Both the large number of new grammars published in the past 20 years and better access to now digitized older literature offer a great opportunity to investigate and re-evaluate linguistic areas in the world. We will present our ongoing project at the University of Zurich, which is part of a larger interdisciplinary project on the peopling and linguistic diversity of the Americas. We use a qualitative and quantitative approach, aiming at discovering linguistic areas by taking into account multiple independent typological features.
Our dataset is based on a fine-grained survey of both well-established and more recently identified typological features with areal potential. It includes a variety of phonological features (e.g. glottalized consonants, tone, syllable structure), as well as morphosyntactic features (e.g. SG-PL verb stem alternation, apprehensional morphology, demonstratives, personal pronouns). Our genetically and geographically balanced language sample covers 215 American languages (117 families and 98 isolates), as well as a global sample of 100 languages for comparison. For the analysis, we use a geographically informed Bayesian clustering approach to detect linguistic areas, while controlling for possible confounding effects due to shared inheritance and universal preference (features being similar due to being globally common).
In this talk we will present the various aspects of the conceptualization of this interdisciplinary project, our approach to data collection and curation, as well as intermediate results.