Comments on Writing
Richard Hughey
This is a list of things I'm tired of correcting -- if too many of
these are violated, I may return the work unread. The list includes
typesetting, grammar, and style, and is not complete.
If you are giving a talk (thesis proposal or defense) provide the text
one month before the talk to the readers after it has been okay-ed by
your primary advisor. Be sure to send a copy of your abstract to the
appropriate staff member at the same time you schedule the talk.
Allow at least two weeks or more after your talk to work on changes
(this is particularly important for taking part in commencement).
Before presenting the proposal or thesis to your committee, ensure
that your advisor has completely read the draft and that you have
integrated all your advisor's comments
Most people use LATEX. I personally use the article style with
some margin changes (fullpage.sty can be a bit much; you may
want to adjust them by hand or, as in this document, use largepage.sty in my tex
directory). Various conferences, or individual esthetics, often lead
to a desire to make LATEX to different things. Many other people
have found this too, thus there is a vast repository of style files
and hacks both on the net and around here. See ~rph/tex/README
for a description of all my favorite style and bibliography files, as
well as answers to a few frequently asked questions and pointers to
archives.
- 1.
- LATEX uses inter-word spacing between words
ending in capital letters, such as U. C. Unfortunately, this means
that when you end a sentence with a capital letter, you do not get
inter-sentence spacing, and must use
C\@.
, as in C. Notice
how the spacing is larger than in the first example. Do this for
every sentence that ends with a capital letter. Regular expression
search and replace will do wonders.
- 2.
- Use as few fonts as possible.
- 3.
- The hyphen is indicated with a single ASCII dash (as in
application-specific array processors), number ranges with two
(97-100), and dashes with three -- like this. Placing spaces next to
the dash is optional, but be consistent. Dashes, as with many other
things, should not be overused. Note the use of a hyphen to form
a compound adjective, above.
- 4.
- Words in mathmode require mboxes. Note the different spacing between
(
$V_{Instr\_count}_i$
) and
($\mbox{\em Instr\_count}_i$
or alternatively {\em Instr\_count}$_i$
). The first is
ugly.
Here are a few trivial errors that often come up.
- 1.
- Be consistent in your use of commas in lists. Choose and
consistently use one form, as A, B, and C; or A, B and C.
Occasionally, for clarity, in either of these schemes the other is
used, but not often.
- 2.
- Know the difference between that and which. The
first is defining and restrictive, while the second provides extra
information that can be eliminated without changing meaning. Clauses
using the second, which can appear in any writing, are usually
separated out by commas.
- 3.
- Proofread your bibliography and ensure all items have dates and
page numbers, and that author and journal names are identical if they
appear more than once. The bibliography is one of the most fruitful
places to look for errors.
- 4.
- Spell out small numbers like one and two. When writing numbers
less than one, always include a leading zero (0.5, not .5). Do not
include an excessive number of insignificant digits. Two or three
significant digits often suffices.
- 5.
- Do not treat citations as nouns; they are not grammatical
elements. For
example, ``Smith and collegues did something exciting [1],'' rather
than ``[1] discusses exciting work.''
- 6.
- Be sure to spell check and proofread. This one often gives me
problems.
- 7.
- In lists, itemizations, and even paragraphs within a section,
use parallel constructions where appropriate -- not necessarily the
same words, but at least similar structures. A trivial incorrect
example would be using ``first'' followed by ``secondly'' in a list.
This is particularly important when the items are closely tied
together. In this document, most items are in imperative without a
pronoun. Adding an item starting with ``You should'' would violate
this idea.
- 1.
- Avoid the use of passive voice, such as ``A machine was built to
solve the problems of the world.'' One alternative is the editorial
we, as in ``We built a machine that solves the problems of the
world.'' Editorial we can be as overused
as passive voice, however, so try to use it only occasionally, if at
all. Instead, write short, snappy, assertive statements such as ``Our
machine solves the problems of the world!'' Passive voice is
particularly bad in abstracts -- the abstract should draw the reader
into the work rather than put the reader to sleep.
- 2.
- Include a solid section on related work.
- 3.
- If you wrote code, spend some time talking about performance and
implementation.
- 4.
- Write short sentences and paragraphs and use lots of sectioning
commands, especially in dissertations. Topic sentences should be
favored as well. Use short words, too!
- 5.
- Remember my favorite passage from Strunk and White (an excellent
book to review before any major writing project): ``Rather, very,
little, pretty -- these are the leeches that infest the pond of prose,
sucking the blood of words.''
Richard Hughey
2004-02-17