\section{Introduction}

\tipa{}\footnote{\tipa{} stands for \emph{\TeX\ IPA} or \emph{Tokyo
    IPA}. The primary ftp site in which the latest version of \tipa{}
  is placed is \texttt{ftp://tooyoo.L.u-tokyo.ac.jp/pub/TeX/tipa}, and
  also it is mirrored onto the directory \texttt{fonts/tipa} of the
  CTAN archives.} is a system for processing IPA (International
Phonetic Alphabet) symbols in \LaTeX. It is based on
\tsipa{}\footnote{\tsipa{} was made in 1992 by Kobayashi Hajime, Fukui
  Rei and Shirakawa Shun. It is available from a CTAN archive.

  One problem with \tsipa{} was that symbols already included in
  \texttt{OT1}, \texttt{T1} or Math fonts are excluded, because of the
  limitation of its 128 character encoding. As a result, a string of
  phonetic representation had to be often composed of symbols from
  different fonts, disabling the possibility of automatic inter-word
  kerning.  And also too many symbols had to be realized as macros.}
but both \MF{} source codes and \LaTeX{} macros have been thoroughly
rewritten so that it can be considered as a new system.

Among many features of \tipa{}, the following are the new features 
as compared with \tsipa{} or any other existing systems for processing
IPA symbols.

\begin{itemize}
\itemsep0pt
\item A new 256 character encoding for phonetic symbols
  (`\texttt{T3}'), which includes all the symbols and diacritics found
  in the recent versions of IPA and some non-IPA symbols.
\item Complete support of \LaTeXe.
\item Roman, slanted, bold, bold extended and sans serif font styles.
\item Easy input method in the IPA environment.
\item Extended macros for accents and diacritics.\footnote{These
    macros are now defined in a separate file called
    `\texttt{exaccent.sty}' in order for the authors of other
    packages to be able to make use of them. The idea of separating
    these macros from other ones was suggested by Frank Mittelbach.}
\item A flexible system of macros for `tone letters'.
\item An optional package (\texttt{vowel.sty}) for drawing vowel
  diagrams.\footnote{This package (\texttt{vowel.sty}) can be used
    independently from the \tipa{} package. Documentation is also made
    separately in `\texttt{vowel.tex}' so that no further mention will be 
    made here.}
\item A slightly modified set of fonts that go well when used with
  Times Roman and Helvetica fonts.
\end{itemize}


\section{\tipa{} Encoding}

\subsection{Selection of symbols}\label{sec:selection}

The selection of \tipa{} phonetic symbols\footnote{In the case of
  \tsipa{}, the selection of symbols was based on ``Computer coding of
  the {IPA}: Supplementary Report'' \cite{Esling:IPAcoding}.} was made
based on the following works.

\begin{itemize}
\itemsep0pt
\item \textit{Phonetic Symbol Guide}~\cite{PSG} (henceforth abbreviated as
\PSG).
\item The official IPA charts of '49, '79, '89 and '93 versions.
\item \ifTUG\begin{sloppypar}\fi Recent articles published in the
    \emph{JIPA}\footnote{\emph{Journal of the International Phonetic
        Association.}}, such as ``Report on the 1989 Kiel Convention''
    \cite{IPA:KielConvention}, ``Further report on the 1989 Kiel
    Convention'' \cite{IPA:FurtherReport}, ``Computer Codes for
    Phonetic Symbols'' \cite{Esling:ComputerCodes}, ``Council actions
    on revisions of the IPA'' \cite{IPA:CouncilActions}, etc.
    \ifTUG\end{sloppypar}\fi
\item An unpublished paper by J.\ C.\ Wells: ``Com\-put\-er-coding the
  IPA: a proposed extension of SAM\-PA'' \cite{SAMPA}.
\item Popular textbooks on phonetics.
\end{itemize}

More specifically, \tipa{} contains all the symbols, including
diacritics, defined in the '79, '89 and '93 versions of IPA. And in
the case of the '49 version of IPA, which is described in the
\textit{Principles}~\cite{Principles}, there are too many obsolete
symbols and only those symbols that had had some popularity at least
for some time or for some group of people are included.

Besides IPA symbols, \tipa{} also contains symbols that are useful for
the following areas of phonetics and linguistics.

\begin{itemize}
\itemsep0pt
\item Symbols used in the American phonetics (e.g.\ \textturncelig,
  \textsce, \textscomega, \textlambda, etc.).
\item Symbols used in the historical study of Indo-European languages
  (e.g.\ \textthorn, \textwynn, \texthvlig, \textcommatailz,
  \textsoftsign, \texthardsign, and accents such as \textipa{\'=a},
  \textipa{\.'e}, etc.).
\item Symbols used in the phonetic description of languages in East Asia
  (e.g.\ \textlhti, \textvibyi, \textctd, \textctn, \textctt, etc.).
\item Diacritics used in `extIPA Symbols for Disordered
  Speech'~\cite{extIPA} and `VoQS (Voice Quality Symbols)'~\cite{VoQS}
  (e.g. \bibridge{n}, \subdoublevert{f}, \crtilde{m}, etc).
\end{itemize}

It should be also noted that \tipa{} includes all the necessary
elements of `tone letters', enabling all the theoretically possible
combinations of the tone letter system. In the recent publication of
the International Phonetic Association tone letters are admitted as an
official way of representing tones but the treatment of tone letters
is quite insufficient in that only a limited number of combination is
allowed. This is apparently due to the fact that there has been no
`portable' way of combining symbols that can be used across various
computer environments. Therefore \TeX's productive system of macro is
an ideal tool for handling a system like tone letters.

In the process of writing \MF{} source codes for \tipa{} phonetic
symbols there have been many problems besides the one with the
selection of symbols.  One of such problems was that sometimes the
exact shape of a symbol was unclear.  For example, the shapes of the
symbols such as \textstretchc\ (Stretched C), \textctj\ (Curly-tail J)
differ according to sources. This is partly due to the fact that the
IPA has been continuously revised for the past few decades, and partly
due to the fact that different ways of computerizing phonetic symbols
on different systems have resulted in the diversity of the shapes of
phonetic symbols.

Although there is no definite answer to such a problem yet, it seems
to me that it is a privilege of those working with \MF{} to have a
systematic way of controlling the shapes of phonetic symbols.

