diff options
Diffstat (limited to 'Master/texmf-dist/doc/latex/linguex')
| -rw-r--r-- | Master/texmf-dist/doc/latex/linguex/linguex-doc.tex | 543 | ||||
| -rw-r--r-- | Master/texmf-dist/doc/latex/linguex/ps-trees-doc.tex | 305 |
2 files changed, 848 insertions, 0 deletions
diff --git a/Master/texmf-dist/doc/latex/linguex/linguex-doc.tex b/Master/texmf-dist/doc/latex/linguex/linguex-doc.tex new file mode 100644 index 00000000000..7d4734392c0 --- /dev/null +++ b/Master/texmf-dist/doc/latex/linguex/linguex-doc.tex @@ -0,0 +1,543 @@ +%% linguex-doc.tex version 3.4 +%% documentation of linguex.sty +%% Author: Wolfgang.Sternefeld@uni-tuebingen.de +%% filedate 02/01/26 +\documentclass{article} +\usepackage{linguex} +%\usepackage{linguho} +%\usepackage[*]{linguho} +%\documentstyle[linguex]{article} +\sloppy\raggedbottom +%\addtolength{\alignSubExnegindent}{.7em} +%%%%%%%%%%%%%%%%%%%%%%%%%% +\begin{document} +\title{linguex.sty Documentation} +\author{Wolfgang Sternefeld\\ +Version 3.4 --- January 2002} +\date{} +\maketitle + +\subsection*{Abstract}% +{\tt linguex.sty} is a \LaTeX-tool written for the lazy linguist. Its +handling of example numbering, indentations, indexed brackets, and +grammaticality judgments reduces the effort of formatting linguistic +examples to a minimum. It also allows for automatic extraction of a +handout. + +The macro runs with \LaTeXe\ only and requires two additional +style files: {\tt xspace.sty} from +the base of and {\tt cgloss4e.sty} (a modification of the +midnight gloss macro) for glosses. + +\subsection*{The Three Basic Commands} The linguex macro defines three +basic commands: \verb-\ex.-, \verb-\a.-, and \verb-\b.-. The first +two generate list environments. The third functions basically like an +\verb-\item-. Here is an example. Type setting + +%\begin{samepage} +\begin{verbatim} +\ex. This is the first level of embedding + \a. This is the second level + \b. This is still the second level, but: + \a. This is the third level + \b. This is not the end. + \b. This is the end +\end{verbatim} +will print: + +\ex. This is the first level of embedding + \a. This is the second level + \b. This is still the second level, but: + \a. This is the third level + \b. This is not the end. + \b. This is the end + +%\end{samepage} +The list environment created by \verb-\ex.- must be closed by a single +blank line (a \verb-\par-). The text following this list will not be +indented. In order to get a \verb|\parindent|, a \textit{second} +blank line must be added, immediately following the first. + +The commands \verb-\c.-, \verb-\d.-, \verb-\e.-, and \verb-\f.- are +equivalent copies of \verb-\b.-; they were defined to produce a kind +of WYSIWYG-effect. + +The blank line at the end of \Last simultaneously closes the three +lists of \Last (opened by \verb|\ex.|, \verb|\a.|, and the embedded +\verb|\a.|). To close just one of the embedded lists (introduced by +\verb|\a.|) put \verb-\z.- at the point of transition from +level $n$ of the embedding to level $n-1$. This is exemplified by the +following example: + +%\begin{samepage} +\ex.\a. {\it Government:}\\ + A governs B iff + \a. A is a governor; + \b. A m-commands B; + \c. no barrier intervenes between A and B + \z. + \c. Governors are lexical nodes and tensed I. + \z. + (from Haegeman 1991) + +% +\begin{verbatim} +This is exemplified by the following example: +\ex.\a. {\it Government:}\\ + A governs B iff + \a. A is a governor; + \b. A m-commands B; + \c. no barrier intervenes between A and B + \z. + \b. Governors are lexical nodes and tensed I. + \z. + (from Haegeman 1991) +\end{verbatim}%\end{samepage} +% + +\subsection*{Cross References}% +can be handled by \verb-\label{...}- and \verb-\ref{...}- in the usual +manner. For example, a label between \verb-\ex.- and \verb-\a.- +stores the main example number for further reference; the same command +following \verb-\a.- or \verb-\b.