Technical Translation: Usability Strategies for Translating Technical Documentation

Text Processing Software 
Text processing software as used here refers to translation memory (TM) 
tools such as 
Trados Translator’s Workbench
, 
STAR Transit
or 
Atril Déjà 
vu
. However, it is conceivable that some other form of text storage and 
recognition software or text database could be used (Buchanan 1992). For 
the purposes of this study, Trados Translator’s Workbench was used. 
Under normal circumstances, TM tools are used to help translators trans-
late from one language to another, but they can also be used to translate in-
tralingually or from one variety of a language to another, e.g. US to UK 
English. When rewriting a text using Trados, we have the source text 
which we will call ST and we have the edited text which we will call the 
target text 
or TT. As each sentence is edited and rewritten, the TT is stored 
211

---

Assessing Usability 
along with the ST in a database. Each subsequent sentence in the text being 
edited is compared against the ST/TT units in the database to see if there is 
a match. As more and more of the text is edited, more ST/TT units are 
added to the database and the likelihood increases that a new sentence will 
at least partly match one or more of the ST/TT units already in the data-
base. 
Relating this to Iconic Linkage, while Iconic Linkage by definition refers 
to making isomorphic, sentences which are non-isomorphic and which 
cannot be detected by TM tools because they analyse surface structure and 
not meaning, TM tools can detect instances of partial matches and isomor-
phic sentences. That TM tools can detect partial Iconic Linkage is clear but 
what makes this useful is that very often, these partial matches can some-
times be turned into full Iconic Linkage by rewriting them.
Also, TM tools can indirectly detect non-isomorphic but semantically iden-
tical sentences thanks to the incidence of placeables. Placeables are words or 
phrases such as product names, toolbar icons, menu options and dialog 
boxes etc. that do not change in either editing or translation. As such, they 
will stay the same regardless of the way in which a sentence is phrased. 
Thus, if an existing unit contains the placeables X, Y and Z, a new sentence 
that has these terms may, depending on the ratio of these terms to the total 
number of words in the sentence, be offered as a partial match solely on the 
basis of these placeables. From preliminary tests using this method, fuzzy 
(partial) matches above approximately 60% can frequently represent in-
stances of non-isomorphic semantic matches. They can be rewritten to in-
troduce full or partial Iconic Linkage where there was none before. Trados 
also provides a concordance search function which allows users to search 
the database for instances of a particular word or phrase. This function can 
also be used to identify potential candidates for partial Iconic Linkage. 
The primary benefit of this method is that TM tools can “remember” and 
analyse a greater number of sentences than a human could ever hope to do. 
In doing so, TM tools capitalise on latent Iconic Linkage, be it full or par-
tial. However, it should be noted that this method on its own can only de-
tect latent Iconic Linkage in the text. This can then be reused throughout 
the text or transformed into full Iconic Linkage depending on the human 
operator’s memory. TM tools alone cannot introduce new Iconic Linkage 
into a text, only repeat existing formulations consistently throughout the 
text. 
212

---

Experiment to Test the Impact of Iconic Linkage
new Iconic Linkage into a text by specifying how something should be 
phrased while TM tools ease the burden of analysing and remembering 
large amounts of text. 
In this study, one version of the user guide was rewritten using Trados in 
conjunction with selected style guide rules. As already stated, it is not feasi-
Version 3.0
was chosen because of its comprehensiveness and because the 
ubiquity of Microsoft products makes its writing style more familiar to us-
ers. It would be unrealistic and impractical to implement every rule con-
tained in the style guide. Indeed many of the rules contained in the style 
guide simply did not apply to this user guide. Rather, the following series 
of rules were selected on the basis of their applicability to DigiLog and how 
easily they could be implemented: 
parallel structures in headings, tables lists, sentences and procedures 
use gerunds in headings and not infinitive verb forms 
avoid anthropomorphism 
avoid use of Latinisms and foreign words and phrases 
short, uncomplicated sentences 
logical procedural syntax 
use the active voice 
use second person to directly address user 
non-sexist language / bias-free pronouns 
do not use possessive form of product names 
use present tense 
use positive constructions where possible 
use consistent terminology and avoid polysemy
213
When style guides and TM technology are used together, however, they 
form a powerful suite of methods with each one effectively cancelling out 
the shortcomings of the other. Thus, style guides can be used to introduce 
it was decided to use a commercially available style guide published by 
ble to develop a comprehensive style guide specifically for this study so
Microsoft Press
. The 
Microsoft Manual of Style for Technical Publications 
•
•
•
•
•
•
•
•
•
•
•
•
•

