Translation as a notion has existed thousands of years ago. Translation is typically outlined as the process of transferring the language of the source text (S.T) into the language of the target text (T.T) considering and appreciating linguistical and cultural discrepancies.
The History of Human and Machine Translation
Arguments of the theory and practice of human translation date back to the ancient world and demonstrate outstanding consistency. John Dryden utilized all the principles of ancient translations and outlined translation as the abstemious combination of the two types of phrasing when choosing analogs or ‘equivalents’ in the target language for the phrases utilized in the source language (Bassnett 2014, p. 69). During the period of the Middle Ages, the Latin language was considered to be the lingua franca (international language) of the western academic world. Thus, all major translations appeared in Latin (Bassnett 2014). In the 9th century, Alfred the Great was a pioneer of his time and accredited “vernacular Anglo-Saxon translations of Bede’s Ecclesiastical History and Boethius’ Consolation of Philosophy” (Bassnett 2014, p. 58). The first splendid translations into English were performed in the 14th century by Geoffrey Chaucer. He actually started a translation of the French-language Roman de la Rose. In addition, he finished a translation of Boethius from the Latin. In the interim, a new era translation started in Renaissance Italy in Florence (Bassnett 2014). Marsilio Ficino translated Plato’s works into Latin. This translation alongside with Erasmus’ Latin edition of the New Testament resulted in a new treatment of translation as such. It was the time when readers required stringency of interpreting due to the fact that philosophical and religious credence hinged on the accurate and precise words of Plato, Aristotle and Jesus. Afterwards, the 19th century introduced innovative criterion of exactitude and manner. The new standard required “the text, the whole text, and nothing but the text”, excluding any salacious excerpts and the augmentation of abundant interpretative footnotes (Bassnett 2014, p. 74).
The first offerings for machine translation utilizing computers were proposed by Warren Weaver in 1949. The offerings were grounded on the information theory, achievements in code fragmentation during the Second World War and considerations regarding global fundamentals incumbent in languages, which developed naturally in use (Bassnett 2014, p 18). Early systems utilized huge bilingual dictionaries and hand-coded regulations for arranging the sequence of words in the eventual translation issue. It was ultimately considered to be too limited and enlargement and evolution in linguistics, for instance generative linguistics and transformational grammar, were utilized to enhance translations quality. Machine translation began insistently in the 1950s. One of the best early plans appeared in 1954. It was known to be the Georgetown experiment. It retracted complete machine translation of approximately sixty Russian sentences into the English language (Bassnett 2014). The experiment had a serious advancement and induced translation in an era of important basis for machine translation research in the U.S. Nevertheless, a genuine progression was much slower-moving. In 1966, the ALPAC report demonstrated that ten years of investigation had failed to accomplish the assumptions of the Georgetown experiment. Afterwards, financing for machine translation was drastically lowered. Beginning in the late 1980s, when computer capacity enlarged and became less costly, researchers became more enthusiastic about the statistical models of machine translation. Currently, there is no system, which was achieved the ‘holy Grail’ of completely automatic superiority translation of unlimited text. Nowadays, limited number of companies utilize statistical machine translation commercially, incorporating Asia Online, SDL/Language Weaver, Google (which utilizes its officinal statistical machine translation system for a number of language conjugations), Microsoft (which utilizes its officinal statistical machine translation system to translate cognizance foundation articles) and Ta with you (which proposes a domain-adjusted machine translation resolution grounded on statistical machine translation using linguistic cognizance) (Bassnett 2014).
Types of Human and Machine Translation
Human translation is subdivided into two types – written and oral. Written translation is subdivided into non-literary and literary (deals with literary texts, i.e. works of poetry or fiction, the major function of which is to create an emotive or esthetic effect upon the reader). Oral translation belongs to non-literary type, which is subdivided into consecutive, simultaneous and sight translation. In consecutive translation the translating begins after the speaker finishes the whole source text speech or its logical part (meaning sentence or paragraph). A translator is supposed to remember huge parts of speech in order to translate it (Palumbo 2009, p. 136). In simultaneous translation the interpreter has to be able to provide the translation at the same time when the speaker is expressing the original speech (Palumbo 2009, p. 137). It can be accomplished while utilizing a particular radio or telephone-kind of equipment. Sight translation can be outlined as the reading of a text by the interpreter from the language of original into the target language, as if the text was written in the target language.
There are three major types of machine translation. The first type is a rule-based machine translation. Such system utilizes huge assemblages of regulations, manually evolved with the time by human adepts, delineating frameworks from the source language to the target language. The human agents in rule-based systems assist in delivering sufficiently good-quality machine translations with anticipated outcomes (Palumbo 2009, p. 73). Nevertheless, in regard to serious manual work, rule-based systems can be very expensive, time-consuming to execute and uphold and, due to the fact that various regulations are appointed and improved, the systems can generate obscurity and translation demission with the time. The second type is the statistical machine translation systems. They utilize computer algorithms to create a translation, which seems to be the best method statistically comparing to numerous rearrangements. Statistical models incorporate words and phrases, which are investigated mechanically from bilingual collateral sentences, originating a bilingual structural set of translations (Palumbo 2009). The appeal of statistical systems arises from the automation level in creating new sets utilizing its machine investigation capacities, resulting in prompt turnabout time and the decreased worth of processing capacity necessary for creating and managing such statistical models. Nevertheless, the main disadvantage of such type of model is the data-dissolution effect, which is provoked by deficiency of appropriate data for ‘educating’ such data-stimulated systems (Palumbo 2009). The third type is a the hybrid machine translation. In order to appeal to caliber and time-to-market restrictions, many rule-based machine translation developers are enlarging the basic technology with a help of statistical machine translation technique to produce hybrid machine translation resolutions (Palumbo 2009). Hybrids present a number of caliber enhancement advantages. Nevertheless, they retain the spending of rule-based systems elevated by augmenting intricacy of operating side-by-side systems (Munday 2012).
