On how to measure phonological complexity of sign languages

Introduction. Having a measure of the phonological complexity of signs is an important challenge for anyone interested in developing assessment tests for evaluating language competences and language impairments in Deaf populations. Traditionally, there are two main types of approaches to phonological complexity, which we can refer to as theory-driven and data-driven, respectively. The theory-driven approach is well illustrated by Clements (1985) and Sagey (1986) where counting distinctive features is the relevant measure. Brentari's Prosodic Model (1998) belongs to this tradition: she assigns to each phonemic class a set of features organized in a hierarchical structure. Each sign can be described in terms of a branching tree where each branch corresponds to a phonemic class (handshape, location, and movement) and each node corresponds to a set of features. The richer the structure, the higher the complexity of a sign. While this model has proven to provide important cross-linguistic generalizations, its validity beyond ASL cannot be taken for granted. Data-driven approaches, on the other hand, typically rely on pattern/order of acquisition, frequency, speech errors and similar measurable facts to assess phonological complexity. While some of these data are available for some sign languages (typically ASL and BSL), they are entirely missing for others.