Sam Lingard and Jaimin Patel.

The Chair: Welcome to this public session of the Public Services Committee. Thank you for coming to speak to us today. This is our inquiry into interpreting and translation services. We are drawing to the end of our inquiry now, as you probably know, so we are looking forward to this session to clarify some of the points that we have heard and to better understand your contribution to the process. The session is being recorded, and you will get a copy of the transcript so that you can make sure that it says what you intended to say. I welcome our two witnesses today and ask them to introduce themselves.

Sam Lingard: Hello, and thank you for inviting us. I am a director at The Language Shop Ltd. We are responsible for the quality assurance contract of language services for the Ministry of Justice.

Jaimin Patel: I am the managing director of The Language Shop Ltd. For a bit of context, we were part of the London Borough of Newham and we spun out as a social enterprise in 2016. Forty-nine per cent of the shares continue to be owned by the London Borough of Newham, and 51% are owned by an employee trust.

The Chair: That is exceptionally helpful. One reason why we were a minute or two late starting was because we were trying to work that out. Lord Laming has a question about it.

Q94 Lord Laming: You must excuse my ignorance about how these things work. Mr Lingard’s biography says that The Language Shop is now incorporated as a company. However, Mr Patel’s says that it is an employee mutual and joint venture with the London Borough of Newham. I do not quite understand the relationship. One appears to be independent, and the other appears to be well and truly part of a local authority.

Jaimin Patel: We are a majority-owned employee mutual, so 51% of the shares are owned by the employees via an employee trust. It is a well-accepted structure. We are an independent company. The London Borough of Newham is a minority stakeholder in every sense of the word, including the control and the controls that it applies in our day-to-day running and management of the company.

Sam Lingard: The Language Shop Ltd is incorporated at Companies House as a limited company. Does that help?

The Chair: Yes. While we are at it, so that we are absolutely clear, you are also a director of Elite Interpreters and Translators. Is that right?

Jaimin Patel: That is a dormant company. The intention behind it was to support, in partnership with the Chartered Institute of Linguists and the National Register of Public Service Interpreters, the creation of a co-operative model of delivery. We have a lot of expertise in systems and processes to deliver. A project was initiated that was intended to gift that to Elite, the idea being that a co-operative model would come out of that stakeholder and that Elite would be owned by a co-operative model. That is still there. It has not gone anywhere.

The Chair: It is not active at the moment.

Jaimin Patel: It is not active for two reasons. First, there has been a change of direction, predominantly from the Chartered Institute of Linguists and the national register, which were the key partners that we were working with to develop that proposal.

The Chair: That has been really helpful, thank you.

Q95 Lord Carter of Coles: Good morning. This question is about you giving us a description for the record of the quality assurance work that you deliver in the court system. We see this as a very difficult area, so we are looking for clarity. Do you do this for any other service apart from the Courts Service? If you do, can you give us a comparator of the difficulty in the Courts Service?

Sam Lingard: I will answer the last part of the question first. There is no comparative assurance quality contract. However, we do other assessment contracts; we do foreign language assessment work for the Foreign Office and the Ministry of Defence, but this is the first of its kind for quality assurance. Does that answer that part of the question?

Lord Carter of Coles: Yes, it absolutely does.

Sam Lingard: Then I will go on to give a description of the services that we provide.

The contract is quite diverse. I will give you an overview of the various bits of it. I will start with the assessments, which are probably the thing you know or have heard most about. Assessments come in two main types. I will start with what our contracts refer to as mystery shopper assessments. They are sampling assessments of live interpreting. So for court interpreting, in the vast majority of cases our assessors observe, from the public gallery, interpreters working live in courts and, where possible, record an assessment against prescribed criteria. We attempt those on 1% of bookings, which equates to about 2,000 assessments a year across all the services.

We also make some of those mystery shopper assessments from recordings. We can talk about that a bit later. We would like to it a bit more from recordings.

The other type of assessment is an in-person assessment—an IPA—again as described by the contract. That is a simulated role-play assessment that we carry out. If somebody fails a mystery shop, they are required to undertake this second assessment. It is effectively an exchange between an English speaker and a speaker of the relevant language in an MoJ context that is then interpreted by the interpreter and assessed by our assessor.

The other way in which an in-person assessment can be triggered is through a complaint. If there is a complaint in relation to the quality of the interpreter, we will trigger an assessment and require that interpreter to undertake that assessment. That is a bit of an overview of the assessments.

We also deliver a trainee scheme, which we can talk a bit about. The idea there is that we support the supplier—in this case, thebigword—to recruit new interpreters at level 6 in languages that they are struggling to fulfil the most. We do that through targeted head-hunting—finding appropriate candidates, asking them to apply, and grant-funding half of their DPSI law qualification to support them financially. We audit thebigword and the other supplier. We carry out annual on-boarding audits to ensure that they are complying with the terms of their contract in respect of qualifications, security clearances and other requirements.

The last one I will mention is the customer survey. We carry out a customer survey when asked by the MoJ. The last one was about a year ago. These are designed to gather feedback on the performance largely of the supplier. It is well responded to: we had over 1,000 responses to that last time from within the MoJ.

Lord Carter of Coles: What have you learned in the course of the contract? How many years have you been doing it now?

Sam Lingard: Eight years.

