Ministry of Justice – Written evidence (ITS0077)
House of Lords Inquiry into Interpreting and Translation Services in the Court
Written responses to the Committee following Officials Evidence Session
Table of Contents
8. Foreign National Prisoners AI Pilot
10. Responses to the committees requests for more information on data
Around 11:27 – We could write to the committee on renumeration after procurement is complete (We will save this as an action for later)
Would be interesting to know if there are any plans to introduce a minimum rate for interpreters (i.e. a minimum amount the provider would have to pass on.
1. MoJ considered setting a minimum rate for interpreters during market engagement based on stakeholder feedback. The findings of that analysis were that the rate paid to interpreters itself, was competitive when benchmarked with other government departments (OGDs) and other public sector authorities. The main difference between MoJ and OGDs was the application of other terms such as the minimum duration an interpreter is paid for. A decision was made not to mandate a minimum rate but instead to increase HMCTS face to face bookings to a two-hour minimum.
2. Other considerations included not wanting to create an internal market among OGDs and other public sector authorities. As MoJ is the biggest user of interpreting in the public sector, any significant changes made to terms would likely have a disproportionate impact on the costs for the public sector as other departments would have to raise and implement comparable minimum rates.
3. In terms of interpreter renumeration expressed as an average annual figure this is particularly difficult when it comes to interpreters as they are freelancers who often operate a portfolio career covering various public sector bodies and other private sector work alongside the work they undertake in the courts.
11:51 – How many languages are used and what are the percentages of cases in each language (wanting to know the major and smaller languages)
Please see attached documents:
4. Appendix A: Q2 response – All requested languages by volume and percentage
Data period Jan23-Dec23
The total language bookings taking into account all MoJ bookings so differ from just the HMCTS bookings captured in the statistics published on Gov.uk.
5. Appendix B: Q2 response – Regional Overall volumes
Data period Jan 23- Dec 23
6. Two of the regions listed are referred to as CTSC and NCES. The National Compliance & Enforcement Service (NCES) is the former name for the part of HMCTS which is responsible for the collection and enforcement of the financial penalties imposed by the Crown Courts and Magistrates Court. For the purposes of the language services contracts, it is treated by HMCTS as a distinct region. The CTSCs (Courts and Tribunal Service Centres) is the name for the range of national operations carried out within HMCTS which sit outside of the regional structure and support HMCTS frontline colleagues by – amongst other things - making all interpreter bookings for first hearing at magistrates’ courts (where a defendant is not in custody), Family Private Law cases and specific tribunals.
7. This information has also been shared publicly as part of the Invitation to Tender documents which can be found here: https://ministryofjusticecommercial.bravosolution.co.uk but does not form part of our routine published statistics series on Gov.uk
12:09 – Write back on the specifics as to whether trials can be affected by interpreters using the wrong word (although this was then answered by Mark Stewart, so let me know if an answer is still needed)
If the QA provider says the interpreting wasn’t up to scratch, what happens with the specific case. I believe Mark/Kris said this would be flagged to the relevant judge. I suppose we want to know how many times this has happened (if any) and if it’s ever led to a judge revisiting the relevant case/how it impacts the case. It would also be useful to understand if such concerns are always flagged to judges, or if it’s decided by the MoJ whether the misinterpreting is bad enough. If the latter, how is this decided?
It would also be helpful to understand how this impacts case records, for example - if a witness was mis-interpreted, and this was identified by the Language Shop’s people, would that then be reflected in the record of the case – or would the misinterpreting that happened on the day be the official record? Would the parties to the case be informed about the concern being raised?
8. There are no recorded examples of mistrials or miscarriages of justice caused by interpreter quality. The Criminal Cases Review Commission have searched their records of cases referred to the Court of Appeal and have found no cases which relate to issues with interpreting or translation. Where an issue regarding the quality or accuracy of interpretation arises at a hearing (whether identified by one of the parties or their representatives) the matter should be raised with the judge at the hearing so that it can be addressed there and then. The judge will consider the nature of the issue being raised and whether it is material enough to impact on live proceedings; where it is felt to be material, the judge is likely to make appropriate directions or other steps to address (for example request alternative interpretation provision is sought). Any such deliberations and decisions would typically be picked up in the court transcript, as a record of proceeding, or be reflected in substantive decision providing a more official record.
