ACKNOWLEDGEMENTS

When I first considered writing my thesis on translation memory, I wasn’t exactly sure on what aspect I should focus. I would like to thank Chris Langewis for helping me to narrow my topics down to the one presented in this thesis. I would also like to thank him for his valuable input as my thesis advisor and as an expert in the field.

My survey would not have been a success without responses from the members of LANTRA-L, CompuServe’s Foreign Language Education Forum (FLEFO), Interlang and from fellow translators who responded to the various personal e-mails that I sent out. Gerald Dennett, Mark Berry and Jeff Allen all deserve a special thanks for contributing valuable information. I would also like to thank the various translation memory software manufacturers, especially TRADOS Corporation, for their input and responses to my questions.

Finally, I would like to thank William Webb for reading and editing and David Sawyer and Frank Austermühl for reading and approving my final thesis. After all, if it weren’t for this audience, I wouldn’t be able to share it with the wider audience—all of you.

The figures and tables that are not displayed in this electronic document can be found in the HTML pages (figures) and Excel spreadsheets (tables) included with this document in the archive zip file.

1 INTRODUCTION

Many articles have been written about translation memory in the last few years. Most of the material is provided by the producers of translation memory systems and only covers a specific product or lists specific features of the technology. Some articles even magnify the negative aspects of translation memory. Despite all that has been written, not much has been said about the actual costs or potential savings involved when using translation memory. It is becoming increasingly clear that translation memory is here to stay and that it is serving a useful purpose, but just exactly how is this technology affecting the translation industry? Who can profit from it? Are there any "losers?” This thesis will attempt to determine the applicability of translation memory technology and illustrate the advantages and disadvantages of translation memory in the form of a cost/benefit analysis from the point of view of the end-user.

For the purpose of this thesis, the “end-user” comprises freelance translators, companies with in-house translation divisions, translation agencies and direct clients. The key considerations for determining the cost-effectiveness of translation memory and the cost/benefit analysis will be covered later in this text.

2 Translation Memory Defined

What is translation memory? Translation memory (TM) is defined by the Expert Advisory Group on Language Engineering Standards (EAGLES) Evaluation Working Group's document on the evaluation of natural language processing systems as “a multilingual text archive containing (segmented, aligned, parsed and classified) multilingual texts, allowing storage and retrieval of aligned multilingual text segments against various search conditions.”[1] In other words, translation memory (also known as sentence memory) consists of a database that stores source and target language pairs of text segments that can be retrieved for use with present texts and texts to be translated in the future. The translator, a different translation memory system or a machine translation system provide the target text segments that are paired with the source text segments so that the end product is a quality translation.

What distinguishes TM from other computer-assisted translation (CAT) tools? There are many CAT tools available to assist the translator, such as bilingual and multilingual dictionaries, grammar and spell checkers and terminology software, but TM goes one step further by making use of these other CAT tools while at the same matching up the original source document stored in its database with the updated or revised document through exact and fuzzy matching. Normally, the basic unit of text in a TM database is a sentence; however, the TM user can define what the unit will be. The basic unit might even be a sentence fragment or a paragraph. The translator does not have to re-translate work he or she has already completed. Figure 1 illustrates the basic translation memory process for creating a target language translation.

Fig. 1: Basic TM Process

How does TM differ from machine translation (MT)? MT creates automated translations and requires an advanced terminology database that includes all grammatical elements of a language. The MT system uses comprehensive dictionaries to translate the source text while at the same time applying the grammatical rules, or rule sets, from the database in order to produce the resulting grammatically correct target sentences. The technology sounds like an excellent solution; however, there is a catch: the source and resulting target text segments are not stored away in a database for future use. If a similar text (such as an automobile user’s manual for the same model but different year) needs to be translated, the MT system would have to start from scratch. On the other hand, a TM system is used as a translator’s aid, storing a human translator’s text in a database for future use. TM can be used a few different ways. One way would be to have a translator or a machine translation system translate the original text, using translation memory to store the paired source and target segments. The translator could then reuse the stored texts to create the revised or updated version of the text. Only the segments of the new text that do not match the old one would have to be translated. The alternative would be to use an MT system or a different TM system to translate the original. The new TM system could then be used by a translator to translate the revision or update by aligning the texts produced by the MT system or other TM system and storing them in the TM database for present and future work. The translator could then proceed to translate only the segments of the new text, using TM as described above.

3 The Effects of Translation Memory on the Translation Process

3.1 THE TRANSLATION PROCESS

How does TM affect the conventional translation process? In order to answer this question, we must have an idea of what takes place during this process. This section will briefly touch on the general translation process and the features of TM used in this process.

Figures 2 and 3 illustrate the conventional translation process and the translation process using TM respectively. If it is possible to analyze quickly what type of text one is translating and if the text suits translation memory, there are about the same number of steps involved in both processes. One of the greatest differences, however, is that once a translation has been performed using TM, not only is there a glossary of terms stored for future recall, individual sentences will also be stored, thus cutting down considerably on the time required for a future translation, update or revision.

Figure 4 illustrates the conventional translation process for a text that is being revised. Figure 5 illustrates the same text using TM. Note that when using TM, fuzzy and exact matching are performed using the translation memory program, allowing for quick access to sections that have changed and permitting the user to focus on translating only those changed sections.

Exact matching is the process by which the TM program pairs text segments in a revised source text that match the original source text exactly; however, any text in the document that does not exactly match the original will not be translated. Fuzzy matching is the process by which the TM program pairs text segments in a revised source text with similar text segments from a previously stored translation based on the original source text. Fuzzy matching will find segments that are very similar to the original and suggest the original translation. This function can be set to different levels of sensitivity, allowing the translator to “match” source text segments that may differ only slightly or segments that vary greatly, but still have some similarities. After exact and fuzzy matching, the translator can modify the remaining segments that reflect the changes between the original and revised texts without having to retranslate the entire document (see Figure 6).

Fig. 6: Exact and Fuzzy Matching

In addition to matching source text segments, fuzzy matching can also be used to find terminology in the terminology database that is very similar to terminology being used for a translation. For example, if the term “communicate” is in the terminology database, the translation of “communicate” will be suggested whenever the terms “communicated” or “communication” appear in the original text. The translator can then enter the correct form of the word accordingly.

Although fuzzy matching is quite useful, the user must also be aware of problems that may arise during post-editing of matched text segments. Gerald Dennett explains in his thesis entitled “Translation Memory: Concepts, products, impact and prospects”:

Take the German sentence pairs:

1 “Ein Messer ist im Schrank. Er mißt Elektrizität.”

2 “Ein Messer ist im Schrank. Es ist sehr scharf.”

Imagine that the translator has translated a document containing sentence pair 1 and has thus stored in his Translation Memory the two segments:

“A meter is in the cabinet.” And “It measures electricity.” The syntactical and contextual information supplied by the second sentence indicates to the translator that the word “Messer” here refers to a meter. The translator then runs a text containing sentence pair 2 through the pre-translation routine in his Translation Memory software. The Translation Memory software will recognise a 100% match in the first part of the pair, and insert “A meter is in the cabinet.” in the translation. A human translator would immediately realise from the syntactical and contextual information supplied in the second part of the pair that here in German word “Messer” is of neuter gender, and hence means “knife”. The translator must hope that he can pick up such mistranslations in his proof-reading.[2]

On the other hand, the likelihood that the above sentences would appear in the same document is probably quite low, especially since they would probably be used in completely different domains or in different types of text.

