|
The central focus of Theoretical Computational Linguistics (TCL) is, on
first approach, human language from the point of view of its computability.
While the computer is the incarnation par excellence of an efficient
computation model, the concerns of this subdiscipline of Linguistics and
the questions it is most interested in addressing go beyond the bounds of a
"Linguistics for computers".
A computation process is, in this context, understood to denote a procedure
whereby, starting with a finite quantity of discrete objects, it is
possible by applying a predetermined set of rules to arrive at a likewise
finite, discrete object. Comprehending a sentence, for example, evidently
involves a computation process in this sense. Based on the findings and
methods of Logic TCL examines the characteristics (features) of such
computation procedures in as far as they can be used to provide an insight
into the structure of natural languages. Thus, for example in the fields of
Logic and Theoretical Computer Science a refined hierarchy of computation
processes has been devised which classifies these processes according to
their increasing capacity. Against this background it becomes a question of
high theoretical and practical interest at what level exactly in this
hierarchy of computable processes the natural languages "per se" rank, as
well as languages modelled on them with the aid of various linguistic
theories. Should it turn out that we can only design our most important
medium of communication by having recourse to abstract computing systems
which lie beyond the practical computing capacity not only of presently
available computers but also of those expected to become available in
future, this circumstance would profoundly call into question the aims of
linguistic research.
The results of studies on the complexity of natural language, besides their
practical relevance as defined by means of theoretical reflections on the
computability of such language, are also closely connected with the basic
concepts of the essence of language established over the past four decades
under the influence of the definitions generated by cognitive science. The
object of cognitive science is the study of the mechanisms of perception
and knowledge guided behaviour in humans, animals and machines. Cognitive
science integrates results and methods of cognitive psychology, artificial
intelligence, linguistics, philosophy, neurosciences and cognitive
anthropology into its research programme. The underlying assumption is that
any answers to problems that fall within the bounds of this discipline must
obey the restriction that cognition be understood in the final analysis as
a computation of (mental) representations. Human language, according to the
precepts of cognitive science, belongs to the mental capabilities
understood to function as a computation of such finite, discrete mental
representations. Thus, if the methods thought to operate in morphology,
syntax, semantics and pragmatics are so extremely complex that they not
only exceed the practical computating capacity of putative computers but
moreover even fail to fit into the framework of the abstract definition of
a general computation process described above, then this would place
hurdles in the way of computer linguistics and its proposed usage and,
furthermore, one of the competing paradigms guiding the way homo sapiens
sees himself would be rocked to the foundations. These involvements make it
clear that, while theoretical computational linguistics, in spite of the
implications of its name, does not aim most notably to reduce human
language to the level of a computer programme, its study area - the
analysis of language from the point of view of computation theory - has at
the same time direct significance for the practical concerns of
computational linguistics and for our own image of ourselves as speaking
beings.
|