Lord Carter of Coles: So, over the eight years, how have you improved it? Do you think that the quality assurance has got better?

Sam Lingard: Absolutely. It was the first contract of its kind and the first time that we had done it—the first time that anybody had done it, really. When we started the contract, we were learning. The MoJ was learning about what it wanted and what did and did not work. We worked with it really closely. It has been quite an involved contract to make it achieve its intended outcomes around making successful assessments, how we make assessments and all the other areas where we support the MoJ with, say, expert advice on quality, qualifications and all those sorts of things. I would definitely say that it has been a real, continuous improvement journey.

Jaimin Patel: It may also be worth mentioning that, when the MoJ designed and set up the scope of the contract for Lot 4 for the first time, it was through extensive engagement that took place across the industry; this included language professionals, language service suppliers and regulatory bodies. That is how the scope that Sam has illustrated came about. It was not that the MoJ internally devised the contract; it followed extensive external stakeholder engagement.

Lord Carter of Coles: Do you have any international comparators? These are challenges that every court system in the world faces.

Sam Lingard: Not that we are aware of.

Q96 Lord Mott: I just want to follow up on your response to the question about other government departments that do not have the same quality assurance system in place. Do you see anything—best practice—in those other government departments that could be applied? More importantly, do you see anything that you currently have in place, in terms of your MoJ contract, that could be rolled out and applied across other government departments?

Sam Lingard: We would love to do more of this across other services, whether in local government, the NHS or other parts of central government. We were the supplier selected to add quality assurance to the Crown Commercial Service’s interpreting and translation framework, which a lot of public service bodies use, but there has been no uptake at all by other parts of government.

Obviously, there are lots of reasons for that. There is a cost to delivering it and budgets are stretched. Practically, it is challenging to deliver—we can talk about that a bit more—but we are here to overcome those practical challenges and work around them. We would like to do more of it; it would be great if we could.

Q97 Lord Willis of Knaresborough: Just one little thing: you seem to be talking about your job as having been a success. What I do not understand is that roughly half of the court cases that involve interpreters fail because they do not turn up or there is a problem, and half of those do not even get reported as a problem, so how can you call it a success?

Sam Lingard: We do not have any role in improving the fulfilment of the contract.

Lord Willis of Knaresborough: That is the point I am making: should you not?

Sam Lingard: What we have done in respect of what is asked of us has been successful, but I recognise that the MoJ has challenges on waste and those sorts of things. We would be happy to assist the MoJ with that side of things but, as you know—we will come on to this, I am sure—waste obviously affects our ability to do our job, in the same way as it does for the suppliers.

Lord Shipley: I have just two facts that I would like to explore in greater depth to understand this better. You do mystery shopping for about 1%.

Sam Lingard: About 1%, yes.

Q98 Lord Shipley: I am not clear how that figure was identified. Why is it not 2%, 3% or 0.5%? In other words, is 1% a safe figure? My second question relates to that. I have not understood how you select the people to be mystery shopped. Who decides where you go?

Sam Lingard: The 1% was specified in the contract from the outset so it was not determined by us. I would say—let us call it by design—that it seems to be around about the right kind of percentage. It is subjective, but I do not think it is far off what would be a reasonable number.

On how we decide which assessments to make, making assessments is challenging in itself because of the court environment. We have to choose bookings where we stand a chance of making an assessment. We have spent a lot of time honing the bookings that we choose to assess. That factors into which assessments we make, but the driving force is about who has been assessed and how recently. We seek to maximise the number of assessments that we make across individuals while at the same time spreading our assessments broadly across the different service types—telephone interpreting, face-to-face interpreting—and across different jurisdictions.

In anticipation of other questions about this, there are some areas where we cannot make assessments. We do not make assessments in closed court; although that has been trialled and we have previously talked to the judiciary about ways of doing it, it is not currently delivered like that, practically speaking. It is a combination of all those things, but the sampling methodology is ours.

Q99 Lord Prentis of Leeds: Good morning. Sam, you mentioned recruitment work and acting as a recruiting agent in your introduction. Much of the evidence that we have received is about the difficulties in getting interpreters to stay in the service. Many are leaving and there are difficulties in recruitment. Are you experiencing those difficulties in recruitment that people have told us about in past meetings?

Sam Lingard: In terms of what we do with the training scheme, the first thing I would say is that it has been a success. We see that the cost of qualifications is a barrier to people entering court interpreting. The amount of it that we can do is limited by what we are allowed to do and what the MoJ ask us to do in that respect. We would like to do more of it; that is anticipated in the new contract because the MoJ recognises the success here and would like to use the scheme more.

It is really language dependent. There are some languages where we have not found it that difficult to encourage new candidates to come forward, take up the grant and do the qualification. Then, in some languages—for example, languages that are scarcer—it is more difficult. In some of those languages, it is quite difficult to articulate a full-time career, for example. There is sometimes an assumption that an interpreter can come to work in the MoJ and get 35 hours of work a week but, actually, it is a fulfilment problem: there might not be anywhere near that number of hours to keep them sustained in a career. There are challenges with it but, as a system and a process to assist with the recruitment of other suppliers, it is definitely useful.

Lord Prentis of Leeds: Do you have any views on why that is?

Jaimin Patel: Can I just add to that? There needs to be a recognition that net pay has an impact in getting people into the system but it is about retaining them, keeping them there and keeping them invested in their skills, their learning and the equipment that they need to purchase if they do remote interpreting. All those factors have a bearing on how long they stay in this sector.

