The present study investigates the lexical and phraseological resources associated with the communicative functions of grade comments written by a group of teachers in the context of a Southern California English language program. This study seeks to establish a core vocabulary and structure for this text type, thereby reducing the overall demands associated with its composition. The study takes a corpus-driven and keyword-centered approach to the lexicogrammatical characterization of a corpus of grade comments totaling over three million words. The corpus contains five sub-corpora, each corresponding to one of the five different institutionally-established letter grades: A, B, C, FP, and F. Keywords for each letter grade were obtained through the comparison of one letter grade sub-corpus against the remaining sub-corpora. The keyword results show that the core vocabulary of each letter grade consists of words commonly used in the U.S. grading system and graded subareas of different classes as described in their corresponding institutional syllabi. The top 20 most-key keywords were then used to qualify lexical bundles found in each subcorpus. The novel methodology adopted in this study has led to the identification of lexical bundles that correspond to the core communicative functions of the grade comments. Furthermore, this particular approach to lexical bundles has suggested a possible connection, at least in grade comments, between keyword-containing lexical bundles and framing devices for generic features, specifically opening and closing moves. The keyword-centered approach requires additional exploration, but it may have an application for researchers taking a corpus approach to genre analysis. The relationship between keywords, lexical bundles, communicative functions, and generic moves also seems to be a worthy area for further investigation.