Umeå University's logo

The digitization of historical records not only creates new research opportunities but also challenges. Documents may be searched online and digital tools may be used for finding patterns in large datasets. Yet, the quality of digitized material is often low and it may be difficult for researchers to evaluate this material when hidden behind interfaces. My aim in this article is to explore a digitized newspaper and the associated problems and opportunities. My exploration is guided by the following question: What was written about the electrical telegraph in Aftonbladet in the 1830s, 1840s and 1850s? The files of the digitized Aftonbladet for the years 1830-1862 contain massive amounts of corrupted words generated by the OCR engine. Another problem is that the tool used for auto-segmentation has merged separate text items into single text blocks and split long texts into smaller parts. The digitization process is not a neutral text transfer from one medium to another, it is a process generating new texts never printed in the original newspaper. In order to find as many (corrupted) versions of the words "electrical" and "telegraph" as possible, the text files were searched with a Levhinstein distance of two (allowing for two letters to be added, replaced or missing). The words that were found were used as search words to identify texts blocks reporting on the electrical telegraph. To find themes in the texts, a co-occurrence search was carried out, identifying clusters of words frequently co-occurring. Previous research has focused on utopian ideas associated with this technology and the telegraph as an immaterial form of communication obliterating physical space. The clusters identified in the digital analysis suggest that more mundane topics dominated the newspaper: the electrical telegraph was described as a material form of communication controlled by bureaucracy, depending on and emphasizing geographical conditions. Digital databases make new research possible, but the low quality of digitized texts results in uncertainties concerning the results. Researchers using these databases need to be aware of the digital forms and interfaces regulating research practices.