Appendix 1: A note on the difficulty of "literal meaning" "Alice finished the book"
The example is suggested by Pustejovsky: to interpret "Alice finished the book" you are compelled, in Pustejovsky's argument, to generate two readings, depending on the restrictions on "book" (a physical object which is read and written):
Apparently, if Alice is a goat and not a woman, you cannot accept prima facie these interpretations (unless the goat Alice is in a fantasy novel). Which answer could be given? Pustejovsky could specify that the "reading/writing" selection are linked to humans. "x finished the book" means "x finished reading/writing the book" if x is a human. This step might make the structure heavier and heavier. A simpler alternative might be to read "x finished y" as
"x terminated a process z regarding y",
where z is unspecified, depending on other information of the context. This would not count against the theory that the first interpretation to be computed is the "literal meaning". The reason is as follows: Pustejovsky claims that the "literal" meaning of "to finish" in the co-text of "to finish a book" is dependent on the object and its relevant properties. We may either abandon this choice or to weaken it. The option suggested above is not an abandon but a weakening. In Pustejovsky's semantics there is a distinction between two kinds of properties of a book: (i) a physical object (ii) something which can be read or written. We might restrict the basic definition of "to finish" to the first level. The more radical alternative is to abandon every link with the object. In this case "to finish" is computed even before arriving at the word "book": "to terminate an operation z". Which operation? It is not really important. Alice has brought to an end some operation. This is the literal meaning of "to finish", to be completed before deciding further specifications given by the context.