Computationally-Assisted Analysis of Select Songs by Sergei Prokofiev

Sergei Prokofiev's songs for solo voice and piano hold a special place in his oeuvre both biographically and stylistically. The works were composed almost continuously throughout his entire lifetime, with only a short lapse in output from 1922-1931. The chronological consistency with which his songs were produced stands in stark contrast to other genres like his concertos, sonatas, symphonies, oratorios, and film music, which were either subject to large chronological gaps between subsequent works or were composed exclusively in his Soviet Period (1936-1953) resulting in separate snapshots of Prokofiev's changing compositional practice instead of a more panoramic and dynamic view. Therefore, his songs for solo voice and piano offer a perspective of his stylistic evolution not available when examining his works from other genres. Even so, very few scholarly works engage with more than a handful of these songs within the same study. Eugenia Joukova attempts to examine all of his solo vocal songs from a historical perspective, noting the subject matter of each song and how it relates to the surrounding political and cultural environment, but her study is unique, as no other scholarly investigations, either historical or analytical in their focus, attempt to address all of Prokofiev's works in this genre in a single project. This dissertation not only demonstrates that an analytical investigation of these songs can lead to a deeper understanding of Prokofiev's compositional style, but also fills a gap in the literature and contributes to the well-established body of Prokofiev scholarship. Producing a thorough, in-depth analysis of a large corpus of music can prove challenging and, at times, unattainable due to various factors. The sheer size of a musical corpus can discourage these studies from being undertaken, and the time and resources required for hand analysis of a large corpus can result in incomplete analyses even after a significant effort has been made. To overcome these challenges, this dissertation will leverage computationally-assisted analytical tools, which are guided by a bi-directional analytical procedure that moves fluidly between the analysis of the underlying song texts and the accompanying song melodies to identify shifts and patterns that distinguish Prokofiev's pre-Soviet (1891-1936) songs from his Soviet (1936-1953) songs. Further, the analysis will contextualize these findings historically, drawing connections between the socio-political environment in which Prokofiev worked and his compositional style to demonstrate that in both stylistic periods, his compositional decisions were often driven by the philosophical dogma of the day. The findings of both the computationally-driven text and music analyses will also be compared to songs by other Russian and Soviet composers, to further identify musical features that are statistically distinctive of Prokofiev's compositional style. The result is a robust, holistic analysis of Prokofiev's songs that not only advocates for the use of computationally-assisted analytical tools in the field of music theory, but also fills a gap in the scholarly literature and provides a clearer picture of Prokofiev's compositional evolution utilizing a corpus of music that, up to this point, has received little analytical investigation.