Select text to annotate, Click play in YouTube to begin
00:00:20
[Music] conf Eng forence um [Music] on man
00:03:15
for thank you very much and good afternoon everybody good afternoon to the interpreters um feel free to wave at me in a panic if I'm going too fast or not speaking
00:03:27
clearly uh thank you very much for the invitation gu Ando um so I've been asked to talk about uh Ai and the title of my presentation is AI or
00:03:39
DIY a pragmatic look at glosser's terminology and AI Tools in conference interpreting uh DIY in case you're not familiar with the term is do it yourself
00:03:50
in English and I'm going to give you a little spoiler and tell you that although it's not a black and white answer my answer to the question leans in the
00:04:03
direction of DIY rather than AI but I'm going to explain uh the details in a minute I'm going to do something very British now as well I'm going to tell you what I'm going to say and then I'm
00:04:17
going to say it and at the end hopefully I will tell you again what I've said which is how we like to set up our uh presentations so I'm going to give you a quick run through of some of the new
00:04:29
tools that have become available for conference interpreters uh over the last 3 4 years it's all very very new um most of them powered
00:04:42
by either narrow AIS or generalized AIS like um chat GDP and and the like then I'm going to have a look at whether or not actually using AI is
00:04:57
better than doing it self and I think if if I were to summarize the main message today it's that AI is neither brilliant nor
00:05:11
awful if you're going to use it it has to work for you so you have to actually examine what you're doing with the AI and you have to say to yourself is it
00:05:23
quicker is it better is it useful and if it doesn't answer those questions for for God's sake don't use it U A lot of people are just using AI because it's AI oh it's lovely you I'll ask chat
00:05:36
GTP you've got to look at whether it's actually helping you but again I'm going to come back in more detail to that and then as a as a little specific uh I'm going to wrap up looking at glossy building because uh I think glossy
00:05:49
building is something that all interpreters do in one form or another and I I hope that it' be useful for you guys as you set out on your
00:06:00
careers to have a little idea of some of the things that you might want to think about um as you start creating glossies so how to how to create glossies how to
00:06:12
organize them so that they're useful so part one uh which button that one some of the new tools I mean I can't tell you about all the new tools uh or any of them in great detail but this is
00:06:27
what we're going to look at today um we're going to look at Aid driven uh PDF readers which offer some functionalities that normal PDF readers don't offer look at some term extraction
00:06:48
software then we're going to look at the idea of autofilling your glossies so most of us have lots of blanks in our glossies AI offers us the opportunity to fill in those blanks without doing any
00:07:00
work I'm going to discuss whether that's a good idea or not and there are also now in Booth live helpers if you like uh some people call them digital Booth
00:07:13
mates but there are there are software tools that can actually help you whilst you're interpreting and we'll have a quick look at those um I would just say I'm not really an expert I would call myself an
00:07:26
enthusiastic amateur and the real experts in this field are very very few in the world of interpreting and if anyone is actually interested in in Reading what real
00:07:39
experts have to say then you should look up uh Josh Goldsmith at tech for word um he does a lot of interesting stuff and then two
00:07:52
German colleagues Ana ruton who runs a Blog called management. I think and Martin Vil who is a terminology expert and he writes about terminology and I'm actually going
00:08:06
to show you later uh a database a glossery that's kind of inspired by his work the other thing that
00:08:18
um I think is important to say is that interpreters are not um interpreters are not early adopters of new technology
00:08:30
so the tools that I'm going to show you these are tools for the Future these are potential tools I don't think there are six people in Europe using these tools at the
00:08:43
moment in the booth they the Pioneers are testing them out people like anel rut and Josh Goldsmith maybe Martin as well they're testing them out to see if they work so what I'm showing you here
00:08:56
is not a reflection of the current state of the profession I'm showing you the possible future of the profession and just to give you a little demonstration of how slow we are to adapt in interpreting to adopt sorry how slow we
00:09:10
are to adopt new technology um colleague of mine Hong Jang from Hong Kong did a survey of interpreters glossery use in 2015 now
00:09:21
2015 might seem like a long time ago to to those of you who are under 25 but 2015 is is yesterday to those who've been interpreting for a few years and in 2015 six 60% of the people who answered
00:09:35
Hong jang's survey were still using paper sheets of paper for their glossies now one thing I can't quite remember is why these numbers don't add up to 100 because usually percentages
00:09:48
would add up to 100 so I would invite you to look at the methodology of the study um but I think regardless of that the proportions are interesting so 45% was a number using paper notebooks as
00:10:00
opposed to loose pieces of paper uh 55% of interpreters who answered the survey were using Word files so one word file per topic
00:10:12
absolute Chaos on on the desktop and only 15% were using actual glossery software so real software made for the purpose of glossies so just to give you an idea
00:10:26
that the technology we look at today is is going to be a while in the coming uh you know people won't be using it immediately even the stuff that does work so the first thing I wanted to show
00:10:38
you was um a PDF reader called read wise and basically um what this you you you
00:10:51
download it as a plugin to to Chrome and I'll just play this silent film whilst you're doing this so this would be my this would be how I'm Pro how I prepare so I've got to prepare something on co2
