The issue of contextual punctuation studies on right dislocation and afterthought

Janina Kalbertodt
IfL Phonetik, University of Cologne

The right periphery is linked to distinct information structural functions. Most researchers agree on the form, e.g. the prosodic and syntactic structure, and the function of right dislocation and afterthought – at least in spoken utterances. But only little research has been done on the use of right dislocation and afterthought in written language. In a previous study we found that listeners are able to identify these structures in written novels correctly. But it remains unclear what the main cue for identifying those constructions is: it could either be the preceding context or the punctuation of the utterance; or even a complex interplay of these parameters.
In the present studies, we concentrated on the contextual aspects of right dislocation and afterthought to investigate whether context represents the main cue for identifying these two types of constructions correctly. The difference between right dislocation and afterthought is that in the context of a right dislocation there is only one continuous topic, while in the context of an afterthought there are two competing topics. Furthermore, theories on German punctuation suggest that – because of this difference, among others – right dislocations are marked with a comma, afterthoughts with a full stop.
With this in mind, in the first experiment participants were given the preceding context and the non-punctuated target sentence and their task was to punctuate the target sentence with the punctuation inventory reduced to full stop and comma. In the second experiment the participants received the preceding context and the already punctuated target sentence; here the participants had to decide whether the punctuation of the target sentences was correct or not. Both experiments did not reveal clear punctuation preferences.
These results suggest that punctuation is not an adequate parameter for measuring the contextual understanding of a text. In many cases, there is not enough competence on the participants’ side and, hence, the results are not fully reliable.