9. Please note - witness interpreters are generally booked via the CPS although they sometimes use MoJ contracts to source an interpreter (their use represents less than 1% of contract bookings).
12:11 – Write on which languages complaints procedures are in and how people can be supported to make complaints
10. In line with the requirements of the Welsh Language Act, HMCTS and the suppliers provide facilities to submit complaints in English and Welsh. Complaints information available on Gov.uk and suppliers’ websites can also be translated through a number of online translation tools such as Google translate which works by entering the website and language translation required.
11. We have noted that there may be issues with court users who are not able to communicate in English or Welsh but wish to make a complaint. We are therefore looking at what amendments could be made to the complaints process to support non-native English speakers.
12. Ease of access to complaints process:
13. It is also easy to google search how to make a complaint. I.e. A search of the words ‘thebigword + MoJ + Complaint’ brings up the supplier’s website as the first search where feedback can be submitted online or via the contact details provided on the page - Ministry of Justice | Feedback
14. We also make internal staff guidance available on how to make a complaint via our intranet.
15. All complaints that are received through official channels enable thorough investigation to take place and provide a source of feedback (alongside our other channels i.e. through surveys and stakeholder groups) to enable us to action changes where necessary to make improvements.
12:11 – A breakdown of who makes complaints (court staff, judiciary, defendants, legal representatives etc)
16. This is not currently data that MoJ routinely collate as part of the published series.
12:12 – Write on the qualification level required for mystery shoppers/ inspectors
17. Whilst there are no qualification levels specified in the current contract, the Language Shop have specifications for the role set out below and a comprehensive training regime for quality assessors.
18. There are qualification levels in the new contract which is currently being procured. Details can be found here https://ministryofjusticecommercial.bravosolution.co.uk and in the appendix attached. Please see Appendix C Q4 Extract of Assessors qualification and requirements.
Current Requirements from The Language Shop
19. Quality Assessor Profile - Quality Assessors will be qualified interpreters and teachers or assessors. Wherever possible they will be required to have the following:
Certify Impartiality and Conflict of Interest
20. All assessors are required to agree and sign terms and conditions in relation to carrying out Quality Assessments. As part of the Mystery Shopping process, the quality assessor will certify both impartiality and that there is no potential conflict of interest in carrying out the allocated assessment. For face-to-face assignments this is done prior to attending the assessment; they are required to sign the Conflict-of-Interest certificate which includes their confirmation of providing an impartial assessment. For telephone and video Mystery Shopping Assessment, this will be completed at the start of the assessment once it is known who the Language Professional (interpreter) carrying out the interpreting is.
12:14 – Write accurately on travel pay
21. We pay the supplier an agreed fixed price to manage the service, which includes travel expenses as per the supplier’s rate card.
22. In practice, the supplier sets its rate card to fulfil bookings, based on their knowledge of the market, and by what the supplier believes will best encourage interpreters to accept in-person assignments. This level of detail falls entirely within their discretion and whilst this could have an impact on their profit, suppliers are incentivised to ensure face to face bookings are fulfilled, via KPIs and service credits (financial penalties).
23. The new contracts are set up in the same way - potential bidders have been provided management information showing the geographical spread of bookings, the percentage of bookings that take place face to face and, within their price submission, have specifically been asked to consider travel costs.
24. The reason for this approach is that suppliers are the experts in the industry and have the knowledge and experience to cost services accordingly with the understanding of what is needed to recruit and retain interpreters at the level required for the contract.
