Skip to main content
;

LANG Committee Meeting

Notices of Meeting include information about the subject matter to be examined by the committee and date, time and place of the meeting, as well as a list of any witnesses scheduled to appear. The Evidence is the edited and revised transcript of what is said before a committee. The Minutes of Proceedings are the official record of the business conducted by the committee at a sitting.

For an advanced search, use Publication Search tool.

If you have any questions or comments regarding the accessibility of this publication, please contact us at accessible@parl.gc.ca.

Previous day publication Next day publication
Skip to Document Navigation Skip to Document Content






House of Commons Emblem

Standing Committee on Official Languages


NUMBER 007 
l
1st SESSION 
l
42nd PARLIAMENT 

EVIDENCE

Monday, April 11, 2016

[Recorded by Electronic Apparatus]

(1530)

[Translation]

    I call the meeting to order.
    Pursuant to Standing Order 108(2), we are continuing our study on the Translation Bureau.
    Today, we will be hearing from Louise Brunette, a professor at the Université du Québec en Outaouais.
    Welcome, Mrs. Brunette.
    Then we will be hearing from another group of witnesses.
    Mrs. Brunette, you have 10 minutes in which to make your presentation. After that, committee members will ask you questions.
    I am not very well versed in procedure, so you will forgive me if I forget to use your proper titles, like Mr. Chair and so on. It could well happen.
    I would like to start by giving you a little biography of myself, to justify my appearance before you today.
    I know this might not seem to be the case, but I have very close to 50 years of experience in translation. I started very young, around 25 years old, working in translation in areas as varied as banking, heating, and the media. So I have a good idea of what a really intense and varied translation practice can be like.
    I have worked as a translator, a terminologist and a translation services manager, which gives me a good overview of the profession as a whole.
    After 30 years of that varied practice, I decided to do a PhD, because I wanted to be listened to when I spoke. I have been teaching since that time. I teach translation, the basic element, as well as revision, which, in translation, is quality control. That led me to machine translation. I have also taught machine translation and post-editing, which I will talk to you about later.
    I am going to let others deal with the political and sociolinguistic issues. I will not be talking about that at all. That is not my goal today. Instead, my goal is to rehabilitate the translation software. I will probably be one of the only ones to do so, but that is not the crux of my message.
    The crux of my message is to show that present circumstances do not really allow us to use what I consider to be an excellent machine translation system in the best way. That does Portage no favours, a system of which Canada can be proud. In international competitions, this system regularly places first. It is one of the jewels in the crown of technological innovation.
    Contrary to what the media has been saying recently, the problem has absolutely nothing to do with the performance of the machine translation software. I would like to point out that all the examples that I have read in La Presse and in other media are completely inappropriate. Let me give you one. People have been laughing themselves silly about the expression “it's raining cats and dogs”. No one uses that expression in the public service. I do not see why we are hung up on it. It is an old-fashioned expression that I feel has no place in the current language of public servants.
    Let me read you a commercial definition of the system that exactly reflects how I feel about it.
    Portage is a statistics-based software system that yields far better results than earlier attempts to automate the highly nuanced art of translation.
    We are not talking about linguistics. Machine translation software does not translate languages; that is not what it primarily does. You end up with a translation, but the system works by statistical analysis. This is about mathematics, not about language. The system understands nothing. It just understands the data it is given and the data it has already stored. It makes comparisons.
    However, I am less in agreement with the words “highly nuanced art”. The software works on binary coding: 1,0,1,0,1,0. There is nothing highly nuanced about it. The program depends on the machine learning statistics. It really is “garbage in, garbage out”. It absorbs what it is given and it gives back what it is given. If what you give it is not good, then the product it gives you will not be good either. Really, it is no more complicated than that.
    Why did I decide to come here? Because, with things as they are at the moment, I was wondering where we are going. This is a three-fold distortion of machine translation.
    First, I feel that the use the Translation Bureau itself had in mind was not the generalist role the system currently has. For example, translating emails comes under the heading of general texts, whereas that is not what the system was intended for. Nor is it what its designers intended it for. They always have been conscious of the fact that, just like at the very beginning and all through the 1960s, machine translation of general texts will never reach the quality that humans can achieve. The designers say so too, they are not kidding themselves. There really has to be preparation from below and from above.
    From below, we have what we call the corpus. I have told you that it is a statistical analysis system. It is going to analyze in terms of what it already has in its memory. There is the machine translation process, the computerized translation process, which ends up as a linguistic product, a text, which then also has to be refined by humans in a process that we call machine text editing, or more commonly, post-editing. So, if there are no humans on both sides, the results are certainly going to be terrible. The software designers recognize that themselves: they never thought that they would end up with quality translations. However, it seems that some other people believe that you can do so, and that is why they want to install the software at all costs.
    The other distortion is that, by implementing the system immediately, we are going to deliver a fatal blow to the development of machine translation, in my opinion, because we are harming it a great deal. But, as I was telling you earlier, it is one of the jewels in the crown of the country's technological innovation. So if we want to kill off all the enthusiasm for machine translation at the outset, as a discipline itself, we have found a good way to go about it. If we put the system into operation now, we will certainly end up with gibberish, just as we have read in the press, and that will harm the reputation of Canadian machine translation researchers.
(1535)
    Now I am going to talk to you about post-editing. I am less familiar with corpus development, which is an area that has much more to do with linguistics, than with quality control. Post-editing is a quality control operation. I will not go into the details, but I also feel that, even with post-editing, we are not completely assured of quality machine translation.
    There are various reasons for that. Among them is that post-editing as it is presently conceived is all about speed. So, the quicker you work, the less freedom you have, for example, to work with the sentences or to make the texts more idiomatic in the target language. I am not even talking about French. It must be said that current software like Portage is not so bad at translating idiomatic expressions. Language that is more idiomatic in general poses the problem, in fact, not idiomatic expressions themselves. However, post-editors are working at such a rhythm that they cannot restore the idiomatic aspects every time.
    Mrs. Brunette, can you perhaps conclude your remarks so that we can take questions from committee members, because we have to hear representatives from a second group now. Then we will—
(1540)
    Can I give my recommendations? I will be quick.
    Go ahead.
    Here are my recommendations. First and foremost, machine translation must be dealt with as a long-term pilot project overseen by professional, certified translators, such as those in the Translation Bureau, for example. Then, post-editing workshops need to be established, because few people know how to post-edit, since my university is the only one to teach it. There must also be investments in corpus development and, above all, experts other than software designers must be consulted, since they are pretty focused. You certainly need to be considering the machine translation experts at the Université de Montréal. I can even provide you with names. Finally, contacts must be established between the software designers and those who use machine translation. That means translators, not the public.
    Thank you.
    Thank you very much, Mrs. Brunette.
    We are now going to hear from the representatives of the second group and then we will move to questions.
    We are pleased to welcome Emmanuelle Tremblay, who is the national president of the Canadian Association of Professional Employees, and its vice-president, André Picotte.
    We are all ears, Ms. Tremblay.
    Forgive me, I hope that I am not going to have a huge coughing fit. I am not in very good shape today.
    Thank you for inviting us.
    The Canadian Association of Professional Employees essentially represents a large group of 12,000 members who work in economics, statistics and policy analysis. We also represent all the government's translators, interpreters and terminologists, who are all employed by the Translation Bureau. I am testifying today on behalf of that group, accompanied by our vice-president, who represents the terminologists, interpreters and translators.
    They are all directly affected not only by the implementation of the machine translation tool, but also by a series of changes made over the years that make us wonder whether the Bureau has lost its way a little. We are wondering whether it is still wondering whether it should be a commercial undertaking or whether it is the institution that should be the guardian of linguistic duality and compliance with the Official Languages Act in Canada.
    There is clearly a great dichotomy between these two visions and it would seem that the first approach has prevailed in recent years to the great detriment of the quality of the Bureau's services and of its members who have seen their numbers decrease drastically in favour of subcontractors and what we call phantom translation units. It is hard to imagine not paying doctors and not hiring new doctors, but that is exactly what is happening in the Translation Bureau. No one new has been hired for five years. Through this inexorable attrition, about 33% or 34% of the translators have disappeared.
    Professor Jean Delisle declared that, in some minds, translation was ”the necessary evil of Confederation”. In his view, the actions taken by the Bureau's management and the economic straightjacket strapped on it by the previous government only reinforces this prejudice.
    The Official Languages Act of 1969 gave the Translation Bureau the mandate to ensure linguistic quality throughout the machinery of government and to develop expertise and tools that would give it an international reputation. That was in 1969, and, at the time, it was a government service that was responsible for all translation activities in the federal government.
    As a result of the program review at the beginning of the 1990s, the Bureau was transformed and, in 1995, became a special operating agency. That is when the dichotomy arose between its mandate to protect Canada's linguistic duality and the need to recover its costs. It must be said in passing that the Bureau was essentially asked to be a special operating agency and to function as a private business, but inside government. It is therefore not a crown corporation, but it is also not a government service in the classic sense. To use a vulgar expression, one could say that it is a bastard organization.
    The Bureau is therefore forced to compete with the private sector, which does not have anywhere near the same operating costs. The Bureau had no choice but to adopt a mercantile approach that led it astray from its core mandate.
    By comparison, Passport Canada, another special operating agency, has a monopoly on passport production. It is therefore free to set prices for its services that allow it to offset its costs. Unlike us, it has no clients.
    The Bureau provides its language tools, such as Termium Plus, at no cost to private translators and translation companies that do not have to contribute to the cost of operating them. The Translation Bureau, for example, pays for the work of the terminologists who keep that tool up to date.
    In addition, the Translation Bureau establishes annual service agreements with the federal departments and agencies. These are not really contracts, just simple agreements. For example, if there is a one-year agreement with Public Safety and Emergency Preparedness Canada for the Translation Bureau to translate its texts, and, right in the middle of the year, the department decides to go to the private sector and tells us about it out of the blue, it means that the Bureau has just lost a client. That prevents the Bureau from planning in the medium and long terms. It also explains why, as Emmanuelle said, we have lost 400 employees since 2002 and why the 140 people who will soon be retiring will not be replaced in 2017-2018. The translators and the administrative staff with whom they work in close cooperation will continue to do quality work, but they often do it at the expense of their physical and mental health.
    Political and economic choices are undermining the ability of the Bureau's employees to produce quality work. They do it, but it is becoming harder and harder. Bureau employees are the victims of an unsustainable model. Sooner or later, the machinery will break down. Our members are devastated by this new reality, even more so because the Bureau management does not hold translators in any regard. For them, we are simply production numbers. There is also a lack of contact between senior management and the employees.
    To save money, the Bureau has also been reorganized into affinity groups. Previously, we used to work with federal departments and agencies. That therefore developed expertise among the translators working with, for example, Employment and Skills Development Canada, Public Safety Canada, or the Department of National Defence. At the moment, with affinity groups, various areas, various departments, are grouped together. So it is much more difficult to develop that expertise and translators have to do it on the job.
(1545)
    Another problem has existed for some time: departments are creating phantom TR positions. They do not call them TRs; they go by other names, such as language quality advisors. Really they are discount translators, in violation of the Treasury Board directive, which gives the Bureau the monopoly on translation in the federal government. Essentially, departments have been told to either go to the private sector or to use the Translation Bureau, but not to create independent departmental translation services.
    Another major blow to the Translation Bureau is the result of all the budget cuts imposed on various departments. For many of them, translation is the first thing to be eliminated due to budget cuts. The number of documents being translated is therefore reduced.
    Francophones are already often as a disadvantage. I come from CIDA and the Department of Foreign Affairs. I had very little opportunity to use my first language in working documents, but now, I have almost no opportunity at all. English is the only language of work in every possible draft imaginable of a working document, thereby reducing the ability of francophones to exercise their right to use the language of their choice in the workplace.
    Since 2000, the population of Canada has increased about 17%. It has gone from 30 million to 36 million. Departments therefore have increased demands. Instead of responding to that with good quality jobs and by hiring translators and interpreters who are recognized for their great skill and their great professionalism, they go increasingly to freelancers and to the private sector. I am not saying that all freelancers are not good. On the contrary, I think some freelancers are excellent, but others are not so good. Not only are we seeing a pernicious deunionization—we have already lost almost a third of our members—but the uneven quality of the freelancers also means that our members now have to correct the mistakes made outside. That ends up costing the Bureau a great deal and forces our members to salvage the institution’s reputation by doing revisions for which they are often not adequately compensated.
(1550)
    Over the years, the Bureau has acquired great expertise in scientific and technical translation. Unfortunately, that expertise is fading away as people retire. As for multilingual translation, the Bureau’s expertise is now external, and provided at discount prices.
    We now come to the machine translation tool, which Mrs. Brunette talked to us about at length.
    I am going to add something to her very eloquent comments. The broad use of a tool designed for language people is having a clear and detrimental impact on the Bureau’s image.
    Ms. Tremblay, could you move to your recommendations right away?
    Sure.
    The Bureau will have to ask itself whether its basic mandate is to achieve savings for the government or to uphold linguistic duality and the Official Languages Act.
    Mr. Picotte will talk about our first recommendation and I will talk about the second.
    The first recommendation is to make the Translation Bureau the sole government agency responsible for translation services within the federal public service. If it were responsible for managing translation on behalf of all departments, the Bureau could plan in the medium and long term, something it cannot do currently because of the permanent state of uncertainty.
    It could also start to hire employees again. This would also lead to a reduction in the administrative costs associated with awarding translation contracts and would ensure that language expertise is maintained.
    Moreover, entrusting the Bureau with the responsibility of overseeing the management of all the government’s translation would make it possible to eliminate the phantom translation units or the phantom translator positions that exist in some departments that the Bureau is aware of, thereby also achieving savings.
    The Bureau plays an important cultural role. It would be good for it to report to Canadian Heritage from now on rather than Public Services and Procurement Canada, because the latter department provides solely utilitarian services, wheras Canadian Heritage has a cultural role to play.
    Second, we suggest that the Translation Bureau be given all the financial and human resources it needs to fulfill its mandate. It is important to put an end to the policy of attrition at the Bureau, to give it the resources required to fulfill its mandate to support Canada’s linguistic duality and to stop making it bear the brunt of untenable budget cuts.
    In many of its mandate letters to new ministers, the present government has indicated that they must help to protect the Official Languages Act. This is just wishful thinking unless this desire is matched by concrete actions, such as restoring the budget so that the Bureau can fulfill its mandate.
    In addition to hiring new employees once again, the Bureau must implement a program to restore its lost expertise in the areas of technical, scientific and multilingual translation. It must also develop a succession plan that will make it possible for experienced employees to pass on their expertise by helping to train new colleagues.
    The time has come to make wise budgetary choices and to stop sacrificing the official languages on the altar of austerity.
    Thank you very much.
    Thank you.
    Because our time is flying, we are going to start the questions immediately, starting with our vice-chair John Nater.
(1555)
    Thank you, Mr. Chair.
    I would like to thank the witnesses for providing us with this useful information.

[English]

     I am a former member of CAPE. I brought my union membership card just to prove it. I do appreciate the work you do on behalf of your members.
    I want to touch on a few questions about the machine translation tool. I'll open it up to both of our groups of witnesses. Is it something where the train has already left the station in terms of machine translation? I use Google Translate all the time. It does not give perfect translation; I acknowledge that.
    Madame Brunette, the comment about needing the people side of things, the pre- and post-editing.... Is that something that the translation bureau should be undertaking, more so in conjunction with a machine translation tool, that pre- and post-editing function on top of it? The President of the Treasury Board talked about encouraging more millennials into the workforce. As a millennial, I think we do need a viable machine translation tool. Maybe you can comment on how you see that.