The alignment tool is an example of a CAT tool that is almost indispensable when initially integrating older translations into TM. Using the alignment tool with TM can save the user time on future projects. Figure 7 illustrates the process of alignment. Alignment involves matching the electronic source/target texts by aligning matching source/target text segments. The translator is essentially building a TM database that is identical to a normal TM database built during the translation process. This process is performed when it is clear that the source text will be revised or updated in the future but was originally translated using the conventional translation process. If the source or target texts are in hard copy, one should seriously consider the likelihood of whether or not the text will require future updates or revisions before performing an alignment.

3.2 MANAGING THE TRANSLATION PROCESS

Perhaps the most intriguing aspect of translation memory is its ability to aid the user in managing projects, coordinating team efforts and building glossaries and dictionaries. Following are some additional features of TM that allow the translator or other user to manage translation projects more efficiently.

3.2.1 INTERNAL ATTRIBUTES

Most TM products not only store language pairs; they also store other information, called attributes, with the pairs. The most common attributes stored include the creation date, the name of the user or creator, the client, the project ID and the main domain or field (e.g., legal, technical, etc.) of the translation. Once this information is stored with the translated segments, the translator or other user can filter the text for the most important attributes. For example, the user can look for similar text segments by project, client, etc. when performing fuzzy matching, or a project manager may have more control over accountability for translated texts by filtering for creation date or the name of the creator of the translated segments. The latter is particularly useful when a number of translators are working on one large project, especially when the translators are all working with the same language pair.

3.2.2 TERMINOLOGY DATABASES

Most TM products come with a terminology database so that the translator can take full advantage of all of the features of TM. Using an integrated terminology database allows a translator to perform fuzzy matching for a specific term or to use a term in the database suggested by TM. Without a terminology database that is compatible with translation memory, the TM user cannot easily obtain suggested translations for individual words without opening a separate electronic dictionary or looking through a conventional dictionary. Naturally, the user must enter the terminology into the database before it can be useful. Once the terms are in the database, however, an individual translator or team of translators can work on a project and receive the suggested terms from the database, maintaining terminological consistency throughout the translation.

3.2.3 ANALYSIS

The ability to estimate in advance approximately how much time a project will take is not always an easy task. If the translation memory system is a good one, it will have the capability of analyzing a document for similar sentences and text repetition. It will also provide raw word counts, ignoring elements like graphics, HTML tags, software code, etc. that could influence the count. This analysis makes it easier for the translator or project manager to assess whether or not translation memory will be useful for the project and also helps him or her determine how much time may be involved in translating the document, depending on the amount of repetition, the word count, etc.

The user may also use the analysis function to compare different documents for similarities. Analysis can reveal if one document that has been translated previously and a newer document are in any way similar. Depending on how similar the two documents are, the user can estimate the time required for translation.

4 Texts That Are Conducive To Using Translation Memory

4.1 REUSABILITY

The most important characteristic of a text that is conducive to translation memory is that the text will be reused in one way or another. Following are examples of how texts can be reused and how translation memory becomes involved in the process.

4.1.1 UPDATES

A not uncommon occurrence during the translation process is when an update of the text being translated is suddenly made available to the translator. An update is a change in a source text that occurs while the translation is still in progress. Receiving an updated text can cause major difficulties for the translator if the text is large and changes have been made throughout the entire document. Figures 3 and 4 illustrate the update/revision process with and without translation memory. Making updates using translation memory has the advantage over the conventional update process in that the translator does not have to physically search through the entire document for changes. Instead, the translator only has to run the updated source text through the translation memory program to identify new or changed segments and any new terminology. New terminology can be entered into the terminology database by the translator for future use.

Keep in mind that in order for translation memory to be effective, all work must be done in TM and saved in TM format. Anything done outside of TM will not be stored in the memory database and therefore will not be a translation that can be manipulated in the future, unless one has access to an alignment tool. The best way to approach TM is to think about it as being an integral part of the main word processor, just like the word processor’s spell checker. If the TM system is a stand-alone product, always keep a copy of the text file that retains the TM product’s file format.

A translator can even begin the translation process before the final original document is completed. If the translator is given drafts of the original document in its early stages of development, the text can be translated and stored in the TM database. Then, as updated sections of the text are made available, the translator can perform fuzzy and exact matching, thus isolating the new parts from the parts that have already been translated or that are similar to the original. Section 6.5 is an example of this process.

4.1.2 REVISIONS

Many translators find that they continually receive revisions from the same clients. A revision is a new project amending a prior translation, reflecting changes made to a prior source text. Often a translator is asked by a client to revise the translation of a manual for the current product model that will be released within a short period of time. The client wants the translated manual to be available at the same time that the product is launched on the market. If the translator were to use the conventional translation process, it could take months before a very large document would be ready, and the client might not have that much patience or time. If, however, the translator uses translation memory, he or she can analyze what has changed within the document and can provide the revised translation of the manual within a shorter period of time than if he or she had used the conventional process. Section 6.4 illustrates this process.

4.1.3 “RECYCLING” PRIOR WORK

At times, a translator may find that he or she is translating a text very similar to one that had been translated in the past. The translator may run across words or phrases that are almost identical to words or phrases in the older document. The odds that a translator will ever translate the same sentence twice in two different texts is very low; however, the odds are higher that a translator will run across similar phrases or words in texts within the same field and/or for the same client. If the translator has an electronic copy of the target and source texts from the previous translation, then he or she can quickly access the files and perform fuzzy matching with the new source text against the old source and target texts.

4.2 REPETITIVE CONTENT

Another important factor is whether or not there is repetitive content within a text. The higher the percentage of repetitive content within a text, the more desirable it is to use translation memory. Repetitive content may include words, phrases or entire paragraphs. There are a number of different text types, but some tend to have more repetitive content than others. The majority of translatable texts fall into the following categories[3]:

- Correspondence

- Journalism/Communication

v Business/Commercial

- Marketing

- Advertising

- Administration

v Legal

v Scientific

v Technical

- Culture

- Literature

The types of texts that are usually suited for translation memory are marked with the "v" symbol. Interestingly, according to the Telecom Observer, “each year 450 million pages of scientific, technical, and commercial materials are translated world-wide.”[4] Some examples of the type of texts that fall into these categories include:

- Patents (Legal)

- Contracts (Legal, Business/Commercial)

- User manuals (Technical)

- Annual reports (Business/Commercial)

Reusability of a text and the amount of repetitive content can help determine whether TM should be used on a given text; however, reusability is the most important factor. Regardless of text type, if a document will be updated, revised or recycled and is fairly large, the use of translation memory is worthwhile. Figure 8 best illustrates this concept.[5]

Zone de Texte: Repetitive Content, Recyclability and Frequency of Updates

5 Key Considerations for Determining the Cost-Effectiveness of Translation MEMORY

The following are brief descriptions of the key considerations for determining the cost-effectiveness of TM for freelance translators, companies with in-house translation divisions, translation agencies and clients requiring translation. Sections 6 and 7 of this paper will provide more detailed examples of how these factors influence the decision to use TM.

5.1 TYPE OF PROJECT: INDIVIDUAL VS. TEAM

How many translators are involved in the project? If a number of people are translating a text, consistency may be difficult to maintain. Many TM products can be used over a network and are capable of suggesting words and phrases to a translator that other translators have already used elsewhere in the document, thus helping the team maintain consistency throughout the document.

