Jan 20
======

Gomez/Gerken:2000
-----------------

How well could infants be expected to do at differentiating rhyming
words in these artificial languages? And in languages in general?

Somewhat related to the paper - have their been studies that test
whether infant acquisition abilities are better when hearing a readl
in person voice vs a recording?

It is said that the amount of time the infant orients toward the
source of sound is taken as a dependent measure for discrimination of
familiarized words. Does the longer amount of time spent mean that it
is the familiarized or unfamiliar word? The familiarized word because
of the familiarity or the unfamiliar word due to confusion of the
infant? How reliable is this procedure?

In the "words in sequence" test, the chosen exposure time to the
grammar was very short (50-120 seconds)? Would a test over a longer
exposure time to the stimulus and with a longer break before testing
be more reflective of the infant's language acquisition experience,
since language acquisition is a long-term process?

In the artificial language studies on category-based abstraction, the
paper explains how it was almost impossible for learners to acquire
the distributional relations between M-N/ P-Q words, but performance
improved when there are cues to mark category membership. The
conclusion notes that the artificial language experiments discussed do
not have a semantic aspect. I would be interested to know if/how a
semantic aspect could be integrated into this kind of category-based
abstraction test with artificial language?

Speaking of systematic cues, that can help acquire the language like
schwa or inflectional endings - can cognates be included in this
group? For some inflected language, where over a hundred words can
have the same root, it may be helpful for kids to distinguish
different cognates.

The paper mentions experiments that have investigated the reaction of
infants to stimuli, for example: "The familiar passage was indeed more
reinforcing, even when read in another woman's voice, suggesting that
infants had learned certain features of their training passage in
utero."  How exactly is it measured which stimulus is more reinforcing
or which stimulus the infants are more interested in? How reliable are
these measurements and statements derived from them?

How do researchers decide on  the duration of the exposure to training
data (50 and 127 seconds seems really short)?

What is the significance of artificial language learning research to
detect infants' sensitivity to linguistic form in the absence of
semantic content?