- stores the label of a +(sub-)subexample for further reference. + +For cross references in the immediate vicinity of an example, you need +not \verb|\label| the examples. \verb-\Next- refers to the following +example number, and \verb-\NNext- to the next but one. Reference to +the previous example is provided by \verb-\Last-; the penultimate one +is referred to with \verb-\LLast-. + +These shorthands always refer to the first level of embedding. +Reference to subexamples can only be obtained by adding an (optional) argument +to the above Next- and Last-commands. For example, saying +\verb.\NNext[g-ii]. right now yields \NNext[g-ii]. The dash between +\addtocounter{ExNo}{2}\arabic{ExNo}\addtocounter{ExNo}{-2} and g is +defined as \verb|\refdash|. It can be suppressed by +\verb|\renewcommand{\refdash}{}|. + + +\subsection*{Footnotes}% +Inside a footnote, the \verb-\Next--command presupposes that the next +example is still inside the footnote rather than within the main body +of the document. Say \verb-\TextNext- to refer to the next example within the +main text. (When you intend to make endnotes instead of footnotes, +all cross references from footnote to text must be handled by +\verb-\label{...}- and \verb-\ref{...}-.) + +In case an \verb-\ex.--environment contains a footnote that contains a +\verb-\par- (= a blank line --- this will necessarily be the case with a +footnote containing itself an \verb-\ex.-), the \verb-\par- will be +misinterpreted as the end of the main text example. It is therefore +necessary to split the command \verb-\footnote{...}- into +\verb-\footnotemark- and \verb-\footnotetext{...}- (see +\LaTeX-manual). The footnotetext must be placed outside the +\verb-\ex.--environment of the main text. + +Some style files (e.g. endnote.sty) modify the definition of +footnotes (as linguex.sty itself does so in order to let \verb|\ex.| know +whether or not it is inside a footnote). Such style files must be +accommodated to linguex.sty by making sure that \verb-\if@noftnote- is +set false at the beginning of each footnote (by saying +\verb-\@noftnotefalse- in the modified footnote definition); otherwise +you will get the arabic style of example numberings (as being used in +the main text) rather than the roman numbers (being used inside +footnotes). + + +\subsection*{Glosses}% +require {\tt cgloss4e.sty}, which is input by {\tt linguex.sty}. +Instead of writing \verb.\gll. (as would be required by the gloss +macro), one should append a `g' to the last letter of an example +command, as shown below: + +\begin{samepage}\begin{verbatim} +\ex.\a. No gloss + \bg. This is a first gloss\\ + Dies ist eine erste Glosse\\ + +\exg. Dies ist nicht die erste Glosse\\ + This is not the first gloss\\ +\end{verbatim} +% +\ex.\a. No gloss + \bg. This is a first gloss\\ + Dies ist eine erste Glosse\\ + +\exg. *Dies ist nicht die erste Glosse\\ + This is not the first gloss\\ + +\end{samepage} +Using \verb.\gll. still works in principle, but should be avoided, +for reasons explained in the next section. + + +\subsection*{Grammaticality Judgments}% +composed out of *, ?, \#, and \%, are automatically prefixed at the +very beginning of a new list. A standard example like \Next +looks like this (to increase readability I also removed mathmode from +subscripting): + +\catcode`_=\active +\def_#1{\ifmmode\sb{#1}\else$\sb{#1}$\fi} + +\begin{verbatim} +\catcode`_=\active +\def_#1{\ifmmode\sb{#1}\else$\sb{#1}$\fi} +\exg. \%*Wen_i liebt_k seine_i Mutter t_i t_k?\\ + Whom loves his mother\\ + `Who does his mother love?' +\end{verbatim} +\exg. \%*Wen_i liebt_k seine_i Mutter t_i t_k?\\ + Whom loves his mother\\ + `Who does his mother love?' + +Automatic prefixing implies that \textbf{nothing} intervenes between the list +opening command and the judgment. In particular, labels must {\it +follow\/} the grammaticality judgment! Likewise, writing +\verb|\ex.\gll| instead of \verb|\exg.| will have the effect of not +prefixing the grammaticality judgment. + + +\subsection*{Labelled Brackets}% + \verb-\exi.