The field of Greek philology, and especially dialectology, represents a distinct and seemingly arcane enclave within the discipline as a whole, rendered all the more bewildering by the fact that universal consensus rarely exists. (Hall 1997: 153)
The problem of the attested dialect geography of Ancient Greek in the first millennium bce has historically long eluded a simple solution. Sharp discontinuities combined with crisscrossing isoglosses point to some combination of prehistoric migrations and in situ network-like evolution, but exactly how these features were distributed from linguistic contact, convergence, or disruption of earlier dialect continua has been one of the most difficult questions to resolve in Ancient Greek linguistics and Aegean prehistory. Aspects of a family tree model of analysis as well those of a wave model are both necessary to explain all the attested data. The main goal of this study is to address one small aspect of this larger problem to try to move the discussion forward from a state of impasse that has existed in some form or another since the beginnings of Ancient Greek dialectology as a discipline, but especially since some of the more polemical works since the 1980s which have doubted whether a phylogeny for the Ancient Greek dialects can, in fact, be established at all.
To this end, this study is about determining subgroupings of dialects in historical dialectology. It also deals with a specific problem in Ancient Greek dialectology, the question of whether or not the Aeolic dialects, traditionally understood as a subgrouping of the Boeotian, Thessalian, and Lesbian dialects of Ancient Greek, form a clade in a phylogeny descended from Proto-Greek, the putative ancestor of all Ancient Greek dialects attested in Archaic and Classical Greece. To this day there remains poor consensus on many aspects of the internal prehistory of the Ancient Greek dialects, making it difficult for non-specialists in other branches of Greek linguistics and Hellenic studies to evaluate competing hypotheses. This is regrettable, since the data from the dialects do provide evidence that can corroborate or problematize alternative historical reconstructions obtained from archaeology and the Greek historiographical tradition.
Much of the previous scholarship on Ancient Greek dialectology has emphasized the qualitative evaluation of dialect isoglosses in establishing their relative value for historical reconstruction. This study takes a new approach, and argues that in this situation a principled quantitative analysis can be fruitfully used to determine the relative the likelihood of competing historical hypotheses in the subgrouping of the dialects. This study argues that the best evidence for testing a proposed genetic relationship between two dialectal sub-varieties is not in the relative quality of a few highly marked linguistic innovations, but rather by the entire bundle of innovatory isoglosses. In probabilistic terms, the more innovatory isoglosses that are shared between distinct dialectal varieties, the less likely they all will be shared due to chance. This study explicitly develops a probabilistic methodology in order to evaluate the relative likelihood of a linguistic innovation within the corpus of the attested dialects. Using this methodology, a probabilistic clade test is developed, and through its application to the Aeolic data, it is argued that the traditional hypothesis of genetic relationship between Boeotian, Thessalian, and Lesbian is statistically likely.
The structure of this work consists of five main chapters. The first chapter consists of a critical literature review, providing an overview of the history of dialectal subgrouping of Ancient Greek in modern scholarship, and a critique of the most recent literature on Aeolic. It argues in addition that Aeolic classification is not just a single problem within Greek dialectology, but is part of a more fundamental debate currently within the sub-discipline, namely whether or not phylogenetic relationships represent a viable model for Ancient Greek dialectal evolution. The second chapter outlines the preliminary assumptions necessary to evaluate the Aeolic hypothesis using a quantitative approach, giving the criteria for the selection of isoglosses to be evaluated using the principles of phylogenetic systematics. Particular attention is drawn here to the use of Hennig’s Auxiliary Principle, which has been hitherto ignored by many dialectologists who take a hypercritical position on classification. This chapter also addresses methodological problems from the disparate source material for the Aeolic dialects.
The third and fourth chapters consist of the main data analysis, evaluating the innovatory isoglosses shared by Boeotian, Thessalian, and Lesbian as to whether they may be used in the probabilistic clade test. The third chapter treats the ‘core’ innovations, shared by all three dialects, while the fourth treats the ‘peripheral’ innovations which are shared by two of the three. The fourth chapter also draws attention to the presence of features that are in areal distribution with their neighbouring dialects, and eliminates these features where convergent development is likely.
The fifth and final chapter contains the mathematical derivation of the probabilistic clade test and its implementation. Its implementation finds a genetic subgrouping of Aeolic dialects statistically likely, but no sub-clades are predicted from the probabilistic method. Following the operation of the clade test it is further argued that most economical relative chronology to account for the linguistic and philological facts is a primary split of Lesbian from the Proto-Aeolic clade, followed by a common period of Thessalian and Boeotian innovation, followed by a secondary split of the latter two dialects.