\subsection{Encoding}

\ifTUG \input{enc-tab1.tex}\fi

\ifTUG\else \input{enc-tab1.tex}\fi

The 256 character encoding of \tipa{} is now officially called the
`\texttt{T3}' encoding.\footnote{In a discussion with the \LaTeXe{}
  team it was suggested that the 128 character encoding used in WSUIPA
  would be refered to as the \texttt{OT3} encoding.} In deciding this
new encoding, care is taken to harmonize with existing other
encodings, especially with the \texttt{T1} encoding. Also the easiness
of inputting phonetic symbols is taken into consideration in such a
way that frequently used symbols can be input with small number of
keystrokes.

Table~\ref{tab:TIPAenc} shows the layout of the \texttt{T3} encoding.

The basic structure of the encoding found in the first half of the
table (character codes \texttt{'000-'177}) is based on normal text
encodings (ASCII, \texttt{OT1} and \texttt{T1}) in that sectioning of
this area into several groups such as the section for accents and
diacritics, the section for punctuation marks, the section for
numerals, the sections for uppercase and lowercase letters is
basically the same with these encodings.

Note also that the \texttt{T3} encoding contains not only phonetic
symbols but also usual punctuation marks that are used with phonetic
symbols, and in such cases the same codes are assigned as the normal
text encodings. However it is a matter of trade-off to decide which
punctuation marks are to be included. For example `:' and `;' might
have been preserved in \texttt{T3} but in this case `:' has been
traditionally used as a substitute for the length mark `\textipa{:}' so
that I decided to exclude `:' in favor of the easiness of inputting the
length mark by a single keystroke.

The encoding of the section for accents and diacritics is closely
related to \texttt{T1} in that the accents commonly included in
\texttt{T1} and \texttt{T3} have the same encoding.

\ifTUG \input{enc-tab2.tex}\fi

The sections for numerals and uppercase letters are filled with
phonetic symbols that are used frequently in many languages, because
numerals and uppercase letters are usually not used as phonetic
symbols. And the assignments made here are used as the `shortcut
characters', which will be explained in the section entitled
``Ordinary phonetic symbols'' (page~\pageref{sec:OrdinarySymbol}).

As for the section for uppercase letters in the usual text encoding, a
series of discussion among the members of the \texttt{ling-tex}
mailing list revealed that there seem to be a certain amount of
consensus on what symbols are to be assigned to each code. For example
they were almost unanimous for the assignments such as \textipa{A} for
\texttt{A}, \textipa{B} for \texttt{B}, \textipa{D} for \texttt{D},
\textipa{S} for \texttt{S}, \textipa{T} for \texttt{T}, etc. For more
details, see table~\ref{tab:shortcut}.

The encoding of the section for numerals was more difficult than the
above case. One of the possibilities was to assign symbols based on the
resemblance of shapes. One can easily think of assignments such as
\textipa{3} for \texttt{3} \texthtb{} for \texttt{6}, etc. But the
resemblance of shape alone does not serve as a criteria for all the
assignments. So I decided to assign basic vowel symbols to this
section.\footnote{This idea was influenced by the above mentioned
  article by J.\ C.\ Wells~\cite{SAMPA}.} Fortunately the resemblance
of shape is to some extent maintained as is shown in
table~\ref{tab:shortcut}.

The encoding of the section for lowercase letters poses no problem
since they are all used as phonetic symbols. Only one symbol, namely
`\textipa{g}', needs some attention because its shape should be
`\textipa{g}', rather than `g', as a phonetic symbol.\footnote{But the
  alternative shape `\textg' is preserved in other section and can be
  used as \texttt{\tbs textg}. }

\ifTUG\else \input{enc-tab2.tex}\fi

The second half of the table (character codes \texttt{'200-'377}) is
divided into four sections. The first section is devoted to the
elements of tone letters and other suprasegmental symbols.

Among the remaining three sections the last section \texttt{'340-'377}
contains more basic symbols than the other two sections. This is a
result of assigning the same character codes as latin-1 (ISO8859-1)
and T1 encodings to the symbols that are commonly included in \tipa{},
latin-1 and T1 encoded fonts.\footnote{This is based on a suggestion
  by J\"org Knappen.} These are the cases of \ae{}, \o, \oe{}, \c{c}
and \textthorn. And within each section symbols are arraneged largely
in alphabetical order.

For a table of the \texttt{T3} encoding, see Appendix~C
(page~\pageref{sec:FontLayout}).

\ifTUG\else
\section{Installation}

First, copy all the package and font description files
(\verb|tipa.sty|, \verb|tone.sty|, \verb|vowel.sty|, \verb|*.fd|,
\verb|*.def|) to a directory which \TeX{} programs can find. In an
ordinary Unix environment, it is recommended to create a directory
such as \verb|/usr/local/lib/texmf/tex/tipa| and copy all these files
into this directory.

\begin{verbatim}
  mkdir /usr/local/lib/texmf/tex/tipa
  cp tipa.sty tone.sty vowel.sty /usr/local/lib/texmf/tex/tipa
  cp *.fd *.def /usr/local/lib/texmf/tex/tipa
\end{verbatim}

Next, copy all the \texttt{TFM} files to a directory which \TeX{}
and device driver programs can find.

\begin{verbatim}
  mkdir /usr/local/lib/texmf/fonts/public/tipa
  mkdir /usr/local/lib/texmf/fonts/public/tipa/tfm
  cp *.tfm /usr/local/lib/texmf/fonts/public/tipa/tfm
\end{verbatim}

It is also recommended to copy all the \MF{} source files into an
appropriate directory.

\begin{verbatim}
  mkdir /usr/local/lib/texmf/fonts/public/tipa/src
  cp *.mf /usr/local/lib/texmf/fonts/public/tipa/src
\end{verbatim}

The final step of installation is that of \texttt{PK} font files. This
procedure may differ according to the user's computer environment.

