Investigating and measuring linguistic areality in the Hindu Kush-Karakoram region

More than 50 distinct ethnolinguistic communities inhabit the mountainous northwestern outskirts of the subcontinent. This region, here referred to as Hindu Kush-Karakoram, is spread over the territories of several countries -- primarily Afghanistan, Pakistan and India -- and comprises languages belonging to six genera: Indo-Aryan (in majority), Nuristani, Iranian, Tibeto-Burman, Turkic and the isolate Burushaski. The linguistic profile of this region and its significance as a contact zone or linguistic area has been the topic of a discussion going on for several decades (Toporov 1970; Èdel’man 1980; 1983: 16; Bashir 1996a; 1996b; 2003: 823; 2016; Baart 2014; Tikkanen 1999; 2008; Koptjevskaja-Tamm & Liljegren 2017: 215–223), but the tendency has been to focus on individual features and phenomena, sometimes based on relatively sparse data, and more seldom have there been attempts at applying a higher degree of feature aggregation with tight sampling. In the present study, comparable first-hand data from as many as 59 Hindu Kush-Karakoram language varieties, was collected and analyzed. The data allowed for setting up a basic word list of 95 comparable meanings (representing close kinship, lower numerals, basic actions, substances and objects) as well as for classifying each variety according to approximately 50 binary structural features (reflecting phonological, morphological, syntactic and lexico-semantic properties). While a comparison of the basic lexicon across the varieties lines up very closely with established phylogenetic classification, structural similarity clustering (visualized by means of NeighborNet) is clearly related to geographical proximity within the region and often cuts across phylogenetic boundaries. The strongest evidence of areality tied to the region itself (vis-à-vis South Asia in general on the one hand and Central/West Asia on the other) relates to phonology and lexical structure, whereas word order and alignment features mostly place the region’s languages within a larger areal or macro-areal distribution, and many morphological features or properties related to grammatical categorization (e.g. gender) display a high degree of genetic stability. The distinctly sub-areal clustering of the many Indo-Aryan varieties (33 out of 59), each along with a set of non-Indo-Aryan languages, suggests multiple centres of diffusion or parallel development, some of them very old and mostly likely of substratal nature, others reflecting contact patterns of a more recent date. Two putative sub-areas of particular interest are: a) a central west-to-east-stretching belt, including many Indo-Aryan languages, the isolate Burushaski as well as most of the Nuristani languages, and b) a northern belt, partly coinciding with the Wakhan corridor, including Indo-Aryan, Iranian and possibly one of the Nuristani varieties. The remaining Indo-Aryan languages, mostly at the southeastern and southwestern peripheries, group structurally either with more typically South Asian languages, such as Indo-Aryan Urdu-Hindi and Iranian Pashto, or with typically Central/West Asian languages, such as Iranian Dari and Turkic. These tentative results lend support to the scenario painted by Morgenstierne (1961: 138; 1974: 2–3), echoed by Strand (1973: 207–208; 2001: 251), according to which the Indo-Aryan languages of the region can be traced back to a cluster of northwestern Indo-Aryan varieties that developed and differentiated in the plains south of the Hindu Kush before gradually penetrating the mountainous area from the south, after which these varieties gradually evolved into the present-day languages and dialects, strongly influenced by adjacent non-Indo-Aryan languages already present in their new environments. These results further refute, once again, the relevance of a “Dardic” level below that of Indo-Aryan (or possibly northwestern Indo-Aryan), whether intended as a phylogenetic label or an areally defined term