5.1.2.2.1. Thematic Hierarchies

Thematic Hierarchies are by far the most widely used method to explain the mapping from semantic representation to syntax. Fillmore (1968) was the first to formulate a hierarchy of "cases" (semantic relations) to help determine subject selection. After him, most theories make use of a mapping between an ordered list of semantic roles and an ordered list of grammatical relations. Given a thematic role hierarchy (agent>theme ...) and a syntactic hierarchy (subject>object ...), the mapping usually proceeds from left to right, mapping the semantic role further to the left onto the first unoccupied position in the syntactic hierarchy. Thus, rather than having invariable correspondence relations, typical of the direct mapping approaches, the mapping is controlled by strategies relative to the hierarchies.

As was the case with thematic roles, what the appropriate thematic hierarchy should be and how it works is a quite controversial matter. There is general agreement that the agent role should be the highest ranking role, but there is no consensus about the ordering of the rest of the roles. Simply to illustrate the diversity of proposals, consider the ones collected by Levin and Rappaport (1996):

It should be taken into account that these proposals are made from widely differing theoretical stands and that at times there are differences in what is being hierarchisized. Nevertheless, most of the disagreement lies in where to locate the Theme with respect to other roles, especially the Goal and Location roles. Two reasons explain the difficulty in locating the theme. First, Theme/Patient arguments can be both subjects and objects. The second reason is that the Theme/Patient competes with the goal argument to be the first object of verbs that take double objects.

One of the advantages of thematic role hierarchies is that they capture generalizations about subject and object selection that apply across many semantic classes of verbs. Their main advantage, however, is that they can be explanatory, if there is independent motivation for the hierarchy and for its relevance in the mapping process. Several types of justification have been given in the literature. Sometimes they have been justified on pragmatic grounds. For example, functional linguists such as Givón (1984) have proposed that the hierarchy reflects the degree of topicality of arguments, with the highest being most topical. The selection of subject becomes important because the subject is usually the primary topic of the sentence. It is logical, then, that the highest thematic role in the hierarchy occupies that position. The object is the secondary topic. Therefore, it should be assigned to the next highest role in the hierarchy. More often, the justification for the hierarchy has been semantic. Scholars, such as Larson (1988), believe that the hierarchy is a reflection of the order of composition of arguments with the verb. The argument lowest in the hierarchy combines with the verb first, whereas the argument highest in the hierarchy combines last. Subject/object asymmetries in English idioms are often used as evidence for this interpretation of the hierarchy.