\documentclass{beamer}
\usepackage{hyperref}
\usepackage{graphicx}
%\usepackage[T1]{fontenc}
\usepackage{verbatim}
%\usepackage{lmodern}
\usepackage[english]{babel}
%\usepackage{natbib}
\mode<presentation>
\usetheme{Warsaw}
\setbeamercovered{transparent}
%\definecolor{blue}{rgb}{0.1,0.1,0.7}
\definecolor{blue}{rgb}{0.51,0.0,0.0} %bordeaux
\setbeamertemplate{navigation symbols}{}

\title[] % (optional, use only with long paper titles)
{On the comprehension-production dilemma in child language \\ Paul Smolensky}

\subtitle[]
{\emph{ in Linguistic Inquiry, Vol. 27, No. 4 (Autumn, 1996), pp. 720-731 }}
\author[] % (optional, use only with lots of authors)
{Andreas van Cranenburgh (0440949)}

\institute[] % (optional, but mostly needed)
{Language \& Optimality, University of Amsterdam}

\date[] % (optional)
{\today}

\subject{Talks}
% If you wish to uncover everything in a step-wise fashion, uncomment
% the following command:

%\beamerdefaultoverlayspecification{<+->}

\begin{document}
\section{Title}

%\begin{frame}
% Can't ... resist ...
%\includegraphics[scale=0.37]{tufte}
%\end{frame}

\begin{frame}
  \titlepage
\end{frame}


\renewcommand{\emph}[1]{\textcolor{blue}{#1}}
\section{Abstract}
\begin{frame}
\frametitle{Abstract} 
\begin{itemize}
\item \emph{Problem:} how to explain that production lags behind comprehension in child language
        \\ two possible answers:
	\begin{itemize}
	\item something outside of grammar restricts children (eg., insufficient motor control)
	\item limitations can be explained by grammar
	\end{itemize}
\item This is a \emph{dilemma} because both answers are undesirable
\item OT to the rescue 
\end{itemize}
\end{frame}




\newpage
\section{Main talk}
\begin{frame}
\frametitle{Issues}

Problems with first answer:

	something outside of grammar restricts children (eg., insufficient motor control)

\begin{itemize}
\item Even if certain words are not produced, they can often be imitated by children
\item Some children make systematic errors, eg. saying `puzzle' for `puddle' and vice versa
\item In general, a solution should explain that children avoid many 
configurations that are marked across languages
\end{itemize}
\end{frame}

\newpage
\begin{frame}
\frametitle{Issues}

Problems with second answer:

	limitations can be explained by grammar

\begin{itemize}
\item In most theories this would mean the child has two grammars 
\item That is bad because it is not parsimonious
%then lots of phenomena can't be explained anymore %which?
\end{itemize}
\end{frame}

\newpage
\begin{frame}
\frametitle{Proposal}

A way out, using one \emph{OT grammar}:

\begin{itemize}
\item Children have rich (adult-like) underlying forms
	\begin{itemize}
	\item These facilitate comprehension
	\item However, during production these are neutralized to unmarked forms
	\end{itemize}
\item The result: a grammar that correctly comprehends but produces \texttt{[ta]} for every input
\end{itemize}
\end{frame}

\newpage
\begin{frame}
\frametitle{Proposal}

This seems paradoxical:
 
\begin{itemize}
\item faithful \emph{comprehension} requires high-ranked faithfulness constraints
\item massive neutralization in \emph{production} requires low-ranked faithfulness constraints
\end{itemize}

Resolution: take kinds of competing structures into account
\end{frame}

\newpage
\begin{frame}
\frametitle{Proposal}

%A structural description consists of a surface form and an underlying form
 
\begin{itemize}
\item In \emph{comprehension} all competitors share the same surface form 
\item In \emph{production} competitors share one underlying form 
\end{itemize}

The result:
\begin{itemize}
\item In \emph{comprehension} the surface forms are equally marked, so
markedness constraints don't differentiate 
\item In \emph{production} neutralization can be achieved by ranking
faithfulness low
\end{itemize}

ie., Smolensky's proposal is that in the initial grammar faithfulness
constraints are dominated by markedness constraints.