Usefulness of Human and Machine Translation
Machine translation shows the best performance with the texts of technical character such as user hand-books or online knowledge bases, where the main objective is the explanation of how to appoint a particular issue or terminate a certain assignment. In such case, there is no strict requirement to be linguistically ideal. However, due to the fact that the translation will still have many linguistic problems, it has to be post-edited by a human translator in a majority of cases. Machine translation can also be particularly beneficial for huge quantities of user-created content, which would not be translated, incorporating online user-created evaluations or appraisals (Munday 2012). Another efficient utilization of machine translation deals with content scanning, which allows receiving a coarse draft or general essence of the text. The great advantage of machine translation in the above-mentioned cases is the fact that it can translate huge quantities of content promptly at a reduced point on a scale of possible prices. It is also useful to utilize machine translation when there is an issue with the time. In case when time is a significant agent, machine translation can be a beneficial option. There is no need to spend hours looking through the dictionaries in order to translate the text (Bassnett 2014). Alternatively, the software can translate the text promptly and provide a high-quality result. Finally, machine translation is a useful tool due to the fact that it is relatively low-priced method. Originally, it may seem as a needless investment but in the end the amount of return will exceed the expectations. The third advantage concerns confidentiality. The providing of classified information to a human translator might be perilous and hazardous, while the data is protected utilizing machine translation (Munday 2012).
There are two major disadvantages of machine translation. Firstly, promptitude is not provided by the machine translation on a sequential substratum. The machine translation only performs word to word translation without comprising the data, which may be emended manually later (Bassnett 2014). Secondly, machine translation follows classified and formal regulations, thus the translation cannot centralize on a contexture and resolve obscurity. It also cannot utilize experience, practice or mental viewpoint as it is performed by a human translator (Bassnett 2014).
Human translation is useful, when it is required to receive high-quality text, especially literary. A human translator will be able to replicate the tone, phrasing, style and nuances of the source text in the target text. Thus, the translation will not look like translation, as it will sound or look as if it has been created in the target language. It is also useful to utilize human translation, when the text or massage is required to be culturally sensitive (Munday 2012). Only human translator will be able to understand the complexities and appreciation of the language. The human translation is also crucial when the literary adaptation is required. Finally, human translation is useful for all oral translations, as it provides prompt accurate translation, utilizing all creative skills, practice, knowledge and sense of language in order to convey the precise message of the speech (Bassnett 2014).
Any tentative to supersede human translation completely by machine translation would obviously fail as there is no machine translation, which can interpret. For example, only the human translator is feasible of interpreting particular cultural constituents, which may appear in the source text and which cannot be translated utilizing reciprocal terms into the target text language, which is done in the machine translation. Moreover, it is widely known that one of the most troublesome objectives of the translation process concerns the ability to preserve analogous influence, which is created by the source text in the target text. The machine translation has demonstrated its debilitation in this regard in the majority of cases when contrasted with the human translation (Munday 2012). The human translator is the only subject in a position able to cognate the distinctive cultural, linguistic and semantic agents promoting the possibility to have analogous influence of the source text in the target text. It is an indisputable fact that machine translation is considered to be an instrument for performing fast and huge amounts of translated texts. In spite of that, the translation quality is still much controversial and disputable (Bassnett 2014).
To conclude, it is important to mention that after the translation was first recognized as an academic discipline, translation studies have notable of the innovative types of translation apparition incorporating the machine translation. Nevertheless, its apparition was not at the outlay of human translation as the human translation manifested itself as the only subject competent of translating not merely by means of superseding words for words, as in the case of the machine translation, but also in terms of esteeming semantic, linguistic and, most significantly, cultural discrepancies between the languages.
Translation is as old as communication itself. Translation is typically outlined as the process of transferring the language of the source text (S.T) into the language of the target text (T.T) considering and appreciating linguistical and cultural discrepancies.
The human translation started thousands years ago in the ancient Rome and Greece. Latin language was considered to be the international language of that time, therefore, the majority of texts were translated into Latin. Nevertheless, it also provoked the appearance of vernacular translations, which demonstrated the importance of each spoken and written language. Human translation constantly changed its principals, and translation experts desired to gain the accuracy of context and style. The development of technology allowed translation to reach a higher level, which caused the development of machine translation in 1950s.
It is obvious that machine translation makes the life easier, however, the fact that it has to be post-edited cannot be ignored. Machine translation is useful, when there is a deficiency of time, as it provides prompt translations. It is useful, when low-cost translations are required. It may seem that the investment in the tool is unnecessary but it saves much money, which can be later spent on human translators. Nevertheless, machine translation will never be capable of providing accurate and culturally sensitive text. Human translators usually interpret into the target language, which means that they use their native tongues, are able to find proper equivalents, which will demonstrate the nature and precise message of the source text. Moreover, human translation is useful for literary and complicated source texts or messages, where it is crucial to utilize the knowledge in the sphere, creative skills, while esteeming semantic, linguistic and most significantly cultural discrepancies between the languages. Machine translation cannot supersede human translation but it can really help save time and efforts.
Bassnett, S 2014, Translation studies, 4th edn, Routledge, New York.
Munday, J 2012, Introducing translation studies: theories and application, Routledge, New York.
Palumbo, G 2009, Key terms in translation studies, Continuum International Publishing Group, New York.