A Cellular Automaton Environment for the Complex System of Speech

Linguistic data collections, such as representative language and dialect corpora collected over several decades, have in the past 15 years become the focus of extensive digitization efforts. The Linguistic Atlas Project (LAP), containing meticulously curated American language survey data from the past 90 years, is a prime example of such a corpus. The scholarly use of such bodies has traditionally relied heavily on the linguist researcher’s intuition in observing patterns and producing maps of the data to answer questions on language use across time and space. Some solutions have been offered by statistical methods and Geographic Information Systems (GIS), but the high learning curves involved in these techniques have limited their appeal in dialectology. At the same time, the spatially and temporally scattered nature of language datasets, and the lack of longitudinal data in particular, has hindered the modeling of language change in dialectology. This thesis aims to provide solutions for both visualizing spatial variation in language and modeling its temporal change.

The framework adopted for this work is the theory of the complex system of speech, and the method of implementation that of the Cellular Automaton (CA). A further distinguishing feature of the work is that it uses the extensive data holdings of the Linguistic Atlas Project as a source of real-world language data to base its simulations on. The results obtained through the work are validated in respect to previous linguistic theory, and the complex systems of speech in particular. The results of the work include the construction of a versatile simulation environment for language, and its successful application to a) the development of a linguistically feasible simulation of language change, and b) the development of an objective, straightforward process for region estimation of linguistic features.