Free translation tools

Translation memory tools are considered one of the most useful tools for professional translators and Language Service Providers (LSP), but also for non-professionals who frequently need to translate texts.

Surprisingly enough, although they’re widely available on the market since the Nineties, not every translator has embraced this technology, while most translation companies do include at least one of them in their workflow. This resistance is partly justified by the high cost of software licences, partly due to legitimate doubts, but also based on a few misconceptions.

Are there any free or open source alternative that can really compete with paid solutions, are easier to use, can really enhance a translator’s or translation company’s productivity and help a non-professional in their translations?

Indeed, there are a few, although they are aimed at the most tech-savvy users. We can speak with full knowledge of the facts, because we have been using one of these tools for more than 10 years.

What are translation memories?

One of the most useful results that the IT+Internet revolution brought to the translator’s office was the advent of translation memories.

A “translation memory”, usually abbreviated as TM, is nothing more than a database of translation units. It’s called a “memory”, because it remembers past translations. Oversimplifying, translation units are segments of text in 2 languages and correspond - more or less - to sentences, according to specified punctuation rules.

Translation memories are used by assisted translation software, or more specifically, by its subset called Translation Memory Managers. Thanks to an algorithm, when translating a text the software matches the sentence to be translated with the sentences found in the database, serving one or more existing translation that can be accepted, edited or discarded.

Within the same program, you can also look up existing terms from a TermBase (TB) and even pretranslate the text with the help of the machine translation engine of your choice.

Advantages of translation memories

Typical advantages of translation memories and CAT/TMM tools for professional translators are:

• No sentence can be left untranslated, because CAT tools don’t allow empty target sentences
• Better coherence of the document: CAT tools retrieve the same target text for identical source text
• Higher productivity, because identical segments need to be translated only once, while similar (fuzzy) segments are also partially translated by the system

This last advantage is clearer when translating the second version of a text, where most of the content remains the same.

What features are included in a CAT tool?

Some prefer to refer to Computer-Aided Translation Tools as Translation Environment Tools (TEnT), because they include more features than just the matching algorithm.

Modern Computer-Aided Translation tools usually include:

• Translation Editor
• Translation Memory
• Term Database (or TermBase, abbreviated to TB)
• Concordance / Search tool, to search within the TM
• TM Editor, to edit translation memory contents
• Machine translation connector, showing suggestions from one or more machine translation engines
• Alignment module, to create a translation memory from translated documents without a CAT tool
• Reference material module, showing documents related to the translation project
• Project management module or functions, for the organization of the project
• Dictionary module or connector, showing suggestions from one or more electronic dictionaries
• Spell checker, similar to that found in common word processors
• Quality assurance module, which reviews the text and flags common errors

Disadvantages of translation memories

Despite all the advantages, a few freelance translators reject the use of translation memories. You can still find complaints on the matter on specialised forums and newsgroups. Detractors base their opinion on several points, some of them quite reasonable, some not so much:

• Full versions of the most used CAT tools have expensive licence costs. Compared to other professional tools, their price seems unjustifiable to many translators. It’s also quite difficult to calculate a proper ROI, because a translator’s productivity depends on several factors and because there are no certified metrics except those supplied by the CAT tools providers themselves.
• The most used CAT tools used to have (and some still have) clumsy user interfaces. Their learning curve is rather steep. Generally speaking, without proper training it is quite difficult to get the most out of these programs.
• CAT tools and translation memories are best suitable for technical texts which contain repetitions and are usually subject to frequent updates. Translators working in different fields are less likely to take advantage of these tools.
• Text segmentation based on sentences is a concept that doesn’t work equally well for all languages. Translators working in exotic language pairs may find TMs not as useful.
• Segmentation and the matching rate found in the database have introduced a new way of rating translation jobs. Segments already present in the TM (100% matches and fuzzy matches) are paid less than no matches. Sometimes 100% matches are not even paid. Some freelancers are reluctant to accept this way of invoicing, seeing this practice as an unfair way of taking advantage of translators.

• The issue related to 100% matches also connects to translation memory maintenance. If a wrong translation has been recorded in the translation memory, due to poor or no proofreading, it can be reused and perpetuate the error. A proper maintenance of translation memories or a proper usage are necessary to avoid such situations. If 100% matches are not paid, they are unlikely to be revised by translators consequently increasing the risk of finding the same mistakes in new translation.
• Ownership of TMs and legal issues. Many translators and translation agencies don’t feel comfortable working with TMs or even creating them from scratch, because it’s not clear at all who will own the TMs.

Free translation tools

In our most recent market survey we found only a few Free and Open Source (FOSS) options which, though not as feature-rich as paid tools, can guarantee a quite complete working environment.

These CAT tools are:

• OmegaT (desktop program, cross-platform)
• Lokalize (desktop program, Linux only)
• BasicCAT (desktop program, cross-platform)
• MateCat (online program)

Now, we’ll only focus on OmegaT, the tool we at Qabiria use on a daily basis; a viable option not only for the professional, but also for non-professionals who need to translate a document for study, work or leisure purposes.

OmegaT, a free and open source translation tool

Although lacking some features found in commercial solutions, OmegaT can be a real alternative to commercial software especially when combined with other FOSS tools (such as the Okapi Framework). Conversion processes between formats can be quite hard to follow for the less tech-savvy users, but with proper training and the help of the community (forums for FOSS CAT tools are characterized by their responsiveness) these difficulties can be easily overcome.

How to choose a CAT tool

However, choosing a FOSS or a commercial tool depends on the scenario you are facing. The translator may found themselves in one of these 3 situations:

1. The translator is forced to use a CAT tool since it is a supplier of a third party company by a larger Multi Language Vendor (MLV).
2. The translator decides to use a CAT tool to leverage their past translations and increase productivity.
3. The translator only want to use a CAT tool for learning or testing

1) Larger MLVs are used to send translation projects to smaller LSPs or freelancers according to the tool of their choice, frequently disregarding potential compatibility issue. That leaves smaller LSPs with little or no choice: they must use whatever CAT tool the larger LSP is using. Negotiations in order to get open standards formats (for instance, requesting to receive XLIFF source files and TMs in TMX format) are often useless. It is unlikely that larger companies modify their processes in order to suit smaller providers’ needs.

If you face this scenario you can:

• use a “satellite” version of the required CAT tool, if any (“satellite” versions are stripped-down versions with fewer features and usually can be only used to work on an existing project rather than create a project from scratch; they can be free or cost only a fraction of the full licence)
• embark in usually complex and not totally fail-proof conversion processes, in order to convert the received packages into open standards usable with FOSS CAT tools
• acquiring a full licence of the required CAT tool

2) If the translator decides to use a CAT tool to leverage their past translations and increase productivity, they should carefully analyse which file formats are the most common in their activity. If those are supported by one of the FOSS CAT tools, our advice is to start with one of them, before investing a considerable amount of money in a commercial solution. The saved money can be invested in training one or more “power users”, who will subsequently train the rest of the team.

3) FOSS CAT tools are also extremely useful to test this technology and personally experience the involved benefits. After a learning and training phase, you can make a more informed choice and buy the most appropriate tool for your specific activity.

Like many other open source projects, open source CAT tools sometimes lack well-written or prompt up-to-date documentation. However, in most cases they provide community-run support. Since inexperienced users may find it difficult to approach this kind of software, our advice is to invest in training. It’s true that there are lots of scattered resources around the web, including good and easy-to-follow tutorials, but personal contact with real-life trainers either on site or online can save time and effort when first coping with the software or when looking for more advanced features.

All in all, FOSS CAT tools, combined with other FOSS solutions, are already an alternative and will be more and more in the future.