25. It is also worth noting that as interpreters are self-employed they are eligible to claim business expenses when collating their tax return which may include travel through HMRC. This will also impact the calculations of renumeration for interpreters and is another example of the complexity that can best be taken into account by market experts. Information on what can be claimed can be found here: Expenses if you're self-employed: Car, van and travel expenses - GOV.UK
12:14 – write on the specifics of whether 2 hours pay is guaranteed even if the case is cancelled
On this, some finer detail – You could plausibly have an interpreter booked for the full day, then on the day it’s decided it’s an ineffective trial because of e.g. prison van issues – do they only get paid for the 1 or 2 hour minimum, or do they get paid for the time they were booked for?
What is the timeline on cancelation and payments – we appreciate the rules on reimbursement might reasonably be different if you’re cancelling 2 weeks before a trial rather than 2 minutes after it was due to start.
There are two types of cancellation that can be made:
Short notice cancellations
26. This refers to when a cancellation of a booking is made by the MoJ after 09:00 on the working day immediately prior to the date for the booking or where the interpreter attends the booking, but the booking does not proceed due to no fault of the Supplier or interpreter.
27. In these cases, a cancellation fee would be provided. 95% of the cancellation fee is provided to the interpreter. The cancellation fee is determined by the supplier in their bid for the services.
Standard Cancellations
28. If a booking is cancelled prior to 9am on the working day prior to the date for the booking no cancellation fee is payable. Cancellations are not linked to the 2-hour minimum booking.
12:14 – write on the details of the contractual remedy/ penalty if lack of interpreter causes an expensive trial to collapse
In general - What can the MoJ do if the provider isn’t meeting KPIs / requirements set out in the contract – what does robust contract management entail here, and have the MoJ ever used penalties or other measures in the 12 years since outsourcing began?
29. Yes, the MoJ have applied service credits as compensation (or a financial penalty) during the life of the contract. The frequency and amounts payable by the suppliers are commercially sensitive.
30. The MoJ monitors suppliers’ performance which is measured through a suite of KPIs which define target performance levels and are linked to remedies in the contracts. Performance points are assigned on the basis of whether a service failure is minor, moderate or critical. Based on the points, a calculation of the service credit is made which is payable by the supplier. Service credits, if applicable, are applied on a monthly basis.
31. Under specific circumstances as defined in the contracts, a supplier’s performance may trigger a notifiable default. If this occurs, the rectification plan process is triggered which requires the supplier to draft and submit a rectification plan to HMCTS. Once the plan is agreed the actions to remedy the default take place immediately.
32. Performance failures as defined in the contract:
“6. PERFORMANCE FAILURES
6.1 Without prejudice to any other rights and remedies the Authority may have, if during the Contract Period:
6.1.1 a Service Failure and/or Non-Fulfilment occurs, Service Credits shall be invoiced to the Supplier by the Commissioning Bodies or the Authority in accordance with Schedule 7;
6.1.2 a Notifiable Default occurs, the Supplier shall comply with the Rectification Plan
Process (in addition to the Service Credits accruing in accordance with Schedule 7);
and/or
6.1.3 a Supplier Termination Event occurs, the Authority may exercise its rights to terminate the Contract in whole or in part pursuant to Clause 51.1.”
33. Full redacted contract here: Contracts for Language Services - Contracts Finder
Specifically – when there is an ineffective trial due to language service failures, are there financial penalties for thebigword / the specific interpreter, and what are they?
34. Within the contract, the financial consequences for the supplier for not providing an interpreter are triggered by failing to meet a specific threshold of fulfilment (not providing enough interpreters to satisfy our bookings). The fulfilment metric is an overall, national performance measure, not one that is applicable to a specific instance such as a trial. If a specific trial was affected, the court may consider making a costs order against a supplier if there has been a serious service failure. A party can apply to court for it to make such an order. If an interpreter was booked and then failed to attend, there may be consequences for that interpreter in line with their terms and conditions for accepting a booking from the supplier.
It would also be good to understand what the average cost of an ineffective trial is in the Crown Court, if possible?
35. An Ineffective Trial is one where on the trial date, the trial does not go ahead due to action or inaction by one or more of the prosecution, the defence or the court and a further listing for trial is required.
