path: root/fonts/junicode/doc/Record_Interpreter_notes.tex

\chapter{Transcribing records}
%\fancyhead[CE]{\scshape\color{myRed} {\addfontfeatures{Numbers=OldStyle}\thepage}\hspace{10pt}transcribing records}

This chapter provides guidance for persons transcribing records (laws and other
public documents) in the style of Charles Trice Martin’s
\textit{The Record Interpreter}, \textit{Statutes of the Realm}, and similar guides
and editions.
Unlike most editions of early texts, these retain (or recommend retaining) the capitalization,
punctuation, and abbreviations of their manuscript sources.

\section{A preliminary note on transcription}
Here are a few observations, based on a long career as a scholarly editor of medieval and eighteenth-century texts.

Before embarking on the task of transcribing an old document, ask yourself what value you want to add to the document as
it already exists, because different kinds of transcription add different kinds of value. The kind of transcription
that adds the least is that which aims at the exact \textit{visual} reproduction of a document. A transcript is not a
facsimile: it needs to do something that a photograph can't do.

Converting a document from visual image to Unicode-encoded text adds a good bit of value all by itself, but only if done
with due regard for the semantics of Unicode characters. Every Unicode character has a meaning, and that meaning is a
help to readers. Using the wrong character is a hinderance to readers, even it if \textit{looks} right.

For example, in transcribing a Middle English text, you may decide that the Unicode ezh (\char"0292,~\unic{U+0292}) looks more like
the yogh in your source than the Unicode yogh (\kern+1.5pt\char"021D,~\unic{U+021D}) and therefore decide to use it for yogh. But the ezh is
not a yogh! It is a character in the International Phonetic Alphabet and a letter in the alphabets of several minor
languages---but not a letter in the Middle English language. If you use it where the yogh is called for, it will make
your text less accessible and less searchable. Indexing, concordance and bibliographical programs may be misled by it;
screen readers will misinterpret it. To solve one problem (that of visual representation), you may well have introduced
a host of far more serious problems.

Fortunately, Junicode offers a solution for this particular problem. The OpenType feature
\textSourceText{cv63}\index{cv63} substitutes for
the yogh a character that \textit{looks} like the ezh but is semantically a yogh and therefore will be
handled correctly by applications. But neither Junicode nor any other font can solve every problem of this kind.
Sometimes you will have to call to mind the important principle stated above: \textit{A transcript is not a
facsimile}. It is much more important that it should have the same \textit{meaning} as the original than
that it should have the same \textit{look}.

This chapter concerns the transcription of texts in Latin (and to some extent, other archaic languages, e.g. Old and
Middle English, Old French). It is long-standing custom, when transcribing certain kinds of documents, to retain marks
of abbreviation---for example, the \textex{\hlig{\char"0A753p\char"0363}} you may find in a manuscript or printed edition representing
the word \textit{propterea}. This is okay---and Junicode can help with the task. But when dealing with the
abbreviations, punctuation, and diacritics of an old text it is more important than ever that you use semantically
correct characters for your transcription, as this will help readers who already face significant challenges.

For example, the abbreviation \textex{\hlig{\char"0A753p\char"0363}} as printed here consists of an underlying sequence of Unicode
characters: \textex{\char"0A753} (\unic{U+A753}, the common abbreviation for \textit{pro}) + \textex{p} + \textex{\char"25CC\char"0363}
(\unic{U+0363}, the combining small \textit{a}). The OpenType feature \textSourceText{hlig}\index{hlig}
 (Historical Ligatures) has been
applied to this sequence, changing its appearance but not its underlying value. That underlying value is intelligible
to computer applications in the sense that they can recognize each character.

This doesn't mean, though, that computer programs can correctly interpret \textex{\hlig{\char"0A753p\char"0363}} as
\textit{propterea}. Many (probably \textit{most}) Latin abbreviations are ambiguous: this one,
for example, can mean \textit{propterea} or \textit{propria}. Some abbreviations (most
notoriously \textex{\char"25CC\char"035B} \unic{U+035B}) can mean many things, depending on context. It takes a human being with a
knowledge of Latin to interpret them correctly.