- identifies ``words'' by looking for space characters + between them; it then checks whether the word following a space + starts with one of the brackets ``['' or ``]''. The material between + such a bracket and the next space character will be subscripted. A + standard example would be the following: +\begin{verbatim} +\exi. *[CP Wen_i [C$'$ liebt_j [IP [NP seine_i Mutter ]% + [VP t_i t_j ]]]] +\end{verbatim} +\exi. *[CP Wen_i [C$'$ liebt_j [IP [NP seine_i Mutter ]% + [VP t_i t_j ]]]] + +Note that above a blank space between {\tt [IP} and {\tt [NP} is +crucial, otherwise the bracket of {\tt [NP} would be subscripted as +well. Note also that a space is required before the grammaticality +judgment, otherwise the beginning of subscripting cannot be properly +identified. + +As with the gloss macro, grouping surpresses subscripting. +Consider the effects of spacing and grouping in \Next: +\begin{verbatim} +\exi.\a. [[NP Fritz ][ snores ]]S + \b. [[NP Fritz][snores]]S + \c. [ [NP Fritz ][snores]]S + \d. [[NP Fritz {][}snores ]]S + \e. {[[+N,--V]} Fritz ][VP snores ] \hfill{} [NP Structure] + \f. *[{}[+N,--V] Fritz ][VP snores ] \hfill [NP-Structure] +\end{verbatim} +\exi.\a. [[NP Fritz ][ snores ]]S + \b. [[NP Fritz][snores]]S + \c. [ [NP Fritz ][snores]]S + \d. [[NP Fritz {][}snores ]]S + \e. {[[+N,--V]} Fritz ][VP snores ] \hfill{} [NP Structure] + \f. *[{}[+N,--V] Fritz ][VP snores ] \hfill [NP-Structure] + +Grouping works the same way as in the gloss macro where it protects +from glossing. E.g., a group such as \verb.{][}. in \Last[d] is +ignored for subscripting so that the following material is protected +from automatic subscripting (but note that something like the opposite +happens in \Last[e] after \verb|\hfill|!). + +Note that commands that look ahead and extend over more than one word +might not work properly unless two or more words are put into a group. +Suppose we are inside an \verb-\exi.--environment and we want to say +something like \verb-\b.[Principle C]-. What will happen? The space +between ``\texttt{e}'' and ``\texttt{C}'' will make the subscripting +device read \verb-\b.[Principle- as a ``word.'' This, unfortunately, +prevents the argument of \verb-\b.- from being interpreted correctly. +Solutions are to write \verb-\b.[Principle~C]-, \verb-\b.[{Principle C}]-, +or \verb-{\b.[Principle C]}-. + +Subscripting outside the \verb|\exi.|-environment can be done by using +\verb-\I- as shown below: + +\medskip +\verb|\I[NP Text| $\Longrightarrow$ \I[NP Text, + + \verb.\I($\alpha$ Text. $\Longrightarrow$ \I({$\alpha$} Text, + + \verb.\I{$\alpha$}NP Text. $\Longrightarrow$ \I{$\alpha$}NP Text, + + \verb.\I{Whom}ACC does John\ldots. $\Longrightarrow$ \I{Whom}ACC + does John\ldots + +\medskip\noindent +For combining glosses with labelled brackets, +say \verb-\exig.- or \verb-\exgi.-. Note that although the +list-generating commands \verb-\ex.- and \verb-\a.- can be combined in +any order, all commands that introduce automatic subscripting can be used +only at the top level, i.e. cannot be embedded!!!. Note also that +$n$ nested \verb-\ex.--commands need $n$ consecutive blank lines +(\verb.\par.s) to properly identify the end of all lists. + +\subsection*{Handouts}% +Since the surface implementation of the list environment does not use +the begin-end-format, style files that suppress printing between the +end and the beginning of certain environments (e.g. {\tt +xcomment.sty}) could unfortunately not be adopted to produce a handout +(I didn't see a way of doing so, probably because I didn't understand +what goes on inside xcomment.sty). + +This deficiency might be compensated for by using {\tt linguho.sty}, +which removes all examples and headings (of sections and subsections) +from the \LaTeX-document and stores them in a file whose extension is +.han (namely \textit{jobname}.han). Putting \verb.\makehandout. +somewhere at the end of the document (e.g. before the bibliography) will +print out \textit{jobname}.han at the position of \verb-\makehandout-. If +\verb-\section*{...}- and \verb-\subsection*{...}- should also go into +the handout, say \verb-\usepackage[*]{linguho}-. Use +\verb-\maketitle- to get the title into the handout. + +Note that all of \LaTeX's \verb|\setlength| and \verb|\settowidth| +instructions automatically go into \textit{jobname}.han and are +executed only there. (The main body of text before +\verb-\makehandout- is considered irrelevant, but I didn't find a way +to suppress writing it into the dvi-file. Any help to solve this +problem will be appreciated; note that I am not a \TeX-wizzard, only a lazy +linguist). + +In case you have defined \verb.