[Translation]

    I am a practitioner, a translator, actually. In our business, we are using computerized tools more and more, which is not a problem in itself. However, you really have to understand that those tools must be in the hands of professionals, people who know the area, not to just anyone at all, as is the case at the moment. With some texts, machine translation does not work at all. It just gives gibberish. For other texts, however, it helps translators to work much more quickly. Whatever the case, this tool must be in the hands of professionals, not of people who are not translators. Otherwise, we are going to be looking at a disaster. We are starting to see it already.

[English]

    My challenge with that, though, is that if it is only in the hands of the professionals, people are going to be using other tools in the workforce on a daily basis. People are going to be using Google Translate, if you are not opening it up to the public service as a whole. I know from my time in the public service that we used alternate means to translate on a fast basis.
    French is not my first language, as you can well tell, as I struggle with using my second language on a regular basis. I rely on things like Google Translate on a daily basis to function. I am a little concerned that we are limiting the use of machine translation only to professionals, which I think we certainly have an important need for.
    Madam Achimov, who spoke to us a few weeks ago, implied that the budget cuts have not had a major impact on her organization, and staff have left because of attrition. Would you like to address those comments?
    I certainly found her comments interesting.

[Translation]

    To be perfectly frank, I do not understand why Ms. Achimov said that. The truth is that people are under extreme stress, both administrative staff and translators. They are being pushed to the limit in order to do the work that they have to do. The organization itself is under stress. If you talk to Translation Bureau employees, you will see that they are constantly complaining about being pushed to the limit and given deadlines that are too tight. Administrative staff is being asked to perform miracles.
    Given the attrition that is still in effect, the situation is only going to get worse. Fewer and fewer people work at the Bureau. A 2014 survey of public service employees revealed that the Bureau has the worst record in terms of workplace satisfaction.
    For Ms. Achimov to say that everything is going well and that people do not have problems is almost unbelievable.

[English]

     Maybe I can specify what she meant when she said that there have been no cuts. What she probably implied was that nobody left involuntarily. There was no one who was given a letter to say that they were out the door, but she used the demographics to basically erode the core staff of the TRs, and to an extent that is seen nowhere else. I have not seen any other department where you have a 34% reduction in personnel over a four-year period. I'm sorry, that doesn't exist. You cannot stand here and say that nobody has been cut.
     No, nobody has been shown the door, but nobody who left has been replaced. That is a profound erosion.
(1600)
    I have a very quick final question. One of your recommendations is to move sole responsibility for any translation in the public service to the translation bureau. Do you have any sense of what the staffing needs to do that would be and what the costs may be to redirect all translation strictly to the bureau?
    We haven't prepared the costing but it's definitely something that we can work on, and we can establish some parameters for making the translation bureau, the service, embody what it should embody for the Canadian public service as a whole. We can work on models. We have plenty of economists to work on that.
    Right on.
    Thank you.

[Translation]

