Paper short abstract:

The past negative in -(a)nanda is widely attested since the 15th century, but no consensus has ever been reached concerning its etymology. Refuting most previous proposals, we will suggest an origin involving a doubled negative suffix -(a)n- and provide corroborating evidence for such a solution.

Paper long abstract:

The past negative in -(a)nanda is widely attested since at least the 15th century and is likewise observed dialectally up to the present day. When it comes to the etymology of this Middle Japanese development, however, no consensus has ever been reached among scholars. Yet, even if "the details of the composition of the form are" thus still deemed "entirely unclear" (Frellesvig 2010 [A History of the Japanese Language]: 336), this is certainly not due to a lack of proposals as to its derivation. In the present paper we will review the more than half a dozen such proposals put forward since the early 20th century by scholars such as Shinmura Izuru, Tokuda Kiyoshi, Hamada Atsushi etc. and refute the majority of them on various grounds: be they chronological, dialectological, morphosyntactic, phonological or sociolinguistic ones. None of the problems the other proposals suffer from occur when we simply assume an etymology involving (besides the obvious past -ta) the negative suffix -(a)n-, however twice, unlike in the functionally identical and dialectally attested -(a)nda with single -(a)n-. In support of this view we will offer a phonologically motivated explanation based on the immediate historical precursor of -(a)nda as well as a variety of parallel cases of what historically appear to be double negatives, double causatives and double passives, ranging from pre-Old Japanese formations (such as causative -(a)sim-) to recent developments in the spoken language (e.g. the so-called sa-ire-kotoba). In terms of its formation, we will argue, -(a)nanda thus ain't not alone at all in the history of the Japanese language.