\active. characters (e.g. Umlaut in +German) these will most likely turn out troublesome (i.e. these +commands should normaly be protected). For a general solution, +inspect the definition of \verb.\MkOthersSpecial. in linguho.sty. +Uncommenting any of the lines there will solve the problem for the +active character contained in the respective line (sometimes it might +in addition be necessary to use t1enc.sty or germanb.sty instead of +german.sty). + + +\subsection*{Customizing Lengths and Margins}% +The user may modify any of the lengths displayed in the following +scheme, which also shows their predefined default values: + +\setlength{\Exlabelwidth}{0pt} +\setlength{\Exlabelsep}{0pt} +\ex. \label{diagramm}\footnotesize\arraycolsep2pt +$\begin{array}[t]{@{}l|l|l|l|l@{}} + \multicolumn{5}{c}{\backslash{\rm Extopsep} \updownarrow + 0.66\backslash{\rm baselineskip}} +\\[1ex] + \underbrace{\backslash\rm Exindent}_{{\rm 0pt}}& + \fbox{$\backslash$\rm Exlabelwidth} + \underbrace{\backslash\rm Exlabelsep}_{{\rm 1.3em}} + & {\rm Text ...} &&\\ + && + \rm a. &\rm Text \ldots & \\ + && \underbrace{\backslash\rm SubExleftmargin}_{{\rm 2.4em}} + &\rm (i) & \rm Text \ldots \\ + && & \underbrace{\backslash\rm SubSubExleftmargin}_{{\rm 2em}} + &\rm\ldots text. \\[2ex] + \multicolumn{5}{c}{\backslash{\rm Extopsep}\updownarrow + 0.66\backslash{\rm baselineskip}} + \end{array}$ + \global\alignSubExtrue + +The value of \verb.\Exlabelwidth. is determined by the width of the +current label. Note that for displaying the table as shown in \Last, +the leftmargin of the list must be zero; accordingly, both +\verb|\Exlabelwidth| and \verb|\Exlabelsep| were set to zero. To +restore all default values of the lenghts shown in \Last +simultaneously, say \verb-\resetExdefaults-. The exact behavior of +\verb.\Exlabelwidth. is explained further below. With some non +standard fonts the default spacing can be defined only +\verb.\AtBeginDocument. (for reasons I do not understand), therefore + the default values for the above lengths are declared + \verb.\AtBeginDocument.. In consequence, any +deviation from the default must be specified +after \verb.\begin{document}. + + +Some journals require a different arrangement of margins, e.g.\ one +which aligns the left edges of examples with the left edges of +subexamples, as shown in \Next and \NNext: + +\resetExdefaults +\addtolength{\Exlabelsep}{.5em}% + \addtolength{\alignSubExnegindent}{.5em} +\ex. \label{align}First example: +\a. Subexample1 + \b. Subexample2 + \a. Subsubexample1 \ldots\global\alignSubExfalse + +\addtolength{\Exlabelsep}{-.5em}% +\addtolength{\alignSubExnegindent}{-.5em}% +By saying \verb.\alignSubExtrue. before \verb.\begin{document}., +the sublabels (a. and b. in \Last) +are given a negative indentation. This is controled by the lengths +\verb.\alignSubExnegindent. and \verb.\Exlabelsep., +which can be customized by \verb-\setlength-. after \verb.\begin{document}.. + + +\subsection*{Customizing Labels}% +The labels of the examples are created by the counters \verb-ExNo-, +\verb-SubExNo-, and \verb-SubSubExNo- for the three levels of +embedding respectively; \verb-FnExNo- is used for the first level +inside footnotes. These counters are the default labels of the list +items. They can be changed anywhere in the document, by using +\verb.\setcounter. as usual. + +There are a number of further possibilities to generate labels that differ +from the default. One is to +use \verb-\a.- without embedding it into \verb|\ex.|. This will yield the +expected result, namely a list that uses the \verb|SubEx| counter at +the top-level: + +\a. First line +\b. Second line +\c. Third line + +Moreover, any of the list commands can take an optional +argument, whose specification replaces the default label: + +\begin{samepage} +\a.[{\bf Principle C} (Chomsky{[81]}): ] An R-expression is free only + \a.[$\alpha$)] with respect to potential binders + in A-positions, + \b.[$\beta$)] within the domain of its chain. + +\begin{verbatim} +\a.[{\bf Principle C} (Chomsky{[81]}): ] + An R-expression is free only + \a.