summary: word learning tasks

[Michael Ullman]

Hi,

Thanks to everyone for their responses to my recent query
on word learning tasks.  The initial query and all responses
are included just below.

Best,

Michael Ullman




Initial query:


>We are looking for a task or tasks that probe word learning. Ideally we would
>be able to use the task (or variants of it) in both cognitively impaired and
>intact kids and adults.
>
>We are *not* looking for episodic memory types of tasks such as
>the AVLT or CVLT, in which the subjects have to remember a list of real words.
>Rather we want to test learning of new words, ideally in a
>(relatively) naturalistic context.
>Note that fast mapping tasks seem to be good in principle, though
>in practice one would likely get ceiling effects for adults.
>
>Any ideas?

Responses:

(1)
Date: Mon, 28 Nov 2005 12:30:49 -0600
From: "McGregor, Karla K" <karla-mcgregor at uiowa.edu>

Hi,

If you have a large enough set of novel words, you can keep the adults
away from ceiling. You might try a "quick incidental learning paradigm,"
the variant on fast mapping that Mabel Rice has used in the past in
which multiple new targets are embedded in a story script.

You might also try multiple dependent variables, the children with
cognitive impairments might demonstrate learning in recognition tasks
only or in production tasks when given multiple retrieval cues; the
children with normal development and the adults may learn well enough
for production without the need of scaffolding.

I'll be interested to see the other suggestions,

Karla K. McGregor, Ph.D.
Associate Professor
Speech Pathology and Audiology
University of Iowa
121c WJSHC
Iowa City, IA 52242
phone 319-335-8724
fax 319-335-8851

(2)
Date: Mon, 28 Nov 2005 12:47:30 -0600
From: "Aleka A. Blackwell" <ablackwe at mtsu.edu>

Dear Michael,

I am teaching a course this spring called "The Science of Words" and I plan to
explore the question you pose below.  I have reviewed some of the literature
on lexical acquisition, and it seems Anglin's work seems most diverse in its
methodology for testing word meaning knowledge.  I understand from your
message that you are interested in word learning, instead, especially since
you refer to fast mapping.  I am also in the process of thinking about ways to
probe adjective learning (property concept word mapping vs. object concept
word mapping), but I am not satisfied with the design I am using.

Would you be so kind to forward the ideas you receive as a result of your
posting.  I would appreciate it.

Best,
Aleka Blackwell

(3)
Date: Mon, 28 Nov 2005 11:41:46 -0800
From: Gedeon Deák <deak at cogsci.ucsd.edu>
Fast mapping is overrated. You tend to get 
shallow, transitory representations of new lexemes.
for example--
Proceedings of the Annual Conference of the 
Cognitive Science Society, 25, 318-323
“Slow Mapping” in Children’s Learning of Semantic Relations
Gedeon O. Deák (deak at cogsci.ucsd.edu)
Department of Cognitive Science, 9500 Gilman Dr.
La Jolla, CA 92093-0515 USA
Jennifer Hughes Wagner
Department of Psychology and Human Development, 
Box 512 GPC, Vanderbilt University
Nashville, TN 372003 USA

Problem is that as I'm sure you know, "word 
learning" is not one thing. If you're looking for 
learning novel phonological patterns, it's one 
thing; exact subcategorization frame is another, 
nuances of meaning is another; pragmatically appropriate usage yet another.

We have a long-term project examining multiple 
measures of learning (comprehension, production, 
& generalization) of object words, over the first 
5 ostensive exposures to a word (or other kinds 
of information), which shows SOME dissociation of 
these measures, though they're all significantly correlated.

As far as "naturalistic," it's a very difficult 
order, and what counts will be quite different 
for older kids than for adults; populations with 
different disabilities that are associated with 
different cultural learning environments further 
complicates the picture (e.g., for a child who 
has a dedicated tutor during much of the school 
day, one-on-one ostensive learning from an 
unrelated adult might actually be more "natural" 
than for a typical 10-year-old, or adult.

In short, no standardized test such as you're 
looking for exists, to my knowledge. Most of the 
experimental tests people have used don't do a 
very good job w/ regard to the constraints you've 
mentioned. That are a few older studies (e.g. 
Nelson & Bonvillian) that did a better job, but 
those tested infants/toddlers. Studies of vocab. 
learning in L2 students (often adults) are 
closer, but the ones I've seen have limited 
ecological validity, shallow testing of what's 
been learned, or both--in short, not valid, 
sensitive tests of individual differences.

Best of luck!

(4)
Date: Mon, 28 Nov 2005 15:17:14 -0500
From: Barbara Pearson <bpearson at comdis.umass.edu>

Dear Michael,

You might look at the DELV-NR (Seymour, Roeper, &
de Villiers, 2005).  There is a long section on Fast-mapping
in the Semantics domain (and we have even more items from
the pilot version, the Dialect Sensitive Language Test).
And Valerie Johnson has even more in her dissertation
(UMass, 2001, "Fast mapping verb meaning from  argument structure.")
There are fairly complex "complement" sentences
as well as easier intransitives.

When we tried it with children 4 to 12, it is one of the few
areas we didn't get a ceiling effect, although I don't know
how it would be for "intact adults."
There's a scaled score for the children 4 to 9, but of course
you can use it with older people.

Let us all hear what you come up with.

Cheers,
Barbara Pearson

Barbara Zurer Pearson, Ph. D.
Project Manager, Research Assistant
Dept. of Communication Disorders
University of Massachusetts
Amherst MA 01003
413.545.5023
fax: 545.0803
bpearson at comdis.umass.edu
<http://www.umass.edu/aae/>http://www.umass.edu/aae/

(5)
Date: Mon, 28 Nov 2005 15:51:20 -0600
From: Margaret Fleck <mfleck at cs.uiuc.edu>

Does it matter what sorts of (unfamiliar) objects you use?   E.g.
does the (relative) performance of adults and children differ
if you make them learn (say)

    -- unfamiliar animals
    -- unfamiliar abstract objects
    -- unfamiliar kitchen tools
    -- new Pokemon characters

Can choice of domain be exploited to help keep the adults off-balance?

Margaret Fleck, U. Illinois, Computer Science

(6)
Date: Thu, 01 Dec 2005 10:52:48 +0000 (GMT)
From: p.monaghan at psych.york.ac.uk

Hello,

Gareth Gaskell and colleagues have looked at testing word learning by
investigating cohort effects resulting from when the novel word is entered
into the lexicon. This only works well if your novel words have late
uniqueness points (like cathedruke interfering with cathedral):

Gaskell, M. G., & Dumay, N. (2003). Lexical competition and the
acquisition of novel words. Cognition, 89, 105-132.

Padraic Monaghan

(7)
Date: Thu, 01 Dec 2005 14:30:02 -0500 (EST)
From: Marnie Arkenberg <mearken at andrew.cmu.edu>

Hi,
   I'm not sure if this will be helpful or not-- 
We taught children (normal-language 4-year-olds) 
450 novel words (150 dogs, birds, and horses)over 
3 months during free play sessions (2-3 per week 
for 45 minutes).  To test their learning we 
assessed the number of items children could 
remember at the next session.  This helped us get 
some learning rate information but we did not 
assess what children remembered after the study 
was over so we do not know what overall retention 
was. Children learned between 35% and 95% of the 
words when assessed during the following session, 
but I doubt that they retained that much.
   I'm currently putting together an experiment 
that will *hopefully* tap into what words 
children remember from story books.  Please let 
me know if any of this sounds like what you are 
looking for and might be helpful.
Cheers,
Marnie


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/info-childes/attachments/20051205/638aa794/attachment.htm>