This is a series of reflections on teaching translation technologies. We've met the issue of technologies at various points in this subject so far. Here, I'm particularly interested in the way it fits into the problems of curricula and that is of sequencing. As we know from our top-down analysis, everything should come from a needs analysis. What do our graduates need when they go out to be employable? Now I'm presenting here some other research done by Yu Hao, when we asked graduates from the Master of Translation at the University of Melbourne, what skill do you wish you had acquired more off? That is, what should we be doing more in our training of translators? And the skills have been categorized in terms of colors.
Here, where technology is that pinkish sort of color. And I'm not too sure that market awareness is really technology unless you define it as awareness of the technologies needed. But the technologies are not really at the top they're in the middle. So although there's plenty of employers who say all we need translators who know their technologies. And the end, we have employees in big companies complaining that graduates don't know enough about technologies.
That's not quite the feedback we're getting here. However, Yu has broken this down, as you can see here. Here are the people who get work in translation and interpreting.
Here are the people who work in languages in a wider sense, and here are the others. And it's very clear that the technology is of much more importance for translation and interpreting at the top here than it is for language work in general, or people who go out and do a range of other things using their skills. So we can say, I think, that if we're talking about employability as translators and interpreters, and that's only about a third of where our graduates go. yes, they are important, but more generally, no, we shouldn't overemphasize the case if indeed we are training for language workers in general. Some other data on the same kind of needs analysis.
This is from the big CIUTI survey that we mentioned previously, where we found, and I looked at this last week, that foreign language training is the most important thing. General translation courses, cultural studies is up there, research skills. These are all soft skills.
So where is the technology here? Well, yes, it's up there. 37% is not too bad. And we could add in software localization and terminology management.
So it is part of the skill set that is needed. But it's not at the top, top level. Ok, here's some other data. This is from the European Language Industry Association 2019. It's a survey of companies and translators, and actually trainers in Europe.
What, what is interesting to me here is that the big translation companies are investing a lot in machine translation and automated workflows than the small companies. So this is telling us that the need for technology skills, is not universal, it's not the same for everyone. and it is great to at the top end of the market where most of the money is being made.
So there is a connection there. The same report makes it very clear that the big money is moving at the top. That the small companies and freelancers are not rapidly increasing their profitability. So machine translation, automated workflow, and the smaller segment, the smaller players in the market, areusing translation memories, although much, much less machine Translation. Okay? So I'm just distributing or trying to present the needs analysis as something that has to be focused. If you know exactly where your translators are going, what they going to do with it.
Then you can focus on certain kinds of technology, I think, and say where it has to be done. What you can't do, I think, and this is the common presupposition in some of the older programs is think, well, when we're training somebody, you go from the easy things to the hard things, and technology is hard for me, for you. the teacher. Therefore we'll do it.
At the end of the training, we found that, I think in Kelly, in the model of the undergraduate program in translation that she presented. Why is it wrong? Well, let's just go and how wrong it is because everyone these days needs technology. Okay? The other thing is you do it in a special place. No, that's wrong. Everybody needs technology and everybody needs a laptop, at least if they're going to be doing translating or really any, any learning with technology. So instead of doing it up here, you want to do it down here, or you want to do some of it down here as some of it up there. Okay, I think a growing solution is to give everybody basic work in pre-editing, post editing, and translation memories down here.
And also terminology work because the interpreters need that as well. And then have advanced technology classes happening up here. And if you like, you can do that for interpreting, but it's not done so much.
In our Master to here at the moment, we do have translation technologies as an elective. What then might you want to do at the initial level down here? Well, I think you should emphasize a range of technologies. I'll come to that in a minute. You're going to have to look at machine translation. Explain what it is explain what it can do and what it can't do.
And then when people understand what it can't do, you go to have to be doing some post editing down it and pre editing as well, since that's one of the major ways we have of working with machine translation technologies. Terminology is a logical thing to do down here at a very basic level, the level of glossaries, because then you can get the translation memories. And the glossaries that can be used in translation memory tools.
And at the basic level down here, quite honestly I would just use the freeware that's available on the web this year we're doing Wordfast and Matecat, for example, both of them. So students get to, to transfer the skills from one to the other. The basic reason for that is that they web-based. So we don't have a problem with PC versus Macs or compatibility of operating systems. Ok. Usually at that basic level, I'd get people to work on two translation tools.
