Senųjų Biblijos vertimų lyginamųjų tyrimų automatizavimas
Volume 19 (2017): Archivum Lithuanicum, pp. 57–98
Type: Article
Open Access
Open Access
Pub. online
31 December 2017
31 December 2017
Abstract
The abundance of Bible citations in old Lithuanian writings makes it difficult to study their relationships. A user-friendly data platform is required to allow the Bible lines to be grouped according to the characteristics interesting to the researcher. The assessment of similarity of citations is the most problematic. An integrated analysis of vocabulary, syntax, and morphology (and in some cases—spelling) allows it, but this kind of philological research is slow and currently impossible to automate. We succeeded in automating the comparison with a cyclic algorithm. The task of the algorithm is to detect identical compounds in two character sequences and to calculate which part of the compared sequences they cover. The effect of the algorithm and the calculated percentages are mainly determined by the length of the lines, the shortest sequence allowed to be compared, and the transliteration accuracy. Different lengths of the lines (a biblical line can be quoted as a fragment that is comparable to the full line) require the consideration of the chronology of the sources and the selection of corresponding comparative calculation parameters. The most optimal shortest comparable character sequence was determined experimentally. The automation of the comparison is hindered by the ambiguous orthography of old writings. It is levelled by the rules of transliteration, the application of which is differentiated according to the most important consistent spelling characteristics of each author. After evaluating the manually and automatically transliterated lines, it was determined that the transliteration method did not have any significant effect on the results. The results can be controversial when comparing the texts written in different dialects (the levelling of
dialectical features was not tested). The obtained similarity results are summarized in two-dimensional diagrams. The data can be sorted according to the chronology of the sources, according to the similarity average, according to the number of compared lines, according to the sequence of the lines in the source, according to the similarity of individual lines, according to the similarity of two lines in two sources, etc.
The results obtained after the software analysis of Bretkūnas Postilė (1591) biblical line comparison are in line with the conclusions reached by the previous researchers that applied different methods. The comparison does not include and cannot assess the influence of translation sources. Automatic calculations do not prove per se the interception of the lines. The percentages of similarity are not absolute; they vary depending on the selected calculation parameters. The most important outcome of the work is considered the tool designed to demonstrate the relative relationship between the sources or their parts, which opens up the opportunities for further, more detailed research.