\title{Connectionism}
\author{Patrick Juola}

\begin{document}
\maketitle


\begin{abstract}
Connectionist networks, also called ``neural'' networks, are the
outgrowth of an attempt to produce human-like behavior by modelling
the architecture of the human brain.  This is a radical departure from
traditional rule-and-object formalisms of the sort that most linguists
are familiar with, but may produce some better results in capturing
regularities (and processes) in the human language system.

This chapter summarizes some of the mathematics underlying common
connectionist systems and shows how these systems have been deployed
to a variety of linguistic problems.  It tries to illustrate both
how connectionism can be helpful as an explanatory and exploratory aid, and
also some of the pitfalls of blindly applying connectionism without
understanding its weaknesses.

\end{abstract}
\end{document}