Fragmentary ancient languages as “bad data”: towards a methodology for investigating multilingualism in epigraphic sources

This is the author accepted manuscript. The final version is available from de Gruyter via the DOI in this recordThe study of language contact in the ancient world has been an area of huge growth over the past ten years. However, in areas of the ancient world where sources are more limited, scholars have been more hesitant to make sweeping claims about the nature of language contact, even in communities where societal bilingualism is likely to have existed for many centuries. Languages only attested in fragmentary epigraphic corpora are considered the ultimate “bad data” and have not always received a great deal of attention in historical sociolinguistics, despite these texts representing our best evidence of many of the communities across the ancient Mediterranean. In response to this problem, this article asks how we should go about interpreting the evidence of ancient language contact in small or fragmentary corpora of texts. This article uses the Oscan corpus from Southern Italy (Lucania, Bruttium and Messina) c. 400-50 BCE as a case study for examining bilingualism in a fragmentary corpus. It outlines the data gathered from a range of different text-types from Southern Italy, the different kinds of contact phenomena which have been found in these texts, and whether there are any discernible patterns in the data. It argues, because of the fragmentary state of the Oscan corpus, that there is little clear evidence of chronological or geographic differentiation in levels of bilingualism. Rather, the evidence shows that in this corpus some text-types are more likely than others to contain contact phenomena. With this in mind, this paper proposes a new model which includes consideration of text-type for the interpretation of language contact and bilingualism in fragmentary corpora