[$\alpha$)] with respect to potential binders + in A-positions, + \b.[$\beta$)] within the domain of its chain. + \end{verbatim} +\end{samepage} +% +The optional argument must immediately follow ``.'', so don't leave +space between \verb|\ex.| or \verb|\a.| and ``['' !!!. By contrast, +it will often be the case that an example starts with a (labelled) +bracket like \I[NP . If so, it is obligatory to put a space +between ``.'' and ``['' . + +As illustrated above, the topmost \verb#\a.#-command picks up the +\verb|\Exlabelwidth| of the preceding example. By contrast, using +\verb-\ex.[{\it Principle C} ...]- instead of % +\verb-\a.[{\it Principle C} ...]- has the effect of adjusting the +indentation of the following text to the width of the optional +argument, as shown below: + +\ex.[{\bf Principle C} (Chomsky{[81]}): ] An R-expression is free only + \a.[$\alpha$)] with respect to potential binders + in A-positions, + \b.[$\beta$)] within the domain of its chain. + +Assuming that regular example numbers do not exceed (999), this +departure from the default behavior only occurs for optional arguments +whose width exceeds that of (999). + +Optional arguments can also be used to simulate other environments. +E.g., \LaTeX-style itemizing at the top level can be simulated as follows: +\begin{verbatim} +\ex.[\hfill$\bullet$\hfill] Line 1 +\b. Line 2 +\c. Line 3 +\end{verbatim} +\ex.[\hfill$\bullet$\hfill] Line 1 +\b. Line 2 +\c. Line 3\global\let\oldalph=\alph\global\let\alph=\arabic + +Temporarily redefining \verb.\alph., as shown in the following +example, yields an enumeration: +\begin{verbatim} +\let\oldalph=\alph\let\alph=\arabic +\a. Text 1 +\b. Text 2 +\c. Text 3 \global\let\alph=\oldalph +\end{verbatim} + +\a. Text 1 +\b. Text 2 +\c. Text 3 \global\let\alph=\oldalph\global\alignSubExtrue + +The stylesheet of \textit{English Language +and Linguistics} requires examples of the following form: + +\setlength{\Exlabelsep}{2em} +\exi. +[v [v broke_i $\emptyset$ ][VP the vase [V$'$ t_i into +pieces ]]] + +\ex.\def\SubExLeftBracket{(}\def\SubExRightBracket{)} +\a. It kept warm + \b. She kept it warm\global\alignSubExfalse + + +Apart form +specifying \verb.\alignSubExtrue. before \verb.\begin{document}. +this requires to redefine the brackets of sublabels, as shown below: +\resetExdefaults +\begin{verbatim} +\renewcommand{\SubExLeftBracket}{(} +\renewcommand{\SubExRightBracket}{)} +\setlength{\Exlabelsep}{2em} +\exi. [v [v broke_i $\emptyset$ ][VP the vase [V$'$ t_i into + pieces~]]] + +\ex.\a. It kept warm + \b. She kept it warm +\end{verbatim}% +In addition, \verb|\renewcommand{\refdash}{}| changes cross +references from eg. \Last[b] to \renewcommand{\refdash}{}\Last[b]. + + +\subsection*{$\backslash$\tt Exlabelwidth}% +By default, the value of \verb.\Exlabelwidth. is determined by the +width of the example numbering. More precisely, for labels between +(1) and (9) the labelwidth is by default that of (11); accordingly, the +actual labelwidth is a bit larger than the natural size of the label. +Similarly, \verb.\Exlabelwidth. has the width of (110) for +\verb-\theExNo-'s between (10) and (99). Finally, +\verb,\Exlabelwidth, has the width of (1100) for example numbers +between (100) and (999). For even wider labels (usually specified by +an optional argument, cf. above), the label retains its natural size. + +Since the space between the \verb.\Exlabelwidth. and the text +remains the same, the transition from (9) to (10) and from (99) +to (100) will cause a small change of indentation, which might look +ugly, particularly in handouts. In order to +supress the default behavior of \verb|\Exlabelwidth|, it must be +assigned a particular length. For example, saying +\verb.