Translation Memory Attributes Wordfast Classic

The TM Attributes tab displays five attributes, four of which can be customised, the first attribute being reserved for the User ID (User initials and name). I recommend reserving attribute #2 for Subject, and attribute #3 for Client, as in the example provided in WFC, to facilitate the interchange of TMs. You remain free, however, to define attributes according to your own needs. Use the Sample button to load a set of typical attributes, which you can then customise.

Click in the desired attribute list to customize an attribute's name using the Enter key to change values.

Click in the lower drop-down list to add attribute items (also called attribute values) using the following keys:

Insert or +, to add an entry;

Enter, to edit an entry;

Delete or -, to delete an entry.

The active attribute value is the one currently displayed by the lower drop-down list.

Entries consist of a mnemonic (an abbreviation, made of 2, 3 or 4 letters) followed by a space, then the narrative. WFC will record only the mnemonic in the individual TUs, to minimise redundancy.

Note: The first attribute is always the "User ID" attribute. By default (if you don't specify a User ID or name), the value for this attribute is the current Ms-Word user initials and name, as they are found in Ms-Word/Tools/User info. You can, however, customize this User ID as you wish. If the TM was used by other users, the drop-down list will show all the translators who have used the TM in the past (maximum number of translators: 60). If you workgroup, this feature lets you see the TM's pedigree.

Attributes are stored in the current WFC setup - the INI file. When working in a translation session, WFC will record the mnemonics of the set of the currently active attribute values into any new, or updated, TU. If you stop the translation session, open WFC and change active attributes values, the TUs generated in the next translation session(s) will receive the new set of attributes values, but the attribute values of the previously existing TUs are not affected.

Applying penalties based on attributes

Penalties are numbers entered between parentheses (see the sample attributes for examples). A penalty lowers the percentage of match rate of a TU when it is found in the TM (if WFC finds a 100% match in the TM, but one of the TUs attribute values has a penalty of 5, the match rate will be lowered to 95%).

There are two types of penalties: absolute penalties and relative penalties.

Absolute penalties: are defined for attribute values (i.e., items in the drop-down list). When WFC proposes a TU which has that attribute value, it will receive the corresponding penalty.

Example: your translator ID is JTB John T. Bisham. You import, in your TM, 200 TUs coming from another translator whose ID is MAT Mark A. Tweed. You wish to unconditionally apply a penalty of 5 to propositions coming from TUs created by Mark Tweed.

Create or edit the

MAT Mark A. Tweed

attribute entry so it reads

MAT Mark A. Tweed (5)

From then on, every time a proposition comes from a TU created by Mark Tweed, it will have a penalty of 5. As a result, a Mark A. Tweed TU will never appear green, it will appear as 95% match at best.

Relative penalties: are defined per attribute (in the attribute caption). These penalties will be applied if the particular TUs attribute value is different from the attribute value of the current session (as you defined it in WFC's TM Attributes section).

Example: you apply a relative penalty of 8 to the User ID attribute. Edit the User ID caption so it reads User ID (8) . From then on, if a TU's User ID is different from the one currently defined - supposedly your ID - then the TU will receive a penalty of 8, regardless of which translator it is.

Absolute and relative penalties are cumulative. So, if Mark A. Tweed already has an absolute penalty of 5, and the entire User ID category has a relative penalty of 8, then a TU with Mark A. Tweed will receive a total penalty of 13.

The basic purpose of penalties is that a TU, which would otherwise appear green, does not appear green but yellow, so that the translator's attention is drawn at that point. Penalties should be modest (a penalty of 2 is enough to prevent a TU from appearing green), because, if they are cumulated, they may actually bring the match rate below the fuzzy threshold. Penalties for TUs created by machine translation, however, are traditionally strong (10 to 15).

One other purpose of the Attribute system, using the TM/glossary editor utility is to manage (extract, merge, classify etc) TMs by taking into account their TUs' individual attributes.

  Back to Wordfast Classic User Manual