00:11:04
emissions from cars and light utility vehicles yes so Vans um so I Google that I find what I'm looking for which is a European commission web page I go to
00:11:18
that page I check that it's you know it's relevant um and then in the top left you see the you'll see the plugin that says open in
00:11:31
reader okay so I I go to the web page I'll I'll click on that and now I'm going to open it in readwise so this is the readwise uh user interface and this is now where readwise
00:11:45
does something a bit different um you you go to that menu and you invoke something called the ghost reader in readwise and you can then do the first
00:11:58
two things either of the first or both of the first two things in that menu which are quite interesting summarize the document or ask the document a question um I can't remember what I've done in this film we'll soon find out I
00:12:11
think I asked for summary and then in the bottom corner this little ghost appears to tell you that the ghost reader is reading and then it produces a four or 500w summary
00:12:37
of the 50-page document which you can copy and paste and read so that is um obviously potentially a very a very
00:12:48
interesting time saver for interpreters prepping um I would say for anything more than 10 pages you know this can save you a lot of time um particularly the the function of
00:13:02
asking questions quite useful you know if the whole document is not really useful but if you want to say as I did with this example I said for example what is the new maximum limit
00:13:14
for parate emission from Cars according to this legislation and it finds the answer in the document for me so that's quite useful on the Assumption the answer is
00:13:27
right and then you have to you have the the trouble is we don't yet know whether we can trust the AI to get the answer right but sometimes it doesn't matter so much
00:13:40
if the answer is wrong it gives you an idea um of what the answer might have been so that's the first Tool uh second tool
00:13:52
is term extraction and again uh this is is a way of dealing with with very big documents so this is um this is a thing called
00:14:06
sketch engine and what it allows you to do is bring in two documents in two language versions and then click extract and it
00:14:18
will find in the documents what it thinks are the matching terms and it will present them to you and this is one of the one of the strong points of sketch engine it doesn't just
00:14:31
give you the matching terms it presents you one language and then gives you three or four options for the other language and you have to kind of Click just to check that it's actually the
00:14:42
right one uh and that that engagement is very important because I'm I'm going to come back to this if the computer does everything for you you haven't done anything for yourself and what doesn't
00:14:55
go through your brain what you don't process yourself is not in there so you can't get it back out so this is why the title of this presentation is AI or DIY that the DIY is is is also thinking for
00:15:09
yourself because you can't remember something that you've never known um so this is quite useful because you do actually have to engage with the terminology um and another nice thing about sketch engine is if that you're
00:15:21
not sure about the terminology you click on one of those terms in the list and it'll show you EX L where it is in the two documents and then you can actually read the two
00:15:34
sentences um and so the first one is probably too small to see on the screen but it's quite interesting that it picked the completely the wrong word for the French for the English but I can
00:15:47
look and read that sentence and I can see what the french should have been and then I can go back to this list and I can I can click the right one
00:15:57
um and then what sketch engine will do you click on export and it'll export the two language versions that you've chosen in a table like that which you can then
00:16:11
use as your glossery cheat sheet or whatever for the meeting um and I don't know if I mean it's not the subject of the presentation but if any of you like using flashcards to
00:16:24
learn vocab those two column tables copy straight into flash cards like Quizlet and you can use them uh to uh to revise your
00:16:35
vocab okay um next up we have glossery autofilling so there's been an awful lot of uh stuff going on
00:16:48
online where people have been saying oh this is brilliant look I can get chat GTP to fill in my glossery or even create a glossery for me and I don't have to to do anything
00:17:01
um and this is a demonstration so this is a tool called interpret bank which was developed fairly close to here I believe where does Mr fantoli lecture here or in
00:17:13
gersheim yeah so um and this is what happens if you this is how you create a glossery um so the first part isn't
00:17:26
necessarily well I'll show you just for the hell of it so you you create a glossery like that with a title of detergents and for example I've got a list of English terms which I'm
00:17:38
going to import from a Word document so I choose my word document this is just a mini glossery of 10 terms for the for the purposes but this is the interesting
00:17:50
bit in before Imports it asks do you want to automatically translate the terms into German so it automatically translates them into German so from a monolingual glossery
00:18:03
gone to a bilingual glossery now the the thing is that people have lots of people online have been saying wow that's absolutely amazing brilliant I never need to look anything up again okay
00:18:17
trouble is I um I obviously did this process backwards so I started with a two two language glossery of my own which which I knew to
00:18:28
be correct because it had been checked and it had been made during real meetings with real speakers and real experts and that is what the 10 terms
00:18:41
look like when I actually compared them to the correct terms so that's 10 terms five of them are wrong one of them is not quite
00:18:53
right and that that is quite a poor return on your effort and the other thing apart from wrong is that it can't offer me two versions of wash liquor in my glossery I
00:19:08
