How valid is preposition incorporation for Kiswahili predicate verbs?

In (1988), Baker came out with a theory of preposition incorporation, inter alia, into verbs. In (2007), Peterson also came out with a book on the same subject. Both writers claim that applicative verbs in many languages, including Bantu languages, derive from the incorporation of prepositions (P) of prepositional phrase (PP) complements of predicate verbs. P-operation allows two things to happen. Firstly, the basic predicate verb into which P incorporates becomes a derived applicative verb with often a valency change, e.g. (in)transitive => (di)transitive. Secondly, after P-incorporation, the complement of P becomes a core argument, e.g. direct/indirect object of the applicative verb. (1-2), from Kiswahili Bantu, illustrate the operation. 1. Mw-anafunzi a-li-kw-end-a kwa mw-alimu. Cl. 1-student Cl. 1 SM-PAST-STRESS AFX-go-MOD Cl. 17/26 SM-COP-n 'A' to Cl. 1-teacher 'The student went to the teacher.' 2. Mw-anafunzi a-li-mw-end-e-a mw-alimu. Cl. 1-student Cl. 1 SM-PAST-Cl. 1 OM-go-APPL-MOD Cl. 1-teacher 'The student went to the teacher.' Our paper explored whether or not a P-incorporation that derives (2) from (1) is, indeed, systematic and valid in Kiswahili Bantu, too. For simplicity's sake, we used the basic verb -enda and a couple of other illustrations. Kiswahili Bantu internal evidence reveals that PPs collocate with applicative verbs all the time. The pattern renders P as the source of applicative morpheme in V untenable. There are also intransitive applicative verbs in the grammar that are not derived via P-incorporation, because Kiswahili Bantu does not have intransitive P, i.e. one that stands after V or a phrase without a complement. We conclude that the theory of P-incorporation as the source of the applicative, at least in Kiswahili Bantu, is weak, in theory and principle.