From Textop Wiki
On this page we will propose and refine some rules for chunking texts. Proposed rules may be found flush left in bold.
A completely chunked text will select out for chunking all definitions, explicit distinction-making, arguments, and so forth that can be found in a text, no matter how much overlapping the text is.
- Rationale: there can be no clear, uncontroversial reason for chunking only some definitions and not others, only some arguments and not others, and so forth. While in the judgment of an expert, some points might be ancillary and not very interesting, other experts might well disagree. In any event, a large body of texts, exhaustively chunked in this way, will provide a comprehensive picture of all the different "resources," so to speak, to be found in the texts. The more selective you are about what to chunk, the less reliable and useful the result will be for researchers.
- Explanation: this rule is admittedly not clear in its present formulation, and perhaps there is no way to make it very clear. One way to explain it is to say that all "significant points" will be made into chunks, but that just leaves open the question what a "significant point" is. As a general rule--in need of study and discussion--we might say that most listed linguistic functions, such as definitions, explanations, etc., if distinguishable, should be made into chunks. Exceptions are most propositions and examples that are parts of arguments, explanations, and descriptions. Propositions and examples are made into chunks only when they are set apart by the author or dwelled upon (elaborated) at some length.