Searching for Double Object Ditransitives

[William Snyder]

Dear Liam,

Some years ago, Karin Stromswold and I did a fairly fine grained analysis
of double-object datives in the longitudinal corpora that were available in
CHILDES at that time.

[Snyder, William and Karin Stromswold. 1997. The structure and acquisition
of English dative constructions. Linguistic Inquiry 28(2): 281-317.]

The type of approach we used is still an option for you.

In current terms, you would use the CLAN program 'freq' (with the +u
switch) to get a list of all words used at least once by the child in a
given corpus. (It would also be possible to combine corpora and run 'freq'
on all the transcripts at once, to obtain a single list of words used at
least once by at least one of the children you are studying.)

The next step would be to hand-code that list to identify all words that
can either function as a double-object verb in adult English, or that have
a meaning that might tempt a child to use the double-object structure (in
error).

The third step would be to enter the words in a text file, and use the CLAN
program 'combo' to locate all child utterances that contain at least one of
the words. Preferably you'd use the -w2 switch to get two lines of context
for each match, so that you could easily identify and discard direct
imitations of other speakers.

The fourth step would be to hand-code the matching utterances to identify
the ones that are relevant to your project.

~~~

If you want to be sure to catch all the child's early uses of the
double-object dative, including the errors that may have occurred, this
strategy would (I think) be a reasonable way to go.

On the other hand, if you need speed more than a high level of accuracy,
using the automatic parses in some way (preferably in the way that Brian
recommended) could be a better choice for you.

~~~

With best wishes,

William

William Snyder
University of Connecticut


In (Snyder & Stromswold 1997)

On Wed, Dec 21, 2011 at 3:39 PM, Brian MacWhinney <macw at cmu.edu> wrote:

> Dear Liam,
>      The best way to do this would be to create a test file.  That file
> would include
> as much variation in the configuration of double object sentences as you
> can think of.
> You would start by collecting about 60 such sentences by hand and eye from
> various corpora.  Then perhaps you would imagine some other possible
> combinations.
> Then you would see if your search strings correctly located each
> occurrence.
>      If you can first do the work of composing a test file, we could go
> from there.
>      Regarding your %mor line attempt, I can easily think of many cases it
> would miss, such
> as sentences with two nouns as objects.  In theory the %gra line should be
> more definitive,
> but the level of accuracy of tagging of objects there is at about 90%, so
> the GRASP tagger
> is itself going to miss some things.
>    Generally, this is probably going to take repeated work and testing.
>
> -- Brian MacWhinney
>
> On Dec 21, 2011, at 1:58 PM, Liam Considine wrote:
>
> > Hey Chibolts Community,
> >
> > I am working on extracting double object ditransitive occurrences from
> > the CHILDES corpus.
> >
> > "John give me the cookie"
> >
> > I've tried a handful of different searches on the %mor and %gra line.
> > I would really like some other people who are familiar with CLAN
> > syntax to check out my searches. I have already made a search for the
> > prepositional dative so I am trying for this search to exclude those
> > instances.
> >
> > Here is my %mor line attempt:
> > combo +t*CHI +t%mor +sv*^(pro*)^(det*+qn*+pro*)^(n*+pro*) +k +r2 +u
> > *.cha
> >
> > My first %gra line form:
> > combo +t*CHI +t%gra +s"1|0|ROOT^2|1|OBJ^((3|4|DET^ 4|1|OBJ2)+3|1|
> > OBJ2)" +k +r2 +u *.cha
> >
> > My best effort %gra line:
> > combo +t*CHI +t%gra +s"(1|2|SUBJ^2|0|ROOT^3|2|OBJ^((4|2|OBJ2)+(4|
> > 5DET^5|2|OBJ2)))+(1|0|ROOT^2|1|OBJ^((3|4|DET^ 4|1|OBJ2)+3|1|OBJ2))" +k
> > +r2 +u *.cha
> >
> > I've selected the same data files from CHILDES as Anat Ninio does in
> > the book "Syntactic Development Its input and output." This seems to
> > be about 75% of all the files available.
> >
> > With my bigger %gra search i'm getting about 1075 hits. Is this
> > consistent with the frequency of occurrence others have seen? Does my
> > syntax have any glaring errors?
> >
> > Thanks for all the time and energy,
> > Liam Considine
> >
> > --
> > You received this message because you are subscribed to the Google
> Groups "chibolts" group.
> > To post to this group, send email to chibolts at googlegroups.com.
> > To unsubscribe from this group, send email to
> chibolts+unsubscribe at googlegroups.com.
> > For more options, visit this group at
> http://groups.google.com/group/chibolts?hl=en.
> >
> >
>
> --
> You received this message because you are subscribed to the Google Groups
> "chibolts" group.
> To post to this group, send email to chibolts at googlegroups.com.
> To unsubscribe from this group, send email to
> chibolts+unsubscribe at googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/chibolts?hl=en.
>
>

-- 
You received this message because you are subscribed to the Google Groups "chibolts" group.
To post to this group, send email to chibolts at googlegroups.com.
To unsubscribe from this group, send email to chibolts+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/chibolts?hl=en.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/chibolts/attachments/20111221/313766c0/attachment.htm>