% LaTeX \documentclass[leqno]{article} \usepackage{url} \usepackage{graphicx} \usepackage{amsmath} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{bm} \usepackage{gellmu} \usepackage[margin=100bp,nohead]{geometry} \setlength{\parskip}{6bp} \setlength{\parindent}{0bp} \thispagestyle{empty} \title{Why \emph{eos}? \ } \newlength{\centerskip} \setlength{\centerskip}{\topsep} \newcommand{\hsf}{\hspace*{\fill}} \newcommand{\tdbc}[1]{\hsf\textbf{#1}\hsf} \newenvironment{menulist}{ \begin{list}{}{ \setlength{\topsep}{0bp} \setlength{\labelwidth}{0.03\linewidth} \setlength{\leftmargin}{0.06\linewidth} \setlength{\itemindent}{0bp} \setlength{\itemsep}{-6bp} \setlength{\parsep}{6bp}} }{\end{list}} \newenvironment{Menulist}{ \begin{list}{}{ \setlength{\topsep}{0bp} \setlength{\labelwidth}{0.03\linewidth} \setlength{\leftmargin}{0.06\linewidth} \setlength{\itemindent}{0bp} \setlength{\itemsep}{3bp} \setlength{\parsep}{6bp}} }{\end{list}} \newenvironment{toclist}{\normalsize \begin{list}{}{ }}{\end{list}} \newenvironment{Toclist}{\large \begin{list}{}{ }}{\end{list}} \newenvironment{citations}{ \begin{list}{}{ \setlength{\topsep}{0bp} \setlength{\labelwidth}{0bp} \setlength{\leftmargin}{0.04\linewidth} \setlength{\labelsep}{0bp} \setlength{\itemindent}{-0.2\leftmargin} \setlength{\itemsep}{3bp} \setlength{\parsep}{0bp}} }{\end{list}} \author{William F. Hammond} \begin{document} \begin{center}\LARGE\bfseries{} Why \emph{eos}? \ \end{center} \begin{center}\Large\bfseries{} \textsl{William F. Hammond} \end{center} \medskip \par{The \textsc{GELLMU} \emph{article} document type has an end-of-sentence mark \emph{eos}, which is a defined-empty \textsc{XML} element, corresponding to the concept of sentence in languages such as English and French. \ But there is no provision for regarding a western sentence itself as an \textsc{XML} element. \ Why? \ } \par{There are two reasons. \ } \par{Sometimes one wants to begin a \emph{display} in the middle of a sentence. \ Then it can happen that the display is the last part of the sentence. \ It is a formal rule of \textsc{XML} that if an element begins inside another element, then the second element must be closed before the first element is closed. \ Following this rule, when a display is the last part of a sentence, the display must be ended before the sentence is ended. \ As a consequence an \textsc{XML} processor must usually work very hard to place the sentence-ending punctuation mark correctly. \ } \par{Is this just a technical \textsc{XML} issue? \ Not really. \ } \par{The second reason for modeling an end-of-sentence mark but not a sentence is that some literary use of a language such as English does not actually resolve into clean sentence units even though end-of-sentence punctuation is used. \ } \par{One could argue that when a sentence is used, it could be marked up with a \emph{sentence} element\footnote{The model would then likely permit each of \emph{sentence} and \emph{display} to contain the other}. \ In that event it is unlikely that authors would want to be required to insert end-of-sentence marks explicitly. \ Moreover, there would be something of a dilemma for the \textsc{XML} processor if it happens to notice an item of \textsc{CDATA} at the end of a sentence that appears to be an end-of-sentence mark. \ There would still be the vexation caused by a display that ends a sentence. \ And would authors use the \emph{sentence} element? \ } \par{Will authors want to use the explicit \emph{eos} rather than the simple \textsc{CDATA} punctuation mark \texttt{`.'}? \ If so, how is the sequence \texttt{".\string"} to be handled by a processor? \ } \par{Authors are the end users, and authors need convenience. \ Reasonable convenience lies in the convention that began with the dawn of the mechanical typewriter: \begin{quote} A sentence is ended with a period followed either by a newline or by two or more blank spaces. \ \end{quote} } \par{Handling this convention is not a reasonably efficient task for an \textsc{XML} processor. \ But it works very well with a \LaTeX{}-like markup interface for \textsc{XML}, i.e., when there is pre-processing from \LaTeX{}-like markup to \textsc{XML} markup. \ } \end{document}