Measuring Dialect Pronunciation Differences using Levenshtein Distance

The work in this thesis has been carried out under the auspices of the Behavioral and Cognitive Neurosciences (BCN) research school, Groningen. Acknowledgements This thesis is attributed to exactly one author as can be seen on both the cover and the title pages. The author is the one who is responsible for the content. But it should be emphasized that many people contributed to the coming about of this thesis. I would like to mention them. First of all, I thank my promotor, John Nerbonne. More than five years ago, he encouraged me to work on the dialectometry project, and during the project he gave invaluable support. Without his support this thesis would never have been published. Probably just the pictures in this thesis will catch the eye of the reader. While I implemented programs for calculating distances between language varieties and for clustering them, Peter Kleiweg developed software for creating dendrograms, multidimensional scaling plots and different types of (color) maps. I am grateful to Peter for developing and making available this excellent software, and for his extensive help when creating the figures. During a visit on a cloudy afternoon, one of my best friends, Martin de Vries suggested that one seek speech segment distances on the basis of an acoustic representation. Becoming the inventor of a new type of voice-producing prosthesis, this approach seemed obvious to him. I thank him for this valuable suggestion. In cooperation with Roberto Bolognesi I worked on the comparison of Sardinian dialects. In this small project, the use of acoustic segment distances was developed and the use of the Levenshtein distance was improved. I thank Roberto for his help and his friendly cooperation. In the field of phonetics and phonology I got the help of many persons. I thank Paul Boersma and David Weenink for making available their excellent PRAAT program. Norwegian dialects play an important role in this thesis. Jørn Almberg made Norwegian recordings and transcriptions of the fable 'The North Wind and the Sun'. I am grateful to him for this his permission to use this material and for his help during the whole investigation. I thank Charlotte Gooskens for the good cooperation in our Norwegian research, and for her permission to use the results of her perception experiment in Norway. Thanks are due to Charlotte Gooskens v vi and Sabine Rosenhart for cutting the Norwegian word samples. I thank Arnold Dalen …