Borrowing non-canonical inverse between Kabardian and Abaza Peter M. Arkadiev Institute of Slavic Studies of the Russian Academy of Sciences / Russian State University for the Humanities, Moscow Abstract Abaza, a polysynthetic ergative Northwest Caucasian language, shares with its neighbour and distant relative Kabardian a typologically peculiar use of the deictic directional prefixes monitoring the relative ranking of the subject and indirect object on the person hierarchy. In both languages, the cislocative (‘hither’) prefixes are used if the indirect object outranks the subject on the person hierarchy, and the translocative (‘thither’) prefixes are used in combinations of first person subjects with second person singular indirect objects. This pattern, reminiscent of the more familiar inverse marking and hence called ‘quasi-inverse’, is observed with ditransitive and bivalent intransitive verbs and is almost fully redundant, since all participants are unequivocally indexed on verbs by pronominal prefixes. I argue that this isogloss, shared by West Circassian (a close relative to Kabardian) but not with Abkhaz, the sister-language of Abaza, is a result of pattern replication under intense language contact, which has led to an increase of both paradigmatic and syntagmatic complexity of Abaza verbal morphology. Keywords: Northwest Caucasian languages, Abaza, Kabardian, polysynthesis, inverse, language contact, pattern-borrowing, morphological complexity 1. Introduction In this paper I discuss a hitherto unreported case of pattern borrowing of a typologically peculiar morphological pattern between two distantly related polysynthetic languages of the Northwest Caucasian family, Kabardian (ISO 639–3: kbd) and Abaza (ISO 639–3: abq). The two languages, which are typologically similar but mutually unintelligible, have been in a state of intense contact for several centuries, which has resulted in Abaza having numerous lexical and some morphological borrowings, as well as morphosemantic calques from Kabardian. The morphological pattern in question involves what I call the Word Structure 14.2 (2021): 148–173 DOI: 10.3366/word.2021.0185 © Edinburgh University Press www.euppublishing.com/word