have vasher for wash liquor but I also have an alternative German word for Vash lior for for wash liquor Vash laa which I can no longer remember but I know
00:19:21
there are other words and of course if I work from German into English I want to know all of the possible words in advance so that I'm not surprised by them so this is you know this this is an
00:19:33
issue with automatic uh filling and you know there's this well we'll come back to this but if you have to check all of the words in a glossery and you do then you
00:19:47
have to ask yourself wasn't it quicker just to go and look them up in the first place this is this is this is the problem um so you you have an advantage with this sort
00:20:01
of tool because this is not the only sort of tool um I'm not picking on interpret Bank you can do this with chat GTP you can do this with notion you can do this with interpreters
00:20:12
help um I think and yeah you can bring in your own glossies that's a great plus but um it has these failings
00:20:27
it it will not only get things it's getting things wrong for a number of reasons um it doesn't recognize technical fields for example yeah so this technical field is about detergent so
00:20:40
scaling when you're talking about detergent you're talking about washing up machines so it's bags building in a wash in a dishwasher yeah it's not
00:20:53
shopen I don't have shopen but you might have shopen in German um so that it it's it's very deceptive
00:21:05
particularly for beginner interpreters it looks like it might be right but you can make uh terrible mistakes if you're not checking um what's going on there ah okay yeah sorry I've got
00:21:20
another autofill for you this is um this is notion so notion I believe is a is like a user interface that you can put on the front of chat GTP it just looks a bit more friendly but I think behind the
00:21:32
interface is still chat GTP so I asked it for uh what did I do I asked it for a glossery so I said can you give me the most frequently used terms in exhaust filter technology
00:21:59
now you might say you might say wow these are crazy technical areas detergents exhaust filter termin technology this is where we need the terms you don't need the terms in
00:22:13
everyday English and German you need them in the very technical stuff so that's the uh the reason for so you can see it's quite a nice process you ask it a question it gives you some ter
00:22:28
terms um I would argue with the term Ash for example the others are actually relevant terminology Ash is is is not it's wrong
00:22:40
um and then I ask it to translate those terms into [Music] German for oops to stop it there so it's not bad as a
00:23:39
list but the trouble is we don't know if we can trust it so I mean I happen to know that most of those terms are actually useful
00:23:53
and I I can't judge whether they're well translated into German but I think they are the English terms seem okay except for ash that Ash is not what comes out of the back of a
00:24:04
car fine partic or fine particulate pollution um but certainly not Ash and now the problem is you have to check the terms if you
00:24:17
can check the terms you're already an expert so you didn't need to do this and if you can't check the terms you're not an expert so you shouldn't be doing this and this is this is where the the Paradox comes um and if you have to go
00:24:31
and I mean like I say I think DIY you you should probably go and find out what these terms mean and and what they are in English yourself because although this might be right you can't be sure
00:24:44
and it's the it's the not being sure which is which is the problem uh okay next tool um actual in Booth help and this is uh you know this
00:24:55
is really quite a cool little tool it's called Simon note and you plug it into the audio and it will do a live transcription of what it
00:25:07
hears it will color the numbers in blue and if you input your own glossery into Simon note before the meeting when it recognizes a glossery
00:25:19
term it will highlight that in red and offer you the translation in Brackets okay so it looks like this this you won't hear but for this demo I'm basically reading to it in English and
00:25:34
I've already input a a glossery into the system and it's about uh cigarette manufacturer this glossery it's a it's a patent in fact so it's quite amusing that it gets
00:25:53
its own name name wrong right at the beginning it's Simon notes um so I'm reading to it at this stage and you can see already the number is in blue so that's quite useful if you're interpreter the number stands
00:26:10
out and now you can see how the red the English is C electronic cigarette tip and then the German is in Brackets after that from my glossery not from the
00:26:21
Internet or anything actually from my glossery so again I mean this is a potentially fabulous tool um you know we we write down numbers for
00:26:44
each other are we humans in the booth so having a machine do it seems like a good idea but it does have some problems um
00:26:55
so if you can see here let's have a look this 180 band right that that was actually in English the word toe band which is the front end of a
00:27:08
cigarette it's the name in manufacturing of the the part where the filter is that end of a cigarette and this was the first term in my glossery which I put
00:27:21
into this tool and it couldn't understand it either in English or find it in the glossery and this term came up three times in this six line extract here so it's called it a
00:27:35
180 band here it's called it a toy band and then further down it calls it a toe bar so it it's again potentially fabulous
00:27:51
but actually unreliable and then you need to ask yourself the question do I have the mental capacity to interpret simultaneously and work out in my head
00:28:04
whether the computer is telling me something useful or not um I don't know if you guys have done simultaneous with text yet so I
00:28:17
cannot interpret simultaneously and site translate at the same time so when I look at a text and start reading it I can't interpret anymore so my strategy for simultaneous with text is to read
00:28:31
the text in advance and then put the text to one side because I can't do both at the same time some people can but I can't so then you have to ask yourself the question can I read the computer screen work out whether the computer
