Some (statistical) considerations on the use of inscriptions for the study of “sociolinguistic” variation in Latin: the problem of “frequency”

This article addresses the issue of defining “frequency” in quantitative/statistical sociolinguistic inquiries of the epigraphic sources. While the subject has thoroughly been discussed in the field of diatopic research, it has received limited scholarly attention in the analysis of diaphasic (and diastratic) variation. To bridge this gap in the literature, the present study will concentrate on the instances of the spelling variation with < E > vs. < AE > in a corpus of 1,650 “Christian” inscriptions from Rome (ca. 300 CE – ca. 600 CE). The chosen phenomenon is particularly well-suited to the problem under discussion, since the monophthongisation of /ae/ was most likely a completed linguistic change by the time of our inscriptions. In the course of the study, the epigraphic documents under scrutiny have been classified into three “sociolinguistic categories” (based on formality), i.e. formal, informal, and diaphasically low texts. The “frequency” of the cases of < E > for < AE > in these texts has been calculated according to two different methods, the “degree of error” and the “relative frequency of faults”. The results of the analysis (implemented by using Conditional Inference Trees) suggest that, in line with the most recent findings in the field of Latin (corpus) linguistics, the sociolinguistic study of inscriptions is better implemented by a cross-referenced use of both methods, even more so than in the case of diatopic inquiries.