The effect of learned culture (e.g., birdsong dialects and human languages) on genetic divergence is unclear. Previous theoretical research suggests that because oblique learning allows phenotype transmission from individuals with no offspring to an unrelated individual in the next generation, the effect of sexual selection on the learned trait is masked. However, I propose that migration and spatially constrained learning can form statistical associations between cultural and genetic traits, which may allow selection on the cultural traits to indirectly affect the genetic traits. Here, I build a population genetic model that allows such statistical associations to form, and find that sexual selection and divergent selection on the cultural trait can indeed help maintain genetic divergence through such statistical associations, while selection against genetic hybrids does not affect cultural trait divergence. Furthermore, I find that even when the cultural trait changes over time due to drift and mutation, it can still help maintain genetic divergence. These results suggest the role of obliquely transmitted traits in evolution may be underrated, and the lack of one-to-one associations between cultural and genetic traits may not be sufficient to disprove the role of culture in genetic divergence.