5.2 PERCENTAGE OF WORK IN HARD COPY VS. ELECTRONIC FORMAT

Obviously, if most of the work a translator receives is in hard copy, translation memory is of little use; however, we are quickly becoming a paperless society. From the recent survey I conducted for this thesis (see section 7.1), 70% of the 37 respondents reported that half of the documents they translated were in electronic format. Out of the 25 freelancers who responded, 60% performed at least half of their translations on electronic documents. Almost all of the agencies and companies with in-house translation that responded worked primarily with electronic files. The high percentage of electronic documentation within translation agencies and companies suggests that freelance translators could probably receive a higher percentage of work in electronic format if they were to ask their clients for electronic files. At some point, almost everyone will be translating only electronic documents. As the number of electronic documents increases, the use of translation memory will be all that much more justified.

5.3 TYPE OF TEXT USUALLY TRANSLATED

As mentioned above, translation memory is most useful when texts will be reused and when working with certain text types. If a translator only translates literature (having very little repetitive content) or translates such a variety of texts that he or she will most likely never encounter similar documents again, then translation memory is not an option. On the other hand, if a translator mainly translates software manuals or other material with repetitive content and which the translator knows will be revised in the future, then translation memory is an excellent choice.

5.4 TIME REQUIRED FOR CONVENTIONAL TRANSLATION PROCESS VS. TM PROCESS

Initially, the time required to produce a quality translation using TM may be the same or may even take a little longer than when using the conventional translation process, especially when there is a learning curve involved. The respondents in my survey estimated that it took them one to two weeks of regular TM use to feel comfortable using it. Entering a new translation into TM may also take longer than using the conventional translation process because the first draft is so critical. Any future work done in the TM database will depend on what was entered initially. Post-editing is also affected by the quality of the first draft. A terminology database may also require additional time to develop.

Once the original translation has been stored, however, subsequent revisions and/or updates of the same text will take considerably less time to translate using TM. Overall productivity using translation memory with machine translation is said to be increased up to 30% or 40% for translators.[6] In addition, dictionary development and more efficient coordination of team efforts as a result of using TM help to speed up the translation process. Examples of how productivity can be increased during the translation process are discussed in sections 6 and 7 of this paper.

5.5 COMPARING RATES

Translators must be made aware of the fact that some agencies or customers may request a discount or may want to pay by the hour instead of by the word if they know that a translation was completed using TM. The translation agencies responding to my survey point out that some clients already request discounts for repetitive content within a text. If this is the case, then one is clearly better off using TM to take care of repetitive content so that more time can be spent focussing on the overall quality of the translation. If a translator is able to complete more work in less time, he or she could also potentially earn much more than what was initially expected over a certain amount of time just from the sheer volume of work that he or she is able to produce.

Customers or agencies using translators are increasingly taking advantage of the lower cost of translation when using translators with access to TM. As a result, the price of translation work—regardless of whether or not one uses translation memory—may eventually be driven down by customers and agencies requesting its use. In the near future, freelance translators may find themselves scrambling to acquire a translation memory product so that they can increase their productivity and remain competitive.

One project manager who responded to the survey works for a company providing TM software. He cited various theories on how agencies and freelancers view the rate situation:

1. Using TM adds consistency to the translation, therefore improving the quality of the translation. The customer should pay for the extra quality.

2. One should charge less per word/line/page, but base the rate on the word count of the entire project. Any repetition would be money in the translator’s/agency’s pocket.

3. Charge standard rates for the parts that require special attention. Full translation rates would apply to sections requiring translation. Full proofreading/editing rates would apply to pre-translated sections. Full layout work rates would apply for any desktop publishing, plus a flat overhead fee would be required for processing the job in TM.

4. The translator receives part of the pay from an agency for segments that are derived from fuzzy matching. This is predetermined, based on the idea that “if 70% is similar, we pay only 30%.”

At least two freelance respondents mentioned that they were paid by the hour when they used TM. In addition, the responses received from LANTRA subscribers revealed that most of them were paid by the hour when pre-translating text using TM.[7]

5.6 NEED TO INTEGRATE PRIOR WORK FOR PRESENT PROJECTS (ALIGNMENT)

Many translators find that they receive revisions of work they have done in the past. Depending on the degree of similarity between the prior translation and the current project and whether or not the prior work is in electronic format, the translator may decide that it is worth aligning the old source text and old translation to assist him or her in the present translation assignment.

5.7 NEED TO INTEGRATE PRESENT WORK FOR FUTURE PROJECTS

TM is quite useful when a translator works on a project and realizes that he or she may be working on a revision of the project at some point in the future. It is important to note, however, that two documents for completely different projects very rarely resemble one another enough to use TM. Translation memory is therefore almost exclusively employed for updates and revisions. When the translator receives a new document and determines that it has similar terminology or sentences that match an older translation done in TM, the translator simply translates the new text in TM, using exact and fuzzy matching to identify previously translated terms (from the terminology database) and/or sentences that either exactly or partially match the current terms/sentences. The more terms and/or sentences that match or are similar, the greater the increase in the translator’s productivity.

5.8 FREQUENCY OF UPDATES AND REVISIONS

The more frequently a translator has updates or revisions, the more sense it makes to implement TM. In fact, even if a translator finds that he or she only receives one substantial update or revision a year, translation memory may pay for itself. For more information, see section 6 of this paper.

5.9 CHARACTER SET ENABLING SENSITIVITY

Despite all of the advantages of TM, translators who work with double-byte source languages such as Japanese or Chinese may determine that translation memory actually causes more problems and slows down their productivity rather than provides a cost-effective solution. In order to create a TM database, translation memory must be able to recognize what makes up a sentence—in particular those elements that indicate the end of a sentence and those that are never found at the end of a sentence. A period is often considered the end of a sentence; however, when combined with “Mr.,” “Dr.” and “vs.,” for example, the period no longer identifies the end of a sentence. Western TM systems can handle these types of differences; however, most of them cannot parse double-byte character set strings in a source text because they do not recognize the characters. This means that the TM programs cannot determine what is the end of a sentence in a language with double-byte characters. TM can be used, however, to translate from a source text that does not use double-byte characters to a double-byte character language. TM is therefore most useful to translators of languages with double-byte characters when the double-byte language is the target language.

6 Examples

This section includes various examples of how clients, corporations with in-house translation divisions, translation agencies and freelance translators can estimate the cost-effectiveness of TM for their specific translation needs. The examples will start from the client side, move to the agency and freelance translator, and end with an in-house translation department. This section is by no means all-encompassing, since every company’s or translator’s situation is unique. The scenarios below are restricted to certain factors and therefore should be viewed only as theoretical applications of translation memory; however, they provide an effective way of illustrating the benefits of translation memory in the working environment.

6.1 INITIAL INVESTMENT

The initial translation memory investment for a translator or company may vary. Translation memory (with a terminology database) costs on average about $2,000. The alignment tool may cost another $2,000. Of course, a translator could create a template or macro to facilitate the integration of prior work, but depending on how much text needs to be integrated, the time required to create the template or macro might not be worth the effort. A 17" monitor or larger is recommended ($400 - $1,300) due to the fact that the translation memory software usually runs using a split screen. If the translator has hard copy documents that he or she would like to integrate into translation memory, then a scanner ($90 - $1,000) and OCR (Optical Character Recognition) software ($100 - $500) are required. The translator may, however, determine that using OCR software to create electronic documents is not worth the effort. Translation memory also uses much of the computer's available memory, so the computers should have between 32 and 64 MB of RAM ($60 - $200).

