Object-iniital languages

[Brian MacWhinney]

Dear Östen,

  I have never seen a sentence like "Ich krank bin" in any of the German
child corpora in CHILDES.  I agree with you that it is difficult in general
to find German child language sentences that cannot be matched to some adult
word order, if one allows for omissions, which are typically possible even
for adults, although at a lower frequency.  What we see mostly in German
child language are lots of SV units, and a distinctly smaller number of VS,
OV and VO units.  Full combinations into SVO or SOV are absent at first and
rare for quite awhile.
  However, arguments based on the child's productions, although empirically
well grounded, may be making a fundamental error.  After all, the child is
spending a lot of time listening to sentences before speaking and it is
likely that she/he picks up a fuller word order in comprehension before it
is demonstrated in production.  This pushes the issue back a few more months
and makes it harder to verify exactly what the child is doing.
  Is the child picking up a set of templates, elaborating item-based
patterns, or selecting values on parameters?  The fact that there are so few
cases of what one could honestly call a word order error in early production
suggests that this earlier comprehension-based tuning works out pretty well.
Thus it is likely that the distinction between SVO and SOV in German is
already in place by the time of the first utterances.  But does this mean
that the child has discovered the "basic word order".  I don't see why one
would argue that.  Rather, simply that the child is doing a good job of
controlling the two major options and their secondary realizations first in
comprehension and then in production.

--Brian