Origins of metathesis in Batsbi

Blevins and Garrett (1998) investigate in detail the origins of CV/VC metathesis in a number of languages and identify two types of metathesis and a “pseudometathesis”. For them, “pseudometathesis” is a synchronic process that does not originate through the historical process of metathesis. They analyze languages in which “pseudometathesis” originates through epenthesis and deletion (1998) or through reinterpretation and generalization of other processes in the language (Garrett and Blevins 2009). I argue here that metathesis in Batsbi originates as a result of grammaticalization, together with regular phonological processes. When a function word, such as an auxiliary, grammaticalizes as an affix on a base, affixes trapped between the base and new affix are often lost (Harris and Faarlund 2006). However, in Batsbi some trapped affixes were not immediately lost, and I argue that this is the source of the variable position of the present tense marker, and that its variable position was reanalyzed as metathesis. I argue further that the reanalyzed process is true metathesis synchronically.