\settowidth{\Exlabelwidth}{(110)}. will cause the labelwidth of +all examles from (1) to (99) to be identical. + +Since wide labels should still retain their natural size, the user's +specification of \verb|\Exlabelsep| should not be wider than (1100), +otherwise the default mechanism is still active, and the value of +\verb|\Exlabelsep| is ignored. E.g., saying +\verb|\setlength{\Exlabelwidth}{3em}| will most likely cause the list +declaration to take up its default behavior, because ``3em'' is wider +than ``(1100)''. (The default value of +\verb|\Exlabelwidth| is 4em). + + +\subsection*{cgloss4e.sty} +I found two problems with cgloss4e.sty. The first relates to AMS-\TeX. +Writing a translation immediated below the glossed material normally +behaves as one would expect, but with AMS-Tex the translation becomes +indented, for reasons I do not understand. To remidy this, one has to +(re-)define the list indentation, e.g. as shown below (from \LaTeX): + +\begin{verbatim} +\makeatletter +\def\@listi{\leftmargin\leftmargini + \parsep 5\p@ \@plus2.5\p@ \@minus\p@ + \topsep 10\p@ \@plus4\p@ \@minus6\p@ + \itemsep5\p@ \@plus2.5\p@ \@minus\p@} +\let\@listI\@listi +\@listi +\makeatother +\end{verbatim} +The second problem concerns the use of \verb.\gll. at a position +other than the beginning of a list. Then the gloss starts a new +paragraph and puts unwanted vspace above and below the list. One way +to suppress this is to change the flushleft commands used within +cgloss4e.sty into raggedright by saying +\verb.\renewenvironment{flushleft}{\raggedright}{}.. + + +%\makehandout +\end{document}
\ No newline at end of file diff --git a/Master/texmf-dist/doc/latex/linguex/ps-trees-doc.tex b/Master/texmf-dist/doc/latex/linguex/ps-trees-doc.tex new file mode 100644 index 00000000000..25377efa567 --- /dev/null +++ b/Master/texmf-dist/doc/latex/linguex/ps-trees-doc.tex @@ -0,0 +1,305 @@ +%% ps-trees-doc.tex version 2 +%% Author: Wolfgang.Sternefeld@uni-tuebingen.de +%% filedate 99/05/04 +\documentclass{article} +\usepackage{ps-trees} +\parindent0pt\sloppy +\parskip\bigskipamount +\raggedbottom +%\overfullrule2pt + +\begin{document} +\title{ps-trees.sty Documentation} +\author{Wolfgang Sternefeld\\ + Version 2 --- May 1998} +\maketitle + +Experimenting with a number of tree-macros I found that complex +real-life examples often lead to problems with controlling the +positioning of nodes in a tree. I therefore decided to use +tree-dvips.sty (by Emma Pease), where the user has full control +over the position of nodes by putting them into a table. The present +macro inputs tree-dvips.sty and presupposes knowledge of its +node-commands. + +One of the shortcomings of tree-dvips.sty is that it does not control +the correct use of the names for nodes. This means that in case of a +user error, no warning will appear and no output will be printed. +The basic idea of the present macro is to regain control +over the use of names by using them as an encoding of the structure of +a tree. More specifically, for a node named $x$ to be connected with +a node named $y$ (i.e. $x$ and $y$ are the first arguments of the +command \verb|\node|) $y$ must have the form $cx$, where $x$ and $y$ +are strings of characters and $c$ is exactly one character. For +ordinary branching trees, this makes the specification of +\verb.\nodeconnect{#1}{#2}. superfluous. Consider the following +example and compare the printed file with the .tex-source of this +file: + +\begin{treetab}{*4c} + &\Node{}{IP} \\[3ex] +\Node{l}{D} & &\Node{r}{VP}\\[3ex] +\Node{ml}{She}&\Node{lr}{V} & &\Node{rr}{D}\\[3ex] + &\Node{mlr}{exonerated}& &\Node{mrr}{him} +\end{treetab} +% +\hspace*{\fill} +\begin{minipage}{.7\textwidth} +\begin{verbatim} +\begin{treetab}{*4c} +&\Node{}{IP}\\[3ex] +\Node{l}{D}&&\Node{r}{VP}\\[3ex] +\Node{ml}{She}&\Node{lr}{V}&&\Node{rr}{D}\\[3ex] +&\Node{mlr}{exonerated}&&\Node{mrr}{him} +\end{treetab} +\end{verbatim} +\end{minipage}\hfill +% + +\verb.\begin{treetab}. is defined in terms of \verb.\begin{tabular}.. +The actual execution of the node connections (based on the node names +as specified by the first argument of \verb|\Node|) is performed as a part of +\verb.\end{treetab}.. In case a node cannot be interpreted as a node +in a tree, a warning will be typed out. + +The remaining problem is to gain control over the width of columns in +a table. This is often tedious, but the effort seems worthwhile as +evidenced by the poor results of macros that calculate the position of +nodes automatically. Being essentially an aesthetic problem, the +present macro offers only limited support towards a uniform solution. +The following, however, might be helpful. + +First of all, the effects of \verb.