reliability, cueing, and neurotic parenting

Tue Mar 4 16:45:01 UTC 2008

Dear All, but especially Peter and Isabelle,
I shouldn't try to speak for Infoture, as I am just an advisor, so let
me just respond to a couple of points that I happen to have good
information on and that might be helpful. I know that if you have
concerns about any aspect of  the company's work, they welcome direct
contacts. Also if you are interested in the database, you can contact
them to see if something can be worked out to make aspects of it
available to you for research -- I am doing that myself. I have been
told that a formal option for licensing is anticipated in the future,
but in the meantime, why not make suggestions about whatever might be
useful to you?

I will speak boldly on one point, which Isabelle raised. I have been
present at enough Board meetings to be certain that no one at Infoture
is seeking to have LENA replace SLP's or psychologists or
pediatricians. On the contrary, the goal is to provide new tools to
supplement and assist professionals in their work -- for example to
provide a basis to obtain naturalistic data on conversation at home
(at least in terms of amount of adult and child talk, and
conversational turns). I am not sure how the impression could have
been given that Infoture might have a goal to replace professionals,
but I would hope that any one who has that impression would voice the
opinion to the company and explain what had been done to make it seem
so -- I'm sure they would want to clarify.

On whether there has been validation research on the LENA as it
currently stands, the answer is yes, and lots of it. Very large scale.
Much of the data are posted on the Infoture website at Lenababy.com
(choose "Researchers Click Here", then "Research" and you'll find a
list of tech reports). Of course there's lots more that can be done,
and there are indeed massive additional efforts going on to improve
the identification algorithms and to validate their performance.

A last point is about CHAT. The research version of LENA (which I am
beta testing at the moment) provides a conversion of processed audio
files that is CHAT/CLAN compatible. I am now categorizing utterances
produced by my daughter (and people talking with her) that were
located by LENA software and then converted to a .cha file. I use the
Sonic mode in CLAN to access the whole waveform (up to 16 hours). This
is a nice aid to locating large numbers of utterances -- the LENA
software indicates for you the periods of time where high vocal
activity occurs, and if you want to transcribe selectively (which I
need to, since I can't categorize every hour of evey day of recording
that I am looking at), you can focus on periods of high activity or
low activity. I'm hoping to report on this work for the next ASHA
convention.

Best wishes,
Kim

On Mar 1, 6:48 am, "Gordon, Peter" <pgor... at exchange.tc.columbia.edu>
wrote:
> I was intrigued by Kim's description of the LENA device and agree that we should not dismiss it out of hand (perhaps just be wary of how it's being promoted).  It sounds like a language equivalent of a pedometer that estimates how many steps people take in a day from their bouncing around.  One of the problems is that some of them keep counting when you're in a car that bounces around too.  I'm wondering if there was any validation research to look at how accurate the counting is under noisy conditions.
>
> On a second note about the "Clever Hans" phenomenon, there is a difference here in that the horse was simply cued to stop stomping from the subtle movements of the trainer.  In the case of the infant videos you have qualitatively differentiated responses that don't seem to be cued to anything the adult is doing with his body.  If the word cards on the video are doing the cuing for the infant to touch it's body parts, then that's not really a case of a Clever Hans effect, because it's the relevant stimulus that's doing the cuing.  But the question again is whether we can say that the shapes of the words are actually acting as symbols in representing body parts for the infants in a non-iconic mode, even if they are not representing the phonetic values of the individual letters (which is why it presumably is not real reading per se).  Perhaps if young infants can learn things like baby signs, then maybe this should not be that surprising, although we generally think of the baby signs as being a bit more iconic in nature.  It might be interesting to see if infants can learn arbitrary baby signs as easily as iconic ones.  
>
> Finally, even though there definitely is an unhealthy culture of competitive parenting, this doesn't mean that everyone who tries to extend the knowledge of their very young children is necessarily doing a bad thing.  We took great pleasure in marveling at what our daughter was able to pick up about language and number at a very early age "before she was supposed to".  I don't think we should be in the position of being educational Luddites and prescribe what the "natural" age is for any particular ability.  Consider the case of the Emperor Charlemagne, who considered it impossible that peasants could ever learn to read.  I think that it's fine if parents want to see what their kids can do; it just shouldn't turn in to a requirement.  But nowadays, when preschoolers have to be interviewed and tested for entry into kindergarten, it's pretty hard to stem that flow.
>
> Peter
>
> Peter Gordon, Associate Professor
> 525 W 120th St. Box 180
> Biobehavioral Sciences Department
> Teachers College, Columbia University
> New York, NY 10027
> Office Phone: (212) 678-8162
> FAX: (212) 678-8233
> Web Page:www.tc.edu/faculty/index.htm?facid=pg328
>
> ________________________________
>
> From: info-childes at googlegroups.com on behalf of Kim Oller
> Sent: Fri 2/29/2008 1:47 PM
> To: Info-CHILDES
> Subject: LENA
>
> I am a member of the Scientific Advisory Board of Infoture, Inc.,
> which is the producer of LENA. Perhaps rather than responding to
> specific interests and concerns that were expressed in the recent
> postings, I'll just offer to talk with anyone who wants to know about
> the company and the tools it has developed. Also I would encourage any
> one who would like to talk with people at the company directly to do
> so. These are very friendly and open people. Perhaps the most
> appropriate contact would be Jill Gilkerson at 303 441 9014 or
> JillGilker... at infoture.org.
>
> I wouldn't be involved if I didn't view the developments at Infoture
> as extremely positive and indeed fundamentally important to our
> futures, both scientific and clinical.  Infoture invested heavily to
> develop a battery-powered recording device that weighs about an ounce
> and yields good quality (16kHz) data for 16 consecutive hours. Further
> they have developed extremely intriguing software that processes the
> data to yield a variety of automatic measures -- in particular, pretty
> reliable counts of adult words spoken in the sample, child
> vocalizations, and conversational turns. Much more is on the way, and
> the software itself is rapidly continuing to be improved and
> enhanced.
>
> These developments are going to be extremely useful for those of us
> who are interested in large scale naturalistic sampling that can be
> done all day long in the home. I am recording now with LENA and intend
> to do longterm, longitudinal research using it, including research
> employing neural network approaches that need really large quantities
> of data. Infoture is not just developing devices -- they already have
> over 40,000 hours of recording, much of it on a carefully stratified
> longitudinal sample. This sample is going to be a tremendous resource
> for research, and Infoture has done very significant research with it
> already. Collaborators from a variety of universities are already
> seeking to use the database for specialized projects, and Infoture is
> making adaptations to encourage that kind of collaboration and
> utilization of the data.
>
> In part my enthusiasm about this is due to my belief that laboratory-
> based research needs to be supplemented with large-sample naturalistic
> research. I don't think it is likely that we will be able to process
> really large samples without automated preprocessing. So the Infoture
> efforts are laying infrastructure for critical research,
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Info-CHILDES" group.
To post to this group, send email to info-childes at googlegroups.com
To unsubscribe from this group, send email to info-childes-unsubscribe at googlegroups.com
For more options, visit this group at http://groups.google.com/group/info-childes?hl=en
-~----------~----~----~----~------~----~------~--~---