Quantitative Methods for Classifying Writing Systems

Despite linguists’ necessary reliance upon writing to present and preserve linguistic data, writing systems have remained a largely neglected corner of linguistics. While the linear and normative aspects of Gelb’s [1963] teleology have been unceremoniously rejected by more recent work on the classification of writing systems, an acknowledgement that more than one dimension may be necessary to characterize the world’s writing systems has not come easily. The ongoing polemic between Sampson [1985] and DeFrancis [1989], for example, while addressing some very important issues in the study of writing systems,1 has been confined exclusively to a debate over which of several arboreal classifications of writing is more adequate. Sproat’s [2000] classification, to our knowledge, was the first multi-dimensional one. While acknowledging that other dimensions may exist, Sproat [2000] arranges writing systems along the two principal dimensions of Type of Phonography and Amount of Logography. This is the departure point for our present study. The goals of our research programme can then be stated as follows: (1) Assuming Sproat’s classification grid as a correct characterization of the structure of writing systems, are there quantitative methods that would allow us to posit a given writing system within this 2-dimensional space? This involves “measuring,” at least in a relative sense with respect to other writing systems, its type of phonography and amount of logography. (2) Given such quantitative methods, can we distributionally classify the world’s writing systems on this grid in such a way that corroborates or casts doubt upon the adequacy of Sproat’s grid, or his placement therein of several specific writing systems [Sproat, 2000, p. 142]? (3) Given such quantitative methods and the putative correctness of Sproat’s grid, can we distributionally classify the world’s writing systems in a way that informs us of other underlying principles or latent variables upon which a writing system’s position in several dimensions might possibly depend? If no such dependence exists, then these other variables would presumably imply additional dimensions for this grid. If such a dependence did exist, it could in principle reduce the dimensionality of the grid, and thus the classification’s complexity. The holy grail in this area, of course, would be a quantitative procedure that could classify entirely unknown writing systems to assist in attempts at archaeological decipherment, but more realistic applications do exist, particularly in the realm of managing on-line document collections in heterogeneous scripts or writing systems. No previous work exactly addresses this topic. None of the numerous descriptive accounts that catalogue the world’s writing systems, culminating in Daniels and Bright’s [1996] outstanding reference on the subject, count as quantitative. The one computational approach that at least claims to consider archaeological decipherment Knight and Yamada [1999], curiously enough, assumes an alphabetic and purely phonographic mapping of graphemes at the outset, and applies an EM-style algorithm to what