     Thank you very much.
    The floor goes to Ms. Lapointe and Mr. Lefebvre, who are going to share their time.
    Go ahead, Ms. Lapointe.
    Thank you very much for being with us today. My questions go to both of you.
    You say that 400 positions have been lost to attrition and that there will be 140 more. Four or five years ago, before the change in direction, how many employees did you have?
    The Translation Bureau had more than 1,200 TR positions, compared to about 800 now.
    Do you still have the same workload?
    The report on plans and priorities was probably one of the factors that led the committee to undertake this study. It seems to claim openly that the bargaining unit representing translators will continue to be eroded, translation will be sent to the private sector and attrition will continue with further positions to be eliminated. Of course, that has an impact because, even if you send texts for translation to the private sector, translators have to reread them. That is work that it difficult for the Bureau to bill for.
    Thank you.
    Mrs. Brunette, you used a humorous image: garbage in, garbage out. If you do not want garbage in, what would you do with this translation tool?
    That is a little technical.
    Even if what we call translation outcomes, machine-translated texts, are corrected perfectly as an outcome—that is a day I dream of—we are still going to have to put them back through the system. As I said, this is an automated statistical system. The machine learns. It will not make the same mistake if it is properly corrected. So the goal is to invest in corpus development, and, at the same time, in post-editing. Investing in corpus development means feeding the software with high-quality texts that it can compare.
    The system will search statistically for what is most often present in its memory. If everything there is bad, it is going to come out bad at the other end. For example, if there are 25 acceptable solutions and two that are not acceptable, the system statistically will search for the 25.
    Thank you.
    Thank you very much.
    Over to you, Mr. Lefebvre.
    Thank you for being here today.
    You mentioned phantom TR positions. Could you explain that in a little more detail?
    It is a chronic problem. It has been talked about since I started my career here 29 years ago.
    When an office in any given department wants a text to be translated quickly, they prefer to have someone do it on the spot. Often, the texts are small, not briefing notes or reports. So they decide to create a position. However, because of the Treasury Board directive, they cannot officially create a TR position. So they call the position “language quality advisor“ or something of the sort.
(1605)
    So the position is internal to the department.
    Exactly.
    Here is what already happened to me. Some time ago, I called a department to get some information and the person who answered said: “Good morning, Translation Bureau.”
    Really?
    It was not the Translation Bureau.
    You thought you had the wrong place.
    Mrs. Brunette, do you want to add anything?
    Here is an example for you. I have students working in departments. They have not finished their studies, but they are in revisers' positions. They have never worked as TRs but their title is reviser even if they are not able to translate.
    That is incredible.
    It gets worse. Once—this was before 2011—, the Bureau held an exam for new TR positions. Afterwards, we learned that the Bureau had contacted those who had failed the exam to offer them positions as revisers or language advisors.
    Thank you for your answers.
    We have talked about phantom translation units, but I would like to bring up another subject, the subcontractors. Are departments free to hire them? What are the rules governing their use?
    The Translation Bureau no longer has the monopoly. Ms. Achimov, uses the example of Fisheries and Oceans Canada as a shining example of the need to continue with a business approach, otherwise departments will no longer use its services. Fisheries and Oceans Canada decided on its own to establish a small group to manage translations and to use freelancers. The Translation Bureau also uses freelancers, but we deplore the practice.
    The Bureau has to translate xmillion words per year and there are always peak periods. For example, when a new government comes in, the work increases. However, there is also a certain continuity. At present, since the Bureau is chronically understaffed by about 40%, it has to continue to rely on freelancers on a systematic basis.
    Co-location is another approach that the Translation Bureau has adopted. There are fewer and fewer specialized TR positions co-located in departments. At CIDA, we had regular access to two or three TRs working with us. They were highly qualified and highly specialized. They were officially Translation Bureau employees.
    Those are not the TRs that we are talking about here, but people who are providing the same kind of service. It goes against the Treasury Board rules.
    Thank you very much, Ms. Tremblay.
    Mr. Choquette, you have the floor for six minutes.
    Thank you, Mr. Chair.
    Thank you all for being here today.
    My first question goes to Mr. Picotte.
    If we consider the Translation Bureau's mandate in terms of linguistic duality and respect for the official languages, what additional advantage does a Translation Bureau give us compared to the private sector?
    Essentially, the private sector produces words. Commercial logic says that words must be produced as economically as possible. It may be that the private sector does very good work, but we should not be under any illusion: the objective is to make money or at very least not to go bankrupt.
    In a normal and ideal situation, the Translation Bureau does more than that. It promotes language, it plays a role in the standardization of language through its terminologists.
    There is also a cultural aspect to our work that we mentioned in our brief. We stand up for linguistic duality. We care about more than producing words and paying our bills.
    Thank you.
    That is exactly why I asked Ms. Achimov when she appeared how the Translation Bureau calculates the cost of the outside services that it uses, for example the Termium Plus terminology service and post-editing. She mentioned that prices still had to go down in order to remain competitive.
    So how can you stay competitive if you do not calculate the costs of outside services?
(1610)
    I would really like to know whether she gave you an answer, because our unit has already taken part in several rounds of negotiations; we are asking the employer to give us an idea of the way in which it calculates costs when it signs an agreement with a department. We are trying to find out the basis on which costs are calculated and how it can then claim that costs are fair for Canadians. In an ideal world, we should exclude all kinds of costs which do not relate directly to the words to be translated—hoping that there is no garbage in or garbage out—in order to find out about the whole infrastructure involving terminology, post-editing, revision and so on. I am not referring to basic revision, which should be part of the per word cost.
     That is where the Treasury Board must intervene in order to protect all those aspects of the Translation Bureau's costs so that they can be excluded. When you come down to it, adding a cost to those services is a disincentive for departments to use them. The intention was clearly to reduce costs, but they went so far as to violate the fundamental rights of employees to work in the language of their choice.
    The Bureau should also comply with the Official Languages Act and protect the linguistic duality of public servants in their workplace.
    How long has it been since new employees were hired?
    Since 2011.
    No one has been hired since 2011?
    I think they hired three or four interpreters.
    No translators have been hired?
    No, not as far as we know.
    No translators have been hired since 2011?
    What consequences is that going to have on knowledge and skills transfer?
    We are losing expertise, especially in scientific and technical translation. As I said just now, we had accumulated tons of expertise over the years and the decades. The veterans had that expertise, but they have left.
    Beforehand, when young people were hired, the experienced people trained them and passed on their knowledge. Now, those people are leaving and nothing is left any more. People talk about outsourcing, using expert freelancers to translate scientific and technical texts. I doubt if you can find those by the dozen, but anyway, that is the direction we are going in. In other words, the Bureau is killing scientific and technical translation.
    In short, quality in translation is also being killed.
    I teach students and I can say that they have nowhere to go for the time being. I often tell them that they can have a life outside the Translation Bureau, that they have to look around and they will find it.
    Who will train them? Medium-sized and large translation companies don't want to hire them because it is too expensive to train them. Energy, training and quality are lost. I don't want to speak for other universities, but I can tell you what the situation is in mine. When students graduate from university, they are just ready to start learning.
    Thank you very much, Mrs. Brunette. I would like to ask you a very short question.
    Thank you very much. We could come back to it.
    It is Mr. Vandal's turn.
    My question is about the quality.
    Mrs. Brunette, have you noticed a decrease in quality because of the attrition over the past few years?
    I cannot answer that question, because we are talking about the Bureau. Since I don't have a lot of contact with the Bureau, I will let my colleagues talk about that.
    You talked about a decrease in the number of employees because of attrition to the tune of 400 and 140.
    That does not affect the quality, Mr. Vandal.
    Has the quality decreased?
    Actually, the quality has stayed the same. As I said, people are going to great lengths to maintain the quality. However, they are increasingly under stress because of the employees retiring. At some point, something's got to give and we will no longer be able to maintain the quality. Right now, we are delivering high-quality work, but there is a limit to what we can tolerate. We are approaching the breaking point. That's the problem.
