Computer programs making use of translation memory (TM) have been mentioned in several articles published in the Translation Journal. These programs were originally developed for the software localization industry to support the translation of high-volume texts that contain a large number of repetitions, are frequently updated, and require quick turnaround times. The use of TM also represents invaluable help in translating technical documentation in the automotive and aeronautical industries, which often use TM projects developed in-house for certain large projects. Most commercial TM programs of today were conceived in the course of such projects.


Reasons for the TM buying frenzy

In late 1999, we were witnessing widespread popularity of TM programs. This phenomenon can be attributed to several reasons:

• Over the years, translation bureaus and translators were led to believe that TM programs represent the cutting-edge technology in the translation industry. The fact that these programs are actually special tools to support special projects somehow got lost in the din of self-promotion. Translators who believed they could effectively work without such aids were suddenly made to feel insecure.
• More and more translation bureaus have looked for translators who used TM programs. The intent behind this has, unfortunately, rarely been to improve quality, but rather to reduce costs.
• Price competition among TM programs toward the end of 1999 has made these programs, which used to be extremely expensive, affordable.
• Instead of modules, complete packages were being offered (Trados, Transit), at an attractive price, which made translators believe that they could purchase an essential tool with a one-time investment.

WordFisher cannot be considered an alternative to TM programs where these are used for their originally intended purpose. But it can be an alternative wherever the nature of the task does not justify the use of a heavy-duty TM program.

According to the information found on Trados's home page, around 4000 copies of the Freelance package were sold worldwide during the few months preceding early December within the company's 15th Anniversary Special Offer.

Does the nature of translation projects justify this volume? Are there this many large projects in the translation market? I am not even sure that many of the truly large projects ever get to the free translation market. If a large company has an ongoing need for translation, is it not more economical for it to purchase or develop the program that best satisfies its specific needs and perform the work in-house or by building up a relationship with a bureau that specializes in that industry?

If the amount and nature of projects do not justify it, what explains this buying frenzy? On the one hand, it is actually due to panic, i.e., fear of technical obsolescence and loss of orders. On the other hand, it is because these programs offer a number of functions that are actually helpful in the translator's work. I would emphasize two factors: the emergence of corpuses and automatic performance of repetitive tasks.

Dictionaries v. corpuses. Dictionaries are the traditional aids to translators. With the spread of computers, electronic dictionaries and glossaries have also emerged. These were initially "copies" of their paper-based counterparts, but had much better search capabilities due to the nature of the technology. With the increase in computer power and the popularity of CD-ROMs, encyclopedias have also become very popular with translators. Encyclopedias specifically designed for computers, with no paper counterparts (such as Microsoft Encarta) were born. Multilingual corpuses (such as the Bible and legal corpuses) also emerged.

Automatic replacement of repetitive texts. Depending on their type, translation projects contain a certain number of repetitive segments. These segments may be words, specialized terms, standard phrases, or even complete paragraphs. While most search and replace operations can be performed with the built-in functions of word processors, TM software promises—and increasingly also offers, although not to extent as many believe prior to purchasing—"intelligent" and "automatic" search and replace over the entire project.

Since no other program offered translators a solution for building and managing corpuses or an intelligent project-level search and replace function, it is understandable that many people considered the purchase of a TM program worthwhile from purely rational considerations.

Many buyers are or will be disappointed. I know several translators who hardly use the program they purchased or do not use it at all. Tunneling through hundreds of pages of a manual, learning the use of a new interface, the concept of databases, etc. are particularly difficult for those who would like to focus on translation, rather than technology.


Could it be simpler?

I've been developing the WordFisher for MS Word macro set since 1994. It started out as a glossary management program, but later, as I became familiar with TM programs, I became "envious" of their functions and attempted to implement as many of them as possible in the macro language of Word. I realized that there is a gap between advanced word processors and heavy-duty TM programs. WordFisher intends to fill this gap, and its current version can be regarded as a full-fledged translation aid program.

The new version of WordFisher requires MS Word 6.0 or above, but it also works under Word 2000. The program is written in the WordBasic language. For the translator, it resembles a TM program, but provides a simpler interface in Word. The main functions of WordFisher are:

• Handling (multi-file) translation projects, automatic indexing to help navigation within the files in the project, and integrity control (existence of files, integrity of indices);
• Searching in the project, context check (this function is usually neglected by TM programs; they normally search in previous translations only, although context within the text to be translated is often more useful);
• Replacing in the project, even multiple replace on the basis of a list (in practice this means the possibility of pre-translation based on a list of expressions);
• Automatic glossary preparation during translation (logging the replacements);
• Automatic building of a bilingual corpus (prepares a table from the source- and target-language sentence pairs);
• Checking the consistency and completeness of the finished translation (on the basis of the sentence pair tables).

When translating by overwriting, the translation's bilingual corpus is built automatically. Corpuses can be later input in any TM program without time-consuming preparatory work. This function alone makes working in WordFisher worthwhile. WordFisher has a built-in aligner to convert previously done translations into corpuses.

Although corpuses can be searched by WordFisher, larger corpuses are better handled using external programs. In my experience, dtSearch (http://www.dtsearch.com) is ideal for this purpose. dtSearch uses indexing, can process almost any file format, and can even search using fuzzy technique if needed. The result of a search can be better viewed in dtSearch than in most TM programs.


Organizing a translation project with the help of WordFisher

It has been shown under actual working conditions that projects theoretically (by the agency's theory) requiring a TM program can be handled using WordFisher and dtSearch without affecting quality or speed.

All TM programs are capable of exporting their translation memories into a parallel text and of importing parallel text pairs into their translation memories. These functions are used in fitting WordFisher to TM programs.

The customer sends the exported translation memory and the glossary of the project. If their combined size does not exceed 1 or 2 MB, WordFisher can handle the reference material by itself. If there is more reference material, it is convenient to combine it with the files received and to handle them together, for example, using dtSearch.

The translation project is created in WordFisher from the files to be translated. This process is automated—no user intervention is required. Translation is then performed by overwriting. During the translation process, the reference material must be constantly monitored, and whatever is found there must be taken into account. Repeating expressions only have to be entered once; WordFisher does the replace in all the files.

Upon completion of the translation, the program prepares the tables containing the source-language and target-language sentences. These can be returned to the customer along with the translated files, and the customer can then enter them into the translation memory of his own TM program.


Finally, an answer to the question

The answer to the question posed in the title of this article: WordFisher cannot be considered an alternative to TM programs where these are used for their originally intended purpose. But it can be an alternative wherever the nature of the task does not justify the use of a heavy-duty TM program, i.e., in smaller MS Word projects, which constitute the bulk of most translators' everyday work.

The home page of WordFisher can be found at:
http://www.wordfisher.com