Difficulty in Anglo-American poetry: a linguistic and empirical perspective

This project sets out to develop a model for a study of difficulty in poetry as systematic and nuanced as possible. In doing so, it endeavours to make a significant theoretical and practical contribution to the fields of stylistics, poetics and literary theory. Throughout the twentieth century, the notion of difficulty in poetry has never ceased to interest linguists, literary theorists, psychologists and researchers in education. The popularity of the notion among non-specialist is equally significant, as it is not uncommon for readers to justify their lack of interest in poetry on the grounds of its supposed difficulty. Notwithstanding this, the dynamics of text-reader interaction – the defining trait of difficulty, as argued in Chapter 1 – remains notably underexplored. This thesis addresses this gap in the literature by (a) providing a psychologically plausible and linguistically sound account of difficulty; and by (b) unifying under a coherent framework the insights offered by a large body of materials – from critical readings of literary works to anecdotal evidence, from psychological models of comprehension to controlled psycholinguistic experiments. In terms of methodology, a linguistic, text-based approach is intertwined with an empirical, reader-based one. This combined effort leads to an in-depth analysis of a set of poems from both perspectives (Chapter 3 to 5). Such a qualitative approach allows for the identification of textual and readerly components typical of difficulty. On the textual side, I identify twenty-four features, called linguistic indicators of difficulty and affecting all the linguistic levels – graphology, syntax, lexis, semantics and text structure. Based on scholarly remarks and experimental evidence, these indicators are likely to hamper readers’ comprehension and thus increase the processing effort they require. These two main readerly dimensions of difficulty I qualify as online (i.e. affecting the processing effort in actual reading) and offline (i.e. affecting the post-reading understanding of a poem). In turn, online and offline difficulty are cued by observable readerly behaviours (e.g. interpretive uncertainty, slowed-down reading, statements of rejection) that are explored in Chapters 4 and 5. Overall, difficulty is viewed as a response phenomenon that has a strong linguistic motivation. For reasons of focus and critical consistency, the model is applied to twentieth and twentieth-first century Anglo-American poems only. This temporal restriction acknowledges the critically established connection between difficulty and modernism (e.g. Adams 1991, Adamson 1999, Diepeveen 2003). The case studies from Chapter 3 to 5 focus on Wallace Stevens, Ezra Pound, Susan Howe and Jeremy H. Prynne as representing different aspects of difficulty. Chapter 6 extends this purview to a larger corpus, featuring Gertrude Stein, T. S. Eliot, E. E. Cummings, Hart Crane, Charles Olson, Dylan Thomas, John Ashbery and Charles Bernstein. All these poets have been deemed ‘difficult’ by other critics, so the corpus rests on an intersubjective agreement that was missing in previous accounts. The hope is that the model proposed will be fruitfully extended and applied to non-Anglo-American literary traditions as well as to poetry written in earlier centuries