Stress in Meter, Rhythm, and Rhyme, Part 2

• In Event: Computational Poetics

Abstract

Our machine-assisted identification of the place of stress in plain-text Russian verse relies on a full-text database version of A. A. Zaliznjak’s Grammatical dictionary of the Russian language that we have been developing. We have previously demonstrated how, when completed, that dictionary will provide efficient lookup of stress placement in all inflected forms of approximately 100,000 Russian lexemes, making it possible to translate input in natural Russian orthography (that is, without indication of stress) into a text that exposes the place of stress for subsequent analysis of meter, rhythm, and rhyme. But what of wordforms that do not appear in the dictionary at all, wordforms that can be stressed in more than one location, and wordforms that were stressed differently in the poetry of earlier periods than in the twentieth-century Russian regarded as authoritative in the Grammatical dictionary? We have thus far only speculated about the methodological approach to these questions, and in this session we demonstrate and discuss new implementations of those methods in an analytical pipeline. The needs addressed in these new resources include the assessment of metrical schemes on the basis of incomplete input information, the projection of those schemes back over the ambiguous data, the machine-assisted identification of dictionary lacunae and ambiguities, and the progressive enhancement and expansion of our dictionary resources based on feedback from those processes.

Authors

• David J. Birnbaum, U of Pittsburgh

• Elise Thorsen, Novetta