Languages & Linguistics
Machine Translation
Machine translation refers to the automated process of translating text from one language to another using computer algorithms. It involves the use of software and artificial intelligence to analyze and translate content. While it can be a useful tool for quickly translating large volumes of text, it may not always capture the nuances and cultural context of language, leading to potential inaccuracies.
Written by Perlego with AI-assistance
Related key terms
1 of 5
12 Key excerpts on "Machine Translation"
eBook - PDF- Chiew Kin Quah(Author)
- 2006(Publication Date)
- Palgrave Macmillan(Publisher)
57 3 Machine Translation Systems Machine Translation is an important technology socio-politically, commercially and scientifically, despite many misconceptions about its success or lack of it over the decades. The emergence of the Internet as one of the main media of modern communication has turned translation into a bridge that connects speakers of different languages. The endless traffic of communication between different language groups requires translation, but when instant translations are needed, human transla- tors are not able to supply them fast enough. A highly skilled profession like translation using human translators is expensive and also slow, especially when a large number of languages and subject fields are involved. In order to meet the growing translation demand, Machine Translation systems are seen as a cost-effective alternative to human translators in a variety of situations. Ever since the first system was built, Machine Translation has been presenting scientific challenges (see also Nirenburg 1996). It became the testing ground for many experiments and applications for natural-language processing, artificial intelligence and even linguistics (Arnold et al. 1994: 4–5). Machine Translation is an interdisciplinary enterprise that combines a number of fields of study such as lexicography, linguistics, computa- tional linguistics, computer science and language engineering (Whitelock and Kilby 1995: 2; see also Wilss 1999). It is based on the hypothesis that natural languages can be fully described, controlled and mathematically coded (Wilss 1999: 140). This chapter begins with a brief history of Machine Translation (see also Slocum 1988; Hutchins 2000b; Nirenburg, Somers and Wilks 2003), followed by a description of the components involved in a Machine Translation system. These components and their configurations are sometimes referred to as the Machine Translation architecture. Hybrid
No longer available |Learn more- (Author)
- 2014(Publication Date)
- Learning Press(Publisher)
________________________ WORLD TECHNOLOGIES ________________________ Chapter 4 Machine Translation Machine Translation , sometimes referred to by the abbreviation MT , also called com-puter-aided translation , machine-aided human translation MAHT and interactive translation , is a sub-field of computational linguistics that investigates the use of computer software to translate text or speech from one natural language to another. At its basic level, MT performs simple substitution of words in one natural language for words in another, but that alone usually cannot produce a good translation of a text, because recognition of whole phrases and their closest counterparts in the target language is needed. Solving this problem with corpus and statistical techniques is a rapidly growing field that is leading to better translations, handling differences in linguistic typology, translation of idioms, and the isolation of anomalies. Current Machine Translation software often allows for customisation by domain or profession (such as weather reports), improving output by limiting the scope of allowable substitutions. This technique is particularly effective in domains where formal or formulaic language is used. It follows that Machine Translation of government and legal documents more readily produces usable output than conversation or less standardised text. Improved output quality can also be achieved by human intervention: for example, some systems are able to translate more accurately if the user has unambiguously identified which words in the text are names. With the assistance of these techniques, MT has proven useful as a tool to assist human translators and, in a very limited number of cases, can even produce output that can be used as is (e.g., weather reports). History The idea of Machine Translation may be traced back to the 17th century. In 1629, René Descartes proposed a universal language, with equivalent ideas in different tongues sharing one symbol.