It is not inexpensive to own translation memory. An individual translator could pay $4,650 to $7,000 if he or she were to purchase the above items all at once. A company or translation agency would pay even more for multiple software licenses, etc. Most translators, however, already own some of these products. Not only that, translation memory can pay for itself within a year or two, depending on what type and size of documents are translated. The following examples depict how translation memory can help save companies or translators money and even help provide them with more work over the long term.

6.2 THE CLIENT

In this section, the term “client” is considered to be a larger corporation. In order to discuss how TM can be beneficial to the client of a translation agency or individual translator, we must first understand what the client needs. Based on personal experience working for a business consulting company and on information obtained from people working in the translation business, the client has at least five requirements:

1. Increase income

2. Keep translation costs down

3. Receive quality translation within a short amount of time

4. Receive the translation in the format requested

5. Have the ability to plan a translation budget and estimate time-to-market

Most clients will do anything they can to keep translation costs down, but at the same time, they do not want to give up quality. They realize that a poorly translated text could reflect badly on them or could cause liability issues. However, clients also have deadlines and budget issues to face. Let us examine the translation process from the client’s point of view. In this example, the client is a coffeemaker company.

The client’s project manager (CPM) for product information knows that a new line of coffeemakers will be coming out in two months. The coffeemakers are marketed in France, Germany, Scandinavia, the U.K., the Netherlands and North and South America. The company is located in the U.S., and the American English user’s manual has just been completed. CPM realizes that when the coffeemaker hits the shelves, the manual that comes with it must be available in at least eight different languages in addition to American English. The manuals must be ready for packaging in fifteen days. Ideally, CPM would like to have the translation finished in five days so that the printing division will have plenty of time to publish the manuals. CPM has a budget of $8,000 for the translation of the 20 page/5,000 word manual. CPM contacts the company’s regular translation agency (TA) to find out how much it would cost and how long it would take for the translations. TA informs CPM that the agency has recently implemented translation memory, but will not have time to align the new manual to the previous one for this project. The translation agency’s project manager (TPM) assures CPM that once the new manual has been integrated into TM, any similar manual in the future will be translated much more quickly.

After TPM examines the manual, it is estimated that the project will cost approximately $8,400 for the initial translation (at $.21 per word). TPM also estimates the turn-around time to be four to five days (three days for translating and editing and about one to two for post-editing and administration, using one translator per language). CPM accepts the conditions, although the cost is slightly above budget, and places an order for the project.

Now let us take a look at the same corporate client dealing with the same agency six months from this last transaction.

CPM knows that a new coffeemaker will be launched on the market in two months. The manual for this coffeemaker is very similar to the manual translated six months ago, except for some changes involving the coffeemaker name and features. CPM’s staff has analyzed the differences in a desktop publishing program and has estimated that 20% of the text has changed.

CPM calls TA and lets TPM know there know that a 20-page manual needs to be translated into eight languages. CPM informs TPM that the manual is very similar to the manual translated six months before, and therefore the company would like a discount for the repetitive content. TPM asks to see the manual to analyze it for the changes. CPM sends the electronic copy to the agency.

TPM finds the last manual translated for this client. TPM then analyzes the English manuals using TM to determine what differences there are between the two. CPM was accurate in the estimate. TPM informs CPM that, indeed, only about 20% of the text had changed. As a result, the estimated turn-around time for the project would be approximately three days (one day for pre-translation using TM, one day for translating and editing the remaining text and one day for post-editing, using one translator per language). TPM also offers a 50% discount to CPM for the 80% repetitive content.

TPM has the two in-house translators perform the pre-translation work in TM. The in-house translators finish the work in less than half an hour. The two translators earn about $35,000 per year, or about $17 an hour/$.08 per word. TPM then has the two in-house translators work on two documents and sends the remaining six documents out to six freelance translators. Each translator must still translate the remaining 1,000 words, plus post-edit both the pre-translated work and the new translation. The freelance translators earn approximately $.12 per word for new translation work. They are paid about $30 per hour for post-editing the pre-translated work. All of the translators require one day (8 hours) for post-editing the pre-translated work and one for translating and editing the remaining 1,000 words.

The cost for the freelance translators would total approximately $2,160 ($720 for 6,000 words and $1,440 for post-editing). The in-house translator cost would be about $449 ($160 for 2,000 words, $17 for pre-translation work and $272 for post-editing). The cost to TA for the entire project would be about $2,609. The cost to CPM would be about $3,360 for the pre-translated text and $1,680 for the remainder, totaling $5,040 for the entire project. CPM has saved about $3,360 on this project.

Here is a table to summarize the company’s savings:

	Without TM	Using TM	Savings
Number of words in project	40,000	40,000	N/A
Number of days for turn-around	4-5	3	1-2
Total translation cost	$8,400	$5,040	$3,360
Cost per word	$.21	$.13	$.08

If the above project had been for a larger manual, e.g., a 200 page/50,000 word manual, and the rate were the same ($.21 per word), the savings would be even more apparent. The first time the project was translated, it would take approximately 33 days to complete (29 days for translating and four days for post-editing) and cost the client $84,000, assuming no volume discount. If 80% remained unchanged in the next version of the manual, the project would take approximately 9 to 10 days to complete (approximately half of a day for the pre-translation using TM, 6 days for the remaining translation work and 3 days for post-editing, using one translator per language). The client would only have to pay $50,400 (assuming the same 50% discount from the previous example)¾a savings of $33,600.

This table illustrates the savings on the 400,000-word project.

	Without TM	Using TM	Savings
Number of words in project	400,000	400,000	N/A
Number of days for turn-around	33	9-10	23-24
Total translation cost	$84,000	$50,400	$33,600
Cost per word	$.21	$.13	$.08

Using translation memory has fulfilled most of the client’s needs. CPM has saved the company money and has received the translation back sooner than originally expected. As a result of using TM, CPM will also be better able to plan future budgets and estimate the time necessary for projects. All of this will help CPM streamline the process on the company’s end and may even help prevent the company from losing money due to the late release of manuals for products that are already ready to hit the market.

The other benefit of translation memory is that the client will see consistent use of company-specific terminology throughout the translations. Depending on the company’s agreement with the translation agency, CPM may also have access to the translation memory database file in the future, giving the company the additional flexibility to decide if it wants to start an in-house translation division or even to choose another agency that uses TM.

6.3 THE TRANSLATION AGENCY

A translation agency has needs that are very similar to those of a client’s. A translation agency wants to:

1. Keep costs (overhead) down

2. Perform quality translation quickly

3. Increase income

4. Plan budget

5. Manage project

If we take the previous example from the client side and look at it from the translation agency’s point of view, we see that the agency also draws benefits from employing translation memory.

As mentioned above, one of the most important concerns for the translation agency is to keep costs down. TPM must constantly be aware of how much a project is costing the company. For this project, the two in-house translators work on a salary of about $35,000 per year, translating to about $17 an hour, or $.08 per word. The freelance translators charge an average rate of $.12 per word. The charge for post-editing averages $30 per hour. Ideally, the translation agency would prefer to have the in-house translators do the bulk of the work, but unfortunately, they only specialize in two of the target languages. The time estimate per translator averages 213 words per hour (1,700 words per day for translating, working over three days, plus one day for post-editing). This translates to an average of $800 (at $.08 per word) for the in-house translators and $3,600 for the freelance translators (translation cost) plus another $272 to $544 for the in-house translators and $1,440 to $2,880 for the freelance translators (post-editing), totaling approximately $6,112 to $7,824 for the project (excluding other costs of doing business).