00:28:44
screen is talking rubbish or not and interpret at the same time I can't so I can't use this tool um or I don't at the moment other other very fundamental
00:28:57
issues with this tool you have to have a good glossery to go into it in the first place okay and if you have a good glossery which I'm going to suggest you will have made yourself if you have made a good
00:29:10
glossery yourself you don't need the computer to tell you what's in the glossery because you will remember it because you made the glossery yourself so it it becomes a sort of Zero Sum game you know it's it it looks nice but is it
00:29:23
worth it um there's also a c I mean this thing costs money um and it it wasn't cheap well I I did an experiment and I think it's it was
00:29:36
something like between 10 and20 EUR for an hour right so over a day I mean you can get monthly subscriptions and what have you and there may be cheaper versions but this this tool was not being given
00:29:48
away for free and quite rightly so I mean you know it's an extra it's a sophisticated tool but it costs money the other thing that you have with this tool is
00:30:02
um confidentiality um does your customer want a written record of the meeting because this is a written record of the meeting and that may not be
00:30:14
desirable uh it may be for confidentiality reasons or it may be for example I work as does T at the European patent office the meetings are not confidential but recording of the
00:30:26
meeting is not allowed and this is a recording of the meeting so this is not allowed at the European patent office where it might be quite useful but it's not allowed okay and one other in Booth
00:30:39
tool this is just a screenshot oh sorry I haven't put on there um oh hello there we go so this uh is taken from the website of a company called ibridge and this is
00:30:53
live transcription and translation at at the same time into text so it's it's almost interpreting there just isn't the voice synthesizer to speak the
00:31:06
translation um so this is a this example which you can find on the website ibridge people um this is a speech by Angela Merkel and you can see you'll see that
00:31:21
you know you'll see the demo on the website it translates it into I guess that's Spanish at the bottom is it uh Spanish and English and it transcribes into German but you can already
00:31:32
see it can't do everything because what she says is she's talking to Mario dragi the chair the president of the ECB who has been her political counterpart
00:31:43
partner for many many years and she turns to him and in German says it's Li Mario because they're friends and the computer can't work that out so it
00:32:00
saysex which is complete nonsense and so again is it is it helping that that's the that's the ultimate question so let me let me just sort of recap now
00:32:14
on uh that's that button isn't it oh hello not that direction here we go so those are the new
00:32:30
tools now let's look at this question just for a moment AI or DIY um so here are I think some of the reasons why DIY
00:32:44
beats AI um I noticed one or two of you taking notes on the computer now I don't want to pick in anybody out but it is scientifically demonstra ated that
00:32:57
handwritten notes go into your brain better than typed notes because you have to think a little bit more about what you write on the page just like in consecutive interpreting and because you
00:33:11
think about where it is on the page you remember it if you don't process you can't remember there's also a very human element of sort of geospatial if you write with your hand
00:33:24
you can actually remember oh yeah I wrote that at the top right hand corner I wrote that bottom left when you type on a computer even on a I find even writing on an iPad you lose the
00:33:36
geospatial element of where you put things so engagement with content is is quite important if you want that content to be available to you as an interpreter and
00:33:49
when I say available I mean can you recall it when you're in the booth can you remember what the threshold limit value for particular
00:34:01
emissions from Cars was in the draft legislation if rewise told you you probably can't remember if you read it yourself and wrote it down you probably can remember
00:34:16
so um and and again the the summary of documents the extraction of terminology it's very good
00:34:28
but um automation I believe in interpreter preparation is a zero sum game so what I mean is zero sum it adds up to the same
00:34:41
if you automate here then you have to activate here so let's if you if your preparation process involved reading something yourself and
00:34:53
then going into the booth and you could remember it it but then you decided right I'm not going to read it myself I'm going to automate the reading and use readwise then over here you have to add
00:35:06
an activation phase you have to do something to activate the knowledge otherwise it's not available to you in the booth so with terminology extraction same thing if you read a document and
00:35:19
you write out the words yourself you're activating them just by writing them out you will remember them because you've written them out if the machine extracts them for you then you
00:35:31
have to add flash cards later on to activate and you have to test yourself on the vocab to activate now that's not a bad thing I don't have a preference but you have to be aware that the
00:35:43
automation is not saving you time the automation is Shifting the hard work from here to here so if you don't like reading get the machine to do it but be aware that later you're going to have to
00:35:56
maybe do some flash cards you can't get round the hard work you can just move it around in your PR preparation process so it's a zero sum game
00:36:07
preparation is not quicker with AI it's different uh and I mean there are so many possibilities you just have to look at them all with a critical eye choose the one that suits you but don't think
00:36:21
of it as a Magic Bullet um reliability so I've mentioned this a couple of times already uh checking whether or not the
00:36:37
output of a machine is reliable or correct whilst you're working in the booth is extra cognitive load and most of us are already maxed out on cognitive
