This dissertation presents two studies on the complexity of qualitative accounting disclosures. Reporting is considered to be more complex as it becomes longer and/or less readable. The first study examines investors' reactions to disclosure readability. The SEC's emphasis on the use of plain English is designed to make disclosures more readable and more informative. Using an experiment, I find that more readable disclosures lead to stronger reactions from small investors, so that changes in valuation judgments are more positive when news is good and more negative when news is bad. Drawing on research in psychology to explain this result, I predict and find that processing fluency from a more readable disclosure acts as a subconscious heuristic cue and increases investors' beliefs that they can rely on the disclosure. While I do not find that more readable disclosures directly increase perceptions of management credibility, I do find evidence of an indirect effect operating through feelings of processing fluency. In supplemental analyses, I find that investors who receive more readable disclosures revise their valuation judgments to be less extreme when they are explicitly made aware of the potential for variation in readability. I discuss potential explanations for these revised valuation judgments. The second study examines managers' reporting choices with respect to disclosure complexity. Prior research finds that the reporting complexity of qualitative disclosures increases as firm performance deteriorates, and that reporting complexity may dampen investors' short-term reactions to news. While some argue that managers intentionally increase reporting complexity to hide bad performance, others suggest that bad performance is instead simply more difficult to describe in fewer and more readable words. I use a controlled experiment to investigate these explanations, and find greater support for the idea that bad performance is inherently more difficult to describe than good performance. Counter to arguments made in the archival literature, I find that individuals provide longer, but more readable, disclosures when performance is bad and they are given a goal of reporting strategically rather than reporting accurately. This contradicts the claim that managers intentionally obfuscate poor performance in order to mitigate negative reactions to bad news. My study also provides evidence in a controlled setting on how other linguistic choices vary with performance and goals.