V-to-I lowering and split IP hypothesis

Question

English is I-lowering language, but on split IP hypothesis([AgrP [TP [VP]]]), verbs move to the head of TP: V-raising.

/John often kissed Mary./ On unsplit IP(IP[VP]), [+tense] which the head of IP has moves to the head of VP for realization; I-lowering. Whereas on split IP(AgrP[TP[VP]]), [+tense] which the head of TP has is realized by verbs moving to the head of TP; V-raising.

Why, on IP split hypothesis, does V-raising occur in I-lowering language?

And in a sentence which has adverbs like above, after V-raising occurred, the verb is before the adverb; *John kissed often Mary. How is the word order correct before spell-out? I would like to know the movement of subjects and verbs. If you know how is the nominative Case assigned, I'd like to know.

Answer 1

At first, this represents how generative syntax was done in the late 80s-early nineties. Many of those ideas were abandoned a long time ago.

Secondly, you grossly misrepresented it. There were two options, V-raising or I-lowering (other variants: Agr-I, AGR-I, Agr-T etc.). French was considered a V-raising language (i.e. with overt verb raising): V moves to Agr, and then it all moves to T. English, on the other hand, was a I (Agr-T) lowering language, which means T lowers to Agr and then T-Agr lowers to V.

What was problematic about I-lowering (and lowering in general) is that it created an improper chain, with an unbound trace. Two theories were proposed back then, Chomsky 1991 in favor of LF V-raising or Bobaljik 1995 in favor of I-lowering at PF.

In MP (in Adger's interpretation), all of the problems above are elegantly solved with the help of strong/weak EPP, strong/weak tense value on v, feature checking and phi-features - see chapter 6 in Adger 2003.