a single short line or motif (e.g. laments)

B) Melodies traceable back to two short sections (III-IV)

III. First line is undulating or ascending (often AkA form and different cadences)

IV. Two short stagnant, descending or hill-like sections with small compass and 2, (b)3 or 4 cadences.

C) Melodies with four short sections and (1) main cadence (V)

D) Melodies with four or more sections (VI-X)

VI. Low melodies with 2/b3 (2) x cadences and higher melodies with 4/5 (2) x cadences

VII. Low and higher melodies with b3(b3)x cadences

Page 6: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

6

VIII. “Psalmodic” and descending melodies with 5/4 (b3) b3/1 cadences

IX. A special “Chanakkale” melody group

X. Melodies with characteristic line or bar-sequences

XI. Disjunct melodies

E) One- or two sectioned tripodic melodies (XII)

F) Melodies with "doomed" form (XIII)

Page 7: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

7

High beginning, high end Low beginning, high end

High beginning, low end Low beginning, low end

We tell the program the name of the database of the folk musics we want to compare, in this case that of the Bektashis and that of the Bulgarians. Then we determine the desired degree of the similarity, and we tell the program to compare whole melodies or only specified sections.

The program generates a map with green point for the Bektashi melodies and purple points for the Bulgarian ones. The program helps to find melody parallels as well.

Then the researcher examines the melody parallels, and if he finds them convincing, he copies them into a file. Based on the quantity and character of the parallels we can draw far-reaching conclusions on the relation between the folk music in question.

The interface of the program

The computer aided comparison

Page 8: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

8

The area of the research: Bulgaria- Thrace (Bektashis) – Anatolia – Azerbaijan - Mongolia

Page 9: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

9

Let start off our journey.

Anatolia

Bektashis are Turks, whose ancestors came from Anatolia to the land of Bulgaria then went back to the European part of Turkey.

A very close musical connection would not be a surprise and really, most of the Bektashi melody types do have Anatolian parallels.

This confirms the Bektashi’s Anatolian origin and the survival of their Turkish cultural identity.

On this and on the next slide I show Anatolian-Baktash melody parallels form the VIIIth melody class (see above)

Anatolian (a)-Bektashi (b) melody parallels

2.nd video

Page 10: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

10

Bektashi -Anatolian melody parallels (cont.)

Page 11: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

11

Azerbaijan Let us go a bit further to the East. Azeri and Anatolian Turks are linguistically related, and there are lots of common event in their ethnogenezis as well. However in Anatolia the Turkish tribes (arriving in different waves from Inner Asia) Turkicised the former Byzantium, while in Azerbaijan these tribes settled on a Caucasian and Iranian substratum.

Azeri folk music joins to Bektashi music only with a few melody parallels. These similar melodies have small compass and very simple melodic line. There are two types of these parallels. One consists of a short sections with narrow compass and of its variants.

The other type is based on two sections moving on C-D-E-F-G pentachord, the second section being one second lower then the first one. This latter structure can be found in the Hungarian, Anatolian and other folk music as well. Let us now listen to an Azeri lament showing these features.

3.rd video: Karabahi sirató

Azeri-Bektashi melody parallel

Page 12: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

12

Mongolia

As Turkic and Mongolian people belong to the Altaic language family, musical connection between them is quite possible to imagine.

However among the more then two thousands melodies at hand only 4 or 5 show lesser or greater similarities in the melody movement. As example we show one very rare Bektashi parallel to a typical Mongolian fifth-shifting melody type on ex.4.

Bektashi-Mongolian melody parallel

Page 13: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

13

Bulgaria

Let us turn back to a people living closer to the Bektashis, to the same place where they came to Turkey from. We know that some 1000 years ago Bulgar-Turkic people took part in the Bulgarian ethnogenezis, and then merged into the Slavic majority. Later Oghuz tribes arrived in many waves, but they did not merged into the Bulgarians.

The comparative research shows that, in spite of the long coexistence, Bulgarian and Bektashi music has almost nothing to do with each other. Bulgarian Turks, among them Bektashis preserved their musical world intact, and did not let Bulgarian influence in.

There are only a few Bektashi and Bulgarian parallel. This kind of songs can be found in great quantity in Bulgaria, but only one or two types represent them in the Bektashi repertoire. We see a representative of these melodies on the next example.

Bektashi-Bulgarian parallel

Page 14: 1 Janos Sipos ( János Sipos (Institute for Musicology, HAS),  Using artificial intelligence for comparative musicology

14

I hope, this paper cold prove that we have a very effective tool with which we may compare whole folk musics. This may open new perspectives in the field of comparative musicology