After the initial project is completed, the agency now has a translation memory database for this client’s manual. When the client comes back in six months requesting the manual revision, the agency will see a substantial increase in profit as a result of a decrease in overhead costs. The following calculations demonstrate how the agency is able to realize the profit.

TPM estimates that only 20% of the original text has changed. Since 80% remains unchanged, one of the options TPM has is to employ the agency’s in-house translators to pre-translate the text, using the previous translation as the template. It would take less than half an hour for two in-house translators to pre-translate eight 5,000-word texts in TM. The cost for this work would be approximately $17 (at $8.50 per half hour). The rest of the translation work comprises about 8,000 words, or 1,000 words per translator. The translators can accomplish this work in one day, including editing and some post-editing. Post-editing the text already entered into translation memory by the in-house translators along with the newly translated segments would an additional day. The total cost for the translation work and post-editing would be about $2,592 ($432 [translation and post-editing] for the in-house translators and $720 [translation] plus $1,440 [post-editing] for the freelance translators). The entire project would cost approximately $2,609 (excluding other costs of doing business) and would take about three days (including administration). By employing only the in-house translators for the pre-translation work, the agency was able to save approximately $3,503 to $5,215 in overhead expenses and speed up the translation process.

Here is a summary of the agency’s savings and resulting profit:

	Without TM	Using TM	Savings
Number of words in project	40,000	40,000	N/A
Number of days for turn-around	4-5	3	1-2
Total translation cost	$6,112 to $7,824	$2,609	$3,503 to $5,215
Cost per word	$.15 to $.19	$.07	$.08 to $.12
Total profit	$576 to $2,312	$5,936 to $7,556*

* Includes savings, although some of the savings may be used to cover any additional costs of using TM.

TPM estimates that only 20% of the original text has changed. Since 80% remains unchanged, one of the options TPM has is to employ the agency’s in-house translators to pre-translate the text, using the previous translation as the template. It would take less than half an hour for two in-house translators to pre-translate eight 5,000-word texts in TM. The cost for this work would be approximately $17 (at $8.50 per half hour). The rest of the translation work comprises about 8,000 words, or 1,000 words per translator. The translators can accomplish this work in one day, including editing and some post-editing. Post-editing the text already entered into translation memory by the in-house translators along with the newly translated segments would take an additional day. The total cost for the translation work and post-editing would be about $2,592 ($432 [translation and post-editing] for the in-house translators and $720 [translation] plus $1,440 [post-editing] for the freelance translators). The entire project would cost approximately $2,609 (excluding other costs of doing business) and would take about three days (including administration). By employing only the in-house translators for the pre-translation work, the agency was able to save approximately $3,503 to $5,215 in overhead expenses and speed up the translation process.

If the translation agency were to take on the 400,000-word project, the savings and resulting profit would have been more like this:

	Without TM	Using TM	Savings
Number of words in project	400,000	400,000	N/A
Number of days for turn-around	33	10	22
Total translation cost	$50,848	$15,784	$35,064
Cost per word	$.13	$.04	$.09
Total profit	$33,152	$69,680*

* Includes savings, although some of the savings may be used to cover any additional costs of using TM.

The agency was able to do something it could not have done without using TM. It managed to keep its costs down, increase its profit margin, speed up the translation process and manage the project much more efficiently.

6.4 THE FREELANCE TRANSLATOR

How is the freelance translator affected by the use of translation memory? What are the needs of a freelance translator? The freelance translator is not really much different from a translation agency when it comes to his or her requirements. The translator wants to:

1. Keep costs down

2. Increase income

3. Speed up translation process while maintaining quality

4. Eliminate repetitive translation tasks

5. Plan his or her budget

Let us take the last example from the agency’s point of view and look at it from the freelance translator’s perspective. We will first look at a translator who did not personally own translation memory.

Translator 1 (T1) receives a call from TA. TPM asks if T1 would be interested in translating a manual similar to one T1 had worked on in the past. T1 does not own translation memory, but has used it on a past project, so T1 is familiar with it. TPM informs T1 that the agency will lend T1 a copy of TA’s translation memory software and the terminology database for the project.

T1 charges $.12 per word to translate the text into French. It takes T1 approximately three days to translate and edit about 5,000 words. T1 charges the agency $600 for the translation. T1 has made about $200 per day, or $25 per hour. Since T1 does not own a copy of the translation memory software, T1 must return the software and TM database that has been created for this project. The agency allows T1 to keep a copy of the terminology database that has been developed, but the database will be of no use to T1 unless it is exported into another format or unless T1 eventually buys a copy of the terminology database software.

Six months later, T1 receives another call from the agency asking for a translation of 1,000 words from the revised manual and for post-editing of the 4,000 words in the translation memory database. T1 is again given a copy of the TM software so the post-editing can be completed. T1 charges $30 per hour for post-editing. It takes T1 one day for the translation and some editing and about eight hours for post-editing the entire document. T1 has made about $360 on this project and has finished the work in two days.

Now let us look at another translator (T2), a German freelance translator who has been using translation memory for at least a year and owns TM software.

T2 receives a similar call from TA. T2’s rates happen to be the same as T1’s. TPM supplies T2 with the CMP’s terminology database, although T2 already has a copy from the last translation project. T2 has also already integrated the last project into a TM database. TPM knows that T2 owns a TM system, but does not know that T2 has already integrated the previous manual. TPM requests that T2 provide the agency with a TM database at the end of the project. Since T2 already has a similar manual in a TM database, T2 analyzes the new text against the old and estimates that about 50% of the text is the same or very similar. It takes T2 less than half an hour to pre-translate the similar or identical text. T2 translates the remainder and edits the completed translation over the next two days. T2 has earned $300 per day, and has an extra day in which to take on new translation work.

Six months later, T2 receives another call from TPM to translate the 1,000 words and post-edit the remainder. T2 accepts the job and makes $120 for the 1,000 words that need to be translated. The rest of the text is identical to what is already in the TM database. Post-editing takes about eight hours, costing the agency another $240. T2 has earned a total of $360 in about two days.

If we look at the two examples just described, we see how the freelancer can benefit from translation memory just as much as a client or translation agency can. T2 initially has a clear advantage over T1 as a result of the investment in TM. T2 is able to complete more work in less time, thus being able to realize an increase in income by taking on more jobs. T2 is also more valuable to the translation agency for the simple reason that T2 is faster and does not need to rely on the agency for translation memory software. These advantages make it more likely that T2 will receive more work from the agency in the future. Coupled with the fact that T2 is now able to maintain client databases and budget time more easily, T2 appears to be on the road to success.