For users unfamiliar with running \MF{} programs, it is possible to
get pre-compiled \texttt{PK} font files from ftp sites. In this
case, just copy them to an appropriate directory.

\begin{verbatim}
  mkdir /usr/local/lib/texmf/fonts/public/tipa/pk
  mkdir /usr/local/lib/texmf/fonts/public/tipa/pk/your_mode_name
  cp *pk /usr/local/lib/texmf/fonts/public/tipa/pk/your_mode_name
\end{verbatim}

If you make \texttt{PK} fonts by yourself, generate fonts shown in
table~\ref{tab:fullset} by the \MF{} program.

\begin{table}
\begin{center}
{\ttfamily
\begin{tabular}{|l|l|}
\hline
\textrm{\itshape Magnification} & \textrm{\itshape font name} \\
\hline
(mag=1)       & tipa8, tipa9, tipa10, tipa12, \\
              & tipasl8, tipasl9, tipasl10, tipasl12, \\
              & tipabx8, tipabx9, tipabx10, tipabx12, \\
              & tipass8, tipass9, tipass10, tipass12 \\
mag=5/8       & tipa8, tipasl8, tipabx8, tipass8 \\
mag=6/8       & tipa8, tipasl8, tipabx8, tipass8 \\
mag=7/8       & tipa8, tipasl8, tipabx8, tipass8 \\
mag=magstep.5 & tipa10, tipasl10, tipabx10, tipass10 \\
mag=magstep1  & tipa12, tipasl12, tipabx12, tipass12 \\
mag=magstep2  & tipa12, tipasl12, tipabx12, tipass12 \\
mag=magstep3  & tipa12, tipasl12, tipabx12, tipass12 \\
mag=magstep4  & tipa12, tipasl12, tipabx12, tipass12 \\
\end{tabular}}
\end{center}
\caption{Full set of \tipa{} fonts}\label{tab:fullset}
\end{table}

The bold font \texttt{tipab10} and fonts of \texttt{xipa} family are
not included in this table. Make \texttt{PK} files for these fonts
by yourself if you need them.

In the Unix environment, all these fonts can be automatically
generated by running the shell script \verb|makepk1.sh|. And a set of
\texttt{xipa} family fonts can be generated by running the shell
script \texttt{makepk2.sh}.

Also, if you are using \texttt{MakeTeXPK}, it is not necessary to
generate all these size variations at once. 

\fi % end of \ifTUG

\section{\tipa{} fonts}

This version of \tipa{} includes two families of IPA fonts,
\texttt{tipa} and \texttt{xipa}. The former family of fonts is for
normal use with \LaTeX, and the latter family is intended to be used
with `\texttt{times.sty}'(PSNFSS). They all have the same \texttt{T3}
encoding as explained in the previous section.

\begin{itemize}
\item \texttt{tipa}
  \begin{description}
  \item[Roman:]
    \texttt{tipa8}, \texttt{tipa9}, \texttt{tipa10}, \texttt{tipa12},
    \Sep \texttt{tipa17}
  \item[Slanted:]
    \texttt{tipasl8}, \texttt{tipasl9}, \texttt{tipasl10},
    \Sep \texttt{tipasl12}
  \item[Bold extended:]
    \texttt{tipabx8}, \texttt{tipabx9},
    \Sep \texttt{tipabx10}, \texttt{tipabx12}
  \item[Sans serif:]
    \texttt{tipass8}, \texttt{tipass9}, \texttt{tipass10}, \Sep
    \texttt{tipass12}, \texttt{tipass17}
  \item[Bold:]
    \texttt{tipab10}
  \end{description}
\item \texttt{xipa}
  \begin{description}
  \item[Roman:] \texttt{xipa10}
  \item[Slanted:] \texttt{xipasl10}
  \item[Bold:] \texttt{xipab10}
  \item[Sans serif:] \texttt{xipass10}
  \end{description}
\end{itemize}

All these fonts are made by \MF{}, based on the Computer Modern font
series. In the case of the \texttt{xipa} series, parameters are
adjusted so as to look fine when used with Times Roman (in the cases
of \texttt{xipa10}, \texttt{xipasl10}, \texttt{xipab10}) and Helvetica
(in the case of \texttt{xipass10}).


\section{Usage}

\subsection{Declaration of \tipa{} package}

In order to use \tipa{}, first declare \tipa{} package at the preamble 
of a document.

\begin{verbatim}
  \documentclass{article}
  \usepackage{tipa}
\end{verbatim}

\subsection{Encoding options}

The above declaration uses \texttt{OT1} as the default text encoding. If 
you want to use \tipa{} symbols with \texttt{T1}, specify the option
`\texttt{T1}'.

\begin{verbatim}
  \documentclass{article}
  \usepackage[T1]{tipa}
\end{verbatim}

If you want to use a more complex form of encoding, declare the use of
\texttt{fontenc} package by yourself and specify the option
`\texttt{noenc}'. In this case the option `\texttt{T3}', which represents
the \tipa{} encoding, must be included as an option to the
\texttt{fontenc} package. For example, if you want to use \tipa{} and
the University Washington Cyrillic (\texttt{OT2}) with the \texttt{T1}
text encoding, the following command will do this.

\begin{verbatim}
  \documentclass{article}
  \usepackage[T3,OT2,T1]{fontenc}
  \usepackage[noenc]{tipa}
\end{verbatim}

By default, \tipa{} includes the \texttt{fontenc} package internally but 
the option \texttt{noenc} suppresses this.

\subsection{Using \tipa{} with PSNFSS}

In order to use \tipa{} with \texttt{times.sty}, declare the use of
\texttt{times.sty} before declaring \texttt{tipa} packages.

\begin{verbatim}
  \documentclass{article}
  \usepackage{times}
  \usepackage{tipa}
\end{verbatim}

Font description files \texttt{T3ptm.fd} and \texttt{T3phv.fd} 
are automatically loaded by the above declaration.

\subsection{Other options}\label{sec:otheroptions}

\tipa{} can be extended by the options \texttt{tone}, \texttt{extra}.

