Lexical diversity (LD) has been shown to be a strong predictor of second language (L2) proficiency. However, most current indices combine all word classes into a single measure and thus only capture the broadest patterns of lexical variation. In addition, researchers interested in examining usage patterns across word classes currently lack access to measures of LD that are both more linguistically interpretable and robust to text length. In response to these issues, this paper introduces a methodology for examining part-of-speech (POS)-specific LD indices (e.g., verb and noun diversity) that apply the moving-average type-token ratio (MATTR), a measure validated for its stability across text lengths and reliability in shorter texts. We also evaluate these measures by comparing them to traditional, so-called omnibus LD measures for interpreting L2 corpus data. The results show that examining LD within different parts of speech can help researchers disentangle the broader developmental trend observed with omnibus measures. These findings highlight how POS-specific LD provides more linguistically interpretable description of L2 lexical development, thus complementing traditional omnibus measures.