The translation process was described to be one of the most effective that is clearly a method of communication especially among civilizations of different dialects. Translation as a concept has existed century ago, but it is merely during the second fifty percent of the twentieth century which it emerged as an independent academic. An awful dependence on translation has prompted specialised and theorists in the field to get for more advanced methods and techniques for quick, cheap and effective translation. Thus, a new type of translation has seemed to compete with People Translation to create Machine translation or the programmed translation.
Nowadays the utilization of machine translation is vital than we might think, because different aspects of modern life have immediate for more efficient ways of translation, thus the demand for translation is not satisfied, because there are not enough real human translators, or because individuals and organizations do not understand translation as a complicated activity requiring a higher degree of skill, and therefore they are not prepared to pay what it is worth.
This research tries to compare the most important linguistic areas of machine translation and also to review its main problems.
The purpose of the given research is to investigate the down sides of machine translation.
The hypothesis that people postulate because of this research is that the interlingua procedure display the greatest degree of difficulty in the process of translation.
The specific aims of the study are:
to define the notion of Machine Translation;
to identify and compare different machine translation techniques;
to analyze the primary problems of machine translation;
The research methods employed in the task are analysis, which was used for the analysis of machine translation and identifying its essential features; diachronic analysis, that focuses on historical development of machine translation; the classification method was used for classifying the strategies of machine translation and their problems of ambiguity.
We selected this theme, because the device translation is an instrument that allows visitors to have information about a variety of things in several languages and to understand it without knowing the terms. Furthermore it allows us to really have the meaning of a term or expressions in a rapid and effective way. As well Machine Translation provides translators useful tools that help them to make their job more successfully and faster.
The most important sources which may have been used are: "Concise History of the Terminology Sciences: from the Sumerians to the cognitivists" by Koemer E. F. , "An Advantages to Machine Translation" by W. J. Hutchins and Harold L. Somers, "Introducing Translation Studies: Theories and Applications" by Munday J. , "Machine Translation" by Maegaard B. , and "Machine Translation: An Introductury Guide" by Arnold D. J,
Language is the major method for folks communicating with one another, but people can only communicate the other person with language they both know. Alas there are around 7000 different sorts of languages on the globe, and these languages may have different writing systems, grammar and pronunciation. Alternatively, the fast grows of international communication (such as international businesses, countrywide diplomacy, and international conferences) making the demand of translation (such as business record translation, legal report translation and methodical and technological documents translation) is also growing speedily, cheap and fast translations are needed. In this case machine translation becomes a solution.
Identifying different explanations of Machine Translation
Machine translation of natural dialects, often called MT, has multiple personalities. Sergei Nirenburg and Yorick Wilks, in their book "Machine Translation" declare that, to begin with, machine translation is a venerable technological enterprise, a component of the bigger section of studies concerned with the studies of individuals language understanding capacity.
They write that MT is also a scientific task of the first order. It includes an possibility to test the knowledge of the syntax and semantics of a number of languages by encoding this vast, though rarely comprehensive, knowledge into a form suitable for control by computer programs. Also in this e book "Machine Translation" they say that MT has a strong connection with the needs of modern societies. It can be realized as an economical necessity, due to the fact the expansion of international communication keeps intensifying both at government, for instance, EU, NAFTA, GATT and business and business levels, for case, the exporters need product documents in the languages of the countries where their products are promoted [12].
In the article "Brief Record of Machine Translation Research" Leon Dostert mentions that the storyline of the genesis of machine translation was traced with care in the first compendium of essays on the subject entitled Machine Translation of Dialects, edited by William Lock and A. Donald Booth. In which they write that the transference of interpretation in one patterned group of signs occurring in a given culture into another group of patterned signs happening in another related culture through an electronic computer [7].
In the record "Language and Machines Pcs in Translation and Linguistics" is explained that machine translation means that it will go by algorithm from machine- readable source text to useful focus on text, without recourse to human being translation or editing and enhancing [1].