And I'll have all this training at a very basic level. And then they have to pick up a third tool learn how to use it to do a translation where, working in pairs, they have hours to do that and they can do it. What does that mean? They've learnt not just how to do all these things, but they've learned how to pick up new technology, make sense of it, use the learning resources that are available on the web and they've learned to learn.
And I think that's the most important skill to develop at the initial level down here. You can or cannot throw in subtitling I do, because it's fun. And I think all language learners should get basic training in subtitling because they can do their favorite video clips and bits of TV series and, and enjoy seeing the product of their work. At the advanced level ,though, it's going to look similar. And here I'm, I'm just giving a list of what we actually do in the elective that we have in the Master, It's the same. but the translation students here are obliged to do Trados which is the market leader in translation memory software.
It's also fairly complicated, it's not easy to learn. It has lots of features, lots of terminology, especially lots of quality control, lots of, well, some project management features built in. So we spend quite a while doing Trados As you can see, the terminology and project management is done in that environment. The thinking here is that if you can use the most complex tool that's out there and the one that looks best on your CV. Those skills can be
transferable to any of the simpler systems that you might want to use. Okay. So it's a different way of thinking. There's also then project work. We do a two-week project where the technologies integrated into group work in a company, a simulated company environment. And we look at website localization and I think a bit of game localization, although there's not a lot of special technology required for game localization.
Okay, so that's what an advanced level training calls might look like. It has some of the features here that are shared. There's got to be repetition, of course, but it's doing them at a more advanced level, at a more professional level. We can also do a little bit more theorizing about the impact of the technologies. The problem with that kind of planning is that there is still resistance to it.
And here is a resistance statement from Brian Mossop. If you can't translate with a pen and paper, then you can't translate. Meaning we have to train people to translate with a pen and paper. Because if you haven't got those skills, that all the technology in the world is not going to make you better.
I don't know if that's still true, but let's take it as a legitimate expression of concern and a focus on, on saying that real translation is not with technology. I would have as a retort these days, if you can't use technology at all, you can't get a job. So what does it mean to get a job? Well, OK. You have to be able to translate and do all these other things as well. Okay, so I don't think these are incompatible statements.
I think we can make them compatible. And the way you do that will give you a lot of clues about how to organize the training. I would also say Tomasa broadly, I think we're only going to use pen and paper. Surely we have to use other technologies such as computers, for example.
If you define the learning outcomes carefully, you can navigate quite successfully between those two positions. It's a question of avoiding the wrong questions, not getting the right outta, just affording the wrong questions. And I get these not questions. I asked statements like get this.
Whenever I do presentations on machine translation will post editing. I, the human can always translate better than a machine. It's an opposition. That's not the question. The question is, is a human with technology kind of translate better than acumen without technology, that is, the machine is not your enemy, it's your friend. And you got to work out the best ways to be friendly with it.
This is just saying that doing post editing is more efficient and gives better quality than translating from scratch. And we can do tests. We do experiments to demonstrate that There are many provides us depending on the quality of the machine translation, depending on your field knowledge, et cetera.
But OK, this is a better statement to pursue then this one, which is just living in denial. To be honest, if you look at the contemporary machine translations, they are not translating, strictly speaking, they are picking up human translates a translations and recycling them in devious and rather clever ways. So we could also retort, the systems are not translating any way, but I won't go into that here. Translators will say, I don't use technology. Most of might say that I, I said, well, I don't know your pen and paper.
That is technology. Writing is technology, using an alphabet? Is technology a computer, a clock, electronic communication, a calculator to figure out how you get paid. All the Internet resources you can, you can incorporate all that is technology. And I think part of the battle here is to make people see that technology is more than machine translation is more than something that's going to take work away from you. It's going to help you work data. What are the things I do at the beginning of any Coles and technology is point them to this rather date of 2016 website where translators talk about the tools that they use.
So that'll free tools around the place, lots of them time-on-task tools, so you keep track of how long you spending on the job. And so you could charge for it if by the hour. And rather than by the word of the things I really useful one that I did pick up from this site was text to speech tools as a revision instrument.