If you want to use the optional package for `tone letters', add
`\texttt{tone}' option to the \verb|\usepackage| command that declares
\texttt{tipa} package.

\begin{verbatim}
  \usepackage[tone]{tipa}
\end{verbatim}

And if you want to use diacritics for extIPA and Vo\-QS, specify
`\texttt{extra}' option.

\begin{verbatim}
  \usepackage[extra]{tipa}
\end{verbatim}

Finally there is one more option called `\texttt{safe}', which is used
to suppress definitions of some possibly `dangerous' commands of
\tipa.

\begin{verbatim}
  \usepackage[safe]{tipa}
\end{verbatim}

More specifically, the following commands are suppressed by declaring
the \texttt{safe} option.  Explanation on the function of each command
will be given later.

\begin{itemize}
\itemsep0pt
\item \verb|\s| (equivalent to \verb|\textsyllabic|)
\item \verb|\*| (already defined in plain \TeX)
\item \verb+\|+, \verb|\:|, \verb|\;|, \verb|\!| (already defined in \LaTeX)
\end{itemize}

\section{Input Commands for Phonetic Symbols}
\subsection{Ordinary phonetic symbols}\label{sec:OrdinarySymbol}

\tipa{} phonetic symbols can be input by the following two ways.

\begin{enumerate}
\itemsep0pt
\item Input macro names in the normal text environment.
\item Input macro names or \emph{shortcut characters} with\-in the
  follwoing groups or environment.
  \begin{itemize}
  \item \verb|\textipa{...}|\footnote{I personally prefer a slightly
      shorter name like \texttt{\tbs ipa} rather than \texttt{\tbs
        textipa} but this command was named after the general
      convention of \LaTeXe. The same can be said to all the symbol
      names beginning with \texttt{\tbs text}.}
  \item \verb|{\tipaencoding ...}|
  \item \verb|\begin{IPA} ... \end{IPA}|
  \end{itemize}
  (These groups and environment will be henceforth refered to as the
  \textsl{IPA environment}.)
\end{enumerate}

A shortcut character refers to a single character that is assigned to
a specific phonetic symbol and that can be directly input by an
ordinary keyboard. In \tipa{} fonts, the character codes for numerals
and uppercase letters in the normal ASCII encoding are assigned to such
shortcut characters, because numerals and uppercase letters are
usually not used as phonetic symbols.  And additional shortcut
characters for symbols such as \ae{}, \oe{}, \o{} may also be used if
you are using a T1 encoded font and an appropriate input system for
it.

The following pair of examples show the same phonetic transcription of 
a English word that are input by the above mentioned two input methods.

\begin{tipaexample}
  \xitem
\ifTUG
    \verb|[\textsecstress\textepsilon kspl|\\
    \verb|\textschwa\textprimstress ne|\\
    \verb|\textsci\textesh\textschwa n]|
\else
    \verb|[\textsecstress\textepsilon kspl\textschwa|\\
    \verb|\textprimstress ne\textsci\textesh\textschwa n]|
\fi
  \xitem {[\textsecstress\textepsilon kspl\textschwa\textprimstress
    ne\textsci\textesh\textschwa n]}
  \xitem \verb|\textipa{[""Ekspl@"neIS@n]}|
  \xitem \textipa{[""Ekspl@"neIS@n]}
\end{tipaexample}

It is apparent that inputting in the IPA environment is far easier
than in the normal text environment. Moreover, although the outputs of
the above examples look almost the same, they are \textsl{not\/}
identical, exactly speaking. This is because in the IPA environment
automatic kerning between symbols is enabled, as is illustrated by
the following pair of examples.

\begin{tipaexample}
  \xitem \verb|v\textturnv v w\textsca w |\Sep
         \verb|y\textturny y [\textesh]|
  \xitem v\textturnv v w\textsca w y\textturny y [\textesh]
  \xitem \verb|\textipa{v2v w\textsca w yLy [S]}|
  \xitem \textipa{v2v w\textsca w yLy [S]}
\end{tipaexample}

Table~\ref{tab:shortcut} shows most of the shortcut characters together
with the corresponding characters in the ASCII encoding.

\subsection{Naming of phonetic symbols}

Every \tipa{} phonetic symbol has a unique symbol name, such as
\textsl{Turned A, Hooktop B, Schwa}.\footnote{The naming was made
  based on the literature listed in the section entitled ``Selection
  of Symbols'' (page~\pageref{sec:selection}).  And users of \tsipa{}
  should be careful because \tipa{}'s naming is slightly modified from
  that of \tsipa{}.}%
Also each symbol has a corresponding control sequence name, such as
\verb|\textturna|, \verb|\texthtb|, \verb|\textschwa|. The name used
as a control sequence is usually an abbreviated form of the
corresponding symbol name with a prefix \verb|\text|. The conventions
used in the abbreviation are as follows.

\begin{table}
\begin{center}
{\ifTUG\tabcolsep.4em\else\tabcolsep1em\fi
\begin{tabular}{llc}
\hline
  \emph{Symbol name}  & \emph{Macro name}  & \emph{Symbol}\\
\hline
  Turned A            & \verb|\textturna|      & \textturna  \\
  Glottal Stop        & \verb|\textglotstop|   & \textglotstop \\
  Right-tail D        & \verb|\textrtaild|     & \textrtaild \\
  Small Capital G     & \verb|\textscg|        & \textscg \\
  Hooktop B           & \verb|\texthtb|        & \texthtb \\
  Curly-tail C        & \verb|\textctc|        & \textctc \\
  Crossed H           & \verb|\textcrh|        & \textcrh \\
  Old L-Yogh Ligature & \verb|\textOlyoghlig|  & \textOlyoghlig \\
  Beta                & \verb|\textbeta|       & \textbeta \\
\hline
\end{tabular}}
\end{center}
\caption{Naming of \tipa{} symbols}\label{tab:naming}
\end{table}

\begin{itemize}
\itemsep0pt
\item Suffixes and endings such as `-ive', `-al', `-ed' are omitted.
\item `right', `left' are abbreviated to \texttt{r}, \texttt{l} respectively.
\item For `small capital' symbols, prefix \texttt{sc} is added.
\item A symbol with a hooktop is abbreviated as \texttt{ht}...
\item A symbol with a curly-tail is abbreviated as \texttt{ct}...
\item A `crossed' symbol is abbreviated as \texttt{cr}...
\item A ligature is abbreviated as ...\texttt{lig}.
\item For an old version of a symbol, prefix \texttt{O} is added.
\end{itemize}

Note that the prefix \texttt{O} (old) should be given in uppercase
letter.

Table~\ref{tab:naming} shows some examples of correspondence between
symbol names and control sequence names.

