MIT CogNet, The Brain Sciences ConnectionFrom the MIT Press, Link to Online Catalog
SPARC Communities
Subscriber : Stanford University Libraries » LOG IN

space

Powered By Google 
Advanced Search

 

Mapping between Concepts and Grammar: The Case of Psych Verbs

 Holden Haertl
  
 

Abstract:
In strictly modular models of language generation, e.g. Levelt (1989), the mapping of non-linguistic, conceptual structures onto language specific, grammatical representations requires a definition which is highly restrictive in nature. A test for an appropriate conception of the mapping mechanisms is provided by the phenomenon of psych verbs which systematically realize their seemingly identical thematic arguments in divergent syntactic positions (henceforth the most adequate English translation of the German examples will be given):

1) E-S-verb: John fears Mary.
2) S-E-verb: Mary frightens John.
(Mary = STIMULUS; John = EXPERIENCER)

The goal of the current study is to motivate this divergence both conceptually and grammatically. It will be shown that the two verb groups given above differ with regard to their thematic - i.e. non-linguistic - structures. This difference is reflected in event-structural features that are encoded grammatically. The argumentation is based on theoretical work as well as (quasi-) experimental evidence.

Contrary to the assumption that only S-E-verbs are causative (cf. Grimshaw (1990)) I will argue that psych verbs generally express (implicit) causality on the conceptual level: Both S-E- and E-S-verbs can be characterised by the causality features of covariation (cf. Brown & Fish (1988); Haertl (1999)). Additional evidence comes from a questionnaire study which shows that for both (German) verb groups subordinate causal sentences are preferably assigned to the STIMULUS-entity:

3) John fears Mary because [she/??he] is untrustworthy.
4) Mary frightens John because [she/??he] is untrustworthy.

However, it is the thematic structure that distinguishes S-E- from E-S-verbs. The following two empirical findings support this. First of all, another questionnaire study was conducted in which the animacy - i.e. a thematic feature - of the STIMULUS varied. The results show an ordinal interaction of verb group and animacy: The causal sentences were more frequently assigned to an animate STIMULUS in E-S-verb complexes. Furthermore, an Event-Related Potential (ERP) study was carried out using the material given under (3) and (4). For S-E-verbs the unexpected assignment of the causal sentence to the EXPERIENCER - indicated by the respective pronoun - evoked an N400 component whereas the same assignment evoked a P600 component in the case of E-S-verbs.

The differences can be interpreted as a reflex of divergent thematic structures: Only S-E-verb complexes preferably denote a temporally homogeneous activity in which an intentionally acting individual is involved. Thus, S-E-verbs can be analyzed as common activity verbs such as 'dance' or 'hunt' that normally realise an AGENT in the subject position. This assumed event-structural difference was tested directly in a third questionnaire study. Subjects were asked to judge on a five-step scale the acceptability of causal sentences that occurred either as temporal activities or as states. The causal sentences were all assigned to the STIMULUS in the matrix clause. Consider the following example of an S-E-verb - note that German does not have explicit aspect markers:

5) S-E-verb: Mary frightens John
because she is doing something now. (= activity)
because she has a certain property. (= state)

In agreement with the hypothesis the results show that S-E-verbs are preferably realized as activities whereas with E-S-verbs a stative reading of the STIMULUS-eventuality is judged to be more acceptable.

 
 


© 2010 The MIT Press
MIT Logo