¶ 2 Leave a comment on paragraph 2 1 The historian sits down at her desk, flicking on the lamp. She begins to pore over a stack of badly photocopied court proceedings from late 18th century London, transcribing the text. As she works, she begins to notice interesting patterns in the language used to describe young female prisoners. ‘I wonder….’. She turns to the Old Bailey Online and begins to search. Soon, she has a corpus of a thousand court proceedings featuring women prisoners. She downloads the complete transcriptions, and loads them into Voyant Tools. Moments later, she has a graph of key words, their collocations, and their frequencies over time. A suspicion grows. She turns to MALLET and begins to look for the underlying semantic structure in the records. The algorithm, after much exploration, seems to suggest that 23 topics account for the majority of the words in each text.
¶ 3 Leave a comment on paragraph 3 1 But what do these topics, these lists of words, mean? She begins to explore the relationship between the topics and the texts, uncovering a web of discourse, seemingly surrounding the moral duty of the state towards women prisoners. She takes this web and begins to explore its formal characteristics as a network – what words, what ideas, are doing the heavy semantic lifting? – while at the same time, she runs the RezoViz tool on the corpus to extract the named individuals and organizations in the document. She begins to query the social network that she has extracted, and is able to identify sub-communities of women and warders, children and men, zeroing in on a smaller set of key individuals who tied the prison community together. Soon, she has a powerful, macroscopic sense of not just the discourses surrounding a century of women’s trials, but also of the key individuals, organizations, their connections. She looks at the clock; two hours have passed. Satisfied, she turns off her historical macroscope, her computer, and turns once again to the transcription at hand.