José Manuel de la Rosa Govantes


Organization of text comprehension in technical systems

Image of Dipl.-Ing. José Manuel de la Rosa Govantes Copyright: © José Manuel de la Rosa Govantes

Summary of the dissertation by Dipl.-Ing. José Manuel de la Rosa Govantes

This dissertation focuses on the success of individual mass correspondence. The investigations are carried out under the special framework conditions that are found in very large organizations such as insurance companies, banks or public authorities: in a dialogue lasting for years to decades, a large number of institutional members communicate with an even larger group of clients and carry out life-relevant ones Actions that are implemented with individualized, but mostly automatically generated, documents.

Empirical studies confirm the widespread, subjective perception; the corresponding letters are usually difficult to understand, are also impersonal and can be used, for example. Do not put off the connotation of their often mechanical production. The results of this dissertation should contribute to the improvement of the situation, so it has al-so a strong practical reference.

The title of the work is deliberately shortened, since comprehensibility is to be regarded as a necessary but not sufficient condition. Rather, the analysis of internal characteristics of texts alone is not sufficient to meet the demand for operationalized concepts or solutions. The latter, in turn, motivates motivation as well as opportunity for this dissertation, since the long-term, mainly professional occupation with the topic - u.a. in technical and technical linguistic and text projects with insurance companies - access to extensive, up-to-date text files as well as contacts to numerous experts from the industry. Last but not least, these possibilities are flanked by a deep understanding of the technical and organizational structures and processes of large organizations.

The methodology of the work reflects its interdisciplinary character: a comprehensive, summary presentation of the multidimensionality of the overall problem goes further than previous studies and touches on numerous fields of research. Although the main focus lies in the networking of the current state of research in linguistics (textual, corpus and structural linguistics, comprehensibility and readability research, speech act and action theory) with the possibilities of computer science (especially linguistic computer science, eg text Mining, in addition to pattern recognition, semantic meshes, etc.) in order to automatically determine relevant correlations in text corpora and to be able to make statements about the quality and style of a text sample. The relevant research is accompanied by empirical studies and supplemented by exchanges with leading institutions. With similar priority, the prior art must also be raised in readability indices and text production systems, for which i.a. Expert interviews and test series are scheduled.

In addition, however, there are findings in psychology and neurobiology (on mental models of cognitive processing), media theory (on communication media), sociology, organization theory and also philology (on the text-producing organization) as well as the language and analytic philosophy and Epistemology (on ontologies and taxonomies), which should at least be considered and investigated for targeted impulses to answer the research question:

Is it possible with probabilistic means to make automated statements about quality and style or even the illocutionary role of a present document?