In multiple Spec AgrP hypothesis (Grewendorf and Sabel 1999), many kinds of scrambling can be explained in terms of adjunction to AgrP and substitution to [Spec AgrP] for [Σ] checking. Also Constraint on Adjunction Movement is necessary. Miyagawa(2001a,b)'s EPP driven Scr is limited to only IP-adjunction position. However, I can explain diverse kinds of scrambling by modifying the above Agr-projection into simplified T-projection model and accepting Kim(2003a,b)'s idea on pure [EPP] driven scrambling. The Theme object substitutes to [Spec2 vP] to check pure EPP feature on v in short scrambling, only object substitutes to [Spec2 TP] for EPP feature checking on T in middle scrambling (IP- adjunction scrambling), and an element may successively scrambles to adjunct TP -position for checking pure EPP on T in long scrambling. However, short scrambling leaves some unsolved problems.

In multiple Spec AgrP hypothesis (Grewendorf and Sabel 1999), many kinds of scrambling can be explained in terms of adjunction to AgrP and substitution to [Spec AgrP] for [Σ] checking. Also Constraint on Adjunction Movement is necessary. Miyagawa(2001a,b)'s EPP driven Scr is limited to only IP-adjunction position. However, I can explain diverse kinds of scrambling by modifying the above Agr-projection into simplified T-projection model and accepting Kim(2003a,b)'s idea on pure [EPP] driven scrambling. The Theme object substitutes to [Spec2 vP] to check pure EPP feature on v in short scrambling, only object substitutes to [Spec2 TP] for EPP feature checking on T in middle scrambling (IP- adjunction scrambling), and an element may successively scrambles to adjunct TP -position for checking pure EPP on T in long scrambling. However, short scrambling leaves some unsolved problems.

LF-undone, multiple Specs, short scrambling, pure EPP, [∑]-feature