No longer available |Learn more- (Author)
- 2014(Publication Date)
- University Publications(Publisher)
____________________ WORLD TECHNOLOGIES ____________________ Chapter- 3 Machine Translation Machine Translation , sometimes referred to by the abbreviation MT , also called computer-aided translation , machine-aided human translation MAHT and interactive translation , is a sub-field of computational linguistics that investigates the use of computer software to translate text or speech from one natural language to another. At its basic level, MT performs simple substitution of words in one natural language for words in another, but that alone usually cannot produce a good translation of a text, because recognition of whole phrases and their closest counterparts in the target language is needed. Solving this problem with corpus and statistical techniques is a rapidly growing field that is leading to better translations, handling differences in linguistic typology, translation of idioms, and the isolation of anomalies. Current Machine Translation software often allows for customisation by domain or profession (such as weather reports), improving output by limiting the scope of allowable substitutions. This technique is particularly effective in domains where formal or formulaic language is used. It follows that Machine Translation of government and legal documents more readily produces usable output than conversation or less standardised text. Improved output quality can also be achieved by human intervention: for example, some systems are able to translate more accurately if the user has unambiguously identified which words in the text are names. With the assistance of these techniques, MT has proven useful as a tool to assist human translators and, in a very limited number of cases, can even produce output that can be used as is (e.g., weather reports). History The history of Machine Translation generally starts in the 1950s, although work can be found from earlier periods.
eBook - PDF- M.A.K. Halliday, Jonathan J. Webster(Authors)
- 2004(Publication Date)
- Continuum(Publisher)
Machine Translation: THE EARLY YEARS PART ONE This page intentionally left blank EDITOR'S INTRODUCTION This first section includes two works from Professor Halliday's early years, the first The Linguistic Basis of a Mechanical Thesaurus, present at the October 1956 Conference on Mechanical Translation, and published in 1957 in MIT's Mechanical Translation. The seco Linguistics and Machine Translation, appeared five years later in Zeitschrift fiir Phonetik, Sprachwissenschaft und Kommunikationsforschu While computer technology has drastically changed since the time when these commentaries first appeared, nevertheless, they continue to be as relevant today to those working in this area as they were nearly half a century earlier. This is because the principles and methodology which they discuss are grounded in a theory of lan-guage as a unique, patterned form of human social activity, oper-ating in the context of other social activities and not dependent on the current state of computer technology. Machine Translation, states Professor Halliday, is a problem in applied linguistics, spe-cifically a problem requiring the application of those parts of General Linguistic theory which deal with the systematic description and comparison of languages. The basic premise is that description must precede translation, in particular at the levels of grammar and lexis on the one hand, and context on the other. Moreover, the description must be anchored in sound and scientific theory, and the methods used in descrip-tion [should be] derived from and answerable to, this theory. In the case of Machine Translation, the computer needs to perform a complex operation of comparative descriptive linguistics, which means that it has to digest not only the descriptions of two lan-guages but also the rules for the systematic relating of these two descriptions one to the other. 3
eBook - PDFAdvances in Empirical Translation Studies
Developing Translation Resources and Technologies
- Meng Ji, Michael Oakes(Authors)
- 2019(Publication Date)
- Cambridge University Press(Publisher)
4 The Evolving Treatment of Semantics in Machine Translation Mark Seligman 4.1 Introduction Language is primarily a way of conveying meaning, and translation is primarily a way of assuring that, so far as possible, a surface-structure segment conveys the same meaning in language B as in language A. In the face of this painfully obvious observation, it is striking how far present-day Machine Translation (MT) systems have come without any consensus among translation researchers on the meaning of meaning. In fact, influential theorists have argued that MT programs have always operated, and can only ever operate, without understanding what they are talking about. Philosopher John Searle has notoriously contended that current translation programs, and more generally current computer programs of all sorts, necessarily function without semantics, as he understands the term (Searle, 1980). 1 In his notorious Chinese Room thought experiment, a locked- in homunculus slavishly follows rules to successfully translate into another language Chinese text passed to him through a slot. He understands neither the design of the rules nor the text which he processes: To him, they are literally meaningless – ergo, sans semantics. Computer programs arguably follow rules just as blindly. It is true that MT and many other natural language processing (NLP) systems have made steady and impressive progress, while use of explicit semantic processing has undergone a rise and fall. With the ascent of the statistical era of MT and the decline of the rule-based era and its symbolic representation, Google research leader Peter Norvig and colleagues have observed “the unrea- sonable effectiveness of data”: given enough data and effective programs for extracting patterns from it, many useful computational tasks – NLP among them – can be accomplished with no explicit representation of the task, and, in particular, no explicit representation of meaning.
