Object-initial languages

Brian MacWhinney macw at CMU.EDU
Fri Oct 25 17:49:31 UTC 2002


On 10/25/02 12:05 PM, "Ellen F. Prince" <ellen at CENTRAL.CIS.UPENN.EDU> wrote:

> Furthermore, while subordinate clauses are generally acquired later
> than main clauses, it appears that German-acquiring children use SOV
> order until they acquire tense. (I don't have the reference handy but
> I bet someone here does.) That seems to me a compelling bit of
> evidence that SOV is not just 'subordinate' word order in German but
> something more basic.
>

The claim that German children begin with SOV order, which was proposed at
about the same time in the late 1970s by Clahsen and by Park, is based
largely on sentences with either SV or OV order and seldom both.  There are
virtually no sentences in the earliest corpora with full SOV order.  Meisel
and Pienemann linked these SOV analyses to parallel analyses for L2 German,
but with similar interpretive problems.  Peter Jordens reviewed these claims
critically about 1990 pointing to the various flaws in the analyses.
Despite the clarity of Jordens arguments, workers in German child language
continue to accept the notion that children begin with SOV.  I can only
assume that they do so because of other theoretical commitments and not
because of the superficial pattern of the data.

My own interpretation of this literature and the relevant data is that
German children start with a competition between fragmentary SV, OV, and VO
item-based constructions.  (VS is rare, and VO seems to arise from
imperatives.)  They then use these during the third year to develop SVO
templates when the main verb is tensed and SOV templates when it is not.
Relative clauses, of course, come in much later and it would seem strange to
me to argue that they would be the basis of learning of these central main
clause patterns.

--Brian MacWhinney

P.S.  I'm still hoping that someone will take a plunge and try to answer
Ron's question.



More information about the Funknet mailing list