Lexical Contribution to Discourse Inference and Interpretation

Experiencer or psych verbs are a notorious problem for the interface between lexical semantics and syntax (e.g., Grimshaw 1990, Pesetsky 1995). This paper shows that the consideration of discourse structure sheds new light on the data observed in the literature. In particular, this paper follows Asher and Lascarides’ (1995, 2003) and Asher and Pustejovsky’s (2000) ideas on the close interaction between lexical semantics and discourse and applies these ideas to object experiencer (OE) constructions in Modern Greek (MGr). Unlike in English, OE constructions inMGr manifest a clear-cut distinction between a psychological and an agentive use. This paper shows that the psychological and agentive reading include two different presuppositions, which are reflected by the main predicate and have to do with the subject and object of the verb. These lexical presuppositions help resolve anaphoric ambiguities and explain the role of clitic doubling observed with Greek psych verbs (Anagnostopoulou 1999). 1 Discourse Relatedness to MGr OE Predicates The analysis and the discussion here are based on Anagnostopoulou (1999), who shows that there are two different readings of OE verbs, the psychological and the agentive readings. Anagnostopoulou argues that the alternation is a matter of syntax. In contrast, this paper argues that the alternation can be understood better by integrating a discourse structure perspective. Example (1) shows the psychological reading, in which clitic doubling of the object experiencer (OE) is obligatory. The interpretation here is that there is a property of “furniture” that “unintentionally” causes the mental state of “bother” to Peter. (1) Ta epipla *(ton) enohlun ton Petro the furniture.nom cl.acc bother the Peter.acc ‘The furniture bother Peter.’ (psychological) In the agentive OE reading shown in (2) the subject is always animate and is interpreted as being deliberately involved in the creation of the emotional state of the experiencer. In this case, clitic doubling is optional and when the clitic doubling phenomenon appears, the agentive reading is obtained by the presence of adverbials like “deliberately” or other modifiers that explicitly state the intention of the agent to create the emotional state of the experiencer. In any other case, the presence of the clitic doubling does not allow the agentive interpretation. (2) O Giannis enohlei ton Petro Giannis.nom bothers the Peter.acc ‘Giannis bothers Peter.’ (agentive) Furthermore, certain discourse conditions imposed by the definite that denotes the experiencer (involving anaphora resolution) have to be respected in order to be able to insert a clitic in such contructions where the agentive reading remains, as is shown in (3b). (3) a. I Maria enohlise ton Petro the Mary.nom bothered the Peter. acc ‘Mary bothered Peter’ b. Prin apo kero dimosiefsa ena neo vivlio. ‘Some time ago, I published a new book.’ #Epitidhes ton enohlisa deliberately cl.acc annoyed/bothered.1sg ton tupografo. the typographer.acc. ‘Deliberately, I annoyed the typographer.’ In (3b) the insertion of a new discourse referent—introduced by “the typographer”—that has not been mentioned in the preceding discourse creates a sense of oddity. This is expected when a definite is introduced because the speaker does not find any direct or indirect connection between “the typographer” of the second sentence and any other discourse referents in (3b).1 Thus, the condition here is that in the agentive uses of OE predicates, with or without clitic doubling, the experiencer should have already been introduced in a previous sentence. 1The connection between “the book” and “the typographer” requires heavy inferencing from the interpreter. I claim that the unusual effort the interpreters are engaged in in order to build this connection makes it quite implausible to build such a bridging relation. Therefore the sentence is odd.