eBook - PDF- Ferenc Papp(Author)
- 2019(Publication Date)
- De Gruyter Mouton(Publisher)
Algorithms for different languages must be worked out and intermediary languages constructed and tested in practice. There is also a great need for linguistic statistical research. The compiling of algorithms will facilitate the clarification of structural problems for the par-ticular languages. The technical problems facing MT are as follows: the construc-tion of special translating machines, the elaboration of systems of operation for these machines, the devising of high-capacity and easily accessible memory storage units for the machines, etc. The mathematical aspects to be solved are as follows: the working-out of rational methods for the coding of information (for each of the phases of work), the increasing of the effectiveness of the algorithms, the studying of abstract language models and model translations, the working-out of a mathematical language for the description of algorithms and the automation of the programming of algorithms. The complex cybernetic tasks include the mechanical solution of the problem of compiling algorithms, the mechanization of linguistic statistics and the construction of language models by means of machines (on the basis of restricted texts). For similar surveys, partly summarizing and partly blueprinting future tasks, see also Andreev, 1960a, Ivanov, 1961c. Machine Translation 103 2. It is interesting to glance at the development of the general attitude to MT in the Soviet Union. When the regular column devoted to MT was opened in VJaz (No. 5, 1956), considerable space was taken up by the authors of the feature article (Kuznecov, Ljapunov and Reformatskij; Kuznecov, 1956) to dispel certain misconcep-tions and prejudices concerning MT.
- Daniel Steom, Georg Rehm, Felix Sasaki(Authors)
- 2018(Publication Date)
- Language Science Press(Publisher)
Chapter 2 Macine translation: Past, present and future Daniel Stein Universität Hamburg, Hamburger Zentrum für Sprachkorpora Te atemp t to translate meaning from one language to another by formal means traces back to the philosophical schools of secret and universal languages as they were originated by Ramon Llull (13th c.) or Johann Joachim Becher (17th c.). Today, Machine Translation ( mt) is known as the crowning discipline of natural language processing. Due to current m t approaches, the time needed to develop new systems with similar power to older ones has decreased enormously. In this article, the history of m t, the diference with computer aided translation, current approaches and future perspectives are discussed. 1 History of macine translation Although the frst systems o f mt were buil t on the frst computers in the years right afer World War II, the history o f mt does no t begin, as ofen stated, in the 1940s, but some hundred years ago. In order to judge current developments in mt properly, i t is important to understand its historical development. 1.1 Universal and secret languages Most likely the frst thoughts on m t emerged ou t of two philosophical schools that dealt with the nature of language and resulted in similar insights, although stemming from di ferent directions. Te frst was directed at creating secret lan-guages and codes in order to communicate in secrecy. Te second evolved from the ideal of a universal language which would allow communication without borders in the times afer Babylonian language con fusion. Daniel Stein. Machine Translation: Past, present and future. In Georg Rehm, Felix Sasaki, Daniel Stein & Andreas Wi t (eds.), Language technologies for a multilingual Europe: TC3 III , 5–17. Berlin: Language Science Press. DOI:10.5281/zenodo.1291924
eBook - PDFPapers in Computational Linguistics
Proceedings of the 3rd International Meeting on Computational Linguistics held at Debrecen, Hungary
- Ferenc Papp, György Szépe, Ferenc Papp, György Szépe(Authors)
- 2020(Publication Date)
- De Gruyter Mouton(Publisher)
The reason this is such an extremely difficult problem is because both the development and the maintenance of Machine Translation systems require the cooperation of personnel with two sets of qualities that are very rarely found in the same individuals. On the one hand, work in Machine Translation 458 AUTOMATIC TRANSLATION requires great originality, expertise, intuitive brilliance, and all the other qualities that make for good researchers. On the other hand, machine trans-lation research also requires extreme intellectual discipline, patience, persist-ence, and willingness to give up one's individual original ideas in favor of the established parameters of the system. One of the more easily resolvable problems of staffing is the decision as to whether the work of linguists and programmers should be combined in the same person, or whether the two competencies should be kept separate. In my experience, no linguist will ever become a good enough programmer to replace a properly trained and experienced professional programmer, and conversely. Therefore, in order to maintain