So another way you can add value in your transcript is by interpreting abbreviations like
\textex{\hlig{\char"0A753p\char"0363}} and
supplying expansions of them. Fortunately, systems for representing texts often offer ways to handle this task
gracefully. For example, in a TEI (Text Encoding Initiative) text, you would use this construction:\\[1ex]

\noindent\verb!<choice>!\newline
\verb!  <abbr rend="hlig">!\textrm{\hlig{\char"0A753p\char"0363}}\verb!</abbr>!\newline
\verb!  <expan>propterea</expan>!\newline
\verb!</choice>!\\[1ex]

\noindent This kind of structure can be approximated in HTML, with supporting CSS and scripting to allow readers to choose between
a ``diplomatic'' version, with unexpanded abbreviations, and a ``reading'' version, with expanded abbreviations and
perhaps other amenities, such as modern punctuation and capitalization.

There are other ways to add value to a transcript---for example, by correcting errors, annotating the content, or
writing textual notes. Each of these operations takes your transcript farther from the facsimile and closer to the
edition.

\section[Common combining marks]{Common combining marks}
A \textbf{combining mark} is a character that combines with another character (called the \textbf{base}) to form a
character with accent (e.g. \textex{\'e}) or an abbreviation (e.g. \textex{\cvd[1]{81}{p\char"035B}} for \textit{prae}).
Unicode and the Medieval Unicode Font Initiative (MUFI) offer code points for many precomposed combinations of base +
combining mark, but it is also possible to place any mark over any base character by entering first the base and then
the combining mark. It is also possible to place a combining mark over another combining mark. For example, to produce
\textex{\cvd[32]{84}{q\char"0363\char"0304}}, enter this sequence: q (\unic{U+0071}) + \unic{U+0363} + \unic{U+0304}.

Junicode 2 contains many variants of combining marks: for example the curly zigzag \cvd[1]{81}{\char"25CC\char"035B} is a variant of
Unicode's angular zigzag {\char"25CC\char"035B} (\unic{U+035B}), produced by applying the OpenType feature 
\textSourceText{cv81[2]}\index{cv81} to
\textbf{both the base character and the combining mark}. Sometimes the combination of base + combining mark + OpenType
feature will not produce the desired effect. When this happens, place \unic{U+034F} (a special invisible combining mark,
included in Unicode for exactly this purpose) between the base and the (visible) mark.\\

\noindent\secletter{a.}\ \ For a straight stroke over any letter, use the \textsc{combining macron} (\unic{U+0304}):
\begin{quote}
\=onis \textit{omnis}; o\={m}is \textit{omnis}; d\=apna \textit{dampna}; dam\={p}a
\textit{dampna}.
\end{quote}
The combining macron can also be applied above superscripts and combining marks. Apply the OpenType feature
\textSourceText{cv84[33]}\index{cv84} for a narrower macron:
\begin{quote}
\cvd[32]{84}{antiqua\char"034F\char"0304} \textit{antiquam}; \cvd[32]{84}{q\char"0363\char"0304} \textit{quam}.
\end{quote}
For the superscript \textit{a}, use the OpenType feature \textSourceText{sups}\index{sups} (see r. below).\\[1ex]

\noindent\secletter{b.}\ \ For a straight stroke through a tall letter, use the \textsc{combining short stroke overlay} (\unic{U+0335}): \textex{f\char"0335\ d\char"0335\ l\char"0335}. But Unicode also has precomposed versions of
\textex{d}, \textex{l} and other characters \mbox{with} stroke, e.g.
\textex{{\dj}} (\unic{U+0111}), \textex{\char"019A} (\unic{U+019A}).\\[1ex]

\noindent\secletter{c.}\ \ For \textex{\~{}} above any character, use the \textsc{combining tilde} (\unic{U+0303}):
\begin{quote}
\~a \textit{ac}, \textit{apud}; \~a \textit{alias}.\newline
d\~ns \textit{dominus}; ca\~{r}ina \textit{carmina}; \cvd[4]{12}{f\~{c}is} \textit{factis}.\newline
p\~oita \textit{posita}.
\end{quote}

