With the rapid advancement of corpus linguistics and natural language processing technologies, the measurement and analysis of textual cohesion have emerged as a focal point in second language (L2) writing assessment. However, the relationship between writing quality and textual cohesion among L2 writers from different cultural backgrounds remains underexplored. This study builds a small-scale corpus based on the ELLIPSE dataset, using “distance learning” as a writing prompt, and employs the TAACO tool to quantitatively analyze 489 L2 writing samples. Results indicate that cultural background does not exert a significant influence on writing quality. Additionally, both global cohesion and text-level cohesion demonstrate a significant positive correlation with writing quality. Among the 168 cohesion features examined, a combination of four key indices: functional word MATTR, adjacent two-paragraph overlap function lemmas, lemma MATTR, and binary adjacent two-paragraph overlap pronoun lemmas, explains 17.2% of the variance in writing quality. These findings highlight the critical role of textual cohesion in L2 writing evaluation from a cross-cultural perspective, offering theoretical insights for English writing instruction and practical implications for the development of automated essay scoring systems.