2. Chapter Overview

Here you will find a brief synopsis of the chapter "Computing linguistic distances between varieties" by April McMahon and Warren Maguire..

At the start of the chapter McMahon and Maguire introduce the reader to the idea of comparing different varieties on the basis of the concept of rhoticity, one of the most prominent differences between English dialects. The realization of the non-pre-vocalic /r/ is, according to the authors, presumably one of the main features with which to distinguish different varieties. But, in turn, the authors also state that it is very important to realize that this relevance of rhoticity is impressionistic and that empirical evidence is hard to come by.

Consequently, McMahon and Maguire introduce their idea of a quantitative study which can be used to prove the importance of rhoticity. However, they also emphasize that it is very important to not concentrate only on rhoticity itself. This means that all (i.e. as many as possible) features which distinguish different varieties from each another need to be considered. To do this, McMahon and Maguire explain the source data they used for their study, a database of 110 Germanic cognates which can be found on the Sound Comparisons website. They compared the realizations of these 110 different words from many different languages and varieties with each other. These values take into account a wide range of different criteria such as location and degree of stricture, voicing, nasality, etc. On the basis of this input, a software programmed by Paul Heggarty created a huge matrix in which each word’s similarity with cognate words in all other varieties is expressed by a value between one (identity) and two (no similarity).

The authors then describe the challenges that come with visualizing a matrix of this size: although linguists are trained at and used to reading family-trees, this specific way of visualization is not expedient in this case, because we already know that all these varieties are part of one family. What we actually want to know is how close the different varieties are to each other. McMahon and Maguire therefore introduce the use of the NeighborNet algorithm implemented in the SplitsTree package, originally designed for the field of bioinformatics.

The next page then shows a comparison diagram, created with SplitsTree, which takes all varieties into consideration. The authors furthermore argue that the big split which can be seen is triggered by realizations of rhoticity. At the same time, they also stress that the diagram still provides no proof of the pre-eminence of rhoticity, showing one of the main disadvantages of the NeighborNet algorithm: once the data has been put in and visualized there is no way to read off any of the features directly from the graphical representation.

By comparing two non-rhotic varieties and one rhotic and one non-rhotic variety with each other the authors then show that rhoticity represents one of the main differences between English varieties. These clues are further strengthened by making Splitstree create a diagram in which rhoticity has been eliminated and only the remaining criteria are used to calculate the difference and distance between varieties. The result is a completely different diagram in which the big split in the middle has been eliminated, suggesting that rhoticity was indeed the reason for this big division between different varieties.

The authors conclude the chapter with a short general discussion of the advantages and disadvantages of using Splitstree4.

Created with the Personal Edition of HelpNDoc: Free PDF documentation generator