\subsection{Ligatures}

Just like the symbols such as ``, '', --, ---, fi, ff are realized as
ligatures by inputting \verb|``|, \verb|''|, \verb|--|, \verb|---|,
\verb|fi|, \verb|ff| in \TeX{}, two of the \tipa{} symbols, namely
\textsl{Secondary Stress} and \textsl{Double Pipe}, and double
quotation marks\footnote{Although \tipa{} fonts do not include the
  symbols `` and '', a negative value of kerning is automatically
  inserted between ` and `, ' and ', so that the same results can be
  obtained as in the case of the normal text font.} can be input as
ligatures in the IPA environment.

\begin{tipaexample}
  \yitem \verb+\textipa{" "" | || `` ''}+
  \yitem \textipa{" "" | || `` ''}
\end{tipaexample}

\subsection{Special macros {\tt\tbs*}, {\tt\tbs;}, {\tt\tbs:} and
  {\tt\tbs!}}

\tipa{} defines {\tt\tbs*}, {\tt\tbs;}, {\tt\tbs:} and {\tt\tbs!} as
special macros in order to easily input phonetic symbols that do not
have a shortcut character explained above.  Before explaining how to
use these macros, it is necessary to note that these macros are
primarily intended to be used by linguists who usually do not care
about things in math mode. And they can be `dangerous' in that they
override existing \LaTeX{} commands used in the math mode. So if you
want to preserve the original meaning of these commands, daclare the
option `\texttt{safe}' at the preamble.

The macro \verb|\*| is used in three different ways.  First, when this
macro is followed by one of the letters f, k, r, t or w, it results in
a turned symbol.\footnote{This idea was pointed out by J\"org
  Knappen.}

\begin{tipaexample}
  \yitem \verb|\textipa{\*f \*k \*r \*t \*w}|
  \yitem \textipa{\*f \*k \*r \*t \*w}
\end{tipaexample}

Secondly, when this macro is followed by one of the letters j, n, h,
l or z, it results in a frequently used symbol that has otherwise no
easy way to input.

\begin{tipaexample}
  \yitem \verb|\textipa{\*j \*n \*h \*l \*z}|
  \yitem \textipa{\*j \*n \*h \*l \*z}
\end{tipaexample}

Thirdly, when this macro is followed by letters other than the above
cases, they are turned into the symbols of the default text font. This
is useful in the IPA environment to select symbols temporarily from
the normal text font.

\begin{tipaexample}
  \yitem \verb|\textipa{\*A dOg, \*B k\ae{}t, |\Sep
    \verb|ma\super{\*{214}}}|
  \yitem \textipa{\*A dOg, \*B k\ae{}t, ma\super{\*{214}}}
\end{tipaexample}

The remaining macros {\tt\tbs;}, {\tt\tbs:} and {\tt\tbs!} are
used to make small capital symbols, retroflex symbols, and implosives
or clicks, respectively.

\begin{tipaexample}
  \yitem \verb|\textipa{\;B \;E \;A \;H \;L \;R}|
  \yitem \textipa{\;B \;E \;A \;H \;L \;R}
  \yitem \verb|\textipa{\:d \:l \:n \:r \:s \:z}|
  \yitem \textipa{\:d \:l \:n \:r \:s \:z}
  \yitem \verb|\textipa{\!b \!d \!g \!j \!G \!o}|
  \yitem \textipa{\!b \!d \!g \!j \!G \!o}
\end{tipaexample}

\subsection{Punctuation marks}

The following punctuation marks and text symbols that are
normally included in the text encoding are also included in the
\texttt{T3} encoding so that they can be directly input in the IPA
environment.

\begin{tipaexample}
 \yitem \verb|\textipa{! ' ( ) * + , - .\ / = ?|\\
        \verb|  [ ] `}|
 \yitem \textipa{! ' ( ) * + , - .\ / = ? [ ] `}
\end{tipaexample}

All the other punctuation marks and text symbols that are not included
in \texttt{T3} need to be input with a prefix \verb|\*| explained in
the last section when they appear in the IPA environment.

\begin{tipaexample}
 \yitem
\ifTUG
   \verb|\textipa{\*; \*: \*@ \*\# \*\$|\Sep
   \verb|  \*\& \*\% \*\{ \*\}}| 
\else
   \verb|\textipa{\*; \*: \*@ \*\# \*\$ \*\& \*\% \*\{ \*\}}|
\fi
 \yitem \textipa{\*; \*: \*@ \*\# \*\$ \*\& \*\% \*\{ \*\}}
\end{tipaexample}

\subsection{Accents and diacritics}

Table~\ref{tab:accent} shows how to input accents and diacritics in
\tipa{} with some examples. Here again, there are two kinds of input
methods; one for the normal text environment, and the other for the
IPA environment.

\begin{table}
\begin{center}
{\ifTUG\tabcolsep.4em\else\tabcolsep1em\fi
\def\TblShrt#1{\hbox to 2em{#1\hss}}
\def\TblMvRt#1{\kern 2em #1}
\begin{tabular}{llc}
\hline
\textit{Input in the normal}& \textit{Input in the IPA} & \textit{Output} \\
\textit{text environment}   & \textit{environment}      & \\
\hline
\TblShrt{\Tt{'a}}               & \TblMvRt{\Tt{'a}}        &\textipa{\'a} \\
\TblShrt{\Tt{"a}}               & \TblMvRt{\Tt{"a}}        &\textipa{\"a} \\
\TblShrt{\Tt{~a}}               & \TblMvRt{\Tt{\ttilde a}} &\textipa{\~a} \\
\TblShrt{\Td{r}{a}}             & \TblMvRt{\Td{r}{a}}      &\textipa{\r{a}}\\
\TblShrt{\Td{textsyllabic}{m}}  & \TblMvRt{\Td{s}{m}}      &\textipa{\s{m}}\\
\TblShrt{\Td{textsubumlaut}{a}} & \TblMvRt{\Tt{"*a}}       &\textipa{\"*a}\\
\TblShrt{\Td{textsubtilde}{a}}  & \TblMvRt{\Tt{\ttilde*a}} &\textipa{\~*a}\\
\TblShrt{\Td{textsubring}{a}}   & \TblMvRt{\Tt{r*a}}       &\textipa{\r*a}\\
\TblShrt{\Td{textdotacute}{e}}  & \TblMvRt{\Tt{.'e}}       &\textipa{\.'e}\\
\TblShrt{\Td{textgravedot}{e}}  & \TblMvRt{\Tt{`.e}}       &\textipa{\`.e}\\
\TblShrt{\Td{textacutemacron}{a}}& \TblMvRt{\Tt{'=a}}      &\textipa{\'=a}\\
\TblShrt{\Td{textcircumdot}{a}} & \TblMvRt{\Tt{\tcircum.a}}&\textipa{\^.a}\\
\TblShrt{\Td{texttildedot}{a}}  & \TblMvRt{\Tt{\ttilde.a}} &\textipa{\~.a}\\
\TblShrt{\Td{textbrevemacron}{a}}& \TblMvRt{\Tt{u=a}}      &\textipa{\u=a}\\
%\Td{}{a}                    & \Tt{}              &\textipa{}\\
\hline
\end{tabular}}
\end{center}
\caption{Examples of inputting accents}\label{tab:accent}
\end{table}