36. We are unable to provide a defined figure for the average cost of an ineffective trial as the impact of a case not taking place as planned can be mitigated by the courts in numerous ways which will vary the costs incurred. Whilst we do know the average unit cost for disposing of a crown court case was £2496 in 2023/24, this is an average of the direct costs of staffing and judiciary costs only and excludes other costs such as estates costs and technology. This also does not determine the cost of an ineffective trial as when a trial is ineffective the costs of judiciary and staff are not lost but rather will be transferred to other activities in the time that the original case would have taken up.
Can we have a copy of the assessment made by the MOJ on different models pre-tender (the assessment that included the proposal for bringing the service back in-house).
37. Please see at Appendix D both the redacted version of the delivery model assessment (DMA) which can be used for publishing purposes and Appendix E the full version. Documents have been appropriately labelled as either ‘official sensitive – recipient only’ or ‘official – for public release’.
38. The DMA was completed in line with best practice guidance published by Cabinet Office. The project did not inherently fall into the category of requiring a DMA but due to best practices and stakeholder feedback, a proportionate model was completed.
39. The DMA consists of three parts, the long list of options that were considered (Appendix F) the evaluation of the short-listed options (Appendix G) and a make vs buy (Appendix H) which considered the financial aspect.
40. The DMA contains figures, costs and dates that were accurate as of the time of drafting. Supplier costs have been redacted due to the commercial sensitivities of the ongoing procurement.
Members were also keen to have more information on the exercise being undertaken with foreign national prisoners and AI. This would include details of a timeline of the exercise, how it will be evaluated and the risk assessment.
41. In 2022, discovery work on language-based AI solutions for the next generation Language Service contracts concluded that technological advances and the use of AI in courts without careful control, could harm people within the justice system. Because of this, responsible use is at the heart of our approach to AI, particularly in such a complex environment with significant risks attached. We did recognise that exploratory and developmental work should continue however to mitigate missed opportunities for future interpreter provision. As a stop-gap, the new suite of language services contracts specifies successful bidders must actively participate in activities led by MoJ to drive future innovation. This includes digital/AI solutions.
42. As part of the subsequent developmental work, we have started a Proof of Concept (PoC) exercise to utilise AI technology supplied by an existing HMPPS supplier to provide an automated AI interpretation tool with transcription capabilities in a sample of prisons which house a high percentage of non-English speaking prisoners. These establishments face operational pressures due to communication barriers, whilst prisoners themselves experience difficulties because they rely on interpretation to be understood. The prison environment will enable us to test the technology in a controlled environment and provide invaluable insight into the readiness of the AI and operational viability in a safe and practicable way.
43. Working closely with HMPPS colleagues we have developed the PoC which will provide a digital interpretation tool loaded onto a hand-held device or existing HMPPS computers. We aim to commence the PoC in January 2025 for a six-month test period, which will be followed by an evaluation period. Interpretation will be provided for safe, non-legal discussions between prisoners and HMPPS officials. For example, a day-to-day verbal discussion where a prisoner wants to raise an issue can be quickly facilitated by the interpretation tool without the need to book an interpreter via the existing contract.
44. The technology has been trained on the 15 most in demand languages currently in prisons with accuracy numbers above 90% for some languages. The interpretation platform will allow for private non-legal discussion such as the initial First Night Assessment process where a prisoner has first arrived at an establishment.
45. We are currently completing an AI risk assessment for the PoC to ensure that the use of the technology is appropriate. Guidelines will stipulate that if at any time during the use of the AI interpretation tool there are challenges in understanding, or there are potential legal ramifications, standard operating procedures are implemented, and officials must immediately halt the use of the AI interpretation tool and acquire the services of a human interpreter from the language services contract.
46. Evaluation will comprise of a wide range of metrics including feedback from the prisons, performance of the system and lessons learned to inform further developmental work for language-based AI.
47. To support the PoC, the MoJ Innovation fund will cover the full cost of the supplier costs for the PoC exercise.
Following the completion of the PoC and assessment, the next steps will be, but not limited to:
48. The PoC will focus on metrics such as potential time savings, value for money, accuracy, and operational effectiveness. The goal is to create a more efficient, cost-effective, and equitable service, providing valuable insights for future implementation across the MoJ.
