engleski

Mining macaronics

The present paper has a twofold purpose: to describe the compilation of an annotated digital corpus of macaronic poems, and to demonstrate the usefulness of the corpus in the context of comparative analyses. The computing tools and procedures described in the first part of the paper can be modified and employed in tagging corpora of other kinds of linguistically mixed texts. Apart from that, some of the methodological challenges would probably at least partly overlap with those dealt with here. The analysis of the monolingual cluster sizes has pointed to two tendencies. First, genetic linguistic groups of the embedded languages seem to play a role in the relative frequency of choices between Latin, hybrid, and pure embedded language lexical material. Second, although longer clusters of monolingual discourse occasionally appear, the increase in the size of clusters corresponds to an exponential decrease in their frequency, which points to the high level of granularity of the macaronics in this respect. The second analysis compared the frequency of hybrid words at verse beginnings and has shown that verse initial words are not more frequently macaronic than would be expected from the overall frequency of macaronic words in individual texts ; on the contrary, the two variables are highly correlated. Moreover, the total frequencies of the macaronic words tend to be slightly higher than their frequencies at the verse opening.

macaronic Latin, Latin, Neo-Latin, corpus linguistics

nije evidentirano