Question words in focus positions

Malayalam, an SOV language, moves its wh-phrases to a Focus position immediately to the left of V (linearly speaking). Multiple wh-phrases are “stacked” in this position. Wh-extraction from an embedded clause is not possible. When a wh-phrase in an embedded clause has matrix scope, scope-marking is done by two movements: the wh-phrase moves to the Focus position in the embedded clause, and the embedded clause is pied-piped to the Focus position in the matrix clause. It is shown that the device of “Attract” by EPP is inadequate (by itself) to describe these movements (or multiple wh-fronting). We suggest a supplementary device. “Association with focus”, the algorithm by which the question operator accesses question words, is (we suggest) a kind of ‘probe’. In languages which employ strong focusing devices, the question operator’s probe “looks at” only Focus positions. In these languages, wh-phrases must cluster in Focus positions in order to be interpreted.