Presentation at Naming the World: From Common Nouns to Proper Names, September 2004; article in the conference proceedings, ed. by Dunja Brozovic Roncevic and Enzo Caffarelli, published as QuadRIOn 1.
The existence of patterns as one of the factors in the toponomastic process has been known for more than a quarter of a century. However, while some onomasticians have suggested that such patterns can play an important role even when the names in question can be adequately explained by other means, such hypotheses have been rather difficult to prove. The present study is an attempt to address the issue: the goals were, first, to find regularities in the naming of Finnish lakes; second, to assess whether such regularities imply the presence of naming patterns; and third, to see if a quantitative study could give new insights about the properties of such patterns.
This was done by applying methods developed in the computer science field of data mining to an electronic corpus consisting of all Finnish lake names found on the 1:20 000 Basic Map. These revealed several groups of names that appear next to each other significantly more often than could be expected, even after accounting for regional variation in the distributions of the names.
Some of the groups can be explained by referring to e.g. cultural history, but in a large number of groups the names have a semantic relationship which suggests that there is a large number of relatively widespread patterns in naming Finnish lakes. However, these patterns are very specific and it is difficult to see a systematically productive general pattern. Some of the phenomena involved can be described using Construction Grammar, but it is evident that the theoretical framework needs some adjustments.