PUB - Publikationen an der Universität Bielefeld

Language is a unique and core human ability. Language is abstract and arbitrary and yet it enables us to communicate with each other. Language allows communication and communication is inherently social. Communicating with and about others is of highest interest for humans, as humans are social beings. This is why receiving human feedback is often extremely emotional. Although we have an extensive knowledge about the neuronal bases of emotional language processing, there are only a few studies yet conducted to investigate socio-communicative influences on language processing.

In my dissertation I examine the influence of a social communicative partner on emotional language processing. Three studies systematically manipulated the expertise and identity of putative interaction partners. These interaction partners gave feedback on positive, negative and neutral adjectives while a high-density Electroencephalogram (EEG) was recorded. Actually, in all conditions random feedback was presented, thus a differential processing could only be attributed to sender characteristics. By means of event-related potentials (ERPs), the influence of sender characteristics, emotional content and their interaction was observed. In studies I and II - as a proof of principle - a 'human sender' was compared to a random computer (unequal expertise, unequal humanness). In this study, both feedback anticipation (study I) as well as feedback presentation was investigated (study II). In study III the 'human sender' was compared to a socially intelligent computer (similar expertise, unequal humanness). Eventually, in a fourth study a 'human expert' was compared to a 'layperson' and a random computer sender (unequal expertise, but the 'expert' and 'layperson' were both 'humans').

During anticipation of 'human' feedback, an extremely early enhanced general processing was found. On later stages a more intense processing of emotional adjectives was found in the 'human sender' condition. In general, effects during feedback presentation were substantially larger than during feedback anticipation. Here, large effects were found on early and late ERP components, for both human-generated and emotional feedback. Further, emotional feedback given by a 'human' was additionally amplified. Eventually, in study IV 'expert-feedback' was processed most intensely, followed by 'layperson-feedback' and finally 'computer-feedback'. Localization methods found enhanced sensory processing for 'human-generated' and emotional feedback. Studies III and IV showed additionally increased activations in somatosensory and frontal effects for 'human senders'.

Overall, these experiments showed that not only emotional content but particularly also communicative context influences language processing. We automatically seem to take context factors into account when processing language. Here, 'expertise' results in an enhanced processing aldready on early and highly automatic stages, while supposed humanness seems to be of highest relevance: 'Human-generated' feedback led to enhanced processing in sensory, but also somatosensory and frontal areas. This shows that in human interactions language is amplified processed, which is especially true for emotional language. This dissertation shows for the first time that in realistic communicative settings (emotional) language processing is altered. Here, it seems that first sender information is processed, while emotional content affects later processing stages. The use of state of the art source localization methods enabled to get next to the extremely high temporal resolution (when something happens), a good and reliable spatial resolution (where something happens) of the cortical generator structures of the ERP effects.