Measuring genre differences in Mark with correspondence analysis
David L. Mealand · Literary and Linguistic Computing · 1997
This article reports a series of tests on samples from the Gospel of Mark. The first set of results show that groups of samples divided by genre display clear between-group differences. Using correspondence analysis, it is possible to see which variables contribute most to these genre differences. These tests are corroborated by further use of other multivariate statistical tests. These tests established that there are differences, and which function words and other high frequency words are most implicated in those differences. A further analysis of the texts was conducted. This used a program called Word Cluster to extract striking examples of clusters of high frequency words characteristic of each genre. These extracts from the texts show in a more traditional literary way just how effective the result of statistical analysis and text searching systems can be. The entire series of tests show that, as with other literature, so in the gospels, before decisions about authorship are made, attention must be paid to differences of genre. This consideration seems to be particularly acute in the case of Mark, but may affect other literary analyses also. We can show not only that style varies with genre, but also which stylistic markers are used most heavily in which passages. We can also discover other stylistic variation which seems not to be explained by the genre differences examined here.