repetition effects for speech-in-noise: summary

[Stuart Rosen <stuart@xxxxxxxxxxxxxxxxxxx>]

Earlier in the month, I asked the following question:

Can anyone refer me to studies of the effects of repeating items when
testing, for example, identification of words in noise or other
degradations? I myself have seen someone do much better with
noise-vocoded sentences because he had been in a study using the same
material more than a year previously! There is a large, somewhat related
literature concerning semantic and repetition priming, but that is not
quite the task I am interested in.

Thanks to all who sent the following responses:
-------------------------------------------------------------------------
Christophe Pallier
www.pallier.org
Steve Goldinger (JEPLMC 1996) studied repetition using a task of word
identification in noise (See also Church & Schacter (JEPLMC 1994).

Anecdotally, we observed that people once exposed to very fast speech
seem to retain an advantage days or months later (Mehler et al.,
Understanding compressed sentences: the role of rhythm and meaning.
Annals of the New York Academy of Sciences, 682:272-282, 1993). Not sure
if we reported it.
-------------------------------------------------------------------------
Bill Yund
yund@xxxxxxxxx

We were concerned in finding a study (Wilson RH, Bell TS, Koslowski JA. 
Learning effects associated with repeated word-recognition measures 
using sentence materials. J Rehabil Res Dev 2003;40(4):329-36.) 
demonstrating that the speech reception threshold (SRT) in quiet 
improved with repeated testing, but just as much with new sentences as 
with sentences repeated over five 1-hr sessions. If similar SRT 
improvement occurs with other sentence tests, and even in the presence 
of noise, that could make such tests useless for tracking speech 
perception changes over time. We are just finishing running the HINT 
sentence test with noise, in the Wilson-Bell-Koslowski (WBK) design. Our 
results are what you would expect, if you didn't know about the WBK 
results. We find improvement for repeated, but not for new sentences. We 
will submit a short report to J Rehabil Res Dev, with some explanations 
for the different results, which should be ready in a week or two. We 
would be happy to send you a copy of the manuscript as soon as it is fit 
for someone else to look at, if you like. I would also be interested in 
any other leads you get on your question; I have not been very 
successful in finding other similar studies.

-------------------------------------------------------------------------
trevoragus@xxxxxxxxx
Trevor Agus
(PhD Student, MRC Institute of Hearing Research Scottish Section)

There's a section about open-set and closed-set speech reception in
Green & Swets "Signal Detection Theory and Psychophysics" that may be
relevant to you. In section 11.3.1 there's a graph showing results of
a listener is listening to speech signals from closed sets of
different sizes.

In my own experiments, I've noticed that people seem to get familiar
with the 'style' (syntax/vocab/length) of the ASL sentence-lists -
their incorrect guesses can be quite wide-of-the-mark near the start,
but after a few sets, their incorrect guesses sound like parodies of
the original sentences.
-------------------------------------------------------------------------



/*------------------------------------------------------*/
Stuart Rosen, PhD
Professor of Speech and Hearing Science
Dept of Phonetics & Linguistics
University College London
4 Stephenson Way
London NW1 2HE
England

Directions to Wolfson House (where I am based):
http://www.phon.ucl.ac.uk/dept/maps.html

Tel:   (+ 44 [0]20) 7679 7404
Admin: (+ 44 [0]20) 7679 7401
Fax:   (+ 44 [0]20) 7679 5107

Email: stuart@xxxxxxxxxxxxxx

Home page: http://www.phon.ucl.ac.uk/home/stuart/home.htm
/*------------------------------------------------------*/