Generally, the term "language software" is used as a common term for any kind of computer program that deals with language processing and international word processing. Sometimes, you can hear the term "electronic translators", "foreign language translation" or "machine translation software" for software that puts words into another language. In this field, there is more confusion rather than transparency.HEISOFT Aktiengesellschaft is a worldwide-known publisher of language and translation software. Here, we would like to show you differences of software solutions and we would like to describe the main advantages of each of these products.
Machine translation challenges software developers. The task to translate entire documents requires a profound grammatical and syntactic analysis.
The market for translation software is very confusing, because pure phrase translators pretend to be a translation program for complete documents. But they only provide a word-by-word or phrase-by-phrase translation.Translation of entire documents cannot be done by a phrase-by-phrase processing. Natural language is too complex to be classified and translated in phrase categories. You should be very cautious when you see a so-called foreign language translation program. Being limited to a small field of applications, phrase translators often use such a name.
Actually, the market for real machine translation programs is quite clear. Probably only half a dozen of programs may be counted in the outstanding circle of machine translators for natural language translation.
However, there is serious confusion about the possibilities of computer-based translation software. Often the user has inadequate expectations regarding the make-up of documents.
Now and then, users assume that translation programs can be used for the economical translation of commercial memos or letters. Surely, such software is able to process this kind of document. But the expenses are disproportionately high. Machine translation only yields economical results when you need to translate large documents with a homogeneous terminology, i. e. contracts, documentations, reports, manuals, instructions...
As far as the usability is concerned, machine translation programs are laid out in different ways. Very often, it is useful to translate within a separate editor that shows the source document and the translation in two windows. This procedure offers the advantage that the user can compare source and target documents at a glance. At first sight, users see a disadvantage in the new environment and commands. Indeed, a human translator does not translate the sentences of a document phrase by phrase or word by word, but the translation is done within an entire paragraph. The user of a machine translation program should proceed in the same way.A comparison of the source document and the translation makes sense when you train your translator when filling up your user dictionary. In two windows (source and target), you can check the correctness of new words in the context of the sentence.
The integration of the final translation process into a word processor is desirable; that means that the user operates in his or her well-known word processing environment. Now the user has the possibility to focus on his or her document to translate and not on the translation software.
Translation for complete documents or automatic translation programs are called Ñmachine translation" (MT) by professionals. They require a powerful software that is able to analyze grammatical structures in a first step. Afterwards, they have to be processed on a sophisticated linguistic basis.
Broad dictionaries have to be equipped with linguistic information for each word. They are needed to ensure a clear classification of the sentence's components. In the context of a sentence, a word must be recognizable as an adjective or as a verb. Only by clear grammatical information he sentence can be processed as a unit. Of course, this procedure requires a sound vocabulary list with grammatical information in the background.
A structural analysis of syntax and grammar, however, by itself is not sufficient. Only a restricted number of modern machine translation programs designed for entire documents know to process specific characteristics of the meaning besides pure grammatical structures. By a content-sensitive division of words, important differences can be made. The translation processing of terms for animals or persons differs from the one for locations or things in a lot of languages. In this way, a certain plausibility check can be made within a translation. Usually you need a 32-bit operating system and an adequate performance of your PC.
On the one hand, a high software status is required. This is due to characteristics of languages that determine the different meanings and uses of a single word.
On the other hand, the user needs some basic information on language and translation process. Thus, it is helpful for the user when she or he understands the foreign language, above all when the target language is the foreign language.Especially with technical texts, a good command of languages is required. In general, translation systems use the integration of additional technical subject dictionaries when a technical text is to be translated; i. e. when the document includes many technical terms.
Usually, the translation system works with a General or Core Dictionary that serves as a data base for general words. In this way, you can obtain a satisfactory translation of documents without special vocabulary.
Subject specific documents can be for instance
By using subject dictionaries, the computer sets priorities for different lexical elements. These elements are fixed in these subject dictionaries. Furthermore, necessary extensions of the terminology base must be made. A professional machine translation process without the implementation of subject dictionaries can only be made on a restricted level.
Maybe it appears as an evident fact that the user has to co-operate with the translation software in a very close way to obtain good results. After all, computers are machines without any intelligence or the ability to think for themselves. But computers may process procedures based on logical structures. According to the state of today's technology, a machine translation software for entire texts cannot understand contents or interpret them. They lack common-sense information for which they depend on their users.