Perhaps more importantly, for the interpreters who are doing the work at the moment, the hourly rate is not the determining factor; it is the net pay that they take home with them on a weekly basis. I cannot give you any empirical evidence to demonstrate this but there is a feeling that that has been reducing at such a rate that they are now asking, “Is it economically viable for me to continue doing this?”

The other part of this—Sam is absolutely right—is that the training scheme has been successful, but will those who have come in stay? We have made a significant investment to pull them into the sector, but are we drawing them in for a short-term fix?

They will come in and see that it is not sustainable—that they cannot get a mortgage on it, pay their expenses, and attend the courts and bookings that they need to attend to make a living out of it. That is the work that we feel has been eroded significantly, but we cannot give you any empirical evidence to demonstrate that, and apologies for that in advance.

The other point that I should make is that it is a false economy to pull people into the sector and register them as interpreters—which is what I think the MoJ recognised in the new contract—if we do not then make a similar investment about keeping them and making it economically viable to keep them.

A final point I should make is that a lot of the interpreters at work in the public sector—the ones that we assess and the assessors that we draw from—are not doing it just for the money. They absolutely love and are passionate about their work. Honestly, they get up in the morning and are not thinking, “How much do I earn?” They are thinking about the problems they are going to fix. When they go in and are exposed to that type of work, the other side of the coin is that they hear really harrowing stories. They take those home with them at the end of the day, at the end of a booking, and have to cope with them because there is no structure for them to be debriefed after. I suppose what I am saying, as regards their well-being and professional counselling, is that they have to pick up the cost of that as well. There is a story that goes beyond the hourly rate, and that feeds into your question about the trainee scheme and how we pull people in and keep them in the industry as a whole, including the MoJ.

Sam Lingard: I echo those points and say that the trainee scheme works within the framework in which we are operating. It is a useful tool. But, to echo exactly what Jaimin says there, it needs to be within a framework of support.

Lord Prentis of Leeds: You have answered my second question, so thank you for that, but just one thing sticks in my mind. Where do the interpreters go if they want to continue working? Are we in competition with anybody else where they can get other interpreting jobs with higher pay and better conditions?

Sam Lingard: The police is an obvious competing area.

Jaimin Patel: There are competing sectors for the MoJ interpreter. The police pay a two-hour minimum and I think that the MoJ is now moving to a two-hour minimum standard, specifically to help with fulfilment. But I should also say that, probably since Covid and remote working, more competing elements have been introduced. An interpreter can pick and choose, and be more selective about the work they can do. Because they are location-agnostic, there is a greater demand and a smaller pool, which is not helping with fulfilment. That is, once again, anecdotal. The interpreters are highly skilled, have degrees, are qualified and can do other work. That is what we are competing with—not necessarily with other sectors. Over the next 24 months, what we are actually competing with is whether they leave and go and do something completely different. They have transferable skills and are highly competent professional, high-performing individuals. They choose to work in this sector because it fulfils their values and gives them satisfaction.

Sam Lingard: And economically—

Jaimin Patel: Economically, there is a there is a pressure to think, “How do we continue to pay for this?”

The Chair: That is really helpful.

Q100 Lord Porter of Spalding: Going back to the mystery shopping bit: I am assuming you have some sort of matrix that you either will not share or might share with us, which shows the type of languages and cases involved, with a sense that it is 1% across the spectrum, not 1% of the easy ones.

Sam Lingard: Yes, we can share. We share all that data with the MoJ, to which we are contracted and, subject to its agreement, we are happy to share. It is a combination. It is not about being easy to do but about successfully getting a result that has a value.

Lord Porter of Spalding: You said something after I was going to ask a question. Do you do exit surveys of people who used to do interpreting and then stopped? Or do you not know that they have stopped?

Sam Lingard: We do not do anything formal.

Jaimin Patel: That is really outside our scope. I should imagine that that would be the contractor, thebigword, which provides that service. It will know whether the interpreters have stayed or left, and the reason for leaving. They might still be registered with thebigword but are doing other work. It is really difficult for us to know those aspects of it.

Q101 The Chair: Just before we go on to the next question, there is a follow-up. You mentioned a survey in your introductory remarks. Who does that go to? One of the recurring themes throughout the evidence sessions has been that people who work in the courts, like judges and barristers, do not feel that they ever have an opportunity to feed back on quality. It is not their job. They are busy people with things to do. Does your survey include asking them how to think about the quality of the service, or is it just interpreters?

Sam Lingard: No, it is a survey within the MoJ. It is not for the interpreters. It is for the staff in courts and anywhere else. We take from thebigword the email addresses of everybody who has made a booking, and ask them to forward the survey on. It is not a closed survey just to the people who have been invited. They can forward it on to anybody else who wants to complete it and anonymously submit feedback through the survey. So absolutely; that survey could be forwarded to a judge who could provide feedback.

The Chair: Do you have a breakdown of who replies and what their role in the courts system is?

Sam Lingard: We do not because it is, at the moment, an anonymous survey with anonymous feedback.

The Chair: Darren, we covered the main question. Do you want to ask the second supplementary? That is a matter that we have not addressed.

