Validating judgments of perspective embedding: Further explorations of a new tool for literary analysis

Previous work ( Whalen, Zunshine, & Holquist, 2012 ) has shown that perspective embedding ("she thought I left" embedding her perspective on "I left") affects reading times for short vignettes. With increasing levels of embedment 1–5, reading times rose almost linearly. Level 0 was as slow as 3–4. Embedment level was determined by the authors, but validation by others is desirable. In Experiment 1, we trained 12 literature students to make embedment judgments. Their judgments correlated highly with ours (.94 on average) and agreed exactly in the majority of cases (74.5%); almost all were within one (94.2%). In Experiment 2, judgments of the first three paragraphs of "To Kill a Mockingbird" ( Lee, 1960 ) yielded a lower level of agreement; literature uses subtle means for introducing perspective embedment, and individuals differ about including them. Assessment of perspective embedding, and exploration of sources of disagreements, provide new tools for analyzing literature.