Controlled Language 2-99 (June 1999) by Ursula Reuther

In the field of technical writing the use of Controlled Language (CL) attracts more and more public interest. However, the merits of controlling language in the context of technical documentation are not uncontroversial.

On the one hand, there is a clear need for correct, consistent, easily readable and understandable, and translatable documentation. This objective undoubtedly can be reached by adopting a Controlled Language approach, restricting "natural language" input to previously defined syntactic and stylistic structures as well as to a restricted set of both general and terminological vocabulary.

On the other hand, there are the reservations put forward by the technical writers about the realisation, the feasibility and even the usefulness of such an approach, since often they feel uncomfortable being restricted in "yet another" domain, i.e. their personal style of writing.

Within the MULTILINT project one of the aims has been to find a reasonable and feasible balance between the two positions above: the prototype application developed within the project takes into account both the users’ needs (i.e. the requirements and expectations by the technical writers) and the needs emerging from the request to produce high quality technical documentation.

The linguistic tools integrated in the system focus on source text control are one of the most prominent points for quality control within the whole documentation process. The approach of controlling as early as possible aims not only at improving translation performance, but also at improving terminology management, information retrieval, and the like.

All tools can be activated from the SGML editor the technical writers are using for producing and editing their documents. The source text control tools comprise spell checking, grammar checking, terminology checking, consistency checking, and style checking. Once a control tool or a set of control tools is applied to a given document, the system highlights the defective text part in a separate control window and gives an explanation of the error and, for some control tools, a suggestion of how to correct the text segment. The tools can be combined and executed in any combination. So the control can be "tuned" according to the personal needs and well-known weaknesses of the technical writer, who then may decide for himself or herself which control tools to apply and which recommendations to follow. Thus the writer to some extent preserves his or her "freedom".

In addition, the system allows authors to specify further control parameters, i.e. they may choose between instructive and informative text types (which has an impact on the "style rules" to be applied), they can specify the (technical) domain the text relates to, and they may choose between various languages and various levels of error explanation, since writers cannot always assume they will write in their mother tongue or to have an in-depth linguistic knowledge.

The most recent feature integrated in the system, which supports the user-friendliness on the conceptual level, is the so-called "shielding" of lexical items and grammatical structures. This functionality allows technical writers to "ignore", in special cases, the error message provided by the system. As for lexical items, the system stores the unwanted item in an author-specific word list for a possible later legitimisation, where appropriate. The unwanted error messages with respect to syntactic structures are stored only for the same document, thus respecting the context sensitivity of syntactic errors.

So, given the modular design and the flexible handling of the system, the MULTILINT approach consists not only of a system which controls the technical writer, but also the technical writer controls individually what is going to be controlled and so this control is realised.  

"transline Deutschland - Übersetzungsdienst für technische Übersetzung"