Q102 Lord Mott: That makes perfect sense. I am also interested in what Lord Willis’s follow-up is because I may have a follow-up on that question. Sam, we drifted into this but could any systems of quality assurance be more effective?

Sam Lingard: The main one is getting more access to recorded material from within courts to make assessments, because a lot of court hearings are recorded in one form or another and we do not have full access to those. There is potential access to live remote hearings as well. That would help us deal with some of the waste around an assessor travelling for three hours to a hearing that is then adjourned when they arrive. Those definitely would be the main things I would suggest. I think that that is recognised by the MoJ in the way in which it has structured the next contract. It anticipates trying to do more of those remote assessments.

Q103 Lord Mott: How does remote interpreting affect quality assurance of interpreting in courts?

Sam Lingard: It makes it more practically deliverable. We can readily assess remote interpreting.

Jaimin Patel: Could I just add to that? Remote live presents similar challenges that you would see from a normal video interaction, when there are mic issues, issues around speakers, issues around the way in which people are laid out in a room, and issues where the interpreter cannot see the participants, which therefore affects their ability to interpret faithfully. Those elements will have an impact on the way in which we do our work from a quality-assurance perspective for a live remote session. It is important that we adapt to that and take note of it in the way in which we issue our results and consider the feedback.

I say that because it is not as if you can take what we are doing and simply apply it without adapting it for a remote live session. We are conscious of that and would work with the MoJ and the necessary parties to make sure that we pick up on those elements of it. It is also important that we try and find a way to adapt things, particularly in those type of settings where the microphone, the audio and the equipment have not supported the interaction as well as they could, because that is not necessarily wholly the performance of the interpreter.

Lord Mott: We have heard that from other sessions. The interesting proposal that was put to us at last week’s session is whether we need to completely think again about the format of courts, rather than just layering on technology in a setting that, in the first instance, is not fit for purpose. We need to reimagine to make sure that remote work could be done and be far more effective.

Jaimin Patel: We wholeheartedly echo that. If you see interpreting taking place, even in a simple interaction with an interpreter, you can see how hard they are working just to hear who is speaking and the context of delivery. It is incredibly challenging. It is not as if it is in this space, when they would be able to see, absorb, hear and transfer in real time. The error rate has to be higher, if that equipment and that foundation are not set properly, so we definitely echo that sentiment. It is easier said than done, but would definitely make a big difference.

The reason I wanted to say that is because there is an assumption that, if you put it on to remote live, we might get a result, but it creates another set of challenges and problems that we need to deal with. We need to be conscious that, just because the interpreter has spoken, it does not mean that we will be able to issue a result from that channel of delivery.

Q104 Lord Bach: You may want more video remote—I understand that we are talking video remote—but what is your preference between having someone in court watching what is going on and someone watching on video? We have heard evidence, and it is common sense, that you have to be in the courtroom to know whether the interpreter is any good or not. I can see how it is easier and perhaps cheaper to do it remotely, but are you not making a sacrifice as well?

Jaimin Patel: We have to trial it for a number of reasons. We have to test the outcome and pilot it carefully, and the reasons are twofold. We talked about keeping people in the industry—that is one of the comments that we made—and remote interpreting is a key way of giving interpreters the hours to make sure that we keep them in the sector. That is really important: the number of hours is critical, not the hourly rate. The hourly rate is important and needs to be looked at, but the net pay that ends up in the interpreter’s pocket is the critical factor.

From my perspective, and I am talking personally here, the only way to do that which is viable in the short term and will deliver results is via remote interpreting. It allows interpreters who are quality assured to get that work, take enough work and hold on to their net pay. If that is the channel that helps support retention, we will need to adapt our quality assurance to make sure that it is effective. Whatever difficulties are presented, it needs to be a channel that we quality assure accurately and effectively, so that we can deliver results.

I am under no illusion: while it might be more challenging, we will find a way to get the right results in place and make the right decisions, and we will put in the hard work to make that happen. We will do it through consultation, partnership and engagement; that is what we need to do to get the right result. The answer to your question is that I do not think there should be a preference. They both need to deliver a result that can be trusted.

Q105 Lord Willis of Knaresborough: Thanks for that positive response to that question and to what Lord Mott said about the issues we heard about and some of the advice that we were given last week. The reality is that new technologies, particularly AI, will dominate virtually every area of public sector operation, whether we like it or not. I would be very interested to know how you, The Language Shop, are considering the use of AI tools in interpreting more broadly. What is your approach to it, before I ask you a more a specific question?

Jaimin Patel: Do you mind if I allow Sam to answer the question from a quality-assurance perspective? For your reference—I am sorry; we should have said this earlier, Chair—

Lord Willis of Knaresborough: I was going to come back to quality assurance, to be honest.

Jaimin Patel: The reason I want to say this is that, within The Language Shop, we operate almost a Chinese wall, because of the role that quality assurance plays in the business. Sam leads as the contract director and senior responsible officer on the MoJ contract and will be able to share that part of it, but I can respond to the wider AI part and on how I see it.

Lord Willis of Knaresborough: That is what I want first.

Jaimin Patel: I will do that first, in that case. There needs to be clear recognition, irrespective of what the public sector feels, that it is being adopted on the front line, at the moment. I am not talking about the MoJ, because it is not my area of expertise, but in councils and the NHS we know that Google Translate, for example, has been used on the front line to provide interpreting. It is here, it will be used and it needs to be incorporated, but I have several concerns.

