• Keine Ergebnisse gefunden

Main effects of object animacy: The results of the study show robust main effects of object animacy, both for sentences in the accusative and in the da-tive condition. Reading times for adverbs and critical verbs were longer if the objects of the sentences were animate than if they were inanimate. I interpret these longer reading times as indicating higher processing costs for the animate conditions. This main effect of object animacy fits the expectations formulated above, and is in line with findings from the literature (Frisch and Schlesewsky, 2001; Grewe et al., 2007; Trueswell et al., 1994; Weckerly and Kutas, 1999). This main effect of object animacy was already measurable on the adverb, indicating the incremental build-up of a sentence context using animacy information.

Main effects of verb class: There were no statistically significant effects of verb class.

Interaction between object animacy and verb class: On the first postverbal

3. Self-paced reading time study

word und, there is a statistically significant interaction between object animacy and verb class for the reading times of this position. In the accusative condition, the reading times for und are longer after an animate than after an inanimate object. The influence of object animacy is thus the same that it was on the two preceding words. In the dative condition, however, there is no statistically signif-icant difference between the reading times for the postverbalund in the animate and inanimate conditions.

This interaction effect supports our initial hypothesis, indicating that the ob-ject animacy effect on sentence processing is indeed affected by the processing of verb class information. The difference between the object animacy effects in the accusative and dative conditions is visible once verb class information has become available (i.e., once the critical verb has been read).

In the General Discussion (see Chapter 6), I will offer some suggestions for the linguistic mechanisms possibly underlying this interaction. Here, I will discuss the experimental method and its possible influence on the time course of the ef-fects, i.e., why the interaction effect was only visible on the postverbal und, not on the critical verb.

One possible explanation is that the interaction between object animacy and verb class reflects a late processing step (e.g., beginning reanalysis) that only happens while the participants have already pressed the key to read the next word. While this makes sense given the results of the self-paced reading time experiment, the results of the remaining experiments presented in Chapters 4 and 5 do not support this interpretation of the current study.

Another possibility is that the time course of the effects is influenced by the unnatural reading paradigm. In non-cumulative reading, all the words have to be memorised while the next words are presented in order to build a representation of the whole sentence (and, in this experiment, to be able to answer the ques-tions). The high memory workload might delay the processing of a word and its integration into the sentence context, so that the reading times of the following words are affected and the measures of increases in processing cost spill over onto the next words.

Another possible explanation is that the readers perform only the access to

the lexical information of the verb during the reading time of the verb. Then, after having pressed the key to see the next word, they incrementally integrate the verb into the sentence context, while at the same time they perform the access to the postverbal und. Under this explanation, it is quite natural for the reading time of a given word to be affected by the syntactic and semantic properties of the preceding words, together with its own lexical properties, without having to assume an influence of memory workload.

Irrespective of the explanation for the time course of the interaction effect chosen, the results illustrate the fact that compared to eyetracking and ERP measurements (used in the experiments described in Chapters 4 and 5), self-paced reading times are a ‘late’ measure. Another example for this is the time course of the main effect of object animacy, which is statistically significant in both the F1 and F2 analyses only on the adverb, but not on the object NP.

This first experiment suports my initial hypothesis. However, some criti-cal points remain that I will adress here. The interaction occurred on the first postverbal word (und). This a very short function word that always appeared at the same position in the critical sentences, and therefore had rather short reading times in general. I assume that after the first few sentences, the participants in the experiments were accustomed to the sentence pattern and therefore pressed the key quite mechanically when this small, highly predictable word appeared on the screen. This might have shortened the reading times for und, making it more difficult to interpret effects found at this position.

The critical sentences were not interspersed with filler sentences in this first experiment. I assumed that the response data would indicate if participants became bored and did not read (and therefore comprehend) the sentences prop-erly. This assumption might have been too optimistic, as the absence of filler sentences might still have contributed to some kind of syntactic priming. Read-ing a sequence of very similar sentences certainly did cause weariness with the participants.

However, if syntactic priming occurred, it should have affected all conditions, and its effects would have been evenly distributed because of the pseudoran-domisation method chosen. I do therefore not assume that the interplay between

3. Self-paced reading time study

object animacy and verb class visible in the results can be explained as a spurious interaction effect caused by the lack of true randomisation and filler sentences.

However, it is possible that the monotony of reading the same sentence pattern again and again could have weakened both the effects of object animacy and the interaction.