fill in the missing word

[Brian MacWhinney]

Dear Info-CHILDES,

I would like to thank Jean Berko Gleason, Annette Karmiloff-Smith, Michael
Thomas, and Virginia Marchman for extremely helpful replies to my questions
and observations regarding the form of the nonce-word "wugs" task.

As Jean and Virginia note, none of the versions of the task should be viewed
as "fill in the blank".  Instead better names would be:
1. finish my utterance (this is the original version from Berko 1958)
2. finish the word I started (this is the version introduced by Tyler and
Marslen-Wilson and used by Moore et al.)
3. answer the question (this is the form used by MacWhinney, 1975, 1978).
These forms of the task then further vary by the nature of the visual or
contextual support given to the child. This support can involve pictures,
objects, or enactments. As I suggested, using objects that the child can
hold and actual enactments can allow the task to measure productivity with
nonce words down to 1;8, whereas the other methods can be used with children
from 3;6, as Virginia notes.

There is another wrinkle to my earlier implementation of the task that I
failed to mention.  This is primarily important for children under about
2;6, but could also be important for children with SLI or WS.  This is the
fact that, if one uses real objects and demonstrates their action a couple
of times, then it is easier to get the child to learn the name of the
action, at least in fast-mapping terms.  If the child has fast-mapped the
noun or verb, then the formation of the plural or past tense is more likely
to go through the normal productive system (whatever that might be).
However, if fast-mapping has not been fully achieved and the child is trying
to dredge up the word under load, then zero-marking is likely to result.

In regard to picture supports, I'm not quite sure how they work in the cases
of nonce verbs.  It would seem easy enough to draw good pictures for things
like "running" and "swimming", but how do you draw "miffing" and "tiving" or
the various novel actions that we see in recent studies by Brooks, Tomasello
and others.  I don't remember seeing displays of the pictures that were used
in any of these studies with "finish my utterance" tasks for nonce verbs.
However, verbs have become increasingly important in recent argumentation
about the specificity of grammatical impairment in language disorders.

I should add that my ability to get nonce forms from age 1;8 using the
"answer the question" procedure was for nouns in Hungarian.  Also it worked
at about 2;3 for nouns in German.  For verbs in Hungarian, the "answer the
question" procedure didn't work until about 2;3 in Hungarian.  So, verbs are
harder than nouns, as we know.

When reading the interesting article from Michael, Annette, and colleagues,
I did of course notice the inclusion of a second task and the marked
difference that the second task had on decreasing errors of omission from
26% to 5% in the WS group, as well as the youngest control group.  In fact,
it was this comparison that triggered me to raise the issue on info-childes
in the first place.

It was crucial in these studies to continue to use the "finish my utterance"
form of the task, since this has been used by Clahsen and Almazan (1998) and
van der Lely and Ullman (2001).  By using the "finish the word I started"
technique of Tyler and Marslen-Wilson, Thomas et al. were indeed able to
reduce memory load and get a fuller picture of morphological productivity.

Michael's further comment was that "currently we are not clear whether the
higher level of zero marking errors in the 'fill in the missing word' task
was due to that task's great memory load or due to a context that
constrained a past tense response less strongly than in the 'finish the word
I started' procedure.  I think that is exactly the right question to be
asking.

Thanks for a useful discussion of these issues and happy wugging.

--Brian MacWhinney