Diachronic Analysis of Italian Opera Librettos

The purpose of this article is to find out which are the most typical words used in Italian opera librettos in each of historical periods of opera and to see the evolution of their use. A word which statistically occurs in a corpus of texts more often than expected is called a keyword. The article shows a new approach in diachronic linguistics, proposing a different examination of keywords found with the Log-Likelihood method in a corpus of Italian opera’s librettos. To get the keywords, the libretto’s corpus was compared with a reference corpus which includes the most representative works of Italian literature. Later the librettos’ corpus was divided diachronically in 5 sub-corpora according to some historical periods. A new analysis of these sub-corpora was performed using a software tool written by the author called CorpStat. It allows to analyze the evolution in the use of lexicon of opera’s librettos. Often the Log-Likelihood statistical method was criticized mainly because some words qualifies as keywords, even if they appear only in one or very few texts. With the software CorpStat it was possible to validate words as real keywords taking into consideration the percentage of number of texts in which they appear. The analysis brings to the conclusion that only 41 keywords out of a list of first 200 keywords could be validated as real keywords. However, the diachronic analysis shows that a word is a real keyword only in a certain historical period, but not necessarily in another. This can be due to the changes of libretto’s plots and language along the history.