[from Joel Judd]
Clark (920501)
Re: the Kuhl et al. paper. I don't know how "expert" will be this
assessment of the Kuhl paper, but I can tell you what I like and don't like
about it.
I like the authors' emphasis on experience and not "meaning" as crucial to
phonetic learning, although at the end they say "Infants demonstrate a
capacity to learn simply by being exposed to language during the first half
year of life..." which smacks of "blank slate" philosophy. The importance
of a learned prototype in explaining older learners' difficulties also
seems to make sense, and could be carried further to help explain why
simply passive listening (the Silent Way techniques such as the Marvin
Brown anecdote mentioned awhile back) is also not sufficient in developing
a L2 phonology.
What I don't like (perhaps predictably) is their reliance on a
cross-sectional design for something that, given a mere six months of time
and some $$$$ (which they obviously didn't lack--see bottom middle of page
607 where they make it clear that all experimental equipment and three
people were transported from Washington state to Sweden), they could have
looked at a number of infants longitudinally (a test they admit needs to be
done in footnote 12). Because of this, they must rely on statistical
significance instead of saying something like "every English child went
from no reaction to either prototype to reacting to /i/ and its prototypes,
and every Swedish...etc." or something like that.
If you look at the percentages for each ring, instead of the averages they
printed, the differences don't seem quite so impressive. For example, the
English infants equated what looks like 69-70% of first ring variants with
the /i/ prototype, the Swedish infants equated what looks like 63-64% of
the variants. The differences become greater as the variants become farther
from the prototype, but this shouldn't be surprising since they apparently
overlap phonetically.
In case anyone reads the summary of the article on page 535 by M. Barinaga,
I think she misses the boat by trying to read into the study something the
authors are providing evidence against--needing meaning to develop
phonological systems--and in addition asking a blatantly encodingism
question at the beginning: "How do infants manage to sort through the
jumble of spoken sounds bombarding them and tease out the ones that encode
meaning?"
The answer, though I'm not the one to elaborate on it, is: they don't!!
I think Clark is probably accurate in saying that children develop
[SENSATION/CONFIGURATION] reference signals for those sounds which are
heard in the environment, although there must an interactive component to
this development, as Molly Mack, a language professor here, has shown that
bilingual children develop "compromise" phonologies to use with similar
sounds in their two languages--acoustically neither L1 nor L2, but
"native-like" in both languages.
There are several interesting issues here but I've said enough.