摘要：We provide a methodological contribution by developing, describing and evaluating a methodfor automatically retrieving and analysing text from digital PDF annual report files publishedby firms listed on the London Stock Exchange (LSE). The retrieval method retains informationon document structure, enabling clear delineation between narrative and financial statementcomponents of reports, and between individual sections within the narratives component.Retrieval accuracy exceeds 95% for manual validations using a random sample of 586reports. Large-sample statistical validations using a comprehensive sample of reportspublished by non-financial LSE firms confirm that report length, narrative tone and (to alesser degree) readability vary predictably with economic and regulatory factors. Wedemonstrate how the method is adaptable to non-English language documents and differentregulatory regimes using a case study of Portuguese reports. We use the procedure toconstruct new research resources including corpora for commonly occurring annual reportsections and a dataset of text properties for over 26,000 U.K. annual reports.
