Inflation in the Indo-Aryan Compound Verb: 1300-2000

The compound verb, a characteristic feature of most South Asian languages, is made up of the conjunctive participial form (CP) of the main verb plus the inflected form of an auxiliary or "vector" verb which is homophonous with a full lexical verb. For example, in (1) from Marathi khatsun is the conjunctive participial form of the main verbs khats 'slump'; while geli is the inflected form of the vector verb dzaa GO:(1) mul.aa-tsaa mrutyu paah-aa-v.aa laag-l.yaa-ne child-Gen death see-Obl-Inf must-Pst-Instr vruddh aai khats-un ge.l-i aahe aged mother slump-CP WENT-Fsg is 'Having to see the death of her son (his) aged mother has become despondent.' Under certain semantic and syntactic conditions the compound verb alternates with the non-compound:(2) nokari gel-yaa muL.e to khats-laa aahe service go.Pst-Obl cause he slump-Pst.Msg is 'He's become despondent from losing his job.' We use the term "inflation" to indicate an increase in the text frequency or "flux" of the compound verb. In our paper we present evidence for inflation in the NIA language Marathi that parallels direct evidence for inflation in Marwari (Hook 1993) and indirect evidence for it in Bangla (Zbavitel 1970) and reflects an expansion in the scope of the compound verb that occurred across the length of northern India over the past five hundred years. In Marathi prose texts from about 1300 CE to the present there is a 1600 percent increase in the text-frequency of the compound verb. This gradual, monotonic climb in the incidence of compound verbs over seven centuries can be compared with parallel inflations in the flux of compound verbs in Bangla over the past 500 years and in Marwari over the past 400:Figure 1. Line graph showing the increase in frequency of use of the Marathi, Marwari, and Bangla compound verb from 1300 CE until the present.As CV flux increases over time there are changes in the degree to which vectors retain their lexical senses, in their privileges of occurrence, and in the ways they "compete" with other vectors. In our paper we instantiate these changes with data from Marathi, Hindi-Urdu, and Kashmiri and discuss them in the context of general properties usually attendant on the historical process of grammaticalization (Hopper and Traugott 1993: 113-126).