00:36:52
load um I don't know you know like I say I can't site translate and do simultaneous interpreting I can't read a book and do simultaneous interpreting I can't check
00:37:04
whether notion is right or not and do simultaneous interpreting so um if you don't believe the machine that's a real problem I mean one thing used to be an
00:37:19
absolute Golden Rule if you were going to write down a number or a bit of terminology for your colleague like human to Human in the booth it absolutely had to be
00:37:31
right you can't present the wrong answer to your colleague that you they they'll just kill you for it afterwards so the problem here is that we have the machine giving us the wrong information and
00:37:43
we're too trusting of the machine um and again I think I would repeat what I said earlier you have to check ai's output from
00:37:54
terminology and if you if you can check then you're doing your own preparation so the EA AI wasn't necessarily useful so if you know nothing about the topic and then you go
00:38:09
and check online then the AI is not saving you time if you can check you know how to check because you're already an expert then you didn't need the AI in the first place and if you're not an
00:38:23
expert and you can't check then you shouldn't be trusting the AI so for me um a lot of the time the DIY is better than the
00:38:37
AI I mean there are other things as well you know AI is strictly limited to the internet and so if it's not published on the internet the AI can't find it but a lot of the meetings that we're doing are
00:38:52
absolutely Cutting Edge so I mean the patent office for example Le there's not a lot of non-patent literature that is published about this stuff because it's brand new but that goes for everything
00:39:04
every every meeting is a technical meeting and it's not all published on the internet um and you know the reasons we had the reason we had glossies 20 years ago was because the information wasn't
00:39:17
available publicly if you work at NATO for example none of that is available on the Internet it's all top secret so everyone has their own glossery because otherwise
00:39:28
there is no glossery um so AI or DIY number three is it actually saving you time and effort that's the question that you should ask yourself when it the answer might be yes
00:39:41
in which case great um if AI is s saving you time and effort and offering you the right answers then go for it use it but do approach it with a with a critical
00:39:55
critical eye um 2 uh 2.3 third yeah so confidentiality and recording restrictions I mean again I've touched upon this but just taking my own personal um situation I work in one
00:40:11
institution where all meetings are confidential so none of it none of these tools would be allowed in um in that one the European Space Agency in Paris absolutely none of these tools will be
00:40:23
allowed because all the documents are confidential at the European patent office none of the documents are confidential so some of those tools will work but the in Booth helper would not
00:40:36
be allowed because that is recording the meeting and uh the European Parliament the European commission um some of the meetings are confidential some of them aren't so
00:40:49
there would be a mixture and my feeling is if you can't use the tool every single time it gets to be a bit of a hassle that oh today I'm using it tomorrow I'm not using it you you you
00:41:01
you sort of it becomes a cognitive effort to work the tool because you're not working the tool every day it'd be like changing from Mac to Windows every two days it's like oh c how do I do it
00:41:15
how okay it's there and that's not the sort of question you need to ask you don't want to ask yourself that sort of question in the booth you don't have time so you want everything to be very quick and the other the other thing about
00:41:29
confidentiality and and recording restrictions in my experience when you ask a customer an institution or somebody who works or has recruited you
00:41:41
whether you can use a tool in their meeting they simply don't know what the tool is they don't know whether the tool is secure or
00:41:55
not um so for example sketch engine has an ISO Standard Security label it has a label but I mean what does that mean and so what I've had happen to me is I ask
00:42:08
if I can use the tool the person opposite has no idea and they either just say no because the no is the easy answer and they don't want to investigate or they send it up the
00:42:20
hierarchy in the institution and then my request disappears into the institution comes back weeks later with a no because I'm a very very interpreters are a very
00:42:31
very small set of fish in the big pond of the institution and They Don't Care About Us using this tool so it's just like no so there are lots of reasons why
00:42:43
these tools end up not being uh very practical uh to use next there is a technical barrier I've already said you know that interpreters are very slow to adopt you have to learn to use a new tool and you have you have to be able to use a new
00:42:57
tool very very quickly without thinking about it because you you got no time or brain space to think in the booth uh so you have to do a lot of playing with a tool outside the booth in order to make
00:43:10
it work in the booth and most interpreters stop at the first barrier I don't know about you but most internet services I'm I'm not even going to log in if I have to log in and
00:43:23
register I that's enough of a barrier and I just stop and there are very few you only need one or two technical barriers most interpreters uh will stop and and cost I mean interpreters
00:43:36
hate to say it we are very very stingy um and most interpreters don't pay for software at all I mean maybe your generation is more used to paying for software but when I was 25 30 software
00:43:50
was all free everything was free and now paying for it seems a bit weird but a lot of people also like I can't be bothered to either learn to use it or pay for it why would I pay for it so
00:44:04
that's stopping interpreters um using these tools as well so to to sum up on the AI or DIY thing I would say um experiment with it by all
00:44:18
