Working with Historical Textual Data: Preliminary Results from Applying Survival Analysis to the Old English Poetic Corpus

Survival analysis (SA), a statistical method traditionally employed in fields such as medicine(e.g., modeling patient outcomes based on treatment efficacy or risk factors), has only recentlybeen applied to linguistic research. Studies, such as Van de Velde and Keersmaekers’ (2020)evolutionary model of lexical longevity and Vogelsanger’s (2023) analysis of rhyming patterns,highlight its potential. However, its applications to the Old English poetic koiné remain virtuallyunexplored. By building on prior applications of SA in evolutionary linguistics, we aim to offer aquantitative framework to interrogate which linguistic features correlate to the obsolescence ofOld English poetic vocabulary over time. To this purpose, we apply SA to investigate arepresentative segment of the Old English poetic corpus, including the Vercelli Book (28.96%lexical density) the Nowell Codex (11.55%), the Exeter Book (23.57%), the Junius Manuscript(24.82%), and the Paris Psalter (18.94%).Our dataset consists of 182 words, of which 119 are classified as “dead” (i.e., no longerin use), yielding a survival rate of 34.62% and a death rate of 65.38%. Key variables, such asword frequency, z-score, syllable count, manuscript provenance, morphosyntactic roles, and thesemantic categories defined by the Historical Thesaurus of English are analyzed to identifypatterns in lexical survival. The results reveal strong correlations: syllable count (p = 0.03) andprovenance (p = 0.03) are significant predictors of word survival, while Historical Thesaurus ofEnglish categories show an even stronger correlation (p = 0.001). These findings suggest thatsemantic fields—particularly those related to the mind (e.g., emotion, cognition), society (e.g.,kinship, governance), and the world (e.g., nature, physicality)—play crucial roles in determiningthe longevity of Old English poetic vocabulary, likely due to their cultural resonance. Byadapting SA to this context, this study introduces a novel quantitative approach to understandinglinguistic change. It also extends the methodological scope of digital philology, addressingNichols and Altschul’s (2012) call to realize the field’s interdisciplinary aspirations.