Here is a table that summarizes the above example:

	Project 1	Project 2		Project 1	Project 2
	T1 (no TM)	T1 (using TM)	T1 (no TM)	T2 (using TM)	T2 (using TM)	T2 (using TM)
Number of words in first project	5,000	Post-editing	1,000	5,000	Post-editing	1,000
Number of days for turn-around	3	1	1	2	1	1
Total earnings	$600	$240	$120	$600	$240	$120
Earnings per day	$200	$180		$300	$180

What if these two translators received work from the 400,000-word project? Their earnings would look more like this:

	Project 1	Project 2		Project 1	Project 2
	T1 (no TM)	T1 (using TM)	T1 (no TM	T2 (using TM)	T2 (using TM)	T2 (using TM)
Number of words in first project	50,000	Post-editing	10,000	50,000	Post-editing	10,000
Number of days for turn-around	30	4	5-6	16 (trans.) plus 3 (pre-trans. & post-edit.)	4	5-6
Total earnings	$6,000	$960	$1,200	$6,000	$960	$1,200
Earnings per day	$200	$240	$200-$240	$316	$240	$200-$240

6.5 COMPANIES WITH IN-HOUSE TRANSLATION DIVISIONS

An in-house translation division within a company is structured similarly to a translation agency with in-house translators. An in-house translation division fulfills the role of a company, translation agency and translator all rolled into one; its needs are therefore similar to all three. An in-house translation division seeks to:

1 Increase income

2. Keep translation costs down

3. Produce a quality translation within a short amount of time

4. Produce the translation in the format required

5. Have the ability to plan translation budgets and estimate time-to-market

6. Eliminate repetitive translation tasks

7. Efficiently manage the translation project

In order to demonstrate how translation memory affects an in-house translation environment, we must look at a different model. In this scenario, we will look at the in-house translation division of an automobile manufacturer.

The in-house translation division has a technical department comprised of 40 translators, two per language. The technical department is involved with translating all of the different user manuals for each make and model of automobile. Translation memory is a standard part of the division’s translation process.

The company is coming out with a new model of its five standard cars in six months. The 50-page/12,500-word manuals for these automobiles must be ready in five months. The translations of the five manuals from the previous models are already stored in translation memory.

The project manager receives the first draft of the five English manuals and distributes them to the 20 translation teams. The translation teams compare the new drafts to the older manuals and determine that 45% of the new drafts is exactly the same as the previous manuals and 20% is very similar. The teams research new terminology and add the new terms to the terminology databases, which takes a couple of days. One of the team members then pre-translates two manuals and the other pre-translates the remaining three using fuzzy and exact matching. The pre-translation work takes less than one half hour per manual.

After the pre-translation work is completed, the translators post-edit what was entered and translate the remaining 35% of the text in each manual. The entire process takes twelve days (two for research and database entry, seven to pre-translate and translate and three to post-edit).

Two months later, the translators receive updates of the English manuals. The translators compare the updated manuals to the earlier drafts and determine that 80% of the text is exactly the same and 10% is very similar. The pre-translation work is completed in one day, but this time it only takes the translators approximately two days for translation and editing work and two days for post-editing the document (remember, more of the text matches exactly).

Two and a half months later, the translators receive last-minute updates to the manuals. The manuals are due in less than one month. This time, the translators analyze the manuals and determine that 90% of the text is the same, 7% is similar and only 3% is new. The pre-translation work takes one day per team. Translating the remaining text and post-editing only takes about one and one-half days, and final editing takes about half a day. The final version of all five manuals in twenty different languages is ready at least twelve days before the due date¾even when the translators were given the last update with only about 15 days to spare.

First Draft

Second Draft

Final

Number of words in project

62,500

Number of days

for turn-around

What has the automobile company gained from translation memory? The answer is pretty clear. The company will have its manuals in time for shipping. There was no need to send any translation work outside, thus theoretically saving the company money. The teams of translators worked in tandem over a network to complete the translations, ensuring consistency throughout each document. The project manager was able to manage the large projects with ease by starting earlier in the year. In short, the company was able to accomplish all of the tasks required to streamline the translation process and thereby make the company more efficient.

7 Survey/Case Studies

This section discusses the results of a survey conducted for this thesis and case studies compiled by others.

7.1 SURVEY

The following questions were asked in the survey for this thesis entitled "Advantages and Disadvantages of Translation Memory: A Cost/Benefit Analysis." The survey was sent out over three Internet language forums—FLEFO (CompuServe’s Foreign Language Forum), LANTRA-L and Interlang—as well as to individuals whom I personally contacted. The responses to these questions can be found in the appended tables 1 and 2.

1. What percentage of your work is done on an individual basis as opposed to as a team?

2. What percentage of your work is received in hard copy vs. electronic format?

3. What type of texts do you usually translate (e.g., manuals, literature, etc.)? What is the frequency of word and phrase repetition in your texts?

4. What type of additional hardware/software have you had to purchase as a result of using TM (e.g., OCR software, scanner, the TM package and tech. support, etc.)?

5. Have you experienced different rates of pay as a result of using TM (for example, have you noticed that the client/agency pays more or less when it is known that you are using TM)? It isn't necessary to tell me how much you actually charge or the agency pays, just give a theoretical example.

6. What kind of effect does TM have on your translation process (estimate of time saved using TM vs. the conventional method; consider the learning curve and quality issues as well)?

7. Do you also use machine translation? How does this influence the translation process when combined with TM?

8. What percentage of your clients request TM? Request a hybrid of TM and MT usage? Also, if you have time, please tell me what you like or dislike about using TM vs. the conventional translation process.

9. Do you have a need to integrate prior work into TM so it will be available for future projects? If so, what percentage of prior work have you needed to integrate?

10. Do you ever need to translate multiple languages simultaneously? If so, how often?

11. How frequently do you have updates and/or revisions?

12. Please indicate whether you are a freelancer, translation agency, company with in-house translation or client who requires translations.

There were 37 respondents in total. Only one of my six personal contacts did not respond to the survey. Due to the fact that the survey was sent over electronic forums, it is not possible to determine the total number of people who actually received and read it.

The respondents varied by occupation: five were agencies, seven were in-house translators, twenty were freelancers, two were freelancers who also worked in-house at one time or another and three were freelance translators who also worked for agencies. Eleven of the freelancers used TM, although two of them only used it when it was supplied by an agency. Another freelancer who did not own TM was planning on buying a TM product in the near future. Four of the agencies and six of the in-house translators also used TM. Both freelancers who were also working as in-house translators used TM. The three respondents who considered themselves freelancers and agencies did not use TM. One of the last group of respondents had never heard of TM.

The majority (92%) of agencies and in-house translators performed at least half of their work in teams. The freelancers performed an average of about 86% of their work alone.

The majority of the respondents (70%) received at least half of their work in electronic format. The translation agencies handled about 84% of translations in electronic format. In-house translators performed practically all of their translation work (99%) in electronic format. The freelance translators averaged about 56% of their translation work in electronic format.

Virtually all of the respondents who had worked with TM reported that they were impressed with the ability to maintain consistency throughout texts and/or were impressed with the ability to increase the amount of text translated in a day. Productivity varied from a 10% to 60% increase, depending on text type and amount of repetition in updated or revised texts.

A few of the respondents noted that when a text was translated for the first time using TM, it actually took longer to translate. Two respondents mentioned technical problems were encountered at some point during their use of TM. One respondent noted a change in work style to adapt to TM.

About 81% of the respondents translated technical or software manuals. Approximately 35% of the respondents reported a moderate to high rate in frequency of revisions and updates.

About 43% of respondents had a need to align prior work for integration into present and future work. One respondent noted that there was no longer any need to align prior work because the work had already been integrated into TM at an earlier time.

Only two respondents mentioned that they had been requested to use TM, and only two stated they were working on integrating machine translation into the translation process alongside TM.

Most of the respondents did not have to purchase any additional hardware or software as a result of their purchase of TM. At least two of the freelance translators mentioned that they did not personally own a TM package, but that they used TM provided to them when working on projects for a specific client. Only two respondents mentioned having to update their operating systems, one mentioned upgrading system memory and another mentioned having to purchase a second dongle for the TM system, two mentioned purchasing scanners and OCR software, and one mentioned upgrading to new computers and purchasing technical support for the TM system. It is very likely that most of the respondents from translation agencies and companies with in-house translation departments were unaware of their companies' extra purchases resulting from the integration of TM.