In "An Launch to Machine Translation" W. John Hutchins and Harold L. Somers clarify that the word Machine Translation is the original and standard name for computerized systems in charge of the development of translations in one natural terminology into another, with or without individuals assistance. Machine translation can be known as as mechanical translation and computerized translation. These conditions are now almost never used in English, but their equivalents in other dialects are widely-used, for example in French traduction automatique, or in Russian † †. Also in this publication is written that the term does not include computer-based translation tools which support translators by providing access to dictionaries and distant terminology databases, facilitating the transmission and reception of machine-readable text messages, or interacting with word processing, word editing and enhancing or printing equipment, but, however, it offers systems in which translators or other users assist personal computers in the creation of translations, including various combinations of content material preparation, on-line relationships and following revisions of end result [16].
M. Kay and Xerox Parc in their article "Machines and People in Translation" write that we should differentiate a narrower and a wider use for the word machine translation. In the narrow sense, the term identifies a batch process when a text is given over to a machine from which a result is collected which is the end result of the device translation process. Whenever we use the word in the wider sense, it offers all the procedure necessary to obtain last translation output in some recoverable format [8].
In the article "Machine Translation Workstation" is explained that the MT is a general tree-manipulation system with several built-in inference strategies. They demonstrate the procedure of machine translation through the next scheme:
And they state that whenever a user is applicable the machine he/she writes a guideline base to control the execution of the device and chooses the correct inference strategy. The machine can take well-defined linguistic trees and shrubs as source and produces as result trees which symbolize meaning-preserving transformations of the source trees and shrubs. Furthermore the MT is vocabulary independent, since it impose limitations on what sorts of transformations are possible [4].
In conclusion we can say that machine translation can be an programmed linguistic translation", particularly, a word-by-word translation and it identifies the utilization of software to translate wording from one terms to another dialect.
Machine Translation Strategies
In this article "Machine Translation and Computer-Assisted Translation" Craciunescu says that Machine translation can be an autonomous operating system with strategies and approaches that can be classified the following:
the direct strategy
the copy strategy
the pivot terminology strategy
She says that the immediate strategy is dependant on a predefined source language-target words binomial in which each word of the foundation terminology syntagm is directly linked to a corresponding product in the target terms with a unidirectional correlation, for example from English to Spanish but not the other way around.
But the copy strategy is based on the amount of representation and involves three stages. The analysis stage describes the foundation file linguistically and runs on the source terms dictionary. The transfer stage transforms the results of the evaluation level and establishes the linguistic and structural equivalents between your two languages. It uses a bilingual dictionary from source words to target terms. The generation level produces a file in the prospective language based on the linguistic data of the source language by means of a target terminology dictionary.
The pivot terms strategy is predicated on the idea of setting up a representation of the text unbiased of any particular words. This representation functions as a neutral that is distinctive from both source language and the target language. This method reduces the machine translation process to only two periods: evaluation and generation. The examination of the foundation text contributes to a conceptual representation, the diverse components of which are matched up by the era module to their equivalents in the target terminology [5].
Another characterization of strategies of MT we find at W. J. Hutchins and Jonathan Sloculn in their articles "Machine Translation: A Brief History" and Its History, Current Status, and Future Leads" separate three basic strategies.
The first strategy is referred to the direct translation approach. Direct translation is quality of something designed from the begin to translate out of one specific terms and into another. For example, Russian is the words of the initial texts-the source terms, and English is the language of the translated texts-the concentrate on vocabulary. Translation is direct from the source language (SL) text message to the target language (TL) wording [14].
Arnold in his book "Machine Translation" presents the direct approach through the following scheme[3]:
Text SL
Direct Translation Text TL
The second basic design strategy is the Interlingua procedure, which assumes that it is possible to convert SL text messages into representations common to more than one vocabulary. Furthermore the Interlingua way is quality of a system where the representation of the meaning of the source language input will be self-employed of any terminology, which representation can be used to synthesize the mark language productivity [14].
In his book "Machine Translation" Arnold presents the Interlingua procedure through the next scheme [3]:
IL
Analysis
Synthesis
Direct Translation
Text SL Content material TL
The third basic strategy is the less ambitious transfer approach. The transfer approach is characteristic of something where the root representation of this is of any grammatical product (e. g. , sentence) differs with respect to the language from which it was produced or into which it is to be generated; this implies the existence of a third translation level which maps one language-specific interpretation representation into another: this stage is called Copy. The transfer approach performs through three levels involving underlying (abstract) representations for both SL and TL text messages. The first level converts SL text messages into abstract SL-oriented representations; the next stage converts these into equal TL-oriented representations; and the 3rd generates the final TL texts. Whereas the Interlingua approach always requires complete resolution of all ambiguities in the SL word so that translation into another language can be done, in the copy approach only those ambiguities inherent in the words involved are tackled; problems of lexical variations between languages are handled in the second stage (transfer proper) [14].
