A computer implemented system, plug-in application and method for composing a formatted text input to improve legibility, readability and/or print economy while preserving the format of the text input and satisfying any user selected aesthetic constraints. This is accomplished by reading in blocks of text input having defined characters including letters and punctuation in a given input format. A language unit such as a lexical or sub-lexical unit, a subset of punctuation or another defined unit for a particular language is examined and an information measure (IM) is assigned to each character in the language unit indicating the predictability of that character to differentiate the language unit from other language units. Typically, multiple different IMs are assigned to each character and combined to form a combined IM (CIM). The process is repeated for at least a plurality of language units and typically until all the text input in the block has been analyzed and information measures assigned to all of the characters. An adjustment to a physical feature is determined for each character in the plurality of units to modify the visual prominence of that character according to the values of the assigned information measures and a permitted range of physical variation for the block. The adjustments are applied to each character to compose the text input consistent with the input format.