\noindent\secletter{d.}\ \ For \textex{\~{}} through a vertical stroke, use the \textsc{tilde overlay} (\unic{U+0334}):
\textex{l\char"0334\ d\char"0334} (\unic{U+0303} would be positioned above the letter, e.g.
\textex{\~{l}}, \textex{\~{d}}). For the ligatures
\textex{l̴l̴}, \textex{b\char"0334b\char"0334}, and \textex{f\char"0334f\char"0334}, type the
sequence for \textex{l\char"0334}, etc. twice.\\[1ex]

\noindent\secletter{e.}\ \ For the tilde positioned above two letters, use \textsc{combining double tilde}
(\unic{U+0360}) between the letters. It is
automatically repositioned to clear tall characters: c\char"0360o t\char"0360o d\char"0360o o\char"0360l. The same is true of
\textsc{double breve} (\unic{U+035D}) c\char"035Do d\char"035Do, \textsc{double macron} (\unic{U+035E}) c\char"035Eo d\char"035Eo,
\textsc{double inverted breve} (\unic{U+0361}) c\char"0361o d\char"0361o, and \textsc{double circumflex} (\unic{U+1DCD})
c\char"1DCDo d\char"1DCDo.\\[1ex]

\noindent\secletter{f.}\ \ The figure used to represent \textit{er\textup{ }}(and other similar combinations) is a common medieval
abbreviation which takes many forms. The semantically correct Unicode character is the \textsc{combining
zigzag} (\char"25CC\char"035B, \unic{U+035B}), but the best match in Junicode 2 for the figure as it appears in the
\textit{Record Interpreter} and the \textit{Statutes} is a gothic variant of this, which MUFI
encodes as \unic{U+F1C8} (the curly form zigzag). However, because for technical reasons many applications will not position
the MUFI character correctly over the base, that code point should be avoided. The best way to access this variant is
to apply \textSourceText{cv81[2]} to \unic{U+035B}, as here:
\begin{quote}
\cvd[1]{81}{deb\char"035Be \textit{debere}; int\char"035B\ \textit{inter}; f\char"035B\hspace{0.2em}r\=u \textit{ferrum}; gn\char"035Bo
\textit{generatio}; p\char"035B; \textit{prae}; seru\char"035Be \textit{servire}}.
\end{quote}
The curly form of the combining zigzag may be attached to any letter, and it may change shape depending on the letter it
is attached to (including caps, for which use the \textSourceText{case}\index{case} feature, and small caps:
\textex{\cvd[1]{81}{A\char"035B B\char"035B \textsc{c\char"035B
d\char"035B\hspace{0.2em}}}}).\\[1ex]

\noindent\secletter{g.}\ \ All letters a--z, and several others too, have combining forms.
You may access these via their code points, when they are standard Unicode, via the
\textSourceText{\textsc{cv84}} feature, or via
Junicode's special entity references. For details, see \hyperlink{ss10}{4.10.3, Character Entities
for Combining Marks}.
\begin{quote}
q\char"0366\ \textit{quo}; q\char"0365\ \textit{qui}; quatt\&\_orr; \textit{quattuor}.
\end{quote}
\section{Spacing characters}
\secletter{h.}\ \ The symbol for \textit{is}, \textit{es} and a number of other abbreviations is the
\textsc{is-sign} (\unic{U+A76D}):
\begin{quote}
for\char"0A76D\ \textit{foris}; o\={m}\char"0A76D\ \textit{omnes}; \char"0A76Ft\char"0A76D\
\textit{competentes}; inf\char"0A76D\ \textit{infortunium}.
\end{quote}
This character will sometimes ligature with the preceding letter. The italic version differs from the roman
stylistically (\ \textex{\textit{for\char"0A76D\ o\={m}\char"0A76D\ \char"0A76Ft\char"0A76D\ inf\char"0A76D}}), but it will be
intelligible to informed readers.\\[1ex]

