Meta-analyses using articles has nothing to do with language, as long as the research topics are similar. you could quantify the variations of the study (both in Chinese or other languages) by adding covariates in your model. Usually, you get better results if the studies you chose follow very similar study designs, then your accuracy or Q and I2 will be minimized. For your reference, take a look at the R package, https://cran.r-project.org/web/packages/meta/meta.pdf, they have examples that you could try.