We need to be cognisant of the impact it will have in a number of ways. One is the error rate that results just from different languages. You can see loads of examples with Spanish or German where it is working really well but, with Kurdish, to support a midwife with a delivery tomorrow morning, it just does not have the accuracy to help at this time.

To answer your question, I think it should be phased. There should be clear milestones that are adopted before the next phase of AI is introduced. My preference would be to see it as an assistive technology in the first instance to help support the work of all the stakeholders, in a controlled environment, so that it is not being used without the type of controls that we would expect. The moment you put it in, it goes up to the cloud and breaches of confidentiality and trust will be taking place.

My answer to your question, which is not very good—I apologise for that—is that I would like to see it focusing wholly on assistive technologies, in the first instance. That is where it is going to end up.

Lord Willis of Knaresborough: Are you basically saying that you are not doing anything at the moment?

Jaimin Patel: We are not doing anything at the moment.

Lord Willis of Knaresborough: Let me move on to quality assurance.

Sam Lingard: I will give you a quick answer on quality assurance. We do not have any plans to use AI in the delivery of quality assurance any time soon. However, the role of quality assurance can be quite powerful in the phased introduction of AI, when we get to that point, because it demonstrates a layer of assurance. This is not something that we are doing at the moment in interpreting, but it could be used as a tool to ensure that, where it is being used, it is being used safely and is not compromising the quality of interpreting. But it is not used in the quality-assurance process.

Lord Willis of Knaresborough: I find that very disappointing, to be honest, because we are clearly in a situation where AI and other technologies—not just AI—will play an ever more significant role in maintaining quality services. If you are not doing anything at the moment to introduce that, either by training people or within the operation, then that is a big issue that we should address to the Government.

Sam Lingard: I accept the challenge. It is something that, as you rightly say, is down the road, and not too far down the road. I am just giving an honest answer about where we use it in our quality-assurance process at the moment.

Jaimin Patel: I would say that we have the first embryonic—there is no point pretending that it is anything other than that—conversation taking place with an AI developer next week, to understand how we might invest in this area to improve our capability. It is the start of the journey and the first conversation we are having. The conversation will be around how we can use it to improve the planning of the work that we are doing. The honest answer is that buying consultant support will help us shape what we should be looking at and how we can use it. That first conversation is taking place next week.

Lord Porter of Spalding: There seems to be a reticence to go anywhere near this, on the basis that this particular service is so important. But the health service is starting to use it and there is no more important service from a user’s perspective than the health service, because “Oh, sorry, we got it wrong and cut your leg off” is not fixable. “Oh, sorry, you’ve got a fine” is fixable. Surely if the health service is starting to embrace this technology, your industry must embrace it.

From a consumer’s perspective, yours is a pretty safe place to put it in. It is just language. The consequences are not as dire as they would be in the health service. I was with somebody last week—I shall not say who they are, but it was international crime stuff. They use this technology to decide whether they are going to start investigations into people. People’s lives will be changed on the basis of it. If you do not engage with it, you will be so far behind the curve that you will always be playing catch-up with it.

I just find this difficult. Phil and I do not normally agree on much, but, on this area, not being in the middle of the pack is a bad place to be. It is fine not to be right at the front of it, but not in the middle.

Jaimin Patel: Can I explain our reticence? First, we are a small business, and it is the amount of money we will have to invest in this. Is it a question of, “Let’s see how this develops before we back a horse and see where that goes?” I will be perfectly honest with you: there is a large part of that in the decisions that we are making in relation to seeing where it goes. We are taking that early step to try to understand how much we can invest and where we should invest and see what we can do. That is starting from next week, as I said, to try to buy in a consultant’s support and skill.

My reticence is that I would like to see the success and efficiency gains that it will deliver so that we can bring it in and have a clear business case and a rationale about how we introduce it into The Language Shop and what benefits we will see as a direct result of that introduction.

Lord Willis of Knaresborough: You are missing the point. The point, which you both accepted earlier, is that unless you are part of that process rather than waiting for someone else to do it so that you can respond to it, you will miss the bus. That is the point that we are trying to make.

Jaimin Patel: I do not think we have missed the bus. That is the point I am making. I think we are having the conversation early enough to start it, so I do not think that we are missing it. I think we are starting at just the right time to try to understand how it will benefit The Language Shop, what investments we need to make, and how we understand that.

Over the next six months, the ideas that we create will formulate a business plan that starts looking at the next three years. A core part of that is the threat—I call it a threat, but it is also an opportunity—that this presents. Early adopters, early stakeholders who come out with the right mix of AI and delivery in the language sector, will be the ones who will survive the AI revolution.

We do not have deep pockets. Last year, 76% of our language revenue was paid straight out to the linguists, because that is the commitment that we make. That does not leave much else for making investments in the future. We are careful about how we do it. That is the honest answer. We are not shying away from it, but it is about careful expenditure, careful planning. My thoughts are that we will do that over the next six months.

The Chair: Okay. That has been really helpful and a good discussion.

Q106 Lord Bach: Can we move on to something else that is perhaps also slightly controversial: the data that you publish?