    So my answer is that quality is in fact being maintained and we are doing a good job, but this will not always be the case.
(1615)
    That is also affecting high-quality employees, students graduating from university, either from University of Ottawa or UQO. Their opportunities are relatively limited. In addition, this means that the jobs available to them are precarious. Our proposal is to strengthen the Translation Bureau so that it can once again become the natural place for top graduates who may look forward to joining. We might then see an increase in quality. Right now, the quality is there to the detriment of our members' health.
    I understand.
    Mrs. Brunette, what are your thoughts on that?
    I am prepared to defend machine translation, but I have to say one thing in terms of quality: even with a solid corpus and proper post-editing, machine translation in itself does not guarantee quality. It ensures speed and understanding, but it is not intended for wide dissemination. A political speech should never be translated with a machine translation tool. That would make no sense. The fact remains that you can obtain minimum quality for a client who wants minimum quality. For instance, machine translation can be used if the document stays in-house or if it has to be translated after hours or during the Easter break.
    Thank you, Mrs. Brunette.
    Thank you very much, Mr. Vandal. I will allot three minutes of your time to Mr. Arseneault.
    Very well.
    This question is for Mr. Picotte or Ms. Tremblay.
    You mentioned the Treasury Board rules. Could you tell us more? What are the rules?
    I can only repeat what I said earlier. The Bureau has been given the monopoly over translation for the departments that don't choose to do business with the private sector. There are therefore two choices: federal departments and agencies can either do business entirely with the private sector or the Translation Bureau, but they cannot set up an internal translation service. That's the rule. But as I said, they do not follow the rule. The Treasury Board often sends a notice to the departments to remind them of their duty, but the notice ends up in the trash and they go back to their old ways.
    I have nothing to add to what Mr. Picotte said.
    When you talk about the obligation of linguistic duality, is that what you are referring to?
    Yes, but at different levels. Clearly, to ensure that the services are available to Canadians in the language of their choice, the translated documents must meet the required standard of quality. Historically, the Translation Bureau has vouched for that quality. It is not certain that this is the case for all departments, including those that have decided to go to the private sector exclusively or those using the so-called phantom translation units.
    That said, I am also talking about linguistic duality and rights of public servants. I am specifically thinking of the decreased use of bilingualism at work. I have experienced it first hand; I used to work for CIDA, which was fairly francophone. However, the agency was folded into the Department of Foreign Affairs, which was predominantly anglophone. Forget about working in French. That was over.
    Actually, the more pressure is placed on departments to reduce costs of things that fall under their responsibilities, the more they take those kinds of shortcuts, obviously. It is no longer realistic for francophones in the public service to speak French or use French in their communications, especially in their written communications.
    In fact, the minister for whom I worked at CIDA was francophone, and we used to write the briefing notes in English for him, because his staff was anglophone.
    Thank you very much.
    We have two more speakers. First, Mr. Fergus for three minutes, and then Mr. Généreux for three minutes.
(1620)
    Thank you very much, Mr. Chair.
    I have two questions.
    For the first question, I would like to go back to a comment made by Ms. Tremblay.
    We have seen that the entire philosophy of the public service has changed. Instead of considering bilingualism as an obligation unique to our country, a decision has been made to consider it a burden and to try to save money in the area.
    Could you go back to your comment about saving money in a field where money should not be saved?
    Yes. That goes to the basics, meaning compliance with the Official Languages Act.
    Has the commissioner already appeared before you? If not, is he scheduled to appear? I am sure that he will speak on the issue more eloquently than I can today.
    Clearly, the Official Languages Act has been sacrificed on the altar of austerity. That does not only apply to that act, but also to the many issues facing minority communities across the country. The focus is on the bottom line only; that's the only thing that matters. I hope that this perspective will change with the new government.
    Since I have only three minutes, I will ask my questions very quickly.
    Do you think the issue could be resolved if we hired 30% more translators?
    It will take more than that. If we want to reposition the Bureau as the go-to place for the public service as a whole, we will need more than just reverting to the status quo ante. We will have to manage growth. As I said, our economists will be able to carry out an analysis to see what that might mean in practical terms.
    It is clear to me that this is the only way to be able to actually enforce the Official Languages Act.
    Is it essential to have a Translation Bureau where all the translators are centralized in one specific place? Would it not be better to assign the translators to the various departments so that they can practise their expertise in their field?
    I will let my colleague André tell you more about that.
    Personally, I would say that they are not mutually exclusive. The role of one place can be to ensure quality control. The centralization of translation in an organization such as the Translation Bureau, which might fall under Canadian Heritage or another department, is important. However, we can also co-locate people. We know that many translators are teleworkers. The Bureau is also saving a lot of money by having people telework. Translation work can be done over long distances. So we don't need a big Translation Bureau in your riding, Mr. Fergus.
    Some hon. members: Oh, oh!
    We can also continue the way we are. The fact remains that the objective of quality and respect for linguistic duality should be the top priority.
    Colocalization has already given results. The approach has been to save money by nickel and diming, and going backwards.
    Thank you, Mr. Fergus.
    Mr. Généreux, you have three minutes.
    My thanks also to the witnesses for being here.
    Mr. Picotte, you referred to your special operating agency in words that have slightly taken me aback. You referred to it as a “bastard organization”. That is quite a loaded expression in various contexts.
    You said that the number of people working for the Translation Bureau has gone down from 1,200 to 800. You just said that you are still providing a high-quality service, but that the organization you are working for is a bastard organization. I will give you the opportunity to elaborate on that. I know that the word clearly does not refer to the people, but it shows that the organization itself is repudiated. It is as if it was no longer a part of the government or the private sector.
    Could you elaborate on that expression? I felt that it was very emotionally charged for you, but it is also very pointed.
    By using the word “bastard”, I was of course talking about the organization, not the people, the managers and senior executives.
    We have one foot in the public service using the government's way of thinking, and another in the private sector using the private sector's way of thinking. The Bureau must fulfill government duties and reach its break-even point.
    Something is not working, and that is why I use the word “bastard”. The Bureau is either a crown corporation and operates exactly as a private business, or it is a government service, which is the ideal scenario for us.
    The Bureau has one foot in the government and the other in the private sector: that is what I call a bastard organization.
(1625)
    I have many questions, but it would be useful to find out the potentially appropriate amount of money according to you and your economists. It would be useful for the committee to review that some day.
    Would you be able to forward the cost-related information to us? Could you send it to the clerk so that he can circulate it to the members of the committee?
    I will look into what we can do in terms of projections as soon as possible and we will be in touch with the clerk.
    That would be very good.
    Mr. Généreux, the floor is yours.
    Mrs. Brunette, I really liked your explanation of the machine translation software. Inevitably, technology is part of our daily lives.
    The number of employees has gone down from 1,200 to 800, and the work still gets done. Undoubtedly, those positions were redistributed somewhere else.
    Based on your example of “garbage in, garbage out”, my understanding is that the software carries out a mathematical analysis.
    Do you agree with me that this is an evolving process?
    Furthermore, I suspect that a piece of software like that is inevitably used by translators and it must increase the quantity of information that can be translated. Do you agree that this is part of the evolution of things, meaning that we might need fewer people because very fast tools are gradually being implemented?
    Yes, I think it is evolving. The more you feed the corpus, the better the results are. In particular, texts that would have otherwise never been translated will benefit from this tool. That said, I cannot see the day when it will replace human translation.
    Has anyone anywhere talked about the ultimate objective of having the machine replace human translators some day in the Government of Canada? Has someone somewhere said that?
    No.
    Has anyone said that they wanted the machine to replace humans?
    No one may have said so, but that is what Ms. Achimov's behaviour suggests.
    Thank you very much.
    Thank you very much, Mr. Généreux.
    Thank you very much, Ms. Tremblay and Mr. Picotte. Thank you very much to you as well, Mrs. Brunette.
    We will suspend the sitting for a few minutes to allow the second panel to settle in.
    Thank you for your presentations once again.
(1625)