means but keep asking yourself the question is it really helping me am I really saving time are the answers really right and and where you guys are at the moment learning to interpret I would really
00:44:31
say Do It Yourself DIY because you want stuff to to you want to process stuff mentally yourselves so that you remember it you understand
00:44:43
it learning how to analyze a 50-page document is part of learning to be an interpreter learning how to read a 50-page document in 3 minutes
00:44:56
that's part of being an interpreter you can ask your computer to do it but you don't understand how the computer has done it so you've missed something about the document but knowing to look at for
00:45:10
example the table of contents in a document knowing that most documents have an executive summary of one page knowing that French documents have the table of contents at the back not the
00:45:23
front these are useful things for interpreters to be able to do instantaneously in the booth um and and again I think if you're learning to interpret it's is probably more useful
00:45:35
to do it yourself so let me now come to the third part of uh this presentation and this is something about
00:45:45
glossery building and glossery building ourselves and for this um this part of the presentation I owe a vote of thanks to Martin will who by
00:46:03
chance became my sort of mentor in things glossies about 25 years ago because we worked a lot together and he was very big into glossies and databases at the time and
00:46:16
um the ideas that you're about to see are really derivative of his ideas and his Publications which go into much more detail than I will today so
00:46:28
um before I get to the list on the screen I think what what's interesting one thing that's interesting about glossies and databases is that the the perhaps the one of the main reasons for having one in the first place has
00:46:41
changed in the 25 years since I started interpreting so 25 years ago you could not get hold of technical glossies anywhere there was barely an internet to
00:46:54
speak of and I would come into the I would come into the booth either with my an dictionary of whatever it's called industry Technic brilliant dictionary or
00:47:08
the other brilliant dictionary schaer V also beautiful purple and silver but there there were only dictionaries there were not online glossies so interpreters
00:47:20
created glossies so they had something relevant to take into the booth with them because they couldn't find the information anywhere else nowadays it's almost completely the opposite there's
00:47:32
so much information available on the internet that the the question is not is there any information it's what is the useful information out of this massive massive pile of information I've got in
00:47:45
front of me um but yeah and the AR the answer can be the same and that is having your own glossery um so um what can a glossery do for you
00:48:05
well if you're prepping for a meeting um well let's um let's have a think yeah so if you if you read Martin
00:48:23
Will's uh work you'll see he he often talks about actually three stages to to glossery building and this might surprise people but there's a preparation stage before the meeting
00:48:37
where you write stuff down and then there is a stage during the meeting and then there's a stage after the meeting and
00:48:50
things happen with your glossery during all of these stages um so when you when you look up brand new words and Concepts Wikipedia and go to technical websites um you're looking you're
00:49:03
reading definitions you're you're actually activating just because you're engaging with the terminology and you're writing stuff down um and you'll be writing down terms that you already know that you want to
00:49:17
activate and terms that you don't know so you create a glossery ahead of a meeting and you're learning stuff as you you go along then during the meeting you'll be
00:49:29
doing two things one you'll be using the the glossery list that you've made looking at looking at it drawing on it from from memory um but you'll also be checking
00:49:44
your glossery list against what's actually said in the meeting because really the the ultimate Arbiter of whether the glossery is right
00:49:56
or not is that meeting the person in that meeting Wikipedia can be wrong an can be wrong uh even duden can be wrong as far as The Interpreter is concerned
00:50:10
the the guy in the meeting if the German delegate says it then for that meeting it's right might not be right in another meeting but it's right in that meeting so you're you're checking your glossery as you go along and you're also
00:50:23
looking stuff up uh um in the booth now if you want to use your glossery to look things up in the booth so actually type in
00:50:40
um the AL English and you type it in what your glossery needs to do for you is give you access to everything that's in
00:50:53
it when you're looking up a word it can't just be looking in a small bit of your glossery it's got to look everywhere otherwise uh it may well not find it
00:51:04
um so you probably want things in one file uh but again there's definitely two different sort of schools of practice some interpreters do look things up in the
00:51:26
booth like actually whilst they're interpreting they'll type a Word into a computer and the computer will give them an answer and there are a lot of other interpreters who use a computer for everything except that because they
00:51:37
don't have the brain space to look something up and interpret at the same time and I'm one of the the latter um and then the third phase is like post
00:51:53
preparation so after the meeting you've got notes that you've written down whilst you in the meeting you've got your glossery you've got what you can remember and you bring them all together to correct the terminology list and put
00:52:06
that into a glossery for next time and then the next time you do the same meeting because usually we do the same meetings again and again you've already got uh the glossery
00:52:18
available so creating a glossery is is learning about stuff it's creating a glossery when no glossery exists for very technical meetings it's activating the terminology so you can actually remember it
00:52:32