Obviously, under the contract there is some very limited data that you have to publish, and that you do, about how many complaints there are, if there are complaints to you in less than 1% of cases where you provide an assessment, the 10% rule on appeals, and all that. Cutting to the chase, are there not other areas where you could publish data that would help this whole system? It is clear from your evidence so far that you are interested in helping the whole system—the MoJ, the interpreters, all of us.

Where are the figures on, for example, the number of times or the percentage of cases where you have identified problems with interpreting services, the severity of those problems and what action you have taken? Why can we not see those published? Why can nobody see that published data? It would be very useful to know that. To call it an accusation is putting it a bit high. The question is that the published data of assessments does not provide a picture of the quality of interpreting across the board. It appears to capture much more the performance of you as provider. Could you not add appreciably to the published data that you provide for the benefit of yourselves and the system?

Sam Lingard: Yes, we could, because we collect the data. If this sounds like I am passing the buck, I apologise, but we provide much more data to the MoJ than it publishes. What gets published is not our decision; it is a decision for the Ministry of Justice. So I would say that we have nothing to hide in respect of the data that we have. If the ministry is happy to publish more data from our service, then absolutely.

The Chair: Do you collect data that you are not asked to collect as part of the contract? Do you only collect what you are required to collect, or do you go beyond that?

Sam Lingard: We probably go beyond that, but we capture a lot more data for the contract than you have seen. We would not need to capture more data for you to see more data. It would just need to be published.

Lord Bach: You would be happy to do that yourselves, but you say that it is not a matter for you; it is the MoJ that decides what data is published.

Sam Lingard: That is right. It publishes the data on our behalf. We do not publish any data ourselves directly. It takes what we give it and it publishes the data.

Jaimin Patel: We do not publish anything outside those direct controls.

Sam Lingard: We do not publish anything. The MoJ publishes all the data on our behalf. We give it lots more data than is published.

Lord Bach: It is not very satisfactory, is it? Do you agree?

The Chair: That is a leading question.

Lord Bach: It is. I am happy to ask questions.

Sam Lingard: I think there would be value in there being more visible information.

The Chair: Very well put.

Baroness Stedman-Scott: Just to support that point, it is common in all government contracts.

The Chair: You mean that more data is collected than published.

Baroness Stedman-Scott: Yes. When I was contracting with the DWP, we were not able to publish data without the DWP’s approval. You may have read Margaret’s Hodge’s book. I went before a committee where she said, “So they stopped you doing it?” and I said yes. So to be fair—.

The Chair: Do you see a pattern in the data that is published and the data that is not published? Does it fall into two groups quite easily, or are you quite surprised at what is and is not published?

Sam Lingard: As I have said, I do not think that a great deal is published, but what data is published and what is not published is all determined by contract beforehand. We supply the information we are asked for by the MoJ.

The Chair: That is very interesting. I have not heard that at all previously, so thank you. That is very helpful.

Q107 Baroness Stedman-Scott: Could you give us your view of the current quality of interpreting services offered in the courts? Can you tell us something about the qualifications that interpreters should have? What is your particular role at The Language Shop in supporting interpreters to gain further qualifications and developing their careers?

Sam Lingard: The quality of interpreting is a very broad question, and little bits of data might help to illuminate it. The quality of interpreting is better than it was at the start of the contact. The pass rates across all services were around 70% when we started and are now closer to 85%. I am giving you a little bit more data there. Hopefully that is useful.

Lots of people have worked hard to improve the performance, but that is in no small part down to the work we have done to quality-assure and improve the services. They have definitely improved. There is definitely room for improvement. We have submitted suggestions to the MoJ on how we might pivot the service more towards supporting continuous improvement rather than just identifying very poor performance. We see our role more as being, “This was good enough, but here are some suggestions on how you might be able to perform better”.

Baroness Stedman-Scott: How did they respond to that?

Sam Lingard: I think it got tangled up in the new contract, but the MoJ has reflected that ambition in the new terms of the contract.

Baroness Stedman-Scott: Could you make our day by sharing with us the recommendations you made?

Sam Lingard: I would have to ask the MoJ for that. I think that has been picked up in the new contract, and there is a sense that we want to be more supportive of interpreters, who do an incredible job in very difficult circumstances. We do not want to be seen as a negative force that is trying to make their job harder, sitting at the back of a court and with the potential to take away their—

The Chair: Cost deadline.

Sam Lingard: Yes. We would prefer to be able to do something that is more supportive of them, in the collective interest of supporting and growing the interpreting community that we know is under pressure for various reasons.

The Chair: You have seen that there are qualifications that are desirable for each of the three categories of court cases, but from the evidence we have been taking we understand that those qualifications are not essential to have if there is no one available who has that qualification.

Has your assessment picked that up? What damage is being done by people without the recommended experience or qualifications acting in particular roles? We know it happens, but we do not have a feel for how often it happens and what the consequences are.

Sam Lingard: Those who have lesser qualifications are definitely more likely to underperform with respect to the standard we would expect. That does not mean that everybody who does not hold a DPSI level 6 qualification is a poor interpreter, but in our view the level 6 DPSI is the right standard for court work. That is the level that the interpreting needs to be delivered at, in the main.

The Chair: You are aware of what percentage of those you assess have that qualification, because that is one of the questions you ask when assessing them.

Sam Lingard: Yes, we have information on who holds what qualification.

The Chair: Does that mean you could tell us or the MoJ how many cases are using an interpreter who does not have the necessary level of qualification? You have that information?

