VocD

VocD, short for Vocabulary Diversity, is a probabilistic measure used in corpus linguistics to quantify lexical diversity in text or speech. It is designed as an alternative to the type-token ratio, addressing sensitivity to sample length and topic. VocD aims to reflect the richness of a speaker’s vocabulary given observed word usage.

The VocD framework analyzes how the number of distinct words grows as more tokens are sampled. It employs a statistical model to estimate the underlying diversity of the lexicon and to predict the rate at which new, unseen words would appear with additional sampling. The resulting VocD score is intended to provide a more robust indication of lexical richness across varying sample sizes.

Applications of VocD include comparing lexical diversity across individuals, genres, languages, or developmental stages. It is used in studies of child language acquisition, second-language vocabulary growth, and cross-cultural linguistic variation. VocD supports longitudinal analyses and cross-sectional comparisons by accounting for differences in token counts and sampling design.

Implementation and interpretation rely on careful preprocessing, including consistent tokenization and sampling strategies. VocD scores are primarily relative measures; comparisons should consider corpus composition, domain, and register. In reporting, VocD is commonly presented with uncertainty estimates such as confidence intervals or bootstrap-based results to reflect sampling variability.

See also: lexical diversity, type-token ratio, vocabulary size, probabilistic models in linguistics.