Controlled language – does
my company need it?
Controlled languages use basic writing rules to simplify sentence structure. Here is how they work and how your
company can benefit from introducing a controlled language.
Photo: Kellie Folkerts
16 APRIL 2009
By Uwe Muegge
EXAMPLES OF ORGANIZATIONS THAT HAVE
CREATED A CONTROLLED LANGUAGE
What is control-
led language? Alcatel: Controlled English Grammar (COGRAM)
Avaya: Avaya Controlled English (ACE)
A controlled language is a natu- Caterpillar: Caterpillar Technical English (CTE), Caterpil-
ral language, as opposed to an lar Fundamental English (CFE)
artificial or constructed language. Dassault Aerospace: Français Rationalisé
Natural languages such as Ericsson: Ericsson English
English or German are languages General Motors (GM): Controlled Automotive Service
that are used by humans for gen- Language (CASL)
eral communication. A controlled IBM: Easy English
language differs from the general Kodak: International Service Language
language in two significant ways: Nortel: Nortel Standard English (NSE)
1. The grammar rules of a Océ: Controlled English
controlled language are typically Siemens: Siemens DokumentationsDeutsch
more restrictive than those of the Scania: Scania Swedish
general language. Sun Microsystems: Sun Controlled English
2. The vocabulary of a control- Xerox: Xerox Multilingual Customized English
led language typically contains
only a fraction of the words that
are permissible in the general
As a result, authors who use a controlled
language have fewer choices available when
writing a text. For example, the sentence “Check Eliminating translation Enhancing comprehensibility
the spelling of a paper before publishing it” is a One of the most widely used controlled lan- Helping authors avoid both semantic and
perfectly acceptable sentence in general English. guages today is ASD-STE100 Simplified Technical syntactic ambiguity has been recognized as a
Using CLOUT™, a controlled language rule set English, also known as Simplified English. Simpli- goal worth pursuing in and by itself, especially
developed by the author of this article, the fied English was originally developed by the in the domain of technical communication.
sample sentence would have to be rewritten as European Association of Aerospace Manufactur- Some organizations are deploying a controlled
“You must check the spelling of your document ers (AECMA) in the 1980s. The main purpose language for the sole purpose of improving the
before you publish that document” to comply of Simplified English was to create a variant of user experience of a product or service on the
with rules regarding vocabulary, active voice, Standard English that aircraft engineers with only domestic market.
use of articles, and avoidance of pronouns. a limited command of English could understand,
thereby eliminating the need to translate mainte-
nance manuals into foreign languages.
Why do we need Common features
controlled languages? Streamlining translation A common characteristic of controlled languages
Within the localization industry, many people is the fact that very little information about their
Facilitating language learning familiar with the controlled language concept rule sets and vocabularies is freely available. This
When C.K. Ogden developed Basic English in associate it with automating the translation proc- is not really surprising considering that a control-
1930 – probably the first controlled language – ess. In fact, it typically comes as a surprise that led language can provide an organization with a
the developer had the explicit goal of dramati- controlled languages can and have been used for distinct advantage over its competitors.
cally reducing the 5+ years it takes to master purposes other than making the translation proc- Apart form that, you will find very few similarities
Standard English. Based on a vocabulary that ess more efficient. By restricting both vocabulary between controlled languages. Nortel Standard
contains 850 essential words (for comparison, and style, using a controlled language typically English, for instance, has only a little over a dozen
the Oxford English Dictionary defines more than improves match rates in translation memory envi- rules, while Caterpillar Technical English consists
600.000 words), Basic English is designed to be ronments and translation quality in (rules-based) of more than ten times as many. And a recent
acquired in a few weeks only. machine translation environments. comparative analysis of eight controlled English
languages found that the number of shared controlled-language source documents typically granularity. In other words: There is too much
features was exactly one, i.e. a preference for have higher match rates when processed in a variability within these topics!
short sentences. translation memory system. Higher match rates
mean lower translation cost and higher transla-
Why should I use a for translation memory
controlled language? Some controlled languages have been specifi-
cally designed with machine translation in mind,
Here is why it makes sense for your company to e.g. Caterpillar Technical English or the Control- Writing in a controlled language reduces vari-
use a controlled language: led Language Optimized for Uniform Translation ability, especially if the controlled language not
CLOUT. Using a controlled language customized only covers grammar, style, and vocabulary, but
Improved usability for a specific machine translation system will also function.
Documents that are more readable and more significantly improve the quality of machine- In a functional approach to controlled language
comprehensible improve the usability of a generated translation proposals and dramatically authoring, there are specific rules for text func-
product or service and reduce the number of reduce the time and cost associated with a hu- tions such as instruction, result, or warning mes-
support calls. man translator or editor. sage. Here are two simple examples for functional
controlled language rules:
Objective metrics and author support
Impacts on translation
Tools-driven controlled language environments
enable the automation of many editing tasks and Even in environments that combine content Text function: Instruction
provide objective quality metrics for the author- management systems with translation memory Pattern: Verb (infinitive) + article + object +
ing process. Controlled-language environments technology, the percentage of untranslated seg- punctuation mark.
also provide authors with powerful tools that ments can still remain fairly high in new projects. Example: Click the button.
give them objective and structured support in This can be a major challenge for organizations
a typically rather subjective and unstructured that wish to reduce the cost and time involved Text function: Result
environment. in the translation of their materials. While it is Pattern: Article + object + verb (present
certainly possible to manage content on the tense) + punctuation mark.
