Which is the better way to do accents in LaTeX?

Question

When you need to use accents in LaTeX like when you need to write the word "Cálculo", that means calculus in Portuguese, you can do in follow ways:

1. write c\'alculo
2. use the package \usepackage[utf8]{inputenc}, and just write "cálculo"

I was just using the second way, but my professor said that the better way to write accents is the first way. I didn't find a good reason for it in google, is there any good reason for that?

Answer 1

The only sane reason I can think of for giving in to your professor's preference concerns bibliographies and, specifically, bibliographies created with BibTeX and a bibliography style that sorts entries alphabetically by authors' surnames. (Aside: The issues raised in the remainder of this answer do not pertain to bibliographies generated with biblatex+biber.)

For deeply historical reasons, BibTeX does not sort "accented" characters such as á, ä, à, and â that may occur in the author and editor fields together with the letter A; instead, they are treated (for sorting purposes only) as coming after the letter Z. This affects how pieces authored by, say, Rädermacher, Ràdon, Rámos, and Râmuz [with some deliberate misspellings -- sorry!!] are sorted relative to pieces written by, say, Randall and Rybczynski. Do you want these entries to be placed before Randall or after Rybczynski? You may expect the former to happen, but BibTeX delivers the latter outcome.

Here's another example: Suppose your bibliography contains single-authored entries by Hasbrouck, Haščič, Hase, and Hayworth. Would your expect Haščič's publication to be listed before Hase's -- or after Hayworth's? If you enter the author's name as Haščič instead of as Ha{\v s}\{v c}i{\v c}, BibTeX delivers the second option.

For more information on how to enter accented characters in BibTeX entries in order while avoiding the sorting-related issues noted above, please see the posting How to write “ä” and other umlauts and accented letters in bibliography? [Shameless self-citation alert!]

Well, I can think of a second reason: If your computer keyboard does not provide a straightforward method for entering certain accented characters -- not just á, ä, and â but also, say, angstroms, ogoneks, and thorns [!] -- it's obviously very nice to know that you can enter them as \'a, \"a, \^a, etc as well.

A separate comment: If you compile your document with pdfLaTeX and if you input accented characters directly into your document, I will assume that you also load the fontenc package with the T1 option. If you compile your document with XeLaTeX or LuaLaTeX, there's no need to load the fontenc package.

---

Addendum: The following MWE, compiled with pdfLaTeX on a MacTeX2020 system, demonstrates that BibTeX places entries by Rädermacher, Ràdon, Rámos, and Râmuz (note the accented characters) after Rybczinski rather than before Randall. Ouch!! Hence, in order to obtain what most people would think is the "correct" sorting outcome, it is necessary to enter these names as R{\"a}dermacher, R{\`a}don, R{\'a}mos, and R{\^a}muz if and when they occur in the author or editor fields of BibTeX entries.

    \documentclass{article}
    \begin{filecontents}[overwrite]{mybib.bib}
    @misc{r1,author="Randall",year=3000,title="Thoughts"}
    @misc{r2,author="Rädermacher",year=3000,title="Thoughts"}
    @misc{r3,author="Ràdon",year=3000,title="Thoughts"}
    @misc{r4,author="Rámos",year=3000,title="Thoughts"}
    @misc{r5,author="Râmuz",year=3000,title="Thoughts"}
    @misc{r6,author="Rybczynski",year=3000,title="Thoughts"}
    \end{filecontents}
    \usepackage[T1]{fontenc} % useful under pdfLaTeX
    \usepackage[authoryear]{natbib}
    \bibliographystyle{plainnat} % use a bib style that sorts entries alphabetically
    \setlength\bibsep{0pt} % optional 
    \begin{document}
    \nocite{*}
    \bibliography{mybib}
    \end{document}

Answer 2

Actually, you do not need type \'a nor \usepackage[utf8]{inputenc} that is loaded by default, so simply:

\documentclass{article}
\begin{document}
Cálculo
\end{document}

Time ago, with people using a myriad of encodings according to the OS and the idiom, a good reason to type \'a was that will be rendered as "á" also with latin1 or cp437 encodings, for example, whilst á will produce "Ã!" with latin1 encoding or an error with cp437 encoding. Today that almost everybody use only utf8, this is not longer an advantage for sharing LaTeX sources .