\tabcolsep. and \verb.\nodemargin. +on a precise positioning of nodes are disturbing, so both values are +set to zero by \verb.\begin{treetab}.. In order to achieve a minimum +of symmetry of branchings, a first idea would be that all rows should +have the same width, implying that the width of each node should be +determined by the widest leave. Since individual nodes that come +later in a table might determine the size of previous nodes, it might +be a good idea to put all relevant nodes (by default: all leaves) into +boxes before starting the table. The minimal size of all nodes (or +columns) can then be determined by the command \verb.\NodeWidthNo{n}., +where $n$ is the number of the box, whose width should become the +minimal width of all other columns. \verb.\NodeWidthNo. is defined +as follows: +\begin{verbatim} +\newcommand{\NodeWidthNo}[1]{\setlength{\MinNodeWidth}{\the\wd#1}} +\end{verbatim} +The length \verb|\MinNodeWidth| will subsequently be used to +determine the minimal width of each node. (In order to give an +individual node the width of a particular prespecified box +$n$, the node command can take an optional argument +\verb|\Node[n]{...}{...}|). + +For example, if we let the word \textit{exonerate} determine the size +of all nodes, the result is the following: + +\begin{verbatim} +\setbox1=\hbox{She} +\setbox2=\hbox{exonerated} +\setbox3=\hbox{him} +\NodeWidthNo2 +% +\begin{treetab}{*4c} +&\Node{}{IP}\\[3ex] +\Node{l} {DP}&&\Node{r}{VP}\\[3ex] +\NodeNo1{ml}&\Node{lr}{V}&&\Node{rr} {DP}\\[3ex] +&\NodeNo2{mlr}&&\NodeNo3{mrr} +\end{treetab} +\end{verbatim} +\setbox1=\hbox{She} +\setbox2=\hbox{exonerated} +\setbox3=\hbox{him} +\NodeWidthNo2 +% +\begin{treetab}{*4c} +&\Node{}{IP}\\[3ex] +\Node{l} {DP}&&\Node{r}{VP}\\[3ex] +\NodeNo1{ml}&\Node{lr}{V}&&\Node{rr} {DP}\\[3ex] +&\NodeNo2{mlr}&&\NodeNo3{mrr} +\end{treetab} + +But now the space between box2 and box3 might strike one as too wide, +being seperated by the box of the VP. One possibility would be to +specify VPs node as \verb|\Node[0]{r}{VP}| which puts VP into a box +whose width is \verb|\wd0|, which is zero because box0 has not yet +been used. A more general way would be to place the VP node right +into the middle between two columns. This saves us one column in the +table; the command \verb.\NodeZ. makes a box of width zero and puts it +at the right edge of a column (i.e. ``between'' two columns): % +\begin{verbatim} +\begin{treetab}{ccc} +&\Node{}{IP}\\[3ex] +\Node{l} {DP}&\NodeZ{r}{VP}\\[3ex] +\NodeNo1{ml}&\Node{lr}{V}&\Node{rr} {DP}\\[3ex] +&\NodeNo2{mlr}&\NodeNo3{mrr} +\end{treetab} +\end{verbatim} +\setbox1=\hbox{She} +\setbox2=\hbox{exonerated} +\setbox3=\hbox{him} +\NodeWidthNo2 +% +\begin{treetab}{ccc} +&\Node{}{IP}\\[3ex] +\Node{l} {DP}&\NodeZ{r}{VP}\\[3ex] +\NodeNo1{ml}&\Node{lr}{V}&\Node{rr} {DP}\\[3ex] +&\NodeNo2{mlr}&\NodeNo3{mrr} +\end{treetab} + +But now we got an asymmetry of the branches at the top level, which +can be fixed by shifting the IP to the left. This can be done by +replacing \verb|\Node{}{IP}| by \verb|\NodeZ{}{IP}\hspace*{0.75\wd2}|: + +\setbox1=\hbox{She} +\setbox2=\hbox{exonerated} +\setbox3=\hbox{him} +\NodeWidthNo2 +\begin{treetab}{ccc} +&\NodeZ{}{IP}\hspace*{0.75\wd2}\\[3ex] +\Node{l} {DP}&\NodeZ{r}{VP}\\[3ex] +\NodeNo1{ml}&\Node{lr}{V}&\Node{rr} {DP}\\[3ex] +&\NodeNo2{mlr}&\NodeNo3{mrr} +\end{treetab} + +For larger trees these asymmetries have to be levelled out +consecutively, as shown in the next tree (an example from real life): + +%%%%%%%%%%%%%%%%%%%%%%%%% +\let\N=\Node +\let\NZ=\NodeZ +\let\NNo=\NodeNo +\let\NTNo=\NodeTNo +\let\MNW=\MinNodeWidth +%%%%%%%%%%%%%%%%%%%%%%%%%% +\newcommand{\End}{\\[4ex]} +\newcommand{\trivtab}[2]{\hbox{\begin{tabular}[t]{@{}c@{}}#1\\ + #2\end{tabular}}} +\newcommand{\twotab}[2]{\hbox{\begin{tabular}[t]{@{}ll@{}}#1\\ + #2\end{tabular}}} +% +\tabcolsep2pt +\setbox0=\twotab{mit & wem$_i$}{with & who} +\setbox1=\trivtab{glaubst}{think} +\setbox2=\trivtab{du}{you} +\setbox3=\hbox{t$_j$} +\setbox4=\hbox{t$_i$} +\setbox5=\trivtab{dass}{that} +\setbox6=\trivtab{Maria}{Maria} +\setbox7=\hbox{t$_i$} +\setbox8=\twotab{gesprochen & hat}{spoken & has} + +% +\NodeWidthNo1% +\newcommand{\adjustA}{\hspace*{.75\MNW}} +\newcommand{\adjustB}{\hspace*{.625\MNW}} +\newcommand{\adjustC}{\hspace*{.5625\MNW}} +\newcommand{\adjustD}{\hspace*{.53125\MNW}} +\newcommand{\adjustE}{\hspace*{.515625\MNW}} +% +\begin{treetab}{*9{c}} +\rlap{\hspace*{.25\wd0}\adjustA\makebox[0pt]{\N{}{CP}}}\End +\N{l}{Spec}&&\N{r}{C$'$}\End +\NTNo0{ml}&\N{lr}{C$^0$}&&\N{rr}{IP}\End +&\NNo1[6]{mlr}&\N{lrr}{DP}&&\NZ{rrr}{V}\adjustE\End +&&\NNo2{mlrr}&\N{lrrr}{V}&&\NZ{rrrr}{CP}\adjustD\End +&&&\NNo3{mlrrr}&\N{lrrrr}{Spec}&&\NZ{rrrrr}{C$'$}\adjustC\End +&&&&\NNo4{mlrrrr}&\N{lrrrrr}{C$^0$}&&\NZ{rrrrrr}{IP}\adjustB\End +&&&&&\NNo5{mlrrrrr}&\N{lrrrrrr}{DP}&&\NZ{rrrrrrr}{V}\adjustA\End +&&&&&&\NNo6{mlrrrrrr}&\N{lrrrrrrr}{PP}&\hspace{\wd1}\NZ{rrrrrrrr}{V}\End +&&&&&&&\NNo7{mlrrrrrrr}&\NodeTZNo8{mrrrrrrrr} +\end{treetab} + +We first define a number of shortcuts which will keep the table +managable: +\begin{verbatim} +\let\N=\Node +\let\NZ=\NodeZ +\let\NNo=\NodeNo +\let\NTNo=\NodeTNo +\let\MNW=\MinNodeWidth +\newcommand{\adjustA}{\hspace*{.75\MNW}} +\newcommand{\adjustB}{\hspace*{.625\MNW}} +\newcommand{\adjustC}{\hspace*{.5625\MNW}} +\newcommand{\adjustD}{\hspace*{.53125\MNW}} +\newcommand{\adjustE}{\hspace*{.515625\MNW}} +\newcommand{\End}{\\[4ex]} +\newcommand{\trivtab}[2]{\hbox{\begin{tabular}[t]{@{}c@{}}#1\\ + #2\end{tabular}}} +\newcommand{\twotab}[2]{\hbox{\begin{tabular}[t]{@{}ll@{}}#1\\ + #2\end{tabular}}} +\end{verbatim} +% +Next we put all leaves into boxes: +\begin{verbatim} +\tabcolsep2pt +\setbox0=\twotab{mit & wem$_i$}{with & who} +\setbox1=\trivtab{glaubst}{think} +\setbox2=\trivtab{du}{you} +\setbox3=\hbox{t$_j$} +\setbox4=\hbox{t$_i$} +\setbox5=\trivtab{dass}{that} +\setbox6=\trivtab{Maria}{Maria} +\setbox7=\hbox{t$_i$} +\setbox8=\twotab{gesprochen & hat}{spoken & has} +\end{verbatim} +And finially we specify the minimal node width and construct the table: +\begin{small}\begin{verbatim} +\NodeWidthNo1 +\begin{treetab}{*9{c}} +&\N{}{CP}\End +\N{l}{Spec}&&\N{r}{C$'$}\End +\NTNo0{ml}&\N{lr}{C$^0$}&&\N{rr}{IP}\End +&\NNo1[6]{mlr}&\N{lrr}{DP}&&\NZ{rrr}{V}\adjustE\End +&&\NNo2{mlrr}&\N{lrrr}{V}&&\NZ{rrrr}{CP}\adjustD\End +&&&\NNo3{mlrrr}&\N{lrrrr}{Spec}&&\NZ{rrrrr}{C$'$}\adjustC\End +&&&&\NNo4{mlrrrr}&\N{lrrrrr}{C$^0$}&&\NZ{rrrrrr}{IP}\adjustB\End +&&&&&\NNo5{mlrrrrr}&\N{lrrrrrr}{DP}&&\NZ{rrrrrrr}{V}\adjustA\End +&&&&&&\NNo6{mlrrrrrr}&\N{lrrrrrrr}{PP}&\hspace{\wd1}\NZ{rrrrrrrr}{V}\End +&&&&&&&\NNo7{mlrrrrrrr}&\NodeTZNo8{mrrrrrrrr} +\end{treetab} +\end{verbatim}\end{small} + +Note that box0 is wider than box1, therefore the root should not be in +the middle of box1 but should appear +exactly in the middle between box0 and box2. We therefore have to +replace the first line of the table by +\begin{verbatim} +\rlap{\hspace*{.25\wd0}\adjustA\makebox[0pt]{\N{}{CP}}}\End +\end{verbatim} + + +Here is another real life example from phonology: + +\setbox1=\hbox{mn} +\NodeWidthNo1 +\begin{treetab}[t]{*4c} +\nodeZ{l}{$\sigma$}&&\nodeZ{r}{$\sigma$}\End +\Node{ll}{V}&\Node{rl}{C}\nodeZ{xx}{=}&\Node{gr}{C}&\Node{lr}{V}\End +\Node{mll}{e}&\NodeZ{mrl}{b}&&\Node{mlr}{e} +\end{treetab} +\nodeconnect{gr}{mrl} + +The main point to be observed is that the first line cannot contain +\verb|\Node|'s (but must contain \verb|\node|'s), otherwise the macro +will complain that it cannot find the root of a tree: + +\begin{verbatim} + \setbox1=\hbox{mn} +\NodeWidthNo1 +\begin{treetab}[t]{*4c} +\nodeZ{l}{$\sigma$}&&\nodeZ{r}{$\sigma$}\End +\Node{ll}{V}&\Node{rl}{C}\nodeZ{xx}{=}&\Node{gr}{C}&\Node{lr}{V}\End +\Node{mll}{e}&\NodeZ{mrl}{b}&&\Node{mlr}{e} +\end{treetab} +\nodeconnect{gr}{mrl} + +\end{verbatim} +One final warning: It seems to me that \TeX\ internally uses certain +boxes to construct its tables. This means that the width of a box +might have changed by \TeX\ internally to zero before you come to the +point where you want to use the lenght prespecified by \verb|\setbox|. It +might therefore be necessary to store this lenght as soon as possible; +e.g. by saying \verb|\NodeWidthNo{n}| immediately after having defined +\verb|\setboxn|. +\end{document}
\ No newline at end of file |