In a sense, the following question is a sequel of this one: Usage of the implicit objective subordinate clause in English.
In that question I asked some information about the usage of the implicit object-subordinate in English in place of the corresponding explicit one. For instance, as I wrote there, while speaking English we usually prefer:
(a) they want me to do this
(b) they want that I do this
My way of analysing the sentence (a) was as follows:
(x) There are two clauses: "they want" is the principal and "me to do this" is an object-subordinate, functioning as the object of "want". The syntactic subject of the verb "to do" of the subordinate "me to do this" is "me" and it is in the accusative case.
Anyway I was told that a more correct way of analysing (a) is:
(y) "me" is in the principal clause and it is the syntactic object of "want". It is just the semantic subject of "to do".
Why did I think that the analysis (x) is sound? Well, as I explained in a comment of Usage of the implicit objective subordinate clause in English, just because this is precisely what we do in the Latin syntax. Indeed, when we say:
(c) vult me ire
(literally, "[he] wants me to go" / "[he] wants that I go") we say that "vult" is the main clause and that "me ire" is its object-subordinate with verb "ire" in the infinitive mood and syntactic subject "me" in the accusative case. Precisely as in English (indeed once I was told that this syntactic construction existed in Proto-Indo-European as well, but I'm not sure of this, I don't have any reference about it).
Now, I noticed that (maybe) the same construction exists in Italian as well. In fact, in Italian we say:
(d) l'ho visto correre
(literally "[I] saw him running" - yeah, in English the syntax is not the one that we are discussing, as "running" is not in the infinitive mood; anyway in Italian it is, "correre" is in the infinitive mood) where, if we reorder the sentence a little bit (see the remark below), we get:
(d') [io] ho visto lo correre
[l' and lo are the same thing in Italian].Then we may analyse (d') as follows (this is just an hypothesis, I don't know if I'm right):
(z) The sentence "[io] ho visto lo correre" is made of two clauses. "[io] ho visto" (I saw) is the main one, while "lo correre" is its object-subordinate (in Italian "ho visto", "vedere", "to see" is usually transitive, so this makes sense). Moreover "correre" is a verb in the infinitive mood and "lo" is its syntactic subject and, again, it is in the accusative case! In fact "lo" in Italian is precisely the accusative of "egli" ("he").
I don't know if the analysis (z) could be considered correct. As I said, it is just an hypothesis. I found it amazing how this typical Latin syntactic construction, usually called "infinitive-accusative subordinate", can be rediscovered in English and in Italian (even if my analyses (x) and (z) may be incorrect), but I don't know if it actually comes from the Latin (or Proto-Indoeuropean) one. I would really like to read some of your opinions about this curious fact. Do you think my hypothesis may be correct or this is just a coincidence?
Remark. LjL and Denis N. (see below through the comments) observe that (d'), which is (d) together with an edited order of its words, is actually ungrammatical. Therefore Denis N. suggests the following sentence:
(d'') [io] ho visto te correre
(that is, "[I] saw you running") as an alternative example of the phenomenon explained above. Here "te" is the italian accusative of "tu" ("you").