The study of syntax has not usually received much attention from historical linguists. Many historical grammars either give very little consideration to syntax or do not include a section on it at all. Historically oriented works that do treat syntax, including important early volumes such as Delbrück 1888 and 1893–1900 and Wackernagel 1920, are largely descriptive and confine themselves to the interaction of morphology and syntax. Another approach has been to employ typology to attempt to explain syntactic change, notable examples including Lehmann 1974, Friedrich 1975, and Miller 1975, the first two of which are fairly extreme in their attempts to motivate syntactic change in the interest of typological consistency (see the telling criticisms of Watkins 1976).

The reason for the lack of serious attention to syntax in historical linguistics, no doubt, is that it is hard to know what to compare between two synchronic states of a language from different time periods (so, recently, Longobardi 2003: 127). One compares phonemes in phonology, morphemes in morphology, but what in syntax? In the history of many Indo-European languages, a shift from Object-Verb (OV) to Verb-Object (VO) or Genitive-Noun to Noun-Genitive order is manifest in the documentary record, but, as noted by Longobardi 2003: 109, such changes have multiple causes and cannot be reduced to a single shift in linguistic structure. The difficulty lies in the fact that syntactic structures and operations are not ‘visible’ in the way that phonemes and morphemes are. As a child acquires a language, as noted by Hale 1998: 9 and Embick 2008: 60, among others, she does not have direct access to the syntactic structures of the speakers of the previous generation as she does for phonemes and morphemes, only the syntactic output that they produce. It is for this reason, then, and because of gaps and other limitations in the historical documentary record, as well as variations due to sociolinguistic factors (which may not be obvious to contemporary researchers), that, as Kroch 2001: 700 puts it, “conclusive results have been hard to come by.”