(1630)
    Please have a seat.
    Pursuant to Standing Order 108(2), we will resume our study on the Translation Bureau.
    Joining us are representatives from the Language Technologies Research Centre: Alan Bernardi, president general director, and Donald Barabé, chairman of the board of directors.
    Welcome, gentlemen. You have about 10 minutes for your presentation and we will then move to questions and comments from the members of the committee.
    You have a presentation in front of you. I will go through it to review the issue of machine translation and to try to answer the following question: why make an imperfect tool available to all federal public servants? I will also talk about the Bureau’s mandate and its financing model.
    Mr. Bernardi and I agreed that I would make the presentation because I spent my entire career—35 years—with the Translation Bureau. I retired from the Translation Bureau just three years ago. My entire career was spent there, from translator to vice-president. I was one of the originators of the machine translation project we’re discussing here. So we thought it might be interesting if I presented this perspective.
    Let’s start with machine translation. I will tell you about the global context. Every day, 400 million pages are translated by Google. That means that Google translates the equivalent of the Translation Bureau’s annual output every five seconds. The Bureau translates 1.5 million pages a year; Google translates 400 million pages a day.
    The European Union already offers European citizens a machine translation system, including for translating diplomas. Even the German authorities are using machine translation to integrate Syrian refugees, for example, because the Syrians don’t speak German, and the Germans don’t speak Syrian. So machine translation is used for that. Whether we like it or not, machine translation is becoming the lingua franca of the world, the most widespread language.
    At the federal level, when Ms. Achimov appeared before you on March 7, she told you that federal public servants had used Google Translate over a million times last year. One of the issues is that by using Google Translate, Government of Canada texts are loaded on the servers of a company subject to the Patriot Act. That isn’t the only issue, but it's one that was considered in the decisions.
    You should also know that machine translation has been used in the public service since the early 1970s. All weather reports have been translated by a machine since the early 1970s. Why? Because they are repetitive and the formulations are repetitive. It’s basic. The implementation of social media, where everything moves very quickly, creates a demand for instantaneous translation, like it or not.
(1635)
    I would like to repeat an expression that was just used.

[English]

    The train has indeed left the station.

[Translation]

    In other words, machine translation is used widely around that world and, like it or not, it is used by federal public servants.
    As Ms. Brunette said so well, the fact remains that this is something that produces a document that is inferior in quality to the original. If the original is at a certain level, the version produced by machine translation will be much lower. As a result, this tool doesn’t produce a quality text. It helps save time and have an idea of what the text is about. That’s why it’s important to control it.
    Four conditions were set to control it. The first condition was not to load any classified texts into the tool. The second condition was that it be for personal use and information. The third condition was that professional revision would be required if it had to be distributed, that is, by a professional translator, not a bilingual clerk. Ms. Brunette just gave a good example of the fourth condition. So that the system isn’t contaminated and an erroneous translation put in the system, these translations must be revised at sufficient intervals and fairly regularly.
    That's machine translation.
    I will now take off my hat as representative of the Language Technologies Research Centre and move on to the Translation Bureau.
    There are three important dates in the Bureau’s history: 1934, 1993 and 1995.
    What happened in 1934? The Translation Bureau was created through the enactment of the Translation Bureau Act by the Parliament of Canada. The act states the following.

[English]

    The bureau shall “act for all departments, boards, agencies and commissions” and “for both Houses of Parliament in all matters relating to the making and revising of translations”.
    That was subsection 4(1). Subsection 4(2) says that all departments, boards, agencies and commissions “shall collaborate with the Bureau in carrying into effect this Act and the regulations.”

