I want to talk about possession and transfer predicates, but first I need to make a digression on the issue of raising. As so often happens with Lojban, the notion of "sumti raising" we inherited from Lojbanic lore is only vaguely related to the notion of raising as understood in linguistics, so it's worth looking into it first:
"In linguistics, raising is the construction where a given predicate/verb takes a dependent that is not its semantic argument, but rather it is the semantic argument of an embedded predicate. In other words, an argument that belongs to an embedded predicate is realized syntactically as a dependent of a higher predicate/verb. Not all languages have raising predicates, but English is one that does."
http://en.wikipedia.org/wiki/Raising_(linguistics)
We might add that Lojban is another language that has raising, much like English. Examples of raising verbs in English are:
Tom seems to have won the race.
That proves Susan to be a jakass.
In the first example Tom is raised to subject of "seem" (compare with "it seems that Tom has won the race"), in the second case "Susan" is raised to object (compare with "that proves that Susan is a jackass".
Raising predicates can be contrasted with control predicates that have some features in common, but in which the argument syntactically selected by the higher predicate is also semantically selected. Compare the bahavior of the raising verb "want" with the behavior of the control verb "tell":
Susan wants John to write the letter.
Susan wants the letter to be written by John.
Susan wants that John write the letter.
Susan tells John to write the letter.
?Susan tells the letter to be written by John.
*Susan tells that John write the letter.
In the case of the raising verb "want", all three sentences mean basically the same thing, "want" doesn't assign any special semantic significance to the raised argument. In the case of the control verb "ask" the meanings are completely different.
In any case, the difference between raising and control predicates is not my main concern. My main concern is the difference between raising and something completely different, which is this:
Susan wants chocolate.
Susan wants to eat chocolate.
That is NOT an example of raising in the linguistic sense. In "Susan wants chocolate" there is no raising involved, chocolate is just what Susan wants, Compare with "Susan wants John to write the letter", where John is not what Susan wants, but it is raised from the subordinate clause that describes what Susan wants.
Lojban has many raising predicates. "simlu" is the obvious one corresponding to English "seems". We can see that it's a raising predicate with these transformations:
la djan cu simlu lo ka ce'u dunda lo tinba la meris
lo tinba cu simlu lo ka la djan dunda ce'u la meris
la meris cu simlu lo ka la djan dunda lo tinba ce'u
lo du'u la djan cu dunda lo tinba la meris cu simlu lo ka ce'u fatci
which all mean basically the same thing. The predicate "simlu" raises an argument from x2 into x1 to put it in focus, but not with much semantic significance.
All predicates in Lojban that have a property place and a place for the thing with the property are candidates to be sumti raising predicates or control predicates. If it makes little or no semantic difference which argument from the subordinate property is raised as a syntactic argument of the superordinate predicate (as in "simlu"), then it's a sumti raising predicate. If it makes an important difference which argument is selected, then it's a control predicate (for example "minde").
mu'o mi'e xorxes