1.2 - CAT Technologies



Translation Memories
TRANSLATION MEMORIES (TMs) are databases that store source sentences and their translations as segment pairs (Figure 1) or TRANSLATION UNITS (TUs). A TM remembers each sentence that has been translated. If an identical or a similar sentence comes up later, it does not need to be translated from scratch. Instead, the translation can be easily retrieved from the TM database.This technology offers two main advantages:
- Less time and money is needed, as repetitive content can be translated much more quickly.
- Better consistency and quality: with TM technology you can make sure that you do not translate repeated occurrences of the same sentence differently, which can be confusing to readers.
__________________
Reflection #1: How do TMs help translators with maintaining high quality output?
Predictive Writing
Predictive writing is an additional helpful CAT feature when translating, that can be used in conjunction with terminology glossaries and automatic texts. Some say, that introduction of this technology is yet another milestone in the history of CAT tools, perhaps the most important after the introduction of “fuzzy matches” (see Unit 2).
While TMs store whole segments, the technology of predictive dictionaries is used for retrieving segment fragments. When you use such a dictionary, it will automatically suggest segment fragments while you are typing. For example, when you start typing the letter "D", it will suggest relevant expressions that fit the current context, e.g. "distribution of wealth in ..." . This feature is rather demanding in terms of processing power and is present only in the more advanced (and mostly commercial) CAT tools. Fortunately, SDL Trados Studio 2009 is one of these applications (from Unit 6 onwards). The only problem with the so-called AutoSuggest dictionary is its activation. It may only be created after your TM contains over 25,000 translation units. Even though this initial hurdle seems too difficult to surpass, there are ways to make the TM larger quickly (e.g. by importing TMs from other translators or using one of text-alignment tools - for more, see Unit 3)
__________________
Reflection #2: Predictive writing is not such a new idea, where else could you come across it? How do you think it works?
Terminology Databases
Terminology databases or TERMBASES (Figure 2) are similar to electronic dictionaries. Rather than storing entry words with lexicographic information, they contain single words or expressions typical for certain language variety. Creating and maintaining termbases is useful for storing organization-specific terms, which cannot be found in standard dictionaries.
Example: a termbase used for software localization may contain terms such as "dialog box", "dropdown list", "taskbar", etc.
Termbases can be accessed from within the translation environment. In this case, the source sentences are automatically searched for relevant terminology in the background. The corresponding target terms can then be easily inserted into the translation. This process is called ACTIVE TERM RECOGNITION (see Unit 8). This saves you the trouble of having to open a separate application in order to do time-consuming research work. Termbases offer the following main advantages:
- Cost and time savings, as you avoid tedious extra work such as researching and typing long-winded expressions.
- Quality increases, as the same expressions are used consistently throughout a document. Termbases help you effectively avoid unnecessary variations, which can be confusing to readers.
Reflection #3: How do you understand the term "software localization"? Why are TERMBASES so useful in the context of software manuals?