\title{Reply}{\rightskip0pt plus1fil\hyphenpenalty100
Thank you for your generous review of my book {\sl Typefaces for
Desktop Publishing\/} in the last issue of the  newsletter. I
hope it will not sound peevish if I take up the issue of
hyphenation which you mention in the review. I think its worth
raising because it is something about which people have very
definite views. Indeed witnesses may be able to tell of
differences of opinion between myself and Paul Stiff, who
designed the book, as we made individual decisions about
hyphenation while working on the make up of the pages (although
those who know us both may not be at all surprised at this). And
anyone who has been involved in editorial work will know that
hyphenation decisions can arouse great passion in the hearts of
authors, whose natural concern is that their words be displayed
in the best manner possible.

That brings me to the first point. If you opt for hyphenation
(in either justified or unjustified text) its unwise to rely on
the hyphenation decisions made by your software. They will do as a
first pass, but you need to check the decisions against a
hyphenation dictionary, unless you have particularly good reason
to be confident in your software. In other words, you have to
override many of the software's decisions. Laborious, but
necessary. Certainly I would not trust the hyphenation decisions
of QuarkXPress, either in its previous or present version (just
as I would not rely on a spelling checker, without checking a
document myself). So any grouses about the hyphenation in my
book have to be addressed to me, and not to Quark.

On to the bigger issue of whether or not to hyphenate
unjustified text. There are two preferences at work here, both
of which can be supported by arguments about enhancing the
readability of the text. But the preferences may also be founded
on what we have grown accustomed to, and what we find pleasing
aesthetically. Hyphenators will argue that using hyphenation to
maintain a relatively smooth, and predictable, right-hand margin
is likely to contribute to the efficiency of the saccadic
movements of the eye, that underlie reading. Smoothing the
right-hand margin seems especially important in narrow columns,
where the visual impact of varying line endings can be dramatic.
(Aesthetic sub-text: hyphenators like smooth right-hand margins,
and are not squeamish about breaking up words, within limits.)
Non-hyphenators will argue that hyphenation breaks up the
profiles of words, which are so vital to word perception in
reading, and so is likely to disrupt reading processes.
(Aesthetic sub-text: non-hyphenators just don't like breaking up
words.)

The `truth' probably lies somewhere between the two: a smooth
right-hand margin helps, and a degree of hyphenation to bring
this about can be tolerated, but excessive disruption of words
through hyphenation can get in the way of efficient reading.

Its always hard to pitch the results of readability testing
against preferences based on experience of one particular way of
working. Nevertheless, for the record, can I quote research by
Jim Hartley and Peter Burnhill (respectively, a psychologist and
a typographer)?  In a pilot study, they found that hyphenation
at every line ending, within a single page text, slowed down
reading for eight out of ten people tested, compared to a single
page text with no hyphenation (none of us will be surprised at
this). They then tested reading speed and comprehension of text,
without hyphenation, and of text with hyphenation on roughly 33\%
of line endings, and found no significant difference between the
two conditions. None of the people tested noticed the difference
in hyphenation between the two texts they had read, until it was
pointed out to them. When they were shown the two texts side by
side, significantly more said they preferred the un-hyphenated
text to the hyphenated text (24 to 10, with a further 8
expressing no preference). Just one of many cases in typographic
research when preference does not map on to performance (another
classic is the serif versus sans serif issue).

As far as I know, no one has done eye-movement studies of the
effects of justification and hyphenation on reading. In fact I
think it might be impossible to prepare materials that would
allow you to isolate those variables, given the sensitivity of
eye movements to other factors that would co-vary with them. But
that's another story.
In sanguine mood (or maybe feeling more vulnerable than I will
admit), I checked a few sample pages of my book and found (phew)
that hyphenation is well within Hartley and Burnhill's 33\%
range. For the non-hyphenator, however, reading hyphenated text
can be like trying not to think of pink elephants once someone
has reminded you not to do so: if you find hyphenation irksome,
every instance will leap out of the page at you.

Well, I have gone on at length. Someone who goes on at greater
length, and makes an engaging read, is Ronald McIntosh, in his
book Hyphenation.  I recommend it to hyphenators and non-hyphenators alike.

}

{\frenchspacing\parindent0pt
\everypar{\hangindent1.5em\hangafter1}


J. Hartley, and P. Burnhill. Experiments with unjustified
text. Visible Language, 5(3), 1971.

R. McIntosh. Hyphenation. Bradford: Bradford Computer
Hyphenation, 1990.