In the IPA environment, most of the accents and diacritics can be
input more easily than in the normal text environment, especially in
the cases of subscript symbols that are normally placed over a symbol
and in the cases of combined accents, as shown in the table.

As can be seen by the above examples, most of the accents that are
normally placed over a symbol can be placed under a symbol by adding
an \texttt{*} to the corresponding accent command in the IPA environment.

The advantage of IPA environment is further exemplified by the
all-purpose accent \verb+\|+, which is used as a macro prefix to
provide shortcut inputs for the diacritics that otherwise have to be
input by lengthy macro names. Table~\ref{tab:accprefix} shows examples
of such accents. Note that the macro \verb+\|+ is also `dangerous' in
that it has been already defined as a math symbol of \LaTeX{}. So if
you want to preserve the original meaning of this macro, declare
`\texttt{safe}' option at the preamble.

\begin{table}
\begin{center}
{\ifTUG\tabcolsep.4em\else\tabcolsep1em\fi
\def\TblShrt#1{\hbox to 2em{#1\hss}}
\def\TblMvRt#1{\kern 2em #1}
\begin{tabular}{llc}
\hline
\textit{Input in the normal}& \textit{Input in the IPA} & \textit{Output} \\
\textit{text environment}   & \textit{environment}      & \\
\hline
\TblShrt{\Td{textsubbridge}{t}}   & \TblMvRt{\Tt{|[t}}   & \textipa{\|[t}\\
\TblShrt{\Td{textinvsubbridge}{t}}& \TblMvRt{\Tt{|]t}}   & \textipa{\|]t}\\
\TblShrt{\Td{textsublhalfring}{a}}& \TblMvRt{\Tt{|(a}}   & \textipa{\|(a}\\
\TblShrt{\Td{textsubrhalfring}{a}}& \TblMvRt{\Tt{|)a}}   & \textipa{\|)a}\\
\TblShrt{\Td{textroundcap}{k}}    & \TblMvRt{\Td{|c}{k}} & \textipa{\|c{k}}\\
\TblShrt{\Td{textsubplus}{o}}     & \TblMvRt{\Tt{|+o}}   & \textipa{\|+o}\\
\TblShrt{\Td{textraising}{e}}     & \TblMvRt{\Tt{|'e}}   & \textipa{\|'e}\\
\TblShrt{\Td{textlowering}{e}}    & \TblMvRt{\Tt{|`e}}   & \textipa{\|`e}\\
\TblShrt{\Td{textadvancing}{o}}   & \TblMvRt{\Tt{|<o}}   & \textipa{\|<o}\\
\TblShrt{\Td{textretracting}{a}}  & \TblMvRt{\Tt{|>a}}   & \textipa{\|>a}\\
\TblShrt{\Td{textovercross}{e}}   & \TblMvRt{\Td{|x}{e}} & \textipa{\|x{e}}\\
\TblShrt{\Td{textsubw}{k}}        & \TblMvRt{\Td{|w}{k}} & \textipa{\|w{k}}\\
\TblShrt{\Td{textseagull}{t}}     & \TblMvRt{\Td{|m}{t}} & \textipa{\|m{t}}\\
\hline
\end{tabular}}
\end{center}
\caption{Examples of the accent prefix {\tt \tbs|}}\label{tab:accprefix}
\end{table}

Finally, examples of words with complex accents that are input in the
IPA environment are shown below.

\begin{tipaexample}
  \yitem \verb+\textipa{*\|c{k}\r*mt\'om +\Sep
    \verb+*bhr\'=at\=er}+
  \yitem \textipa{*\|c{k}\r*mt\'om *bhr\'=at\=er}
\end{tipaexample}

For a full list of accents and diacritics, see Appendix~A 

\subsection{Superscript symbols}

In the normal text environment, superscript symbols can be input by a
macro called \verb|\textsuperscript|, which has been newly introduced
in the recent version of \LaTeXe. This macro takes one argument which
can be either a symbol or a string of symbols, and can be nested.

Since the name of this macro is too long, \tipa{} provides an
abbreviated form of this macro called \verb|\super|.

\begin{tipaexample}
  \xitem
    \verb|t\textsuperscript h |\Sep
    \verb|k\textsuperscript w|\\
    \verb|a\textsuperscript{bc}|\\
\ifTUG
    \verb|a\textsuperscript{b%|\Sep
    \verb| \textsuperscript{c}}|
\else
    \verb|a\textsuperscript{b\textsuperscript{c}}|
\fi
  \xitem
    t\textsuperscript h k\textsuperscript w a\textsuperscript{bc} 
      a\textsuperscript{b\textsuperscript{c}}
  \xitem
    \verb|\textipa{t\super{h} k\super{w}|\\
    \verb|  a\super{bc} a\super{b\super{c}}}|
  \xitem
    \textipa{\textipa{t\super{h} k\super{w} a\super{bc}
      a\super{b\super{c}}}}
\end{tipaexample}

These macros automatically select the correct size of superscript font 
no matter what size of the text font is used.

\subsection{Tone letters}\label{sec:tone}

\tipa{} provides a flexible system of macros for `tone letters'. A
tone letter is represented by a macro called `\verb|\tone|', which
takes one argument consisting of a string of numbers ranging from 1 to
5. These numbers denote pitch levels, 1 being the lowest and 5 the
highest. Within this range, any combination is allowed and there is no
limit in the length of combination.