\noindent\secletter{i.}\ \ There are two characters for \textit{{}-us} in Unicode:
\textsc{spacing us} \unic{U+A770} (do not
confuse this with \textsc{con} \unic{U+A76F}) and \textsc{combining us} \unic{U+1DD2}. The \textit{Record
Interpreter} and \textit{Statutes} appear to use only the spacing character:
\begin{quote}
\cvd[1]{81}{i\~{p}i\char"0A770\ \textit{ipsius}; u\char"035Bs\char"0A770\ \textit{uersus}; p\char"0A770tea
\textit{postea}; p\char"0A770\ \textit{post}.}
\end{quote}
\noindent\secletter{j.}\ \ The three-like sign is the \textsc{et sign} (\char"25CC\char"0A76B, \unic{U+A76B}, also used for
\textit{us} in the Latin ending \textit{-ibus}). Do not use the numeral three (\textex{3})
or the Middle English yogh (\textex{\char"021D}, \unic{U+021D}):
\begin{quote}
quib\char"0A76B\ \textit{quibus}; lic\char"0A76B\ \textit{licet}; s\char"0A76B\ \textit{sed}.
\end{quote}
\noindent\secletter{k.}\ \ For \textit{{}-rum} the Unicode \textsc{rum rotunda} (\unic{U+A75D}) is like the one in MUFI/Junicode.
The one in the \textit{Record Interpreter} and \textit{Statutes} is a late stylized version of
this. Use \unic{U+A75D} and apply OpenType feature \textsc{\textSourceText{cv80}}\index{cv80} to obtain the correct shape:
\begin{quote}
\cvd{80}{a\~{i}a\char"0A75D\ \textit{animarum}; co\char"0A75Dpere \textit{corrumpere}; beato\char"0A75D\
\textit{beatorum}}.
\end{quote}
\noindent\secletter{l.}\ \ For \textit{cum}, \textit{con}, etc. use \textsc{small letter con} (\unic{U+A76F}):
\begin{quote}
\char"0A76Fputus \textit{computus}; \char"0A76Fa \textit{contra}; \char"0A76Fnouit \textit{cognouit}.
\end{quote}
\noindent\secletter{m.}\ \ For \textit{per} (or sometimes \textit{par} and other similar sequences), use
\textsc{p with stroke} \unic{U+A751}:
\begin{quote}
\char"0A751s\={o}a \textit{persona}; \char"0A76F\char"0A751et \textit{comparet}.
\end{quote}
\noindent\secletter{n.}\ \  For \textit{pro,} use \textsc{p with flourish} \unic{U+A753}:
\begin{quote}
\char"0A753ceres \textit{proceres}.
\end{quote}
\noindent\secletter{o.}\ \ For \textit{prae}, \textit{pr{\ae}}, \textit{pre}, there is no separate character;
use a variant of the \textsc{zigzag} (f. above) with \textex{p}:
\begin{quote}
\cvd[1]{81}{p\char"035Bs\~{e}s \textit{praesens}}.
\end{quote}
\noindent\secletter{p.}\ \ For \textex{q} with stroke through the descender, there are two Unicode points: \unic{U+A757} for a
straight stroke, and \unic{U+A759} for a diagonal stroke (the \textit{Record Interpreter} appears to use only the
former, and neither is listed among the \textit{Statutes} abbreviations):
\begin{quote}
\char"0A757\ \textit{quod}; \char"0A757d \textit{quid}; \char"0A757b\char"0A76B\ \textit{quibus}.
\end{quote}
\noindent\secletter{q.}\ \ For \textit{quae}, \textit{que}, use \textex{q}
followed by \textsc{et} (\unic{U+A76B}) with or without \textSourceText{hlig}\index{hlig}: \textex{q\char"0A76B}
\textex{\hlig{q\char"0A76B}}. For the semicolon-like \textsc{et} sign (\textex{\cvd{83}{q\char"0A76B}}), use
\textSourceText{cv83[1]}\index{cv83}; for the subscripted version (which can also form a ligature via \textSourceText{hlig}), 
use \textSourceText{cv83[2]}:
\textex{\cvd[1]{83}{q\char"0A76B\ \hlig{q\char"0A76B}}}.\\[1ex]

