Edited by Lars Heltoft, Iván Igartua, Brian D. Joseph, Kirsten Jeppesen Kragh and Lene Schøsler
[Current Issues in Linguistic Theory 345] 2019
► pp. 233–251
Old Japanese (largely 8th century AD; “OJ”) is, like Middle and Modern Japanese, a typical SOV language and also shares with them a complex predicate construction consisting of two adjacent verbs, V1 V2, of which V2 has some grammatical function, (i). However, OJ in addition has a complex predicate construction in which V1 is grammatical and V2 is the main verb, (ii), which seems anomalous in an SOV language and which is not found in Middle or Modern Japanese.
NJ nomi-au (lit. ‘drink-meet’) ‘drink together’
OJ api nomu (lit. ‘meet drink’) ‘drink together’
Situated within a classical version of Henning Andersen’s language change theory, this paper offers a diachronic interpretation of the OJ construction in (ii) as a transient stage in the emergence of the complex predicate construction in (i), which may be understood as having arisen through categorial reinterpretation of preverbal adverbial material as grammatical, reflected in (ii), followed by a structurally motivated shift to postverbal position, reflected in (i). This proposal is further generalized to account for several grammatical suffixes in Japanese as having originated in similar sets of innovations, specifically the prohibitive final particle na, negative -(a)n-, conjectural -(a)m-, necessitive be‑ and negative potential masizi.