---

Assessing Usability 
Once the rewritten version of the user guide was completed, both ver-
sions were carefully proofed to ensure correct spellings, consistency of for-
matting, completeness of information, correct references and terminology as 
well as general style and register. To ensure that both user guides were ade-
quate, fully functioning documents from a technical communication point 
of view, Flesch readability tests were carried out on each text using the 
Mi-
crosoft Word
readability statistics function. The results of these tests showed 
that both user guides were of virtually the same level of readability and that 
either one would have been perfectly acceptable for users (see Section 2). 
Figure 1 and Figure 2 show the results of the Flesch readability tests. Ac-
cording to Microsoft Corporation (Microsoft Word Help 2002), the ideal 
score for documents is between 60 and 70 for the Flesch Reading Ease test 
and between 7.0 and 8.0 for the Flesch-Kincaid test.
Figure 1: Readability Statistics for Original User Guide 
214
Rewriting was carried out using the two methods in parallel. Thus, as each 
sentence was analysed in Trados, various style rules were implemented 
where appropriate and the edited sentence added to the database.

---

Experiment to Test the Impact of Iconic Linkage
Figure 2: Readability Statistics for Edited User Guide 
The readability data also shows that, in addition to reducing the overall 
word count by approximately 12.7% from 3,287 words to 2,870 words, 
there are no passive sentences in the rewritten user guide. This is an obvi-
ous result of the rewriting process when we consider that one of the rules 
selected to implement Iconic Linkage explicitly states that the active voice 
should be used instead of the passive voice.
Nevertheless, it is possible that, even before we conduct the experiment, 
the elimination of passive sentences could be regarded as a confounding fac-
tor. This may or may not be a valid proposition. One argument to support 
this may be that eliminating passive sentences merely makes the text more 
ity
tributes
ments
attempt
menting IL. 
In any case, if the results of the experiment do show that users using the 
rewritten version of the users guide perform better, the nature of the im-
provement in performance will indicate whether eliminating passives is a 
genuine confounding variable. If, for example, participants in the experi-
mental group only perform better in terms of the speed with which they 
work, then it is possible that the missing passives are a confounding variable 
because we can attribute the increased speed with ease of comprehension 
caused by improved readability or a shorter text. However, if improve-
ments take place across a range of usability criteria, e.g. speed, error rates, 
215
readable, not usable. However, referring back to the discussion of readabil
in Chapter 2, it is apparent that readability is just 
one factor
which con
to usability, and as such is not distinct from it. Thus, any improve
in readability (which, in this case are negligible) are as a result of an 
to improve usability, i.e., the selected strategies aimed at imple

---

Assessing Usability 
retention of information over time, numbers of errors and satisfaction, then 
it would be difficult, if not impossible, to attribute such an improvement to 
improved readability alone. Thus, the issue of whether the elimination of 
passives can only be answered by the results of the experiment. 
These readability scores, while showing that both versions were of a rela-
tively equal standard from a traditional readability point of view, do not 
show the deeper and more fundamental textual differences between the 
documents. Instances of full Iconic Linkage can, however, be uncovered 
using the “Analyse” function in 
Trados Translators Workbench
. 
Match Type 
Original Version 
Edited Version 
Repetitions - Segments 
10 
28 
Repetitions - Words 
21 
291 
IL Percentage 
0 
10 
Table 3: Comparison of Repetition in Both User Guides 
Table 3 contains results from the “Analyse” function run on the original 
and edited versions of the user guide. While both versions are virtually the 
same in terms of content and readability, the results clearly show that the 
amount of repetition or Iconic Linkage in the two versions varies quite sig-
nificantly. We can see that in comparison to the original version, at least 
10% of the total word count of edited version consists of Iconic Linkage as 
suggested by the fact that 28 out of the 306 segments are repetitions of 
other segments. (Note: Some partial repetitions could be instances of IL but 
they are not detected by the analysis tool in Trados). 
The finished user guides were proofed one final time before being 
printed in colour on high-quality paper. The user guides were spiral bound 
to make them easier to open and place on a desk while reading.