Detecting mental illness in 18th century writing: the case of King George III

Language is seen and heard everywhere, as people communicate with one another in spoken and written format, apparently without effort. Ease of use, however, belies the complexity of the tasks of combining letters or sounds into words, words into grammatical sentences, and sentences into coherent discourse. Because using language involves so many interacting aspects of brain function, the study of spoken or written language output has shown it to be a reliable marker of brain malfunction. Analyses of language using automated, computer-based methods in schizophrenia have demonstrated the impact of disordered thinking on language use. Computer-based analyses of the written language output of literary and political figures have found evidence of declining intellectual function in individuals with degenerative dementia, years before symptoms first appeared.

Such approaches have not previously been extended to the study of remote historical figures, yet the survival, from the 17th century onwards, of large quantities of written output raises the possibility that novel insights into the mental states of prominent historical figures could be derived using modern language analysis techniques.

The recurrent mental illness of King George III of Great Britain has been well documented in popular culture. Recent reappraisal of the historical evidence suggests that it was most likely due to a psychiatric condition (mania) rather than, as previously thought, to the inherited metabolic disorder porphyria. Large volumes of correspondence between King George and senior parliamentarians survive in manuscript form and transcripts of the majority of these letters have been published. Gaps in the correspondence correlate with known periods of mental derangement, but letters immediately preceding these intervals and those in the early weeks and months after writing resumed, may shed light on the occurrence, and perhaps eventually the nature, of the king’s mental ill health.

IMG 0155 1316 edited 1 copy edited jpg

One of many hundreds of documents written and signed by King George III, and preserved in the Royal Archives at Windsor Castle. In common with all the King’s communications, this one is marked not only with the place (Windsor) and day (May 9th, 1789), but also the time (6:35 pm) of composition.  

The principal aim of this project is to create a mathematical model of 18th century written language based on a large sample of published writing that survives from that period. The model will be constructed using a novel computer-based language analysis technique known as latent semantic analysis. The model can then be used to obtain accurate measures of textual characteristics known to be sensitive to mental illness, and to plot their changes over time. Preliminary results comparing a measure of similarity between sequential pairs of letters before and after a well documented period of illness in 1788-89 have suggested a loss of coherence between sequential pairs of King George’s letters, in comparison with those of his correspondent (the Prime Minister of the day, William Pitt the Younger).

This project not only crosses boundaries between different branches of science (medicine, psychology and computer science) but will provide historians with a novel scientific methodology. The techniques and resources developed will therefore enrich historical study, while addressing a question that has already gained considerable public interest: the nature of King George’s illness.

Dr Peter Garrard
St George's, University of London

Peter was awarded a Research Project Grant in March 2012; providing £105,082.