What Characterizes Pictures and Text?

This paper addresses an apparently trivial question: what is the difference between graphics and text? It appears to be trivial because there appears to be several alternative and simple ways of answering it. For example, 'text is made up of letters whereas graphics are not', 'one can create text using a keyboard', 'one can read text aloud', etc. However, none of these provides robust conditions to differentiate graphics from text, e.g. cases such as typewriter art and gobbledygook can be identified. The paper approaches the problem of identifying content conditions by analysing boundary cases which lie on the margins and are difficult to classify. It considers examples that arise in the production of materials, including bitmapped text, graphics consisting of letters and words, text used as patterns or in tables, etc. It also considers examples that arise from the consumption of materials, including a comparison of the methods used for reading and interpreting text and graphics. This paper concludes that current XML specifications, e.g. TEI guidelines, for the integration of graphics into text are primarily made on the basis of form rather than content. This is incompatible with a content-based markup scheme. Before such