Arnold also signifies the third methodology, the transfer way, through a scheme as follow [3]:
Analysis IS SL
Transfer ISTL
Synthesis
TEXT SL Text message TL
Direct Translation
In simple, the interlingual machine translation is one of the traditional solutions to machine translation. In this approach, the source language - the written text to be translated is transformed into an interlingua - an abstract language-independent representation. The prospective terms is then generated from the interlingua. Furthermore, the interlingual way is an option to the direct way and the copy approach.
Main problems of machine translation
The major problems of all MT systems concern the quality of lexical and structural ambiguities, both within dialects (monolingual ambiguity) and between dialects (bilingual ambiguity). The lexical ambiguity is whenever a word has several meaning, however when a term or phrase can have significantly more than one structure it is called structural ambiguity [3].
Hutchins in his article "Machine Translation: Record and General Ideas" mentions that any monolingual ambiguity is a potential difficulty in translation since you will see several possible equivalent. For example, homographs and polysemes (English cry, French voler) must be fixed before translation (French pleurer or crier, English fly or steal); ambiguities of grammatical category (English light as noun, adjective or verb, face as noun or verb) must moreover be settled for choice between lumiЁre, clair or allumer, etc. He declares that the examples of monolingual structural ambiguities happen when a word or phrase could modify more than one element of the phrase. And he points out this through the next example, old men and women, the adjective old may send and then men or to men and women [15].
Prepositional phrases can adjust almost any preceding verb or noun phrase,
e. g. (a) The car was motivated by the tutor with great skill.
(b) The automobile was powered by the tutor with defective tyres.
(c) The automobile was influenced by the tutor with red hair.
Lexical and structural ambiguities may and often incorporate: He observed her shaking hands, where shaking can be either an adjective hands that have been shaking or a verb part that she was shaking hands [15].
Bilingual lexical ambiguities happen generally when the TL makes distinctions absent in the SL: E. g. British river can be riviЁre or fleuve (Fluss or Strom);
English eat can be German essen or fressen;
English wall membrane can be French mur or paroi, German Wand, Mauer or Wall membrane.
Hutchins implies that an example which can inllustrate this is actually the translation of wear from British to Japan. Although there's a common verb kiru it is normal to make use of the verb appropriate to the kind of item worn: haoru (jacket or jacket), haku (shoes or trousers), kaburu (head wear), hameru (band or gloves), shimeru (belt, tie or shawl), tsukeru (brooch or clip), kakeru (spectacles or necklace), hayasu (moustache) [15].
Also in this article is remarked that the bilingual structural variations cover both basic facts, for occasion, in English the adjectives generally precede nouns but that they often follow them in People from france, and dissimilarities conditioned by specific lexical differences. A familiar example occurs when translating the English verb like She loves to play tennis, as a German adverb gern Sie spielt gern Tennis games [15].
Other illustrations are:
simple verbs (trust) rendered by circumlocutions (avoir confiance );
single clauses He pushed open the entranceway restructured as a subordinate clause Il a ouvert la porte en la poussant [14].
The structural distinctions combine with lexical variations, for instance the translation of know into French or German, where selection of connatre (kennen) or savoir (wissen) influences both structure Je connais l'homme. (Ich kenne den Mann); Je sais ce qu'il s'appelle. (Ich weiss wie er heisst) and the translation of other lexical items (what as ce que and wie) [14].
The morphological examination can be involved with the id of base varieties from infected forms of nouns, verbs and adjectives (irregular forms being came into as items in dictionaries), with the identification of derivational forms (e. g. English -ly as an adverb produced from an adjective, German -heit as a noun from an adjective), and with the segmentation of substance forms in dialects like German (Dampfschiff, Dampfhammer) [14].
In the "An Release to Machine Translation" Hutchins unveils that MT systems have problems with 'unfamiliar' words, especially with the neologisms and new combinations. He says that if derivational elements and components can be properly discovered then can be translated with the 'international' equivalences of several elements, for case, French demi- and English semi-, France -ique and English -ic) [16].
