Runic Inscriptions in an Unrecognized Foreign Tongue?: Methodological Preliminaries to Language Identification

Some allegedly nonsensical runic inscriptions from the late Viking Age and the early Middle Ages in Scandinavia might appear to be written in an unencrypted natural language. This paper discusses prerequisites to determining their possible linguistic meaningfulness and to identify- ing the language or languages concerned. Major indications of meaningfulness, both linguistic and non-linguistic, are cross-inscriptional textual parallels, particularly if the inscription- carriers are geographically far apart. Clues which jointly suggest linguistic structure include the employment of essentially all the signs of the basic 16-symbol futhark, the occurrence of special ‘rune-like signs’ that appear to augment the futhark inventory, the linguistically natural frequency of graphs, linguistically natural graphotactics, the recurrence of specific runes or runic combinations e.g. in ‘quasi-word-final’ position, and seemingly natural usage of separators. A crucial pre-condition of identifying the language used consists of the establishment of a basis for the comparison of inscription and language, either through recourse to actual contemporary language data or through historical-comparative and language-internal reconstruction. Factors suggestive of successful language identification include (a) a high degree of inscription-internal structural and lexical consistency vis-à-vis the language under investigation (barring borrowings and code-mixing), (b) parallels with other internally consistent inscriptions, (c) a markedly bet ter match to one particular language than to others, and (d) the applicability of the hypothesis to cases beyond the one for which it was originally devised. Due to the complex nature of the enterprise (highly impoverished runic script, brief texts, meager corpus of inscriptions, absence of contemporary records for many languages, lack of socio-cultural information etc.), the analyst normally operates with varying degrees of probability rather than definitive proof of language identification.