This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The aim of this article is to analyse the role played by computers in language learning and in developing students' translation competence. I will deal with the elements involved in computer-assisted language learning: student/translator workstation, corpora and intelligent tutoring. I will also explain and give examples of using concordancers, translation memory software, and machine translation techniques in translation. Through this paper I hope to demonstrate that translation and language learning can be successfully integrated within a computer-aided learning/teaching framework, especially by involving the use of corpora.
Keywords: computer-assisted language learning, translation, translator workstation, translation software, machine translation, corcondancers
Computers nowadays provide great help to linguists, especially if we are to think of text manipulation techniques: large storage, quick search and retrieval as well as networked communication. As a result, the styles and efficiency of text processing, language learning and translation can be greatly improved. Especially the existence of extensive electronic corpora offers examples and generalisations of language usages, which in turn will contribute to better language teaching and produce high-quality translation. Electronic tools such as dictionaries, concordancers, encyclopaedias and Web page search engines offer instant returns to inquiries about word or phrase usages. Secondly, upgraded machine translation technology has turned into a helpful tool, not only to business and industry, but also for translators and language learners. We can therefore say that the machine-aided translation technology also has beneficial effects both on translation and on language learning.
Different computer programs, combining the characteristics of human intelligence in providing guidance with the machine capacities to manage resources and solve tasks, have the potential to become good assistants for translators and language users and learners in the future. Nowadays educators' needs in language and translation teaching cannot be met without orienting the courses and the curricula in the same direction as the new computer era takes us in all other aspects of life. I will in the following make a short presentation of computers' benefits to translation and language learning, from the translator's workstation to the translation memory applications, the machine translation technology and computer assisted language learning. I will try to analyse the interrelatedness of concepts such as translating, language learning and the role the computer can play in this environment.
The reality of modern times is such that a translator does not have to go to the library to find reference materials any more: a computer with Internet access can provide answers to the majority of questions arising in the translation process. Among the reference tools readily available on the Internet or on different specialised software mention should be made of: electronic dictionaries, terminology banks, encyclopaedias, concordancers, collocation finders, Web search engines, etc. All these tools contribute to the creation of a computerised working environment, which has been traditionally called the translator's workstation (Melby, 1992; Hutchins, 1997; Freigang, 1998) or the translator's workbench (Trujillo, 1999).
According to Melby (1992) there are three levels of functions for a translator workstation. Level one functions comprise word processing, terminology management and telecommunications. Level two functions consist of text analysis, automatic dictionary look-up, and synchronized bilingual text retrieval. Level three "provides an interface to machine translation systems".
We have all encountered and are familiar with the use of electronic dictionaries, encyclopaedias and bilingual terminology files at the lower levels; therefore we will not spend too much time on this aspect. A less obvious tool, but similarly helpful for the beginner translator is the Web search engine. For example, in attempting to translate the expression Bakugan battle brawlers into Romanian, the Romanian equivalent may not be found in the dictionary or in a terminology bank. The translator may know what the word means, but it is another thing to translate it into Romanian, because an established translation could exist which the translator may not know.
(1) Bakugan Battle Brawlers are action-figure warriors, that are tucked into spheres that pop open when they're rolled onto a game card, where they gather points.
The definition we can find on TheFreeDictionary is brawl intr.v. brawled, brawling, brawls 1. To quarrel or fight noisily. 2. To flow noisily, as water; brawler n. a fighter (especially one who participates in brawls). It stands to reason that a warrior brawls in a battle.
In such cases it is useful to resort to a localised Web search engine like Google Romania, which can help retrieve relevant national Web pages with the accepted translations for the term we need.
(1) LuptÄƒtorii Bakugan sunt figurine ascunse în sfere care explodeazÄƒ atunci când sunt aruncate pe cÄƒrÅ£ile de joc, unde acumuleazÄƒ puncte.
Level two functions include text analysis which includes mostly the tool of concordancers (Melby 1992). A concordancer is useful to the translator mainly in the target language domain, as it primarily assists the translator in the processes of rendering target texts (as opposed to understanding the source texts). The concordancer can search a large native-speaker corpus and find lines of texts containing the query word whose usage the translator, as a non- native speaker, is unsure about. The concordance lines provide the translator with a set of contexts that the target word is used in so that she can be more certain about how to integrate the word in her own sentence. For example, suppose our task is to translate the Romanian fragment in (2).
(2) DacÄƒ în strÄƒinÄƒtate cerÅŸim de foame, acasÄƒ am ajuns sÄƒ "negociem" cu marile companii din domeniul petrolului de frica frigului. Åži asta pentru cÄƒ guvernanÅ£ii s-au trezit în prag de iarnÄƒ surprinÅŸi de avalanÅŸa preÅ£urilor la gazele naturale, care tinde sÄƒ îngroape speranÅ£ele multora dintre români de a trece cu bine peste sezonul rece.
In the following there is a 'literal' translation of the passage above, which doesn't seem to sound well, at least to a native speaker.
(2') If abroad we beg of hunger, at home we've come to "negotiate" with the big companies in the oil field for fear of cold. And this because governors woke up at the beginning of winter surprised by the avalanche of prices for natural gases which tends to bury the hopes of many Romanians to get well over the cold season.
First of all, the literal translation of â€žcerÅŸim de foame" - "we beg of hunger" may sound awkward, and the translator may want to check in a concordancer based on large corpus how the verb 'to beg' is used by native speakers. The following is a snapshop of the BNC output:
Fig. 1. Snapshop of 'beg' concordances
As can be seen from the above selection, the phrase 'having to beg for food' is used in a similar context, referring to starvation and poverty. Conversely, the phrase 'governors woke up at the beginning of winter surprised by the avalanche of prices for natural gases' does not seem to sound too native-like. A simple search through the BNC corpus would return more natural associations: 'surprised to find/see'.
Fig. 2 Snaposhot of 'surprised' concordances
Moreover, the expression 'avalanche of prices' needs improvement too. Our search through a concordancer would return the collocation 'soaring prices', which would retain the meaning in Romanian, and would at the same time sound more appropriate.
Fig. 3 Snaposhot of 'prices' concordances
The same strategy may be applied for all the word associations which seem awkward-sounding. We have not included all snaphots of the concordancers we have used, but the final revision of the translated text would rather look as in the following:
(2'') If abroad we beg for food, at home we've come to "negotiate" with the big companies in the oil field for fear of cold. And this is because at the beginning of winter our governors were surprised to see the soaring prices of natural gases, which tend to crush/kill off the hopes of many Romanians to successfully withstand the cold season.
Another level-two function in Melby's frame of a translator workstation may be what he dubs the 'synchronized retrieval' of bilingual text files. A widely accepted contemporary term for this process is the 'translation memory', which is much more complicated than what we have covered before and therefore would deserve special attention in another context.
According to Hutchins (1997), translation memory represents a facility which "enables the storage of and access to existing translations for later (partial) reuse or revision or as sources of example translations". In his opinion, "the sales of translator workstations incorporating translation memories are increasing rapidly, particularly in Europe". On the other hand, in Asia, some translation companies also advertise the use of translation memory software as a useful tool on their Web pages. The practical value of translation memory software to translators is proved by its world-wide use, especially by international enterprises, freelance translators and translation companies.
One example of modern translation memory software is the TRADOS Translator's Workbench, usually has the following modules:
- Translation memory
- Terminology management
- Sentence alignment
The translation memory module will record the user's translations sentence by sentence in a translation memory file. It also has a fuzzy-match algorithm which h will search the memory to find any (partial) match for a new source sentence and retrieve the corresponding target sentence for the translator to analyse. In contrast, the terminology management module will keep a bilingual terminology bank and, upon the activation of a source sentence, will search and retrieve the target language parts of any terms that exist both in the source sentence and in the database. Finally, the sentence alignment module will enable the translator to import an existing piece of translation (from a source text file and a corresponding target text file) that has not been previously treated by the translation memory module, and align the source texts with the target text on a sentence-to-sentence basis. The results of the alignment can then be exported to a translation memory file in order to enhance its contents.
The use of such software means that several translators working together can pool all their translation texts and terms together, including those from the present and the past, in the form of translation memory files and terminology database files. In this way, the tasks of translation and maintenance will be easier and more efficient, while the translation of terminology and recurring patterns will be more consistent across persons and organisations.
The history of machine translation (MT) has been rather contradictory since its origin in the 1940's (Hutchins, 1995; Somers, 1998b), from the following perspectives:
- the methodology it has used: from the direct translation method to the transfer method, the interlingua method, to the more recent corpus-based method (see Somers, 1998b, Arnold et al., 1994; Trujillo, 1999);
- the expectations of MT: from FAHQMT (fully automatic high-quality MT), to no prospect at all in the late 1960's, to the current limited and realistic goals and expectations (see Hutchins, 1999); and also
- the platforms and users for MT: from the translation companies' desktops to the real-time Internet service for the general public (Yang & Lange, 1998) to its possible roles in the fully-fledged "teletranslation industry" conceived by O'Hagan (1996).
Researchers in MT have underlined that the success of MT depends on the restrictions of the domain, the input, the readers and purposes, and the language pair used (Hutchins, 1995, 1999; Somers, 1998a; O'Hagan, 1996: 26-35; Manning &Schütze, 1999: 463).
Research into MT from Romanian into English and vice versa is still in its incipient stages. Free translation software is at the same time inefficient. Let us consider a part of the extract we analysed above:
DacÄƒ în strÄƒinÄƒtate cerÅŸim de foame, acasÄƒ am ajuns sÄƒ "negociem" cu marile companii din domeniul petrolului de frica frigului.
The Google Translate facility (©2009 Google) will provide the following translation:
If abroad begging hunger at home have come to "negotiate" with big oil companies in the field of cold fear.
As can be seen, some problems are to be found at grammatical level. First, in Romanian we don't usually express the subject when this is a pronoun, as it is implicit in the verb endings ('cerÅŸim de foame' = noi cerÅŸim de foame; 'am ajuns' = noi am ajuns -- 2nd person sg.). However, the translation in itself is a faulty one, and does not convey any meaning whatsoever. The combination 'in the field of cold fear' is hilarious, to say the least. Interestingly enough, the idea of oil companies was rendered, although the works field was preserved and wrongly associated with the following semantic unit.
Fig. 4 Google Translate translation
Although in the future MT will play a more and more important role in linguistic services, it will definitely not replace human translators in the foreseeable future. According to specialists, automatic translation cannot, unaided, produce anything better than 'rough' (occasionally barely comprehensible) 'translations' (see above). In order to get higher (publishable) quality, the solution will remain with human translators, aided by all the computer aids that are appropriate (Hutchins, 1998).
O'Hagan shares the same view, considering that "users are recognising the potential benefits of MT and are beginning to make use of the technology within restricted environments" (1996: 35).
This is to say that MT "must no longer be put forward as a 'solution' to people's translation needs, but it must be seen as no more than a 'useful aid'" (Hutchins, 1999).
Increasingly, not only is MT used by the general public for dissemination (of information), communication, entertainment and language learning purposes (Yang & Lange, 1998), it is also used by translation professionals. It doesn't come as a surprise that students also find that MT can help them do their foreign language homework. Moreover, even for a language teacher, MT can become a useful teaching aid too. One useful exercise would be for the teacher to use the faulty translation provided above and have the students correct it and find explanations for why the machine has generated such
If we consider all the educational and translational technology aids mentioned above, it appears that we can find bridges between computers, translation and language learning. We have seen how translation can be combined with computers in the implementation of a translator's workstation. We have also seen how see language learning can be integrated with a computer system in the CALL environment. What is more, the possible combination of translation (especially translating into L2) and language learning is also strongly implied from the general working environment language learners and trainee translators share (i.e. both benefit greatly from the use of computers), the tools and resources they use (e.g. concordancers and corpora) and their common goal of interaction (i.e. to produce high-quality written language, close to native speakers' proficiency).
Therefore, we would like to suggest that translating into L2 and L2 learning activities should be carried out in an integrated computerised environment and both will benefit doubly as a result of mutual reinforcement and other factors such as varied resources and increased interest.