[Translation]

    Therefore, when the Bureau was created, it was an agency that all departments were required to use. Under the legislation and regulations, it was to do all translations requested by the departments and by Parliament.
    From 1934 to 1995, it was financed through parliamentary appropriations and, as a result, free for the departments.
    In 1993, the Bureau was transferred from the Secretary of State, now known as Canadian Heritage, to the Department of Supply and Services, now Public Services and Procurement Canada. The Bureau’s mandate and financing model remained unchanged, but while it was a crucial component of the country’s social fabric, translation became an administrative service. It’s common knowledge that the general aim is to try to reduce the cost of administrative services as much as possible.
    In 1995, it was decided that the Bureau would be made a special operating agency, and the departments would no longer be required to use its services. From then on, the departments could choose to do business with the Translation Bureau or the private sector, but could not have their own translation service. However, the Bureau had to start billing all its costs.
    I’d like to come back to linguistic duality. You saw earlier that the Bureau went from Canadian Heritage to Public Services and Supply Canada. The presentation contains an article that made the headlines of La Presse in 2007. The article, titled “La survie du Canada repose sur le bilinguisme”, appeared following the publication of a book called Sorry, I Don’t Speak French by Graham Fraser, who was not yet the Commissioner of Official Languages.
    The map on page 10 details linguistic duality in Canada. Using statistics from the last census, it shows that 17.5% of Canada’s population is bilingual and that translation is an essential bridge between cultures and communities. In fact, translation plays a fundamental role in the sense that it guarantees each Canadian the constitutional right to be unilingual. It also guarantees that public servants have the right to work tools and documents in their official language.
    Page 11 states that the Bureau became a special operating agency, that the departments are no longer required to use its services and that it must recover all its costs. Page 12 outlines that this has led to some difficulties. The first is that the Bureau must continue to meet the demand, but that the departments are not required to feed it. Even if they decide to use it, they fairly often change their mind during the year.
    Full cost recovery means that the Bureau must bill the departments for costs for which they are not appropriated. For example, rents for the departments and insurance for employees are financed centrally. Since the Bureau must assume these costs, it must recuperate them through its clients. These costs are non-negotiable.
    At the same time, the rules that are imposed on the Bureau are such that if a department decides to use the services of the private sector and put out a call for tenders, the Bureau does not have the right to submit in response to these tenders.
    The departments even have more procurement authorities in translation over the Translation Bureau, which is the designated translation authority. The Bureau’s procurement authorities for translation are $25,000, while those of the departments are $2 million. Unfortunately, this led to unintended consequences, including delayed and cancelled translations, and we saw the re-emergence of something we no longer see: on-demand translation. Some departments indicated that they would no longer have documents translated, unless translation was requested. It isn’t widespread, but it still happens fairly frequently.
    Given that the departments don’t all have the money to pay for translation, some of them have decided to create an internal translation service, thinking that it would cost them less. Independent studies have shown that it sometimes cost three times more. That led to consequences for the industry. It’s important to note that the Government of Canada is the largest translation client in Canada and one of the largest in the world. The way it provides work has an impact on the industry’s development. By giving the departments the authority to conclude contracts, federal buying power in translation is fragmented, which contributed to the fragmentation and vulnerability.
    In conclusion, machine translation is an imperfect but useful tool. However, it needs to be used properly. The Translation Bureau is a key component of the infrastructure that Canada has established to operate as a bilingual country, but unfortunately it is underused.
    I have a few recommendations.
    In terms of machine translation, it is important to ensure that the four conditions are applied for a successful implementation of the machine translation software. Departments need to be educated on the benefits and limitations of machine translation and on issues relating to the Official Languages Act.
    As for the Translation Bureau, it might be necessary to review the Bureau's location within the federal government. Certainly, the difficulties and unintended consequences need to be corrected and the Bureau’s expertise needs to be used to eliminate duplication of costs and that texts that are written by public servants and that need to be translated is indeed translated by public servants.
    The private sector doesn’t want to translate everything. I can assure that it does not want to translate unprofitable texts. So the Translation Bureau must be equipped to translate them. The Bureau’s expertise should also be used to consolidate the federal government’s buying power in translation to promote the development of Canada’s translation industry.
(1645)
    Thank you very much for your presentation, Mr. Barabé.
    We will now move on to Mr. Généreux and Ms. Boucher, who must share their allotted time.
    Thank you, Mr. Chair.
    I would like to thank both witnesses for being here today.
    Could you explain what the Language Technologies Research Centre is? To use a popular expression, what do you call that when it's at home?
    I'll make a brief comment and then pass things over to my colleague.
    It's a centre that was created at the urging of several federal agencies, including the Translation Bureau.
    Mr. Bernardi will give you the facts.
    Three organizations created the CRTL: the Translation Bureau, the NRCC and the Université du Québec en Outaouais. We are set up just across the river at UQO. Our role is to do research and develop technology in the language technologies sector.
    In 2014, we expanded our scope. For lack of funding, we are now focusing on business intelligence, on large archives of data and on entrepreneurship in the Outaouais region.
    Did you work directly or indirectly on creating the software or on its supervision? Were you connected to this tool in one way or another? Did you analyze it? I imagine that you are doing an ongoing analysis of this tool, which was to be introduced on April 1.
    No. We work in the same building and we have used this software for three years, but we had no contact with the version that was deployed for the government.
    I should note that there are two parts, as Ms. Brunette explained. There's the software and there's the corpus or what the software learned. This software translates the language of the public service for the public service. It was trained with official texts from the public service that had already been translated in both languages.
    What happens when it's removed from its context? When it is asked to translate emails and it hasn't learned to translate emails, the tool doesn't perform nearly as well because the language is different.
    As for meteorology and Environment Canada, translation of weather reports comes from the same software that learned the specific language of weather. So it performs very well because the texts vary very little.
(1650)
    Thank you.
    Mrs. Boucher, you have three minutes.
    There are a lot of new technologies, but you explained to us a little earlier that this tool is for us. It isn't a tool for writing texts that we can send. It's for the personal use of parliamentarians and public servants.
    That's right.
    To do a translation, the person who sees what is produced by the machine has the know-how required to ensure that the tool performed properly and to make corrections if necessary. That's what Ms. Brunette called post-editing. If used the other way, you have to have the skills to do so.
    It's a little like Google Translate.
    Exactly, and perhaps that is the issue. Public servants are currently using Google Translate a million times a year. Do we continue to use public tools, with the risks that may involve, or do we create an internal tool that does not have security risks and that, with training, can perform much better because it will know the terminology used in the federal public service?
    How can we use this tool when we want to apply the Official Languages Act? Does this tool perform well enough to protect both official languages? The way it works, is it pretty much anything?
    The tool has its limitations, which depend on the person using it. If used properly, it protects the official languages to some degree. It helps to translate, to understand someone who is speaking to us in another language. We can use it for the purposes of understanding, but it won't strengthen the legislation; it's only a tool.
    Mr. Lefebvre, you have the floor for three minutes.
    Thank you, Mr. Chair.
    Ultimately, this translation tool could be used for any internal communication. But external communications should go through the Translation Bureau.
    More specifically, I would say that this tool should not be used for any communications. We know that where there is a communication, there is distribution.
    This is a tool that helps someone who does not know the other language to have an idea of what a document contains.
    So it's a basic tool.
    It's basic and that's the issue of official languages.
    The fear is that people are using this tool to communicate with colleagues internally.
    Exactly.
    One of your recommendations was to educate departments on the benefits and limitations of the translation tool. How do you suggest that be done? This is what would happen in a perfect world.
    It's important to explain that machine translation presents advantages in terms of speed. The tool helps to give an idea of a document's content, but it is important to stress that it cannot be used in any communications because it does not respect the principle of official language equality. That is key.
    That's where we hit a snag.
    That's where the issue lies.
    Do you think that there will be more translation with this tool, or will it stay the same?
    An enormous amount of translation and communications are done in the world.
    Will the machine translation tool help to make gains?
    Clearly, it contributes to gains in productivity.
