There are basically three versions of prompt worthy of consideration by professional translators or translation companies. The most expensive, known as NET Professional, is currently on special offer for approximately £405. As the name suggests, the distinguishing feature of this version is that it is the networkable version of the most fully-featured product, which is something only the very largest translation companies are likely to find of use, unless they specialise in the mass provision of gisting services. That's not to say this version has no use, but that it is more likely to be of interest to large companies (not translation companies) who want to carry out all their gisting themselves. Such companies may of course find it a very cost effective solution compared to buying a large number of individual licenses.

 

@promt Professional 7.0 is the basic "full" version and comes in at £167, although it is currently on special offer at £133. What you lose out on in this version are the more advanced dictionary and terminology management tools offered in the next version up, known as @promt Expert 7.0, which also includes a tool for integrating @promt with TRADOS. Taken altogether, these tools make the Expert version an excellent choice for anyone wanting to use Machine Translation to build up a comprehensive series of terminology databases for their Translation Memory system. Although @promt Expert has been designed to support TRADOS and Transit users in particular, computer literate users of other systems, including Déjà Vu and SDLX, will also benefit from coughing up the approximately 50% extra required. @promt Expert sells at £336 and is currently on special offer for £268. Overall, it provides a more comprehensive array of tools than Systran.

 

Unfortunately, however, the story doesn't end there, because whereas Systran is supplied with a pack of eight languages, each version of PROMT translates for a single language pair only. What's more, as explained above, you'll also find yourself paying extra if you want any of the specialist dictionaries, either buying them individually for some £66 each or getting them all as a pack for around £170 (PROMT's prices are in dollars so these are rough conversions).

 

Just a little quick maths and you can see that if you work in two language combinations and require a set of specialist dictionaries for each, Prompt can quite happily work out more expensive than Systran once you've packed all the features you require into your admittedly very attractive Russian doll.

 

Features and usability

 

On the plus side, you actually get a printed manual with @promt. That said it frequently betrays the fact that it has either not been written or not been translated by native English professionals. Indeed, it's sometimes difficult to understand the instructions as a result. The same criticisms apply to the Help files, although @promt certainly scores higher marks here than Systran, since the Help files do actually offer help and, indeed, context-sensitive help into the bargain. Similarly, the manual is generally also more useful than Systran's for the simple reason that it is better at providing step-by-step instructions.

 

The interfaces themselves are more cluttered than in Systran but no more difficult to use. Speed of use is also largely comparable. Unlike Systran, @promt doesn't automatically install any toolbars in other programs, instead offering a small program called SmarTool that allows you to install and uninstall this support at will. That's actually a very handy little touch, particularly where Word is concerned, since it's prey to having so many different toolbars installed, including those for Workbench and MultiTerm if you're a TRADOS user, that it can sometimes be impossible to start Word and open a file in it automatically, by simply double-clicking on that file in Windows Explorer, without hitting timeout problems that result in an error message erroneously telling you your file can't be found.

 

@promt's Word toolbar provides easy access to all the functions you require and is intuitive to use. Unfortunately, however, @promt does sometimes get confused by the formatting in Word documents. Like Systran, @promt has a feature to prevent you from mistakenly overwriting your only copy of a source document with its translation, but unlike the former offers you a series of different options according to your preference.

 

@promt's TRADOS integration functions work well enough, speeding things up a little although not actually doing anything that couldn't be carried out manually using another system. Unfortunately we weren't able to test this module as thoroughly as we'd have liked since the TRADOS dongle decided we'd been overworking it and committed suicide halfway through the evaluation routines. Suffice it to say that you can use it to obtain the segments for which a TRADOS Analysis is unable to find any matches and get them translated by Prompt automatically.

 

Another problem with evaluating @promt is that the company does not currently provide specialist dictionaries for the language combination we were using, so we were unable to test their impact on translation quality. With over twice as many dictionaries in a set as are offered by Systran, however, one might at least in theory quite reasonably assume that because the dictionaries are more specialised, they would have a greater impact on quality. As we know, however, theory and practice often don't match up, and without being able to verify the size and quality of these dictionaries, there's really no way of knowing how things would work out.

 

Whilst on the subject, though, it is worth noting that the dictionary set can be added to any version of @promt. Also, if the quality of the dictionaries corresponds to the quality of the basic dictionaries and machine translation engine, they will at the very least compare favourably with Systran's offerings.