\end{frame}


\begin{frame}
\includegraphics[scale=0.1]{table1}
\end{frame}

\begin{frame}
\includegraphics[scale=0.1]{table2}
\end{frame}

\begin{frame}
\includegraphics[scale=0.18]{table3}
\end{frame}

\begin{frame}
\includegraphics[scale=0.16]{table4}
\end{frame}

\newpage
\begin{frame}
\frametitle{Comments}

\begin{itemize}
\item Very extreme case, \emph{all} forms are produced as \texttt{[ta]}. 
\item Not clear (to me) if it would work when many forms are produced correctly
\item For example, wouldn't it require special constraints for the child that
confused `puddle' and `puzzle'?
\item What if an unfaithful input has to be comprehended? The proposal can't
deal with the rad-rat problem! (always comes out faithfully as `rat')

\end{itemize}

In general, OT would predict that an improved ranking of constraints produces a
general impromevent, whereas evidence indicates that children improve
selectively (eg., only inflect most frequent verbs correctl)

\end{frame}

\newpage
\begin{frame}
\frametitle{Comments}

\begin{itemize}
\item That children have rich, adult-like underlying structure might be plausible for phonology
\item But not for semantics / pragmatics, I think
\begin{itemize}
\item Acquiring the right representations seems the most crucial part of language acquisition
\end{itemize}
\item Two-word stage seems qualitatively different from adult-language
\begin{itemize}\item Childreem seem to \emph{gradually} move from concrete experience to abstract usage
\end{itemize}
\end{itemize}

But OT would suggest that children are merely fine-tuning their ordering of constraints.

\end{frame}

\newpage
\begin{frame}
\frametitle{Comments}

My hypothesis:

\begin{itemize}
\item In the two-word stage children's meaning representation is limited (limited abstraction)
\item The mental \emph{lexicon} is very incomplete, mainly contains content
words for objects and actions relevant to the child (toys, food, etc.) 
\item Utterances are interpreted and produced by extrapolating from previous
linguistic experience (exemplars) 
\item Production can only use words which have already been \emph{abstracted}
from experience, whereas comprehension can make use of the whole situation to
make an educated guess
\end{itemize}

\end{frame}

\newpage
\begin{frame}
\frametitle{Conclusion}


\begin{itemize}
\item OT can model a lot of phenomena with little mechanism
\item but OT's strong assumptions about representations make the problem of
language acquisition too easy $\rightarrow$ little explanatory power
\item The comprehension-production dilemma benefits from being viewed in a
broader cognitive perspective
\end{itemize}

\end{frame}


\end{document}

\begin{comment}
\newpage
\begin{frame}[fragile]
\frametitle{Results: simulate dialogue from Childes}
\begin{verbatim}
*MOT:  that's the cow .
*CHI:  cow
*MOT:  what's this ?
*CHI:  
*MOT:  is that a donkey ?
*CHI:  donkey
*MOT:  right .
*CHI:  
*MOT:  that's a donkey .
*CHI:  donkey
*MOT:  what's this [=duckie] ?
*CHI:  duckie
*MOT:  what does a duckie say ?
*CHI:  quack@o
\end{verbatim}

\end{frame}
\end{comment}

\newpage

\section{Bibliography}
\begin{frame}
{\bf Bibliography}
%\bibliographystyle{plainnat}
\begin{thebibliography}{9}

\bibitem{Smolensky} 
	Paul Smolensky,
	\emph{On the comprehension-production dilemma in child language},
	in Linguistic Inquiry, Vol. 27, No. 4 (Autumn, 1996), pp. 720-731

\end{thebibliography}
\end{frame}