As Mico well pointed, using bibtex, composed characters of authors names will be wrongly sorted in some languages.

Have the wrong keyboard is also a good reason, and even some others. There also limitations in math mode: you cannot use $á$ (nor $\'{a}$) bot $\acute{a}$ and sometimes I have had to avoid composed characters in other situations, using this or that command of the package <wathever>. That I cannot remember a concrete scenario illustrated that is a rare situation (or that I have a very bad memory).

But nothing of this obscure the fact that write, read, or check the spelling of a long text with escaped tildes is a pain. So, whenever you can, type á and left \'{a} as the plan B.

Answer 3

Unicode-Native (La)TeX Engines

Some questions here assume LaTeX=pdfLaTeX, however there are other LaTeX engines out there with native Unicode support, namely LuaLaTeX and XeLaTeX. If you are free to choose, you can simply use these, which will remove many headaches, especially when using accents in listings and the like, they simply work.

Bibliography

The bibliography is not handled by the (La)TeX engine, but a different tool, classically bibtex. If you are free to choose, you don't need to use bibtex anymore but an alternative bibliography backend, biber with another frontend (package), called biblatex.

\documentclass{article}
\usepackage[backend=biber]{biblatex}
%\addbibresource{bibliography.bib}

\begin{document}
•
\end{document}

Glossaries and the like (makeindex)

If you use anything that uses makeindex, you might want to switch to xindy, at least if you want to use non-latin characters for the collated list and it doesn't really matter which TeX engine you use, e.g.:

\documentclass{article}
% xindy option, s.t. a .xdy file is generated
\usepackage[xindy]{glossaries}
\usepackage{iftex}
\ifPDFTeX
  \usepackage[T1]{fontenc}
\else
  \usepackage{fontspec}
\fi

%\makenoidxglossaries
\makeglossaries

\newglossaryentry{Radermacher}
{
  name=Rädermacher,
  description={foo}
}

\newglossaryentry{Radon}
{
  name=Ràdon,
  description={bar}
}
\newglossaryentry{Ramos}
{
  name=Rámos,
  description={baz},
}
\newglossaryentry{Ramuz}
{
  name=Râmuz,
  description={buzz},
}
\newglossaryentry{Rybczinski}
{
  name=Rybczinski,
  description={basse},
}

\begin{document}

\glsaddall
\printglossaries

\end{document}

Running latex mwe, texindy mwe and latex mwe will output the entries in the correct order. Actually, the sorting is language-specific (some languages would sort "a, ä, á, ...", some sort "a, ..., z, ä, á, ...". xindy (and LuaTeX/XeTeX) uses -L to specify the language.

Again, you can circumvent this issue and use {\'a} or the sort=Ramuz key to make it somewhat work, but this will hardcode one sorting and not allow for a flexible change in language as xindy allows.

In Journals

Some journals force you to use pdfLaTeX which, by now, at least uses utf8 as input encoding by default. However you still need the correct mapping of input characters to font glyphs and thus the fontenc package, and might have issues with other packages such as listing when using Unicode symbols. I usually simply use iftex in my preamble and check for pdfTeX to load these helper packages. That way I can mostly just write in Unicode and only have to think a bit harder about the problem when submitting to journals.

ConTeXt

This is not LaTeX, but an alltogether different TeX engine which uses LuaTeX in the backend (thus UTF-8 native) and doesn't need any external tools like bibtex/biber or makeindex/xindy to work. Basically it is the "no frills, no need to worry" variant:

\definesynonyms[glossary][glossaries]
\glossary[Rädermacher]{Rädermacher}{foo}
\glossary[Ràdon]{Ràdon}{bar}
\glossary[Rámos]{Rámos}{baz}
\glossary[Râmuz]{Râmuz}{buzz}
\glossary[Rybczinski]{Rybczinski}{basse}

\starttext
\completelistofglossaries[criterium=all]
\stoptext

Switching to ConTeXt isn't all that easy though, even if many things are arguably better "designed"