英语论文网

Imagine you have worked for decades on a project that promised to deliver nothing less than one ofhumanity’s dreams. Suddenly you begin to realisethat, despite all of your efforts, you will not be able to
留学生作业写作指导deliver. Imagine – and this would surely be an evenmore difficult situation – you have alreadyannounced that your efforts to realise this dream have beensuccessful when, in reality, you know that you are going to failmiserably. What are your options?
According to a paper recently published in the proceedingsof ASLIB, Translating and the Computer 25, by one of the most
respected and eminent industry figures, Jaap van der Meer,“Machine Translation (MT) has long been a controversialtopic, the source of illusions, jokes and even serious disputes.Research and development in fully automatic translation hasbeen carried out for fifty years. At regular intervals,researchers (…) have heralded the big breakthrough”.
Jaap van der Meer is right. MT developers announced againand again that they had finally proven their fully automatedhigh quality machine translation systems could make humanity’sdream a reality. They claimed to have solved the dilemma,created by the tower of Babel, with the help of a clever computerprogramme.
Because executives with purchasing power (and littleunderstanding of the issues involved) were easily impressedby ‘hard figures’, it became fashionable to randomly quotepercentage values as a measure of success in MT. 95% accuracy,a minimum of 40% savings, 100% consistent use of terminologyand so forth.
Blinded by the enormous potential savings, these executivesbought high-powered MT systems and employed linguists to finetuneand programme them. This managed to create short-lived,cyclical surges of interest (and investment) in MT. However, theywere always soon followed by a depression (and redundancies)because customers quickly realised that, once again, over-enthusiasticsales executives had taken them for a ride.
At ASLIB, Jaap van der Meer announced that this time “thebreakthrough is true” and that it is “market-driven ratherthan technical”. He conceded that “MT is not perfect” but that“it has become an economic necessity”. He presented figuresto prove that, finally, MT can be employed in such a way that itmakes economic sense.
Another respected professional, Ross Smith, from
PricewaterhouseCoopers, provided “An Overview ofPwC/Systranet on-line MT Facility” at the same conference,offering use statistics and feedback, in addition to some practical
examples around areas such as gisting.The user feedback reported by Ross Smith is positive overall.(Frankly, looking at the ever so slightly skewed questionnaire,one cannot entirely discard the suspicion that the positiveuser response was at least partially pre-programmed.)
The only negative example reported came – to the amusementof the presenter and his audience – from a user who had
attempted to translate a document written in Dutch intoEnglish, using a German-English MT engine. No wonder hewas not happy with the translation!So, it seems that there is not just one, but many reasons to
push the boat out and organise celebrations for the new dawnof MT, the final breakthrough.
Or are there?
It took smart MT developers years to comprehend that,even in a sales situation, the correct answer to the question“Will MT help us to save money on translation?” is “Itdepends”. They realised that although this is probably theTHIS ARTICLE WAS FIRST PUBL