The high-altitude Hindu Kush-Karakoram region covering north-eastern Afghanistan, northern-most Pakistan and Kashmir, is for the Eurasian context particularly diverse with its approximately 50 languages belonging to six genera (Indo-Aryan, Iranian, Nuristani, Turkic, Tibeto-Burman, Burushaski). While it has been claimed to constitute a significant linguistic convergence area (Toporov 1970; Èdel’man 1980; 1983:16; Bashir 1996; 2003:823; Tikkanen 1999; 2008; Baart 2014) largely overlapping with a distinct religious-cultural sphere (Jettmar 1975; Cacopardo & Cacopardo 2001), relatively little systematic research has been carried out, often based on small, non-representative samples. Those studies were also largely limited to a few phonological and grammatical features.

The present project aims at revisiting those claims by: a) using very tight sampling, b) applying a high degree of feature aggregation, and c) striving for high granularity in determining the nature of areality and long-term contact patterns in the region. While utilizing available descriptions, priority was given to obtaining comparable sets of first-hand data, gathered in five “collaborative elicitation workshops” arranged in various locations in the region with invited native-speaker consultants. The elicitation package consists of: a) a 40-word list, based on the Automated Similarity Judgment Program (Wichmann, Holman & Brown 2016); b) a list of numerals; c) a 96-item kinship list; d) a sentence questionnaire, based on the Leipzig Valency Classes Project (Hartmann, Haspelmath & Taylor 2013); e) a translation of the ‘Northwind and the Sun’ fable; d) the Pear Story video (Chafe 1980) used as a stimulus for obtaining a natural narrative; and, e) an abbreviated and slightly adapted version of Wilkin’s demonstrative questionnaire (1999). The resulting data sets (comprising 54 data points as of late 2017) were used as a source for exploring multiple features: lexical, phonological, morphological and syntactic. The analysis uses WALS-features, as well as a number of novel WALS-like features, as its starting point, but allows for discovering fine-grained distinctions made by individual languages, thereby allowing for higher resolution in typological classification.

The features subject to study so far are e.g., the inventory and size of retroflex and affricate subsets, kinship terminology, numeral bases/composition, alignment patterns (case marking and verbal agreement), the presence and nature of gender/animacy distinctions, the presence and nature of spatial and geomorphic coding, basic word order, and the order of adposition and noun phrase. The emerging distribution shows partly contradictory results, but at the same time gives evidence to historical contact patterns in the central parts of the region while reflecting ongoing encroachment of surrounding macro-areas (such as the South Asian linguistic area and a Persian-dominated area of West and Central Asia) on the present-day region. A possible interpretation of those patterns is that different features reflect separate stages of ongoing fragmentation of an old “refuge” zone (possibly a continuous one extending throughout the entire Himalayan region (Nichols 1992:21)), in which the language isolate Burushaski probably played a non-trivial role (Hock 2015; Tikkanen 1988) along with other languages or language families now only detectable as substrata.