Whereas the previous chapter examined structural differences within one language from the diachronic viewpoint, this chapter presents a synchronic analysis of structural variation across languages. However, the basic approach remains the same as there is no principled difference in the comparison of two synchronic stages of the same language and that of different languages at the same moment in time. The model outlined in the introductory chapter makes clear and testable predictions regarding the range of cross-linguistic variation. Generally speaking, languages are expected to vary to a greater extent in the structural than in the content domain. By virtue of the fact that a great many degrees of freedom are involved in how the elementary building blocks may be put together, structural units may be predicted to exhibit a certain variability across languages. By contrast, no decisions as to the type of content unit (and their strength) have to be taken in preparing to speak. The stock of content units is on speakers’ shelves, as it were, and allows them to retrieve items as appropriate. This retrieval process admits of little variation even across languages as everything is “prefabricated.” We would therefore expect content units to be relatively invariant across languages.