instantly when you need it it's also there for looking stuff up in the booth and then when you get to this post preparation stage it's even going to help you
00:52:45
remember in the long term most of us are only worried about being able to remember terminology that day and we forget 90 95% of stuff the the immediately afterwards but if
00:52:58
you do Post preparation and you write stuff into your glossery and correct it uh you tied it up sort of thing uh you actually find that you you remember stuff for a bit longer few more
00:53:12
weeks um so if you do a similar meeting a couple of weeks later you have to do less preparation and then of course the glossery gives you preparation for next time uh so very often I I I do quite a
00:53:25
lot of meetings about detergents so um washing up powder and washing powder for clothes so dishwashing and washing machines um and
00:53:40
so I can just pull up a glossery and see immediately sort of 20 key terms like T seed which is surfactant and remind myself because
00:53:52
it's not a word I use in my everyday life life and okay next time I'm ready to go before I've even looked at the before I've even looked at the documents for the meeting I've got 10 15 key terms that I've just
00:54:06
reactivated so [Music] um how we doing time wise got lots of time left so let's have a quick look a very practical step by step how this is
00:54:18
how my glossery is databas is built up 10 um and as I say very much based on the ideas um that I was shown by by Martin
00:54:33
Ville um so it seems perfectly logical for an interpreter to organize their vocabulary by customer so one day um one day I might be working for an
00:54:48
aircraft manufacturer and another day I'm working for a patent lawyer or the patent office and I've got two different meetings two different sets of
00:55:06
vocab but let's say they what I find for any single meeting is that you you have actually two separate types of terminology in a
00:55:17
given meeting you have a technical topic area and then you have the inhouse administrative stuff that is specific to that institution to that
00:55:30
company to that customer and that is not about the technical topic so for example uh I do a Works Council which is a union representative body for an aircraft manufacturer so they talk about
00:55:44
two things they talk about aircraft and they talk about workers rights and how the company is set up very different things but always the the same too and then when I'm doing patents for example I might be talking about
00:55:58
aircraft again but the in-house stuff is very different then it's patent law and how patent law works so um like I was saying a lot
00:56:14
of oh I've lost a I've lost a slide have I um there we go yeah so a lot of interpreters will have two Word files this is supposed to represent two Word
00:56:28
files on a computer and it doesn't matter whether you can read the terms or not um but the blue is about aircraft and the the black is about other stuff but if I've got my aircraft what happens
00:56:41
next time I've got a meeting I know for a third customer about aircraft but I've got my aircraft stuff in two glossies so that that that's not helpful
00:56:53
is it I want to put them together so how do I put them together I create a single table and in that table I have two extra columns I have German I have English
00:57:06
then I have topic area whatever you want to call it subject topic and then I have customer name so that's what these two mini glossaries look like there were 10 words in each
00:57:18
glossery and now I put them into a single table and then what um what you can do what have I done there is filter by filter by aircraft so filter means I
00:57:45
go into this column here and I ask the computer to remove everything that doesn't say aircraft and because computers are good at this
00:57:59
stuff it does that for me and it brings together the 10 terms from the two different meetings about aircraft and there I have my technical aircraft
00:58:12
terminology um so I use uh Microsoft Access which allows me to look for aircraft in one column and
00:58:28
then also arrange the French column alphabetically for example so that but you can do any any column alphabetically um I can also of course do it the other
00:58:43
way around I can still filter by customer I can still go to my aircraft manufacturer choose all the vocab for my aircraft manufacturer
00:58:55
I can also choose aircraft manufacturer and aircraft filter by the two columns because as we'll see in a minute sometimes different institutions have different words for the same
00:59:10
thing so the advantage of of a s having all your stuff in a single table is when you search for something if it's there it'll find it because everything is in
00:59:21
there um you can filter so you can still narrow it down and you can sometimes you can see whether there are uh two terms I'll I'll show you this in a
00:59:35
second so in um again I'm I'm I'm selling the method here not the tool I use Microsoft Access
00:59:48
I think Microsoft Access is very powerful in terms of filtering uh but there are other tools that may be as powerful I think noox is
01:00:01
one that I've just been told about let me just check that yeah noox is an online database um interpreters help has a database as does interpreting interpret bank but
01:00:15
they don't allow quite so many columns to filter so in Microsoft Access you have these icons which mean filter and you can filter away what I think is nice
01:00:32
um as I was saying sometimes the same word means different things in different places and because you have this metadata this information that's attached to the vocabulary so who's the
01:00:46
customer you can see that for example cman in the European Space Agency is AB orong in German but in the European institu the EU institutions it's en zendon and that's
01:01:00
useful if you work in both places and only by having all of your terminology in one place and having the terminology Associated to a customer
01:01:13
can you see that uh at a glance so this this format obviously is expandable pretty much add infin item so if you imagine here are our two original customers the
01:01:29
aircraft manufacturer the patent customer then we have another customer an international institution that decides to talk about intellectual property so we got intellectual property in green we've got aircraft in
