No Proto-Celtic: word order

[Thomas McFadden]

> 2) Explanations based purely on pragmatic criteria risk being logically
> circular (first position is emphatic: therefore what is in first position
> must be emphatic; similarly for theme-first). Pragmatic analyses might be
> more informative if they attempted to integrate surrounding clauses in their
> explanations: for example, a following relative clause modifying the main
> clause subject will encourage VS, while a complement may encourage VO.

of course this is absolutely correct.  good recent work in pragmatics is
actually primarily concerned with how a sentence fits into the
discourse.  so for example using newness/oldness of a particular NP to
help explain its position in the clause implies that you read the
surrounding sentences to determine whether the referent of the NP has
already been mentioned in the discourse.  work of this type does not start
with assumptions like `first position is emphatic', but rather tries to
figure out, based on corpus data, what characteristics first position
has in a particular language under particular circumstances.  circularity
is not a danger if the work is done with a bit of care.

> In sum, word order appears to reflect 'weight to the right' (cf. Behaghel
> 1909 and many subsequent commentators).

absolutely.  another argument for the position that `free-word' order is
flexible, but principled.