Traditional readability formulae use features of wording (e.g. sentence length and vocabulary frequency) as if these features related directly to the comprehensibility of the text. They seem to assume, for example, that long sentences are harder to understand than short ones. Unfortunately, the relation between a text and its context is such that even the best scheme for counting features of the wording of a text can provide only indirect guidance as to the comprehensibility of that text. This is true, because although people interact using language, the text produced does not contain the interaction. Rather, it encodes the interaction. Texts are semantic units. Therefore, examining the words and grammar of a text to evaluate the coherence or readability of that text must involve interpreting the language in context. One must consider the interactional context because a given meaning may be valued highly in one interactional context, and may be valued quite differently in another interactional context. One must also consider the linguistic context since texts create patterns of meanings which influence the perception of the parts. No single feature or set of features (e.g. cohesive ties) make the text coherent.

[Close window to return to Publication]