In terms of rates of pay, most of the freelance respondents had not encountered a situation in which they were paid less when using TM. One of them reported that he actually charged more for TM because he was the one who mentioned to his client that he could use it to improve consistency. Two freelance translators mentioned that when they use TM (supplied by an agency), they are paid by the hour instead of by the word. In a separate survey sent over the LANTRA-L Internet mailing list regarding this subject, most of the respondents were paid by the hour when they used TM. At least two of the respondents mentioned that they do not tell their clients or agencies that they use TM for fear of a decrease in their current rates. Here is what one freelance respondent had to say:

"The agencies pay less when they ask you to do the translation using TM. I don't tell the agencies that I use TM when it isn't part of the agreement (Subject 21).”

When asked what they liked or disliked about using TM, responses varied:

"I don't like that you basically have to stick to segments. In a Word document, a translator can rearrange a paragraph if it makes more sense in the target language. With [TM], it takes a considerable amount of time to merge the segments, and that defeats the purpose of the memory (Subject 1)."

Subject 1 does have a point. In certain cases, a translator may want to combine or break up sentences or text segments. The difficulty comes when combining more than one sentence or segment. Most translation memory tools allow access to one sentence or segment at a time. For example, when a translator wants to combine two sentences into one, he or she will most likely translate both sentences when presented with the first original sentence and will skip entering a translation when presented with the second sentence. This technique might not work the way the translator would like, since the current translation would appear correct, but the sentence pairs stored in the translation memory would not be correct. Instead of combining the two sentences into one translated sentence, the translator would be better off making the changes outside of the translation memory.

"[The] time saved is directly proportional to percentage of matches... Time spent learning the program is negligible compared to benefits attained (Subject 2)."

"The throughput on standard translations (highly redundant software manuals, help texts) increases from 3000 to about 5000 words per day if the text is completely new for us. If we can use memory databases built up during the translation of earlier versions or similar products, we can reach a rate of 8000 words per day. However, a certain amount of preparation is necessary, i.e., about one day for processing, converting and adapting the source files, generating databases and glossaries from reference material supplied with the texts, etc.

"A skilled translator will need about five work days to reach his full level of productivity using a TM tool... (Subject 8)"

"I must point out that the 'search and replace' function in a common word processor also can be used to speed things up. Before buying my TM program, I sometimes could process as much as 700-800 words an hour by making extensive use of previously translated material [and] search and replace. My 'normal' capacity would be 250 to 350 words an hour (Subject 18)."

Subject 18 is correct in pointing out that a translator can use cut and paste and search and replace functions to mimic some of the tasks that translation memory can handle, but Subject 4 seems to have a better understanding of where translation memory pays off:

"I talked to the owner of a German translation bureau this morning on the question of whether his purchase of [TM] has paid off for him. He answered 'yes' and then went on to tell me that... his use of [TM] had enabled some of his translators to increase their throughput to up to 120 pages/day from their normal 10-12 pages/day. Granted, that must have been a highly repetitive project, but nevertheless imagine how long they would have been cutting and pasting had they not had [TM] available to them... (Subject 4)"

7.2 CASE STUDIES

TRADOS is a German company that is well known for its translation tools. The company has recently begun conducting case studies among users of its TM product Translator's Workbench. One of the companies with in-house translation, Bernhard Beumer Maschinenfabrik KG, reports that the use of the TRADOS product has "improved quality through standardization of terminology used for translations,” "increased efficiency through reduced turn-around time,” and the company has witnessed a “30% translation cost reduction."[8]

8. TM DATABASE OWNERSHIP

One important area of discussion in recent years is the question of who owns a translation memory database. The question is not an easy one to answer. Everyone has an opinion. The issue is relatively new and consequently there are no known legal precedents covering TM database ownership. Alison Rowles, business manager for LISA, says: “Based on our previous experience with similar legal issues in this industry, …this area is one that is normally defined specifically in client/vendor contracts. Since the localization industry is so new there is not a great body of law and precedents for reference.”[9]

Mark Berry from MCB Systems states that a TM database is “a sentence-by-sentence log of the work, ... translation memories are arguably a ‘by-product’ of the translation process…the ‘by-product’ argument cuts both ways. The translation vendor may maintain, ‘Translation memory is simply a by-product of the work we do. Therefore we own it.’ But the client may argue, ‘Translation memory is just a by-product of the work we commissioned, so it belongs to us.’”[10] He believes that the best solution is to “make translation memory ownership a specific point in contract negotiations, so that both sides come away knowing that their respective investments in this asset are both protected and compensated”.[11]

On the other hand, when it comes to compiling the terminology database for a specific project, using TM may actually work to the translator’s advantage, as Suzanne Falcone points out in her article on translation aid software in the Translator’s Journal: “…for once, the client is ‘technically forced’ to provide some terminology. The glossaries of the projects may not be that big, but they still help, as well as the text possibly contained in translation memories or repetition files.”[12]

Donald Plumley, senior vice president of marketing for Bowne Global Solutions states: “From our point of view, the Client owns the [database] if they are paying for it to be developed and maintained. However, there are really tricky issues if [there is a] mixing of company resources (which could be domain specific) and client-specific [databases].”[13]

If a TM database has been created for a specific client, then it is best for the translator to assume that the database is the client’s property, unless an agreement has been made otherwise. This also means that the translator should not use this client’s database on projects for other clients. The bottom line appears to be that the translator and client need to agree on who owns the database by making it part of the business contract.

9. Drawbacks of translation memory

Translation memory, like every other good invention, is bound to have drawbacks. Four particular issues come to mind:

1. Any post-editing of a translation cannot be easily integrated into a TM database if performed outside of the database. If the translator creates a draft translation in TM and then exports the document in a different format, any corrections made to the exported document cannot be captured by TM. This problem can be remedied by making all corrections to the document within the TM system or by aligning the post-edited text with the source text.

2. The tendency for a translator to create only one or two drafts of a translation in TM is higher, possibly affecting the quality of the final translation. Many translators find that when they use TM, they are less likely to “fine-tune” the translation. They may feel that the TM system is less flexible than a word processing program. It may be that once the translator uses the TM system more often, the less likely this will continue to be an issue. It is also possible that these translators may tend to work in fields that award speed over accuracy.

3. Post-editing of pre-translations performed in TM can be quite time-consuming and may take longer than if the translator had used the conventional translation process to translate and edit the document. This issue is largely related to instances in which many text segments appear to be exact or fuzzy matches, but in reality are expressing completely different ideas (see Geral Dennett’s example in section 3.1).

4. Learning how to use the TM program thoroughly may take some time, which the translator may not have, especially if the translator is borrowing the TM program for a specific project. Suzanne Falcone comments:

One problem with these programs is that they usually come at the same time as the first job, and the deadline for delivery obviously doesn’t take training into account. The translator rarely has time to read the user’s manual…, and sometimes can’t even install the program properly on the first (and even second) attempt… This is one point to take into account before accepting work with a program that the client is offering to provide.[14]

10. Translation Memory Products

There are many different types of products on the market that feature translation memory. Translation memory technology can be found as a product in and of itself or as part of a localization or machine translation product. Following are lists of the various types of products in which translation memory technology can be found. See Table 3 in the Appendices for more detailed information on standard translation memory software.

