Enclitic -(m)a 'but', clause architecture and the prosody of focus in Hittite.

We discuss the extraordinary syntax of the enclitic -(m)a: as different from other enclitics it does not cliticize to a set of words like nu, mān, etc. However, as different from -pat with its free distribution within the clause -(m)a attests a very clear tendency to be positioned after the first stressed word. To make things even more complicated, in the focusing function it attests a seemingly free position in the clause. The paradox of -(m)a is that most of the details of its aberrant syntactic behavior, save some clause internal usage which we are demonstrating for the first time, are perfectly well known, but several pieces of evidence have never been brought together. The main one is the fact that the set of words which does not host -(m)a as well as -ya is precisely the set of words which is not taken into consideration when the second position of such constituents as relative pronouns in determinate clauses, subordinators maḫḫan and kuit is determined. The paper also provides a unitary explanation of both topicalising -(m)a in the left periphery and focusing -(m)a in the preverbal position: the common feature might be the prosodic boundary to the left of both hosts triggered by information structure to the left of the host of both topicalizing -(m)a/additively focusing -ya in the left periphery and focusing -(m)a/ya preverbally. [ABSTRACT FROM AUTHOR]