 

Translation quality

 

The translation quality of @promt without its specialist dictionaries was at least the equal of Systran where accuracy was concerned and superior, everything considered, when it came to intelligibility, having a transformer that is more adept at creating the appropriate target language constructions rather than simply repeating what it finds in the source. The difference in quality wasn't massive, though, and it should never be forgotten that we are talking about Machine Translation quality, meaning that all but the very simplest of generic sentences are likely to have some sort of problem regarding either intelligibility, accuracy or both.

 

Is it all worth the cost and trouble?

 

Whichever system takes your fancy, neither are cheap enough to be considered a lightweight investment. You could buy a very nice printer for the same sum of money, for example. Not only that, but the money is only half the problem, because if you buy either of these systems, you're going to need to invest a good few hours in working out how to get the best out of them for all but simple "raw" gisting, which is a "click and go" affair. If you want to do more than that, you really ought to invest some time in evaluating both systems before you buy either of them (assuming the language combinations you need are offered by both). That's because any of the above comments on translation quality only apply to @promt's English/French engine and Systran's English/French and English/Italian engines; (@promt doesn't currently cover Italian, although English_-> Italian will soon be available, while English, French and Italian are the only languages I'm qualified to evaluate myself). In other words, if you work in other language combination/s, you may well find your mileage varies.

 

The good news here is that free evaluation versions are available for both programs, although you'll need to make your mind up quickly if you want to take advantage of PROMT's New Year special offers, which are about to expire. That said, there's no need to rush if you're in the UK, because both @promt and Systran are available all year round with a 20% discount to ITI members from Omega First.

 

So, assuming you haven't already decided to order three cases of wine online and dedicate your free time to drinking them instead of learning how to use a machine translation system, the question is what would you gain from making this investment?

 

There are only three possible scenarios. If fiddling around with computer programs tends to make your eyes glaze over and encourage you to put off what you can do today until tomorrow, or preferably even later, then forget Machine Translation for your personal use. You might as well throw the money away. Unless you have the interest and energy for one of the remaining two scenarios to be yours, then you'll never get a return on your investment. It certainly isn't currently the case that MT can in any way be considered an essential tool for the modern professional translator, although depending on your subject area and familiarity with the systems, it can in some circumstances certainly be a useful productivity tool .

 

The second scenario is specific to Translation Memory users, particularly those who are interested in maximising the size of their terminology databases and/or Déjà Vu users who have discovered and become addicted to the benefits of using the Lexicon and its terminology extraction functions. The latter will stand to benefit most thanks to Déjà Vu's sub-segment matching technology unless they also dictate their translations using speech recognition software and are highly experienced, so that they don't generally bother adapting fuzzy matches or assembled segments, preferring to simply dictate in their own translation without even taking note of the terminological support provided.

 

The final scenario regards translators who work with "direct clients" and have already proven to themselves that they are dynamic marketeers and can therefore be confident that they will offer and exploit the benefits of a structured MT service of the type described in the introduction. If that describes you, one of the more sophisticated versions will be what you’re after. Alternatively, if you just want to offer a simple gisting service without getting involved in too many technicalities, buy the cheapest version with a decent set of ready-made specialist dictionaries.

 

There is of course nothing to say that the second and third scenarios are mutually exclusive which, if you want to be picky, suggests there is a fourth scenario - the techie marketeer - who could find that cunning marketing, TM and MT take them as close to paradise as it's possible to get, especially if they also use speech recognition, which would undoubtedly fit their character profile.

 

Before signing off, I don't want to forget existing MT users. All in all, the answer for them is much simpler: if you're happy with your existing program, don't bother upgrading.

 

As a last word, if you do decide to take the plunge into MT, please don't forget to keep us posted. Hearing of your successes and failures will help us provide the best possible information in the future.

 

 

 

 

 

Michael always welcomes feedback on his articles. You can contact him at: michael@michaelbenis.com

 

 

 

Useful addresses

 

PROMT: http://www.e-promt.com

Systran: http://www.systransoft.com

Omega First: http://www.omega-first.co.uk

 

 

 

[Captions]

 

[Systran Word.jpg]

Systran's Word toolbar allows you to translate any document at the click of a button, while also giving you instant easy access to all its main options.

 

[STPM.JPG]

Systran's Translation Project Manager provides you with a customisable twin-window editing interface that can also consult its own translation memories.

 

[Prompt interface.jpg]

@promt also offers a parallel window interface for post-editing machine translation output. The interface is slightly more powerful than Systran's, but also more cluttered.

 

[Prompt Word.jpg]

@promt provides a comprehensive Word toolbar, but can unfortunately sometimes get thrown by even quite simple formatting.

First published in ITI Bulletin, 2005.