Sam Lingard: Absolutely. The MoJ could do that, and thebigword could do that. They are aware of the qualifications their own interpreters have. That information is available.

The Chair: We have not seen that.

Q108 Lord Mott: My question goes back to Lord Willis’s question and the witnesses’ response to that. Has anybody in the MoJ discussed with you whether, in quality assurance terms, they would put an AI pilot alongside your existing contract so that the Government themselves—not just you or the MoJ—can understand this? I am becoming more concerned, and I completely understand the commercial reality that you described, that when we talk about two years, three years, five years, the reality is that we will be in a completely different space. So I just wonder whether anybody at any stage is starting to have that conversation with you.

Sam Lingard: Nobody has had a conversation with us about AI and how we might use it in the performance of quality assurance. That is a straight answer to that question. I know we have covered this, but we would certainly be very open to having that conversation. It is definitely not a reluctance on our part. It is just that it is not a conversation that has been had yet.

Jaimin Patel: Whether it is with us or someone else, there is a role for the quality assurance to validate the results of the AI-driven outcome, not from a fluency perspective but from the perspective of whether it actually transferred the words correctly. That is really important in this space particularly. You would be assessing against a different set of measures according to the subject area specialism. That would be a useful thing to start planning alongside. I think that is what you are saying: that the introduction of AI should have some quality framework underpinning it as part of the planning process before they even start. That makes absolute sense, because the outcomes still need to be validated.

Lord Willis of Knaresborough: So it does with languages as well.

Jaimin Patel: Absolutely.

Lord Willis of Knaresborough: If you are doing a language from central Africa, one of 200 different dialects, I do not believe you have anybody who could go into a court and say, “I listened to that and it was all right”.

Jaimin Patel: I do not think either solution is perfect, but you do not want to transfer one set of problems from one environment into another. All I am suggesting is that while you have the opportunity you should create the framework to understand it.

You would have the same problem assessing the AI in a particular rare language that you describe, because you would be looking to buy in a skill set that is in short supply from the outset. There are the same challenges, but it should not prevent us from trying to put in a high-quality assurance framework to make sure we are heading in the right direction and are not seeing unintended consequences of the direction of travel that we are proposing. That is what I would say.

Q109 Lord Willis of Knaresborough: I am glad that Lord Mott asked that question, because I had hoped I would have a minute at the end to ask you about it. In the health service we have Lord Darzi, who back in 2008 to 2010 was producing an NHS system that looked at highly specialised experimental areas for doing work, particularly in the assessment of disease and conditions. That was picked up again by the Conservative Government, particularly in the last couple of years, and particularly under the last Prime Minister, who agreed totally with what Lord Darzi had said earlier. We now have Lord Darzi doing the new NHS research in exactly the same way. If the Government decided to create AI and technology courts—in other words, specific courts in specific areas that just used AI and technology—would you be supportive of that?

Jaimin Patel: That is a really difficult question.

Lord Willis of Knaresborough: Forget the MoJ. Just give us an answer.

Sam Lingard: From a quality assurance perspective, we would be really interested in using our skills to make sure that that worked. I made the point earlier about using our quality assurance. Yes, a role for AI in quality assurance might come down the line, but given that using AI for savings and the rest of it is about where the money is spent—in the interpreting sector, that is on the interpreting—if we can provide the assurance by having, in the first instance, human quality assurance over something innovative like AI, it could be really powerful in that context

Jaimin Patel: From our perspective it all boils down to building the right controls in the right space. It is almost too easy sometimes to look at this as a homogenous group. Sometimes we are better off saying that it is out of scope for that language group, and we will not use machine translation because that undermines the rest of the programme where we can be really efficient, where we have the dataset, where the languages can work and where the stakeholders can engage with it in an effective manner.

I would go back to the quality control framework that says that these are the languages we should try and these are the languages we perhaps stay away from. So yes, is the answer, but the learning and the thinking should be language-specific and subject-specific so that we can be confident in the results and not undermine the efforts of the AI by having it do something where it is set up to fail from the outset because we have given it too big a challenge to get right. We want to embrace it, but we would want to put some controls in.

Lord Willis of Knaresborough: I knew we would end up as friends.

The Chair: We have started this discussion, and Lord Laming will continue it.

Q110 Lord Laming: In your guide to quality assurance, which you kindly sent us, you rightly put an emphasis on assessment, learning from assessment and the work of the assessors. Could you tell us how you choose your assessors, how they are trained and how they are supported?

Sam Lingard: All the assessors have experience of working as court interpreters. We think that is really important, because if they do not have that experience they cannot evaluate it. Lots of them come from some assessment background, so they might be assessing for the DPSI, for example, and they are the most experienced of the language professionals in the sector in their particular language. Obviously, we have to have that cross-section of languages. That is generally their background.

We train them ourselves. It is a thorough training process, so we give them a training course and then we spend quite a lot of time with them one to one. The whole process, before they work on live assessments, may take two or three weeks. In that time, they have at least two or three one-to-one sessions. We give them test assessments, before they work on live assessments, to make sure they understand the process that we are applying and that they can do that work effectively. If we are not happy with that, we will redo it until we are happy that the assessor can do the job that we ask of them.