the highest possible level of professional competence in the research staff, the two competencies should be kept separate but should learn to work in close coordination. This again is an extremely difficult objective to achieve in practice, although it is much talked about in theory. State University of New York at Buffalo AUTOMATIC TRANSLATION 459 R E F E R E N C E S Andreyev, N. D. (1967) 'The Intermediate Language as the Focal Point of Machine Translation', Machine Translation, ed. by A. D. Booth, Amsterdam, North Holland, pp. 1-27. Automatic Language Processing Advisory Committee (1966) Language and Machines: Com-puters in Translation and Linguistics, Publication 1416. Washington, D. C., National Academy of Sciences, National Research Council. Bar-Hillel, Yehoshua (1970) 'Position Paper on Machine Translation in 1970', Univer-sity of Texas at Austin, Multilith.
- Steven Simske, Marie Vans(Authors)
- 2021(Publication Date)
- River Publishers(Publisher)
This example simply illustrates how translation system can be made more robust by placing them in context of failsafe approaches and by allowing the possibility of automatic apology and a second attempt at translation. With this introduction, we now consider the current state of automatic translation. 4.2 General Considerations 4.2.1 Review of Relevant Prior Research Brief History: The search for the universal language that would uncover the language spoken by biblical Adam, lost at the Tower of Babel, gathered steam in the 17th century by well-known philosophers and mathematicians such as Francis Bacon, René Descartes, Marin Mersenne, Isaac Newton, Gottfried Leibniz, and John Wilkins. It was thought that a system could be discovered that mapped a language-independent symbol to every unique concept. In fact, Bacon believed that Chinese characters were “neither letters nor words” but “things or notions” [Lacr18, p. 163]. Fast forward to the 20th century and 4.2 General Considerations 141 we find that many approaches to Machine Translation (MT) boil down to a similar search for a language-independent universal grammar. In the 20th century, interest in MT naturally increased as a result of the success of cryptanalysis during World War II and progress in information theory [Shan48]. The beginning of contemporary MT is typically understood to be a result of a memorandum by Warren Weaver [Lacr18, p. 165]. Weaver believed that it was possible to build MT systems that, while not perfect translators, could be within “only X percent error” rate using a common basis between languages. This call for research resulted in 10 years of in-depth investigation and the first MT conference held at Massachusetts Institute of Technology (MIT) in 1952. By 1966, however, the large grants for research from the US government had dried up as a result of slow progress in the field and the lack of evaluation metrics for both MT and human translation.
eBook - PDFSoftware Engineering, COINS III
Proceedings of the Third Symposium on Computer and Information Sciences Held in Miami Beach, Florida, December, 1969
- Julius T. Tou(Author)
- 2014(Publication Date)
- Academic Press(Publisher)
Many Machine Translation experts have abandoned their work, having discovered that lan- guage translation is not a straightforward "mechanical" task; and, although information retrieval workers have frequently recognized the potential value of linguistics research to their field [31], specific collaboration seems to be lacking. The concepts of intermediate languages appear to provide a founda- tion for concurrent treatment of these two areas of interest. The manipulation of information within a single language is really only a special case of language data processing. INTERMEDIATE LANGUAGES 115 In simple systems, information may be stored as strings of characters accessed by highly formalized commands. As greater processing complexity is required for the stored information, it becomes necessary for the machine to "understand" storage contents, as well as input commands, in order to output the requested information. According to Ivanov [32]: In the projected information machine, all the aggregates of information, pertaining to a definite field of knowledge, should be written in abstract machine language. The linguistic problems of creating such a language for an information machine are closely related with the problem of translation into this abstract language from separate specific languages. Many scientists feel that humans may even find it useful to use the resulting abstract machine languages for aiding normal thought processes. This is consistent with the belief that language shapes thoughts as well as giving expression to them, ALGOL (the "Algorithmic Language") is an example of such a formal language, particularly in its use as a standard of communications (e.g., in The Journal of the Association for Computing Machinery). Computer-oriented information retrieval may be directed toward direct access to information elements or access to pointers to information elements.
eBook - PDF- Ivo Ipsic(Author)
- 2011(Publication Date)
- IntechOpen(Publisher)
Within this framework, two basic components should be distinguished: a translation model, and a retrieval model that may work as in the monolingual case. The translation can be faced either in the query, or in the document. In the case of document translation, statistical Machine Translation systems can be used for translating document collections into the original query language. In the case of query translation, the challenges of deciding how a term might be written in another language, which of the possible translations should be retained, and how to weight the importance of translation alternatives when more than one translation is retained should be considered. Here, we use the query translation approach. Then, a segment of text in a given source language is used as query for recovering a similar or equivalent segment of text in a different target language. Given that we are using complete sentences which provide a certain context for the terms to be translated, we do not have the disadvantages mentioned in the above lines. Particularly, when using the query translation approach, we investigate if using either a rule-based or a statitical-based Machine Translation system influence the final quality of the sentence alignment. Additionally, we test if standard automatic MT metrics are correlated with the standards metrics of the sentence alignment. Rule-based Machine Translation (RBMT) systems were the first commercial Machine Translation systems. Much more complex than translating word to word, these systems develop linguistic rules that allow the words to be put in different places, to have different meaning depending on context, etc. RBMT technology applies a set of linguistic rules in three 2 2 Speech Technologies different phases: analysis, transfer and generation. Therefore, a rule-based system requires: syntax analysis, semantic analysis, syntax generation and semantic generation.
eBook - PDF- Einar Haugen, Werner Winter, Einar Haugen, Werner Winter(Authors)
- 2019(Publication Date)
- De Gruyter Mouton(Publisher)
When the proposed linguistic descriptions are completed, hard-ware is to be designed for the individual matrices. If computer logic can be arranged in accordance with linguistic patterns, interpretations encountered during analysis will initiate searches whose results will indeed be found with the speed of light. 3. MT GROUPS USING SYNTACTIC INFORMATION Three centers have been involved in mechanical translation by means of a language-dependent algorithm without using a semantic component: 1. The National Physical Laboratory, Teddington, Middlesex, England. The report by McDaniel and colleagues (1967) indicates that an MT system requiring post-editing from Russian to English was produced. The level of the output was favorably evaluated by W.L. Price (1967) in his article Computer translation — is it worth-while? Further work, however, was discontinued through lack of support. 2. IBM-Deutschland, Sindelfingen, Germany, has produced an MT system for translating IBM manuals from English to German. 3. The Joint Nuclear Research Center, Ispra Establishment, Italy, Scientific Data Processing Center (CETIS) has been using the most sophisticated of language-dependent algorithms. The underlying theory is discussed more fully below. In addition to these three centers a fourth may be in the process of establishment: the Deutsche Forschungsgemeinschaft (the German equivalent of the National Science Foundation) has sponsored a program to translate from Russian into German, which was developed by Peter Torna, formerly of the Georgetown Automatic Translation Group. Torna programmed the algorithm to analyze the Russian input; the program to produce the German output is being written at the University of Saarbrücken. IBM-Deutschland and CETIS are continuing their research in MT, using in general the following theoretical approach. Language-dependent algorithms like those used by these groups exhibit a similar logical structure.
Index pages curate the most relevant extracts from our library of academic textbooks. They’ve been created using an in-house natural language model (NLM), each adding context and meaning to key research topics.
