Confirming the themes and interpretive unity of Ghazal poetry using topic models

We apply topic modeling to classifying the genre of Ghazal, a form common in Persian poetry. We show that a classifier based on automatically-generated topics exposes important information with only a small performance penalty: the top discriminative topics can be manually aligned with themes prevalent in the associated genres, as identified by scholars of literature. We also weigh in on a long-standing debate about the interpretive unity of Ghazal. In particular, we show evidence that, on the average, Ghazals seem to have interpretive unity at the level of the full poem, as opposed to just at the level of the couplet. Our dataset is a collection of almost 18,000 Ghazal, comprising over 3 million words. The collection contains poems from 30 different poets, and spans nearly 900 years (1080 A.D.–1968). Generative models tend to perform worse than discriminative models in classification tasks. The hope is, however, that a generative model provides additional insight, by allowing decisions to be understood with reference to the generated explanation. We show that this hope holds true for topic models [1] in two specific cases in the domain of Ghazal, a form common in Persian poetry. First, we show that, for a 3-class genre classification task, the most discriminative topics correspond to known themes prevalent in the associated genres. These themes have been identified by scholars of Persian poetry, and we manually align computationally-extracted topics with these human-identified themes. Second, we show that classification via topics can provide some insight into actual scholarly debates. Here we focus on interpretive unity, a debate among scholars of Ghazal regarding whether the poems are best interpreted at the level of a whole poem, or at the level of a couplet. [2, 3] We train a 30class topic-model classifier on the Ghazal and use it to do poet attribution (the identification of which poet wrote which poem). As expected, poet attribution is poor when the model is trained using couplets as the document unit, but reasonably good when the model is trained using poems as the document unit. Interestingly, however, when couplets are rearranged (within a poet’s poems) to create “scrambled” Ghazals, and the model is trained using these fake Ghazal as the document unit, performance is extremely poor. We take this as evidence that, on the average, Ghazal are coherent, and have interpretive unity, more at the level of the poem than the couplet.