Diversity, stablity and diffusion in the Hindu Kush region of Inner Asia

The mountainous Hindu Kush, on the northwestern edge of the Indian subcontinent, offers a promising “test site” for questions relating to language contact, diffusion and stability, considering its high linguistic density and diversity, with languages from six genera: Indo-Aryan, Iranian, Nuristani, Tibetan, Turkic and Burushaski (Masica 2001: 225; Liljegren 2017: 107–108). While traces of several substrata suggest that the region in a distant past served as an important accretion zone (Tikkanen 1988: 304), akin to Caucasus (Nichols 2003: 306), the region as of today bears witness of several waves of small- or larger scale diffusion (Bashir 2003: 823; Liljegren 2014: 162–167), mainly related to Indo-Aryan northward expansion within the last few millennia (Morgenstierne 1932: 51; 1961: 138; Strand 2001: 200), and in more recent times by superstratal influences (Bashir 2007) from a few languages of wider scope (e.g. Pashto, Urdu and Dari). The main question asked in the present study is to what extent properties of some linguistic subsystems are more prone to diffuse than others, and whether they cluster similarly (geographically or genealogically) or significantly differently. We also ask to what extent individual properties show a higher or lower degree of intra-genealogical stability. A set of comparable first-hand data was collected from 59 varieties (representing all six genera) in a handful of collaborative workshops held in the region. The data was coded and analysed for approximately 50, mainly binary, linguistic features belonging to five different subsystems: phonology, word order syntax, grammatical categories, simple clause properties, and lexical structure, with an even distribution of features belonging to each of these five. In addition, a basic word list was cognacy-coded and used as the basis for measuring genealogical relatedness. It was primarily Indo-Aryan, due to its high representation (33 of the 59 varieties) that was analysed for feature stability. The preliminary results, visualized with NeighborNet representations (Huson & Bryant 2006), indicate differential clustering depending on subsystem. The clearest examples of diffusion affecting substantial parts of the region were found within phonology, word order and lexical structure, while simple clause features (alignment in particular) display more limited (subareal) clustering, and grammatical categories (especially gender) a relatively high degree of intra-genealogical stability. Language contact and relatedness in the Hindukush regio