19.4.3 Problems with Thematic Roles

Representing meaning at the thematic role level seems like it should be useful in dealing with complications like diathesis alternations. But despite this potential benefit, it has proved very difficult to come up with a standard set of roles, and equally difficult to produce a formal definition of roles like AGENT, THEME, or INSTRUMENT.

For example, researchers attempting to define role sets often find they need to fragment a role like AGENT or THEME into many specific roles. Levin and Rappaport Hovav (2005) summarizes a number of such cases, such as the fact there seem to be at least

two kinds of INSTRUMENTS, intermediary instruments that can appear as subjects and enabling instruments that cannot:

(19.28) The cook opened the jar with the new gadget.

(19.29) The new gadget opened the jar.

(19.30) Shelly ate the sliced banana with a fork.

(19.32) *The fork ate the sliced banana.

In addition to the fragmentation problem, there are cases where we'd like to reason about and generalize across semantic roles, but the finite discrete lists of roles don't let us do this.

These problems have led most research to alternative models of semantic roles. One such model is based on defining generalized semantic roles that abstract over the specific thematic roles. For example PROTO-AGENT and PROTO-PATIENT are generalized roles that express roughly agent-like and roughly patient-like meanings. These roles are defined, not by necessary and sufficient conditions, but rather by a set of heuristic features that accompany more agent-like or more patient-like meanings. Thus the more an argument displays agent-like properties (intentionality, volitionality, causality, etc) the greater likelihood the argument can be labeled a PROTO-AGENT. The more patient-like properties (undergoing change of state, causally affected by another participant, stationary relative to other participants, etc), the greater likelihood the argument can be labeled a PROTO-PATIENT.

Finally, it has proved very difficult to formally define the semantic roles. Consider the AGENT role; most cases of AGENTS are animate, volitional, sentient, causal, but any individual noun phrase might not exhibit all of these properties.

In addition to using proto-roles, many computational models avoid the problems with thematic roles by defining semantic roles that are specific to a particular verb, or specific to a particular set of verbs or nouns.

In the next two sections we will describe two commonly used lexical resources which make use of some of these alternative versions of semantic roles. PropBank uses both proto-roles and verb-specific semantic roles. FrameNet uses frame-specific semantic roles.