Various views at dichotomy in sentence information structure from the perspective of an annotated corpus
Abstract
We present examples how the data from the Prague Dependency Treebank can be used in linguistic research, allowing to study the relationship among surface syntactic structure, deep syntactic structure including the sentence information structure, and the overall text structure. Thus we could have come to the following conclusions: (1) In general, it cannot be claimed that the topic-focus dichotomy (in any terminology) can be based on the difference between the old (given) and the new information. (2) In addition to the global division of a sentence to topic and focus, it is appropriate to work with local segmentation applied at all layers of the sentence structure. (3) For linking the sentences in the text in terms of their division into topic and focus, it seems to be more typical (for Czech, unlike English) linking of the topic of the given sentence to the focus of the preceding sentence (so-called linear sequence).
