Computational Methods and Russian Verse

The tasks of quantitative prosodic analysis—characterizing the metric and phonetic constants and tendencies that allow the reader to recognize poetry and its norms of well-formedness, and subsequently to identify departures from those norms as points of semantic interest—seem ideal for adaptation to machine-assisted analysis, which provides powerful tools for finding and counting what a user tells it to count, given appropriate markup. “Appropriate markup” is a serious caveat—the markup of poetry has traditionally taken a great deal of human intervention. However, Russian poetry, by virtue of the structure of the Russian language and its history of versification, lends itself relatively more readily to automatized markup.

The Anna Akhmatova Markup Project has been a testing ground for the development of such tools. Akhmatova’s position within a transitional period of Russian poetry provides a range of forms against which to test the tools at hand, in preparation for the “distant reading” of more diffuse literary fields—for example, newspaper poetry or epigonal schools. Given a digitized text and a reference work that allows one to mark the position of stressed syllables, it is possible to derive rhythm and a phonetic representation of the text, which in turn allow for the computer-assisted derivation of meter, rhyme, and other forms of phonetic repetition. This proof-of-concept site has underscored the challenges presented by the hierarchical structure of Extensible Markup Language and its related specifications, which requires significant translation from the human reading experience into appropriate algorithms. Indeed, such translation is likely to always remain incomplete, but the results significantly reduce human labor and direct attention to places of interest, and the process itself is informative about the logic of Russian versification and reading it.

Our Resources