49. Medium to long term testing will provide us critical data on the practicality, effectiveness and value of Digital and Technological solutions in language services which will be used to roll out the service wider with significant opportunities to reduce cost and improve the quality of interactions between people and the MoJ.
50. The preparatory work to engage staff with this pilot is underway with focus group discussions with ‘first night’ staff having taken place recently.
Members also had some questions on the published data:
Criminal court statistics - GOV.UK
As this data has not yet been published we are unable to share this yet but hope to publish this shortly.
We are not currently able to separate these out based on the existing raw data that we currently receive.
If the former, can they give us the courts data specifically?
51. The KPI’s reported to the Cabinet office can be reviewed using this link: Key Performance Indicators (KPIs) for government’s most important contracts - GOV.UK
52. For quarterly data please see Published KPI data spreadsheet forwarded as Appendix I.
53. Contracts for spoken, translation and transcription, visual and tactile, and quality assurance are all reported on separately. KPIs are specific to individual contracts.
54. KPIs are measured holistically across the MoJ and are not disaggregated to court level.
55. To note Clarion (the Visual and Tactile service) are missing from the last quarter due to an admin error but were rated as ‘good’ across all three KPIs. They will be included in future publications.
Other data questions from the committee:
56. Data published on interpreting and language services in the courts is typically found in the criminal court statistics, although it covers all jurisdictions. The data is published quarterly.
57. These have been postponed so the latest data available covers Oct-Dec 2023.
Can the MOJ provide any more up to date info?
58. Not at this time. The release of the latest statistics has been postponed but we are hoping to publish updated figures shortly.
59. In June, we notified users of the cancellation of the Criminal Court Statistics quarterly publication, highlighting work that was necessary to assure the quality of source data for the key series. That initial work highlighted some necessary changes to court processing systems to enable robust and accurate data to be produced. These fixes do not affect the operation of the courts and are purely an issue for the data reporting.
60. These are being implemented alongside work to align the MoJ Official Accredited statistics and HMCTS management information methodologies. This will benefit users by providing greater transparency and coherence in court data.
61. While the fixes being implemented will lead to some changes in headline measures of the Criminal Court Statistics, they are unlikely to have a material impact on trends of the key series, for example Crown Court receipts, disposals and outstanding caseloads. The work will affect the back series, and the largest impacts are likely to be seen since 2021.
62. We are working as fast as possible to deliver robust data series at the earliest opportunity and by the end of the year and will keep this page updated with our future plans.
The data reports interpreter services and translation services together as ‘language services’ – are MOJ able to separate this out?
63. Not based on the existing raw data that we currently receive.
The data includes reports of complaints from different courts – but not the type of person that lodged them e.g. judge, HMCTS staff, legal representative, non-English speaker – do MOJ gather this data.
64. This is not currently data that MoJ routinely collate as part of the published series.
Do MOJ gather any data from The Language Shop on their quality assurance work, and is it published anywhere?
65. We receive a quarterly return detailing the complaints received by The Language Shop.
66. We ensure that complaints received from The Language Shop are included in the complaints data from theBigWord.
67. These complaints are added to those made to suppliers, with any duplicates being identified and removed, and are published in the quarterly MoJ statistics.
Are MOJ able to disaggregate any data by geographic area?
68. Data is centrally collated to the geographic region. These can be found in Appendix B by language and region.
What KPIs are they currently collecting data on? What’s in the new contract?
69. In the current contract we measure:
70. In the new contracts, following the Cabinet Office’s model form terms and conditions we have both KPIs and SPIs (subsidiary performance indicators). We will still measure fulfilment, portal availability, response to the helpdesk and number of complaints. The changes made to these KPIs are in the calculation, performance levels and service credits. We have added further measures as SPIs for complaint response time, portal functionality, social value and fulfilment by an alternative (meaning if any of the requests made within the booking are changed to an alternative e.g. face to face is requested but remote is delivered).
71. Please see Published KPI data spreadsheet forwarded as Appendix I