The Letters Column: A Remediation of the Daily NATION Newspaper, 1974 ‒ 1978

Abstract

Digitizing newspapers has a decades-long history, particularly within the library settings that cater to needs of researchers and scholars. However, as scanned images, navigating such archives to locate news articles of interest, mobilize items by topic, authorship or correspondences is often disconnected and tedious without bibliographical indexes. Yet, digitization of this resource has benefited even less from such indexing due to a number of reasons including the volume of collections and nature of its content. The implication of which this resource and its inherent variety of genres remain obscured and underutilized. Here, the reputed letters-to-the-editor column becomes the focal interest onto which content is examined under bibliographical, media and technological perspectives. The end result of this thesis demonstrates the desired navigation strategies for users of digitized newspapers that takes the textual content, topics, authors, and the image itself into consideration. The thesis is interdisciplinary by design and therefore examines the data from a variety of fields. Its historical context and material not only provides arguments that have generated and sustained past digitization projects, but reaffirms the newspaper as an important research resource. Tied to the research goals is automated indexing and how this approach is complicated by the type of content under exploration. To frame this properly, the exploration process for the user has to incorporate elements of the newspaper that made it useful in contemporary times as well as the interconnected interface found in web pages. The process of realizing this uses a number of computational tools and techniques (including computer vision, page layout analysis, and text analysis) to extract letters in the Mailbox column which are then indexed using a subject vocabulary. This research attempts to give full treatment to an obscured text in the expectation that besides preservation, the digitized content is also useful already. The outcome illustrates interfaces enriched with referential metadata including bibliographic references, and capable of connecting threads of correspondence within a given collection.