TranslationManager is designed to address the serious problems faced today by commercial enterprises, public-sector organizations, translation companies and many freelancers in connection with natural language translation. These include: high unit costs, rising volumes, and long overall production times. TranslationManager is designed for use in a single-user PC or LAN environment. Documents can be processed by different translators or terminologists simultaneously, whereby each member of the team has access to the latest translation and terminology information, thus assuring highest translation consistency and integrity.

A customer value analysis can be used in conjunction with TranslationManager which allows the customer (by applying system, time, and cost based evaluation metrics) to determine a priori productivity gains, quantifiable savings per translator and per document as well as in terminology creation and updating.

TranslationManager provides the following six major components:

An editor that is full-featured yet easy to use for the translation process.

The source language text, the Translation Memory, and the dictionary are displayed in three adjustable windows on the screen of the user's computer.

The user creates the target language text by superimposing the translation on a copy of the source text in the target window. In two additional windows, the Translation Memory and dictionary run parallel to the source text segments. In these windows, the matching Translation Memory and dictionary entry translations are, respectively, automatically displayed. On request, the complete dictionary entry of a term can be displayed using the keyboard or mouse. In addition to translation-specific editing commands, the user can "paste" information by one hot key stroke from the Translation Memory and/or dictionary window into the target text window. Any newly translated segment is automatically added to the Translation Memory, thereby increasing the knowledge base of the system.

Markup language tags are recognized during translation. These formatting tags may be protected, unprotected or hidden during the translation or editing, and can be used unchanged in the target text, thus saving re-layouting and simplifying the translation process.

Document formats currently supported are:

• WordPerfect
• AmiPro for Windows
• RTF, e.g. for Microsoft Word
• Ventura Publisher
• Interleaf
• Plain ASCII
• Text in program files
• IBM DCF and BookMaster
• IPF

For the support of additional formats, the IBM development group can be contacted. In addition the above formats, the user can specify his own markup language tables for the system to recognize the tags of other markup languages.

A terminology management tool designed to support the creation and maintenance of up-to-date terminology data bases.

These files can be personalized glossaries tailored to the requirements of individual translators and/or specific texts, master dictionaries developed for the use of an entire translation staff, or dictionaries compiled with an aim directed toward publication and dissemination to a wider audience. TranslationManager allows the use of read- write dictionaries, that is, dictionaries that can be modified by any user, as well as dictionaries that are in read-only mode (password-protected).

The user can specify a dictionary hierarchy, that is, an order in which the dictionaries are to be searched for when looking up a term. He can also switch between different dictionaries, that is, display dictionary entries from the same or different dictionaries in a "cascading" panel sequence.

Dictionary and terminology data bases that use formatting tags based on SGML can be imported into the TranslationManager environment or exported from it.

The text analysis component allows the generation of (1) a list of terms found in the specified dictionaries, a so-called text-related glossary, and (2) a list of terms which are not found in the specified dictionaries and which can serve the terminologist as the principal source of input for dictionary updating.

A Translation Memory which renders automatic retrieval of previously translated segments whose source text is identical or similar to the current text to be translated, permitting the user to reuse the previous translation.

This Translation Memory helps to ensure the consistent translation of identical or similar text and thereby expedite the translation process. When translating a segment, the Translation Memory database is searched to find exact or similar (fuzzy) segments matching the current (active) one.

If one or more matches are found, the corresponding target segment is displayed in the Translation Memory window. The user can retrieve the appropriate segment with one keystroke, that is--copy the selected Translation Memory proposal into the current translation.

The user can also have the system automatically replace the source text globally with the translations from the Translation Memory, interactively during editing or in batch mode prior to his translation task.

A component to generate initial Translation Memories from existing translations.

This tool is used to align parallel bilingual texts of previously translated documents in a variety of formats, creating the Translation Memory automatically. This allows productive work with TranslationManager from the beginning of an installation, since Translation Memories already exist for the new source files to be translated.

Administrative functions to control the job flow within TranslationManager ranging from source text preparation to delivery of a finished translation. Administrative functions include:

• Creation of folders with source texts.
• Creation of a Translation Memory database.
• Text analysis to create text segmentation and list of terms found in the dictionary as well as lists of words not found.
• Importing and exporting text files, dictionaries, and Translation Memories.
• Spell check operations.
• Word counting and translation completion percentages.

On-line tutorial and help to enhance the user-friendliness of the system. TranslationManager offers an integrated tutorial, which teaches you how to perform the tasks of a translator, terminologist, and translation coordinator using TranslationManager, and provides extended help as well as specific field help at any time during translation.

Language Support

A unique capability of TranslationManager is the built-in language support. This includes automatic reduction to stem forms, automatic sentence segmentation, and spell checking for the following source and target languages:

• Catalan
• Danish
• Dutch preferred
• Dutch modern
• English, U.K.
• English, U.S.
• French, Canadian
• French, national
• German, Swiss
• German, national
• Greek
• Icelandic
• Italian
• Norwegian, Bokmol
• Norwegian, Nynorsk
• Portuguese, Brazilian
• Portuguese, national
• Spanish
• Swedish

As a target language, any operating-system-supported language can be used including Russian, Japanese, Korean, and Chinese.

Translation Quality Measurement

This program operates in conjunction with TranslationManager to perform functional and linguistic quality assessments of the completed translations, and allows users to evaluate the quality of different categories of translations, such as publications or software text, on the basis of given quality guidelines buttressed by parameterized statistics. Its goal is to determine the quality of individual translations using a quality assurance system according to ISO 9001 and to monitor the quality trend over a period of time.

Interface to Automatic MT

TranslationManager allows interfacing to automatic translation systems. To accomplish this, the segmented source text is passed on to the machine translation system for automatic translation. Subsequently, the target text is converted by the system into the SGML format of the Translation Memory, passed back to TranslationManager and made available as an MT-flagged Translation Memory.

Interface to Information Retrieval

TranslationManager can be linked to an information retrieval program, such as IBM's SearchManager, to facilitate precise searching and rapid retrieval of electronically stored documents preparatory to a translation task. Thus, former source documents can be retrieved together with their translations on the basis of one or more search keys, and an initial Translation Memory (see item 4 above) can be generated from those related documents ad hoc to be used for translation with TranslationManager.

Interface to Speech Recognition

An interface exists to the IBM Personal Dictation System as well as the IBM Speech Server Series allowing a combination of automatic replacement of source language text by previous translations in the Translation Memory and dictation of the as yet untranslated text passages by the translator. That is, after TranslationManager has matched the source text against the Translation Memory and replaced the source by the corresponding target text found, the translator dictates the translation of the remaining source text into a microphone connected to the PC. The speech recognition system recognizes the dictated words and displays them on the screen, overlaying the corresponding source text. After it has been dictated and checked for accuracy (by issuing corrections or oral commands to the text displayed), the translation can be passed directly to other applications on the computer, such as printing or transmission over a telecommunications line.

operating System Requirements

The following system configuration is required to run TranslationManager:

• Any 386 system unit (including DBCS supported systems)
• System memory of 8 MB
• Disk storage of at least 25 MB free

TranslationManager will run on different operating system platforms. When using TranslationManager in a client/server mode in a LAN environment, the IBM LAN Server and Requester software of OS/2 Version 2 is required.