[Corpora-List] a worldwide word association test

Fri Feb 17 11:17:27 UTC 2012

Sabine,

Many thanks for this. Maybe the reply of Storms to a previous remark (which was not on the list) is informative for those who are interested in this discussion: 

yes, indeed, we are aware of this so called chaining effect.  doug nelson investigated this in detail ... and in our data there are indeed traces of chaining.  But it is also clear that the second and third response provide a lot of semantic information.  Predictions based on three responses are qualitatively better (when predicting similarity or lexical decision etc) than measures derived from just the first response.  Take a word like 'blood'.  The vast majority of first associations is 'red'.  Only few other responses are given.  For such words, the second responses are much more interesting ('wound', 'pain', 'hurt, etc.) because they open up a lot of semantic information, and they are clearly (in this case) not mediated by the 'red' response. Oo the bottom line is: yes, there's chaining, but this can be filtered out if necessary and the advantages of multiple responses seem to outweigh the disadvantages.  

Best, marc

-----Original Message-----
From: Sabine Schulte im Walde [mailto:schulte at ims.uni-stuttgart.de] 
Sent: 17 February 2012 12:01
To: corpora at uib.no; marc.brysbaert at ugent.be
Subject: Re: [Corpora-List] a worldwide word association test

Hey,

this effect that the nth associate response is associated to the (n−1)th 
response rather than the stimulus is called 'association chain', cf. 
McEvoy and Nelson (1982).

I'm sure that Gerrit Storms and his collegues take this effect into 
account. For example, if one wants to be sure that there is no chaining 
effect in the collected data, you can just consider only the first 
association. If the effect doesn't matter for some research reasons, you 
can still use all responses, though.

Ref: McEvoy, C. L., & Nelson, D. L. (1982). Category name and instance 
norms for 106 categories of various sizes. American Journal of 
Psychology, 95, 581–634.

Best,
Sabine.

On 15.02.2012 13:22, Justin Washtell wrote:
> Hi,
>
> I found this incarnation of the word association experiment a little unnatural. **Perhaps it is best not to read on at this stage if you're planning on having a go at it, just in case I influence you!!**
>
> For each word the user is asked to provide three responses, one below the last, which arise "spontaneously" from the cue word. I think its fair to say that personally I never had three responses all come to mind simultaneously, such that I could hold them in my mind and then fill out the boxes. Rather, I'd have one arrive (which I might then fill in) and then I'd have to "allow" my mind to come up with another and so on.
>
> What I found however was that my latter responses tended to be influenced by my previous responses, and in some cases were only very tenuously associated with the original cue word, or not at all (e.g. awake ->  alive ->  kicking; idea ->  thought ->  provoke). I then found myself in the peculiar situation of consciously fighting this urge... and also of course wondering whether perhaps I shouldn't do such a thing. To be fair, the experimenters do provide the option of writing "no response", and I did this a few times when I was in doubt.
>
> I wonder then if the experimenters are accounting for this effect (I hope it is not just me!) Presumably the box to which a response belongs is being recorded, so if the second and third responses are indeed coloured by this sort of effect then it can be observed, and perhaps even factored out if it is not in keeping with the aims of the study. But what if users fill the boxes in some other arbitrary or even random order, as opposed to top-to-bottom like I did? More generally, I wonder whether the task of trying to be alert for three simultaneous responses may make this a more of a conscious task and colour the types of responses garnered.
>
> I'd be interested to hear others thoughts having attempted the task.
>
> On a related note, there is a similar experiment which has been running for some years now at www.wordassociation.org<http://www.wordassociation.org>. I did make a few attempts to tried to contact the creator, to try and obtain the (substantial) data for my PhD thesis, but had no luck. Does anybody know anything about this?
>
> Justin Washtell
> University of Leeds
>
>
>
> ________________________________
> From: corpora-bounces at uib.no [corpora-bounces at uib.no] On Behalf Of Marc Brysbaert [marc.brysbaert at ugent.be]
> Sent: 15 February 2012 10:07
> To: corpora at uib.no
> Subject: [Corpora-List] a worldwide word association test
>
> Dear all,
>
> Gert Storms and Simon De Deyne are running a worldwide word association test in English. Thus far they have over 1.25 million responses, but they require many more in order to have enough spontaneously produced associates to all known English words. The data will be made available to all researchers, just like the Florida norms, so that we can use them for our studies and include them in our computational models. Would it be possible to forward the call to your colleagues and students? The task itself only takes 5 minutes and involves giving associates to a few target words.
>
> Many thanks in advance, marc brysbaert
>
> From: Gerrit Storms [mailto:Gert.Storms at ppw.kuleuven.be]
> Sent: 15 February 2012 10:55
> To: Marc Brysbaert
> Subject: word associations
>
> Dear Marc,
>
> Can I ask you a little favor?
> Over the past few months, we have been trying to set up a scientific study that is important for many researchers interested in words, word meaning, semantics, and cognitive science in general.  It is a huge word association project, in which people are asked to participate in a small task that doesn't last longer than 5 minutes. Our goal is to build a global word association network that contains connections between about 40.000 words, the size of the lexicon of an average adult. Setting up such a network might teach us a lot about semantic memory, how it develops, and maybe also about how it can deteriorate (like in Alzheimer's disease). Most people enjoy doing the task, but we need thousands of participants to succeed. Up till today, we found about 40,000 participants willing to do the little task, but we need more responses. That is why we address you.  Would it be possible to forward this call for participation to graduate and undergraduate students who are fluent in English?
>
> The task can be found at
>
> http://www.smallworldofwords.com
>
> Of course the network will be freely available to all interested language researchers when it becomes substantial enough.
>
> We thank you in advance.
>
> If you want more information, don't hesitate to contact me.
>
> With kind regards,
>
> Prof. G. Storms and Dr. S. De Deyne
> Department of Psychology
> University of Leuven
> Tiensestraat 102
> 3000 Leuven
> Belgium
>
> _______________________________________________
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing list
> Corpora at uib.no
> http://mailman.uib.no/listinfo/corpora

-- 

PD Dr. Sabine Schulte im Walde

Institut für Maschinelle Sprachverarbeitung
Universität Stuttgart
Azenbergstraße 12
70174 Stuttgart, Germany

fon:   +49 (711) 685-84584
fax:   +49 (711) 685-81366
email: schulte at ims.uni-stuttgart.de

_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora