COMPUTER AIDED TRANSLATION

The computer-assisted translation (CAT, or computer aided translation) is language translation performed with the help of a software program. The program does not do all the work, but it can create some shortcuts and also grows with the translator and can develop more facility over time. This can be useful for linguists preparing a variety of texts, transcripts, and other contents. Software companies offer a range of computer-assisted translation programs.

How can the CAT Tools help us?

The computer-assisted translation (CAT) Tools may be just what you need to take your translation work to the next level. Here are some of the benefits of using a CAT Tool:

The Quantity Aspect: Take in More! Localization requires fast turnaround of high volumes of content, with rigorous consistency of terminology and style. CAT tools, especially major ones like SDL Trados, make this kind of performance possible. Linguists who can transfer translation memories (TM), update them while they localize, and send them back, are crucial to the process.

The Quantity Aspect: Produce More! Even in traditional environments, a CAT tool can increase your production. With a tool pre-translating from a reasonably-stocked translation memory, you could finish twice as much work in the same time. Warning: this depends on how similar the new content is to what is in the TM. My own work data shows an increase of over 50%.

Some programs can start the translation for the translator. The program is loaded with spelling and grammar data for two or more languages and may be able to render sentences in reasonable translations. The linguist can skim for errors and may correct sentences that are obviously incorrect, fail to capture the intent of the source text, or read stiffly. These corrections are part of the teaching system for the program, which can learn from them to prevent future errors. Other computer-assisted translation may not perform rough pass translation, but it can still be helpful. Spelling and grammar checkers are available to assist linguists as they move between languages. The program can flag errors and may offer suggestions to fix them. Some intelligent programs may also identify homophones and alert the translator when a word appears to be inappropriate for the purpose. Translators can also add idioms and terms, important for technical translation where words unfamiliar to a base dictionary may be used.

The proper role for human translators (which is slow and expensive) and machine (which produces low-quality translations) in a collaboration to produce good quality translations efficiently is still up for debate. It has long been argued that machines should only play an assisting role when professional linguists craft their translation. The tide has turned towards placing man as post-editor of output of software translation systems, but this argument is not over yet.

While human translation is still necessary, computer-aided translation can assist speed the process. Human linguists need to check the computer's work and may need to perform some translation tasks, depending on the text and the program. Translation skills are critical, as someone who do not possess a thorough knowledge of both languages could make errors or might not recognize problems with the software translation. Software companies work with linguists and consultants to improve the technology, and some create consumer products that offer rough translation to people like Internet translators who want a quick overview of the content on a page in a foreign language.

While statistical software translation system often provide useful or good enough translations, demands for high quality, publishable translations still require human linguists. Tools for linguists to improve their productivity can be build using conventional software translation methods.

The Post-editing - Several studies have shown increase in productivity by post-editing software translation output instead of translating unassisted directly by human translators. Skadiņš et al. show a 30 percent increase for English-Latvian translation with a slight but acceptable degradation in quality. Plitt and Masselot (2010) compare post-editing machine translation against unassisted translation in custom web-based tool for a number of language pairs on information technology documents, showing productivity gains of up to 80%. Guerberof compared the benefits of translation memory matches and software translation for a subset of sentences that lie within the 80-90% fuzzy match range, showing higher productivity gains and better quality (according the LISA standard) when using machine translation. Federico et al. assess the benefit of offering machine translation output in addition to translation memory matches (marked as such) in a realistic work environment for linguist working on legal and information technology documents. They observe productivity gains of up to 20-50%, roughly independent from the original translator speed and segment length, but with different results for different language pairs and domains. Garcia also measured higher productivity when bilingual native-Chinese students localize between English to Chinese in both directions. Vazquez et al. (2013) find higher productivity for post-editing software translation than using translation memory matches in a fuzzy match range of 80-95%. Pouliquen et al. (2011) showed for a patent translation task that non-professional post-editors may be able to create good quality translations, comparable to a professional translation house. In an experiment on translating English into three languages with a very restricted web interface used by professional linguists, Green et al. carry our more sophisticated statistical analysis using ANOVA and show that post-editing leads to better and faster translations. Läubli et al. stress the importance of testing post-editing software translation under realistic working conditions, and found lower productivity increased (15–20%) than reported elsewhere, on a German-French translation task. Bogaert and Sutter show productivity increases in the range of 20% to 134% for 10 linguists in an English-Dutch task on financial European Commission publications, with slightly higher quality. Karamanis et al. investigate the impact of introducing post-editing software translation on the work practices of professional linguists. Pointing out that trust plays a great role when relying on previously localized segments found in translation memories (e.g., preferring work from close colleagues over freelancers), such trust lacks when assessing output from machine translation systems.

Analysis of the post-editing efforts - Koponen examined the relationship between human assessment of post-editing efforts and objective measures such as post-editing time and number of edit operations, finding for instance that segments that require a lot of reordering are perceived as being more difficult, and that long sentences are considered harder, even if only few words changed. Koponen finds relatively little difference between post-editors when given a choice of output from multiple computer translation systems, albeit in a controlled language setting.

Man aiding computer - By giving human linguists access to the inner workings of software translation system, they may fix errors at various stages, such as changing the source sentences or its linguistic analysis (Varga and Yokoyama, 2007). Conversely, the input to a translation system may be automatically examined for phrases that are difficult to localize (Hwa and Mohit, 2007).

Interactive software translation - The Trans-Type project developed an interactive translation tool which predicts the most appropriate extension of a partial translation by quick re-translation based on translator input. Word graphs allow for quicker re-translations and confidence metrics indicate how much should be presented to the translator as reliable prediction.

Translation options - In addition to interactive predictions, human linguists may be aided by the display of word and phrase translations. Showing multiple such translation options may even allow monolingual translators to localize from unknown source languages Koehn.

Other assistance - Various types of data may be beneficial for a translator of a translation tool, such as suggested translations for idioms, unknown words, and names. Large word-aligned parallel corpora such as the billion word French-English corpus may be superior to traditional terminology databases.

Translation memory - Another valuable tool in computer-assisted translation is translation memory. The system can retain phrases and chunks of data that the linguist has already localized. It may substitute the translations for convenience to allow the linguist to focus on new content. This can save time on a translation, and the computer will flag its suggestions so the translator can check them to ensure they are accurate. Some terms may not always localize in the same way because they can reflect different intents on the part of the speaker or writer.

Translation memories are a widely accepted and popular tool for active linguists. When translating a new sentence, these tools retrieve the most similar source sentence and its translation from what software translation researchers would call a parallel corpus. Furthermore, the mismatch can be corrected by letting a statistical software translation model localize it. The fuzzy match from the translation memory may be encoded as a large hierarchical rule or other methods. The alignment between input sentence and the source part of the translation memory sentence pair may be aided using syntactic structures.

Usage analysis - Macklovitch et al. presented a tool that visualizes the linguistic interactions. Human post-editing data may be mined to improve the performance of software translation, as shown for transfer-based systems. Software translation may also be used for interactive tutoring tools for foreign language learners. Macklovitch shows how alignment methods could be used to identify error in human translations.

The bilingual concordancer - Professional linguists may wish to search translation memories not only for fuzzy matches of full source sentences, but also for exact matches of words and phrases. The widely used TransSearch system returns full sentence pairs, allowing the professional linguist to examine which translation is more customary given the sentence context. Linguists mostly search for 2-3 word terms, especially highly polysemous adverbials and prepositional phrases. Translation spotting is the technique to highlight the search term and its translation. Translation spotting may be improved by filtering, merging of variants, and pseudo-relevance feedback. Pastor and Alcina (2009) argue for the training of linguists in search techniques in monolingual and bilingual corpora. Bai et al. (2012) present a normalized correlation method for translation spotting, which overcomes weaknesses of both word alignment-based and association-based translation spotting.

Automated reviewing - computer-assisted translation methods may be also used to detect errors in professional translations. Experts suggested added or missing content, consistent use of terminology, and present work on spotting translation of deceptive cognates, word pairs that have similar surface forms but different meanings.

We are a team of veteran Mandarin translators dedicated to delivering top-quality English to Mandarin translation service. We make best use of CAT to enhance our productivity and ensure maximum consistency.

________________________________________________________