We encourage dialogue and support our assessors all the way through the process. We look at all the assessments that they send back to us and we will engage with them if we want to discuss anything in the assessment. We also encourage them to engage back whenever they want and we will always make ourselves available for discussions and further support of the assessors.

Lord Laming: That is most helpful. Are you able to share with us how your assessors are paid?

Sam Lingard: We are happy to share with you. We were discussing this and I think we would prefer to share it afterwards, because it is quite a compact service and we have a range of pay. That sort of suggests that there is something to hide, but I would say that we pay our assessors really well. As well as the hourly rate being what we think supports a language professional, the terms are better than they might get across the rest of the interpreting sector.

Baroness Stedman-Scott: How do you know that? How do you know your pay is better?

Sam Lingard: As a provider ourselves, we understand exactly what is being paid in the industry. There is also some public information out there about what interpreters are being paid.

Baroness Stedman-Scott: I am not trying to catch you out. I am sorry; I should have asked the Chair to come in.

The Chair: Herbert, did you get an answer to your question?

Lord Laming: From what you have said, would we be right in assuming that you have no difficulty recruiting assessors?

Sam Lingard: We are very flexible in what we pay, so we do not struggle generally to recruit assessors, but it is very difficult to recruit them in certain very rare languages. That is a challenge; it is just a difficult thing to do. However, because of our pay and conditions, we generally do not have a problem finding assessors.

Lord Laming: May I ask a final question? The picture that I have—please correct me if I am wrong—is that, if you needed to do an assessment in court of someone interpreting in French or German, you would have no difficulty getting an assessor. But, if you were assessing somebody speaking in one of the African languages, you would have difficulty.

Sam Lingard: Yes, we would.

Lord Laming: So your assessment is geared towards the European languages and away from the others.

Sam Lingard: I would not say that it was geared towards the European languages, but it is a fair statement that it is geared towards languages that are used more frequently, as it is easier for us to get those assessors. However, we have a broad spread of assessors. For some of the very rarest languages, there might be only two or three working interpreters in the country, so it would not be practical for us to find an assessor.

The Chair: They know each other as well, do they not?

Sam Lingard: They know each other as well, exactly.

Jaimin Patel: The challenge sometimes is not just that they know each other, but that you are asking them to assess each other. They are happy to do the interpreting work, but not the assessment work, if that makes sense.

Baroness Stedman-Scott: I am sorry; I apologise for interrupting. I am just interested. I do not dispute that you are paying more and do not have trouble recruiting people, but the issue is that you are losing people, as everyone in the industry is. You are not alone in that. Would you be better putting a premium on the more difficult languages?

Sam Lingard: We do. We pay more for languages that are more difficult to recruit. There is flexibility across our rates of pay to account for that.

Baroness Stedman-Scott: And despite your efforts, you still lose people. What would it take to stop losing people?

Jaimin Patel: I think we have high retention rates among assessors. When I was talking earlier about retaining and holding on to people, I was talking about drawing them from the training scheme to the interpreting pool that we then use. It was about the retention of those interpreters; our retention rate for assessors is above 90%. It is very good.

The Chair: Are they on contract with you—you perhaps do not need to answer on this level of detail—or do you pay them per assessment? Do you employ them directly?

Sam Lingard: It is per assessment—a freelance model.

Q111 The Chair: This is the last question. As you know, the report will make recommendations to Government and we want to give you a chance to say what recommendations you think we ought to include. We would be grateful for one or two each; we do not need a long list.

Sam Lingard: Go on, then. I will be a stuck record with some of the other people who have said that it is about recognising interpreting as a profession, treating interpreters as professionals and rewarding them as professionals. That is the only way that we will have a sustainable service going forward. Pay and, where required, interventions in the public system of procurement will make sure that it is not a race to the bottom. That is my recommendation.

Jaimin Patel: I have a couple. The fixation on hourly rates of pay takes away from the discussion about what actually ends up in their pockets. I would try to move the conversation to what recommendations can be made to ensure that a DPSI interpreter, who spends time, money, effort and CPD to hold on to that learning, can maximise their revenue. It is about the safe introduction of remote sessions and remote interpreting, making sure that somehow it is mandated—and I do not know how you mandate it—to the level 6s, because they are the ones who we want to retain.

At the same time, we already have a pipeline in the system so that they aspire to that level 6. We are trying to say, “If you put the learning and effort in, we will give you more work”. We cannot do that overnight. It is about creating that pathway to success, so that that keeps working effectively.

Another thing—and this may be controversial—is that it is easy to say that people who do not have level 6 do not have a role to play, but the reality is that they are here. The opportunity should be to look at which are the best and try to nurture the training development opportunities so that they can become level 6. We need to create the aspirational journey that creates that throughput and pipeline.

The Chair: It is a sense of ambition.

Jaimin Patel: Absolutely, but it needs to be a cohesive model that works together with and involves the MoJ, the stakeholders, the voice of the interpreters and the language service supplier, so that bookings are allocated based on an outcome that is not purely about fulfilment in the short term. It is about fulfilment in the long term and using the right interpreters over a period of time. To protect the industry and to safeguard the work that the court system does, that is my suggestion.

The Chair: That is very helpful and it reflects the comments that you have made throughout. That was a fascinating session and we have all improved our understanding of what happens. We are very grateful to you for your time; we know it has been particularly busy for you recently with the tender, so we are grateful to you for agreeing to come. Thank you.