Reconstructed syntax?

Morphology, as we have seen, poses problems for historical reconstruction that make straightforward storytelling difficult. There are no ‘regular morph-changes’ quite like ‘regular sound-changes’ (except of course insofar as changes of morph-shape reflect regular changes in constituent phonemes). In addition, analogy and semantics are constantly interfering with what might otherwise be reasonably ‘regular’ developments. Syntax poses these problems as well, in addition to special ones of its own.

The ‘vocabulary’ (set of elements) is much more complex than those for phonology or morphology, and the idea of ‘ancestry’ is conceptually difficult. Since syntax is the ‘creative’ part of grammar, we do not normally think of a given sentence in a language as the ‘descendant of’ an equivalent sentence in an earlier stage, itself the descendant of some yet older sentence. It makes sense to say that OE cyning is the descendant of WGmc */kuninγ/, which descends in turn from PGmc */kuninγ-α-z/. But it appears not to make the same kind of sense to say that the OE sentence se cyning þās word gehīerde ‘the king heard those words’ is the descendant of the Proto-Gmc sentence */se kuninγαz θαis worðu γαxauziðα/.

It shouldn't be hard to see why, in a general way. Given sentences' high specificity of meaning, their embedding in discourses, texts, etc. – as well as the fact that the ‘set of sentences’ in a language is at least indefinitely large – we would not expect this kind of descent.