Contrary to most syntactic theories which adopt hierarchical constituency structure as a self-evident matter like an axiom of mathematics, which place the so-called INFL(ection) over the VP (consequently INFL c-commanding VP, thus, IP dominating VP), and where the whole clause's representation is a VP-within-IP schema, what I propose now for the clause structure is an IP-within-VP schema, under which a clause will be regarded as a VP, the maximal projection of a verb which governs not only the object and necessary oblique complements on the one hand, but also the IP dominating the potential INFL and the candidate NP for subject, on the other. Its typical example is illustrated as follows:
Fig.
[where Pd and AG are abstract adjectival predicates which mark the NP with the same index depending on whether it is undergoer or actor(typically), respectively. Note that the same verb appears twice: once as governor of inner arguments including NP; (the object), and secondly as governor of IP involving as its part the subject-candidate NP; which is not yet case-marked with any case.]
When in the stage of this figure, the IP is still covert, implicit or (as it were) submerged: not allowed of its expression. But when it is promoted, and grafted to the D-marked node, IP turns overt as if a film were developed into images in photography, and motivates the verb into a finite form, at the same time assigning nominative case to NP; , which finally moves into the status of SpecvP i.e. clausal subject.
I applied this notion of clausal composition to the analysis of the Frenchor Japanese-type causative construction, which, unlike English, comprises serial verb structure such as faire lire or yo111-aseru , with different treatment of case marking between transitive and intransitive causees. In the process of the study, I came to the conclusion that such an idea is important that one clause is juxtaposed(spatially) and adhered to another and imprints informations on it. This notion explains the above-mentioned type of causative construction elegantly. I would like to call that process 'juxtahesive impression.'