In this chapter we want to highlight the importance of taking the factors source language and register into account when trying to make sense of linguistic differences between translated and non-translated texts. More specifically, we investigate how the aforementioned factors affect so-called normalization behavior of both translators and writers. This will be achieved by verifying how translated and non-translated texts in the Belgian Dutch context deal with (accepted) English loanwords when there is a synonymous, more endogenous alternative available. Furthermore, we draw attention to the added value of applying multivariate statistics in corpus-based translation studies together with more qualitative analyses. Therefore, three complementary analyses were carried out, viz. a correspondence analysis, a qualitative analysis of the source text lexemes, and a logistic regression analysis, which not only allows us to determine how the various factors under investigation behave, but also if and how they affect one another. Our results show that all factors under investigation do indeed have an influence on whether a loanword or an endogenous alternative is used and should therefore not be ignored in future inquiries.