The Translator’s Tool Box - © International Writers’ Group, LLC
These tools applied three different theories to the count of these files:
• Word merely counts the words displayed in a browser, omitting all hidden
text, such as keywords or pop-up texts for graphics, etc.
• PractiCount, Déjà Vu and Trados are specifically geared toward counting or
translating HTML files so they have to be able to reveal and count all
• The text editor UltraEdit counts all words, including a lot of non-
translatable coding information. It’s not a very useful number to present to
your client in an invoice (unless you are not interested in keeping the
client, that is. . .).
Interestingly, even within these groups there are fairly significant differences.
These are due to differences in the counting parameters. These include
questions of delimiters (how are \ and - counted and how many words is
C:\Program Files\Firefox or format-specific?) and numbers (are those
to be counted and, if so, how many words is 255.255.255.0?). And it
becomes very hairy, of course, when it comes to non-alphabet-based
languages or languages without spaces between words.
It seems that there are two main strategies for dealing with these problems.
You can avoid word counts altogether and either go with an hourly rate or a
character count (such as the 55 characters per line that many European
translators do business by), or you can make a special point with your client
to agree on a certain program for the word count.
For most text- or word-processor-based documents (DOC, RTF, WPD or TXT),
Word’s word counts have become a de-facto standard. This means that if you
plan to count your words in any of these documents using other applications,
you had better inform your client about it. The same accounts for most other