\noindent\secletter{r.}\ \ All of the letters a-z are available in superscript form. Access with the \textex{sups}
OpenType feature:
\begin{quote}
q\sups{o}s \textit{quos}; c\sups{i}lo \textit{circulo}; cap\sups{i} \textit{capituli}.
\end{quote}
The basic Latin letters a--z have anchors that allow you to position combining marks over them (see a. above)\\[1ex]

\noindent\secletter{s.}\ \ Tironian \textsc{et} sign \char"204A\ \unic{U+204A}, cap \char"2E52\ \unic{U+2E52}. 
With \textSourceText{cv69[1]} \cvd{69}{\char"204A\char"2E52}; with
\textSourceText{cv69[2]}\index{cv69} \cvd[1]{69}{\char"204A\char"2E52}.\\[1ex]

\noindent\secletter{t.}\ \ For \textit{est}, use \textex{\char"223B} \unic{U+223B} \textsc{homothetic}. Use of a mathematical sign for this
purpose is not ideal, but Unicode offers no better solution.\\[1ex]

\noindent\secletter{u.}\ \ For \textit{tz} (Old French), use \textex{\char"01B6} \unic{U+01B6} \textsc{z with stroke}.\\[1ex]

\noindent\secletter{v.}\ \ For an abbreviation for \textit{Rex},
use \textex{\textrecipe} \unic{U+211E} or \textex{\char"211F} \unic{U+211F}.\\[1ex]

\noindent\secletter{w.}\ \ At least one edition uses a spacing version of the \textsc{combining zigzag}
(\textbf{f.} above).
Neither Unicode nor MUFI has a matching character: with Junicode, apply \textSourceText{cv67}\index{cv67} to the spacing
\textsc{macron} (\unic{U+00AF}): \textex{\cvd{67}{\char"25CC\char"00AF}}.

\section[Other formatting]{Other formatting}
\secletter{x.}\ \ For underdotted text, use Stylistic Set 7, Underdotted. For letters that lack an underdotted form, use \unic{U+0323} \textsc{combining dot below}.

\section[On the web]{On the web}
Because Junicode is a very large font, web pages should use a subsetted version to speed loading
(see \hyperlink{OnTheWeb}{Chapter 9, Junicode on the Web}, for instructions).
The variable version of the font is better for web use than the
static fonts, since one variable font file can do the work of many static font files.

All major web browsers (Firefox, Chrome, Safari, Edge) are capable of accessing all of Junicode's characters via
OpenType features, use of which promotes accessibility and searchability. When building a web page, study which
features will be needed and write them into the appropriate element or class definition of the page's CSS style sheet.
For example, if you use the curly form of the zigzag (\unic{U+035B}) anywhere, you are likely to want it everywhere, and so it
should be included in the CSS styling for the <body> element:
\begin{verbatim}
body {
  font-family: Junicode;
  font-feature-settings: "cv81" 2;
}
\end{verbatim}

\noindent But the \textSourceText{hlig}\index{hlig} feature, if applied to the whole text, 
will produce many unwanted effects, so it should be
included in a class definition to be used in a <span> applied just to the target sequence:
\begin{verbatim}
.que {
  font-feature-settings: "hlig" on;
}
filio<span class="que">q&#xA76B;</span>
\end{verbatim}

\noindent The illustrations here use the low-level CSS font-feature-settings property.
There are higher-level properties for some
OpenType features, but as these are not (yet) universally supported by browsers, and some implementations are buggy, it
is best to stick with font-feature-settings for now.

For the purposes addressed in this document, the font-feature-settings for the <body> element
should probably be as follows:
\begin{verbatim}
font-feature-settings: 'cv69' 2, 'cv80' 1, 'cv81' 2;
\end{verbatim}
\noindent And the following classes should be defined:
\begin{verbatim}
.super {
  font-feature-settings: 'sups' on, 'cv84' 39;
}

.que {
  font-feature-settings: 'hlig' on;
}

.deleted {
  font-feature-settings: 'ss07' on;
}
\end{verbatim}