    Let's take the case of weather reports. I don't know how many people are on this team currently, but there were seven translators when I was there. If humans had to do this work without using machine translation, it would have taken about 70 translators. There are enormous savings, but it is an extremely precise area.
    And specific.
    Absolutely. It's mainly because meteorologists are heard describing the day's gloomy weather only one way, and the machine regurgitates that description.
(1655)
    Thank you very much.
    Mr. Fergus, you have three minutes.
    I am relieved to hear you say that machine translation should only be used as a tool for understanding, not for translation. Indeed, it would lead to erroneous communications that would not be revised by a professional. I fully agree with you.
    My question has to do with one of your slides that says that the "Bureau has to recover its full costs, except for services to Parliament".
    You didn't mention in your recommendations whether this model should be reviewed. Should it?
    We recommend correcting the difficulties and unintended consequences.
    By coincidence, half of my career was spent in the Translation Bureau when its services were mandatory and free, and the other half when they were optional and not free. You might expect this not to be the case but, strangely, I preferred the second half, because it required being as efficient as possible. However, it has probably gone too far.
    The best parallel I can make is with Justice Canada's legal services. The departments have legal issues and turn to lawyers from the Department of Justice to help them. Each department has a legal service that reports to Justice Canada, not the department it is in. Only a portion of the costs are billed by Justice Canada to the departments.
    When departments are billed the full costs, including for items for which the departments have no budget, such as rent and insurance, this causes undue pressure on them and results in undesirable actions, such as the creation of full ghost services.
    Some independent studies have been done to see what these services cost. It turned out that it may cost up to three times more than the Translation Bureau. We have probably gone a little too far, and these aberrations need to be corrected.
    Thank you, Mr. Fergus.
    Mr. Choquette, you have the floor.
    Thank you, Mr. Chair.
    I would like to thank the witnesses for being here today.
    I am really pleased that Mauril Bélanger and I requested this study. Indeed, the deeper we go into the subject, the more we realize the importance of the Translation Bureau in connection with linguistic duality and respect for official languages.
    When Ms. Achimov appeared before this committee, she stated, as she did in the media, that the translation tool could be used to draft short emails and non-official messages.
    Why do you think she said that?
    Perhaps because of a lack of understanding of the tool and how exactly it can be used.
(1700)
    I asked her if she was aware that experts had stressed, as you did as well, that it was extremely important never to use this tool for drafting purposes. I asked her if she did not find that this situation had led to confusion, a mess, and whether she thought it would be preferable to postpone introducing this tool. She responded that it wasn't necessary.
    Like the associations of translators of New Brunswick and Ontario, I recently sent a letter to Minister Judy Foote to ask that this implementation be postponed. Fortunately, it is for now.
    You might have learned in the media why it was postponed. Comments made by the Department of Public Services and Procurement included the following:
    Changes aimed at "increasing response speed" and "clarifying information for the user" will have to be made "in the coming weeks", before the machine translation tool is rolled out across the federal public service.
    Based on your knowledge of the matter, would you say that this response is satisfactory?
    It's difficult for us to comment on this. However, on the second point you raised, about the need for clarification, I think the fact that we are all here discussing this issue shows that clarification is needed.
    Great.
    You covered the Translation Bureau and some recommendations you made. You also spoke about what the private sector should not or would not translate.
    Could you expand on that?
    Earlier, union representatives spoke at length about using outside contractors. The Translation Bureau Act clearly states that the decision to use contractors lies with the CEO of the Translation Bureau. Simply from a management perspective, it is an absolute necessity because demand fluctuates. Staffing positions to have everything done internally when there are peak periods would be very expensive for taxpayers and would not be efficient. So it is important to use the private sector.
    What was the next part of your question?
    You spoke about what the private sector should not or would not translate.
    The issue is determining what we send to the private sector. We included in our recommendations the fact that the private sector should not be given texts that should not be entrusted to it, such as texts that the government uses to make decisions. That is also the case for all texts with a security classification. In fact, even though a supplier has security clearance, it may end up in a conflict of interest.
    Let me give you an example. During my entire career as a public servant, I had "top secret" clearance. Every six months, I had to complete a declaration of interests to show my employer, the Government of Canada, that I was not in a conflict of interest. This is part of the Values and Ethics Code for the Public Sector, which is included in the employment conditions of public servants. However, this may not apply to a supplier. Therefore, texts with a classification should probably not be entrusted to the private sector.
    The other point concerns what the private sector would not do. Since I am very active in the association that represents the language industry in Canada, I can tell you that the private sector does not want to translate texts that are not profitable. That's normal. These people want to ensure their profitability. Texts that are very short, scientific or highly specialized are examples of non-profitable texts. It isn't profitable for these people to put together the resources for them.
    An analysis may be done by considering these factors.
    Thank you very much.
    Ms. Lapointe, you have the floor.
    Thank you very much for being here today.
    I will come back to your recommendations.
    You said that the translation tool should only be used for comprehension purposes. You said you want to educate the departments about that.
    In an ideal world, how should the departments be educated?
(1705)
    There needs to be a lot of communication. I wasn't there when things began to be put in place. For each translated text, there could be a warning that it is for the purposes of understanding, not communication. A whole communication exercise is needed.
    You also mentioned the risk that Google Translate poses for security. You spoke earlier about security classifications. Could you speak about the problem that Google Translate presents so I can ensure that we fully understand what it is.
    Everything put in Google becomes its property. It's in your contract with Google. By using the service, you sign a contract. That's the way it is for all texts put in Google, either personally or as a public servant. Every time a public servant puts a government text in Google Translate, the text becomes Google's property, whether it has a security classification or not.
    Should the warnings to the departments contain anything specific? Should we tell them not to use it to translate government or department texts and to use only the tool provided to them?
    It wouldn't hurt.
    A warning like that would be as important as the warning you mentioned.
    Absolutely.
    It should be specified with what the translation system produces that this is what is preferred for one reason or another.
    Absolutely.
    Thank you.
    Mr. Chair, I'm going to share my time with my colleague.
    You have six minutes now and six minutes later.
    Right.
    I would like to add something.
    There's one thing that might be explored. The tool could be made available to Canadians, too, not just public servants. That's what the countries in the European Union do. Their machine translation tool was developed for the government, but now everyone can use it. It might keep Canada's trade secrets from being revealed.
    I find that very interesting. When the tool is operated by computer software open to everyone, advertising revenue might be possible.
    It could happen.
    So because people would use the tool, advertising revenue might be generated.
    Potentially, yes.
    Okay.
    It's interesting because we would go for more. And it would stay in Canada. I imagine the servers are in Canada?
    For this system, of course.
    Not for Google?
    No, absolutely not.
    We just want to make sure that the people in the back of the room are taking note of that.
    Even if Google decided to put its servers in Canada, they would still be its property. They are included in what I said.
    Okay, thank you.
    I will now turn things over to my colleague, Mr. Arseneault.
    I have a practical question about this small and wonderful translation engine. I'm sorry, but I am a doubting Thomas, and even he would be an amateur compared to me.
    I am trying to draw an analogy with the French checker Antidote. I suppose French may be one of the most difficult western languages to learn. It may not be alone, but with its agreements, genders and whatnot, it is really difficult to learn.
    You are familiar with this small and wonderful tool. With the automatic creation of a database of words and phrases, is translation from French to English as quick or effective as translation from English to French? I assume that most of the databases of words and phrases are created from English at the outset, as the majority of Canadian users are anglophones.
    I want to preface my answer by saying that there is a huge difference between Antidote and this kind of a tool. Antidote makes very little use of statistics; it uses very specific rules. I am quite familiar with the Antidote people. It is easier to make a French checker than an English one. The Antidote people just released an English checker. French is more structured as a language than English. So things are easier for the tool in that sense.
    Automated translation software uses statistical models and does not differentiate between two languages. It is loaded with well-translated texts, and that's the base. Using those texts, the software creates a language model. The process is exactly the same regardless of the language.
(1710)
    I understand the practical side, the basics. If you enter peanuts, it gives you arachides, and if you enter arachides, it gives you peanuts.
    Voices: Oh, oh!
    Exactly.
    I would like to come back to Antidote. It may be easier for the machine with Antidote, which I use regularly. I am told that it is one of the most sophisticated softwares on the planet.
