Contradictions are inherent in the evaluation of placement test writing, contradictions that at once value and devalue writers and writing, readers and reading. In testing, the evidence for the essay's effectiveness rests almost entirely on the writer's choice of linguistic forms. The characteristics that distinguish evaluation in competency testing are (1) the final evaluation of what are, in effect, first drafts; (2) the use of those first drafts to assess both text quality and writer ability; and (3) the centrality of linguistic forms as evidence for both. These characteristics create complex dilemmas for writers and readers in negotiating appropriate placements. A taxonomy of new, inferable, and old information in noun phrases (developed by E. Prince) was used to evaluate essays written by 99 incoming students at Temple University (Pennsylvania). While analysis showed that the more complex forms of unused and of brand-new information influenced reader judgments significantly, these entities did not seem to function the way the taxonomy predicts, had writers and readers actually been cooperating with each other in the exchange of information. (A copy of the information taxonomy as well as samples of evaluated essays are included.) (NKA)