Lower translation costs sentence/segment level, the current best practice Example: The window “Expense Report”
As controlled-language texts are more uniform seems to be to chunk at the topic level. This appears.
and standardized than uncontrolled ones, means that reuse occurs at a fairly high level of
+7 (495) 913 6653
Russia Ukraine Kazakhstan Germany
ISO 9001:2000 certified
18 APRIL 2009
Do I have to develop
Implementing functional controlled language or French > English and watch how Systran’s free
my own controlled
rules will enable authors to write text in which machine translation system turns a complete
sentences with the same function have a very website into a fully navigable, highly compre-
high degree of similarity. This not only makes hensible virtual English site in real time. Click on
sentence modules reusable within and across Not at all! Today, many organizations that wish to the link Controlled Language/Rules for Machine
topics in a content management system, it also reap the benefits of controlled-language author- to see ten sample CLOUT writing rules that have
dramatically improves match rates in a translation ing opt for a software-driven solution that comes a high impact on the comprehensibility and
memory system. with a built-in set of grammar and style rules. (machine) translatability of instructional text in
Software-driven rules checking English.
Systems like acrolinx IQ Suite, IAI CLAT, or Tedopres
Controlled authoring HyperSTE have enabled thousands of organiza-
for rules-based machine tions to improve the quality and productivity of References:
their authoring and translation processes. In a
translation systems software-driven authoring environment, organi- Ogden, Charles Kay. 1930. Basic English: A
zations don’t have to maintain the staff of highly General Introduction with Rules and Grammar.
While uniformity is the decisive factor in improv- trained linguistic experts needed to develop London : Treber, 1930.
ing efficiency in a traditional translation memory and deploy a proprietary controlled language.
environment, reducing ambiguity in the source Instead, the organization simply selects the rules Basic English Institute 1996. Ogden’s Basic
text makes machine translation more productive. that are most suitable for a given content type English Word List. Ogden’s Basic English. [Online]
The problem that rules-based machine transla- from a set of preexisting writing rules. Typically, [Cited: February 3, 2009.] http://ogden.basic-
tion systems like Systran struggle with, is the fact these checking tools support the definition of english.org/words.html.
that in uncontrolled source texts, the (grammati- multiple sets of rules for multiple types of content
cal) relationship between the words in a sentence (e.g. stricter rules for user documentation than for AeroSpace and Defence Industries Associa-
is not always clear. To enable rules-based knowledge base articles). tion of Europe. 2005. ASD-STE100 - Simplified
machine translation systems to generate better Terminology management support Technical English - International Specification for
translations, the controlled language needs to From a technology standpoint, it is relatively the Preparation of Maintenance Documentation
have rules like the following that helps the ma- easy to implement the rules part of a controlled in a Controlled Language. ESSAS Electronic Sup-
chine translation system to successfully identify language; the terminology part is typically more porting System for ASD Standardization. [Online]
the part of speech of each word in a sentence: labor intensive. It is certainly true that many 2005. [Cited: February 3, 2009.] http://www.asd-
controlled language software solutions include stan.org/sales/asdocs.asp.
a module for collecting terminology, but still the
task of creating a corporate dictionary, which is O’Brien, Sharon. 2003. Controlling Control-
what this job amounts to, might be a daunt- led English: An Analysis of Several Controlled
Write sentences that have articles before ing one. Not only will all synonyms among the Language Rule Sets. Machine Translation Archive.
nouns, where possible. possibly thousands of terms in use within the [Online] 2003. [Cited: February 3, 2009.] http://
Do not write: Click button to launch organization have to be identified, but these www.mt-archive.info/CLT-2003-Obrien.pdf.
program. synonyms will also have to be categorized into
Write: Click the button to launch the preferred and deprecated (do not use) terms.
program. While creating a corporate dictionary may be a
challenge, once it is available, that dictionary may
Write sentences that repeat the noun also be the feature most valued by the users of Uwe Muegge is the Direc-
instead of writing a pronoun. the controlled language system. tor of MedL10N, the life
Do not write: The button expands into a science division of CSOFT.
window when you click it. He is currently a member
Examples of controlled
Write: The button expands into a window of TC37 at the International
when you click the button. Organization for Standardi-
zation (ISO) and teaches
To see an implementation of a simple controlled graduate courses in Terminology Management
language designed for machine translation, visit and Computer-Assisted Translation at the
the author’s website at www.muegge.cc. The Monterey Institute of International Studies
With rules in place that mitigate the weaknesses entire site was written in CLOUT, the Controlled
of rules-based machine translation systems, Language Optimized for Machine Translation. On email@example.com
the quality of their output is bound to improve the home page, click on any of the language www.medl10n.com
dramatically. combinations into English, i.e. German > English