Akademik Veri Yönetim Sistemi
academia.edu, cilt.1, sa.1, ss.1-6, 2025 (Hakemsiz Dergi)
Through the application of the “trigram melodic contour analysis” first described in [Yarman & Sethares et al. 2019, “An investigation of the role of diatonic functions in the seyir of Turkish makam music: case of ‘Hicaz family’”] to the SymbTr-2025 corpus [Karaosmanoğlu 2025, “SymbTr v.3: Türk makam müziği sembolik bilgiler veritabanındaki yenilikler”], we extracted the context- independent stylistic signatures of 3199 digitally transcribed traditional Turkish music pieces in 4 different tuning resolutions. The concomitant date-label metadata tagging for 1781 of these works were done by hand, since the time period in which a composition or improvisation came to be is gen- erally not indicated on the score. After this, a date-label we manually associated with a given piece was checked by 27 distinct WEKA algorithms against similar date-labels associated with other piec- es’ trigram feature vectors to yield confusion matrices. These consitute algorithmic guesses on what the machine-learning thinks the date-labels should be instead. We then claim to have identified a per- sistent trend of overemphasis in the clustering of diversely dated Ottoman-Turkish works under par- ticularly the 1940-1959 period bin. This strongly suggests a widespread bout of ‘Hüseyin Saadettin Arel era’ revisions that we recall herein as “Arelization”. Our present communication outlines the in- novative music information retrieval and symbolic music processing technicalities behind these re- sults, insofar as the proposed methodology is poised to detect and challenge ideologically motivated counterfeiting across all musical cultures in the world.