However, segmentation can be problematic, e. g. extradition analysed as both extradit+ion and former mate+tradition, cooperate as both co+operate and cooper+ate. He suggests that these would be solved by dictionary discussion, but sometimes alternate segmentations are equally valid (German Wachtraum could be shield room (Wacht+Raum) or day goal (Wach+Traum), until you are eliminated at a later stage [16].
In his article "Machine Translation: A BRIEF OVERVIEW" Hutchins writes that in MT there are three basic methods to syntactic structure research. The first aim is to identify reputable sequences of grammatical categories, for case, in British article + adjective + noun. This process is dependant on predictive analysis, which really is a sequence of categories forecasted that the following
category would be one of a comparatively limited set. The second aim to identify groups of
categories, for example, as noun phrases, verb phrases, clauses, and eventually sentences. These are based on expression framework or constituency grammar. The third aim to identify dependencies among categories, for example, reflecting the actual fact that prepositions determine the truth types of German and Russian nouns, that the form of a French adjective is determined by the noun it modifies. The basis is dependency sentence structure [14].
He also remarks that SL constructions are changed into comparative TL constructions by conversion rules, regarding phrase composition or dependency trees and shrubs by 'tree transducers', which may apply either unconditionally, for example, British adjective+noun to French noun+adjective or conditionally, followed by specific lexical items, for example, English prefer to German gern [15].
Another problem which identifies Arnold is the multiword products like idioms and collocations. The real problem with idioms is they are not generally set in their form [3].
Hutchins in his article "Machine Translation: Background and General rules" highlights that MT systems can are unsuccessful for many practical reasons, for case, mysterious words neologisms or new ingredients, misspellings supercede, persue, British orthography rather than expected American traveller for traveller, typographical problems from rather than form, wrong usages principle as an adjective, ungrammaticalness do not require were present. Although full disambiguation cannot be achieved, a crude translation may be obtained with basic phrase structure identification. It is now common for systems to hold on to information from all degrees of analysis; thus copy (or interlingual) representations will combine morphological, syntactic, semantic and thematic information [15].
Historically, MT systems have steadily introduced 'deeper' levels of analysis and
transfer. Early on word-for-word systems were restricted to bilingual dictionaries and simple
morphology. Later 'direct' systems presented syntactic evaluation and synthesis. Term composition and dependency analyses provided the basis for simple copy systems with little semantic examination.
Conclusion
The use of machine translation is more important than we may think. It could be said that the resources available to the translator through it imply a big change in the partnership between your translator and the written text, that is to say, a fresh way of translating. However, you have the development of new features, which leads us to indicate a number of essential areas of the current situation. Translating by using the computer is certainly not the same as working exclusively in some recoverable format and with newspaper products such as classic dictionaries, because computer tools provide us with a relationship to the text which is much more flexible when compared to a solely lineal reading. Furthermore, the Internet with its common access to information and instant communication between users has created a physical and geographical freedom for translators which were inconceivable before. Translators need to accept the new solutions and understand how to utilize them to their maximum potential as a way to increased productivity and quality improvement. Once we stated there are problems of ambiguity whenever using MT, and the ones problems are also common for all of us. A clear example would be translations from Spanish to Basque. In those translations, aside from ambiguity problems, there would be structural problems, because structurally Spanish and Basque are very different.
Having analyzed some theoretical options we came to the following conclusions:
Machine translations enable visitors to have information in many dialects, assisting to understand it without knowing the dialect;
MT provides translators useful tools that help them to make their job more successfully and faster;
It can end result much larger quantities of translation than any team of translators;
Machine translation almost never reaches precision levels above 70%;
Machine translation is a venerable technological enterprise, a scientific concern of the first order and it can be understood as an monetary necessity;
Machine translation is an automatic linguistic translation", namely, a word-by-word translation;
Machine translation refers to the use of software to translate text from one dialect to another language;
In the procedure of translation Machine Translations face some problems of ambiguity that make that their use to be hard.
This research could be a good basis for an additional development of this topic, specifically, a profound research of different machine translation and their reliability in translating. We consider that the given study might be of great use to research workers in neuro-scientific translation and linguistics. It may serve as a reference point for the elaboration of season and graduation documents.
Finally, we ought to explain that machine translation comes with an important role in the process of translation which is very helpful for translators.
