Identifying speech acts in a corpus of historical migrant correspondence

A full account of the pragmatics of personal correspondence requires speech act annotation, and as manual annotation of large datasets can be extremely difficult, this study proposes to use an automated speech act tagger developed by the first author. It was originally designed for use with business emails; however, the latest iteration of the tagger can be applied to other datasets–such as personal correspondence–providing a useful resource for the corpus linguistics community. In this study, the speech act tagger is tested on a collection of letters written by Irish migrants at the end of the nineteenth century. After discussing issues to do with the digitisation, transcription and annotation of historical migrant correspondence, the article will report on the results of this trial study, demonstrating how the tagger can perform with some success even on corpora with very different characteristics. Although the dataset used for this trial study is small, the findings show the potential for carrying out this type of analysis across larger digital archives allowing for different datasets to be compared, taking into consideration sociobiographic variables such as the author’s sex, class and role within the notional familial hierarchy.