This study is a discourse analysis and deconstruction of public documents published electronically in connection with the evaluation of the Advanced Placement Language and Composition Examination, found on the educational website: apcentral.collegeboard.com. The subject of this dissertation is how the characteristic of writing identified as Voice functions covertly in the calibration of raters' evaluation of student writing in two sets of electronic commentaries: the Scoring Commentaries and the Student Performance Q&A;'s published between the years 2000-2010. The study is intended to contribute to both socio-linguistic and sociological research in education on the influence of inherited forms of cultural capital in educational attainment, with particular emphasis upon performance on high-stakes examinations. Modeled after Pierre Bourdieu's inquiry into the latent bias revealed in the "euphemized" language of teacher commentary found in The State Nobility, lists of recurrent descriptors and binary oppositions in the texts are deconstructed. The result of the deconstruction is the manifestation of latent class bias in the commentaries. Conclusions: discourse analysis reveals that a particular Voice, expressive of a preferred social class identity, which is initiated to and particularly deft in such academic performances, is rewarded by the test evaluators. Similarly, findings reveal that a low-scoring essay is negatively critiqued for being particularly unaccustomed to the form(s) of knowledge and style of writing required by the test situation. In summation, a high score on the AP Language Examination, rather than a certification of writerly competence, is actually a testament to the performance of cultural capital. Following an analysis of the language of classification and assessment in the electronic documents, the author provides several "tactics" (after de Certeau) or recommendations for writing the AP Language and Composition Examination, conducive to the stylistic performances privileged by the rating system.