01:01:41
blue we can still put them all into the same single table but this time I filtered for patent vocab and now I have the patent vocab from customers two and
01:01:54
three instead of the aircraft vocab from patent from customer 1 and two and that basically goes on for as many customers as you like
01:02:10
um one thing I would I would tell you as you as you set out on glossery building is that you can design your glossery or your database carefully but you can't really
01:02:23
know how much aircraft terminology is going to be in it in 10 years time or how much intellectual property terminology is going to be in it in 10 years time so for example what happened to me I had
01:02:36
lots of patent stuff and then I realized okay it's not all about patents some of it is broader my intellectual property so I have another column so I have
01:02:48
English German topic subject topic actually the other way around in this table I have subtopic then topic then customer and then I can
01:03:01
I can filter intellectual property so I have something quite broad or I can filter by P patents and have something quite narrow um and actually over the years because I'm an obsessive
01:03:15
collector things have got bigger and bigger and bigger and I have a third I have a sub sub topic because what I want to filter out when I'm prepping is a list of 30 odd words some I don't want 500 words to
01:03:28
look at when I'm prepping I want 30 because that's manageable uh let me just give you another not sure that slide's in the right place but look just in case you didn't believe me
01:03:41
um and again this is why the internet and glossies are uh not always here's an example of some words that mean different things in different places so the Polish word
01:03:57
scon means two different things in the European court of justice in Luxembourg if it's in an antitrust case it's called the that is called the
01:04:11
complainant if it's in more General European court of justice procedure that word is translated as applicant and because it's very legal you have to get it right so this is the value of the
01:04:23
metadata the extra columns tell you oh yeah okay I can't just see chat GTP will just say applicant to you and you say applicant and it's wrong you don't know why it's
01:04:35
wrong and no one's told you it's wrong so this the metadata is really useful similarly if you were to talk about bag's buildon on the blades of a gas turbine in an airplane you would be
01:04:48
laughed out of the room because it is the rather beautiful word fund in German and not but in English they're both scaling so again it's useful to know that one comes from HL the detergent the
01:05:01
German detergent Giant and the other one comes from Airbus the aircraft manufacturer and similarly uh Esa means at least three different things in different
01:05:16
places so look this is just to uh to show you quickly what my a bit of my glossery look like I actually have another four or five columns so I have
01:05:29
um I have English French German and polish then I have subtopic no sorry I have sub subtopic subtopic and topic then I have the meeting which is
01:05:43
the customer then I have the source for the English of the German I have the source for the French or the Polish and that's really useful as well because in 10 years you won't remember where the word
01:05:54
came from in 10 years you can go oh no that word it was my colleague Gilles told me that not sure I believe that but you might have written anst comes from the an dictionary ah okay
01:06:08
that's good if it's from an it I can believe it the date is good as well rather embarrassingly if you look at the top right hand corner here you can see that my glossery was active in
01:06:20
1997 so I might be a little bit more suspicious about using those words because they might be out of date but my glosser is going to tell me that and then I also have a little column for
01:06:31
definitions so extra just text and um this even allows you to link to files so you might just have a a word file or or some company document that's useful that you can link to from the the
01:06:47
database um this if we got still got two three minutes no I can skip this one two so this is just to show you how this how this works so if I I want to prep
01:06:59
something okay I had an I had a meeting about the electricity Market a couple of weeks ago and I know that I've got electricity in here so I go to my detail column and I look for electricity and I see that I've got I've got a topic or
01:07:13
subtopic called electricity generation so I click filter by selection I get all my electricity generation terminology and you can see there's 91 terms in the bottom left hand corner but as I'm
01:07:26
looking through it I noticed that some of them are not about the electricity grid they're very much about wind turbines and so I can go into the advanced
01:07:37
filter um and I go into another column a subject column and I'm going to write not wind turbines and then the glossery is going to show me everything that says electricity in one column but it's going
01:07:51
to take out wind turbines cuz I don't want wind turbines and so there you have and I've got that down to 70 terms about the European electricity market so as I say there's three phases
01:08:08
to glossery building there's before the meeting there's during the meeting when you're checking making sure they're right and maybe looking stuff up and there's after the meeting where you can correct your notes tidy your notes up
01:08:21
enter them into your glossery and give free reign to your obsessive compulsive disorder as I do when I'm organizing things in my glossery um but you know
01:08:33
just to give you an idea the power of the human brain because I've done all this myself if you give me a word in English or German I'm 99% certain I can tell you whether it's in my glossery or
01:08:46
not I might not be able to translate it I might not be able to translate it but I can tell you whether it's in there because I did the D iy and not the AI so AI is good could be good be critical
01:09:00
about it and my recommendation initially would be DIY thank [Music] you
End of transcript