Bio-inspired Grammatical Inference - Foundations on Natural and Artificial Computation - page 315

3 Linguistic Studies
The question of how children acquire their native language has been tradition-
ally addressed by linguists. Their approach is specially focused on making exper-
iments with children that are learning their native language. In that way, they
try to collect data about the process of natural language acquisition (e.g., first
sentences produced by children, errors made, etc.). Their final goal is to investi-
gate the mental processes that occur during children's language acquisition, and
try to describe this process.
There are different types of experiments. Depending on the way in which the
data is collected, we have a naturalistic approach (i.e., samples of child lan-
guage is collected or recorded in a comfortable environment. Data is collected
regularly) or a experimental approach (i.e., the researcher proposes a work hy-
pothesis and design specific tasks that have to be performed by the child to
use specific language structures. A statistical analysis of the data is done at the
end). Depending on the number of children used to do the experiments, we have
longitudinal studies (i.e., experiments are focused just on one child, and they are
done over a long period of time. Such approach is often combined with the natu-
ralistic approach) or transversal studies (experiments are made with a group of
children of different ages. Such approach is often combined with an experimental
approach). And finally, depending on the kind of tasks performed by the child,
we have experiments based on comprehension, production or imitation.
The CHILDES database (Child Language Data Exchange system) provides a
large amount of useful data for linguistic studies of children's language acquisition.
In this database we can find transcript and media data collected from conversa-
tions between children and adults. It has content in over 20 languages from 130
different corpora, all of which are available in http://childes.psy.cmu.edu/
Studies carried out in the field of Linguistics have helped to better understand
some aspects of natural language acquisition [14]. However, there is still a lot of
questions that do not have a clear answer; for example, what factors really have
an effect in the process of children's language acquisition. Therefore, despite
all investigations conducted so far, it has not been possible yet to understand
all the rules, strategies, and other processes that underlie children's language
acquisition.
Next we review some of the works and results obtained in this field, that are
relevant for GI studies.
3.1 Data Available to Children
The question of what kind of data is available to children during the learning
process is still a subject of discussion in Linguistics. It is generally accepted that
children receive sentences that are grammatically correct, that is, positive data .
However, the availability of another kind of data, usually called negative data ,
is a matter of controversy.
There are three different proposals to this question. The first proposal is that
children do not receive negative data and they must rely on innate information to