As an example of the usage of tone letter macro, the four tones of
Chinese are show below.

\begin{tipaexample}
  \yitem
    \verb|\tone{55}ma ``mother'', |\Sep
    \verb|\tone{35}ma ``hemp'',|\\
    \verb|\tone{214}ma ``horse'', |\Sep
    \verb|\tone{51}ma ``scold''|
  \yitem
    \tone{55}ma ``mother'', \tone{35}ma ``hemp'', \Sep
    \tone{214}ma ``horse'', \tone{51}ma ``scold''
\end{tipaexample}

\section{How easy to input phonetic symbols?}

Let us briefly estimate here how much easy (or difficult) to input
phonetic symbols with \tipa{} in terms of the number of keystrokes.

The following table shows statistics for all the phonetic symbols that
appear in the '93 version of IPA chart (diacritics and symbols for
su\-pra\-seg\-men\-tals excluded). It is assumed here that each symbol
is input within the IPA environment and the \texttt{safe} option is
not specified.

\begin{center}
{\ifTUG \tabcolsep.4em\fi
\begin{tabular}{c|c|l}
\emph{keystrokes} & \emph{number} & \emph{examples} \\
\hline
1            & 65 & \textipa{a, b, @, A, B, etc.} \\
2            & 2  & \textipa{\o, ||} \\
3            & 30 & \textipa{\ae, \:t, \;B, \!b, etc.} \\
5            & 1  & \textipa{\c{c}} \\
more than 5  & 7  & \textipa{\textcloseepsilon, \textbarglotstop,
  \textdoublebarpipe, \textturnmrleg, etc.}
\end{tabular}}
\end{center}

As is shown in the table, about 92\% of the symbols can be input
within three keystrokes.

\section{Changing font styles}

\begin{table}
\begin{center}
\ifTUG
{\tabcolsep.4em
\begin{tabular}{lll}
\hline
\emph{Roman} & \\
\verb|\textipa{f@"nEtIks}|           &\textipa{f@"nEtIks} \\
\hline
\emph{Slanted} & \\
\verb|\textipa{\slshape f@"nEtIks}|  &\textipa{\slshape f@"nEtIks} or\\ 
\verb|\textipa{\textsl{f@"nEtIks}|   &\textipa{\textsl{f@"nEtIks}} or\\ 
\verb|\textsl{\textipa{f@"nEtIks}|   &\textsl{\textipa{f@"nEtIks}}\\ 
\hline
\emph{Bold extended} & \\
\verb|\textipa{\bfseries f@"nEtIks}| &\textipa{\bfseries f@"nEtIks} or\\
\verb|\textipa{\textbf{f@"nEtIks}|   &\textipa{\textbf{f@"nEtIks}} or\\
\verb|\textbf{\textipa{f@"nEtIks}|   &\textbf{\textipa{f@"nEtIks}}\\
\hline
\emph{Sans Serif} & \\
\verb|\textipa{\sffamily f@"nEtIks}| &\textipa{\sffamily f@"nEtIks} or\\ 
\verb|\textipa{\textsf{f@"nEtIks}|   &\textipa{\textsf{f@"nEtIks}} or\\ 
\verb|\textsf{\textipa{f@"nEtIks}|   &\textsf{\textipa{f@"nEtIks}}\\ 
\hline
\end{tabular}}
\else
\begin{tabular}{@{\hspace{0pt}}lll}
\emph{Roman}&\verb|\textipa{f@"nEtIks}|        &\textipa{f@"nEtIks} \\
\emph{Slanted}&\verb|\textipa{\slshape f@"nEtIks}|&\textipa{\slshape f@"nEtIks}\\ 
or          &\verb|\textipa{\textsl{f@"nEtIks}|&\textipa{\textsl{f@"nEtIks}}\\ 
or          &\verb|\textsl{\textipa{f@"nEtIks}|&\textsl{\textipa{f@"nEtIks}}\\ 
\emph{Bold extended} & \verb|\textipa{\bfseries f@"nEtIks}| &
  \textipa{\bfseries f@"nEtIks}\\
or          &\verb|\textipa{\textbf{f@"nEtIks}|&\textipa{\textbf{f@"nEtIks}}\\
or          &\verb|\textbf{\textipa{f@"nEtIks}|&\textbf{\textipa{f@"nEtIks}}\\
\emph{Sans serif}& \verb|\textipa{\sffamily f@"nEtIks}| &
  \textipa{\sffamily f@"nEtIks}\\ 
or          &\verb|\textipa{\textsf{f@"nEtIks}|&\textipa{\textsf{f@"nEtIks}}\\ 
or          &\verb|\textsf{\textipa{f@"nEtIks}|&\textsf{\textipa{f@"nEtIks}}\\ 
\end{tabular}
\fi
\end{center}
\caption{Examples of font switching}\label{tab:fontswitch}
\end{table}

This version of \tipa{} includes five styles of fonts, i.e.\ roman,
slanted, bold, bold extended and sans serif. These styles can be
switched in much the same way as in the normal text fonts (see
table~\ref{tab:fontswitch}).

The bold fonts are usually not used within the standard \LaTeX{} class 
packages so that if you want to use them, it is necessary to use
low-level font selection commands of \LaTeXe.

\begin{tipaexample}
  \yitem \verb|{\fontseries{b}\selectfont| \Sep
    \verb|abcdefg \textipa{ABCDEFG}}|
  \yitem {\fontseries{b}\selectfont abcdefg \textipa{ABCDEFG}}
\end{tipaexample}

Note also that slanting of \tipa{} symbols should correctly work even
in the cases of combined accents and in the cases of symbols made up
by macros.

\begin{tipaexample}
  \yitem \verb|\textsl{\textipa{\'{\"{\u*{e}}}}}|
  \yitem \textsl{\textipa{\'{\"{\u*{e}}}}}
  \yitem \verb|\textsl{\textdoublebaresh}|
  \yitem \textsl{\textdoublebaresh\/} {\small(This symbol is
    composed by a macro.)}
\end{tipaexample}

\section{Internal commands}

Some of the internal commands of \tipa{} are defined without the
letter \texttt{@} in order to allow a user to extend the capability of 
\tipa{}.