In this step, source documents are prepared for the machine translation process. Complicated phrase constructions have to be simplified and cleared. Ambiguous words are to be eliminated and to be replaced by unambiguous ones. Furthermore, spelling and punctuation are checked thoroughly, because the sole twisting of two letters can distort the result.
Moreover, the text becomes aligned as far as its terminology is concerned. By a terminological classification, we can let the software know to which technical field or special subject the text belongs and which terminology is mainly used. In practice, this is done by the use of external subject dictionaries or user dictionaries.
Step 2: The actual machine translation process
Now, the text is being translated automatically by the computer. In the beginning, this process is to be repeated several times. Unknown words will be identified by the program and prepared for being added to the user dictionary. After this step, you only have to repeat this procedure if you have a document with an obviously different terminology. In the course of your work with the program, your dictionaries will get so extensive that this step will be skipped later on.
Step 3: Post-editing (review)
Finally, the text has to be reviewed. Phrase constructions are altered if necessary, and single words are exchanged within the target text.
In step 2 the text was translated according to logical criteria. Now in step 3, it is refined by the user. The translation is ready.
Though the translated text is grammatically correct, it can happen that native speakers find the text strange or awkward when reading. The logic of the machine cannot fully grasp a natural language.
Hence, there is a serious conflict between real life and the nature of a machine. This conflict could only be resolved if the computer were able to think like a human being. This can be simulated by artificial intelligence. Nevertheless, according to scientists, we will definitely have to wait some decades for this.
In order to show you some problems of machine translation, we collected some demonstrative texts for you, done by Transcend G -> E.In practice, upon using a machine translation program, you will have problems you should know in order to obtain satisfactory results and to evaluate questionable results competently.
This error occurs very often. It is related to the ambiguity of words
or phrases. Being unable to understand a text document, the computer needs
the user's terminological classification. This can be achieved by editing
a user dictionary.
Here are some examples that show challenges for a good machine translation program. To make things easier, we would like to suggest some german sentences that can be translated to English. The sentences can be tested as they are without extension of the Core Dictionary.
Vielen Dank für Ihr Interesse an dem Übersetzungsprogramm Transcend.
German compound nouns must be written separately in English.
Ein wichtiges Merkmal von Transcend ist seine Integration in Textverarbeitungsprogramme.
In German, the position of the verb in subordinate clauses is different to its position in main clauses.
Manchmal müssen Dokumente übersetzt werden, die in Microsoft Word oder WordPerfect erstellt worden sind.
Some German prefixes of verbs (prepositions) are separated when declined.
Transcend behält das Format des Ausgangstextes bei. Das übersetzte Dokument sieht wie das ursprüngliche Dokument aus.
The machine translation program should observe retain the structure of complex phrases.
Die in den einzelnen Phasen der Entwicklung und Fertigung anzuwendende Sorgfalt darf nicht unterschätzt werden.
A personal pronoun used in the second clause of a sentence can refer to a noun within the first clause.
Der Mann trägt den Hut, den er gestern gekauft hat.
Here you can see some examples for imperatives as you can find them in technical instructions.
Starten Sie Microsoft Word.
Warten Sie, bis die Auswahl angezeigt wird.
Im Datei-Manager löschen Sie TRANSCND.WPT und TRANSCND.WPD aus dem WordPerfect Vorlagenverzeichnis.
Nach diesem Schritt entfernen Sie die Schrauben.
Bevor Sie das Programm verlassen, schliessen Sie die Datei.
Schliessen Sie die Datei, bevor Sie das Programm verlassen.
Vergessen Sie nicht, den Stecker zu ziehen, bevor Sie Ihre Arbeit beginnen.
Würden Sie ein anderes Material bevorzugen?
Gibt es eine bessere Lösung?
Gibt der Mechaniker unserem Mitarbeiter die Information?
Hätten wir ihnen dieses Projekt ankündigen sollen?
Werde ich die Firma anrufen, obwohl ich keine Zeit habe?
Rufe ich diese Firma an, weil sie meine Nummer nicht haben?
All products quoted are trademarks or registered trademarks of their respective holders.