Volume 42, Issue 2

The present article contributes to the discussion of methods of corpus studies in the field of text and discourse analysis. Starting with a brief overview of bottom-up and top-down methods of corpus analysis, the article focuses on a critical examination of Rhetorical Structure Theory (RST), a framework for describing natural text that has been widely used for the annotation of texts, especially in the field of computational linguistics. This critical assessment, which is based on an action-theoretic theory of text organization, is, at the same time, used as a means of discussing some quality criteria for textual analysis. After a brief introduction to basic tenets of RST, the article discusses various features of this framework that appear problematic from a theoretical point of view and that tend to cause difficulties for practical analysis. The problems connected to these features include problems of granularity associated with the “classic” set of “rhetorical relations” as well as problems connected to the adjacency and nuclearity conditions. Furthermore it is shown that RST lacks adequate instruments for the description of the topic structure of texts and aspects of linearity and the dynamics of texts, e.g. the discontinuity of functional and thematic elements. The final section of the paper summarizes the results of this critical analysis, briefly reflecting on conceptions of multi-level annotation and suggesting some further steps in the corpus-based analysis of textual organization.

One of the most significant tasks in forensic analysis of written discourse is to examine whether a text is genuine or in some way falsified. In order to deceive the addressee of a blackmail letter, authors often try to disguise their mother tongue by introducing different kinds of artificial errors. The paper firstly deals with the difficulties one has in judging the authenticity of an error, the cognitive conditions of deliberate error production, and its intended purposes. This is followed by a presentation of the results of a study carried out with 98 German native speakers who were asked to write a blackmail letter simulating a non-native speaker identity. The focus of the corpus analysis lies in the description and interpretation of the morphological and syntactical errors produced by the participants. It shows that many participants singled out identical grammatical categories, and produced similar patterns of errors, resembling the defective structures of German foreigner talk . The results from this study suggest that deliberate error production is affected by the cognitive load caused by the asignment, and is automated to a higher degree, and that its patterns are noticeably shaped by the grammatical structures of the mother tongue.

Drawing both on conversation analysis and text linguistics, this article retraces the emergence of a new communicative practice in an online discussion forum: Based on data from an academic learning environment, we demonstrate how peers in a student study group start using the “edit button” which allows them to modify in retrospect not only their own but also other people’s posts. This communicative practice of post editing in online discussion forums develops in four stages: It starts out as simple postings of messages on a discussion board. Next, the collocutors make use of the edit button to change their own posts, and, in a dialogical manner that of their discussion partners. Finally, it comes down to a complex form of exchange between the interlocutors who innovatively use the edit button within a single post. By using the edit button in innovative ways the participants bring together sequentially related messages in a single post that are usually spread over multiple posts. We argue that the emergence of this innovative use of the strategy of “sequential compression” ( sequenzielle Verdichtung ), as we shall call it, may be understood as an answer both to the affordances of asynchronous communication in discussion forums and to the learning situation which is characterized by pressure of time calling for new and innovative strategies.

In this article a research programme is sketched, together with an outline of the theory of Instruction Grammar embedded within it. In the research programme, verbal action is viewed as a means for organizing and coordinating lifeworld interaction. Verbal interaction is a means to bring about action on the part of the interlocutor in order to realize purposes. Both the well-formedness and the situational appropriateness of utterances have to be determined with respect to the success and efficacy of that purpose. The competence of well-formed and appropriate verbal interaction (the “linking competence”) is the subject matter of the theory of Instruction Grammar. This competence draws on a number of sub-competences, ranging from action, perception, conceptualization and attribution to symbolization. In the research programme, all these sub-competences are constitutive of the linking competence whose exploration requires a multidisciplinary approach based on converging evidence. The basic claims of the theory of Instruction Grammar are then deduced using an example scenario. Utterances are characterized as instructions for an interlocutor to simulate perceptions, reenact attributions, and engage in action by himself.