\subsection{{\tt\tbs ipabar}}

Some \tipa{} symbols such as \verb|\textbarb| \textbarb,
\verb|\textcrtwo| \textcrtwo{} are defined by using an internal macro
command \verb|\ipabar|. This command is useful when you want to make
barred or crossed symbols not defined in \tipa{}.

This command requires the following five parameters to control the
position of the bar.

\begin{itemize}
\itemsep0pt
\item {\tt\#1} the symbol to be barred
\item {\tt\#2} the height of the bar (in dimen)
\item {\tt\#3} bar width
\item {\tt\#4} left kern added to the bar
\item {\tt\#5} right kern added to the bar
\end{itemize}

Parameters \texttt{\#3}, \texttt{\#4}, \texttt{\#5} are to be given in
a scaling factor to the width of the symbol, which is equal to 1 if
the bar has the same width with the symbol in question. For example,
the following command states a barred b (\textbarb) of which the bar
position in the y-coordinate is \verb|.5ex| and the width of the bar
is slightly larger than that of the letter b.

\ifTUG
\begin{verbatim}
  % Barred B
  \newcommand\textbarb{%
    \ipabar{{\tipaencoding b}}%
      {.5ex}{1.1}{}{}}
\end{verbatim}
\else
\begin{verbatim}
  % Barred B
  \newcommand\textbarb{%
    \ipabar{{\tipaencoding b}}{.5ex}{1.1}{}{}}
\end{verbatim}
\fi

Note that the parameters \texttt{\#4} and \texttt{\#5} can be left
blank if the value is equal to 0.

And the next example declares a barred c (\textbarc) of which the bar
width is a little more than half as large as the letter c and it has
the same size of kerning at the right.

\ifTUG
\begin{verbatim}
  % Barred C
  \newcommand\textbarc{%
    \ipabar{{\tipaencoding c}}%
      {.5ex}{.55}{}{.55}}
\end{verbatim}
\else
\begin{verbatim}
  % Barred C
  \newcommand\textbarc{%
    \ipabar{{\tipaencoding c}}{.5ex}{.55}{}{.55}}
\end{verbatim}
\fi

More complex examples with the \verb|\ipabar| command are found in
\texttt{T3enc.def}.

\subsection{{\tt\tbs tipaloweraccent}, {\tt\tbs tipaupperaccent}}

These two commands are used in the definitions of \tipa{} accents and
diacritics. They are special forms of the commands \verb|\loweraccent|
and \verb|\upperaccent| that are defined in \texttt{exaccent.sty}. The
difference between the commands with the prefix \verb|tipa| and the
ones without it is that the former commands select accents from a T3
encoded font while the latter ones do so from the current text font.

These commands take two parameters, the code of the accent (in
decimal, octal or hexadecimal number) and the
symbol to be accented, as shown below.

\begin{tipaexample}
  \yitem \verb|\tipaupperaccent{0}{a}|
  \yitem \tipaupperaccent{0}{a}
\end{tipaexample}

Optionally, these commands can take a extra parameter to adjust the
vertical position of the accent. Such an adjustment is sometimes
necessary in the definition of a nested accent. The next example shows
\tipa's definition of the `Circumflex Dot Accent' (e.g.\ 
\textcircumdot{a}).

\ifTUG
\begin{verbatim}
  % Circumflex Dot Accent
  \newcommand\textcircumdot[1]%
    {\tipaupperaccent[-.2ex]{2}%
    {\tipaupperaccent[-.1ex]{10}{#1}}}
\end{verbatim}
\else
\begin{verbatim}
  % Circumflex Dot Accent
  \newcommand\textcircumdot[1]{\tipaupperaccent[-.2ex]{2}%
    {\tipaupperaccent[-.1ex]{10}{#1}}}
\end{verbatim}
\fi

This definition states that a dot accent is placed over a symbol
thereby reducing the vertical distance between the symbol and the dot
by \texttt{.1ex} and a circumflex accent is placed over the dot and
the distance between the two accents is reduced by \texttt{.2ex}.

If you want to make a combined accent not included in \tipa{}, you can
do so fairly easily by using these two commands together with the
optional parameter. For more examples of these commands, see
\verb|tipa.sty| and \verb|extraipa.sty|.

\subsection{{\tt\tbs tipaLoweraccent}, {\tt\tbs tipaUpperaccent}}

These two commands differ from the two commands explaind above in that
the first parameter should be a symbol (or any other things, typically
an \verb|\hbox|), rather than the code of the accent. They are special
cases of the commands \verb|\Loweraccent| and \verb|\Upperaccent| and
the difference between the two pairs of commands is the same as before.

The next example makes a schwa an accent.

\begin{tipaexample}
  \yitem
\ifTUG
    \verb|\tipaUpperaccent[.2ex]%|\\
    \verb|  {\lower.8ex\hbox{%|\\
    \verb|    \textipa{\super@}}}{a}|
\else
    \verb|\tipaUpperaccent[.2ex]%|\\
    \verb|  {\lower.8ex\hbox{\textipa{\super@}}}{a}|
\fi
  \yitem
    \tipaUpperaccent[.2ex]{\lower.8ex\hbox{\textipa{\super@}}}{a}
\end{tipaexample}

\section{Acknowledgments}

First of all, many thanks are due to the co-authors of \tsipa{},
Kobayashi Hajime and Shirakawa Shun. Ko\-ba\-yashi Hajime was the main
font designer of \tsipa{}. Shirakawa Shun worked very hard in deciding
encoding, checking the shapes of symbols and writing the Japanese
version of document. \tipa{} was impossible without \tsipa{}.

I would like to thank also J\"org Knappen whose insightful comments
helped greatly in many ways the development of \tipa{}. I was also
helped and encouraged by Christina Thiele, Martin Haase, Kirk Sullivan
and many other members of the \texttt{ling-tex} mailing list.

At the last stage of the development of \tipa{} Frank Mittelbach gave
me precious comments on how to incorporate various \tipa{} commands
into the NFSS. I would like to thank also Barbara Beeton who kindly
read over the preliminary draft of this document and gave me useful
comments.

\ifTUG\else
 \bibliographystyle{plain}
\fi
\bibliography{tipa}

\endinput