So these days, if I finish a translational, when I finish a translation, I just get the computer to read it out with a voice and I sit back and listen to it and I pick up so many mistakes or in Felicitas, stylistic things that could have been better. I'm, I pick it up because it's sort of more natural listening to the voice, the MAY trying to read a text on the screen. Okay, so it's just just making that step is egg look guys, you're using technology all over the place. Don't just limited to machine translation. Don't just limited to a thing you're seeing as your enemy.
A lot of the debates, I really have been in these debates. I've been in a translation training institution with my fellow teachers and I have had all these very good professionals say that top-level translators do not use translation. Memories all do not post out. And they've just said that two beats. For at that stage, it was a few years ago. I
could get out livelihood. Jackie's research, she did a survey. It was basically in the UK and Europe, but eight showed that 72% of his apple Of the people surveyed were using translation memories all throughout. I say, look, they are using it, pets you or not. But hey, let's face it, we the teachers, we are an older generation and the younger generation and using it, it's not the same for role. And that's a very hot vestige together across the younger you are, the more you have grown up with these technologies.
Not as an add-in, not as something on might do in the second year of my milestone program and we're in the fourth year of my undergraduate program. You've grown up with these things all the way. It's being part of your learning experience.
And I think that's the message we have to get across to our fellow teachers, the people who define curricula and syllabi. If you define the learning outcomes carefully, you can incorporate technologies into the learning process for virtually any learning outcome. Not as an outcome in itself, but as a tool, as something that can be incorporated into the activity. And that's how I think we should be looking at technologies and translate a trainee. And I think in language, language teaching, especially for second language teaching, that is, use it as a learning tool. Not just to empty, but, but empty is the big thing that's advancing.
Going has been radically bouncing since 2016. It can be used as a dictionary, as the comparator to grab a tool as a source of suggestions. And I shouldn't have to show you this, but I can. Okay. This is from I took this from you how again, some 2 thousand, but it's just very useful different ways that people can interact with computers.
Okay? So from humans must make all decisions and actions. At number one and number ten, I don't know why. Ten is at the top where the computer does everything and there is no human interaction with it. Okay? And most of what people who dislike machine translation, most of what they're objecting to is this where the machine gives you one translation and you accept it, right? That, that is not a good way to use machine translation.
It's, it's, it comes from this false belief that this is the translation. It's what people who don't know about translation think that machine translation is doing for them, but you can use these other ways of interacting with computers. Okay? You can narrow the selector, you can use it to narrow your selection down to a few, okay? Or it can give you a whole list of alternatives. And I would be interested in moving downwards on that list.
Okay? But you can also move upwards if you want this. Give you some examples. This is an example I've used elsewhere when you use, here's what Google Translate as an online dictionary, which you can.
It is a dictionary in that it gives you one. Alternative. But if you read down here, you've got a whole lot of information such as what you would find in a normal dictionary and it became big. It can be used in that way as a source of suggestions rather than dictating one translation, JPL does the same thing over here. It gives you alternative translations. And if you're not happy with it, you click on that and you get all these alternatives to choose from.
So it becomes a very quick and easy online dictionary being used in context. Langue, which is associated with the Pell, does the same, but you have previous translations to look at and you can test what you think works by looking at sentences similar to the one you're working on. I shouldn't have to show you these things.
I'm sure you all know them. And then it's just one step to how you're going to use that in a translation memory. Okay? When, what you don't want to do is just accept that as the only viable translation. You can use all of these other tools to think of other things to do. This comes back to a very basic proposition about what translating is and what is the fundamental skill that we're trying to get people to improve in. Now, I proposed this.
Translating is a two-step process. It's a very simple model. You've got a problem in your st, and you generate alternative ways of solving that problem. All these tt's, okay? And then if you're translating, you have to select 1.5 to do it quickly. Now all the technologies we've got, all of them help us with this generation process. It helps us get the alternatives and get some information out there quickly.
And then some of that information, usually the explanatory and contextual information can help us select the one we want. But the main use of the technologies is in the generation of alternatives. The selection still requires soft human skills because there are no rules for a translation problem as such. And it has to be done with justified. That means trustworthy, that means ethical confidence. And this, this is the hard part, this human skill I'll selecting between alternatives in order to achieve a communicative value that is successful.
That is the hot skillet translating. And that's the part that you really have to do. And if you can't do it without a pen and paper, you can't do it.
That technologies should be used across the board at that part of the process. And we have to get people to use very human skills. And the second selective pot so much for teaching with technologies
2021-07-05