Abstract

The paper shows how semantic grammars (also known as text or story grammars) can be used as content analysis coding schemes for narrative texts. The grammar should be of particular interest to investigators involved in socio-historical research. After all, its simple linguistic structure subject-action-object and possible modifiers (such as time and space for action) fundamentally deals with events (indeed, social actors doing something pro or against other actors). When implemented in a computer environment, a grammar has several desirable properties compared to traditional content analysis coding schemes: 1. it preserves much of the original narrative text; 2. it preserves the relations between the various elements of the grammar (e.g., subjects related to actions, actions to objects, time and space to action); 3. it allows investigators to collect far richer and more reliable data; 4. it allows a range of statistical analyses of what are essentially words: from network models to Geographic Information System models, along more traditional regression-based models. Despite its advantages, a story grammar approach to content analysis does have its limitations: it works well only with narrative type of texts. It does not fare well with description, evaluation, or abstraction.