When defining project activities, mark these dependencies clearly.

Sequence

Once all activities and dependencies have been listed, put the activities in a logical sequence, carefully considering all dependencies. An example of a typical localization project sequence can be found in the Localization Process section on page 17 of the Introduction.

Here is a typical localization project sequence:

1. Identification of terminology and of target language term equivalents

2. Software translation

3. Translation of online help and documentation

4. Engineering and testing of software

5. Screen captures

6. Engineering and testing of online help

7. Desktop publishing

8. Quality assurance and delivery

Sometimes it may be necessary to start translating help and documentation before the software is translated. In this case, instruct the translators to translate all software references and to create a glossary of these translated references. This glossary can then be used to translate the software resource files automatically.

If the project contains sample files or templates, it is advisable to start translating these in conjunction with the software files. These types of files typically contain sample data that can be translated by a translator other than the translator working on the software.

Screen captures can be started early in the project because screen captures in help and documentation often contain graphics with data from these sample files.

Note that many activities can also be performed simultaneously. Project preparation can be done while terminology lists are being prepared, for example, online help engineering can be done while documentation is being translated, and screen captures can be created while the functionality of the software is tested.

Workload and Duration

Define the workload and duration of each project activity. If project evaluations have concluded that the online help file contains 10000 words for translation, for example, estimate the length of time it would take to translate these words, edit these translations, and proofread the final output.

There are no standard metrics for productivity in translation and localization, because too many factors can influence productivity levels, such as:

• Complexity of the source material

• Quality of the source material

• File formats of the source

• Tools used

The use of translation memory or machine translation tools can increase productivity significantly. Productivity normally increases during the project, because translation memories generate an increasing number of matches as each component is translated, or as additional products are localized.

Most localization vendors assume that a translator can translate approximately 1500 words of software per 8 hour day, and 2000 to 2500 words of running text in online help or documentation per day. For Asian languages, these numbers would generally be lower, because of the complexity of entering and processing Asian characters. These estimates are only guidelines on new text, and exclude fuzzy or full matches retrieved from translation memory. Editing translations accounts for 20 to 30% of the time required to produce the translations. Most proofreaders can work through approximately 2000 to 3000 words per hour. These estimates may vary depending on the complexity of the subject matter and the target language.