    That's true, and it was developed in Montreal.
    My hat goes off to Montreal, but Antidote is not perfect—far from it. However, that software is the global expert on French grammatical correction.
    I am a doubting Thomas. It is difficult for me to imagine translating from English to French and correcting the grammar on top of that. It seems to me that this is an almost impossible mission.
    Have you experienced it, have you seen it?
    I have experienced the same things as you. As this is a statistical instrument, variance is important. For texts with little variation, such as weather reports....
    Weather reports are too easy.
    That's true, but there is very little variance in weather reports. So the translation can be of high quality. The more variance increases, the more fields are involved, the more difficult it becomes to maintain the same level of quality.
    What do the translation companies that use automatic translation do? The best practices involve limiting the translation tool's learning to the specific field in which the translator is working and to edit texts afterwards.
    I will use the example of manuals for military members. Translators would learn from previously translated manuals. Afterwards, when they started translating the manual and revising everything using automatic translation, that information would be put back into the tool every week. So the translation tool would improve its translation of that manual, since the beginning was already understood. That is one example of the best practices.
    Thank you, Mr. Arseneault.
    Mr. Vandal, you have four minutes.
    In what year did Canada start using translation tools?
    It first started using them in the 1970s to translate weather reports. The automated translation tool Portage started being used very recently. I would say three or four years ago.
    Three or four years ago?
    Yes.
    Is yours a non-profit organization? Were you consulted before this tool was introduced?
    No, we were not consulted before the tool was deployed.
    Is it a translation tool that may encourage the use of both official languages within the Government of Canada?
    It's a tool that may help unilingual public servants get an idea of a text's meaning in the other official language.
    So it may encourage....
    It may encourage the use of both official languages as long as it is limited to understanding and not communication, and as long as it remains limited to personal use. If it went beyond personal use, it would raise the issue of compliance with the Official Languages Act.
    That's still an important role.
    In any case, it's already widely used.
    Your colleague who was here earlier said that he was already using Google Translate when he was a public servant.
    We are not talking about Google Translate. It's not the same thing.
    Or is it the same thing?
    Yes, it's the same thing, but this software is Canadian.
    This software has been loaded with federal government documents.
    So it could be very important for bilingualism across Canada among people outside the public service.
    Yes, as long as everyone understands their limits.
    That's right.
(1715)
    Go ahead, Mr. Arseneault.
    Ultimately, can we say that a language would have to be fairly robotic for a translation to be accurate?
    That is a prerequisite.
    So there shouldn't be too many metaphors or sarcasm.
    You raise an interesting point. One of the things we have recently started looking into is the life cycle of information. Let's use the example of document production. If documents are written in a much simpler manner, such as in simplified English, those documents will be much easier to translate, but also much easier to read for the vast majority of Canadians with literacy problems. So not only is the cost of the document's translation reduced, but the document is also made more accessible in both official languages to a large proportion of Canadians with reading difficulties.
    That may be something to consider. It is not just a matter of official languages, but also of issues related to access to the documents and their transparency, and that is something that can be achieved by simplifying some texts. This would not apply to legal documents, but other documents intended for a general audience could be simplified from the outset.
    Thank you very much.
    Mr. Généreux, you have four minutes.
    Thank you. Speaking time is increasing, so that's good. Thank you, Mr. Chair.
    On page 6 of your document, we can see three important dates in the history of the bureau, including 1995, when costs for the Translation Bureau were considered. Are development costs part of that? There is a comparison between private firms, or translation agencies or phantom services, and the archaic service—I dare not use the term that was used earlier.
    Me neither.
    Does that include the cost of developing software?
    Yes.
    So it is unavoidably more expensive for the Bureau than for private companies.
    Yes. There are various reasons behind that. You have to understand what the costs for the bureau are when it develops these kind of tools.
    The question was asked earlier about Termium. The bureau's costs for developing and updating Termium, for example, are not part of the costs for which departments are billed, as parliamentary votes cover that.
    There are development cost attached to a translation tool, so that is part of the costs that will passed on to departments. At the same time, those costs are not shared by the private sector, but the sector does benefit from those tools.
    So the private sector does benefit from them?
    Absolutely.
    It benefits as much as if the software was Canada-wide, as much as it can benefit from Google.
    I see.
    It is basically paid by Canadians, through taxes.
    Yes.
    For example, when it was decided, with Parliament's approval, to make Termium and the entire language portal available to all Canadians, a cost was attached to the decision. In fact, anything posted on the web is accessible not only to Canadians, but to the whole world. So there is a cost to making the servers capable of absorbing the future demand.
    So if the Translation Bureau does the same for automatic translation, there will clearly be a cost attached.
    Do I have any time left, Mr. Chair? That's very nice.
    I'm also a doubting Thomas, and the accent used on the islands will probably never come through in translation.
    I am not from the islands.
    Voices: Oh, oh!
    You mentioned benefits on page 15. Can you please explain what exactly are the biggest advantages in your opinion?
    I would say the biggest advantages are quickness and understandability.
    I will give you an example that has nothing to do with the government. I have food allergies. If I travel to China, I will be very glad to be able to take a photograph of a Chinese menu to get an idea of what it contains, as I do not understand Chinese.
    There will be an application that will make it possible.
    There already is one, Google Translate.
    It's quick. It's a matter of 400 million pages per day, while the bureau produces 1.5 million pages per year. I can assure you that the quality of the 400 million pages cannot hold a candle to the quality of the bureau texts.
    That gives you a bit of an idea. It's quick, and we get a general sense of what the document is talking about.
(1720)
    What you just said is extremely important. The software's deployment is being delayed for various valid reasons. However, we are in 2016, so it has to get done eventually.
    You just said that everyone in the European Union has access to the software the EU invented and implemented. I assume that EU governments did that.
    We are talking about 22 languages in that case.
    And not two.
    I have covered everything I wanted to.
    Thank you, Mr. Généreux.
    Mr. Choquette, go ahead for three minutes.
    Regarding the implementation of the translation tool, I asked Ms. Achimov whether any risk analyses had been carried out on the respect of official languages, and she told me that there was a pilot project. Unless I'm mistaken, you participated in that project.
    No.
    You have used the tool....
    We have not touched the current version. All that we know comes from newspaper articles.
    I put the question to Ms. Achimov, and she told me about a pilot project. I thought you had participated in it. I asked her whether a report was published, and she said no. I also asked the Commissioner of Official Languages, with whom we will be meeting in a week or two, whether he has assessed risk. He said that his office did not have the expertise needed to do so.
    Who do you think would have the expertise to assess the risks related to implementing this software in the government machine? As you explained, it would be a matter of determining whether instructions are provided for the tool to be used appropriately.
    It is probably possible to see who within the department is using the tool, to see whether it is being used for emails, and so on. In terms of the technology, that is possible, but the issue should be analyzed more thoroughly before I can give you an answer.
    My question has to do with the Official Languages Act. As we know, the introduction of this tool was bungled a bit, particularly by Ms. Achimov.
    As time is running short, I will wrap up by asking you whether the Translation Bureau's management should include some translators who know how to implement such a tool.
    Absolutely.
    Thank you.
    Thank you, Mr. Choquette.
    Are there any further comments or questions? No?
    Thank you for your excellent presentation.
    We will continue our work next Wednesday afternoon. We will then hear from the representatives of the University of Ottawa, the Fédération des communautés francophones et acadienne, and the Language Industry Association.
    Next Monday, we will welcome the representatives of the Université de Moncton.
    Would you like us to hear from the Translation Bureau representatives again?
    Mr. Choquette, go ahead.
    Thank you, Mr. Chair.
    I don't want us to hear from them again. We have already heard from them. They have spent about two hours with us. In my opinion, they said what they had to say, and I think we listened to them.
    Our time is limited, since other studies have to get off the ground. I also think that Mr. Vandal has an interesting study to propose.
    For those reasons, I would prefer for us to focus our efforts on the upcoming witnesses.
    I also propose that we submit a report to the House when we complete the study. I am proposing that right now.
    I suggest that, during the second hour, after we hear from the Université de Moncton representatives, we discuss the content of the report we will submit.
    Thank you very much, Mr. Chair.
(1725)
    Is that okay with you? It is? Very good.
    Thank you for you excellent presentation and your contribution to this study.
    Thank you for inviting me.
    The meeting is adjourned.
Publication Explorer
Publication Explorer
ParlVU