The Interplay of Synonymy and Polysemy

23 | 
2.4.1 Jackendoff (1990) and Pustejovsky (1995) as Schemas 
First of all, I want to note that the major aspects of the theories proposed in 
Jackendoff (1990) and Pustejovsky (1995) can be captured using schemas and links. Both 
authors include argument structure as part of their lexical structure. This can be easily 
captured by schemas. For example, the (simplified) argument structure for the verb 
kill
is 
shown in the following schema.
Figure 3
. Schema capturing the argument structure for the verb 
kill
If we limit our discussion to the most prototypical uses of 
kill
, the argument structure 
is shown in figure 3. The entire box is the schema for the verb. The boxes inside would be the 
e-sites (Langacker 2008), the participant roles (Goldberg 1995) or the arguments 
(Pustejovsky 1995 and Jackendoff 1990). The exact choice of terminology is irrelevant
2
. Part 
of the meaning of the verb is the implication that there are two participants. The first is some 
type of agent, while the second is an animate being. To build a sentence with this verb, one 
must simply fill in those slots with a noun that fulfills the semantic constraints imposed. In 
other words, for a sentence to be correct the second element must be an animate entity. All 
this information is captured in the schema.
The type of information that Pustejovsky (1995) captures in his Qualia Structure and 
that Jackendoff (1990) includes in his 3D model are certainly compatible with the semantic 
information that a speaker will associate with the semantic side of a construction. That means 
that even from a CG perspective, part of the knowledge of the noun 
book
will be that it is 
made of paper, that it is usually rectangular, that it sometimes has a hard cover and 
sometimes a soft one, that it is meant to be read, that it is written by people termed authors, 
and so on and so forth. Though, each author may have different views about the type of 
information that should be captured, my CG perspective can include many of the same types 
of semantic information as both Jackendoff (1990) and Pustejovsky (1995).
Jackendoff argues that some meaning extensions apply to all his primitives. For 
example (§2.3.2), he describes that the primitive BE can mean different things depending on 
the semantic field that is under discussion (1990:26). This can be captured with a schema. 
2
Throughout this paper I often borrow terminology from the various authors and use them as synonyms. For 
example, sometimes I talk of schemas and then use constructions to refer to the same objects. Participant 
roles and arguments are also used interchangeably. The exact terminology does not affect the analysis.
A
GENT
kill 
A
NIMATE

---