Аннотация
Corpus linguistics has moved from a specialised empirical method to a data infrastructure that underpins language technology, quality education and responsible innovation, yet the structure of its rapidly expanding literature remains poorly mapped. This study examines the productivity, collaboration patterns, citation impact and thematic structure of Scopus-indexed publications on corpus linguistics and tests whether growth, international partnerships and thematic composition follow systematic patterns. The study employs bibliographic records of 14,323 documents retrieved from Scopus for 2020–2025, processed with the bibliometrix package and Biblioshiny in R and complemented by descriptive statistics, correlation analysis, log-linear growth regression, heteroskedasticity-robust and bootstrapped regression models, non-parametric tests and Zipf rank–frequency estimation. The results show that output rose from 2,285 documents in 2020 to 2,664 in 2024, a compound annual growth rate of 3.9% (log-linear trend 4.2% per year, p = 0.004, R² = 0.95); conference papers account for 52.5% of documents; China leads with 1,830 corresponding-author publications but only 16.3% multi-country publications; and the share of multi-country publications falls with national output (Spearman ρ = −0.595, p = 0.006). Citation impact is highly concentrated (Gini = 0.672), and computational and artificial-intelligence terms account for 70.9% of the occurrences of the fifty most frequent keywords. The findings suggest that corpus linguistics is growing steadily, that collaboration intensity is inversely related to scale, that impact is dominated by a few language-model studies, and that the field’s intellectual core has shifted toward computational linguistics, with consequences for quality education, innovation and partnerships for the goals.