10.1 STANDARD TRANSLATION MEMORY SOFTWARE

The following are considered standard translation memory products because their main function is to create translation memory databases. They do not provide machine translations and are less frequently used for localization than the products in section 10.2.

¨ Déjà Vu from Atril Software

¨ Joust (TSS) from Alpnet International

¨ Eurolang Optimizer from LANT Technology

¨ Translation Manager (TM/2) from IBM

¨ TRANSIT from STAR AG

¨ Translator's Workbench from TRADOS Corporation

10.2 LOCALIZATION SOFTWARE WITH TM

The following products were specifically designed for localization; however, translation memory is part of the process and is included in the software.

¨ Amptran from SDL

¨ CATALYST™ from Corel

¨ XL8 from GlobalWare

10.3 TM/MT HYBRIDS

These products consist of translation memory and machine translation systems. The first is a true hybrid of the two technologies. The second currently works with localization tools, but will soon have translation memory integrated into the software.

¨ T1 translation memory from Langenscheidt

¨ LOGOS (with XL8) from Logos Corporation

11. Finding Common Ground

Currently, there is no standardized format among the different translation memory solutions. LISA, the Localization Industry Standards Association, is developing a standard data exchange format for TM called TMX. TMX stands for Translation Memory eXchange. "OSCAR (Open Standards for Container/Content Allowing Re-use) is the LISA Special Interest Group responsible for its definition. The purpose of TMX is to allow easier exchange of translation memory data between tools and/or translation vendors with little or no loss of critical data during the process."[15]

Although a standardized format seems to be a good solution for translators who may use various TM systems throughout their careers, the developers of TM software are not as pleased. They feel that such a file format would have an adverse affect on selling their products. There are two major reasons for their concern:

1. Loss of brand loyalty among users

2. Increased competition among product developers[16]

If there were a universal file format, end-users could easily upgrade to a different TM system. Not only that, but then anyone could develop a translation memory system, knowing that they would only have to conform to one specific format instead of being forced to create filters to accommodate other TM products. These two issues are also reasons why TM developers do not condone the development of export utilities. Software developers prefer the idea of filtering TM databases or developing a converter to allow users to integrate TM databases from other products into their product.

12. FUTURE TRENDS

The future looks bright for all kinds of CAT tools. In fact, the future of translation is heading toward automated solutions. This doesn’t necessarily mean that machine translation will dominate the translation industry. While machine translation is improving, it is only part of the bigger picture. The trend is to integrate all of the different CAT tools and elements into one continuous process.

Translation memory will have a strong role in this process. Some companies such as Caterpillar are already using CAT tools and MT together in the translation process.[17] These companies send translations through TM for exact matches, then on to MT to fill in the blanks and then back to TM for integration into the TM database. From that point, the translators take over to post-edit and perfect the translation.

Other tools that fit into this process of integration include terminology databases, Web site management systems, word processing programs, graphics programs, etc. There may be a time when most of the non-translation related work that a translator does is no longer a major issue. Imagine a world in which one program automatically separates graphics and other non-text elements from the actual text. The graphics are sent off to a graphics program. Translation memory and machine translation programs work seamlessly with one another on the source text to produce a decent translation. The translator perfects the output and then reincorporates all of the non-text elements back into the fully translated, localized and formatted document.

Going one step further, imagine a program on the Internet that acts like a virtual translation management program. The translator or customer provides the source text for review. The “virtual manager” analyzes the document, determines what resources are best for translating the project in a cost-effective, efficient manner, provides the translator or customer with information on costs, Internet security and estimated translation time, and arranges for the document to be translated by various resources available over the Internet at the click of a button. Alis Technologies has already developed a very similar type of program called “Alis Translator for Lotus Domino.”[18] This on-line translation program offers a translation solution for Web and Lotus Notes^® applications. This solution can automatically publish Web pages in multiple languages and translate e-mail and other documents quickly using integrated machine translation engines. Other optional tools include dictionaries, accent tools, HTML filters and pre-editing tools. At some point in time, products like this will probably also include translation memory.

Some other solutions are being developed as we speak. One of these solutions is the TransRouter (or Translation Router) project. The TransRouter project “aims to build on existing translation technology and standard integration techniques in order to develop a tool which will help decision makers in translation agencies, service providers and other prospective user categories to make the most effective and appropriate use of translation technology tools and the best mix of human and computer-aided resources for a given set of documents. The tool will partly rely on data supplied by users, but will do much of its work automatically, based on computer analysis of characteristics of the text.”[19] This project is being coordinated by Berlitz Ireland Ltd. and includes the involvement of the University College Dublin, the Gesellschaft für Multilinguale Systeme, the University of Edinburgh, the University of Regensburg, the Center for Sprogteknologi and Suissetra (Institut Dalle Molle pour les études sémantiques et cognitives).

Does this mean that translators will be out of a job? According to Jeff Allen, a research linguist and translation lab supervisor at the Center for Machine Translation at Carnegie Mellon University, it is true that many large companies are looking into automated translation solutions; however, “this does not necessarily mean that the companies are trying to replace their valuable and experienced translation personnel with computers, but rather that the companies are trying to reconsider if they are using the time of their translation specialists most effectively and how they can improve it.”[20] Translators, agencies and corporations will have to adapt to the new technologies and solutions in order to remain competitive over the long term.

13. Conclusion

Translation memory is a technology that is here to stay. It has proven to be useful, especially in technical fields where electronic documents are constantly being updated and revised. Individual translators, translation agencies, clients, and companies with in-house translation divisions can all benefit from this technology. Translation memory saves the user time and money under the appropriate circumstances. The key for success is knowing when, where and how to use it.

Translation memory should be considered one of the many tools of the translation trade. A painter has a brush and palette; a pianist has sheet music. How painters and pianists use their brushes, palettes and sheet music is what counts. The same can be said of translators and translation memory.

Unlike the brush, palette and sheet music, however, not everyone recognizes that translation memory is an invaluable tool for the translator and that the translator, like the artist, is the one who must fit all the pieces together to form a work of perfection. The more that is written about this technology, the more translation professionals and their clients will be made aware of its important role in the translation process.

[1] EAGLES <http://issco-www.unige.ch/ewg95/node152.htm#spptmdesign>.

[3] EAGLES <http://issco-www.unige.ch/ewg95/node162.html>.

[4] Telecom Observer <http://telobs.com:8011/TelecomObserver/Spotlight/5/Welcome.E.html>.

[7] LANTRA <LANTRA@SEGATE.SUNET.SE>, “Survey.”

[8] Trados Corporation <http://www.trados.com/prod/casest/casest.htm>.

[10] Berry <http://www.mcbsys.com/html/news/10/tmowner.html>.

[12] Falcone <http://accurapid.com/journal/03TM2.htm>.

[13] Plumley, “RE: translation memory ownership.”

[14] Falcone <http://accurapid.com/journal/03TM2.htm>.

[17] Subject 36, “Survey: Advantages and Disadvantages of Translation Memory.”

[18] Alis <http://www.alis.com/atld/index.html?AlisFramesTgtDoc>.

[19] TransRouter <http://www2.echo.lu/langeng/en/le4/trouter/summary.html>.

[20] Allen, “TM and repetitive content (part A).”

BA, San Francisco State University, 1992

Fig. 1: Basic TM Process

Fig. 6: Exact and Fuzzy Matching

4.1.1 UPDATES

4.1.2 REVISIONS

4.1.3 “RECYCLING” PRIOR WORK